Preeclampsia-Like Symptoms Induced in Mice by Fetoplacental Expression of STOX1 Are Reversed by Aspirin TreatmentNovelty and Significance
Preeclampsia (PE) is a common human-specific pregnancy disorder defined by hypertension and proteinuria during gestation and responsible for maternal and fetal morbimortality. STOX1, encoding a transcription factor, was the first gene associated with PE as identified by positional cloning approaches. Its overexpression in choriocarcinoma cells mimics the transcriptional consequences of PE in the human placenta. Here, we created transgenic mouse strains overexpressing human STOX1. Wild-type female mice crossed with transgenic male mice reproduce accurately the symptoms of severe PE: gestational hypertension, proteinuria, and elevated plasma levels of soluble fms-like tyrosine kinase 1 and soluble endoglin. Placental and kidney histology were altered. Symptoms were prevented or alleviated by aspirin treatment. STOX1-overexpressing mice constitute a unique model for studying PE, allow testing therapeutic approaches, and assessing the long-term effects of the preeclamptic syndrome.
See Editorial Commentary, pp 561–563
Preeclampsia (PE) is a pervasive complication of human pregnancy characterized by a gestational hypertension associated with proteinuria occurring from midgestation.1 Worldwide, this syndrome affects ≈5% of pregnant women and is a leading cause of maternal mortality, especially in low- and middle-income countries2,3; it is also a direct cause of iatrogenic prematurity,4 because the only definitive therapeutic act is the extraction of the fetoplacental unit.
STOX1 is a transcription factor belonging to the enlarged Forkhead Box gene family that has been associated with PE,5 able to modulate trophoblast proliferation and migration.6 STOX1 is maternally expressed in a specific cell type of the placenta, column extravillous trophoblasts.7 Overexpression of STOX1 in human choriocarcinoma cells induces transcription alterations that mimic those of preeclamptic placentas.8,9 Founds et al10 showed in a transcriptomic analysis that STOX1 is overexpressed (×2.1; P=0.013) during the first trimester in pregnancies that will have a preeclamptic outcome.
In the present work, we generated mice overexpressing human STOX1. Wild-type (WT) female mice crossed with transgenic males overexpressing STOX1 have a severe gestational hypertension, proteinuria, an increased plasma level of soluble antiangiogenic factors, as well as kidney and placenta histological alterations. We also demonstrate a beneficial effect of low-dose aspirin treatment on the maternal symptoms. Hence, STOX1-overexpressing mice constitute a unique model for studying severe PE.
More details of the methods are provided in the online-only Data Supplement.
Animals and Transgenesis and Aspirin Treatment
The complete open reading frame (ORF) of the human STOX1 (A isoform) cDNA was used to generate transgenic animals.11 The number of copies of the transgene was assessed by quantitative polymerase chain reaction (qPCR) from the mouse DNA relative to a single-copy gene. Transmission analysis through generations showed that there was a single locus of insertion on the 3 strains that were analyzed.
Animals (FVB/N) were bred in the animal facility of INRA (Institut National de Recherche Agronomique, Jouy en Josas, France) in a controlled environment (light/dark cycle, temperature, free access to food and water). Blood collection and urine collection, blood pressure (BP) measurements kidney perfusion and aspirin supplementation in the drinking water were performed according to standardized protocols (see Methods in the online-only Data Supplement).
RNA Extraction and Quantitative Reverse Transcriptase PCR Conditions
Total RNA from collected tissues was extracted and quantified by spectrophotometry. Quality of the RNA was systematically verified by Bioanalyzer and RIN to be consistently >8. After reverse transcription, quantitative reverse transcriptase PCR was performed. The efficiency of the PCR was checked for every primer couple and estimated (Methods in the online-only Data Supplement). The Cts were normalized by the Ct values obtained for the murine succinate dehydrogenase subunit A (Sdha) used as a normalizing gene and shown previously to be stable and highly normalizing in the placenta.12 All primers are presented in Table S1 in the online-only Data Supplement.
BP was measured repeatedly in the tail artery in mice before, during, and after pregnancy (for a total of ≈35 consecutive days). Measurements were performed using a computerized, noninvasive tail-cuff plethysmography method (Letica 5001; Bioseb), using thermostatically warmed restrainers designed for mice and adapted to the size of the animal. This system uses volume-pressure recording technology to detect changes in tail volume that corresponds to systolic and diastolic pressures during each measurement cycle. Unanesthetized mice previously accustomed during 1 week to the manipulation (systolic BP stabilized) were placed in plastic holders. The protocol consisted of at least 6 satisfactory measurements daily, meaning without detectable movement of the mice. Thirty-six gestations were followed, which represents >5000 independent BP measurements by 2 independent researchers (systolic, diastolic, and median always displayed a highly similar profile). In addition, invasive measures were performed on 14 WT mice mated with WT or transgenic males at the end of gestation (14.5–17.5 days postcoitum [dpc]). The catheter was inserted into the left common carotid artery under xylazine/ketamine anesthesia. Mice were kept on a heating pad during experimentation. BP was recorded and analyzed from continuous measurements using the BIOPAC MP36 system. The extraction of the systolic BP started when cardiovascular parameters were stabilized.
Assessment of Albumin/Creatinine Ratio
Urine samples were collected daily during gestation and were pooled at early (1–5 dpc), middle (9–12 dpc), and late (15–18 dpc) gestation times separately for each mice. Albumin/creatinine ratio was measured in these urine specimens using the ELISA kit of Exocell (Philadelphia, PA).
sFlt1 and sEng Plasmatic Concentration Assessment
ELISAs were used to determine the concentration of soluble fms-like tyrosine kinase 1 (sFlt1) and soluble endoglin (sEng) in mouse plasma with commercial kits (R&D Systems).
Placentas and kidneys were sampled from 10 transgenic and WT mice (after heart perfusion by 20 mL of PBS in the case of kidneys). Antifibrin immunostaining of slides from paraffin-embedded kidneys was performed using the anti-human fibrinogen at 1/400 from Dako (A0080) and the LSAB revelation system (Dako, K0679) according to the manufacturer’s instructions (standard conditions). Placental histology was performed after hematoxylin–eosin staining or periodic acid-schiff (PAS) staining. The number of vacuolated foci present in the labyrinth was counted after hematoxylin–eosin staining. A grade of 1 to 5 was given according to the level of PAS intensity, and the PAS-positive areas were measured in the labyrinth.
During this study, parametric tests were used using ANOVA, followed by post hoc Dunnett t tests for comparison with controls. The StatistXL add-in of Excel was systematically used. P<0.05 was considered significant.
Generation of STOX1 Transgenic Mice
Three independent transgenic lines (called TgSTOX13, TgSTOX34, and TgSTOX42) were generated by introducing the human STOX1A ORF (complete isoform) under the control of a ubiquitously expressed promoter (the cytomegalovirus promoter). TgSTOX13 and TgSTOX42 strains contain a unique copy that was transmitted in a Mendelian fashion. Because the expression level of TgSTOX34 was very similar to that of TgSTOX13, we focused our attention on the TgSTOX13 and TgSTOX42 strains.
The expression levels of human and mouse STOX1 were tested by quantitative reverse transcriptase PCR in the maternal kidney, liver, brain, and placenta of pregnant TgSTOX13 and TgSTOX42 females, as well as in the testis and spermatozoa of transgenic males of both strains, using species-specific primers (Table); hSTOX1 expression was comparatively higher in placental tissue than in any other tissue or spermatozoa. Quantitative reverse transcriptase PCR efficiency was maximal for mouse and human STOX, allowing to suggest that the placental expression level of the human hSTOX1 was roughly similar to that of mStox1 in TgSTOX13 transgenic mice (Table) and ≈13-fold higher in TgSTOX42. In the other tissues tested (kidney, brain, testis, and liver), the transgene expression level was lower than that of the endogenous gene but always higher in TgSTOX42 than in TgSTOX13 (Table). Because there is a unique transgene insertion in both strains (as tested by quantitative PCR), we assume that the difference in the expression level was strictly regulated by the genomic position of this insertion. In transgenic placentas from WT×TgSTOX42 crosses, where the transgene expression is the highest, the endogenous gene was overexpressed 4-fold (P=2×10–11 by Student test compared with WT×WT crosses).
STOX1 Transgenic Mice Develop a Strong Gestational Hypertension, Proteinuria, and Present an Increased Level of Circulating sFlt1 and sEng
WT females were crossed with males of 3 genotypes (WT, TgSTOX13, or TgSTOX42) to ensure that the transgene expression would be restricted to the fetoplacental unit. BP was monitored daily in the various crosses (Figure 1A). An increase in BP during gestation was observed in all cases of crosses involving transgenic males. The BP curves reached a plateau in both transgenic crosses around embryonic day 10.5, until days 14.5 to 15.5, when the BP tends to increase again by an additional 15 to 20 mm Hg. The maximal increase reached ≈80 mm Hg from a normal pressure of ≈116 mm Hg (Figure 1A). BP returned to normal rapidly (in ≈48 hours) after parturition. In addition, we carried out measurements by catheterization of one of the carotid at late gestation times, and consistently with human BP data under anesthesia,13 the increase in mice carrying transgenic fetuses was present, albeit reduced; the BP was 12 mmHg more elevated in the transgenic crosses (P=0.008; Figure S1).
Proteinuria was evaluated as a ratio between albumin and creatinine (Figure 1B). After individual normalization, we observed an increase in the albumin/creatinine ratio, which comprised between ≈1.7- and 1.9-fold at midgestation (P=0.02 and 0.03 for TgSTOX13 and tgSTOX42, respectively) and between ≈1.8- and 3.8-fold at late gestation points (P=0.026 and 0.002, respectively), in females crossed with transgenic males compared with control crosses.
sFlt1, considered as a biomarker for PE in the human disease,14–16 was monitored (Figure 2A). The average concentration in the plasma of pregnant mice was 2075 pg/mL when TgSTOX13 males have fathered and 2455 pg/mL when crossed with TgSTOX42 males, an increase of 1.9- and 2.3-fold compared with control (1076 pg/mL), respectively (P=0.02 and 0.03). sEng was also measured in the plasma (Figure 2B), and it was found to be increased in the crosses involving TgSTOX13 (×2.0; P=0.04) and TgSTOX42 males (×2.3; P=0.01) compared with crosses with WT males.
WT Females Carrying Transgenic Fetuses Have Kidney Anomalies That Are Characteristic of PE
In the kidneys of females crossed with TgSTOX13 mice, glomerular changes were observed: first, a glomerular tuft swelling with expansion of the capillary wall, up to the point that the tuft could occasionally make contact with the Bowman’s capsule; second, we observed a slight increase in the glomerular cellularity (42.3±10.1 in TgSTOX13 and 48.1±11.9 in TgSTOX42 versus 34.0±11.8 in WT; Figure 3A; P=3.4×10−3 and P=6.4×10−7, respectively) attributable to a mild mesangial cell proliferation. Immunohistochemistry revealed subendothelial fibrin deposition, with a predominant peripheral pattern. These changes are strongly reminiscent of the renal pathology observed in human PE17–19 and could not be observed in the kidneys of WT pregnant mice crossed with WT males (Figure 3B, top and middle photographs). No major abnormality was found in the tubulointerstitial compartment, or in the larger vessels, in either group.
Aspirin Treatment Efficiently Improves the Maternal Phenotype
WT females crossed with TgSTOX13 or TgSTOX42 males were treated with aspirin at low doses in the drinking water during the whole gestation. As a control, WT pregnant females received aspirin all along their pregnancy after a cross with WT males. Aspirin completely prevented the onset of hypertension and brought the albumin/creatinine ratio to normal values at midgestation and late gestation times (Figures 1A and 1B). Fibrin immunolabeling of the kidneys showed that the deposit was considerably reduced in aspirin-treated WT females crossed with TgSTOX13 or TgSTOX42 males, compared with nontreated females (Figure 3B).
STOX1 Transgenic Placentas Present Structural Alterations
The anatomic alterations of hemizygous TgSTOX13 and TgSTOX42 (validated as transgenic by PCR) placentas were studied by histology (Figure S2). No difference was observed in gross morphology at 16.5 dpc, such as in the ratio between the labyrinth and the junctional zone. Because deregulated imprinted gene expression is known to alter glycogen trophoblast location, we counted the number of vacuolated foci, graded the PAS staining, specific of glycoproteins, and measured the surface of PAS area in the labyrinth. Although in hemizygous TgSTOX13 placentas no significant difference was detected compared with measures in WT placentas, in hemizygous TgSTOX42 placentas a significant diminution of vacuolated foci in the labyrinth, as well as PAS grade and PAS area (Figure 4A–4C), was consistently observed.
Litter Size Is Affected by the Transgene
Litter size seemed to be linked to the genotypes of the males used in crosses (Figure S3; P=0.0016 by 1-factor ANOVA), with 8.4 (±2.0) pups on average in WT crosses. In crosses with TgSTOX13 males, the litter size was reduced to 6.9 (±3.4; P=0.048). Consistent with the higher transgene expression in TgSTOX42, the litter size was reduced to 6.0 (±2.6, P=0.001) in crosses involving a TgSTOX42 male. When aspirin was given to females during gestation, the litter size was 7.8 (±2.4) and 9.3 (±3.1) in crosses involving TgSTOX13 and TgSTOX42 males, respectively. In the latter case, this was significantly different from nontreated crosses (P=0.022), showing that the drug was able to correct the reduced litter size phenotype.
Herein, we show that mice overexpressing STOX1 in the embryo–fetoplacental unit reproduce in extenso and without further manipulation the major features of PE: proteinuria and hypertension. A peculiarity of the present mouse model of PE is the intensity of the gestational hypertension, which reaches values corresponding to severe human PE at the end of gestation. The lower difference in BP observed under anesthesia compared with vigil animals may presumably be explained by a relaxing effect of the anesthesia, moderating as well the BP level. We also observed an elevation of both sFlt1 and sEng, key antiangiogenic factors of PE. These mice also harbored PE-specific kidney alterations.
In summary, a unique transcription factor (STOX1) expressed in the fetoplacental unit is sufficient to trigger PE-like symptoms in WT females in a dose-dependent manner (illustrated by the more or less severe phenotype observed in crosses involving TgSTOX13 or TgSTOX42 males). Although the involvement of STOX1 in human PE was debated after its discovery,20 the present study confirms a growing corpus of evidence indicating an actual contribution of STOX1 to the preeclamptic phenotype.
Rodent models of PE have been considered very important for studying PE pathophysiological mechanisms. Various models of reduced uterine perfusion were generated by inducing placental ischemia, administration of angiotensin antagonists, proinflammatory drugs, or vasoconstrictive agents or by using the BPH/5 hypertensive mouse strain21 and crosses between the CBA×DBA strains.22 Gene manipulations have also been used, such as gene invalidation experiments for p57Kip2 or Comt23,24 additive transgenesis of antiangiogenic factors, such as s-flt1,25 or transgenesis leading to manipulation of the renin/angiotensin system.26 In the handful of existing mouse models of PE, gestational hypertension is mild,27 far below the one of our model and far below what is observed in severe PE in women. Furthermore, we observed that the principal maternal symptom, gestational hypertension, seems very early, from 3.5 dpc. In humans, it has been shown that PE can exceptionally occur very early with a complete phenotypic profile before 15 weeks of gestation,28 whereas in general symptoms are reported to appear from midgestation (21–22 weeks) at the earliest. The occurrence of the maternal hypertension at the very beginning of gestation in our model suggests that soluble substances produced by the blastocyst, the embryo, or even from the ejaculate may initiate the maternal preeclamptic phenotype. It is interesting to compare hypertension curves obtained here with those of most studies24,25 that show a raise during the last 5 days of gestation or so. This increase is also visible in our model, but is superimposed on a highly elevated BP triggered by gestation. In sum, our model cumulates presumably the effects of an early message from the embryo–fetoplacental unit to the mother systems controlling BP (endothelial system and kidney) and the effect of an increased placental/fetal demand occurring at the end of gestation, when the rhythm of weight gain is the highest for the fetuses, this effect being the one reproduced in most mouse models of PE.
Under anesthesia, the difference in BP measurement between the control pregnant mice and the ones crossed with transgenic males was present, albeit less pronounced than on vigil animals. This is completely consistent with recent human data showing that under anesthesia there is a steep drop in maternal BP but much more elevated in preeclamptic women than normotensive pregnant women (−10% in normotensive versus −30% in preeclamptic women).13 This suggests that BP variations in our model are very close to what is observed in human clinics.
To the best of our knowledge, we demonstrate here for the first time, in a mouse model, that an aspirin treatment at low doses administered in drinking water from the beginning of gestation is efficient for preventing the maternal syndrome. In humans, it seems that a major issue is the timing of aspirin administration: after 16 weeks of gestation it has no effect at all.29 A consensus tends nevertheless to emerge, with the idea that aspirin is quite efficient when administered early (before the end of the first trimester of pregnancy).29–33 Aspirin at low doses may change the hemodynamic properties of the blood thus preventing blood clotting, an issue that is sometimes associated with disseminated intravascular coagulation in severe PE with HELLP (hemolysis, elevated liver enzyme, low platelet count) syndrome.34
Unexpectedly, mild anomalies of mature placenta (decrease of glycogen trophoblast surface) were only observed in crosses with TgSTOX42 males. Although we did not perform earlier histological examination of developing placenta, focusing on spiral artery invasion for instance, the observation that hypertension occurs before the formation of a placenta indicates that the mechanisms underlying the observed phenotype cannot be fully explained by placentation defects. This opens the possibility that different causes of PE may explain the grade of severity of PE symptoms in humans, as well as the early or late onset of the disease.
Imprinted genes and, among them, the paradigmatic Igf2, are known to have a crucial role in placental and fetal growth.35 In particular, Igf2 invalidation leads to a lack of glycogen cells.36 The observed defaults in glycogen trophoblasts in hemizygous TgSTOX42 placenta may be related to the fact that STOX1 is an imprinted gene in humans and could thus regulate the other members of the imprinted gene network.37
The mouse model presented here can help in the understanding of severe PE, for which until now very few experimental alternatives to the direct study of the patients are available. These mice provide the community working on PE with a tool applicable for cognitive (understanding PE pathogenesis) as well as applied (drug screening) usage, and even potentially deciphering how these drugs act.
Dr Guillermina Girardi is warmly acknowledged for her help and advices for studying kidney histology and preparing the slides. Emma Walton is acknowledged for her advice concerning the manuscript.
Sources of Funding
This work was supported by INSERM.
The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA.111.202994/-/DC1.
- Received July 26, 2012.
- Revision received November 25, 2012.
- Accepted January 5, 2013.
- © 2013 American Heart Association, Inc.
- Sibai BM
- van Dijk M,
- van Bezu J,
- van Abel D,
- Dunk C,
- Blankenstein MA,
- Oudejans CB,
- Lye SJ
- Fenstad MH,
- Johnson MP,
- Løset M,
- Mundal SB,
- Roten LT,
- Eide IP,
- Bjørge L,
- Sande RK,
- Johansson AK,
- Dyer TD,
- Forsmo S,
- Blangero J,
- Moses EK,
- Austgulen R
- Maynard SE,
- Min JY,
- Merchan J,
- Lim KH,
- Li J,
- Mondal S,
- Libermann TA,
- Morgan JP,
- Sellke FW,
- Stillman IE,
- Epstein FH,
- Sukhatme VP,
- Karumanchi SA
- Hu YC,
- Wang PH,
- Yeh S,
- Wang RS,
- Xie C,
- Xu Q,
- Zhou X,
- Chao HT,
- Tsai MY,
- Chang C
- Fogo AB
- Davisson RL,
- Hoffmann DS,
- Butz GM,
- Aldape G,
- Schlager G,
- Merrill DC,
- Sethi S,
- Weiss RM,
- Bates JN
- Kanayama N,
- Takahashi K,
- Matsuura T,
- Sugimura M,
- Kobayashi T,
- Moniwa N,
- Tomita M,
- Nakayama K
- Kumasawa K,
- Ikawa M,
- Kidoya H,
- Hasuwa H,
- Saito-Fujita T,
- Morioka Y,
- Takakura N,
- Kimura T,
- Okabe M
- Falcao S,
- Stoyanova E,
- Cloutier G,
- Maurice RL,
- Gutkowska J,
- Lavoie JL
- Imasawa T,
- Nishiwaki T,
- Nishimura M,
- Shikama N,
- Matsumura R,
- Nagai M,
- Soyama A,
- Koike K,
- Kitamura H,
- Joh K
Novelty and Significance
What Is New?
The first mouse model of severe preeclampsia
Strong clues to prove a role of STOX1 in preeclampsia
What Is Relevant?
Aspirin treatment is efficient to revert the syndrome.
The mice are easy to breed and can be distributed to the researchers interested in preeclampsia or more generally in hypertension.
Mice carrying transgenic embryos overexpressing the human version of the STOX1 transcription factor develop all the symptoms of severe preeclampsia. Aspirin at low doses in the drinking water efficiently corrects these symptoms. This work proves that STOX1 is an important actor in preeclampsia and suggests that the mice generated during this project are an interesting model for understanding some aspects of preeclampsia pathogenesis and testing drugs.