The response of the innate immune and cardiovascular systems to LPS in pregnant and nonpregnant mice^†

Abstract

Sepsis is the leading cause of direct maternal mortality, but there are no data directly comparing the response to sepsis in pregnant and nonpregnant (NP) individuals. This study uses a mouse model of sepsis to test the hypothesis that the cardiovascular response to sepsis is more marked during pregnancy. Female CD1 mice had radiotelemetry probes implanted and were time mated. NP and day 16 pregnant CD-1 mice received intraperitoneal lipopolysaccharide (LPS; 10 μg, serotype 0111: B4). In a separate study, tissue and serum (for RNA, protein and flow cytometry studies), aorta and uterine vessels (for wire myography) were collected after LPS or vehicle control administration. Administration of LPS resulted in a greater fall in blood pressure in pregnant mice compared to NP mice. This occurred with similar changes in the circulating levels of cytokines, vasoactive factors, and circulating leukocytes, but with a greater monocyte and lesser neutrophil margination in the lungs of pregnant mice. Baseline markers of cardiac dysfunction and apoptosis as well as cytokine expression were higher in pregnant mice, but the response to LPS was similar in both groups as was the ex vivo assessment of vascular function. In pregnant mice, nonfatal sepsis is associated with a more marked hypotensive response but not a greater immune response. We conclude that endotoxemia induces a more marked hypotensive response in pregnant compared to NP mice. These changes were not associated with a more marked systemic inflammatory response in pregnant mice, although monocyte lung margination was greater. The more marked hypotensive response to LPS may explain the greater vulnerability to some infections exhibited by pregnant women.

Introduction

In the last CMACE (Centre for Maternal and Child Enquiries; now called MBRRACE-UK [Mothers and Babies: Reducing Risk through Audits and Confidential Enquiries]) report, genital tract sepsis was the leading cause of direct maternal mortality [1]. In addition, pregnant women have a three times greater risk of dying from flu and malaria than nonpregnant (NP) women [2], suggesting that pregnant women are particularly vulnerable to some types of infection. This may be related to various pregnancy-induced adaptations: (i) suppression of the acquired immune system [3]; (ii) excessive activation of the innate immune system, which is thought to be responsible for preterm labor (PTL) [4] and some cases of cerebral palsy [5]; and (iii) higher levels of nitric oxide may exacerbate sepsis-induced hypotension and enhance the excessive innate immune response [6, 7]. Although septic shock is a major cause of mortality in intensive care units worldwide, remarkably little is known about the impact of sepsis on cardiovascular function during pregnancy [8], especially the adaptation and regulation of the peripheral vascular system.

Administration of the bacterial wall component, lipopolysaccharide (LPS), to rodents causes marked hypotension. Yamashita et al. showed that 80 mg/kg intraperitoneal (i.p.) LPS caused a greater than 90% mortality in WT (C57bl/6) male mice and triggered a severe and progressive drop in mean arterial pressure (MAP) of about 60 mm Hg, a 40% reduction from pretreatment values (mean 97 mm Hg) [9]. The effect is dose dependent and associated with increased expression of markers of inflammation [10–12]. Similar effects have also been seen in humans, where LPS induces a hypotensive response, associated with an increase in circulating cytokines [13], although females are more resistant, with a lesser increase in temperature, fall in blood pressure, and in vitro release of cytokines [14–16]. Few studies have investigated differences in the peripheral blood mononuclear cells responses in pregnant and NP individuals and found that the cytokine (interleukin (IL)–6, IL-10, and IL-17) response was increased in the cells from pregnant women exposed to a variety of agents (umbilical cord blood (UCB) of the mother's own child, third-party UCB, phytohemagglutinin, and anti-CD3 antibody) [17]. Similarly, monocyte-derived dendritic cells from pregnant women exposed to cytokines (IL-1β and tumour necrosis factor alpha (TNFα)) showed a marked increase in IL-10, and a similar trend in response to LPS [17]. The cytokine response per se and the balance (proinflammatory, IL-1β and TNFα, vs. anti-inflammatory, IL-4, IL-10) are critical as not only are they thought to play a key role in the adverse response to infection leading to septic shock [18], but, in pregnancy, they may mediate the adverse effects of maternal inflammation including PTL, preeclampsia, and cerebral palsy [4, 5, 19].

The administration of LPS has also been widely used in pregnant mice to model PTL [20]. Intraperitoneal administration of LPS provokes preterm delivery in pregnant females, often with marked maternal morbidity and even mortality [20–23]. However, only one paper has compared the response of pregnant and NP mice to LPS; this paper found that in late pregnant compared to NP mice, there was a greater TNFα and reduced IL-10 response to LPS [24]. The effects of LPS-induced inflammation on the maternal cardiovascular system are not well understood and no comparisons have been made between pregnant and NP individuals or animals. In this paper, we tested the hypotheses that, during pregnancy, the cardiovascular derangement in response to endotoxemia is more marked due to a greater activation of the innate immune system.

Methods

Female CD1 mice (Charles River, UK) aged 8–12 weeks were used for all experiments. All animal procedures were carried out in accordance with the Home Office Animals (Scientific Procedures) Act 1986 (PPL 70/7372), and were previously approved by the Animal Welfare and Ethical Review Board (AWERB) at Imperial College London. All mice were housed in individually ventilated cages with access to normal rodent chow and water ad libitum, and maintained on a 12:12 light-dark cycle.

Time mating

Female mice were placed overnight into cages with male studs. The following morning, the female mice were inspected for a copulatory plug. The day the plug was detected was designated as gestational day 0 (E0). From previous work in our group, we defined E16 to be equivalent to 34 weeks gestation of a human pregnancy. All experiments in pregnant animals were conducted on day 16 of pregnancy (E16). This period was chosen as most cases of severe sepsis occur during the third trimester of human pregnancy [25].

Mouse blood pressure telemetry

Telemetry probes (PA-C10, Data Sciences International) were surgically implanted. The technique allows continuous measurements of blood pressure, heart rate (HR), and activity to be made. Mice were anaesthetized using isoflurane (5% for induction and 2% maintenance of isoflurane, 2 L/min O₂ flow rate), and surgically cleaned and draped. Preoperatively, mice received 5% enrofloxacin and 0.3 mg/mL buprenorphine. A small ventral incision on the neck was made, and the left common carotid artery was isolated prior to cannulation. The catheter was advanced until placed in the aortic arch and secured in place with sutures [26]. The transmitter body was then advanced subcutaneously to lie over the right flank before the incision was closed. Mice were then placed into a heated recovery chamber (30°C). Buprenorphine was given the day after surgery, and animals were allowed to recover for a minimum of 1 week prior to mating or any experiments.

All experiments included light and dark periods to account for diurnal variations. Following recovery, continuous 24-h recording (data collection) began using the Dataquest ART Acquisition System (Data Sciences International, v4.1) for at least 48 h to collect data in NP animals (NP baseline). These data allowed animals to act as their own controls. Radiotelemeters were magnetically activated and deactivated. While recording, animals were kept in their standard cages with access to food and water ad libitum. These cages were placed in a separate room to ensure a calm environment for recording.

For this study, data were collected 1 week after insertion to give baseline data in all animals and then for 24 h after PBS/LPS administration from NP and pregnant mice at E16 and compared. The data collected included MAP, systolic arterial pressure (SAP), diastolic arterial pressure (DAP), HR, and activity. Hemodynamic data were collected for 10 s every 30 s. Labor time (latency to labor after i.p. injection of LPS or control) and litter sizes were also recorded. Accurate labor time was enabled by the use of infrared IP cameras in conjunction with the AVerDiGi (NV3000 Series) HYBRID Surveillance Platform (RF Concepts, Dundonald, UK).

Lipopolysaccharide administration

LPS endotoxin (serotype 0111: B4, Sigma-Aldrich) and sterile vehicle control (PBS) were administered by the i.p. route. All pregnant mice were anaesthetized under isoflurane for i.p. injection to minimize potential harm to fetuses and carried out between 8 am and 10 am.

Time points

To avoid the effects of handling animals during telemetry recordings, serum and other tissues were taken from separate groups of animals at two time points: 6 and 12 h after i.p. LPS injection or vehicle control. Mice were terminally anaesthetized using 5% isoflurane (2L/min O₂ flow rate), and a cardiac puncture was performed to collect plasma or serum samples. Subsequent to schedule 1 culling, lung, left ventricle of the heart, myometrium, and placenta were collected and snap-frozen on dry ice. These samples were then stored at –80°C until RNA and protein extraction.

Multiplex assay

Mouse cytokine/chemokine and angiogenesis/growth factor magnetic bead panel (Merck Millipore) were used according to the manufacturer's protocol. Analytes were quantified using a Luminex MAGPIX instrument with xPonent 4.2 software (Luminex Corp).

RNA extraction and quantitative RT-PCR

RNeasy mini kit by Qiagen Ltd (Crawley) was used to extract RNA from the whole heart and from a second series of animals from the left ventricle only following the manufacturer's instruction. Less than 20 mg of tissue was used. To convert the extracted RNA to cDNA, 6 μg of the RNA was reverse transcribed using oligo dT random primers using MuLV transcriptase (Applied Biosystems Ltd). A unique set of primers specific to mouse were designed and obtained from Life Technologies Ltd (Paisley, UK, see Supplementary Table S1). These were validated prior to their use. Quantitative real-time PCR was done using SYBR Green (Roche Diagnostics Ltd) and amplicon yield was monitored by Rotor Gene R-G 3000 (Corbett research Ltd). The results of the PCR analysis were expressed relative to the constitutively expressed ribosomal protein s18.

Protein extraction and western blotting

Western blotting was used to quantify protein changes taking place. Tissue samples were homogenized in lysis buffer using a homogenizer (Tissue Tearor, BioSpec Products Inc.). The supernatant was centrifuged at 3000 × g for 3 min at 4°C to remove any remaining cell debris and stored at –80°C. Protein concentrations were obtained using a standard protein assay protocol. The protein supernatant used at a dilution of 1:3 or 1:5 was sufficient for this process. Protein samples were denatured at 95°C, separated by electrophoresis and transferred onto a polyvinylidene difluoride membrane (Bio-Rad Laboratories). The membranes were incubated in 5% milk solution for 1 h at room temperature, washed with tris-buffered saline and tween 20 (TBS-T), and incubated with specific primary antibodies overnight at 4°C. After washing, the membranes were incubated with secondary antibody for 1 h at room temperature. Western Blotting Luminol Reagent (Santa Cruz Biotechnology) was used to expose protein bands, which were expressed as a ratio to glyceraldehyde 3-phosphate dehydrogenase (GAPDH).

Wire myography

Aorta and uterine vessels were used to assess function and sensitivity to vasoactive substances. Samples were taken from NP and pregnant E16 mice. Samples were collected 12 h after the administration of i.p. LPS or vehicle (PBS). In brief, vessels were mounted onto the vessel myograph (DMT, Aarchus) and tested for viability [27]. Cumulative concentration response curves were performed for phenylephrine (PE; from 10⁻⁶ M to 10⁻² M), acetylcholine (ACh; from 10⁻⁶ M to 10⁻² M), and sodium nitroprusside (SNP; from 10⁻⁶ M to 10⁻³ M).

Flow cytometry

Fluorescence-activated cell sorting technique was used to further investigate the innate immune response following LPS to assess leukocyte cell densities in blood, lungs, and liver. Tissues were weighed and single-cell suspensions were prepared in 1 mL of Intracellular (IC) Fixation Buffer (eBioscience) using a gentleMACs dissociator (Miltenyi Biotec Ltd) and M tubes as described previously [28]. Cell suspension and whole blood were incubated with the following fluorophore-conjugated rat anti-mouse mAbs antibodies used: Ly6C (HK1.4) (BioLegend), Gr1 (RB6-8C5) (BioLegend), NK1.1 (PK136) (BioLegend), F4/80 (CI: A3-1) (BioLegend), CD45 (30-F11) (BioLegend) CD11b (M1/70), MHCII (AMS-32.1) (BioLegend), CD11b (M1/70) (eBioscience) (Supplementary Table S2). Cell counts were determined using 20 μL AccuCheck counting beads (Invitrogen, UK) per sample. Samples were run using the BD LSR II flow cytometer (BD biosciences). Data were analyzed using FlowJo software (Tree Star, USA). Monocyte subsets in blood were identified as CD11b+, F4/80+, and Ly6C^lo or Ly6C^hi [29]. In lung tissue, Ly6C^lo subset monocytes were distinguished from interstitial macrophages by low expression of MHC II (Supplementary Figure S1) [30]. Analysis of Ly6C^lo monocytes in liver was not performed due to overlap of their surface markers used here with those expressed on Kupffer cells.

Statistical analysis

Data collected in the 24-h period after LPS injections were shown as hourly mean values ± SEM as a change (Δ) from baseline alongside area under the curve analysis, maximum increase, and maximum decrease (where appropriate). Normally distributed data were analyzed using t test or ANOVA with correction for multiple comparisons as indicated. Where the data were not normally distributed as determined by Kolmogorov-Smirnov tests, nonparametric equivalent tests such as Kruskal-Wallis followed by the Dunn post hoc test, or Mann-Whitney U were employed. Differences in the cardiovascular parameters at different time points throughout pregnancy were analyzed by repeated measures ANOVA followed by Bonferroni correction for multiple comparisons between the means. Two-way ANOVA followed by Bonferroni correction was used to compare the response to LPS in pregnant and NP animals when normalized to vehicle controls.

The changes in blood pressure, HR, and activity over time were analyzed using mixed models for longitudinal data. To capture change over time, a quadratic trend was fitted. In some of the outcome measures, it was necessary to adjust for before–after treatment to get a better fit. SPSS 22 and Stata 13 were used to obtain the estimated effects.

Results

Maternal intraperitoneal lipopolysaccharide administration

We used an LPS-based murine model of endotoxemia to compare the response to inflammation in pregnant and NP mice. Initially, we performed a dose–response curve to i.p. administration of LPS to determine an optimal dose of LPS, the latency to labor, and pup survival after injection on E16 as end points (Supplementary Table S3). Previously, we and others have used intrauterine LPS to reliably induce preterm labor [31], but the hemodynamic impact of LPS administration in pregnancy has not been assessed. Time to labor and pup survival were assessed in a vehicle (PBS) group and demonstrated no difference to untreated control data in our group (mean time to delivery in PBS-treated group 57.2 ± 6.6 compared with untreated group 55.6 ± 0.8). LPS administration demonstrated a dose-dependent reduction in time to delivery (Supplementary Table S3), but at high doses of 20 μg and above LPS caused maternal deaths (20, 25, 50, or 100 μg LPS; at least one maternal death per group, numbers shown in Supplementary Table S3). The dose of 10 μg caused no maternal deaths, was most reproducible preterm delivery, and had least variation in labor time, pup survival, and caused no maternal mortality (mean time to delivery 23 ± 2.07, Supplementary Table S3). Subsequent experiments were therefore conducted using this dose.

Hemodynamic responses to lipopolysaccharide

To investigate the changes in maternal hemodynamic parameters after LPS administration, telemeteric recordings were taken from NP and pregnant mice at E16 following administration of LPS and PBS for 24 h and compared. In all outcome measures, an interaction between time (hours) and group was found. This means that the trajectory of the outcome measure varies across groups. It also means that the comparison between groups will depend on time. Curves have been fitted to the data (Figure 1A–E) and the raw data presented in the supplementary section (Supplementary Figure S2A–E).

Overall, the response of MAP to LPS was statistically different between NP and pregnant mice (Figure 1A, P = 0.031; Supplementary Figure S2A), with MAP readings falling in the pregnant mice while being maintained or even rising in the NP mice (Figure 1A, Supplementary Figure S2A). A similar, but nonsignificant response was observed in both SAP and DAP (Figure 1C and D, P = 0.48 and 0.1 respectively and Supplementary Figure S2C and D). The HR tended to decline in pregnant mice and rise in the NP (Figure 1B, P < 0.001 and Supplementary Figure S2B). The time spent inactive was similar in both groups, although overall it tended to increase in NP and reduce in pregnant mice (Figure 1E, P = 0.35, Supplementary Figure S2E). At specific time points, differences in the response between NP and pregnant animals were found (Supplementary Table S4).

Is the response of the innate immune system different in pregnant mice?

Circulating cytokines and chemokines

Absolute concentrations of cytokines/chemokines in maternal serum from pregnant and NP mice after PBS and LPS at 6 and 12 h were measured to evaluate the systemic inflammatory response. Treatment with LPS increased cytokine/chemokine levels both in NP and pregnant animals (Figure 2). Only IL-4 levels were differentially increased in pregnant mice and only at 12 h (P < 0.05, Figure 2I), the response of all other cytokines to LPS was similar between pregnant and NP mice (Figure 2).

Leukocyte densities

Blood: Baseline levels of total monocytes and of Ly6C^lo monocytes were greater in pregnant as compared to NP mice (P < 0.05–0.01, Figure 3A–E). Neutrophils and NK cells numbers were similar in both groups (Figure 3A and B). After LPS, neutrophils increased similarly in pregnant and NP mice (Figure 3A), while the levels of NK cells, total monocytes, and both monocyte subsets declined to a similar level (Figure 3B–E), although significant baseline differences were lost in total and Ly6C^lo monocytes implying the decrease may have been more marked in pregnant animals (Figure 3C and D).

Lungs: Baseline levels of neutrophils and total monocytes were similar (Figure 3F and H), but NK cells were higher in NP mice (P < 0.05, Figure 3G). After LPS, neutrophils declined similarly in both NP and pregnant mice (Figure 3F), the decline in NK cells was more marked in NP mice at 6 h so that levels were similar at this time point; this reversed by 12 h and NK cells were again higher in NP mice (P < 0.05, Figure 3G). In pregnant mice, the increase in total monocytes, at 6 h (P < 0.05, Figure 3H), and in Ly6C^lo monocytes, at 6 and 12 h (both P < 0.01, Figure 3I), were greater compared with NP mice. In NP mice, monocyte numbers generally remained stable or declined (P < 0.01, Figure 3H–J).

Liver: Baseline levels of neutrophils (Figure 3K) were similar to those of total monocytes (Figure 3M). NK cells were higher in NP mice at baseline and after LPS at both time points (all < 0.05, Figure 3L). After LPS, neutrophils increased in both pregnant and NP, but the increase was greater in NP mice at 6 h (P < 0.05, Figure 3K); total monocytes increased to a similar extent in both pregnant and NP mice (Figure 3M).

Is cardiac function differentially affected by lipopolysaccharideadministration?

Initially, we assessed the activation of the AP-1/NFκB systems in cardiac tissue to detect any differential activation. Levels were very similar in pregnant and NP tissue, with the exception of greater p38 and reduced PI3K phosphorylation at 12 h after LPS in pregnant mice (Supplementary Figure S3A–K). In addition, we assessed the phosphorylation of KDR (VEGFR2), as this indicates the activity of the VEGF system, but found no differences (Supplementary Figure S3L).

To assess the cardiac response to LPS, the expression of various markers of inflammation, apoptosis, and cardiac dysfunction were assessed in left ventricular tissue.

Inflammation

Although circulating cytokines were similar (Figure 2), we studied the expression of inflammatory cytokines in myocardial tissue to understand whether local inflammation was greater in the heart. Baseline IL-1β, IL-6, and cycloxygenase (COX)-2 mRNA expression was greater in pregnant mice (P < 0.05–0.01, Figure 4A–C); after LPS, TNFα at 12 h (P < 0.05, Figure 4D) and IL-10 at 6 h (P < 0.01, Figure 4E) mRNA expression was greater in NP mice.

Cardiac dysfunction

Atrial natriuretic peptide (ANP) mRNA expression was lower before and after LPS in pregnant mice in all cases (P < 0.05–0.001, Figure 4F). The expression of brain natriuretic peptide (BNP) mRNA declined in NP mice and was significantly reduced at 12 h after LPS vs. pregnant mice (P < 0.01, Figure 4G). The expression of c-tropinin-1 was greater in pregnant vs. NP mice before LPS and at 12 h after LPS (p < 0.05 and p < 0.05 respectively, Figure 4H). The expression of α-myosin heavy chain (α-MHC), β-MHC, sarco/endoplasmic reticulum Ca2⁺-ATPase (SERCA), and phospholamban (PLN) were similar in pregnant and NP mice (Figure 4I–K). The expression of inducible nitric oxide synthase (iNos) mRNA, which has been implicated in cardiac dysfunction in sepsis, was greater in pregnant mice at baseline, but similar thereafter (P < 0.001, Figure 4L). Expression of the homeobox gene NK2 Homeobox 5 (Nkx2-5) (involved in cardiac development and function) and of the α_1/3 isoforms of Na–K ATPase mRNA levels did not change after LPS administration (Supplementary Figure S4A–C).

Cardiac markers of apoptosis

The baseline expression of B-Cell lymphoma 2 (bcl-2) was reduced and of Bcl-2 associated death promoter (Bad) increased in pregnant compared to NP mice (both P < 0.05, Figure 4M and N), the expression of both were similar in pregnant and NP mice after LPS, although there were no significant differences in caspase-3 protein levels (Supplementary data Figure S5A–C). There were no differences in the expression of Fas Ligand (FasL) or pro-apoptotic marker Bcl-2 associated X protein (Bax) mRNA before or after LPS administration (Figure 4O&P).

Is vascular function differentially affected by lipopolysaccharide administration?

Circulating levels of vasoactive factors (Figure 5A–E)

The circulating levels of endothelin-1 were consistently lower in pregnant mice before (P < 0.01, Figure 5A) and after LPS administration, at 6 and 12 h (P < 0.01 and P < 0.05 respectively, Figure 5A). The increase in nitric oxide levels after LPS was greater in NP mice (P < 0.05 at 6 h, Figure 5B). Surprisingly, basal levels of placental growth factor (PlGF) were greater in NP mice (P < 0.05, Figure 5D); the levels of both PlGF and vascular endothelial growth factor (VEGF) increased after LPS in NP mice only and were significantly greater at both 6 and 12 h (P < 0.01, Figure 5D and E). We measured circulating cGMP levels and found no difference between NP and pregnant mice (Figure 5C). We were surprised that PlGF levels were similar in NP and pregnant mice and measured the levels through pregnancy; there appears to be a nadir at around day 16 (Supplementary data, Figure S6A and B).

Tissue levels of vasoactive factors (Figure 5F–K)

We measured the levels of vasoactive factors in left ventricular samples to assess whether tissue levels were different from circulating levels. At baseline, PlGF and cGMP were higher in pregnant cardiac tissue (both P < 0.05, Figure 5J and K). After LPS, at 12 h, the levels of endothelin-1, VEGF C, and VEGF D were all greater in NP cardiac tissue (all P < 0.05, Figure 5F, H and I), but the levels of LV cGMP were higher at 12 h in pregnant tissues (P < 0.01, Figure 5K).

In vitro analysis of vascular function

We examined the possibility that the blood vessels themselves might become more sensitive or resistant to vasoactive factors, studying vascular reactivity and vessel relaxation using wire myography in the aorta and uterine arteries.

Aorta data (Figure 6): In vehicle-treated groups, vessels from pregnant animals contracted less than those from NP animals in response to PE (c, P < 0.001, Figure 6A) and this was similar although less marked after LPS (P < 0.05, Figure 6B). Overall, the response to ACh was the same in vehicle-treated NP and pregnant animals (Figure 6C). After LPS, the vasodilator response was less in pregnant mice (b, P < 0.01; Figure 5D). The vessels from pregnant mice were less responsive to the NO donor, SNP, in both PBS- and LPS-treated groups (a, P < 0.05 and c, P < 0.001 respectively; Figure 6E and F).

Uterine artery (Figure 7): The uterine arteries from vehicle and LPS-treated pregnant mice contracted more in comparison to the NP animals (c, P < 0.001 and a, P < 0.05 respectively). In vehicle-treated mice, the pregnant uterine artery was significantly more sensitive to ACh (c, P < 0.001; Figure 7C), but this effect was reversed in LPS-treated animals (a, P < 0.05; Figure 7D). In vehicle-treated mice, the uterine artery response to SNP was similar (Figure 7E); but after LPS, uterine arteries from pregnant animals were less sensitive (c, P < 0.001; Figure 7F).

Discussion

In contrast to a significant literature relating to the cardiovascular and immune (dys)-regulation in septic NP experimental animals and humans, little is known about the impact of sepsis during pregnancy. This is all the more surprising given the significant body of literature that suggests that pregnancy is a risk factor for an adverse outcome to infection [32, 33]. To begin to address this issue, we used a mouse model of sepsis involving low-dose LPS to test the hypotheses that during pregnancy the cardiovascular derangement in response to endotoxemia is more marked as a consequence of a greater activation of the innate immune system. In support of our hypothesis, a more marked hypotensive response in pregnant mice and subtle alterations in monocyte lung margination were observed and no marked difference in the innate immune system response between pregnant and NP mice. These data question the role of the innate immune system in sepsis-induced hypotension.

Innate immune system: Several authors have suggested that the maternal innate immune system may be more active in pregnancy to compensate for a possible repression of adaptive immune system and that this overactivity may be responsible for pregnancy complications including PTL and preeclampsia [34]. We studied the response to LPS measuring the circulating cytokines and although there was a marked increase in the most cytokines/chemokines in both pregnant and NP mice, only the response of the anti-inflammatory cytokine, IL-4, was greater in pregnant mice. This is in contrast to the only previous study to compare the response to LPS in NP and pregnant mice, in which the increase in IL-6, IFNγ, and TNFα was greater and of IL-10 was reduced in pregnant mice; these changes were associated with a greater mortality in the pregnant mice. The authors concluded that the greater cytokine response to LPS might explain the increased severity of some diseases in pregnant women, including septic shock, and that they may play a role in the development of disseminated intravascular coagulation and preeclampsia [24]. The dose of LPS used was much greater than the dose we used, 1.6 mg, compared with 10 μg, but the serotype was not stated, meaning that direct comparison is not possible; however, it seems very likely that the strength of the inflammatory stimulus was much greater, as they experienced a 75% maternal mortality [24]. Our data suggest that the hypotensive component of septic shock may not be caused by a “cytokine storm” as previously thought.

Consistent with the lack of change in circulating cytokines, circulating leukocytes in pregnant mice showed no differences in the response to LPS, although at baseline, differences in some cell types were apparent as previously reported [35]. We measured leukocytes levels in lungs and liver as these can contain a substantial reservoir of organ-‘marginated’ cells that are a major part of the intravascular pool. Indeed, neutrophil and monocytes densities within circulating and marginated pools are very dynamic, often inversely related in size, modified by stress and systemic inflammation, and a key measurement-related sepsis pathogenesis and diagnosis. In the lung, the margination of total monocytes and of the Ly6C^lo monocyte subset was greater after LPS in pregnant mice, but other subsets tended to decline or remain static. In the case of the Ly6C^lo monocytes, numbers in the NP mice declined, while they remained static or slightly increased in pregnant mice, resulting in significantly higher levels in the pregnant mice. Neutrophils declined to a similar extent in both groups, while the NK cells were higher basally and after 12 h in the NP mice. In the liver, the LPS-induced increase in neutrophils was more marked in the NP group; as for other cell types/subsets, basal levels of NK cells were higher in NP mice and this difference was maintained after LPS while monocytes showed no significant differences. Lung inflammatory cell profiles have been described in several mouse models of sepsis, but none have studied the impact of LPS in pregnant animals. We considered these measurements might provide potentially novel insights into the vascular responses to sepsis during pregnancy. The total monocyte numbers were increased and may reflect a greater activation of this population. In terms of the more marked hypotensive effect of LPS in pregnant mice, if the greater monocyte margination observed in the lung was reproduced in the arteriolar bed, then the increase in local cytokines and vasoactive factors might influence local vessel tone, resulting in the greater hypotension observed in pregnant mice. Indeed, leukocyte arteriolar adhesion is increased in sepsis via endothelial activation [36] and sepsis-induced arteriolar dysfunction linked to greater NO production [37], supporting this potential mechanism and suggesting that local NO inhibition may reverse the arteriolar dysfunction. However, these possibilities await further study.

These data are important as they show that the response of the innate immune system to LPS is similar in pregnant and NP mice. This is particularly true for the neutrophil response, which has been implicated in septic shock. An excessive innate immune response has been suggested to play an important role in pregnancy complications, including PTL and preeclampsia [34], and while the current data suggest that there are differences in leukocyte dynamics, we have not identified a marked difference in systemic inflammatory response during pregnancy, which not only refutes the suggestion of a greater innate response to sepsis during pregnancy, but also questions the role of innate immune system in the genesis of septic shock.

Cardiovascular response to LPS: As we described in the introduction, LPS has been shown to cause a marked hypotensive response in rodents previously [10–12]. Consistent with these observations in pregnant mice, we observed a progressive decline in MAP, in contrast, in NP mice, the blood pressure rose. The hypertensive response to LPS in NP mice probably reflects the relatively low dose of LPS used in this study. The HR changes were marginal, but in a second model of sepsis, using cecal ligation and puncture, we found that the HR declined in both pregnant and NP mice in parallel to a marked decline in MAP [38]. These data suggest that the murine cardiovascular response to sepsis may be different from that observed in the human, where hypotension is usually associated with a tachycardia.

Cardiac function: Since blood pressure is a reflection of cardiac output and peripheral resistance, we next investigated the impact of LPS administration on the heart. We found no evidence of a differential activation of inflammatory transcription factors and although estradiol has been shown to inhibit the NFκB response in cardiac myocytes in vitro LPS [39], this would not explain the lack of NFκB activation in the NP mice. However, as our earliest time point was 6 h after LPS, any transcription factor may have been missed. After intrauterine LPS administration, we observed NFκB and MAPK/AP-1 activation in the myometrium at 3 h and this continued for NFκB at 7 h after LPS [28], this suggests that the failure to detect activation in the heart is unlikely to be due to timing, and is more likely to be due to a limitation of the method or variation in the timing of response between mice.

Cardiac cytokine mRNA expression increased in response to LPS and was similar in both groups, although the expression of IL-10 and TNFα was greater in NP mice. Indeed, sepsis-induced cardiac dysfunction has been suggested to be caused by the local release of cardiac cytokines [40]; however, we found that the increase in cardiac cytokine expression was similar in both groups of mice. Similarly, there was no evidence of greater myocardial dysfunction or apoptosis in response to LPS. Interestingly, basal IL-1β and IL-6 were greater in pregnant mice and this was associated with higher Troponin1 expression and a pro-apoptotic pattern, with lower Bcl-2 and higher Bad mRNA expression. ANP mRNA expression was consistently higher in the NP mice in all conditions, while the expression of BNP was higher in the pregnant mice at 12 h only. However, the expression of ANP and BNP did not increase in response to LPS, rather they tended to decline, implying that cardiac function was not impaired and consistent with the lack of change in other cardiac markers of dysfunction. Circulating BNP and troponin have been reported to markedly elevated in patients with sepsis and septic shock [41], and to relate to the prognosis. Circulating troponins have been related to cardiac dysfunction in sepsis [42], but, in the case of BNP, levels have been found to be elevated in presence [43] and absence of cardiac dysfunction [44], suggesting that other factors may influence BNP circulating levels. Intriguingly, the expression of both pro- and anti-apoptotic factors tended to decline in mice receiving LPS. This is counter to the existing literature, which is based on much higher doses (250 μg–3 mg) of variable types of LPS administered to male and unspecified sex rats [45–47]. Data from the mouse are variable, one paper reporting increased bax and reduced bcl-2 expression in cardiac tissue from NP mice after the injection of 30 μg of LPS [47]. A reduction in expression of Bcl-2 and Bad mRNA in response to LPS is unusual. However, the reduction in the expression of both apoptotic factors in mice receiving LPS may offer an explanation for the better outcome from sepsis observed in NP females as compared to previous studies [48].

Our data suggest that LPS does not compromise cardiac function in pregnant mice, but rather that pregnancy itself may drive both cardiac inflammation and apoptosis as demonstrated by the higher basal cardiac cytokine, troponin, and pro-apoptotic factor expression.

Vasoactive factors and vessel function: We explored whether the hypotensive response to LPS reflected changes in circulating vasoactive factors or an intrinsic difference in vascular function. On the other hand, in a mouse model of sepsis using cecal ligation and puncture, the endothelin-1 antagonist, bosentan, improved the outcome [49], suggesting that endothelin-1 may play an active role in the outcome of sepsis. Endothelin-1 levels were consistently higher in NP mice, but since they did not change in response to LPS they are unlikely to be responsible for the LPS-induced hypotension observed in the pregnant animals. We had anticipated the hypotension associated with LPS administration may be caused by a greater release of NO, as a critical role for NO has been demonstrated in septic shock [50, 51] and pregnancy is associated with higher NO levels [52]. Intriguingly, although LPS administration in this study was associated with a marked increase in NO, the increase was actually less in pregnant mice. Basal cardiac iNOS mRNA expression was greater in pregnant mice, perhaps accounting for the higher basal levels of cGMP, but the response to LPS was similar. Even if the higher levels of cGMP were reproduced in vessels, it is unlikely to explain the greater hypotensive effect in pregnant mice. Similarly, the levels of VEGF increased in the NP mice and were greater than those in the pregnant mice at both time points. The changes in NO and VEGF would be expected to result in a greater fall in blood pressure in the NP mice in contrast to our observations. We measured circulating levels of cGMP to understand whether there was a greater vasodilator response in pregnant animals, but we found no difference. At the tissue level, the increase in endothelin-1 and VEGF levels were higher in NP mice at 12 h after LPS and although cGMP levels were greater in pregnant animals, this was primarily because cGMP levels declined more in NP mice.

We investigated vascular function in response to vasopressors and vasodilators, but again found no evidence of vascular dysfunction to explain the observed response to LPS. Pregnancy seemed to enhance the pressor response in both vessels and to induce conflicting responses to the vasodilators. In the aorta, pregnancy had no effect on the ACh response, but induced a lesser response to NO, while in the uterine arteries, pregnancy induced a more marked vasodilator response to ACh, but had no effect on the NO response. After LPS, the enhanced pressor response remained in both vessels from pregnant animals, while the vasodilator response in both vessels became greater in NP mice. In the literature, LPS has been shown to enhance the pressor response to PE in the aorta taken from LPS-treated male mouse aorta and to reduce the dilator response in both aorta and mesenteric arteries, consistent with our observations [53]. Overall, these data suggest that the changes in large vessel function in response to LPS do not explain the greater hypotensive response observed in pregnant mice. Changes in the arteriolar bed may be more important in understanding our observations.

Data from Elovitz et al. reported that progesterone administration to pregnant mice increased maternal morbidity and mortality [20]. The authors suggested that progesterone, in this case MPA, which has both progesterone and glucocorticoid agonist effects, enhanced the maternal inflammatory response causing IL-6 levels to be significantly higher [20]. However, generally, progesterone acts to repress inflammation [54], so the findings of Elovitz et al. are a little surprising. Nevertheless, progesterone levels are much higher in pregnancy and if they act to increase the IL-6 response in sepsis, they may contribute to the poor prognosis. Further studies are needed to understand whether progesterone plays a role in the hypotensive response to LPS observed in this study and more broadly in the adverse response to sepsis during pregnancy.

These data are important as they show that pregnancy is associated with a more marked hypotensive response to endotoxemia. This may explain the greater susceptibility to sepsis observed in pregnant women. Furthermore, the data show that the response of the systemic innate immune system to endotoxemia is similar in NP and pregnant mice, counter to previous suggestions [34]. The potential importance of our observation of a more marked monocyte margination in the lung vasculature in response to endotoxemia in pregnant mice awaits further study and if these changes are reproduced in the arteriolar bed, then it may explain the greater hypotensive response to LPS observed in pregnant mice in this study.

Conflict of Interest: The authors have declared that no conflict of interest exists.

Supplementary data

Supplementary data are available at BIOLRE online.

Supplementary Figure S1:Gating strategy for flow cytometry experiments showing density plots of innate immune cells in the lung. Neutrophils, monocytes, and macrophages are classified by SSc low, CD11b+, NK1.1- cells, neutrophils are positive for Ly-6G. The monocyte population has an intermediate F4/80 expression and positive for Ly-6C.

Supplementary Figure S2A-E: The effect of 10 μg LPS administered IP to pregnant (E16) and nonpregnant mice on mean arterial pressure (MAP, A), heart rate (HR, B), systolic arterial pressure (SAP, C), diastolic arterial pressure (DAP, D), and % of time spent inactive (%TSI, E; n = 4-7). The data represent the delta from the baseline readings obtained after 7 days post telemetry probe insertion before pregnancy.

Supplementary Figure S3 (A–L): Multiplex assessment of MAPK-AP-1, NFκB, PI3K-Akt pathways, and VEFGr2 activation in the left ventricle of nonpregnant or pregnant mice. *P < 0.05, after two group analysis using an unpaired Student t-test or Mann Whitney U tests depending on the data distribution. Data are expressed as median ± interquartile range (n = 6). NP, non-pregnant; E16, pregnant.

Supplementary Figure S4 (A-C): Messenger RNA expression of markers of cardiac function (NKx2.5, NKAα1, NKAα3) in the left ventricle of nonpregnant or pregnant mice. *P < 0.05, after two group analysis using an unpaired Student t-test or Mann Whitney U tests depending on the data distribution. Data are expressed as median ± interquartile range (n = 6). NP, nonpregnant; E16, pregnant.

Supplementary Figure S5 (A-C): Total caspase-3 levels in the left ventricle of nonpregnant or pregnant mice. *P < 0.05, after two group analysis using an unpaired Student t-test or Mann Whitney U tests depending on the data distribution. Data are expressed as median ± interquartile range (n = 6). NP, nonpregnant; E16, pregnant.

Supplementary Figure S6 (A and B): Serum PlGF levels in nonpregnant and pregnant mice measured by ELISA. Data are expressed as mean ± SEM (n = 3–7).

Supplementary Table S1: Mouse primer sequences for use in PCR/RT-PCR.

Supplementary Table S2. Flow cytometry antibodies.

Supplementary Table S3. Summary of the LPS dose response. The table gives the labor time, percentage pup survival (both mean ± SEM), and the number of mice studied at each dose level. Maternal mortality was observed in at least one mouse receiving doses of 20 μg or greater.

Supplementary Table S4. Treatment group comparisons between pregnant and nonpregnant hemodynamic responses to LPS. The overall trajectory of hemodynamic responses recorded over the 24 h period after LPS (or PBS control) was compared between pregnant and nonpregnant CD1 mice (overall). Comparisons at 1, 12, and 24 h were also made (n = 4–7). The p-values are shown in this table. NP PBS, nonpregnant PBS-treated; NP LPS, nonpregnant LPS-treated; E16 PBS, pregnant PBS-treated; E16 LPS, pregnant LPS-treated; ns, not significant.

References

Author notes