Activation of the Annexin A1 pathway underlies the protective effects exerted by estrogen in polymorphonuclear leukocytes

Abstract

Objective

The anti-inflammatory properties of the female sex hormone estrogen have been linked to a reduced incidence of cardiovascular disease. In the present study, we addressed whether estrogen could activate vasculo-protective mechanisms via Annexin A1 (AnxA1) mobilization in human polymorphonuclear cells (PMN).

Methods and Results

Using whole blood flow cytometry, we demonstrated that pre-menopausal women expressed higher levels of surface AnxA1 on circulating PMN, compared to males. This correlated with high plasma estrogen during the menstrual cycle. The addition of estrogen in vitro to male PMN induced rapid mobilization of AnxA1, optimal at 5ng/ml and 30 min incubation period; this effect was abolished in the presence of the estrogen receptor antagonist, ICI182780. Estrogen addition to human PMN induced a distinct AnxA1^hi CD62L^lo CD11b^lo phenotype, and this was associated with lower cell activation as measured by microparticle formation. Treatment of human PMN with E₂ inhibited cell adhesion to an endothelial cell monolayer under shear, which was absent when endogenous AnxA1 was neutralized. Of interest, addition of estrogen to PMN flown over the endothelial monolayer amplified its upregulation of AnxA1 localization on the cell surface. Finally, in a model of intravital microscopy, estrogen inhibition of white blood cell adhesion to the post-capillary venule was absent in mice nullified for AnxA1.

Conclusion

We unveil a novel AnxA1-dependent mechanism behind the inhibitory properties of estrogen on PMN activation, describing a novel phenotype, with a conceivable impact on the vasculo-protective effects of this hormone.

Introduction

The effects that sex hormones have on the processes regulating vascular inflammation have recently been re-evaluated, identifying novel cell targets and pathways. These efforts are prompted by epidemiological evidence that gender affects the incidence of cardiovascular and inflammatory diseases. Compared to age-matched males, pre-, but not post-, menopausal females are protected against cardiovascular disease¹, implicating a protective role for female sex hormones. Indeed, there is unequivocal evidence for vasoprotective and anti-inflammatory properties of estrogen (17-β estradiol, E₂)^{2, 3}, begging the issue of its cellular targets and molecular mechanisms. In the vasculature, estrogen has a direct positive effect on vascular function, improving blood flow⁴, and inhibiting vascular injury through regulation of endothelial-derived nitric oxide⁵. It is interesting to note that the timing of estrogen administration is a key factor in determining beneficial outcomes: when estrogen is given prophylactically, the extent of disease and/or its incidence is reduced, but this protection is lost if administered following disease onset^6-9.

In the context of acute inflammation, estrogen has been shown to be anti-inflammatory, suppressing the NF-κB pathway¹⁰, as well as down-regulating specific adhesion molecules on the vascular endothelium ¹¹, including the recently reported modulation of CD62P expression in the vessel wall¹². Regulation of adhesion molecules by estrogen would lead to modulation of the infiltration of leukocytes into the vascular tissue¹². Furthermore, estrogen reduces generation of the polymorphonuclear cells (PMN) chemoattractant CINC-2β (CXCL3, GRO3), thereby inhibiting PMN migration¹³. It is accepted that estrogen modulation of cytokines and chemokines might underlie its regulatory properties in vascular inflammation, providing a potential explanation to clinical observations.

Annexin A1 (AnxA1) is a 37-kDa protein with anti-inflammatory properties in the context of experimental inflammation¹⁴. In human resting PMN, AnxA1 is abundant in the cytoplasm, with only a small proportion present on PMN surface. Upon PMN activation, AnxA1 is rapidly mobilized to the cell surface, where it binds to its receptor, the G-protein coupled receptor, FPR2/ALX¹⁵. Following receptor ligation, AnxA1 is able to evoke a number of anti-inflammatory mechanisms, both in an autocrine and paracrine fashion¹⁴, inhibiting distinct stages of the leukocyte transmigration cascade^16-19. AnxA1 is also endowed with pro-resolving properties including promotion of efferocytosis of apoptotic PMN^{14, 19}. Upon activation, PMN are also capable of releasing microparticles (also known as ectosomes). These are small (<1μm) membrane-bound structures that can be released from most cell types and express surface molecules derived from its parent cell. Recent evidence suggests microparticles can induce cellular cross-talk and inhibit inflammatory responses ^20,21. Of interest to us, PMN-derived microparticles expressing AnxA1 are capable of evoking anti-inflammatory actions²².

Currently there is little evidence indicating that estrogen affects PMN reactivity and most studies have focused on the endothelial cell as the primary target to evoke vasculo-protective properties. Here, we started with the observation that pre-menopausal women express higher levels of AnxA1 on the PMN cell surface as compared to age-matched males, and this correlates with estrogen levels during the menstrual cycle. On these bases, we have studied the association between estrogen engagement of the AnxA1 pathway in human PMN and cell activation in vitro and in vivo. We propose this novel mechanism might at least contribute to the protective effects of estrogen in vascular inflammation.

Materials and Methods

Detailed information on protocols and ethics are in the Supplemental Methods file.

Cell culture and reagents

Unless otherwise stated, all cell culture reagents were obtained from Sigma-Aldrich (Poole, Dorset, UK). Human PMN were isolated via density gradient as described¹⁷. PMN (10⁶ cells per test) or whole blood aliquots (50μl) were incubated with E₂ prior to stopping reactions on ice. In selected experiments, the estrogen receptor (ER) antagonist ICI182780 was added for 10 min prior to E₂. In other cases, E₂ was added at 5 ng/ml (30 min) prior to stimulation with formyl-Met-Leu-Phe (fMLF; 10nM).

Flow cytometric detection of surface AnxA1 on human PMN

Cells or whole blood were incubated for 1 h at 4°C with mouse anti-human AnxA1 (mAb1B, produced in-house²³ or anti-human FPR2/ALX (Genovac, Freiburg, Germany), using a three-step protocol (Supplemental Methods file). Briefly, after the first 1-hour incubation, cells were washed and incubated with a rabbit, anti-mouse IgG FITC-conjugated. Then a further 30 min step with conjugated antibodies to CD16 (PE, clone eBioCB16) or to L-selectin (PE-Cy5,clone DREG-56) or, in some samples, to CD11b (APC, clone ICRF44) was run.

For detection of plasma microparticles, after incubation with E₂ in the presence or absence of fMLF, platelet-free plasma was double stained for CD66b (PE, clone G10F5) (or IgM isotype control) to identify PMN-derived plasma microparticles and co-stained for either AnxA1, FPR2/ALX or CD62L. Beads (1-μm each; Becton Dickinson, San Jose, CA) were also run, in selected samples, to control for microparticles size.

In all cases, 20,000 events were acquired by using a FACSCalibur flow cytometer (Becton Dickinson), and analysed using FlowJo analysis software (Version 9.2, Treestar Inc, Stanford, CA).

Western blot analysis

PMN were separated into cytosolic and membrane fractions as reported^{24, 25} and subjected to standard SDS–polyacrylamide gel electrophoresis (PAGE), transferred onto PVDF membranes (Millipore, Watford, United Kingdom). These were incubated with mAb1B (1 μg/ml) and HRP-conjugated goat anti–mouse IgG (Dako, Cambridge, United Kingdom). Proteins were detected using enhanced chemiluminescence (ECL) detection kit and visualized on Hyperfilm (GE Healthcare).

Determination of plasma of 17-β estradiol, progesterone, and cortisol concentrations

Blood was taken from females who were not on any oral contraceptives (for demographics see Supplemental Table 1) at days 2, 12, 19 and 26 of individual menstrual cycles. All blood samples were taken between 10 and 11 am. Plasma prepared from blood of female volunteers was tested for 17-β estradiol and progesterone concentrations using specific enzyme-immunoassays (estradiol EIA kit, Invitrogen; progesterone and cortisol EIA, Cayman Chemical Co, Inc, Ann Arbor, MI, USA)

Flow chamber assay

Human umbilical vein endothelial cells (HUVECs) were cultured until confluence and stimulated with 10 ng/mL tumor necrosis factor-α for 4 hours (Sigma-Aldrich). Isolated PMNs were incubated with vehicle, E₂ or E₂ plus mAb1B) for 30 min at 37°C, before flow over HUVEC monolayer at a rate of 1 dyn/cm², for 8 minutes, as previously described ¹⁷. PMN/HUVEC interaction in the flow chamber was monitored on 6 random fields recorded for 10 seconds. Analysis of total cell capture, rolling and firmly adherent PMN was carried out off-line by manual quantification. In some cases, cells flowed over the HUVEC monolayers were collected and compared with resting (pre-flowed) or adherent PMN for the extent of cell-surface AnxA1, CD62L and CD11b expression using the flow cytometric protocol described above.

Intravital microscopy of leukocyte recruitment on mesenteric post-capillary venules

Male C57Black6 mice (AnxA1^+/+ and AnxA1^−/−; 4 week old)²⁶ were injected with either PBS or estradiol (100ng/mouse, i.p.) 30 min before administration of IL-1β (10ng/mouse, i.p.). After 2 hours, animals were anesthetized and the mesenteric microvascular bed was exposed for analysis. Intravital microscopy was performed as previously reported¹⁸. A cautery incision was made along the abdominal region and the mesenteric vascular bed was exteriorized, placed on a viewing Plexiglas stage, and mounted on a Zeiss Axioskop “FS” with a water-immersion objective lens (magnification 40; Carl Zeiss, Welwyn Garden City, United Kingdom) and an eyepiece (magnification x10; Carl Zeiss). Mesenteries were superfused with thermostated (37°C) bicarbonate-buffered solution at a rate of 2 mL/min. When a suitable post-capillary venule was selected (diameter 20 - 40 μm; 100 μm straight vessel length), recording was started and kept for a 3-minute period, where adherent and emigrated leukocytes were counted.

Statistical analyses

Data are expressed as mean ± SEM. Student’s t-test was used to compare two groups with parametric data distribution. For multiple comparison analyses, one- and two-way ANOVAs were carried out, with appropriate Dunnet’s, Tukey’s or Bonferroni post correction tests. P-values <0.05 were considered to be significant.

Results

17β-estradiol (E₂) modulates AnxA1 expression in human PMN

Analyses of AnxA1 expression in circulating cells of healthy volunteers revealed a consistent and significant increment in the fraction of AnxA1^+ve PMN in female donors (Figure 1A). Out of 6 donors per gender, there was almost a doubling in cell surface AnxA1 on female PMN (P<0.01). This difference was not seen in the expression of the AnxA1 receptor FPR2/ALX (Figure 1A). Analysis at the single cell level revealed a significant increase in AnxA1 extent of expression on PMN taken from female donors (2-fold increase in MFI units), whilst no differences between gender could be measured for FPR2/ALX or CD62L (Table 1). These discrete, reproducible changes observed in PMN purified from females, including the higher degree of AnxA1 surface expression at the cellular level, prompted us to investigate E₂ effects on human PMN in vitro.

Figure 1. Gender differences in AnxA1 mobilization on PMN surface.

A) Different degree of expression of AnxA1 on PMN cell surface between males and females. Freshly prepared cells were stained and gated on CD16^+ve PMN, then double stained for AnxA1 and FPR2/ALX. Data report the % of positive cells, and are mean ± SEM of 6 subjects per group. * P<0.05 (Student’s t Test).

B) Dose-response and time-course of E₂-mediated mobilization of AnxA1 on the PMN cell surface. Healthy male blood was incubated at 37°C with the indicated concentrations of E₂ for 30 min (left graph), or at 5 ng/ml E₂ for different times (right graph). Cells were gated as described in panel A. Data are mean ± SEM of 5-6 analyses with distinct cell preparations. **P<0.001 (One-way ANOVA with Dunnett’s post-correction [left graph] or two-way ANOVA with Bonferroni post correction [right graph]).

C) Western blotting analysis for AnxA1 expression in PMN cytosolic and membrane fractions in the absence or presence of E2 (5ng/ml; 30 min). Blots are representative of three experiments.

D) Involvement of ER in E₂-induced AnxA1 mobilization. Whole blood aliquots were pre-incubated for 10 min at 37°C with the non-specific ER antagonist ICI182780, followed by a 30-min incubation with E₂ (5 ng/ml). AnxA1 mobilization was analysed by flow cytometry as in panel A. Data are mean ± SEM of 5 experiments. *P<0.001 compared to vehicle without E₂ (veh); # P< 0.001 compared to vehicle + E₂ (two-way ANOVA with Bonferroni post correction).

Table 1.

Expression of AnxA1, FPR2/ALX, CD62L and CD11b on male and female PMN.

Cell Surface Antigen	Male	Female
AnxA1	12.3 ± 1.9	22.9 ± 3.9 *
FPR2/ALX	12.9 ± 1.5	14.9 ± 0.3
CD62L	188 ± 19	165 ± 16
CD11b	41 ± 8.7	55 ± 9.0

Values show mean fluorescence intensity (MFI) units for AnxA1, FPR2/ALX, CD62L and CD11b on PMN from age-matched healthy male and females. Data are expressed as mean ±SEM from 6 subjects per group.

^*P<0.01 vs. respective gender value.

Male donors were used for these experiments. Concentration-response and time-course analyses revealed that 5ng/ml E₂ (18 nmol/L) for 30 min yielded optimal mobilization of AnxA1 on the PMN cell surface (P< 0.001; Figure 1B), and were selected for subsequent experiments. To validate FACS analysis data of in vitro mobilization of AnxA1, Western blot analysis was also carried out. Figure 1C shows that presence of E₂ mobilized AnxA1 to the membrane fraction. Such an effect was due to engagement of the ER, since pretreatment with the antagonist ICI182780, significantly reduced the ability of E₂ to mobilize AnxA1 on PMN surface: ~ 30% of PMN were AnxA1^+ve cells in the presence of E₂, whereas, only 11% of PMN were AnxA1^+ve in the presence of the estrogen receptor antagonist (10-100μM ICI182780; P<0.001) (Figure 1D).

Estrogen levels correlate with AnxA1 expression during the menstrual cycle

Whole blood FACS analysis revealed peak AnxA1 mobilization on PMN surface occurred at day 12 of the cycle (22.6% AnxA1^+ve vs. 8.56%, 7.02%, and 2.30% at day 2, 19, and 26, respectively; P<0.001) (Figure 2A). No statistically significant fluctuation of FPR2/ALX or CD62L expression could be quantified across the cycle (Figure 2A). Plasma E₂ progesterone and cortisol levels were also measured confirming that E₂ levels were highest at day 12 of the cycle (325 pg/ml vs. 91 pg/ml at day 2), whereas progesterone peaked at day 26 (12.3 ng/ml). This peak in progesterone levels coincided with a resurgence in E₂ plasma levels at day 26 (~150 pg/ml) (Figure 2B). Plasma cortisol levels did not change throughout the time-course of the menstrual cycle.

Figure 2. AnxA1 mobilization on PMN in females varies during the menstrual cycle.

A) Fluctuations in AnxA1, FPR2/ALX and CD62L cell surface expression of circulating PMN from healthy females on day 2, 12, 19, and 26 of their menstrual cycle. Data are mean ± SEM of 6 subjects; **P<0.001 vs. day 2 (two-way ANOVA with Bonferroni post correction). Dotted line indicates values obtained in cells from male volunteers.

B) Plasma levels of estrogen, progesterone and cortisol during the menstrual cycle. Time points as in Panel A.

C) Correlation between circulating estrogen levels and AnxA1 expression on the PMN cell surface. Estrogen levels were correlated with AnxA1^+ve PMN (r²=0.5628; left graph). Exclusion of Day 26 values, denoted by arrows, yields the graph on the right with a calculated r² = 0.7655.

D) Progesterone counteracts the effect of estrogen on AnxA1. Progesterone (5 and 10 ng/ml) was added to whole blood in the absence or presence of E₂ (5 ng/ml) for a 30 min incubation period. The degree of AnxA1 and CD62L expression on the PMN was quantified by flow cytometry. Data are mean ± SEM of three distinct experiments. *P<0.05 vs. control (dose 0 group); #P<0.05 vs. E₂ alone (one-way ANOVA with Dunnett’s post correction).

We obtained a good degree of correlation between PMN cell surface AnxA1 expression and plasma E₂ concentrations, with an r² value of 0.5628 (Figure 2C, left panel). To test whether the correlation was affected by the presence of progesterone, day 26 values (peak for progesterone) were removed from the correlation graph, yielding an even higher r² value of 0.7655 (Figure 2C, right panel). To assess the potential effect of progesterone on the ability of E₂ to mobilize AnxA1, we incubated whole blood from male donors with progesterone (10ng/ml; menstrual cycle levels) and E₂, observing a reduced ability to provoke AnxA1 mobilization by E₂ (P<0.05) (Figure 2D).

Profile of E₂-AnxA1 pathway in PMN activation in vitro

The results so far have shown the effects of E₂ on AnxA1 mobilization in the absence of cell activation. In order to investigate the potential effects of E₂ associated AnxA1 externalization, we carried out a series of in vitro experiments where whole blood was either pre-treated with fMLF (10nM, 20min) followed by E₂ treatment (5ng/ml, 30min), or pre-treated with E₂ followed by fMLF (in comparison to vehicle and single treatment controls).

Figure 3A shows box-and-whisker plots of this set of experiments. As previously observed, E₂ treatment significantly mobilized AnxA1 to PMN surface, compared to vehicle control (P<0.001). AnxA1 externalization was not observed in fMLF treatment alone or when PMNs were pre-treated with fMLF prior to E₂ (Figure 3A). In contrast, pre-treatment of PMN with E₂ prior to fMLF resulted in significant mobilization of AnxA1 similar to levels seen in E₂ treatment alone. Analysis of FPR2/ALX expression indicated lack of externalization upon E₂ treatment but an up-regulation post-fMLF, irrespective of the presence of E₂ (P<0.001 for all 3 treatment groups).

Figure 3. Estrogen confers an anti-inflammatory phenotype to human PMN.

A) Male whole blood was treated with either fMLF (10 nM) for 20 min, E₂ (5 ng/ml) for 30 min, or a combination of treatments, that is fMLF, followed by E₂ or pre-treatment with E₂ followed by fMLF. All blood aliquots were kept at 37°C for a total 50 min incubation time. Flow cytometry afforded quantification of the degree of expression of AnxA1, FPR2/ALX, CD62L, CD11b, and CD66b on PMN cell surface. Data are mean ± SEM of 6 experiments conducted with distinct blood preparations. *P<0.001 compared to vehicle control (one-way ANOVA with Dunnett’s post correction).

B) Representative histograms depicting the mean fluorescence units (MFI) for AnxA1, CD66b and CD62L out of the experiments presented in Panel A. Grey histograms indicate isotype controls (IgG1_κ for AnxA1 and CD62L, and IgM_κ for CD66b).

Figure 3B reports representative histograms from these experiments, showing AnxA1 expression alongside that of CD66b and CD62L. Analysis of CD66b expression revealed a significant augmentation in PMN surface expression, when cells were treated with fMLF alone or given as pre-treatment. This was not observed, however, with vehicle or E₂ (alone or as pre-treatment). Vehicle and E₂ alone had no effect on CD62L shedding, whereas fMLF treatment caused significant CD62L-shedding, which was unaffected by with E₂ addition. Interestingly, pre-treatment of PMN with E₂ (5 ng/ml) did not rescue the cells from CD62L shedding, but did maintain low expression of the β₂-integrin CD11b when compared to vehicle and E₂ treatment alone; (Figure 3A and 3B). Therefore, E₂ can selectively modulate PMN stimulation induced by fMLF. Next, we measured neutrophil-derived microparticles as a test for E₂ to inhibit downstream PMN activation, and used the plasma from blood treated as in Figure 3. Microparticles were identified according to their size (and using 1μm beads– see Supplemental Figure 1).

Treatment of PMN with fMLF promoted a significant release of CD66b^+ve microparticles compared to vehicle control (P<0.001). Addition of E₂ prevented microparticle generation elicited by fMLF and, this time, the hormone was active even when added after the stimulus (Figure 4A). These responses are shown in representative histograms (Figure 4B). Microparticles generated in these experimental conditions were characterized further quantifying specific antigens on a per microparticle basis, reporting it as MFI units. Treatment of fMLF alone led to a two-fold increase in the expression of AnxA1 on CD66b^+ve microparticles compared to vehicle (P<0.001). This response was insensitive to cell exposure to E₂ (5 ng/ml) following fMLF, whereas estrogen was ineffective when given alone (Figure 4C). In contrast, and in line with the global data on microparticle formation, pretreatment with E₂ abrogated fMLF-induced generation of AnxA1/CD66b double positive microparticles (Figure 4C). The extent of expression of FPR2/ALX on PMN-derived microparticles was marked even in vehicle-treated cells and only marginally modulated by the treatments. In contrast, CD62L expression (MFI units) was reduced by E₂ only when added before fMLF (Figure 4C). Collectively these data indicated that E₂ can modulate PMN activation recorded via microparticle generation and can modify, at least in part, some of the proteins displayed on these microstructures.

Figure 4. Release of PMN-derived microparticles in plasma is regulated by estrogen.

Plasma was collected from blood treated as described in Figure 4.

A) PMN microparticles in the plasma were identified with CD66b staining and calculated as percentage of the total microparticles. Data are mean ± SEM of 6 experiments conducted with distinct blood preparations. *P<0.001compared to vehicle control (one-way ANOVA with Dunnett’s post correction).

B) Representative histogram plots for CD66b in relation to the different treatments. Grey histograms indicate isotype control (mouse IgM_κ).

C) Further characterization of CD66b^+ve microparticles with double staining for AnxA1, FPR2/ALX and CD62L, reporting variation in mean fluorescence intensity (MFI) units for each antigen. Data are mean ± SEM of 6 experiments conducted with distinct blood preparations. *P<0.001 compared to vehicle control (one-way ANOVA with Dunnett’s post correction).

E₂ modulates PMN reactivity under flow via endogenous AnxA1

In the next series of experiments we wanted to determine whether E₂ could modulate PMN activation, and hence inflammatory responses, in an AnxA1 dependent fashion. To model the impact of E₂-treated PMN in vascular inflammation, we used the flow chamber assay. Following HUVEC activation by TNF-α, PMN from healthy males treated with vehicle, E₂ or E₂ plus an anti-AnxA1 neutralizing Ab were flowed over the monolayer.

Cell incubation with E₂ did not affect the ability of the PMN to come in contact with the monolayer, hence did not affect cell capture (the total number of interacting cells; Figure 5, top panel). However, a selective interference with the extent of PMN adhesion could be quantified with approximately 3-fold decrease (Figure 5, middle panel; P<0.001). This effect was significantly lost when PMN were co-administered with the AnxA1 neutralizing Ab and E_2. There was no significant difference in the capture or rolling of PMN between the 3 treatments (Figure 5A; bottom panel). Figure 5B displays representative images captured from these experiments.

Figure 5. E₂ inhibits PMN adhesion to inflamed endothelium in an AnxA1-dependent fashion.

Human umbilical vein endothelial cells (HUVEC) were stimulated with TNF-α (10ng/ml) for 4 hours. Human PMN were treated with either vehice, E₂ (5ng/ml) or with E₂ plus AnxA1 neutralizing Ab (20μg/ml) for 30 min. PMN were then flowed over the stimulated endothelial monolayer for 8 min, after which 6 random fields were recorded.

A) PMN capture, adhesion and rolling were measured. *P< 0.001 E₂ compared to vehicle. # P< 0.001 E₂ compared to E₂ plus AnxA1 neutralizing Ab.

B) Representative images of PMN interactions with endothelium under flow. Following analysis using Image-Pro, neutrophils that have firmly adhered to the endothelium are situated within the dotted circles. Empty dotted circles indicate non-adhered neutrophils.

The experiments of flow chamber assay allowed us to investigate the phenotype of the E₂-treated PMN that had encountered the inflamed endothelium in the flow chamber, but did not adhere; as well as cells that had adhered to the endothelium. Addition of E₂ (5 ng/ml; 30 min) to PMN that had flown over the HUVEC monolayer resulted in higher levels of AnxA1 on their surface (Figure 6B) compared to hormone addition to resting PMN (pre-flow in Figure 6A; P< 0.05). These changes in AnxA1 were not reflected by any detectable change in CD62L (Figure 6A and 6B). In contrast, whilst post-flow PMN displayed higher CD11b levels to static (pre-flow) PMN, addition of E₂ (5 ng/ml; 30 min) significantly attenuated this cellular response (last set of data in Figure 6A and 6B).

Figure 6. E₂- treated PMN post flow have a marked increase in AnxA1 surface expression, but low CD11b.

PMN were treated as described in Figure 5. Pre-flow (A), post-flow (B) and post-adherent PMN were stained for AnxA1, CD62L and CD11b expression on the plasma membrane. P< 0.001 compared to vehicle (Student’s T test). Representative histograms are also shown.

We also recovered post-adherent PMN (dislocating cells by increasing shear) and detected much higher AnxA1 levels in vehicle treated cells (reminiscent of our old observations¹⁶) and this could be partially modulated by E₂ (5 ng/ml; 30 min) (Figure 6C). CD11b was higher in post-adherent PMN and was no longer susceptible to the addition of the hormone.

E2 modulates in vivo PMN reactivity via endogenous AnxA1

The effect of E₂ on PMN recruitment under flow was also assessed in vivo using the inflamed microvasculature of the mesentery. This model was chosen because of the tonic buffering role exerted by endogenous AnxA1²⁵.

Treatment of WT mice with IL-1β alone caused a significant increase in cell adhesion and emigration when compared to vehicle control, as assessed at 2 hours post-cytokine (Figure 5A, P<0.001). Pre-treatment of the animals with E₂ (100 ng i.p.; -30 min) significantly attenuated the effect of IL-1β, both in terms of cell adhesion and emigration (P<0.001). Figure 5B illustrates representative images of the mesenteric microcirculation of wild type mice treated with IL-1β in the presence or absence of E₂. Importantly, modulation of this inflammatory event by E₂ was dependent on endogenous AnxA1 since this hormone was ineffective when tested in AnxA1^−/− male mice (Supplemental Figure 2). Hemodynamic parameter can be seen in Supplemental Table 2.

Discussion

In the present study we provide evidence that the female sex hormone estrogen exerts inhibitory effects on human PMN, which translate to anti-inflammatory properties in vascular inflammation, and provide a strong indication for a functional association with endogenous AnxA1. Together, these results re-address the target for the anti-inflammatory mechanisms of estrogen indicating that, besides the vascular wall, this hormone can have marked effects on the blood-borne leukocyte.

The anti-inflammatory nature of female sex hormones has long been noted and it is well established that, for instance, inflammatory and cardiovascular diseases are exacerbated in post-menopausal women¹. More specifically, recent studies have shown post-menopausal women present increased levels of adipokines (e.g. serum amyloid protein A ²⁷) as well as pro-inflammatory cytokines (e.g. IL-8, ²⁸) in plasma – all of which increase the risk of cardiovascular diseases. The present study was prompted by the observation that a higher degree of cell surface AnxA1 could be measured on circulating PMN of pre-menopausal female volunteers compared to age-matched males.

A previous study reported the detection of the cell surface AnxA1 in blood leukocytes, specifically PMN, which correlated with plasma cortisol ²⁹.These latter results presented a functional, well-characterized association between glucocorticoids and cellular expression of the anti-inflammatory protein, AnxA1³⁰. Here our unexpected observation of sexual dimorphism, specifically for the blood PMN, prompted us to perform more detailed analyses during the course of the menstrual cycle. These new data indicate that cell surface AnxA1 in circulating PMN is selectively modulated by levels of circulating hormones, with a significant and robust correlation to plasma estrogen levels, but neither progesterone nor cortisol. Such an association was not consequent to a global, non-selective change in PMN phenotype, since AnxA1 cell surface expression was discretely augmented in correlation with peak of estrogen level without changes in cell surface FPR2/ALX or CD62L.

Estrogen is known to modulate a variety of responses in target cells that could impact on the intensity of an experimental inflammatory response. For example, estrogen regulates PMN recruitment by inhibiting the chemokine CINC-2β¹³ and other cytokines, likely consequent to a suppression the NF-κB pathway¹⁰. Such a complex network of effects is not surprising in view of the pleiotropic action of this hormone. In the context of vascular inflammation, the majority of studies have focused on the endothelium. Estrogen maintains vascular tone by mediating key endothelial-derived factors. For example, in vitro treatment on endothelial cells with estrogen augments the production of the anti-inflammatory autocoids nitric oxide⁵, prostacyclin (PGI₂)³¹ and endothelial-derived hyperpolarizing factor (EDHF)³². The dichotomy between gender made the PMN the focus of the present study.

We found that there was a distinct and reproducible mobilization of AnxA1 in human PMNs upon incubation with estrogen; the effect was non-genomic, optimal at 30 minutes, and mediated by classical ERs as demonstrated by the use of the non-selective estrogen receptor antagonist ICI182780. The data were re-assuring in view of the well documented presence of ER-alpha and ER-beta in human PMN^{33, 34}. Moreover, a microarray study in the rat has reported a directly associated ER binding with gene regulation of AnxA1 in blood leukocytes³⁴. Together with our new data, we can propose that this hormone produces non-genomic and genomic regulation of this mediator of endogenous anti-inflammation.

Our ex vivo measurements showing that high AnxA1 correlate with high E₂ (but not cortisol ) during the menstrual cycle added support to our in vitro findings. In the time-course analyses, we chose Day 12 as E₂ is the only hormone that is high during this period of the menstrual cycle. However, E₂ levels remain high for a few days during the menstrual cycle. It is tempting to propose that the anti-inflammatory effects of E₂ (via AnxA1) could be explained by the fact that at days 13 and 14, when E₂ levels remain high, follicle-stimulating hormone (FSH) and luteinizing hormone (LH) are also high. Since both FSH and LH have been positively correlated with AnxA1²⁹ E₂ could act in concert with FSH and LH and exert their anti-inflammatory actions by promoting AnxA1 mobilization. Of course, this hypothesis would need to be substantiated by future investigations.

In the human PMN, AnxA1 represents a substantial amount of the intracellular proteins, being calculated to be between 2% and 4%^{16, 35}. We have gathered evidence for the existence of at least two pools of the protein in resting PMN, one cytosolic and the other associated with sub-cellular organelles, such as gelatinase and azurophilic granules³⁶. More recent studies indicate that these distinct pools of intracellular AnxA1 can be mobilized in a stimulus dependent fashion; thus, cell activation consequent to outside-in signaling (e.g. after cell adhesion) exports the granule/vesicular pool of the protein^{16, 38}. In contrast, pharmacological treatment with dexamethasone ³⁷, cromones ³⁸ or other soluble compounds^{25, 39} appears to engage the cytosolic pool of the protein. The latter process seems relevant to the effects of estrogen upon the PMN treatment, because analyses of specific markers for intracellular granules (e.g. CD66b or CD11b) failed to detect modulation of cell surface expression at concentration, and incubation time, optimal for E₂ promotion of AnxA1 mobilization. Of interest, recent work has demonstrated existence of a partial similarity between glucocorticoids and estrogens with respect to gene modulation in the human PMN, observing also how an ER antagonist can impact on the biological properties of glucocorticoids in vitro⁴⁰ and in vivo⁴¹. The present study extends the potential for this crosstalk, or at least shared properties, between these two classes of hormones to rapid non-genomic modulation of AnxA1 expression.

The functional relevance of estrogen-induced AnxA1 externalization was initially investigated in vitro, monitoring classical PMN activation markers. Pre-incubation of human PMNs with estrogen inhibited cell activation by formylated peptides, in a selective fashion, since significant inhibition was quantified against fMLF-induced CD11b and CD66b upregulation, but not CD62L shedding. Chronologically, the mechanism activated by estrogens must precede the cell simulation with fMLF since the hormone was inactive when added after the PMN activator. Indeed, a distinct PMN phenotype (AnxA1^hi CD62L^lo CD11b^lo) is apparent in estrogen pre-treated PMN, as seen in several settings (see below), prompting us to suggest the existence of an E₂-generated anti-inflammatory PMN phenotype that may impact on further cell recruitment to sites of inflammation^{42, 43}.

The fact that estrogen could alter specific pathways in the human PMN can also be deduced from the lack of modulation of fMLF-induced FPR2/ALX up-regulation. This receptor is known to be expressed in a variety of granules⁴⁴ and to be up-regulated upon cell trafficking⁴⁵; such a pattern is grossly similar to what described for AnxA1, yet it was not modulated by estrogen. This apparent lack of modulation of cell surface FPR2/ALX expression was also evident in circulating cells taken from healthy subjects.

Our analyses of PMN activation were concluded by quantifying the production of PMN-derived microparticles, a response which is gaining momentum both with respect to biological functions^{22, 39, 46} and suitability as disease biomarker^{46, 47}. In line with the results obtained with CD11b and CD66b, addition of estrogen prior to fMLF reduced the production of PMN-derived microparticles, quantified as CD66b^+ve microstructures. Collectively these results indicate that estrogen application to human PMNs - prior to stimulation - afford significant attenuation of cell activation. We propose that modulation of PMN responsiveness by estrogen could be at least contributory, in association with the reported effects on vascular wall reactivity, to gender specificity with respect to gender-specific disease penetrance observed for cardiovascular pathologies.

This proposition, and its link to endogenous AnxA1, is corroborated by the experiments of flow chamber where the specific effect of E₂-treated PMN on inflamed endothelium was tested. The anti-adhesive effect of AnxA1 on PMN under flow has been well documented by our group^{17, 38}. We found that in comparison to vehicle control, there was significantly lower degree of PMN adhesion to the inflamed endothelium with E₂-treated PMN – an effect that was lost when AnxA1 was neutralized. These data were supported by in vivo experiments where the effect of estrogen on the inflamed microcirculation was tested. A real time protocol of intravital microscopy was applied whereby the mesenteric vasculature was inflamed by administration of IL-1β. Intraperitoneal treatment of male mice with estrogen produced a marked attenuation of the extent of leukocyte adhesion and emigration. This finding is aligned with the anti-migratory and anti-inflammatory properties of this hormone as demonstrated in ovariectomized animals, where leukocyte recruitment was significantly increased¹². These experiments provide a crucial functional link between the variety of analyses performed with human subjects and cells and the potential pathophysiological relevance, noting also that the significant attenuation of white blood cell recruitment onto the post-capillary venule vessel wall, afforded by estrogen, was absent in AnxA1^−/− mice.

Finally, we wanted to determine whether E₂-treated PMN that had encountered the inflamed endothelium under flow, but did not adhere, had a similar phenotype observed in our in vitro experiments. Analysis of AnxA1 on post-flow PMN suggest that we might have underestimated the effect of E₂ on AnxA1 mobilization: there was a 3-fold increase in AnxA1 expression on E₂-treated post flow PMN, compared to the pre-flow PMN (static incubation). This is a very novel observation and it stresses the importance of testing modulation of the AnxA1 pathway¹⁴ in the presence of shear. It is therefore plausible that the externalization of AnxA1 induced by the classes of drugs listed above (e.g. glucocorticoids, cromones and more) might be much more pronounced in intravascular settings, where the PMN is subjected to shear stress.

In-line with our in vitro results in static settings, there was a significant attenuation of CD11b expression on post-flow E₂-treated PMN, compared to their adherent counterparts. A similar effect has been observed in eosinophils, where glucocorticoid-induced AnxA1 mobilization, resulted in inhibition of β₂ integrin-ICAM-1 interactions in eosinophils, and to a lesser degree PMN⁴⁸. Taken together, these results suggest that E₂ has two distinct effects on the PMN: under non-inflamed “normal” conditions, E₂ induces a modest increase of surface AnxA1 with no modulation of CD11b, which could be thought of as priming mechanism. However, under inflamed, non-static conditions, E₂ is capable of mounting a pronounced protective anti-inflammatory response by augmenting AnxA1 surface expression and modulating CD11b expression.

In conclusion, we unveil a novel mechanism, whereby AnxA1 externalization can drive estrogen-regulated anti-inflammatory processes in PMN biology. It remains to be seen whether such an association could also be extended to other properties recently ascribed to AnxA1 such as, for instance, modulation of the life span of migrated neutrophils^{37, 49} and of their removal by phagocytes^{19, 50}.