Preeclampsia is associated with long-term adverse maternal health, such as cardiovascular and metabolic diseases. The objective of this study was to determine whether preeclampsia in a well-characterized animal model that was induced by overexpression of soluble fms-like tyrosine kinase-1 (sFlt1) results in alterations in the maternal circulating proteome that persist long after delivery.
CD-1 mice at day 8 of gestation were injected with adenovirus that carried sFlt1 or the murine immunoglobulin G2α Fc fragment as control. Depleted maternal plasma was analyzed 6 months after delivery by label-free liquid chromatography–mass spectrometry assay. The tandem mass spectrometry data were searched against a mouse database, and the resultant intensity data were used to compare abundance of proteins across disease/control plasma pool. Results were analyzed with ingenuity pathways analysis. Right-tailed Fisher exact test was used to calculate a probability value.
Of 150 proteins that are common for both groups, ingenuity pathways analysis determined 105 proteins that were ready for analysis. Diseases and disorders analysis showed significant enrichment of proteins that are associated with cardiovascular disease. Within this cluster, the most abundant proteins were associated with vascular disease, atherosclerosis, and atherosclerotic lesions. Other top disease clusters were inflammatory response, organismal injury and abnormalities, and hematologic and metabolic disease.
Exposure to sFlt1-induced preeclampsia alters multiple biologic functions in mothers that persist later in life. Our results suggest that some of the long-term adverse outcomes that are associated with preeclampsia actually may be a consequence rather than a mere unmasking of an underlying predisposition. If similar results are found in humans, the development of preventive strategies for preeclampsia should also improve long-term maternal health.
The increasing rates mortality and morbidity from cardiovascular diseases (CVD) in women warrants intensifying research into gender-specific mechanisms. The past decade saw an increase in studies that reported on the association between preeclampsia and later CVDs. Indeed, multiple studies and metaanalyses confirm that women, whose pregnancy is complicated by preeclampsia, have higher susceptibility to CVD later in life.
The pathogenesis and genetics of preeclampsia still must be elucidated. However, key pathophysiologic factors (eg, endothelial dysfunction, oxidative stress, inflammatory responses) are common to both preeclampsia and CVD. Moreover, both conditions share similar risk factors (ie, obesity, diabetes mellitus, and dyslipidemia).
Whether preeclampsia directly influences the development of maternal CVD later in life or preeclampsia uncovers a preexisting condition that would have led to CVD later in life anyway remains undetermined.
To dissect the associations between preeclampsia and CVD, studies in animal models are needed. In our laboratory, we have established a mouse model of preeclampsia that is induced by overexpression of soluble fms-like tyrosine kinase-1 (sFlt1), which has been described elsewhere . Briefly, pregnant mice that are injected with sFlt1-carrying adenovirus on day 8 of gestation experience hypertension, endothelial dysfunction, and other features that are characteristic of preeclampsia. We have also examined these mice 6 months after delivery and found that exposure to sFlt1 during pregnancy does not lead to increased blood pressure or vascular dysfunction later in life. These results favor the hypothesis that increased CVD in women after preeclampsia is the result of preexisting conditions that are common to both preeclampsia and CVD. At 6 months after delivery, however, mice are still young adults and may not show direct evidence of CVD. Subtle changes that would lead to hypertension later than 6 months after delivery may still have occurred.
To further investigate consequences of preeclampsia on long-term maternal health, our objective in the current study was to determine whether preeclampsia that is induced by the overexpression of sFlt1 in a well-characterized animal model results in alterations of the maternal circulating proteome that persist long after delivery.
Materials and Methods
The study protocol was approved by the Institutional Animal Care and Use Committee at the University of Texas Medical Branch, Galveston, Texas. The animals were housed separately in temperature- and humidity-controlled quarters with constant 12:12–hour light-dark cycles and were provided with food and water ad libitum. Female and male CD-1 mice were obtained from Charles Rivers Laboratories (Wilmington, MA) and bred in our facility. The animals were killed by CO 2 inhalation according to the Animal Care and Use Committee and the American Veterinary Medical Association guidelines.
At day 8 of gestation, pregnant mice were divided randomly into 2 groups and injected in the tail vein with adenovirus-carrying sFlt1 (10 9 plaque-forming units in 100 μL; n = 6) or adenovirus that carried the murine immunoglobulin G2α Fc fragment as the adenovirus control (10 9 plaque-forming units in 100 μL; n = 6). The procedure to produce adenovirus-carrying sFlt1 and the adenovirus control has been described elsewhere. No other procedures were performed during pregnancy (blood/urine collection) not to intervene with gestation and delivery. Pregnant mice were allowed to deliver. Pups were weaned from their respective mothers 3 weeks after delivery.
Food/drink intake, animal care, and other activities were not regulated throughout the study. Experiments were performed in mothers at 6 months after delivery. We started with 6 mice per group; however, 1 animal that was exposed to sFlt1 died before 6 months after delivery. Mice were anesthetized with a mixture of ketamine (Ketalar; Parke-Davis, Morris Plains, NJ) and xylazine (Gemini; Rugby, Rockville Center, NY); telemetric blood pressure transducers (PA-C10 model; Data Sciences International, St. Paul, MN) were implanted. Arterial blood pressure was monitored in the unrestrained conscious mice as described previously. After a rest period of 3 days, blood pressure was recorded continuously for 4 consecutive days with the Dataquest A.R.T. data acquisition system (Data Sciences International). Then, the mice were killed, and blood was collected by heart puncture at the time of death in microtubes that contained ethylenediaminetetraacetic acid and was spun down at 4°C for 20 minutes at 10,000 rpm. Plasma was removed and stored at −80°C until analysis. The carotid arteries were dissected for in vitro vascular reactivity studies. Two-millimeter segments of carotid arteries were mounted in a wire myograph (Model 410A; J.P. Trading I/S, Aarhus, Denmark) with 25-μm tungsten wires. The preparations were bathed in physiologic salt solution that was maintained at 37°C, pH approximately 7.4. A mixture of 95% O 2 and 5% CO 2 were bubbled continuously through the solution. The force was recorded continuously by an isometric force transducer and analyzed with the PowerLab system and Chart 5 data acquisition and playback software (AD Instruments, Castle Hill, Australia). After stabilization of the tone, the vessels were contracted twice with 60 mmol/L KCl for 10 minutes to enhance reproducibility of responses. Vascular reactivity to vasodilator acetylcholine (10 −9 to 10 −5 mol/L) after vessels were precontracted with phenylephrine (10 −6 × 10 −7 mol/L) was assessed.
The sFlt1 level in the blood was measured with mouse soluble vascular endothelial growth factor R1 immunoassay (R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions.
Plasma preparation for mass spectrometry
Plasma was analyzed for each mouse separately. Whole plasma (10 μL) was depleted with Seppro Mouse Spin columns (Sigma-Aldrich, St. Louis, MO) according to the manufacturer’s instructions. Protein concentration was detected by Bradford assay (Bio-Rad, Hercules, CA). Plasma was denatured and reduced by 6 mol/L urea with 20 mmol/L dithiothreitol in 150 mmol/L Tris buffer (pH 8.2) with subsequent alkylation by iodoacetamide (40 mmol/L). Samples were diluted with Tris buffer (50 mmol/L; pH 8.2), and trypsin (1 μg/μL) was added at a 20:1 substrate:enzyme ratio. Digestion was carried out for 16 hours at 37°C and stopped by acidification. Samples were desalted with C18 columns (Waters, Milford, MA) according to the manufacturer’s instructions and lyophilized.
After reconstitution in 2% (volume/volume) acetonitrile, 0.1% (volume/volume) formic acid samples were analyzed on a LTQ Orbitrap XL (Thermo-Fisher Scientific, Bremen, Germany) that was interfaced with an Eksigent nano-LC 2D plus ChipLC system (Eksigent Technologies, Dublin, CA). Approximately 0.5 μg of sample was loaded onto a ChromXP C 18 -CL trap column (200 μm internal diameter × 0.5 mm length; 3 μm particle size; Eksigent Technologies) at a flow rate of 3 μL/min. Reversed-phase C 18 chromatographic separation of peptides was carried on a ChromXP C 18 -CL column (75 μm internal diameter × 10 cm length; 3 μm) at 300 nL/min with the column temperature controlled at 60°C. Solvent A with 0.1 % formic acid in water and solvent B with 0.1% formic acid in acetonitrile were used for high-performance liquid chromatography gradient. Gradient conditions were 3-8% B for 5 minutes; 8-33% B for 120 minutes; 33-90% B for 10 minutes; 90% B held for 10 minutes; 90-3% B for 5 minutes. The total run time was 150 minutes. The LTQ Orbitrap was operated in the data-dependent mode to measure simultaneously full scan mass spectrometry spectra in the Orbitrap and the 5 most intense ions in the LTQ Orbitrap by collision-induced dissociation, respectively. In each cycle, spectra 1 was acquired at target value of 1E6 with resolution of 100,000 (mass to charge ratio, 400) followed by top 5 spectra 2 scan at target value 3E4. The following mass spectrometric setting was used: spray voltage was 1.6 kV, and charge state screening and rejection of singly charged ion were enabled; ion selection thresholds were 8000 for spectra 2; collision energy was normalized at 35%; activation Q was 0.25, and dynamic exclusion was used for 30 seconds. Each sample was analyzed in triplicate.
Data analysis was performed with MaxQuant software, supported by Mascot as a database search engine for peptide identification. Average label-free quantification intensity values were used to calculate sFlt1/mFc protein ratio.
Ingenuity pathways analysis
Data were expressed as spectra intensity ratio sFlt1 group over mFc group (sFlt1/mFc). Molecules with ratio outside the range of 0.8–1.2 were included in the final analysis. We used ingenuity pathways analysis (IPA; Ingenuity Systems Inc, Redwood City, CA) to determine whether any peptides can be mapped to different biologic or disease functions. For the final analysis, we used the IPA content version 14197757 that was released on Aug. 11, 2012. The dataset was filtered for species (mouse) and confidence (experimentally observed) and included molecules with direct and indirect relationships. The ratio between the 2 groups was converted to fold increase/decrease by IPA. Our objective was to identify the biologic functions that were affected by exposure to sFlt1 during pregnancy.
The functional analysis identified the biologic functions and/or diseases in Ingenuity’s knowledge base that were specific to our dataset. IPA has 3 primary categories of functions: molecular and cellular function; physiological system development; and diseases and disorders. There are 85 high-level functional categories that are classified under the primary categories. Lower level functions and specific functions are classified within the 85 high-level functions, and each may be under multiple high-level categories. In other words, each function is not discrete. However, IPA ranks associated functions based on their significant differences and gives an overall assessment of primary function. If a higher order disease/disorder category contains ≥2 specific functions that reach statistical significance, the software displays the most statistically significant value on the y-axis of a bar graph. Right-tailed Fisher exact test was used to calculate a probability value that determined the probability that each biologic function and/or disease that was assigned to that dataset is due to chance alone. A probability value of ≤ .05 was considered statistically significant.
Using IPA, we have also generated networks from our dataset. A dataset that contained gene identifiers and the corresponding ratio values was uploaded into the application. Each identifier was mapped to its corresponding object in Ingenuity’s knowledge base. With the use of a range outside 0.8 and 1.2 for the ratio, molecules with significantly differentially regulated expression were identified. These molecules, called network eligible molecules, were overlaid onto a global molecular network that was developed from the information contained in Ingenuity’s knowledge base. Networks of network eligible molecules were then generated algorithmically based on their connectivity.
Canonical pathways analysis identified the pathways from the IPA library of canonical pathways that were most significant to our dataset. The significance of the association between the dataset and the canonical pathway was measured in 2 ways: (1) A ratio of the number of molecules from the dataset that map to the pathway divided by the total number of molecules that map to the canonical pathway is displayed. (2) Fisher exact test was used to calculate a probability value that determined the probability that the association between the genes in the dataset and the canonical pathway is explained by chance alone.
The probability value that is associated with functional analysis for a dataset is a measure of the likelihood that association between a set of functional analysis molecules in our experiments and a given process or pathway is due to random chance. The smaller the probability value, the less likely that the association is random, which means the more significant the association. Probability values were very small in our study; thus, IPA converted it into a more illustrative number -log( P value).
A network is a graphic representation of the molecular relationships between molecules. Molecules are represented as nodes, and the biologic relationship between 2 nodes is represented as an edge. All edges are supported by at least 1 reference from the literature, from a textbook, or from canonical information stored in the Ingenuity Pathways knowledge base. Human, mouse, and rat orthologs of a gene are stored as separate objects in the Ingenuity Pathways knowledge base but are represented as a single node in the network. The intensity of the node color indicates the degree of up- or down-regulation. Nodes are displayed using various shapes that represent the functional class of the gene product.
Data analysis for blood pressure, vascular reactivity, and blood sFlt1 levels
Blood pressure and vascular reactivity data passed Shapiro-Wilk normality and Equal Variances tests. The blood pressure data that were obtained from telemetry system were plotted as mean values over 24-hour period, expressed as mean ± SEM, and analyzed with a t -test ( P < .05). In the vascular reactivity studies, phenylephrine precontraction was used to obtain the percentage of relaxation that was induced by the acetylcholine. Results were expressed as mean ± SEM. The areas under the concentration response curve were compared with the use of a t -test ( P < .05).
The sFlt1 levels were calculated with a standard curve that was derived from known concentrations of the recombinant protein. Data passed Shapiro-Wilk normality test, but not the Equal Variance test; therefore comparisons between sFlt1 and mFc groups were made with the Mann-Whitney Rank Sum Test ( P < .05). When nonpregnant age-matched female mice were added to the analysis, Kruskal-Wallis one-way analysis of variance on ranks ( P < .05) was used.
There was no difference in maternal weight between the sFlt1 and mFc groups on day 1 of gestation, at delivery, and at 6 months after delivery (data not shown). Adiposity, which is a percent of adipose tissue from total body weight, was also not different between the groups at 6 months after delivery (mFc group, 8.16% ± 1.9%, vs sFlt1 group, 7.3% ± 1.9%; P = .8). There was no statistical difference in the levels of sFlt1 between the 2 groups of mice ( Table 1 ). The levels of sFlt 1 were similar to age-matched nonpregnant mice (1.5 ± 0.1 ng/mL). Blood pressure and carotid artery responses to acetylcholine that reflected endothelium function were not different between the 2 groups ( Table 1 ). Although SEM for blood pressure results was large, the results that are presented in Table 1 had passed the Equal Variance test with a probability value of .27 for systolic blood pressure and .56 for diastolic blood pressure. The power of the performed test was below the desired power of 0.800. Less than desired power indicates that we are less likely to detect a difference when one actually exists. Thus, negative results should be interpreted cautiously. However, we have observed similar results in a different set of animals in our previous study.
|Parameter||mFc group (n = 6)||sFlt1 group (n = 5)||P value|
|sFlt1, ng/mL||1.03 ± 0.22||1.8 ± 0.6||.4|
|Blood pressure, mm Hg|
|Systolic||129.6 ± 3.7||138.3 ± 24.7||.7|
|Diastolic||110.9 ± 7.3||115.5 ± 23.83||.9|
|Acetylcholine area under the curve, arbitrary units||330.0 ± 22.0||343.5 ± 28.4||.7|
We identified a total of 150 peptides that are expressed in the plasma of mice that previously had been exposed to sFlt1-induced preeclampsia and controls. Of these, IPA determined 105 molecules that were ready for analysis. These peptides were spread among 75 high-level functional categories (of 85 available). The top 5 high-level functional categories that are associated with exposure to preeclampsia are CVD, protein synthesis, hematological system development and function, organismal functions, and tissue morphology ( Figure 1 ).
The top biologic function category was CVD. Of 56 diseases/disorders in this cluster that was found in our dataset ( Figure 2 ), the top 5 were vascular disease (22 molecules), atherosclerosis (14 molecules), atherosclerotic lesion (12 molecules), size of atherosclerotic lesion (9 molecules), and area of atherosclerotic lesion (7 molecules).
For the protein synthesis category, 7 biologic functions were identified. Of these, the most significant ( Figure 3 ) were quantities of protein lipid complex (11 molecules), high-density lipoprotein (HDL) cholesterol (10 molecules), low-density lipoprotein (LDL) cholesterol (7 molecules), protein in blood (17 molecules), and lipoprotein metabolism (4 molecules).
Changes in 53 biologic functions that belonged to the hematological system development and function category were identified. The top 5 in this category ( Figure 4 ) were blood coagulation (12 molecules), quantity of phagocytes (15 molecules), cell movement of leukocytes (17 molecules), quantity of macrophages (9 molecules), and blood platelets aggregation (7 molecules).
The fourth most significant category was organismal functions ( Figure 5 ) with changes in blood coagulation (12 molecules) and flow (4 molecules), wound healing (6 molecules), and infarct healing (1 molecule).