Mechanisms of vascular dysfunction in the interleukin-10-deficient murine model of preeclampsia indicate nitric oxide dysregulation.

Abstract

Preeclampsia is a pregnancy-specific hypertensive disorder characterized by proteinuria, and vascular injury in the second half of pregnancy. We hypothesized that endothelium-dependent vascular dysfunction is present in a murine model of preeclampsia based on administration of human preeclamptic sera to interleukin −10−/− mice and studied mechanisms that underlie vascular injury. Pregnant wild type and IL-10−/− mice were injected with either normotensive or severe preeclamptic patient sera (sPE) during gestation. A preeclampsia-like phenotype was confirmed by blood pressure measurements; assessment of albuminuria; measurement of angiogenic factors; demonstration of foot process effacement and endotheliosis in kidney sections; and by accumulation of glycogen in placentas from IL-10−/− mice injected with sPE sera (IL-10−/−sPE). Vasomotor function of isolated aortas was assessed. The IL-10−/−sPE murine model demonstrated significantly augmented aortic contractions to phenylephrine and both impaired endothelium-dependent and, to a lesser extent, endothelium-independent relaxation compared to wild type normotensive mice. Treatment of isolated aortas with indomethacin, a cyclooxygenase inhibitor, improved, but failed to normalize contraction to phenylephrine to that of wild type normotensive mice, suggesting the additional contribution from nitric oxide downregulation and effects of indomethacin-resistant vasoconstricting factors. In contrast, indomethacin normalized relaxation of aortas derived from IL-10−/−sPE mice. Thus, our results identify the role of IL-10 deficiency in dysregulation of the cyclooxygenase pathway and vascular dysfunction in the IL-10−/−sPE murine model of preeclampsia, and point towards a possible contribution of nitric oxide dysregulation. These compounds and related mechanisms may serve both as diagnostic markers and therapeutic targets for preventive and treatment strategies in preeclampsia.

Graphical abstract

CONCLUSION:

IL-10 deficient murine model of preeclampsia is associated with vascular reactivity dysfunction, due to dysregulation in cyclooxygenase pathway, with a possible role of NO downregulation.

INTRODUCTION

Preeclampsia (PE) is a pregnancy disorder characterized by hypertension accompanied by proteinuria and/or organ dysfunction during the second half of pregnancy.¹ It remains a major cause of maternal and fetal morbidity and mortality worldwide.² It has been increasingly recognized that PE is a heterogeneous disease.³ “Placental” PE has been associated with placental ischemia and impaired angiogenesis, as demonstrated by elevated levels of the soluble fms-like tyrosine kinase receptor-1 (sFlt-1), a soluble receptor of placental origin, which may bind and neutralize pro-angiogenic vascular endothelial growth factor (VEGF), and placental growth factor (PlGF).⁴ This form of PE is commonly associated with an increase in the sFlt-1/PlGF ratio compared to normotensive pregnancies and is clinically characterized by early presentation (<34 gestational weeks), and intrauterine growth restriction (IUGR). “Maternal” PE occurs in the setting of preexisting maternal comorbidities such as chronic hypertension and diabetes mellitus, whereby pregnancy acts as a physiological stress test that exacerbates preexisting endothelial dysfunction. These classifications (placental vs. maternal, early vs. late, severe vs. mild) are likely overly simplistic as these forms tend to overlap. However, their converging point seems to be systemic endothelial dysfunction, resulting in vascular injury, inflammation, and platelet activation.⁵ The underlying molecular mechanisms are still poorly defined, despite recent advances in the understanding of the pathophysiological processes related to this disorder. Normal pregnancy requires a maternal state of immune tolerance to the semi-allogenic fetus. Consequently, Th1 responses are suppressed, while Th2 responses are enhanced (commonly referred to as “Th2 polarization”). ⁶ Th2 polarization results in shifting of the maternal immune system away from cell-mediated, Th1 immunity, which could be detrimental to the fetus. In contrast, PE is characterized by a Th1/Th2 predominant state.^6–8 A key mediator of Th2-mediated immune balance during pregnancy is interleukin 10 (IL-10). Its role in normal pregnancy and dysregulation in PE is supported by several lines of evidence. First, IL-10 plays an important role in normal pregnancy as an anti-inflammatory cytokine,^9,10 with vasoprotective properties,^11–16 and its downregulation may contribute to the known pathophysiological features of PE, its pro-inflammatory milieu and vascular injury. Second, decreased levels of IL-10 are identified in PE and PE-like syndromes.^5,17–19 Third, IL-10 protects against endothelial dysfunction, a hallmark of vascular injury in PE.²⁰ Fourth, a seminal study that used pregnant IL-10^−/− mice, showed that the mice injected with sera from PE patients ²¹ demonstrated classic signs of PE, including hypertension and albuminuria, along with glomerular endotheliosis, placental hypoxic injury, and increased anti-angiogenic factor levels, including fms-like tyrosine kinase receptor-1 (sFlt-1) and soluble endoglin (sEng).This model was further validated by experiments showing that hypoxia, which has long been implicated in the pathophysiology of PE, induced placental injury and increased placental expression and release of sFlt-1 and sEng, with a particularly severe phenotype in the absence of IL-10²². This model of PE, dependent upon IL-10 deficiency along with the plethora of vasoactive factors in PE sera, offered the unique opportunity to investigate clinically relevant features of this disease. However, vascular function and reactivity, which are significantly impaired in PE, have not yet been studied in this model.

We hypothesized that the IL-10 knock-out model of PE would exhibit endothelium-dependent macrovascular dysfunction and that these studies might provide mechanistic data regarding vascular dysfunction in preeclampsia, thus identifying potential therapeutic targets.

RESULTS

1. Characterization of the murine PE model

The following experimental groups were studied: pregnant wild-type (WT) or IL-10 knock-out (IL-10^−/−) mice injected with either normotensive (NT) or pooled severe PE sera (sPE). All four groups were analyzed: one modeling the “normal pregnancy state” (pregnant WT_NT), two intermediate phenotypes, WT_sPE and IL-10^−/−_NT, and the “PE pregnancy state” (pregnant IL-10^−/−_sPE). This design allowed us to study the development of a PE-like state as a function of both the isolated and combined effects of the components of this animal model (i.e., IL-10absence/presence and type of serum). IL-10^−/−_sPE experienced elevated blood pressures and higher albumin/creatinine (Alb/Cr) ratios compared to pregnant IL-10^−/− mice injected with normotensive (NT) sera (IL-10^−/−_NT), as well as pregnant WT mice injected with either sPE (WT_sPE) or NT sera (WT_NT) (p<0.05 for all) (Figure 1A and 1B). The effect of sPE serum injection was pregnancy-specific, producing higher blood pressures and Alb/Cr ratios in pregnant compared to non-pregnant IL-10^−/−_sPE mice (p<0.05 for all) (Figure 1A and 1B). This was accompanied by elevation of the sFlt-1/PlGF ratio in mice sera (Figure 2a), with electron microscopy of the kidney sections in IL-10^−/−_sPE, showing foot process effacement (Figure 2b) and endotheliosis (Figure 2c), which were absent in the other 3 phenotypes (Figure 2d). Furthermore, placentas from IL-10^−/−_sPE showed an increase in glycogen content, similar to glycogen accumulation that has been demonstrated previously in preeclamptic placentas, both in human studies and in animal models ²³ (Figure 3). Descriptive litter data are given in Figure S1. Of note, the pups from the IL-10^−/−_sPE pregnant mice did not show a decrease in the total litter weight, which would argue against placental ischemia, despite abnormal placental findings (e.g., an increase in the placental glycogen content).

Figure 1:

Blood pressure and albuminuria at harvest in WT and IL-10^−/− mice injected with human NT or sPE sera. Systolic blood pressure (A) and urinary Alb/Cr ratios (B) at harvest, n=5 per group. Dots represent individual level data, and lines represent means. One-Way ANOVA with least significant differences as a post hoc test was used for the analysis. (A) Pregnant IL-10^−/−_sPE mice had significantly elevated SBP compared to pregnant WT_NT (p=0.012), WT_sPE (p<0.001), and IL-10^−/−_NT (p=0.002), as well as to non-pregnant IL-10^−/−_sPE (p=0.002) mice. (B) Pregnant IL-10^−/−_sPE mice had significantly increased urine albumin to creatinine ratios compared to pregnant WT_NT (p=0.018), WT_sPE (p=0.007) and IL-10^−/−_NT mice (p=0.003), as well as to non-pregnant IL-10^−/−_sPE (p=0.004) mice.

SBP, systolic blood pressure; Alb/Cr, albumin to creatinine ratio; WT, wild type mice; IL-10^−/−, interleukin-10 deficient mice; PG, pregnant; nPG, non-pregnant, NT, normotensive, sPE, severe preeclamptic

Figure 2: Angiogenic markers and renal histology in WT and IL10^−/− mice injected with either NT or sPE sera.

There was a statistical difference between WT_NT and IL-10^−/−_sPE mice (p=0.016), as well as between WT_sPE and IL-10^−/−_sPE mice (p=0.029). Dots represent individual level data, and lines represent median and IQR. Data were analyzed using a nonparametric Mann-Whitney test (A) Significant foot process effacement (B, black arrows) and endotheliosis (C, black arrows) were observed in IL-10^−/−_sPE mice compared to control group (D). Images were taken by electron microscopy and magnifications were 15,000x, 6,000x and 15,000x, respectively.

WT, wild type mice; IL-10^−/−, interleukin-10 deficient mice; NT, normotensive, sPE, severe preeclamptic

Figure 3: Glycogen content in placental tissue.

Placenta from WT+NT (A) demonstrates decidua (white arrow), junctional zone (grey arrow) without foci of glycogenation, the labyrinth is also represented (black star); Placenta from WT+PE shows small islands of glycogenation black thick arrow) in the junctional zone (B). The placenta from IL10^−/− _NT also shows small islands of glycogenation (black thick arrow) (C). The placenta from IL-10^−/−_sPE shows large confluent areas of glycogenation, replacing much of the junctional zone (D) The WT_NT had 7.8% [0.16–12.3], whereas IL10^−/−_sPE were with 26.72% [16.86–34.48] of glycogenation area in relation to the total area of the junctional zone. There was a significant difference between aforementioned mice (p=0.0095), as well as between WT_sPE (7.9% [3.9–13.1]) and IL10^−/−_sPE groups (p=0.043). No statistical difference was observed between IL10^−/−_sPE and IL10^−/−_NT mice (15.2% [1.2–30.9]). Values presented as median with IQR.

WT, wild type mice; IL-10^−/−, interleukin-10 deficient mice; NT, normotensive, sPE, severe preeclamptic

2. Vascular reactivity in WT and IL-10^−/− aortas after injection with patient sera

We studied phenylephrine-induced contraction; endothelium-dependent relaxation (EDR) to acetylcholine, which is mediated by endothelium-derived NO; and endothelium-independent relaxation (EIR) to diethylammonium (Z)-1-(N,N-diethylamino)diazen-1-IM1,2-diolate (DEA-NONOate), a nitric oxide (NO) donor, which exerts a direct effect on smooth muscle cells. Its relaxation effect, unlike relaxation to acetylcholine, is independent of the presence of intact endothelium. To determine the possible contribution of arachidonic acid metabolites produced by activation of cyclooxygenase (COX) pathway to vascular reactivity, we also studied EDR and EIR in the presence of the COX inhibitor, indomethacin.

Responses to known vasoactive molecules in pregnant IL-10^−/− and WT mice injected with either NT or sPE patient sera are illustrated in Figure 4. Phenylephrine (10⁻⁹–10⁻⁵ M) induced concentration-dependent contractions in all groups (Figure 4A). Pregnant IL-10^−/−_sPE mice exhibited enhanced aortic contractions to phenylephrine compared to pregnant WT_NT (p<0.001), WT_sPE (p=0.002), as well as IL-10^−/−_NT mice (p=0.022). E_max (measure of efficacy, see Methods), and pEC₅₀ (measure of potency, see Methods) for phenylephrine were lowest for WT_NT and highest for IL-10^−/−_sPE mice (22±5 vs. 57±7 for E_max, and 6.9±0.03 vs. 7.0±0.02 for pEC₅₀, respectively, Table 1).

Figure 4:

Aortic reactivity in pregnant WT and IL-10^−/− mice injected with NT and sPE human sera. Concentration-dependent contractions in response to phenylephrine in the absence (A) and presence of indomethacin (D), relaxation in response to acetylcholine in the absence (B) and presence of indomethacin (E), and relaxation in response to DEA-NONOate in the absence (C) and presence of indomethacin (F). Results are expressed as mean ± SD of percent contraction to 80 mM KCl/relaxation to submaximal contraction induced by phenylephrine, n=5 per group. Data were analyzed by repeated-measures ANOVA (between-subjects factor: group; within subjects factor: vasoactive substance concentration). Phenylephrine and acetylcholine had 9 concentration categories (10⁻⁹–10⁻⁵). DEA-NONOate had 11 concentration categories (10⁻¹⁰–10⁻⁵); Indomethacin (10⁻⁵ mol/L; 30 min).

(A) Pregnant IL-10^−/−_sPE mice had significantly higher contractions to phenylephrine compared to the pregnant WT_NT (p<0.001), WT_sPE (p=0.002) and IL-10^−/−_NT (p=0.022) mice. (D) Indomethacin significantly decreased contraction to phenylephrine (p<0.05 for all groups before and after exposure to indomethacin). Even in the presence of indomethacin, the pregnant IL-10^−/−_sPE mice exhibited enhanced contraction to phenylephrine compared to the pregnant WT mice (p=0.001 and p=0.013, in comparison to the WT_NT and WT_sPE mice, respectively). (B) Pregnant IL-10^−/−_sPE mice had significantly decreased endothelium-dependent relaxation to acetylcholine compared to pregnant WT_NT (p=0.004). When compared to pregnant WT_sPE and IL-10^−/−_NT mice, there was a trend towards a difference, although not statistically significant (p=0.140 and p=0.057, respectively). (E) Indomethacin significantly increased relaxation to acetylcholine in all groups (p<0.05 for each groups before and after exposure to indomethacin). There was no significant difference between the groups after exposure to indomethacin (p>0.05 between the groups). (C) Pregnant IL-10^−/−_sPE mice had significantly impaired endothelium-independent relaxation to DEA-NONOate compared to the WT_NT (p=0.007) mice, but this effect was less pronounced than for relaxation to acetylcholine. (F) There was no significant difference between the groups after exposure to indomethacin. Of note, the relaxation before and after indomethacin for WT_NT remained the same (p=0.35) , while all three other groups demonstrated increase in EIR (p<0.05 for each groups before and after exposure to indomethacin).

WT, wild type mice; IL-10^−/−, interleukin-10 deficient mice; PG, pregnant; NT, normotensive, sPE, severe preeclamptic.

* IL10^−/−_sPE vs. WT_NT, ^# IL10^−/−_sPE vs. WT_sPE, ^† IL10^−/−_sPE vs. IL-10^−/−_NT

Table 1:

Efficacy and potency of vasoactive agents in isolated aortic rings

Group	Phenylephrine		Acetylcholine		DEA-NONOate
	pEC50 (−log mol/L)	Emax (%)	pEC50 (−log mol/L)	Emax (%)	pEC50 (−log mol/L)	Emax (%)
WTPG + NT	6.88±0.03	22±5	7.10±0.05	82±3	8.00±0.18	109±1
WT PG + sPE	6.85±0.04	33±2	6.90±0.14	79±4	7.77±0.05	101±1
IL-10−/− PG + NT	6.97±0.03	39±5	6.94 ±0.08	77±4	7.65±0.07	103±2
IL-10−/− PG + sPE	7.00±0.02	57±7	6.68±0.11	65±7	7.84±0.07	100±1
IL-10−/− nPG + sPE	7.02±0.08	57±13	7.17±0.07	88±2	7.89±0.03	103±1

E_max indicates efficacy; pEC₅₀, potency; WT, wild-type; IL10^−/−, interleukin10-knockout; PG, pregnant; NT, normotensive; sPE, severe preeclamptic. Data are means ± SEM (n=5 per group).

Endothelium-dependent relaxation (of aortic segments in response to acetylcholine (10⁻⁹ – 10⁻⁵ M) is presented in Figure 4B. The relaxation to acetylcholine was concentration dependent in all groups. Pregnant IL-10^−/−_sPE mice displayed decreased EDR compared to the pregnant WT_NT (p=0.004) mice, and similar trends in the pregnant WT_sPE and IL-10^−/−_NT mice (p=0.140 and p=0.057, respectively). EDR in the pregnant IL-10^−/−_NT mice was similar to that of the WT_sPE (p=0.629) mice, suggesting similar EDR responses in the intermediate phenotypes. E_max and pEC₅₀ for acetylcholine were highest in the WT_NT and lowest for the IL-10^−/−_sPE mice (82±3 vs. 65±7 for E_max, and 7.1±0.05 vs. 6.7±0.11 for pEC₅₀, respectively, Table 1).

Figure 4C illustrates the EIR of aortic segments for the groups in response to DEA-NONOate (10⁻¹⁰–10⁻⁵ M), which acts directly on vascular smooth muscle and induces relaxation. Pregnant IL-10^−/−_sPE mice displayed decreased EIR compared to the pregnant WT_NT mice (p=0.007), while there was no difference in contrast to intermediate phenotypes, WT_sPE and IL-10^−/−_NT mice (p>0.05). However, the mean difference *(MD) of EIR in the pregnant IL-10^−/−_sPE compared to pregnant WT_NT mice was less pronounced (MD: 8%) compared to the differences seen in EDR (MD: 15%) for the same groups. These findings suggest that in this PE model, reactivity of aortic smooth muscle cells to NO is slightly impaired.

Additional mechanistic insights into the observed vascular dysfunction were gained through ex vivo exposure of murine aortas to indomethacin, a potent COX inhibitor (Figure 4D, 4E, 4F). Treatment with indomethacin (10⁻⁵ mol/L; 30 min) significantly decreased phenylephrine-induced contractions (4D) in both the pregnant IL-10^−/− and pregnant WT mice (p<0.05 for all groups before and after exposure to indomethacin). Even in the presence of indomethacin, the pregnant IL-10^−/−_sPE mice exhibited enhanced contractions to phenylephrine compared to the pregnant WT mice (p=0.001 and p=0.013, compared to WT_NT and WT_sPE, respectively. Indomethacin significantly increased acetylcholine-induced relaxation of aortic rings (4E) in both pregnant IL-10^−/− and pregnant WT mice (p<0.05 for each group before and after exposure to indomethacin). However, the previously noted differences in relaxation between the groups (4B) were no longer seen after indomethacin treatment (p>0.05 between the groups) (Figure 4E), suggesting that the observed alteration in EDR is predominantly caused by enhanced activation of the COX pathway. Similar to EDR, the previously noted differences in EIR relaxation between the groups (4C) were no longer evident after exposure to indomethacin (4F).

In order to assess whether the vascular dysfunction was pregnancy-specific, we analyzed vascular function in pregnant vs. non-pregnant IL-10^−/−_sPE mice. As shown in Figure 5, contractions to phenylephrine and EIR responses did not vary between the pregnant and non-pregnant groups (phenylephrine: p=0.971, Figure 5A; DEA-NONOate: p=0.138; Figure 5C), in contrast to the impaired EDR found in the pregnant IL-10^−/−_sPE mice (p=0.007; Figure 5B). Incubation with indomethacin reduced contractions to phenylephrine, while simultaneously improving EDR in the IL-10^−/−_sPE mice, irrespective of pregnancy status (p<0.05 for all comparisons, Figure 5D and 5E). There were no significant differences in relaxation to DEA-NONOate in aortas treated with indomethacin (p=0.124).

Figure 5:

The effect of pregnancy in IL-10^−/− mice injected with sPE sera on aortic reactivity. Concentration-dependent contraction to phenylephrine in the absence (A) and presence of indomethacin (D), relaxation to acetylcholine in the absence (B) and presence of indomethacin (E), and relaxation to DEA-NONOate in the absence (C) and presence of indomethacin (F). Results are expressed as mean ± SD of percent contraction to 80 mM KCl/percent relaxation to submaximal contraction induced by phenylephrine, n=5 per group. Data were analyzed by repeated-measures ANOVA (between-subjects factor: group; within subjects factor: vasoactive substance concentration). Phenylephrine and acetylcholine had 9 concentration categories (1E-9 – 1E-5) and DEA-NONOate had 11 concentration categories (1E-10 – 1E-5); Indomethacin (10⁻⁵ mol/L; 30 min).

(A) There were no significant differences in contraction to phenylephrine between the pregnant and non-pregnant IL-10^−/−_sPE mice (p=0.971). (D) Indomethacin significantly decreased contractions to phenylephrine in both the pregnant and non-pregnant IL-10^−/−_sPE mice (p<0.05 for both). (B) Pregnant IL-10^−/−_sPE mice had significantly impaired endothelium-dependent relaxation compared to the non-pregnant IL-10^−/−_sPE mice (p=0.007). (E) Indomethacin improved endothelium-dependent relaxation in both the pregnant and non-pregnant IL-10^−/−_sPE mice (p<0.05 for both), and there were no differences between the groups after exposure to indomethacin (E). There were no significant differences in relaxation to DEA-NONOate between the pregnant and non-pregnant IL-10^−/−_sPE mice (p=0.138) before (C) and after indomethacin (p=0.124) (F).

WT, wild type; IL-10^−/−, interleukin-10 knockout; PG, pregnant; nPG, non-pregnant; NT, normotensive; sPE, severe preeclamptic, * Pregnant IL10^−/−_sPE vs. non-pregnant IL-10^−/−_sPE

To further elucidate the molecular mechanisms responsible for the observed vascular dysfunction, we analyzed the aortic gene expressions of COX (Table S2). There were no differences in the expressions of COX isoforms, COX1 and COX2, between the IL-10^−/−_sPE and WT_NT groups (p=0.281 and p=0.281, respectively; Table S2).

DISCUSSION

To the best of our knowledge, our present study reports several novel findings obtained in an IL-10^−/− mouse model of PE that provide mechanistic data for the role of IL-10 deficiency in human disease. First, we demonstrate that pregnant IL-10^−/−_sPE exhibits several features of human disease, including angiogenic imbalance (i.e. elevated sFlt-1/PlGF ratio); foot process effacement and endotheliosis, which are considered to be classical renal histological findings in PE; and placental glycogen accumulation which is increasingly recognized and considered to be the result of either impaired utilization or compensation for placental dysfunction in PE. Second, we detected altered vasomotor function of isolated aortas, including enhanced contractions to phenylephrine, as well as impaired EDR to acetylcholine and, to a lesser extent, EIR to DEA-NONOate. Third, the impairment of vascular function was pregnancy-specific and predominantly derived from impaired EDR. Fourth, indomethacin improved vasoconstriction and corrected vasodilator dysfunction, thus suggesting enhanced activation of the COX pathway, with an imbalance between vasodilatory and vasoconstricting prostaglandins, in favor of the latter.²⁴ Taken together, these results characterize vascular dysfunction and identify the COX pathway as the underlying mechanism in the setting of decreased IL-10 levels. Based on the data showing that indomethacin improved, but failed to normalized contraction to phenylephrine in the murine model of PE, our study further points towards a possible contribution of indomethacin-resistant vasoconstricting factors and/or NO dysregulation in vascular damage in PE.

Both a preeclampsia-like phenotype and vascular dysfunction in the present model require a combination of a) pregnant state; b) IL-10 deficiency; and c) exposure to sPE sera. The use of either of these alone failed to recapitulate the extent of vascular dysfunction observed. This combination establishes the clinical fidelity of this model because IL-10 levels are reduced in clinical PE⁵. The relevance of these findings with respect to the clinical PE phenotype is further underscored by the effects of pregnancy on the severity of the phenotype: hypertension and albuminuria were the most severe in pregnant IL-10^−/− mice injected with sPE sera; non-pregnant IL-10^−/− mice injected with sPE sera behaved similarly to pregnant IL-10^−/− mice injected with NT sera. We recognize, however, that our results obtained in isolated aortas (large conduit vessels) may not reflect changes in small resistance arteries.

The IL-10 knock-out model is commonly used for studies of aging, and senescence, an irreversible cell-cycle arrest mechanism that leads to senescence-related injury that is, in part, mediated by the release of pro-inflammatory markers.²⁵ Previous reactivity studies in non-pregnant IL-10^−/− mice suggest that IL-10 deficiency predisposes to EDR impairment in response to vascular stressors such as angiotensin II (ANGII),^26,27 endothelin-1 (ET-1),¹³ tumor necrosis factor α (TNF-ɑ),^11,28 lipopolysaccharide (LPS),^14,16 chronic hyperglycemia,¹⁵ and aging.^20,29 Taken together, these data imply that a “second hit” is needed to induce vascular dysfunction in the setting of IL-10 deficiency. Indeed, PE sera contain many vasoactive mediators, including TNF-ɑ, ANGII, stimulating ANGII receptor-1 auto-antibodies (AT1-AA), and ET-1.^30–33

Vascular reactivity studies demonstrated heightened contractions to phenylephrine and both impaired EDR and, to a lesser extent, EIR in the pregnant IL10^−/−_sPE mice. With EDR demonstrating a greater effect, the observed impaired relaxation in this model primarily reflects endothelial dysfunction. The clinical corollaries to our observations are human studies supporting both exaggerated responses to vasoconstrictors and endothelial dysfunction in women with PE. For example, studies of omental arteries from women diagnosed with PE demonstrate increased sensitivity to vasoconstrictors, ANGII and phenylephrine, compared to those measured from the omental arteries of NT women.³⁴ Furthermore, these ex vivo observations are consistent with in vivo reports of differential responses to ANGII between NT and PE women.^35–37 Finally, endothelial dysfunction, as measured by flow-mediated dilation, was lower in women with PE compared to those without PE, both before the development of PE (≈20–29 weeks gestation), and at the time of PE³⁸ .

To explore the underlying mechanisms for the observed vascular dysfunction, mouse aortas were incubated in the presence of the nonselective COX inhibitor, indomethacin, as previously performed in hypertensive rats.³⁹The rationale is based on a prior study showing that, under pathological conditions, such as hypertension, COX-derived vasoconstrictor products could be responsible for impairment of EDR to acetylcholine.^39,40 Indomethacin prevented both the increased contractions to phenylephrine, as well as the impairment of both EDR and EIR in IL-10^−/−_sPE mice (Figure 4). However, in contrast to EDR and EIR, where the differences between the groups were no longer present after treatment with indomethacin, the IL-10^−/−_sPE mice continued to demonstrate significant residual, indomethacin-resistant contraction. As for the mechanism(s) for failure to normalize phenylephrine-induced contraction, we propose the possible roles of at least two mechanisms. First, we suggest that decreased NO bioavailability/production may be playing a role in the residual contraction based on the previous studies indicating that inactivation of basal production of NO by eNOS enhances the vasoconstrictor effect of phenylephrine⁴¹. In addition, decreased NO bioavailability/production may be driven by the increased production of superoxide anion by cyclooxygenase,⁴² which inactivates NO. Second, PE has been associated with several vasoconstricting mediators (e.g., endothelin-1 and angiotensin II), which are COX-independent, and thus may be contributing to the indomethacin-resistant component of phenylephrine-induced contraction. Angiogenic disbalance may potentiate the vasoconstricting effects of these mediators as shown by the action of sFlt-1(elevated levels in IL-10^−/−_sPE mice), which leads to enhanced angiotensin II sensitivity,⁴³ one of the hallmarks of human disease. The link between these two mechanisms that may contribute to indomethacin-resistant contraction has been supported by studies demonstrating that angiotensin II sensitivity can be reversed by sildenafil, a phosphodiesterase inhibitor, which enhances NO signaling,⁴³ thus further supporting the role of decreased NO bioavailability/production in vascular injury in PE.

Current observations, interpreted in the context of the previous studies, suggest that both the COX and the NO pathways are involved, but that the primary abnormality is in the COX pathway, principally because indomethacin blocked the effect of PE sera and restored EDR in the pregnant IL-10^−/−_sPE mice. Comparable mRNA expression levels of COX1 and COX2 were detected, however, in the IL-10^−/− mice injected with sPE sera and the WT mice injected with NT sera. Several considerations are relevant to these clear-cut functional effects of COX pathway inhibition on the one hand, and yet comparable expressions of COX1 and COX2 on the other: i) Vascular dysfunction may derive from factors upstream or downstream in the prostaglandin synthesis pathway; ii) significant differences in COX1/2 protein and mRNA expression levels may exist; and iii) enzymatic activity of COX1/2 may be altered. An additional consideration is that COX activity may generate superoxide anion which, in turn, inactivates NO thereby causing impairment of EDR.⁴⁴ Indeed, PE plasma provokes EDR impairment in normal pregnant rat uterine arteries, and this adverse effect is abolished by a COX inhibitor, meclofenamide.⁴⁵ Our present findings, in conjunction with such studies, support the beneficial effects of aspirin usage in pregnant women who are at high risk for PE.⁴⁶

Several limitations are inherent in this particular PE model. One consideration is the variability of the PE patients whose sera were used. In order to better control for confounding variables, we employed pooled sera from several donor samples. Future studies would greatly benefit from using sera obtained from patients diagnosed with both mild and severe PE phenotypes, in order to characterize vascular injury across the spectrum of PE severity.

While reactivity studies with large conduit vessels were novel, the use of smaller resistance vessels may constitute a better indicator of peripheral resistance in this model. In addition, we did not measure relevant prostaglandin classes, but, based on the response to indomethacin, we concluded that there is a relative excess of vasoconstrictors, such as thromboxane, in our model. This would be consistent with human studies that have shown an imbalance in vasoactive prostaglandins, further characterized by reduced prostacyclin (vasodilatory agent) and unchanged levels of thromboxane, a vasoconstrictor, in preeclamptic compared to normotensive pregnancies.²⁴ Regardless, important insights into the endothelial dysfunction typified in PE were observed with the use of aortas, the latter representing an established approach generally corroborated by the use of animal⁴⁵ and human³⁴ resistance vessels.

In summary, our results emphasize the role IL-10 deficiency in exaggerated vasoconstrictive responses and impaired vascular relaxation as mediated through both EDR and EIR, although the relative contribution of the EDR component appears greater. Based on responses to indomethacin, we conclude that the COX pathway is the primary abnormality, thus providing mechanistic data to support aspirin use for PE prevention. In addition, down-regulation of NO may contribute to impaired vascular responses, thus supporting the use of sildenafil in PE. Of note, the use of sildenafil and resultant increase in cyclic GMP (a second messenger for NO) has proven beneficial in reversing vasoconstriction in an ex vivo human placental model.⁴⁷ In a subsequent step, we will decipher the role and relative contribution of NO dysregulation in vascular injury in the IL-10 deficient mouse model of PE. Of note, previous studies have indicated that IL-10 induces the production of NO ⁴⁸ and that IL-10 infusion blunts the rise in blood pressure in response to angiotensin II.⁴⁹ Future studies should explore the use of rIL-10 and/or other means of IL-10 administration as therapeutic options for PE, as well as COX and NO compounds and related mechanisms, which may serve both as diagnostic markers and therapeutic targets for prevention and treatment of preeclampsia.

METHODS

Patient sera

A convenience sample was obtained from each of five normotensive (NT) and five women diagnosed with severe PE (sPE) who were admitted to the Family Birth Center of the Mayo Clinic Department of Obstetrics and Gynecology. Preeclampsia was diagnosed as an elevated blood pressure (≥ 140/90 mm Hg) and either proteinuria (≥ 300 mg protein in a 24-h urine or urinary protein/creatinine ratio > 0.3), and/or organ dysfunction during the second half of pregnancy, according to accepted criteria.¹ The diagnosis of severe PE was based on a systolic BP≥160 or diastolic BP≥110 mm Hg and/or signs and symptoms of end-organ damage,¹ and further confirmed by an elevated sFlt-1/PlGF compared to normotensive controls (Table S1 and Figure S2). Experimental Details are provided in Supplementary Methods.

Animals and Interventions

All animal studies were approved by the Mayo Clinic Institutional Animal Care and Use Committee (no. A-0000 2139–16), and in accordance with NIH guidelines). WT and IL-10^−/−C57BL/6 mice were obtained from The Jackson Laboratory (IL-10^−/− mice stock no. 002251; WT mice stock no. 000664, Bar Harbor, ME). Experimental Details are provided in Supplementary Methods.

Albuminuria assessment and blood pressure measurement

Urine albumin concentration , as a marker of glomerular injury, was determined using a mouse-specific ELISA kit (cat no. 41-ALBMS-E01, ALPCO Diagnostics, Salem, NH) as directed by the manufacturer. Results were standardized by measuring urine creatinine with a creatinine colorimetric kit (cat no. 500701, Cayman Chemical, Ann Arbor, MI). Urine Alb/Cr ratios were reported as μg of albumin/mg of creatinine. Blood pressures were measured using a noninvasive computerized tail-cuff system (CODA, Kent Scientific Corporation, Torrington, CT).⁵⁰

Flt-1 and PlGF in serum

Both Fms-related tyrosine kinase 1 (Flt1) and placental growth factor (PlGF) were determined in human and mouse serum samples according to manufacturer protocol. The following ELISA kits were used (i) Human PlGF ELISA kit (ab100629, Abcam) (ii) Human VEGF R1 ELISA Kit (FLT1) (ab195210, Abcam) (iii) Mouse SimpleStep PlGF ELISA kit (ab197748, Abcam) (iv) Mouse Flt1 ELISA Kit (ab231935, Abcam).

Electron microscopy

Tissue was fixed in Trump’s fixative (1% glutaraldehyde and 4% formaldehyde in 0.1 M phosphate buffer, pH 7.2.⁵¹ Experimental Details are provided in Supplementary Methods.

Glycogen deposits in placenta

Sections of placenta were stained with PAS/HE. Experimental Details are provided in Supplementary Methods.

Vascular reactivity studies

The aortas were dissected and cleaned of fat and surrounding connective tissue in cold Krebs solution containing (in mmol/L): 118.6 NaCl; 4.7 KCl; 2.5 CaCl₂; 1.2 MgSO₄; 1.2 KH₂PO₄; 25.1 NaHCO₃; 10.1 glucose; and 0.026 EDTA. Five mm-long rings were suspended in organ chambers for isometric force measurements (PowerLab, AD Instruments, Colorado Springs, CO) and the data were acquired with Labchart Pro software (AD Instruments) as previously described.⁵² Experimental Details are provided in Supplementary Methods.

qPCR studies

mRNA expression in aortas was assessed in the pregnant IL-10^−/− and WT mice as described in our prior studies.^53,54 Experimental Details are provided in Supplementary Methods.

Statistical Analysis

Patient data are presented as means with standard deviations for continuous normally distributed variables and analyzed using an unpaired t test under the assumption of equal variance. Continuous non-normally distributed patient data are presented as medians with 25–75th percentiles and analyzed using the Mann Whitney U test. Data from animal studies are presented as individual level data, with corresponding summary measures, i.e. means, as well as standard deviations or standard errors. Blood pressure and albuminuria data were analyzed using One-way ANOVA (Post hoc: least significant difference). The effects of serum type, IL-10 availability, and prostaglandin pathway dysfunction on aortic vascular reactivity were examined by repeated measures ANOVA, with experimental animal group as a between subjects factor and vasoactive substance concentration as a within subjects factor, in the absence or presence of indomethacin. Aortic gene expression data were evaluated using an unpaired t test under assumption of equal variance. Analyses were performed using SPSS for Windows (v25.0; IBM SPSS). Data are presented according to recently proposed guidelines for basic science data visualization ⁵⁵A power calculation was not performed in advance, as this study is regarded as exploratory due to the lack of published data in this particular model.

Supplementary Material

1

SUPPLEMENTARY MATERIAL

Supplementary methods: Patient sera, Animals and Interventions, qPCR, Electron microscopy, Glycogen deposits in placenta, Vascular reactivity studies

Table S1: Characteristics of patients providing serum samples

Table S2: Aortic COX gene expression analyses in WT and IL-10^−/− mice

Figure S1: Fetal characteristics in WT and IL-10^−/− mice injected with human NT and sPE sera. Total litter weights (A) and fetal lengths (B) in pregnant female WT and IL-10^−/− mice. n = 5 per group. WT indicates wild type; IL-10^−/−, interleukin-10 knockout; PG, pregnant; nPG, non-pregnant; NT, normotensive; sPE, severe preeclamptic. Data were analyzed by One-Way ANOVA with least significant differences as post hoc test: Total litter size: F=4.937, df=3, p=0.014 (IL-10^−/− PG + NT had lowest total litter weight in comparison to other groups); Fetal length: F=1.500, df=3, p=0.258).

Figure S2: Individual sFlt-1/PlGF ratio values for human serum samples used to inject the animals. There was a significant difference between women with preeclampsia and normotensive women (Table S1).

TRANSLATIONAL STATEMENT.

Interleukin 10 (IL-10), a key mediator of Th2-immune responses, plays a major role in in normal pregnancy as an anti-inflammatory cytokine with vasoprotective properties. Clinical studies have provided data that IL-10 levels are decreased in preeclampsia (PE) compared to normotensive pregnancies. To study the role of IL-10 in the vascular injury associated with PE, we used pregnant interleukin-10^−/− mice injected with severe preeclamptic sera, an animal model that proved to successfully recapitulate key clinical features of PE. Our results characterize vascular dysfunction in PE as enhanced vasoconstriction, impaired endothelium-dependent and, to a lesser extent, endothelium-independent relaxation, with a possible contribution from nitric oxide (NO) pathway dysregulation. Treatment of isolated aortas with indomethacin, a cyclooxygenase (COX) inhibitor, which improved contraction and normalized relaxation, provides mechanistic data that justify aspirin use for PE prevention. Additional studies exploring the role of NO down-regulation in vascular dysfunction in PE may provide preclinical data that will support the use of sildenafil, (a phosphodiesterase 5-inhibitor, which leads to increased eNO-mediated vasodilation) in PE. Finally, future studies exploring the incorporation of rIL-10 and/or other means of IL-10 administration as therapeutic options for PE will be of particular clinical interest and, based upon our study that utilized severe preeclamptic sera only, should discriminate between different forms of PE based on their severity.