Prior Exposure to Experimental Preeclampsia Increases Atherosclerotic Plaque Inflammation in Atherogenic Mice

Abstract

Background

Women with a history of preeclampsia (PE) have evidence of premature atherosclerosis and increased risk of myocardial infarction and stroke compared to women who had a normotensive pregnancy. Whether this is due to common risk factors or a direct impact of prior PE exposure has never been tested in a mouse atherosclerosis model.

Methods

Pregnant low density lipoprotein receptor knockout (LDLR-KO, n=35) female mice were randomized in mid-gestation to sFlt1-expressing adenovirus or identical control adenovirus. Post-partum, mice were fed high fat diet for 8 weeks to induce atherogenesis. Comparison between control and PE model was made for metabolic parameters, atherosclerosis burden and composition by histology, plaque inflammation by flow cytometry, and aortic cytokines and inflammatory markers using a cytokine array.

Results

In pregnant LDLR-KO mice, sFlt1 adenovirus significantly induced serum sFlt1, blood pressure, renal endotheliosis and decreased pup viability. After 8 weeks of post-partum high fat feeding, body weight, fasting glucose, plasma cholesterol, HDL, LDL were not significantly different between groups with no change in aortic root plaque size, lipid content or necrotic core area. Flow cytometry demonstrated significantly increased CD45+ aortic arch leukocytes and CD3+ T cells and aortic lysate contained more CCL22 and fetuin A and decreased expression of IGFBP6 and CCL21 in PE-exposed mice compared to controls.

Conclusion

In atherogenic LDLR-KO mice, exposure to sFlt1-induced PE during pregnancy increases future atherosclerotic plaque inflammation, supporting the concept that PE directly exacerbates atherosclerotic inflammation independent of pre-existing risk factors. This mechanism may contribute to ischemic vascular disease in women after PE pregnancy.

Graphical Abstract

Introduction

Preeclampsia (PE) is a syndrome of high blood pressure combined with target organ damage which occurs late in pregnancy.¹ PE affects approximately 5-8% of pregnancies and is one of the leading causes of global maternal mortality.^2,3 The incidence of PE is rising, in part because pre-gestational obesity is a risk factor for PE.⁴ Despite resolution of PE symptoms post-partum, clinical studies reveal women with prior PE have increased atherosclerosis as measured by cardiac computed tomography⁵, increased carotid artery intimal-medial thickness^6-9 and higher coronary artery calcium scores.^10-13 Moreover, there is a 2 fold^4,14 increased risk of atherosclerotic morbidity¹⁵, myocardial infarction,^16,17 and stroke^18,19 in women who have had PE versus women with normotensive pregnancies. The development of cardiovascular disease (CVD) also occurs sooner in women with a history of PE and can be evident 5-15 years post-partum.²⁰ Pre-pregnancy obesity and hypertension are shared risk factors for PE and CVD. Hence, the high post-partum CVD risk after PE is often attributed to these pre-existing risk factors. However, confounder adjustment shows residual increased CVD risk not explained by obesity, hypertension or traditional risk factors.^5,11 Whether PE identifies an at-risk population due to the cardiovascular stress test of pregnancy or directly enhances atherosclerosis development and promotes future ischemic events is still unclear and is difficult to discern from clinical data and meta-analyses.

PE is likely due to impaired placental development leading to placental ischemia. The ischemic placenta is a source of anti-angiogenic proteins, including soluble vascular endothelial growth factor (VEGF) receptor-1 (sFlt1), found in high levels within maternal circulation.²¹ Circulating sFlt1 levels are significantly increased during pregnancy in women who develop PE and this rise in sFlt1 precedes clinical PE diagnosis and correlates with PE severity^22-24, suggesting a pathogenic role. The relevance of high sFlt1 is underscored by the recent FDA approval of an sFlt1 risk stratification test in suspected preeclampsia.²⁵ sFlt1 binds and sequesters VEGF and placental growth factor (PlGF), inducing hypertension and end organ damage, including renal endotheliosis. After placental delivery, sFlt1 levels drop significantly.²¹ In rodent models, adenoviral sFlt1 overexpression during pregnancy is sufficient to reproduce the PE phenotype.^21,26-28 Using the sFlt1-induced PE model in C57Bl/6 mice during pregnancy, we have demonstrated that high circulating sFlt1 results in increased blood pressure and kidney damage in late pregnancy, and the high sFlt1 and resulting syndrome all normalize post-partum, consistent with the human phenotype.²⁹ Several months post-partum, mice with prior PE have an increased blood pressure response to high salt intake and enhanced vascular remodeling in response to injury.^29,30 These findings suggest persistent changes, making the vasculature more susceptible to dysfunction even after PE is resolved.

Atherosclerosis is a chronic inflammatory disease initiated by vascular damage due to hyperlipidemia and other cardiovascular risk factors.³¹ Vascular damage allows monocytes to infiltrate the vascular wall where they take up lipids, form foam cells, recruit additional leukocytes thereby forming a thrombogenic plaque core. The amount of inflammatory cells within an atherosclerotic plaque correlates with the risk of plaque rupture in humans³¹, which is the cause of heart attack and acute ischemic stroke, which are exacerbated after PE. Low density lipoprotein receptor knockout (LDLR-KO) mice are a well-known atheroprone mouse model which develop atherosclerosis only on a high fat diet.³² Conversely, the PE phenotype resolves post-partum in the sFlt1-induced PE mouse model.^29,30 Combining these models can be leveraged to study whether prior exposure to PE directly contributes to atherosclerotic responses to hyperlipidemia initiated after pregnancy, without confounding of pre-existing risk factors. We hypothesized exposure to the sFlt1-induced PE model would exacerbate plaque inflammation in response to hyperlipidemia-induced atherosclerosis post-partum.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request. Please see the supplemental materials for Major Resources Table and expanded methods.

All procedures were performed in accordance with National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Tufts University Institutional Animal Care and Use Committee. Female LDLR-KO on a C57Bl/6J background were mated and gestation day (GD)0 was determined by the presence of a vaginal plug. On GD9, mice were consecutively randomized 1:1 to injection with sFlt1 versus control adenovirus (an identical empty adenoviral vector; CMV-null, Vector BioLabs). Facial vein blood was collected GD17 to measure sFlt1. From a subset of mice, blood pressure was measured via radiotelemetry and kidneys were harvested for glomerular endotheliosis measurements. Post-partum, mice were immediately started on high fat diet for 8 weeks and then plasma was collected for lipid measurement, aortic root processed for histology, and arch for flow cytometry, as previously described.^33,34 Flow cytometry counts for each sample were normalized as fold change compared to controls for each experimental day. In a subset of mice (n=4/group), protein was extracted from the thoracic/abdominal aorta with perivascular fat and adventitia intact and cytokines quantified with the Mouse XL Cytokine Array Kit (Bio-Techne ARY028).

Statistics

Using Graph Pad Prism-8 (for 1-2D) or R (2E-I), normality was determined using Shapiro-Wilk test and data that passed normality (p>0.05) analyzed by unpaired t-test with Mann-Whitney test for non-normally distributed data. For the cytokine array, the Benjamini-Hochberg corrected p-values were also calculated to adjust for multiple testing. Error bars represent standard error of the mean. P<0.05 was considered statistically significant.

Results

Exposure of pregnant LDLR-KO to the sFlt1-induced PE model recapitulates PE signs without altering post-PE atherosclerosis risk factors

Pregnant LDLR-KO mice were randomized to sFlt1 or identical control adenovirus (Figure 1A). At the end of pregnancy, sFlt1 was significantly increased in mice injected with sFlt1 adenovirus (Figure 1B). In a subset of mice with telemetric blood pressure monitors during pregnancy, a significant increase in blood pressure was confirmed after sFlt1 injection (Figure 1C), along with evidence of glomerular endotheliosis (Figure 1D). Exposure to sFlt-1-induced PE significantly reduced the number of live pups on postnatal day 1 (Figure 1E). In the mice with telemetry, blood pressure normalized and was not significantly different between groups by approximately one week post-partum (Figure 1F). Post-partum, mice were fed high fat diet for 8 weeks (Figure 1G) and as expected, body weight (Figure 1H), fasting glucose (Figure 1I) and cholesterol (Figure 1J) were elevated, with no significant differences in those traditional risk factors, including HDL, LDL and triglycerides, between mice exposed to prior control pregnancy versus sFlt1-induced PE.

Figure 1. sFlt1 induces preeclampsia phenotypes in LDLR-KO mice without altering post-preeclampsia metabolic parameters or atherosclerotic plaque size.

(A) Pregnant low density lipoprotein receptor knockout (LDLR-KO) mice were randomized to control or sFlt1 adenovirus injection to induce the preeclampsia phenotype at gestation day 9. At the end of pregnancy (gestation day 17) in sFlt1-injected mice; (B) Plasma sFlt1 (***p<0.0001, unpaired t-test with Welch’s correction), (C) systolic blood pressure (**p=0.006, unpaired t-test), and (D) glomerular endotheliosis (**p=0.008, unpaired t-test) were significantly increased. (E) On postnatal day 1, the total number of live pups were decreased (**p=0.01, unpaired t-test). (F) By about 1 week post-partum, blood pressure was not different between groups. (G) Post-partum mice were then fed high fat diet for 8 weeks and metabolic parameters measured: (H) body weight, (I) fasting glucose and (J) total cholesterol, triglycerides, HDL and LDL were not different between sFlt1 and control groups. (K) Representative images of aortic root atherosclerotic plaques stained with Oil Red O. Scale bar=100μm. Quantification of; (K) plaque size (example of plaque tracing indicated by green dashed line), (L) relative plaque lipid content and (M) plaque necrotic core size after 8 weeks of high fat diet. F: p=0.23, H: p=0.86, I: p=0.99, J: p=0.83 (Cholesterol), p=0.83 (triglycerides), p=0.94 (HDL), p=0.99 (LDL) K: p=0.60, L: p=0.98, M: p=0.81 via Mann-Whitney test.

Prior exposure to the sFlt-induced PE model during pregnancy does not affect atherosclerosis plaque size or plaque lipid and necrotic core content

Aortic root sections were stained with Oil-Red-O and plaque size (Figure 1K), the relative amount of plaque containing lipid (Figure 1L), and the percent of the plaque composed of necrotic core (Figure 1M) were not different between control and PE-exposed mice.

Prior exposure to the sFlt-induced PE model during pregnancy results in increased aortic arch plaque inflammation and descending aorta inflammatory cytokine expression

Plaque leukocytes were quantified in aortic arch single cell suspensions using flow cytometry. Mice exposed to prior PE had a significant 1.6 fold increase in total plaque leukocyte content as measured by CD45+ cells (Figure 2A). Both CD11b+ myeloid cells (Figure 2B) and F480+ macrophages (Figure 2C) tended to be increased with prior PE exposure but this was not statistically significant. The number of CD45+, CD11b−, CD3+ T cells was significantly increased in the aortic arch of mice previously exposed to sFlt1-induced PE versus control pregnancy (Figure 2D).

Figure 2. Prior preeclampsia increases vascular inflammation in atherosclerosis.

LDLR knockout mice were exposed to control virus (N = 11) or sFlt1-induced preeclampsia (N =10) during pregnancy followed by 8 weeks of post-partum high fat diet and aorta were harvested. Flow cytometric quantification of aortic arch plaque inflammatory cells: (A) total plaque leukocytes (CD45+ cells), *p=0.032 via unpaired t test; (B) myeloid cells (CD45+, CD11b+, CD3−), p=0.062 via unpaired t test; (C) macrophages (CD45+, CD11b+, Ly6G−, F4/80+), p=0.16 via Mann-Whitney test. (D) T cells (CD45+, CD11b−, CD3+), *p = 0.031 via unpaired t test. Tissue lysate from N = 4 descending aortas in each group were subjected to aortic cytokine array to quantify inflammatory mediators. (E) The waterfall plot displays the mean of log₂ (fold change) for each protein in the array that achieved a threshold fold change (dashed line) of ±log₂(0.25) comparing aortas from sFlt1 versus control pregnancy. *p<0.05 by unpaired t test and ** q<0.25 using Benjamini-Hochberg correction. The fold change of the raw signal for; (F) CC motif chemokine ligand 22 (CCL22) *p=0.048, (G) Fetuin A, *p=0.018, (H) Insulin-like growth factor binding protein 6 (IGFBP6) *p=0.029, and (I) CCL21 *p=0.034. Dots represent individual aortas and bars represent the mean +/− SEM. Unpaired student’s t test was used for F, G, and I. Mann-Whitney test was used for H.

Protein lysate from the remaining aortic tissue was analyzed for cytokine expression using an array. The waterfall plot displays the mean fold change between sFlt1 and control mice for each analyte with a fold change of at least log₂(0.25) (Figure 2E). Overall, 38 proteins met this criteria, of which 23 proteins increased and 15 proteins decreased comparing mice previously exposed to sFlt1 versus control pregnancy. CC motif chemokine ligand 22 (CCL22, Figure 2F) and Fetuin A (Figure 2G) were significantly increased while Insulin like growth factor binding protein 6 (IGFBP6, Figure 2H) and CCL21 (Figure 2I) significantly decreased in aortas from mice exposed to prior sFlt1-induced PE.

Discussion

This study examines post-partum atherosclerosis in an established PE mouse model that recapitulates the human phenotype of increased blood pressure, plasma sFlt1, and renal endotheliosis with decreased pup viability in LDLR-KO mice. The results indicate that exposure to sFlt1-induced PE during pregnancy in LDLR-KO mice promotes future atherosclerotic plaque inflammation independent of changes in metabolic risk factors. Specifically, prior exposure to sFlt1 during pregnancy: (1) is sufficient to result in increased plaque CD45+ leukocytes in response to hyperlipidemia; (2) the increase in CD45+ cells is primarily driven by a significant increase in CD3+ T cells; (3) the plaque burden, lipid content and necrotic core content are unchanged; (4) this associates with increased descending aortic expression of CCL22 and Fetuin A and decreased expression of IGFBP6 and CCL21; (5) increased plaque inflammation occurred in the absence of differences in post-partum blood pressure, body weight, lipids, and fasting glucose. This is the first evidence to our knowledge that prior PE exposure enhances future atherosclerotic plaque inflammation in a preclinical atherosclerosis model. Since inflamed plaques are more likely to rupture in humans³², this mechanism could contribute to the significantly increased risk of myocardial infarction and stroke in women after PE.²⁰

Our lab and others have studied prior PE in preclinical models of hypertension, vascular remodeling, and vascular function. We previously demonstrated sFlt1-induced PE results in an exacerbated post-partum response to hypertensive stimuli including high salt or angiotensin II infusion²⁹ and to wire-induced vascular injury,³⁰ despite post-partum resolution of the PE phenotype. Vascular dysfunction has been noted in PE mouse models post-partum in microvessels^35,36 but not in large arteries like the carotid.³⁷ Six months post-partum, the plasma proteome of mice with prior sFlt1-induced PE showed increased proteins related to atherosclerosis signaling, atherosclerotic lesions, and cell movement of leukocytes via ingenuity pathway analysis.³⁸ These published data, combined with the current findings, support the concept that prior exposure to high sFlt1 results in heightened atherosclerotic plaque inflammation potentially via persistent alterations in expression of pro-inflammatory and pro-atherogenic molecules.

A role for sFlt1 in atherogenesis has been considered in a small number of studies with disparate results that require further mechanistic exploration. The Dallas Heart Study found sFlt1 levels correlate with subclinical aortic atherosclerosis, inflammation markers, and increased risk of CVD in humans.³⁹ Since sFlt1 sequesters circulating VEGF and PlGF, data has suggested sFlt1 may abrogate atherosclerosis by preventing atherogenic effects of PlGF, although sFlt1 injection into male mice without nephrectomy did not have an effect on atherosclerosis measurements.⁴⁰ When sFlt1 was administered for two months to male apolipoprotein E (ApoE) knockout mice with nephrectomy, this decreased atherosclerotic plaque size and macrophage infiltration. Conversely, a follow-up study in male sFlt1 and ApoE double knockout mice found increased plaque size and macrophage infiltration.⁴¹ In male rabbits with iliac artery balloon injury, local sFlt1 gene therapy delivery significantly reduced plaque area and circumference.⁴² In contrast to the current work, these prior studies used only male animals with different atherogenic backgrounds and found sFlt1 given during atherogenesis seems to reduce plaque size. It remains to be seen if sex differences, effects of pregnancy, animal model differences, or duration and timing of delivery of sFlt1 relative to hyperlipidemia induction may explain the differences in our study. This study remains the only one to explore sFlt1 during pregnancy in female mice followed by hyperlipidemia, thereby specifically modeling post preeclampsia atherosclerosis.

While the detailed mechanism of atherosclerosis plaque inflammation after PE is not known, the cytokine array suggests potential pathways. In the sFlt1-induced PE model, aortic tissue CCL22 expression was significantly increased. CCL22 is a chemoattractant for T cells and is increased in serum from patients with ischemic heart disease.⁴³ Both macrophages and T cells can secrete CCL22. The cellular source of CCL22 and its receptor CCR4 remains to be determined. Fetuin A was also significantly increased and is known to be produced by the liver, placenta, adipose tissue, vascular cells and monocytes. Fetuin A is atherogenic, highly expressed in human atherosclerotic lesions and co-localizes with macrophages and collagen fibers.⁴⁴ IGFBP6 co-localizes with macrophages and endothelial cells within human atherosclerotic plaques⁴⁵, was downregulated in histologically unstable plaques, plasma^45,46 and during acute myocardial infarction.⁴⁷ CCL21 is constitutively expressed in lymphatic endothelial cells and can mediate leukocyte trafficking to lymph nodes,⁴⁸ but the impact of significantly decreased IGFBP6 and CCL21 remains to be determined.

It is still unknown why prior PE results in a heightened risk for atherosclerosis-induced pathologies and the loss of pre-menopausal CVD protection. While women with prior PE have increased CVD risk factors post-partum, even after for adjusting for higher rates of diabetes mellitus, hypertension, hyperlipidemia and obesity, PE remains a significant risk factor for coronary atherosclerosis.⁵ Approximately 64% of increased coronary artery disease risk after PE has been attributed to hypertension⁴⁹, leaving a significant amount of risk unexplained. Female atherosclerotic plaques tend to be less inflammatory and exhibit more erosion.⁵⁰ Our findings reveal exposure to sFlt1-induced PE is sufficient to increase atherosclerotic plaque inflammation independent of pre-existing or post-partum risk factors.

Several study limitations should be acknowledged. While we confirmed blood pressure normalized post-partum, we were unable to measure sFlt1 in hyperlipidemic plasma at the end of the study due to lipids interfering with the sFlt1 ELISA. In multiple published studies, sFlt1 levels decline to normal post-partum in this model, resulting in no difference between sFlt1 values, systolic blood pressure, kidney damage, or cardiac function at 2 months post partum^29,30, the equivalent time to our atherosclerosis measurement. Also, this study uses one model of PE, so future studies are needed to investigate post-partum atherosclerosis using other mouse PE models. To eliminate the impact of litter size and lactation, pups were removed one day after birth. As lactation associates with decreased subclinical CVD in normotensive women but not in women with PE history⁵¹, we might expect a greater difference in atherosclerosis had the mice been allowed to lactate, providing additional cardioprotection to the control group. The necessity of high sFlt1 alone versus the combination of high sFlt1, high blood pressure and end organ damage to increase atherosclerotic plaque inflammation cannot be determined from this study. Additional studies are warranted to determine whether atherosclerotic inflammation would be modulated by sFlt1 exposure in males or non-pregnant females, situations relevant to those receiving anti-angiogenic VEGF trapping cancer therapies.⁵²

In conclusion, exposure to the sFlt-1 induced PE model in mice is sufficient to exacerbate atherosclerosis inflammation without altering traditional CVD risk factors. Future studies will investigate the molecular mechanism driving vascular inflammation after PE and explore therapeutic strategies to mitigate this. This information is important in the care for the growing population of young women at high risk for adverse cardiovascular events due to exposure to preeclampsia during pregnancy.

Highlights.

• Epidemiological studies indicate that women who have preeclampsia, a hypertensive complication of pregnancy, are at significantly greater risk of cardiovascular disease compared to women who had a normotensive pregnancy.

• Exposure to preeclampsia in an atheroprone mouse model significantly increased future atherosclerotic plaque inflammation without impacting plaque size or traditional cardiovascular disease risk factors.

• Since plaque inflammation associates with rupture and ischemic event in humans, enhanced atherosclerotic inflammation despite similar plaque size may explain the 2 fold increased risk for heart attack and stroke in women after a preeclamptic pregnancy.

• This study suggests that a component of the enhanced cardiovascular risk after preeclampsia may be a direct result of preeclampsia, independent of shared risk factors.

Non-Standard Abbreviations and Acronyms

CCL

CC motif chemokine ligand

CVD

cardiovascular disease

IGFBP6

insulin like growth factor binding protein 6

LDLR-KO

low density lipoprotein receptor knockout

PE

preeclampsia

sFlt1

soluble fms-like tyrosine kinase 1

VEGF

vascular endothelial growth factor