Research ArticleReproductive Biology

Nrf2 inactivation enhances placental angiogenesis in a preeclampsia mouse model and improves maternal and fetal outcomes

See allHide authors and affiliations

Science Signaling  16 May 2017:
Vol. 10, Issue 479, eaam5711
DOI: 10.1126/scisignal.aam5711

ROS against preeclampsia

Preeclampsia is the onset of high blood pressure and proteinuria that acutely develops after about 20 weeks of pregnancy, which impairs fetal growth and can result in maternal organ damage. There are few treatment options available. Oxidative stress mediated by reactive oxygen species (ROS) has been proposed to inhibit blood vessel formation (a process called angiogenesis) in the placenta, which would be expected to increase the risk of preeclampsia; however, clinical trials have found antioxidants to be ineffective in preventing preeclampsia. Nezu et al. used a mouse model of preeclampsia in which the antioxidant system Nrf2 had been genetically or pharmacologically manipulated. Unexpectedly, genetic ablation of an inhibitor of Nrf2 or treatment with an activator of Nrf2 decreased placental angiogenesis, suppressed fetal growth, and worsened maternal survival. In contrast, deficiency of Nrf2 increased placental angiogenesis and improved fetal and maternal outcomes. These results indicate that ROS are necessary for the placental angiogenesis that reduces the risk of preeclampsia, and help to explain the negative results of the clinical trials of antioxidants.


Placental activation of the renin-angiotensin system (RAS) plays a key role in the pathogenesis of preeclampsia. Reactive oxygen species (ROS) are thought to affect placental angiogenesis, which is critical for preventing preeclampsia pathology. We examined the role of ROS in preeclampsia by genetically modifying the Keap1-Nrf2 pathway, a cellular antioxidant defense system, in a mouse model of RAS-induced preeclampsia. Nrf2 deficiency would be expected to impair cellular antioxidant responses; however, Nrf2 deficiency in preeclamptic mice improved maternal and fetal survival, ameliorated intra-uterine growth retardation, and augmented oxidative DNA damage. Furthermore, the placentas of Nrf2-deficient mice had increased endothelial cell proliferation with dense vascular networks. In contrast, the placentas of preeclamptic mice with overactive Nrf2 showed repressed angiogenesis, which was associated with decreased expression of genes encoding angiogenic chemokines and cytokines. Our findings support the notion that ROS-mediated signaling is essential for maintaining placental angiogenesis in preeclampsia and may provide mechanistic insight into the negative results of clinical trials for antioxidants in preeclampsia.


Preeclampsia is a clinical syndrome characterized by the acute onset of hypertension and damage to multiple organs (mainly kidneys, brain, and liver) in late pregnancy (13). It affects 3 to 8% of all pregnant women with varying severity (4), accounting for 10 to 15% of perinatal mortality and representing the major cause of intra-uterine growth retardation (5). Although preeclampsia has a marked influence on maternal and infant health, there are no effective treatments that prolong pregnancy without resulting in adverse consequences to the fetus. The only cure for preeclampsia is delivery of the placenta. However, this is associated with iatrogenic prematurity (3, 5). Furthermore, even after delivery of the placenta, a substantial proportion of patients suffer from postpartum complications such as hypertension, cardiomyopathy, and long-term renal disease (69). Although the discoveries of placental-derived anti-angiogenic molecules, such as soluble FLT1 (sFLT1) [a splicing valiant of vascular endothelial growth factor (VEGF) receptor 1] (10, 11) and soluble endoglin (12, 13), have greatly facilitated the understanding of preeclampsia pathogenesis (14), the precise mechanism underlying the abnormal placentation in preeclampsia is still unclear, which has hindered the development of additional effective strategies to prevent or treat this disorder.

Angiotensin II (ANGII) is a vasoactive octapeptide that is generated by enzymatic reaction cascades, starting with the cleavage of angiotensinogen by renin (15). ANGII activates its major cognate receptor, ANGII receptor type 1 (AT1R). Because the first report showing that the pregnancy-specific activation of AT1R signaling by transgenic expression of both human angiotensinogen and human renin in mice [referred to as pregnancy-associated hypertension (PAH) mice] induces a preeclampsia-like syndrome, the renin-angiotensin system (RAS) has garnered considerable attention regarding the pathogenesis of preeclampsia (16). There are multiple lines of evidence supporting an increase in AT1R signaling in preeclampsia patients: (i) heterodimerization of AT1R with bradykinin B2 receptor increases the responsiveness to ANGII (17), (ii) an activating autoantibody against AT1R is detected in patients (18) and induced a preeclampsia-like phenotype in mice (19), and (iii) an oxidized form of angiotensinogen that is more efficiently cleaved by renin is frequently detected in preeclampsia patients (20). RAS-induced preeclampsia phenotypes have been characterized in several lines of experimental models in rodents (2123).

Augmented AT1R signaling activates NADPH (reduced form of nicotinamide adenine dinucleotide phosphate) oxidase (NOX), a major source of reactive oxygen species (ROS) (24) in the vasculature of preeclamptic patients (25, 26). In addition, tumor necrosis factor–α (TNF-α) and other proinflammatory cytokines that are increased in preeclampsia are also potent inducers of ROS production (27). Excess ROS accumulation induces oxidative stress that modifies and damages proteins, lipids, and nucleic acids, thereby causing cell death or inflammation (28, 29). Oxidative stress has been considered a major aggravating factor for preeclampsia pathogenesis because it induces poor placental perfusion and maternal endothelial dysfunction (3, 14, 25, 30, 31). Therefore, therapeutic strategies to reduce oxidative stress have been repeatedly tested in large clinical trials (32, 33). However, these trials have been largely negative and have reported worsened fetal outcomes, thereby leading researchers to question the causal role of oxidative stress in the pathogenesis of preeclampsia (32, 33). A potential underlying mechanism could be the cellular function of ROS as an essential second messenger (29, 34); thus, it is essential to better understand the role of ROS signaling in preeclampsia.

Nrf2 (nuclear factor erythroid 2–related factor 2) is a master transcriptional regulator of the cellular antioxidative stress response (28, 35). Under normal conditions, Nrf2 is constantly degraded through the ubiquitin-proteasome pathway mediated by the oxidative stress sensor protein Keap1 (Kelch-like erythroid protein with CNC homology-associated protein 1) (36). Under conditions of oxidative stress, Keap1 loses its Nrf2-sequestering activity, resulting in Nrf2 stabilization. Stabilized Nrf2 then translocates into the nucleus and transactivates target genes that encode a battery of antioxidant proteins and detoxifying enzymes (28, 37, 38). The placental accumulation and activation of Nrf2 have been reported in preeclampsia (3941). Nrf2 accumulation has been detected in villous cytotrophoblasts (41), syncytiotrophoblasts, endothelium, and villous stromal cells (40). On the basis of the cytoprotective functions of Nrf2 in diverse diseases, it has been highlighted as a potential therapeutic target for preeclampsia (31).

Here, we used a mouse genetic approach to elucidate the role of ROS signaling and the potential therapeutic impact of interfering with the Keap1-Nrf2 system in preeclampsia. Unexpectedly, we found that augmented ROS signaling as a result of genetic deletion of Nrf2 improved maternal and fetal morbidity and mortality and enhanced placental vascular plexus formation in a RAS-driven animal model of preeclampsia. We further demonstrate a role for ROS signaling as an essential regulator of placental angiogenesis.


Nrf2 deficiency reduces perinatal morbidity and mortality of PAH mice

To gain insight into oxidative stress and preeclampsia pathology, we first ascertained whether the placenta is exposed to extensive oxidative stress in preeclampsia. Placental samples from patients with preeclampsia showed stronger immunoreactivity to malondialdehyde, a lipid peroxidation marker, in comparison with normal pregnancy placentas. This observation indicates that preeclampsia induces oxidative stress accumulation in the placenta, in agreement with previous reports (Fig. 1A) (42, 43).

Fig. 1 Accumulation of oxidative stress in placentas from patients with preeclampsia and improvements in maternal and fetal outcomes in PAH mice with Nrf2 deficiency.

(A) Lipid peroxidation status of human placentas. Immunohistochemical detection of malondialdehyde (brown) was performed using placentas at 32-week gestation from normal pregnancies and those affected by preeclampsia. Hematoxylin (blue) was used to stain nuclei. The pictures show representative data from six normal pregnancy and six preeclampsia patients. (B) Maternal mortality rate of PAH mice between E13.5 and delivery. n = 10, 24, 26, and 25 dams for NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO, respectively. (C) Neonatal mortality rate in 0.5 to 7.5 days after birth. n = 38, 45, 66, and 77 neonates for NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO, respectively. (D and E) Fetal body weight at E16.5 (D) and E18.5 (E). n = 49, 69, 72, and 50 fetuses at E16.5 for NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO, respectively; n = 43, 53, 60, and 62 fetuses at E18.5 for NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO, respectively. (F and G) Representative photomicrograph of fetuses from pregnant dams of each genotype at E16.5 (F) and E18.5 (G). Scale bars, 50 μm (A) and 5 mm (F and G). Data in bar graphs are means ± SD (B to E). *P < 0.05 and **P < 0.01 (B to E).

To elucidate the pathophysiological role of oxidative stress in preeclampsia, we used a RAS-induced preeclampsia model, the PAH mouse (16, 44). In this model, the breeding of female mice bearing a homozygous human angiotensinogen transgene (hAGTTg/Tg) to male mice harboring a homozygous human renin transgene (hRENTg/Tg) leads to pregnancy-specific activation of the RAS and results in a preeclampsia-like phenotype (fig. S1). We then bred these transgenic mice with systemic Keap1 knockdown (Keap1KD) mice or Nrf2 knockout (Nrf2KO) mice (35, 45) and generated PAH mice with Keap1 or Nrf2 gene modification, which resulted in Nrf2 overactivation (referred to as PAH-Keap1KD) or Nrf2 loss (referred to as PAH-Nrf2KO), respectively. We referred to PAH mice with intact Nrf2 and Keap1 gene loci as PAH-WT (wild-type) mice (fig. S1). Because murine renin does not cross-react with human angiotensinogen (16), we used mice bearing the homozygous human angiotensinogen transgene as controls; they are referred to as NCP (normal control pregnancy)-WT, NCP-Nrf2KO, and NCP-Keap1KD mice (fig. S1).

To determine the effects of genetic modulation of the Keap1-Nrf2 system on preeclampsia pathologies, we first investigated mortality and morbidity of PAH mice after Keap1 or Nrf2 gene modulation. As previously reported, overactivation of the RAS in PAH-WT mice led to significantly higher perinatal maternal and neonatal mortality rates (16). Surprisingly, against our expectation, PAH-Nrf2KO mice and their fetuses showed the lowest mortality rate among the three PAH mouse genotypes (Fig. 1, B and C).

Overactivation of the AT1a receptor (one of the major AT1R isoforms in rodents) is responsible for intra-uterine growth retardation in this mouse model (44, 46). Consistent with maternal and neonatal mortality rates, the fetal body weight of PAH mice inversely correlated with the genetic status of Nrf2 expression from embryonic day 16.5 (E16.5) onward but not at E13.5 (Fig. 1, D to G, and fig. S2A). Genetic alteration of the Keap1-Nrf2 pathway did not affect fetal body weight in normal pregnant mice (NCP-Nrf2KO and NCP-Keap1KD) at E16.5 and E18.5 (fig. S2A). These data suggest that Nrf2 specifically affects perinatal complications in PAH mice but not in normal pregnant mice.

Because ROS accumulation in response to RAS activation has been implicated in the pathogenesis of essential hypertension (47), we evaluated whether the Keap1-Nrf2 system affects the blood pressure of PAH mice. The systolic blood pressure of PAH-WT mice increased after E13.5, as reported previously (fig. S2B) (16). Loss of Nrf2 did not alter the systolic blood pressure of PAH mice (PAH-Nrf2KO). The systolic blood pressure of PAH-Keap1KD mice increased similarly to that of PAH-WT and PAH-Nrf2KO mice until E16.5 and then increased further at E18.5 (fig. S2B). Because the systolic blood pressure increase in PAH-Keap1KD was mild, we surmise that the observed effects of worsened maternal and fetal/neonatal mortality and intra-uterine growth retardation in PAH-Keap1KD mice may not be directly related to blood pressure but may be related to altered ROS concentrations secondary to RAS activation.

Nrf2 deficiency improves placental angiogenesis in PAH mice

In PAH mice, perinatal complications occur late in pregnancy, which is the period when fetal growth is driven by the placental feto-maternal exchange system (16, 44, 46, 48). Our study demonstrated that the severity of fetal growth restriction correlated with Nrf2 activity in PAH mice. Therefore, we surmised that genetic modulation of the Keap1-Nrf2 pathway might affect feto-maternal exchange in PAH mice. Because feto-maternal exchange occurs through fetal and maternal blood in the placenta, we analyzed placental angiogenesis by staining for isolectin B4 (ILB4), which specifically binds to endothelial cells (49). We found that at E16.5, the placental vascular network in the labyrinth zone in PAH-WT mice was sparse compared to that in NCP-WT mice (Fig. 2, A and B, and fig. S3A), in agreement with a previous report (48). The placental vascular plexus was sparser in PAH-Keap1KD mice but was dense in PAH-Nrf2KO mice at E16.5. These differences were not found in normal pregnant mice (fig. S3A). These changes in vascular plexus formation were most prominent at the labyrinth zone, which is close to the junctional zone (# in Fig. 2A).

Fig. 2 Nrf2 deficiency improves placental angiogenesis in PAH mice.

(A) ILB4 immunostaining (brown) of labyrinth zone sections at E16.5. Hematoxylin (blue) was used for counterstaining. Lower panels are magnified images of upper panels. # indicates junctional zones. The pictures show representative data from n = 4, 5, 3, and 4 of placentas from NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO mice, respectively. (B) Quantification of ILB4-positive vascular areas in sections from E16.5 placentas. ILB4-positive areas in labyrinth zone sections were measured using ImageJ software (National Institutes of Health), and the relative positive area in PAH mouse samples is shown, with the average value from NCP-WT mouse samples set as 1.0. n = 4, 5, 3, and 4 placentas from NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO mice, respectively. (C) Endothelial cell proliferation in the labyrinth zone. Immunoreactivity for pHH3 was used to detect cellular proliferation in sections from the E16.5 labyrinth zone. Lower panels are merged images of pHH3 (red), ILB4 (green), and DAPI (blue, nucleus) staining. White arrowheads indicate pHH3 and ILB4 double-positive (yellow) proliferating endothelial cells. The pictures show representative data from five placentas for each group. (D) Number of pHH3-positive nuclei in a section from the E16.5 labyrinth zone. n = 5 placentas for each group. Scale bars, 100 μm (A) and 50 μm (C). Data in bar graphs are means ± SD (B and D). *P < 0.05 and **P < 0.01 (B and D).

We next analyzed endothelial cell proliferation using phosphorylated (Ser10) histone H3 (pHH3) immunoreactivity, which detects cells in mitosis. The number of pHH3-positive nuclei was markedly higher in PAH-Nrf2KO placentas than in PAH-Keap1KD or PAH-WT placentas (Fig. 2, C and D, and fig. S3B). Moreover, placental vascular plexus formation was not appreciably affected by genetic modulation of the Keap1-Nrf2 pathway in NCP mice (fig. S3, C and D). Together, these data indicate that placental angiogenesis and feto-maternal exchange in the PAH model is improved by genetic deletion of Nrf2 but dysregulated by increased Nrf2 activity.

Nrf2 deletion augments ROS accumulation in the PAH mouse placenta

ROS have been suggested to promote angiogenesis (47, 50, 51). Because there is extensive oxidative stress in preeclamptic placentas (Fig. 1A), we reasoned that enhanced angiogenesis under conditions of Nrf2 deficiency might result from augmented placental ROS accumulation in PAH mice. At E18.5, the labyrinth zone of PAH-WT placentas tended to have more foci positive for 8-hydroxy-2′-deoxyguanosine (8-OHdG), an oxidative DNA damage marker, than that of NCP-WT placentas (Fig. 3A). The number of foci was further increased by Nrf2 deficiency (PAH-Nrf2KO mice) and decreased by Nrf2 overactivation (PAH-Keap1KD mice) (Fig. 3B). 8-OHdG–positive foci were mainly located in the placental labyrinth zone, where impaired angiogenesis was observed in PAH mice (Fig. 3A).

Fig. 3 Nrf2 deficiency augments oxidative DNA damage in the labyrinth layer of PAH mice.

(A) 8-OHdG staining (dark purple) of placentas at E18.5. The lower panels are magnified views of the dotted rectangles in the upper panels. Nuclear Fast Red (pink) was used for counterstaining. Red arrows indicate 8-OHdG–positive foci, which are distributed in the labyrinth zone. Scale bar, 50 μm. The pictures are representative images from n = 4, 5, 3, and 4 of placentas from NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO mice, respectively. (B) Number of 8-OHdG–positive foci per section in E18.5 placentas. n = 4, 5, 3, and 4 placentas for NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO, respectively. (C) Nrf2 and Nqo1 protein abundance in placentas at E18.5 was examined by Western blotting. n = 3 placentas for each group. Lamin B and α-tubulin were used as internal controls for nuclear extracts (upper) and cytoplasmic extracts (lower), respectively. The pictures show representative data from three independent experiments, each of which used three placentas per group. (D) Relative quantification of Nrf2 and Nqo1 protein abundance by densitometry analysis of the Western blotting data in (C) after normalization to the internal controls. n = 3 placentas for each group. (E) Expression of Nrf2 target genes in the labyrinth zone at E18.5 were determined by qRT-PCR. n = 5, 5, 6, and 6 for NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO, respectively. Data in bar graphs are means ± SD (B, D, and E). *P < 0.05, **P < 0.01 (B, D, and E).

Consistent with the pattern of oxidative DNA damage accumulation and the genetic status of Nrf2 activity, we observed abundant nuclear accumulation of Nrf2 in PAH-Keap1KD placentas at E16.5 and E18.5. Nuclear Nrf2 was not detected in PAH-Nrf2KO placentas at E16.5 and E18.5 (Fig. 3, C and D, and fig. S4A). The enhanced accumulation of Nrf2 in PAH-Keap1KD placentas was associated with a significant increase in the cytosolic protein abundance of Nqo1, which is encoded by an Nrf2 target gene (Fig. 3, C and D, and fig. S4A) (35, 45, 52). The mRNA expression of antioxidant-related Nrf2 target genes such as Nqo1 (which encodes NADPH:quinone oxidoreductase 1), Srxn1 (which encodes sulfiredoxin 1), Txnrd1 (which encodes thioredoxin reductase 1), and Gstm1 (which encodes glutathione S-transferase μ 1) (35, 45, 52) was also robustly increased in placentas from PAH-Keap1KD mice at E16.5 and E18.5 (Fig. 3E and fig. S4B). In contrast, antioxidant gene expression was markedly reduced in PAH-Nrf2KO placentas compared to PAH-WT placentas. These results support our contention that increased ROS accumulation in the PAH-Nrf2KO placenta may alter the PAH milieu in a manner favorable for placental angiogenesis, thereby improving maternal and fetal outcomes.

Expression of mRNAs encoding angiogenic chemokines is induced in the Nrf2-deficient PAH placenta

To identify genes responsible for the improved outcomes in PAH-Nrf2KO mice, we performed a genome-wide transcriptome analysis using E16.5 labyrinth zone samples. A set of genes with greater than 1.5-fold changes in expression in both NCP-WT and PAH-Nrf2KO samples compared to PAH-WT samples was identified as the Nrf2-dependent PAH-specific gene signature (fig. S5A and data file S1).

The gene set contained mRNAs encoding various angiogenic chemokines and their receptors (Fig. 4A). Pathway analysis of the gene set identified the cytokine-cytokine receptor interaction as the top overrepresented KEGG (Kyoto Encyclopedia of Genes and Genomics) pathway (data file S1 and fig. S5B). Quantitative reverse transcription polymerase chain reaction (qRT-PCR) analyses of E16.5 placentas confirmed that Nrf2 deficiency induced the expression of mRNAs encoding angiogenic chemokines—Ccl2 (which encodes C-C motif chemokine ligand 2), Ccl5 (which encodes C-C motif chemokine ligand 5), Cxcl9 (which encodes C-X-C motif chemokine ligand 9), and Cxcl10 (which encodes C-X-C motif chemokine ligand 10)—in E16.5 PAH placentas (Fig. 4B). In addition, mRNAs encoding inflammation-related cytokines, such as Il1a (which encodes interleukin-1α), Il1b (which encodes interleukin-1β), Cxcl1 (which encodes C-X-C motif chemokine ligand 1), and Tnfa (which encodes TNF-α), were highly expressed in the PAH-Nrf2KO mouse placenta (Fig. 4B), whereas the expression of mRNAs encoding classical angiogenic factors—Angpt1 (which encodes angiopoietin-1), Angpt2 (which encodes angiopoietin-2), Vegfa (which encodes vascular endothelial growth factor A), Kdr (which encodes kinase insert domain receptor for VEGF), Flt1 (which encodes fetal liver tyrosine kinase 1), and Tek (which encodes endothelial tyrosine kinase receptor for angiopoietin-1)—were not significantly affected (fig. S5C). The plasma concentrations of PLGF2 (placenta growth factor 2) and sFLT1, both of which have been implicated in the pathogenesis of preeclampsia (11, 30, 53, 54), were not altered by Keap1 or Nrf2 gene modification in PAH mice at E16.5 (fig. S6). The chemokines and cytokines that showed increased expression in PAH-Nrf2KO placentas are critically important for angiogenesis and placental vascular remodeling (44, 55), suggesting that the altered or restored expression pattern of mRNAs encoding angiogenic cytokines and chemokines might play a role in the enhanced angiogenesis in the PAH-Nrf2KO placenta through ROS accumulation in the labyrinth zone. It is also likely that the chemokines and cytokines may contribute to immune cell infiltration that occurs in preeclampsia (56, 57).

Fig. 4 Nrf2 deficiency induces the expression of angiogenic and inflammatory factors in the PAH placenta.

(A) Heat map of the microarray data for the labyrinth zone samples from NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO mice (n = 2 placentas for each group) at E16.5 shows the relative expression of genes related to angiogenesis or inflammation, which were selected from the list of genes in the Nrf2-dependent PAH-specific gene signature (data file S1). Blue letters indicate genes for which multiple probes are included in this signature. (B) qRT-PCR analyses of the expression of mRNAs encoding angiogenic chemokines [Ccl2, Ccl5, Cxcl9, and Cxcl10; highlighted in red letters in (A)] and inflammation-related genes (Il1a, Il1b, Cxcl1, and Tnfa) in the labyrinth zone of PAH mice at E16.5. n = 5, 5, 6, and 5 placentas for NCP-WT, PAH-Keap1KD, PAH-WT, and PAH-Nrf2KO, respectively. Data in bar graphs are means ± SD. *P < 0.05, **P < 0.01.

Pharmacological activation of Nrf2 worsens the perinatal morbidity and mortality of PAH mice

The benefits of Nrf2 inducers have been highlighted as an antioxidative therapy (28). We asked whether pharmacological activation of the Nrf2 pathway recapitulates the genetic induction of Nrf2 or whether there is a therapeutic window for pharmacological induction. Therefore, we treated PAH-WT and NCP-WT mice after E13.5 with an Nrf2 inducer, CDDO-Im {1-[2-cyano-3,12-dioxooleana-1,9(11)-dien-28-oyl] imidazole} (28, 52). The expression of Nrf2 target genes (Nqo1, Srxn1, Txnrd1, and Gstm1) was increased in placentas from CDDO-Im–treated NCP-WT and PAH-WT mice (fig. S7). In PAH-WT mice, CDDO-Im treatment increased maternal mortality and decreased fetal body weight (Fig. 5, A and B). In addition, PAH-WT mice treated with CDDO-Im experienced abortion and vaginal bleeding before dying (Fig. 5C), although CDDO-Im administration did not affect the mortality or fetal body weight of NCP-WT mice (Fig. 5, A and B).

Fig. 5 Pharmacological activation of Nrf2 worsens perinatal complications in PAH mice.

(A) Maternal mortality rate of NCP-WT and PAH-WT mice treated with CDDO-Im every other day from E13.5 to E17.5. n = 3 NCP-WT dams with or without CDDO-Im administration; n = 7 PAH-WT dams with or without CDDO-Im administration. (B) Fetal body weight of E16.5 and E18.5 embryos in NCP-WT or PAH-WT mice treated with CDDO-Im. n = 18 and 9 fetuses at E16.5 in NCP-WT and PAH-WT dams treated with vehicle, respectively; n = 15 and 23 fetuses at E16.5 in NCP-WT and PAH-WT dams treated with CDDO-Im, respectively. n = 18 and 28 fetuses at E18.5 in NCP-WT and PAH-WT dams treated with vehicle, respectively; n = 22 and 33 fetuses at E18.5 in NCP-WT and PAH-WT dams treated with CDDO-Im, respectively. (C) Placental abortion and vaginal bleeding in PAH-WT mice treated with CDDO-Im. These features were observed in every PAH mouse after CDDO-Im administration. Arrows and arrowheads indicate the fetus and placenta, respectively. Data in bar graphs are means ± SD. *P < 0.05, **P < 0.01.


The mammalian placenta undergoes extensive angiogenesis to develop efficient feto-maternal exchange of oxygen, nutrients, and waste products (1, 14). Defects in placental angiogenesis are believed to play a central role in the pathogenesis of preeclampsia and fetal growth restriction (1, 10, 11, 13, 58). Here, we used gene-modified mouse lines to provide insights into the role of the Keap1-Nrf2 system in placental angiogenesis. Our results illustrate the essential and indispensable role of ROS as placental proangiogenic factors in preeclampsia (Fig. 6).

Fig. 6 Schematic of Nrf2-mediated angiogenic suppression in the PAH placenta.

Overproduction of human ANGII (hANGII) stimulates AT1R signaling and creates an anti-angiogenic milieu in the labyrinth zone of PAH mice. ROS accumulation counteracts the anti-angiogenic milieu by inducing the expression of genes encoding angiogenic chemokines and cytokines. Because Nrf2 decreases ROS concentrations, Nrf2 inactivation ameliorates the intra-uterine growth retardation (IUGR) and PAH complications by inducing ROS-mediated angiogenesis in PAH-Nrf2KO mice. In PAH-Keap1KD mice, placental angiogenesis is further worsened due to ROS quenching by activated Nrf2.

Placental angiogenesis is critical for establishing feto-maternal exchange and is accelerated from E14.5 to delivery in mice (44, 46, 48). The roles of the placental anti-angiogenic milieu in the pathogenesis of preeclampsia have been highlighted by evidence that placental-derived anti-angiogenic factors cause a preeclampsia-like phenotype in mice (58, 59). In the PAH model, the plasma concentration of ANGII is increased from E13.5 and contributes to impaired fetal growth and placental vasculature formation (16, 44, 48). In our mouse model, disease severity correlated well with the degree of impaired placental angiogenesis in the labyrinth zone, further supporting the pathogenic importance of placental angiogenesis.

Genetic and pharmacological manipulation of the Keap1-Nrf2 system can reduce oxidative stress and frequently results in beneficial outcomes in various disease models (28). Nrf2 plays context-dependent roles in angiogenesis: Systemic Nrf2 deletion augments angiogenesis after hindlimb ischemia (60), whereas endothelial cell–specific deletion of Nrf2 leads to impaired physiological retinal angiogenesis by reducing tip cell formation and vascular branching (61). Consistent with the former observation, there are many lines of evidence that support the notion that appropriate amounts of ROS are proangiogenic factors that act through various mechanisms, including enhancing VEGF and hypoxia-inducible factor signaling (47, 50, 51, 62, 63). In addition, physiological amounts of ROS may be important in other systems to maintain normal cellular signaling, whereas high amounts of ROS are associated with cell toxicity (64).

Our study demonstrated that systemic deletion of Nrf2 increased oxidative DNA damage and enhanced placental angiogenesis, as evidenced by the higher density of the capillary plexus and the increased proliferation of endothelial cells in the labyrinth zone of PAH-Nrf2KO mice. Conversely, genetic and pharmacological induction of Nrf2 resulted in increased maternal mortality and worsened fetal growth restriction, with reduced capillary density in the labyrinth zone, despite the reduction in oxidative DNA damage. These results are consistent with a large clinical trial that tested the ability of antioxidants to prevent preeclampsia (32); high-risk women that received antioxidant vitamins during pregnancy had worse outcomes, including an increase in the number of low birth weight babies and stillbirths. On the basis of these observations, we posit that appropriate ROS accumulation in the placenta creates a placental milieu that is favorable for endothelial proliferation and angiogenesis, thereby alleviating preeclampsia pathologies that may otherwise be associated with inappropriate amounts of ROS. Although our microarray analyses of placental labyrinth zone samples showed no significant changes in the expression of mRNAs encoding classical angiogenic factors, we observed increased expression of mRNAs encoding angiogenic chemokines, cytokines, and their receptors in PAH-Nrf2KO mice, similar to that which occurs in the hindlimb ischemia model of Nrf2KO mice (60).

Because oxidative stress activates inflammation that produces intracellular and extracellular ROS (65, 66), the oxidative stress response and inflammation pathways are tightly interconnected. Activated inflammatory cells generate a battery of chemokines and cytokines that further potentiate inflammation and can induce angiogenesis (60, 65). Furthermore, these chemokines, such as Ccl2 and Ccl5, regulate placental vascularization (55, 67). We have shown that Nrf2 directly suppresses the expression of inflammatory cytokine–encoding genes in activated macrophages (67). These data support our contention that inactivation of Nrf2 induces ROS accumulation and angiogenic chemokine expression in PAH placentas, whereas Nrf2 overactivation impairs angiogenesis and enhances preeclampsia-mediated adverse outcomes. The mechanisms of excess maternal mortality in the Nrf2 overexpression studies remain to be explored in future studies.

Heme oxygenase-1 (HO-1) is an antioxidant enzyme that protects cells from oxidative stress and may have protective roles against preeclampsia (68). We observed a modest induction of Hmox1 mRNA (which encodes HO-1) in PAH-Keap1KD placentas compared to PAH-WT placentas (fig. S4B). However, this induction did not rescue the PAH outcomes, despite the marked reduction in oxidative DNA damage. Fetuses of Hmox1-knockout mice have extremely low birth weights and intra-uterine growth retardation (69), whereas fetuses of NCP-Nrf2KO and NCP-Keap1KD mice have normal birth weights. In addition, partial deficiency of HO-1 results in impaired placental angiogenesis (70, 71). Together, we speculate that HO-1 may have a unique role(s) in placental angiogenesis or systemic protective effects on the maternal endothelium independent of placental oxidative stress.

The plasma concentration of sFLT1 has also been suggested to correlate with ROS accumulation, and sFLT1 has been implicated in the pathogenesis of preeclampsia in rodents and humans (11, 30, 53, 54). However, sFLT1 concentrations were not altered by Keap1 or Nrf2 gene modification in PAH mice. This observation may suggest that the induction of sFLT1 is not fundamentally dependent on placental ROS accumulation. Simultaneous overexpression of renin and angiotensinogen induces the preeclampsia phenotype in our mouse model by activating the angiotensin signal during pregnancy. Also, increased ANGII sensitivity and enhanced angiotensin signaling have been reported in preeclampsia patients, although the circulating concentrations of renin are low compared to those in women experiencing normal pregnancies (1720, 7274). To generalize our conclusions to human preeclampsia, additional animal models, such as sFLT1 overexpression models, are needed.

In summary, we showed that augmented ROS signaling in response to Nrf2 deletion induced a proangiogenic milieu in the preeclamptic placenta (Fig. 6). Augmented ROS production during preeclampsia may represent an adaptive response to hypertension and systemic vascular damage. Further studies are needed to characterize the role of appropriate amounts of ROS signaling in preeclampsia, and the use of additional models may lead us to a better understanding of the role of ROS signaling in humans with preeclampsia.


Human subjects

The Institutional Review Board at Beth Israel Deaconess Medical Center approved the collection and use of discarded human placentas. Immunohistochemical analysis was performed as described in the Supplementary Materials.


All mouse experiments were approved by the Regulations for Animal Experiments and Related Activities of Tohoku University. hAGTTg/Tg and hRENTg/Tg transgenic mice were obtained from RIKEN BioResource Center, Japan. Male hRENTg/Tg and female hAGTTg/Tg mice were bred to obtain PAH mice, as previously reported (fig. S1) (16). Nrf2KO (Nrf2−/−) (35) and Keap1KD (Keap1KD/KD) mice have been described previously (45). Because Keap1-null mice die around weaning due to malnutrition caused by hyperkeratosis in the digestive tract (75), Keap1KD mice with a homozygously hypomorphic mutant Keap1 gene were used in this study to investigate effects of Nrf2 overactivation in mice. PAH mice bearing Nrf2−/− or Keap1KD/KD alleles were generated as depicted in fig. S1. Homozygous hAGTTg/Tg mice with modified Nrf2 or Keap1 genes were used as controls. Mice were backcrossed to C57BL/6J mice for more than five generations with PCR genotyping using primers listed in table S1. Mice were maintained in a specific pathogen–free facility.

Tissue preparation for histological analysis

Tissues for frozen sections were fixed with 4% paraformaldehyde (PFA) for 2 hours and embedded in an optimum cutting temperature compound (Sakura Finetek) after an overnight incubation in 20% sucrose. Tissues for paraffin-embedded sections were fixed with 4% PFA for 48 hours and processed by conventional procedures. Tissues for 8-OHdG staining were fixed with Bouin solution (DAKO) for 24 hours.

Immunohistochemistry of human placental tissues

Paraffin sections (5 μm thick) were deparaffinized and incubated with citrate buffer (DAKO, pH 6.0) for antigen retrieval. Slides were rinsed with phosphate-buffered saline (PBS) and incubated for 15 min with 3% hydrogen peroxide to quench endogenous peroxidase. To block nonspecific antibody binding, slides were incubated at 37°C with 2.5% normal horse serum in PBS for 40 min. Slides were then incubated with rabbit anti-malondialdehyde polyclonal antibody (1:500, ab6423, Abcam) for 40 min at 37°C, rinsed in PBS, developed using ImmPRESS horseradish peroxidase (HRP) Anti-Rabbit IgG Polymer Detection Kit (Vector Laboratories), and counterstained with hematoxylin.

Immunohistochemistry of mouse tissues

For paraffin-embedded tissue sections, citrate buffer (DAKO, pH 6.0) was used for antigen retrieval. Sections were incubated with primary antibodies after blocking with Protein Block Serum-Free (DAKO). Primary antibodies against pHH3 (1:300 dilution, 9701S, Cell Signaling Technology), biotinylated ILB4 (1:50, B-1205, Vector Laboratories), and 8-OHdG (1:50, MOG-20P, Japan Institute for the Control of Aging) were used. Fluorescence images were obtained using an LSM780 confocal imaging system (Carl Zeiss). For ILB4 and 8-OHdG staining, signals were developed using diaminobenzidine (DAB, DAKO) or a DAB solution containing nickel chloride hexahydrate (DAB-Ni, Wako). Sections were counterstained with DAPI (4′,6-diamidino-2-phenylindole), hematoxylin, or Nuclear Fast Red (H-3403, Vector Laboratories), as appropriate.

Measurement of systolic blood pressure

Systolic blood pressure was measured using a tail-cuff blood pressure system (BP-98a, Softron) following the manufacturer’s protocol (46).

Western blotting

Nuclear extracts and whole-cell lysates of placental tissue were prepared separately using sucrose buffer (0.25 M sucrose in 10 mM Hepes-KOH buffer). Samples were separated by SDS–polyacrylamide gel electrophoresis; transferred to polyvinylidene difluoride membranes; and immunoblotted with antibodies against Nrf2 (1:200 dilution, in-house clone #103), Nqo1 (1:1000, ab2346, Abcam), lamin B (1:1000, M-20, Santa Cruz Biotechnology), and α-tubulin (1:1000, T9026, Sigma). HRP-conjugated secondary antibodies (DAKO) and enhanced chemiluminescence reagents (GE Healthcare) were used to detect signals.

Quantitative reverse transcription polymerase chain reaction

Total RNA was prepared using Isogen (Nippon Gene) and reverse-transcribed using random hexamers and a SuperScript III Polymerase Kit (Invitrogen). Primer information is available in table S2. Gene expression was calculated on the basis of the threshold cycle values and the efficiency of each primer set, and the data were expressed relative to 18S ribosomal RNA.

Microarray analysis

Total RNA was isolated from the labyrinth zone at E16.5 using Isogen (Nippon Gene) and hybridized to whole-mouse genome microarrays (4 × 44 K, Agilent Technologies) following the manufacturer’s procedure after assessing RNA quality with a Bioanalyzer (Agilent Technologies). Data were analyzed using GeneSpring software (Agilent Technologies) and submitted to GEO (Gene Expression Omnibus; accession no. GSE76669). Genes with a greater than 1.5-fold change in average expression compared to that of PAH-WT samples are listed in data file S1, and these selected genes were further analyzed using DAVID (Database for Annotation, Visualization, and Integrated Discovery) Bioinformatics Resources 6.7 software ( with a KEGG pathway database (

CDDO-Im administration

CDDO-Im (Mochida Pharmaceuticals) was dissolved in sesame oil (Wako) and orally administered to mice at 30 μmol/kg body weight at E13.5, E15.5, and E17.5 using disposable feeding needles (Fuchigami).

Statistical analysis

Data in bar and line graphs are means ± SD. Statistical analyses were performed by using R software (The R Foundation, Comparisons between groups were analyzed using Fisher’s exact test, one-way analysis of variance, or an unpaired t test, as appropriate. P values were adjusted by Holm method for multiple comparisons of Fisher’s exact test. The Tukey-Kramer test was used as a correction method for multiple comparisons. Data were considered statistically significant at P < 0.05.


Fig. S1. Mouse models used in this study.

Fig. S2. Fetal body weight, systolic blood pressure, and gene expression profiles of RAS components in NCP and PAH mice with Keap1 or Nrf2 genetic mutation.

Fig. S3. Genetic deletion of Nrf2 improves vascular plexus selectively in the PAH placenta.

Fig. S4. Expression profiles of Nrf2 target genes in the placenta of PAH mice.

Fig. S5. Microarray analysis to identify an Nrf2-dependent PAH-specific gene signature.

Fig. S6. Genetic modification of the Keap1-Nrf2 system in PAH mice does not significantly change the plasma concentrations of sFLT1 and PLGF2.

Fig. S7. Induction of Nrf2 target gene expression in the labyrinth zone of PAH-WT mice by pharmacological activation of Nrf2.

Table S1. Primer sequences for genotyping mouse lines.

Table S2. Primer sequences for qRT-PCR analysis.

Data file S1. Microarray data revealed candidate genes for an Nrf2-dependent PAH-specific gene signature.


Acknowledgments: We thank I. Hirano, A. Konuma, S. Inomata, K. Kato (Tohoku University), and the Biomedical Research Core and Center for Laboratory Animal Research at Tohoku University for technical support. Funding: This work was supported by grants-in-aid from Ministry of Education, Culture, Sports, Science and Technology/Japan Society for the Promotion of Science KAKENHI (grant numbers 24249015 and 26111002 to M.Y., 26116702 and 15H04691 to N.S., and 12J07924 to T.S.); by the Core Research for Evolutional Science and Technology from the Japan Agency for Medical Research and Development (M.Y.); and by Sakakibara Memorial Research Grant from the Japan Research Promotion Society for Cardiovascular Diseases (N.S.). Author contributions: M.N., T.S., N.S., and M.Y. designed the research. M.N., T.S., L.Y., H.S., Z.K.Z., T.N., A.H., and N.S. performed the research and analyzed the data. M.N., T.S., A.Z.-S.W., Z.K.Z., S.A.K., N.S., and M.Y. wrote the manuscript. N.T., S.I., A.F., S.A.K., N.S., and M.Y. supervised the research. Competing interests: The authors declare that they have no competing interests.

Stay Connected to Science Signaling

Navigate This Article