Heme Oxygenase Protects against Placental Vascular Inflammation and Abortion by the Alarmin Heme in Mice

Both infectious as non-infectious inflammation can cause placental dysfunction and pregnancy complications. During the first trimester of human gestation, when palatogenesis takes place, intrauterine hematoma and hemorrhage are common phenomena, causing the release of large amounts of heme, a well-known alarmin. We postulated that exposure of pregnant mice to heme during palatogenesis would initiate oxidative and inflammatory stress, leading to pathological pregnancy, increasing the incidence of palatal clefting and abortion. Both heme oxygenase isoforms (HO-1 and HO-2) break down heme, thereby generating anti-oxidative and -inflammatory products. HO may thus counteract these heme-induced injurious stresses. To test this hypothesis, we administered heme to pregnant CD1 outbred mice at Day E12 by intraperitoneal injection in increasing doses: 30, 75 or 150 μmol/kg body weight (30H, 75H or 150H) in the presence or absence of HO-activity inhibitor SnMP from Day E11. Exposure to heme resulted in a dose-dependent increase in abortion. At 75H half of the fetuses where resorbed, while at 150H all fetuses were aborted. HO-activity protected against heme-induced abortion since inhibition of HO-activity aggravated heme-induced detrimental effects. The fetuses surviving heme administration demonstrated normal palatal fusion. Immunostainings at Day E16 demonstrated higher numbers of ICAM-1 positive blood vessels, macrophages and HO-1 positive cells in placenta after administration of 75H or SnMP + 30H. Summarizing, heme acts as an endogenous “alarmin” during pregnancy in a dose-dependent fashion, while HO-activity protects against heme-induced placental vascular inflammation and abortion.


Introduction
Cleft lip and palate (CLP) is a serious orofacial birth defect affecting approximately 1/750 newborns worldwide with ethnic and geographical variation [1]. CLP is associated with various problems related to facial growth, feeding, speech, hearing and psychosocial status [2]. Palatogenesis is a dynamic and The observed total abortions corresponded with the body weight of the plugged mice between Days E12 and E15 (see Figure 2A). As expected, compared to the non-pregnant mice, the pregnant mice showed an increase in body weight (p < 0.001), whereas mice that demonstrated total abortion showed body weight reduction (p < 0.01). No adult mice died after heme and/or SnMP administration. The treated plugged mice demonstrated normal behavior, and the different experimental groups scored only a low to mild degree of discomfort.
Fewer fetuses per pregnant mouse in the 75H group (p < 0.01) were obtained compared to the controls (see Figure 2B). In the other experimental groups (30H, SnMP and SnMP + 30H), the fetal number was also lower compared to the controls, but this trend did not reach statistical significance (p > 0.05). The observed total abortions corresponded with the body weight of the plugged mice between Days E12 and E15 (see Figure 2A). As expected, compared to the non-pregnant mice, the pregnant mice showed an increase in body weight (p < 0.001), whereas mice that demonstrated total abortion showed body weight reduction (p < 0.01). No adult mice died after heme and/or SnMP administration. The treated plugged mice demonstrated normal behavior, and the different experimental groups scored only a low to mild degree of discomfort.
Fewer fetuses per pregnant mouse in the 75H group (p < 0.01) were obtained compared to the controls (see Figure 2B). In the other experimental groups (30H, SnMP and SnMP + 30H), the fetal number was also lower compared to the controls, but this trend did not reach statistical significance (p > 0.05).

Placental Weight Increase after HO-Activity Inhibition
Considering that downregulation of HO-1 and HO-2 expression in humans' placenta is associated with placental pathology [34,[36][37][38], and placentas from HO-1 +/− mice were lighter than those from wt mice [43], we studied the effect of heme administration and the absence or presence of HO-activity at the mouse placental weight. The placentas were heavier in the SnMP group (p < 0.001) compared to the groups without SnMP administration, i.e., control, 30H and 75H groups (see Figure  2C). A higher placental weight was found in the SnMP + 30H group (p < 0.01) compared to the control and 30H groups. Adult mouse body weight comparison, fetal number, fetal body weight and placental weight affected by HO-activity inhibition and heme administration. (A) Bar chart for the adult mouse body weight comparison, the percentage (%) of the weight change between Days E12 and E15. Unfortunately, the weight at Day E15 was unintentionally not collected from two pregnant mice (75H group) and two mice that demonstrated abortion (SnMP + 75H and 75H groups). Eventually, the body weight comparison of the pregnant mice (n = 22; control n = 6, SnMP n = 4, 30H n = 4, SnMP + 30H n = 5, 75H n = 3), not pregnant mice (n = 5; SnMP = 1, 30H n = 2, SnMP + 30H n = 2) and mice that demonstrated total abortion (n = 7; 75H n = 1, SnMP + 75H n = 3, 150H n = 3) was calculated. Data are shown as mean ± SD. (B) Box-and-whisker plot with 10-90 percentiles of quantitative assessment of the fetal number (number of fetuses per pregnant mouse of the different groups). Number of pregnant mice per group; control n = 6, SnMP n = 4, 30H n = 4, SnMP + 30H n = 5, 75H n = 7. Number of fetuses per group: control n = 91, SnMP n = 56, 30H n = 43, SnMP + 30H n = 58, 75H n = 46. (C) Box-and-whisker plot with 10-90 percentiles of quantitative assessment of the placental weight of the different groups (control n = 91, SnMP n = 56, 30H n = 43, SnMP + 30H n = 58, 75H n = 46). Box-and-whisker plot with 10-90 percentiles of quantitative assessment of the (D) fetal body weight and (E) fetal body length of the fetuses of the different groups (control n = 91, SnMP n = 56, 30H n = 43, SnMP + 30H n = 58, 75H n = 46) were measured. * p < 0.05, ** p < 0.01, *** p < 0.001.

Fetal Body Size Increase after Inhibition of HO-Activity and Heme Administration
Because downregulation of HO-1 expression in human placentas is associated with pregnancy complications and mouse fetuses from HO-1 +/− pregnancies were smaller than those from wt pairings [43], the effect of HO-activity inhibition and/or heme administration on mouse fetal body size was examined. Administration of SnMP increased the fetal body weight (p < 0.001) and fetal body length (p < 0.05) compared to the controls (see Figure 2D,E, respectively). Administration of 75H also significantly increased fetal body weight (p < 0.001) and fetal body length (p < 0.001) compared to the controls. A representative fetus per group is shown in Figure 3A. [43], the effect of HO-activity inhibition and/or heme administration on mouse fetal body size was examined. Administration of SnMP increased the fetal body weight (p < 0.001) and fetal body length (p < 0.05) compared to the controls (see Figure 2D,E, respectively). Administration of 75H also significantly increased fetal body weight (p < 0.001) and fetal body length (p < 0.001) compared to the controls. A representative fetus per group is shown in Figure 3A.

Palatal Fusion despite Inhibition of HO-Activity and Heme Administration
As activation of pro-inflammatory pathways increased the risk of palatal clefting in humans [6,7,44], the effects of inhibition of HO-activity and/or heme administration on palatal fusion was studied in mice. The palatal shelves were fused in the fetuses at Day E16, and disintegration of the MES was found in palatal sections from fetuses of both the control and the different experimental groups: 30H, 75H, SnMP and SnMP + 30H. A representative HE stained palatal section per group is shown in Figure 3B. In the surviving fetuses, no effects on palatal fusion was found despite inhibition of HO-activity and/or exposure to heme.

Palatal Fusion despite Inhibition of HO-Activity and Heme Administration
As activation of pro-inflammatory pathways increased the risk of palatal clefting in humans [6,7,44], the effects of inhibition of HO-activity and/or heme administration on palatal fusion was studied in mice. The palatal shelves were fused in the fetuses at Day E16, and disintegration of the MES was found in palatal sections from fetuses of both the control and the different experimental groups: 30H, 75H, SnMP and SnMP + 30H. A representative HE stained palatal section per group is shown in Figure 3B. In the surviving fetuses, no effects on palatal fusion was found despite inhibition of HO-activity and/or exposure to heme.

Inhibition of HO-Activity Reduces IL-1β Expression in Placental Blood Vessels
Others found that IL-1β induction is counteracted by HO-1 activity during acute inflammatory arthritis in mice [42]. Here, the effects of exposure to SnMP and/or different heme doses, respectively, five and four days after administration, on IL-1β expression in mouse placenta were studied. Lower numbers of IL-1β positive blood vessels were found after administration of SnMP (median = 2.91/mm 2 ) compared to the control (median = 11.0/mm 2 ; p < 0.01) and 30H group (median = 13.1/mm 2 ; p < 0.001) (see Figure 4A,B). Lower numbers of IL-1β positive blood vessels were also found after administration of SnMP + 30H (median = 4.13/mm 2 ) compared to the 30H group (median = 13.1/mm 2 ; p < 0.01). arthritis in mice [42]. Here, the effects of exposure to SnMP and/or different heme doses, respectively, five and four days after administration, on IL-1β expression in mouse placenta were studied. Lower numbers of IL-1β positive blood vessels were found after administration of SnMP (median = 2.91/mm 2 ) compared to the control (median = 11.0/mm 2 ; p < 0.01) and 30H group (median = 13.1/mm 2 ; p < 0.001) (see Figure 4A,B). Lower numbers of IL-1β positive blood vessels were also found after administration of SnMP + 30H (median = 4.13/mm 2 ) compared to the 30H group (median = 13.1/mm 2 ; p < 0.01).  ; SnMP + 30H n = 7; 75H n = 5). * p < 0.05, ** p < 0.01, *** p < 0.001. A minimum of five sections from a minimum five placentas per group was assayed.

Heme-Induced ICAM-1 Expression in Placental Blood Vessels Is Counteracted by HO-Activity
Since we and others previously demonstrated that heme can induce adhesion molecule expression in vascular endothelial cells [26,45,46], we examined whether heme exposure and/or HO-activity inhibition could alter the expression of ICAM-1 in placenta in mice. Higher numbers of ICAM-1 positive blood vessels were found after administration of 75H (mean= 6.6/mm 2 ; p < 0.05) compared to the controls (see Figure 4A,C). In addition, more ICAM-1 positive blood vessels were found after administration of SnMP + 30H (mean = 10.3/mm 2 ; p < 0.001) compared to the control (mean = 1.43/mm 2 ), 30H (mean = 2.76/mm 2 ) and SnMP groups (mean = 2.87/mm 2 ).

Placental Macrophage Recruitment Increases after Heme Administration and Decreased HO-Activity
Since ICAM-1 expression was increased following exposure to 75H and SnMP + 30H, we assessed whether there was also enhanced influx of macrophages in the activated endothelium. Downregulation of HO-activity in human placenta is associated with inflammatory cell influx [34,36]. We therefore studied the effects of HO-activity inhibition and heme administration at the placental macrophage influx in mice. Higher numbers of macrophages were found after administration of 75H (mean = 4.6/mm 2 ) compared to the control (p < 0.05), 30H (p < 0.001) and SnMP groups (p < 0.001) (see Figure 4A,D). Increased macrophage influx was also found after administration of SnMP + 30H (mean = 5.3/mm 2 ; p < 0.001) compared to the control (mean = 1.6/mm 2 ), 30H (mean = 0.8/mm 2 ) and SnMP groups (mean = 0.6/mm 2 ).

Placental HO-1 Expression Increases after Heme Administration
Since heme, the substrate of the HO system, was found to induce HO-1 expression [32,34,38], we investigated whether heme administration resulted in upregulation of placental HO-1 expression. Higher numbers of HO-1 positive cells were found after administration of 75H (median = 10.0/mm 2 ) compared to the control (median = 0.7/mm 2 ; p < 0.001) and 30H group (median = 1.5/mm 2 ; p < 0.05) (see Figure 4A,E). Higher numbers of HO-1 positive cells were also found after administration of SnMP + 30H (median = 7.4/mm 2 ) compared to the control (p < 0.001) and 30H group (p < 0.05). Notably, HO-1 positive cells were predominantly found in the perimetrium, the thin outer epithelial cell layer of the placenta.

Discussion
This study demonstrated that mimicking intrauterine hemorrhage/hematoma by i.p. administration of the endogenous "alarmin" heme, at Day E12 in pregnant mice, causes abortion/resorption in a dose-dependent fashion. No effect on the fusion of the palatal shelves was found in the surviving fetuses. Since we found a fetal loss rate of 50% in the 75H group, we were only able to study palatogenesis at Day E16 in the limited number of surviving fetuses.
Harmful effects of heme exposure have also been found in humans since uterine hematomas are associated with increased risk at intrauterine growth restriction, preterm delivery and miscarriage [47][48][49][50]. In our study, HO-activity rescued fetuses from heme-induced abortion. Administration of 75H led to fetal loss in half of the fetuses, however, blocking the HO activity by SnMP prior to 75H administration resulted in total abortion. Blocking of HO-activity by i.p. injection of zinc mesoporhyrin in mice earlier during pregnancy at Days E0 and E3 [51], E4 and E6 was previously found to increase the abortion rate [36,51]. In mice undergoing abortion, downregulation of both HO-1 and HO-2 was found in placental tissue compared to normal pregnant mice at Day E14 [52]. We previously found fetal growth restriction, severe malformations and craniofacial anomalies in HO-2 KO fetuses at Day E15 [40]. By contrast, upregulation of HO-1 by cobalt-protoporphyrin [36], or exposure to carbon monoxide [53] a product of heme break-down, rescued abortion-prone CBA/J × DBA/2J fetuses from abortion.
Notably, in our study no increased abortion was found after i.p. injection of SnMP at E11 alone. By contrast, the fetal body size was even increased after HO-activity inhibition, suggesting that HO-activity might be more crucial during implantation compared to fetal development at a later stage. Interestingly, others showed that adult HO-2 KO mice were obese, induced by disrupted metabolic homeostasis, caused by insulin resistance and elevated blood pressure [54]. Furthermore, in HO-2 KO mice increased brain edema was observed after intracerebral hemorrhage [55]. In rats resuscitated from cardiac arrest, induction of HO-1 by hemin reduced brain edema, improved neurologic outcome [56]. In contrast, SnMP administration reduced intracerebral mass in an intracerebral hemorrhage model in pigs by decreasing both hematoma and edema volumes [57]. Although the literature shows conflicting results, we cannot rule out that edema formation contributed to the fetal body size increase after HO-activity inhibition. However, we show here that HO-activity is also crucial for protecting against the injurious actions of heme at later stages during embryonic development.
In human placental endothelial cells, exposure to LPS was found to induce IL-1β expression within 24 h in vitro [72,73]. Increased levels of IL-1β were found in placental endothelial cells following exposure for 24 h to trophoblast debris from preeclamptic placentae in vitro [74]. In human pregnancies with reduced fetal movement increased placental expression of IL-1β was observed compared to the controls, although some controls also demonstrated some IL-1β expression [75].
Expression of the pro-inflammatory cytokine IL-1β was observed in blood vessels in the placentas of the controls, as well as in placentas exposed to 30H and 75H. Recent data show that heme exposure triggers the NLRP3 inflammasome pathway, inducing IL-1β production in human endothelial cells in vitro [76]. Intrauterine infections are able to trigger the expression of IL-1β, which stimulates uterine contractions [77], by elevating PGE 2 levels and myometrial contractility [78]. By contrast, HO-1 was found to inhibit IL-1β induction during acute inflammatory arthritis in mice [42], however, we observed that HO-1 inhibition by administration of SnMP decreased placental IL-1β expression in mice. Unfortunately, a limitation of our study was that the effects of SnMP and heme administration were only evaluated five and four days, respectively, after administration. This means that we were not able to evaluate the expression of IL-1β shortly after SnMP and/or heme administration, with the risk that this expression has been dampened.
Both administration of 75H and SnMP + 30H increased placental vascular ICAM-1 expression and macrophage influx. However, no increases in ICAM-1 and macrophage influx were found after administration of heme 30, showing that heme-induced ICAM-1 expression is dose-dependent and likely related to scavenging by Hpx or HO-mediated breakdown at lower concentrations. When ICAM-1 expression together with the number of macrophages in placenta was elevated in sound-stressed mice, increased abortion was found [79]. Moreover, decreased abortion was observed following i.p. injection of monoclonal antibodies to ICAM-1 in mice, indicating that ICAM-1 facilitates immunologically-mediated abortion [80]. Increased levels of ICAM-1 were found in placental endothelial cells following exposure to trophoblast debris from preeclamptic placentae in vitro [74]. In women suffering from Morbidly Adherent Placenta (MAP), an abnormal adherence of the placenta to the myometrium, massive hemorrhage can occur, resulting in severe morbidity and mortality and increased ICAM-1 expression [81]. In addition, in serum [82][83][84][85] and placenta [85,86] from preeclampsia patients, increased ICAM-1 expression was found.
Placental inflammation in uncomplicated human gestations was only present in 4% of the cases [87]. Intrauterine infection induces pro-inflammatory cytokines, particularly IL-1β and TNFα, and influx of macrophages in placenta and uterus, and it may lead to premature birth [77]. In mice, initiated preterm labor by intrauterine infusion of LPS provoked a massive influx of neutrophils in the myometrium [88].
Macrophage influx in placenta and myometrium is associated with placental inflammation and preterm labor in human, mouse and rat [89][90][91].
Our data show that HO-1 expression in placenta increased after administration of both 75H and SnMP + 30H. HO-1 positive cells were predominantly found in the perimetrium [92], indicating that besides the vasculature free heme molecules were able to reach the developing fetus through diffusion. In HO-1 + / − mice induction of HO-activity in placenta by administration of Pravastatin was shown to improve placental function and fetal survival at E14.5 [93]. In human pathological pregnancies, low expression of HO-2 stimulated migration of inflammatory cells to the feto-maternal interface, possibly caused by enhanced serum levels of free heme [34]. We speculate that in our experiment low levels of free heme were scavenged by Hpx [31], while the heme-Hpx complex binding to LRP1 would activate cytoprotective signaling by Nrf2 [94] and upregulation of placental HO-1 expression [32,95].
It is likely that in our experiment the sudden exposure to high doses of free heme overwhelmed the scavenger Hpx and the HO system [96], allowing binding to toll-like receptor 4 (TLR4) [46] and activation of nuclear factor kappa B (NF-κB) [97], driving expression of pro-inflammatory cytokines, such as IL-1β [98][99][100].
The novel hypothetical model ( Figure 5) shows heme-induced endothelial cell activation [101], as exemplified by the expression of ICAM-1 [102], facilitating the recruitment of macrophages into placental tissue [102,103], resulting in pathological pregnancy, increasing the risk of abortion. Furthermore, heme-induced inflammation might cause craniofacial abnormalities before resorption of the fetuses would occur. However, our studies could not provide evidence for this statement. Inhibition of HO-activity will increase the heme-induced inflammatory response, albeit HO-activity will attenuate heme-induced inflammation, promoting normal fetal development. binding to LRP1 would activate cytoprotective signaling by Nrf2 [94] and upregulation of placental HO-1 expression [32,95]. It is likely that in our experiment the sudden exposure to high doses of free heme overwhelmed the scavenger Hpx and the HO system [96], allowing binding to toll-like receptor 4 (TLR4) [46] and activation of nuclear factor kappa B (NF-κB) [97], driving expression of pro-inflammatory cytokines, such as IL-1β [98][99][100].
The novel hypothetical model ( Figure 5) shows heme-induced endothelial cell activation [101], as exemplified by the expression of ICAM-1 [102], facilitating the recruitment of macrophages into placental tissue [102,103], resulting in pathological pregnancy, increasing the risk of abortion. Furthermore, heme-induced inflammation might cause craniofacial abnormalities before resorption of the fetuses would occur. However, our studies could not provide evidence for this statement. Inhibition of HO-activity will increase the heme-induced inflammatory response, albeit HO-activity will attenuate heme-induced inflammation, promoting normal fetal development. This study demonstrated that heme acted as an endogenous "alarmin" during pregnancy in a dose-dependent fashion, while HO-activity protected against hemeinduced placental vascular inflammation and abortion. We postulate that heme-induced inflammatory response promoted endothelial cell activation, which upregulated the expression of ICAM-1, facilitating the recruitment of macrophages into placental tissue, resulting in a pathological pregnancy, increasing the risk of abortion and possibly craniofacial abnormalities. Inhibition of HOactivity will increase the heme-induced inflammatory response, albeit HO-activity will attenuate heme-induced inflammation, promoting normal fetal development. This study demonstrated that heme acted as an endogenous "alarmin" during pregnancy in a dose-dependent fashion, while HO-activity protected against heme-induced placental vascular inflammation and abortion. We postulate that heme-induced inflammatory response promoted endothelial cell activation, which upregulated the expression of ICAM-1, facilitating the recruitment of macrophages into placental tissue, resulting in a pathological pregnancy, increasing the risk of abortion and possibly craniofacial abnormalities. Inhibition of HO-activity will increase the heme-induced inflammatory response, albeit HO-activity will attenuate heme-induced inflammation, promoting normal fetal development.

Mice Selection, Mating, Housing, Ethical Permission
To obtain fetuses for this study, plugged CD1 outbred mice, 12-17 weeks of age, were purchased from Envigo, Venray, The Netherlands (n = 31), and from Charles River, Sulzfeld, Germany (n = 7). At the facility of the animal supplier, the female mice were mated with CD1 outbred male mice for 1 day and checked for the presence of a vaginal copulation plug on the next morning, taken as Day 0 of pregnancy (Gestational/Embryonic Day 0, E0) [104]. The plugged animals were transported to our animal facility and housed under specific pathogen-free housing conditions with 12 h light/dark cycle and ad libitum access to water and powdered rodent chow (Sniff, Soest, The Netherlands). The animals could acclimatize for at least 1 week before the start of the experiment. All mice were randomly assigned to the control or experimental groups by drawing lots. Ethical permission for the study was obtained according to the guidelines of the Board for Animal Experiments of the Radboud University Nijmegen (Ethical permission # RU-DEC 2012-166; date: 20 August 2012).

Heme and/or SnMP Administration, Sample Size, Animal Welfare Monitoring
To be able to study palatal clefting, fetuses should be studied beyond the time point palatal fusion takes place. In wt mice the palatal shelves fuse between embryonic Day E14.5 and E15.5, and the capacity to fuse is lost after Day E16 [105]. Fetuses of Day E16 were therefore considered to be suitable for studying palatal clefting.
Tin mesoporphyrin (SnMP) can bind strongly to the HO-1 and HO-2 enzyme, but cannot be broken down by both isotypes, therefore acting as a competitive inhibitor of HO-activity [106,107]. To abrogate the HO system prior to heme administration half of the animals of the 30H and 75H groups received SnMP 30 µmol/kg body weight (later referred to as SnMP) via i.p. injection at E11. In addition, the control group received neither heme nor SnMP, while another group received SnMP only. Heme and SnMP were purchased from Frontier Scientific, Carnforth, UK.
Heme and SnMP were freshly dissolved with Trizma base. The pH was adjusted to pH 7.6-8.0 with HCl. The heme and SnMP solution were filter sterilized before administration.
To detect an effect size of 0.25 (generalized estimation of the reduction in fetal and placental weight in the experimental groups) and a significance level of 0.05 for the 7 groups with power of 0.80, a total sample size of n = 231 fetuses for this study was calculated by the one-way ANOVA power analysis a priori (G*Power 3.1 software) [108]. This indicated that the mean sample size per group should comprise approximately 33 fetuses. We estimated that the litter size could range from 14 to 18 pups, suggesting that each group should contain at least 3 pregnant mice. The chance of conception was assumed to be around 70%, indicating that for each group at least 5 mice should be mated.
Animal welfare was monitored daily during the stay at our animal facility, and the degree of discomfort after administration was scored according to the guidelines of the Board for Animal Experiments of the Radboud University Nijmegen.

Adult Mouse Body Weight Comparison
At embryonic Days E12 and E15, the plugged mice were weighed. The body weight comparison, the percentage of the weight change between Days E12 and E15, of the pregnant mice, not pregnant mice and the mice that demonstrated total abortion was calculated. The status pregnant, not pregnant or total abortion was confirmed by examination of the uterus.

Fetal Loss Rate and Fetal Number Calculation
At Day E16, the animals were killed by CO 2 /O 2 inhalation for 10 min. All fetuses were isolated from the uterus (see Figure 7A). The fetuses were separated from the placentas. For the fetus-carrying mice, the fetal loss rate, the percentage of non-viable and hemorrhagic embryonic implantations to the total number of embryonic implantations (non-viable or hemorrhagic embryonic implantations + fetuses), and the fetal number were calculated.

Placental Weight and Fetal Body Weight and Length
All fetuses and placentas were weighed. The body length of the fetuses was measured on photographs using Fiji Image J 1.51n software (National Institutes of Health, Bethesda, MD, USA) (see Figure 7B). injection. All plugged mice were killed by CO2/O2 inhalation for 10 min at Day E16; the uteri were removed; and the fetuses and placentas were isolated. The first animals that received administration of SnMP + 75H (n = 4) or 150H (n = 3) demonstrated resorption of all fetuses. Since no fetuses could be obtained, the remaining animals of both groups (n = 4) were reassigned to the SnMP+ 30H group and 75H group. The plugged animals were eventually divided into 7 groups as follows: control n = 6, SnMP n = 5, 30H n = 6, SnMP + 30H n = 7, 75H n = 7, SnMP + 75H n = 4, 150H n = 3. In total, 38 uteri were removed. In 1 mouse from the SnMP group, 2 mice from the 30H group and 2 mice from the SnMP + 30H group, no non-viable/hemorrhagic embryonic implantations or fetuses were found in the uterus, and those animals were regarded as not pregnant. Furthermore, in 2 mice of the 75H group, total abortion had occurred. Finally, from 24 mice (control n = 6, SnMP n = 4, 30H n = 4, SnMP + 30H n = 5, 75H n = 5) in total, 294 fetuses and placentas were obtained (control n = 91, SnMP n = 56, 30H n = 43, SnMP + 30H n = 58, 75H n = 46).

Adult Mouse Body Weight Comparison
At embryonic Days E12 and E15, the plugged mice were weighed. The body weight comparison, the percentage of the weight change between Days E12 and E15, of the pregnant mice, not pregnant mice and the mice that demonstrated total abortion was calculated. The status pregnant, not pregnant or total abortion was confirmed by examination of the uterus.  (30,75 and 150 µmol/kg body weight) at Day E12 by i.p. injection. All plugged mice were killed by CO 2 /O 2 inhalation for 10 min at Day E16; the uteri were removed; and the fetuses and placentas were isolated. The first animals that received administration of SnMP + 75H (n = 4) or 150H (n = 3) demonstrated resorption of all fetuses. Since no fetuses could be obtained, the remaining animals of both groups (n = 4) were reassigned to the SnMP + 30H group and 75H group. The plugged animals were eventually divided into 7 groups as follows: control n = 6, SnMP n = 5, 30H n = 6, SnMP + 30H n = 7, 75H n = 7, SnMP + 75H n = 4, 150H n = 3. In total, 38 uteri were removed. In 1 mouse from the SnMP group, 2 mice from the 30H group and 2 mice from the SnMP + 30H group, no non-viable/hemorrhagic embryonic implantations or fetuses were found in the uterus, and those animals were regarded as not pregnant. Furthermore, in 2 mice of the 75H group, total abortion had occurred. Finally, from 24 mice (control n = 6, SnMP n = 4, 30H n = 4, SnMP + 30H n = 5, 75H n = 5) in total, 294 fetuses and placentas were obtained (control n = 91, SnMP n = 56, 30H n = 43, SnMP + 30H n = 58, 75H n = 46). The fetuses were separated from the placentas, photographed and a scale bar was drawn at the centimeter ruler of 10 mm, and the total number of pixels of the bar was determined (e.g., 30H; 100 pixels). A line depicting the length of the body of the fetus was drawn and the number of pixels was recorded (e.g., 205 pixels). The body length was calculated (e.g., 205/100 = 2.05 cm). (C) Placental sections surface area measurement. Coronal sections through the middle of the placenta (e.g., SnMP and ICAM-1 immunostaining). A scale bar was drawn at the ruler in each photograph of 1000 μm and the total number of pixels of the bar was determined (e.g., 201 pixels). The surface of 1000 μm × 1000 μm (1 mm 2 ) included the total number of 40,401 (201 × 201) pixels. The outline of the total placenta was drawn, and the number of pixels was recorded (e.g., 530,523 pixels). Next, the total section surface was calculated (e.g., 530,523/40,401 = 13.1 mm 2 ).

Placental Weight and Fetal Body Weight and Length
All fetuses and placentas were weighed. The body length of the fetuses was measured on photographs using Fiji Image J 1.51n software (National Institutes of Health, Bethesda, MD, USA) (see Figure 7B).

Paraffin Embedding and Section Cutting of Head and Placenta Samples
The fetuses were decapitated, the head samples and placentas were fixed for 24 h in 4% paraformaldehyde and further processed for routine paraffin embedding. Serial coronal sections of 5-μm thickness, through the secondary palate in head samples and through the middle part of the placenta, were mounted on Superfrost Plus slides (Menzel-Gläser, Braunschweig, Germany). (B) The fetuses were separated from the placentas, photographed and a scale bar was drawn at the centimeter ruler of 10 mm, and the total number of pixels of the bar was determined (e.g., 30H; 100 pixels). A line depicting the length of the body of the fetus was drawn and the number of pixels was recorded (e.g., 205 pixels). The body length was calculated (e.g., 205/100 = 2.05 cm). (C) Placental sections surface area measurement. Coronal sections through the middle of the placenta (e.g., SnMP and ICAM-1 immunostaining). A scale bar was drawn at the ruler in each photograph of 1000 µm and the total number of pixels of the bar was determined (e.g., 201 pixels). The surface of 1000 µm × 1000 µm (1 mm 2 ) included the total number of 40,401 (201 × 201) pixels. The outline of the total placenta was drawn, and the number of pixels was recorded (e.g., 530,523 pixels). Next, the total section surface was calculated (e.g., 530,523/40,401 = 13.1 mm 2 ).

Paraffin Embedding and Section Cutting of Head and Placenta Samples
The fetuses were decapitated, the head samples and placentas were fixed for 24 h in 4% paraformaldehyde and further processed for routine paraffin embedding. Serial coronal sections of 5-µm thickness, through the secondary palate in head samples and through the middle part of the placenta, were mounted on Superfrost Plus slides (Menzel-Gläser, Braunschweig, Germany).

Hematoxylin-Eosin Staining of Palatal Sections and Immunohistochemical Stainings of Placental Sections
Selected paraffin embedded palatal and placental sections were deparaffinized using Histosafe (Adamas Instrumenten B.V., Rhenen, The Netherlands) and rehydrated using ethanol (100-96%). Then, the endogenous peroxidase quenching step with 3% hydrogenperoxide in methanol was performed, followed by further rehydration by ethanol 96% till PBS. The palatal sections were routinely stained with Hematoxylin and Eosin (HE). Antigens were retrieved with citrate buffer at 70 • C for 10 min, followed by incubation in 0.015% trypsin in PBS at 35 • C for 5 min. Next, the placental sections were pre-incubated with 10% normal donkey serum (NDS) in phosphate-buffered saline with glycine (PBSG). Primary antibodies (Table 1) for cytokine IL-1β, vascular adhesion molecule-1 (ICAM-1), macrophage marker F4/80 [109] and HO-1 were diluted in 2% NDS in PBSG and incubated overnight at 4 • C. After washing with PBSG, sections were incubated for 60 min with a biotin-labeled secondary antibody (Table 2), as previously described [110]. Next, the sections were washed with PBSG and treated with avidin-biotin peroxidase complex (ABC) for 45 min in the dark. After extensive washing with PBSG, diaminobenzidine (DAB)-peroxidase staining was performed for 10 min. Finally, the nuclei were counterstained with Hematoxylin for 10 s and sections were rinsed for 10 min in water, dehydrated and embedded in distyrene plasticizer xylene (DPX). Microscopic photographs of the HE and immunohistochemical stained sections were taken using a Carl Zeiss Imager Z.1 system (Carl Zeiss Microimaging GmbH, Jena, Germany) with AxioVision 4.8 v software (Zeiss, Göttingen, Germany).
The palatal sections were classified into 4 stages of palatogenesis based on the anatomy of the palatal shelves, as previously described by Dudas et al. [105]. Per individual fetus, the presence of palatal fusion was studied on multiple transversal sections.

Quantification of IL-1β, ICAM-1, F4/80 and HO-1 Immunoreactivity in Placental Sections
Because the transversal sections through the middle of the placenta showed a significant variance in size, immunoreactivity was adjusted to surface area. The specific surface area per placental section was measured using Fiji Image J 1.51n software (see Figure 7C). The number of IL-1β and ICAM-1 positive blood vessels and the number of F4/80 (macrophages) and HO-1 positive cells within the outline of the placental sections were counted, and the mean number per mm 2 per group was calculated. The counting was performed twice, by two observers (C.M.S. and F.A.D.T.G.W.), independently and blinded for the groups. The inter-and intra-examiner reliability was determined.

Statistical Analysis
The data for fetal loss rate, adult mouse body weight comparison, the number of ICAM-1 positive blood vessels/mm 2 and macrophages/mm 2 in placental sections were quantitatively scored and showed a normal distribution as evaluated by the Kolmogorov-Smirnov test (KS-test). The data were analyzed using ANOVA and Tukey's multiple comparison post hoc test to compare the differences between the groups.
The data for fetal number, placenta weight, fetal body weight, fetal body length, the number of IL-1β positive blood vessels/mm 2 and HO-1 positive cells/mm 2 in placental sections were quantitatively scored and showed a non-normal distribution as evaluated by the KS-test. The data were analyzed using the non-parametric Kruskal-Wallis ANOVA on ranks test and Dunn's Multiple Comparison post hoc test to compare differences between the groups.
To determine the inter-and intra-examiner reliability, the coefficient of determination (R 2 ) was calculated by the square of the Pearson correlation coefficient, and acceptable scores > 0.80 were obtained for the counting.
Differences were considered to be significant if p < 0.05. All statistical analyses of the data were performed using GraphPad Prism 5.03 software (GraphPad Software Inc.©, San Diego, CA, USA).

Conclusions
Our results show that heme acts as an endogenous "alarmin" during pregnancy in a dose-dependent fashion, while HO-activity protects against heme-induced placental vascular inflammation and abortion in mice. More research is necessary to unravel the precise relation between hemorrhage-induced placental inflammation, pathological pregnancy and palatal clefting.