Estrogen-mediated impairment of macrophageal uptake of environmental TiO₂ particles to explain inflammatory effect of TiO₂ on airways during pregnancy

Abstract

Innate defenses against environmental particulate exposures can become deficient when physiological background of the organism is unbalanced. Even those exposures considered innocuous may then become harmful. For example, one of the important inherent risks of pregnancy is increased inflammatory responsiveness in the airways, which extends to exposures considered otherwise innocuous: it has been observed that normally “inert” particulates become inflammatory in pregnancy. They lead to enhanced airway inflammation associated with increased asthma risk in the offspring in the BALB/c model. It was hypothesized that pregnancy hormones alter macrophageal uptake and clearance of particles. This study shows that the phagocytic activity of alveolar macrophages (AM) and RAW264.7 cells against titanium dioxide (TiO₂) was inhibited in pregnancy by ∼10% and in vitro by estradiol by ∼20%; progesterone potentiated this effect. Hence, enhanced inflammation in pregnancy as an outcome of exposure to the “inert” TiO₂ may be due to an effect of pregnancy hormones which decrease the ability of the airways to clear the particles. AM (at 10⁶ cells/recipient) isogenically transplanted from pregnant mothers into airways of recipients were able to confer the phenotype of inflammatory response to TiO₂ (PMN counts of 1.62 [± 0.19] × 10⁵/ml versus 0.61 [± 0.13] × 10⁵/ml in control). Because this small amount of transferred AM could not replace the AM population in the recipients’ lungs, it is postulated that the effect is mediated by inhibitory signaling factors that AM produce and release; hence, a list of probable molecules was identified via genome-wide microarray.

• Airway inflammation
• particles
• pregnancy
• Titanium dioxide
• TiO₂

Introduction

Pregnancy is an immunologically ‘unusual’ condition; due to a multi-fold elevation in pregnancy hormones and a variety of other changes throughout the organism, pregnancy presents a model of immunological variability and enhanced susceptibility to disease, including abnormal inflammatory responses. Epidemiological data suggest pregnancy increases influenza mortality (Memoli et al., 2012; Rasmussen et al., 2008), exacerbates asthma (Kircher et al., 2002), and leads to lung injury (Sheng et al., 2012) and deleterious effects of stress (Coussons-Read et al., 2007). This risk extends to commonplace environmental exposures like inert respirable particulates without soluble components considered non-toxic. Previously, we demonstrated in an in vivo model that in pregnancy airway inflammatory activity is exacerbated in response to an immunologically “inert” environmental particulate titanium dioxide (TiO₂), which served in the past as a proto-typical negative control particle in immunotoxicology studies. This exposure of pregnant dams was associated with increased risk of asthma in their neonates (Fedulov et al., 2008). The mechanism could be linked to an increasing asthma incidence and mediate the connection between environmental exposures and asthma risk (Miller & Ho, 2008; Strachan, 2000). The long-term goal of our studies is to identify potential therapeutic targets and preventative strategies to curb the risk of asthma and other inflammatory airway diseases.

In the current study, we focused on the critical step in the pathogenesis - the interaction of the particles with alveolar macrophages (AM), the first cells to capture and eliminate the agent in the airways (Bowden, 1987; Brain, 1985; Lohmann-Matthes et al., 1994; Thepen et al., 1994). While many factors in a pregnant organism can affect airway responses, we hypothesized that immuno-endocrine interactions lead to alterations in macrophage ability to clear particulates, potentially as a result of action of the pregnancy hormones estrogen and progesterone produced in gradually increasing levels throughout gestation (Albrecht et al., 2000; Cahill, 1995) up to 5-fold or more during pregnancy (Zhang et al., 1999). This prominent concentration increase and the known ability of these hormones to affect immune regulation have served as a rationale to consider their role in this study. To prove the critical role of the AM, adoptive transfer of these cells from pregnant to control mice were performed; we report here that this led to carryover of the inflammatory response to prototypical inert particle TiO₂ in the recipients. We report further that phagocytosis of TiO₂ by AM was inhibited in ex vivo cells from ‘pregnant’airways and that estradiol treatment “mimics” a similar inhibition of the uptake of TiO₂ by primary AM and – for a more detailed investigation – in macrophageal cell lines RAW264.7 and J774 in vitro.

Taken in combination, these data suggested that the deficiency in particle uptake and clearance during pregnancy could be explained by production of yet unknown estrogen-stimulated factors that inhibit phagocytosis. Hence, microarray analyses were performed and a short list of genes were identified that were concordantly up-regulated in AM by pregnancy and by estradiol in the RAW cell line, and that likely include the candidate inhibitors.

Materials and methods

Animals

Balb/c mice were obtained commercially (Charles River Laboratories, Cambridge, MA) as time-pregnant E14 dams or age-matched controls. All mice were housed in a clean barrier facility where animals are maintained at 22–24 °C with a 12-h dark/light cycle, with independent pressure-gradient enabled ventilation system. All mice had ad libitum access to standard rodent chow and filtered water. All studies were approved by the local IACUC, the Harvard Center for Comparative Studies.

Particles exposure

Respirable-size TiO₂ particles (with mean particle size of ∼1 µm) were a gift from Dr. J. Brain (Harvard School of Public Health, Boston, MA); these were used previously in our studies (Fedulov et al., 2008; Palecanda et al., 1999) and characterized in Beck et al. (1982). Particle samples were baked at 200 °C for 3 h (to eliminate endotoxin), aliquoted, and then stored frozen at −80 °C. Particle suspensions (50 µg in 50 µl) or PBS solution (vehicle) were administered by a single intra-nasal (IN) insufflation of pregnant and control mice under light (2%) isoflurane anesthesia. The exposure was performed 1 day after the mice arrived at the facility to allow recovery from travel stress; the E14–15 exposure window is based on prior success in studies that showed that airway exposures to particulates led to increased asthma risk in their offspring (Fedulov et al., 2008).

Alveolar macrophages (AM)

AM were obtained by bronchoalveolar lavage (BAL). In normal mice, both pregnant and control, the BAL cell population is ≈99% AM according to microscopy and flow cytometry; hence, no isolation of cells was required. BAL was performed five times with a total of 0.6 ml sterile LPS-free PBS instilled and harvested gently. Lavage fluid (recovery volume was ≈90% of instilled) was collected and centrifuged at 1200 rpm (300 × g) for 10 min, and the cell pellet was re-suspended in 0.5 ml PBS. (1) For adoptive transfer (isogenic transplantation) experiments, the cells were counted via a hemocytometer, diluted to the required concentration with PBS and transferred by IN insufflation at 10⁶/recipient, followed 24 h later by the TiO₂ particle challenge. Recipients received cells from pooled (not individual) donor samples. (2) In analytical BAL, the volume was normalized to 100 µl of PBS after centrifugation, and total and differential cell counts were performed on cytocentrifuge slides prepared by centrifugation of samples at 800 rpm for 5 min (Cytospin 2; Shandon, Pittsburgh, PA). These slides were fixed in 95% methanol and stained with Diff-Quick (VWR, Boston, MA), a modified Wright-Giemsa stain, and a total of 200 cells are counted for each sample by microscopy. (3). For in vitro hormone treatments, the BAL AM were isolated as above and then re-suspended in cell culture media (see below).

Experimental protocol for adoptive transfer

To test the hypothesis that AM are responsible for the effect observed earlier in pregnant versus control mice (Fedulov et al., 2008), a series of adoptive transfer (isogenic transplantation) experiments were performed by IN instillation of AM from BAL of pregnant or control donors into naïve recipients of similar age. As depicted in a schematic (Figure 1A), transfer of 10⁶ macrophages/mouse was followed the next day by an IN instillation of TiO₂ particle suspension and by analysis 48 h later.

Figure 1. Adoptive transfer of alveolar macrophages. (A) Schematic of adoptive transfer protocol. AM from BAL of pregnant E14 and control mice were introduced by intra-nasal instillations to intact recipient female mice, followed by exposure of the recipients to TiO₂ or vehicle the next day, and analysis by BAL 48 h afterwards. Thick red dotted line indicates intra-nasal TiO₂ exposure; thin dotted lines indicate intra-nasal cell instillation. (B) Macrophages from pregnant mice confer the phenotype. Alveolar macrophages from BAL of pregnant E14 but not control mice confer increased responsiveness to TiO₂ to intact recipient female mice. See schematic in Figure 1 for group assignments. Positive controls were pregnant mice with TiO₂ exposure, as in prior studies. n = 6/group, three repeats. *p < 0.05 against any of the negative groups.

Table

Abbreviation	Rcp:NormTi	Rcp:PregTi	Rcp:NormPBS	Rcp:PregPBS	POS Ctrl
Donor	Normal mouse	Pregnant mouse	Normal mouse	Pregnant mouse	N/A
Treatment	TiO2	TiO2	Vehicle	Vehicle	TiO2

Cell culture

RAW 264.7 and J774 cells were obtained from ATCC (Manassas, VA). For maintenance, cells were cultured in 100-mm Petri dishes in DMEM with stable L-glutamine (Lonza, Hopkinton, MA) complemented with 10% FCS, penicillin (100 U/ml), streptomycin (100 µg/ml) and HEPES. For hormone treatment, these cells (or, in some experiments, AM) were serum-starved for 72 h in Macrophage-SFM media (Invitrogen, Carlsbad, CA), then detached and plated in 12-well TC plates at 2 × 10⁵/ml per well overnight, then detached by pipetting and transferred to 96-well zero-adhesion plates (Corning, Corning, NY) for the TiO₂ uptake assay. 17β-estradiol (Sigma, St. Louis, MO) was used at concentrations of 10–0.0001 µg/ml overnight. In several experiments, progesterone (Sigma) at concentrations of 1.0–0.1 µg/ml was used alone or in combination with estradiol. Controls included naive (untreated) cells and cells treated with an equal volume (100% amount) of the vehicle (ethanol) only. Stock solutions of the hormone in pure ethanol or ethanol alone were serially diluted in culture media to the necessary concentrations; hence, in the highest 10 µg/ml estradiol sample (and the respective ethanol alone control), the concentration of ethanol did not exceed 1:10 000 (0.01%).

Flow cytometry and particle uptake

AM from pregnant and control mice were cultured in vitro in zero-adhesion 96-well plates in 100 µl PBS supplemented with 1% bovine serum albumin (BSA; tissue culture grade, Sigma) solution, and treated with 15 µl of 1 mg TiO₂/ml particle suspensions for 2 h. Particle uptake was registered by flow cytometry (FACS Canto II, BD) based on side scatter (as in Stringer et al., 1995). The data represent the percentage of cells engaged in phagocytosis at a particular moment in time (1.5 h), as evident by their increased granularity. Voltages for FSC and SSC were set relatively low to optimize the detection of macrophages with particles (see Figure 2A).

Figure 2. Pregnancy hormones inhibit particle uptake. (A) Detection of phagocytosing cells based on side-scatter. Left: Scattergram and histogram of vehicle control (PBS) sample. Right: Test sample with macrophages phagocytosing TiO₂ particles for 2 h. Cells in charts were gated from general sample scatter to exclude debris. (B) Uptake of TiO₂ particles in primary murine AM was reduced in cells from pregnant mice compared to controls. n = 12. *p < 0.01. Mean ± SEM. Representative experiment of three is shown. (C) Estradiol inhibits TiO₂ particle uptake. Overnight pre-treatment of normal primary murine AM with 17β-estradiol (1 and 10 µg/ml) led to inhibition of particle uptake similar to that seen in cells from pregnant mice. Left (black bars): females, right (grey bars): males. Pooled average of two experiments; mean ± SEM. Effect significant at p < 0.05 for 10 µg/ml. (D) Dose-response inhibition of phagocytosis of TiO2 particles by RAW264.7 cells due to 17β-estradiol overnight pre-treatment. Average values of three representative repeats. Mean ± SEM. *p < 0.05, **p < 0.01. (E) Dose-response inhibition of phagocytosis of TiO₂ particles in J774 cells by 17β-estradiol overnight pre-treatment. Average values from two representative repeats. Mean ± SEM. *p < 0.05. (F) Cell viability in estradiol-treated cultures determined by Annexin V and Cytox labeling to exclude both apoptotically- and necrotically-compromised cells. Representative sample from routine measures. Mean ± SEM.

Additionally flow cytometry was used to evaluate purity and viability (by Annexin V staining for apoptosis and Cytox staining for necrosis) of cells for adoptive transfer experiments and to monitor viability in estradiol treatment experiments. The cell viability kit was obtained from Invitrogen and used in adherence to manufacturer recommendations. Positive control was accomplished by a short incubation of respective cells with 20% methanol.

Microarray and real-time PCR

RNA was extracted using an RNEasy kit (Qiagen, Valencia, CA) including the optional step of DNAse digestion; quality of the RNA was validated by Nanodrop analysis and Bioanalyzer scans to assure no genomic DNA contamination had occurred. Gene chip microarray analysis was performed at the Dana Farber Cancer Institute Microarray Cores using Affymetrix mouse 430A 2.0 arrays. Real-time PCR validation was used to confirm microarray data. Raw microarray data were submitted to the NCBI GEO database in compliance with the MIAME protocol (Series GSE52649).

Real-time PCR was performed using SYBRGreen Sso Advanced Supermix (Bio-Rad, Hercules, CA) and custom primers (Integrated DNA Technologies, Inc, Coralville, IA) on a Bio-Rad CFX96 real-time system. Primer sequences used were (5′ → 3′ direction): Arg1-FF: ATGGAAGAGACCTTC-AGCTAC, Arg1-RR: GCTGTCTTCCCAAGAGTTGGG, alternative Arg1 pair: GAACACGGCAGTGGCTTTAAC, TGCTTAGCTCTGTCTGCTTTGC; Fizz1-FF: TCCCAGTGAATACTGATGAGA, Fizz1-RR: CCACTCTGGATCTCCCAAGA, alternative Fizz1 pair TCCCAGTGAATACTGATGAGA, CCACTCTGGATCTCCCAAGA; YM1-FF: GGAGTAGAGACCATGGCACTGAAC, YM1-RR: GACTTGCGTGACTATGAAGCATTG; KC-FF: GCTTGAAGGTGTTGCCCTCAG, KC-RR: AAGCCTCGCGACCATTCTTG, alternative KC pair: GTAACGGAGAAAGAAGAC, CAGAACTGAACTACCATC; Marco-FF: TCTGGGAACATCTGGCTGGACAAT, Marco-RR: GCATTCCACACCCGGATCTTCATT; SRA-FF: AGAATTTCAGCATGGCAACTG, SRA-RR: ACGGACTCTGACATGCAGTG; LAIR1-FF: GCTCCAGTGCACACATCCTGCC, LAIR1-RR: TCAGGGAGAGAACCCTCC-TGTGTG; IL6-FF: CTCTCCTAACAGATAAGC, IL6-RR: CAACATAAGTCAGATACCT; Actb-FF: AGC-CTTCCTTCTTGGGTATG; and, Actb-RR: CTTGCTGATCCACATCTGC. A typical protocol included 1 min of 95 °C incubation, followed by 35 cycles of denaturation at 95 °C, annealing as required by the primer (typically 60 °C), and elongation at 72 °C for 10 s each; specificity of amplification was confirmed by melt curve analysis which indicated that a single product has been produced by each primer pair. Negative controls in each PCR plate included a no-template control (NTC) and a no-reverse transcriptase control (NRT), which remained routinely negative throughout the study (data not shown).

Statistical analysis and microarray data

Data analysis was performed using Microsoft Excel from Microsoft Office 2010 Pro (Microsoft Corporation, Seattle, WA), GraphPad Prism version 6.0 for Windows (GraphPad Software, San Diego, CA) and Statistica 5.5 (StatSoft, Tulsa, OK). Statistical significance was accepted at p ≤ 0.05. Data are typically presented as mean ± SEM. Data presented here were either normalized average values from several repeats, or a representative experiment from at least three repeats. Data were tested for normality and dispersion to determine the choice of parametric versus non-parametric statistical criteria. To estimate significance of differences between groups, a non-parametric Mann-Witney U test, an analysis of variance (ANOVA) with a Tukey’s Honest Significant Differences for Unequal N post-hoc test, and a Student-Newman-Keuls post-hoc test were used as appropriate. Microarray data, after validation with real-time PCR for a small set of randomly selected genes (see Appendix), were extracted using RMA Express with quantile normalization, background correction, and median polish adjustment, assembled into matrices and analyzed via TIGR MeV 4.8 using significance analysis for microarrays (SAM) and ANOVA at various stringency settings (p-value threshold and/or Bonferroni adjustment). Lists of significant genes were uploaded for pathway analysis. Pathway analysis was performed using Metacore online pathway portal (GeneGo), which allows the construction of networks and maps of interacting factors based on an extensive proprietary curated database. The list of gene names was processed to evaluate any direct and indirect interactions, allowing a maximum of two intermediary factors that are not on the list (Dijkstra algorithm); the analysis was performed with default settings.

Results

Alveolar macrophage is the key cell in enhancement of inflammatory airway responses to particles in pregnancy

Transfer of AM was able to convey the phenotype of inflammatory response to TiO₂ seen as elevated PMN counts in the BAL. In Figure 1(B), Rcp:Preg/Ti were the recipients of AM from pregnant mice followed by TiO₂, and Rcp:Norm/Ti were control mice receiving AM from non-pregnant donors; both groups received the same particle exposure. Controls included recipients of both ‘pregnant’ and normal AM treated with vehicle (PBS) alone (minimal response). Finally, a group of positive controls included pregnant mice that did not receive any AM and were just exposed to particles to reproduce the earlier finding that pregnant mice respond with high neutrophil (PMN) counts (see PosCtrl:PregTi); in normal mice this particle is non- (or minimally-) inflammatory.

While recipients of control AM from non-pregnant mice produced minimal PMN counts after challenge with TiO₂ particles (0.61 [± 0.13] × 10⁵PMN/ml normalized BAL, similar to other negative controls), those who received AM from pregnant donors responded with 1.62 [± 0.19] × 10⁵PMN/ml (p < 0.05), an outcome similar to the response in the positive control group (TiO₂ exposure during pregnancy). Thus, the effect is rather robust, with over a 2.5-fold change. We hypothesized that estrogen and/or progesterone, hormones that increase steadily throughout the course of pregnancy, were the likely factors to affect uptake and clearance of particles, and the mechanism included alterations of AM production of cytokines/other signaling molecules.

Pregnancy hormones inhibit particle phagocytosis

To test the effect of pregnancy hormones on macrophage particle phagocytosis, an in vitro assay of TiO₂ particle uptake was performed wherein freshly extracted BAL AM from pregnant and control mice were incubated for 2 h with the suspension of TiO₂ particles (Figure 2A). AM from pregnant mice showed less avid uptake of the TiO₂ particles: at 2 h incubation, ≈10–20% less of the ‘pregnant’ AM were engaged in phagocytosis (Figure 2B). Side-scatter MFI values were consistent with the percent positive (data not shown). We hypothesized the main role in this inhibition was played by estrogen. To test this, murine primary AM were first treated ex vivo with high doses of 17β-estradiol overnight to mimic a pregnancy milieu. Results (Figure 2C) demonstrated that estradiol was inhibitory of particle uptake. On average, over three experiments (n = 7 per group), 70.7 (± 1.6)% of vehicle-treated cells were actively phagocytosing at 2 h, while among those treated with 10 µg 17β-estradiol/ml only for 24 h, 64.8 (± 1.4)% were active (p < 0.05). In AM from male mice, there was only a trend to uptake inhibition. Mean scatter values were consistent with the percent values.

Even though the effect was consistent and statistically significant, the magnitude of the response prompted us to seek an assay with an improved signal:noise ratio. To establish an in vitro model to study this phenomenon, the effects of pregnancy hormones on prototypical murine macrophageal cell lines RAW264.7 (genetically female) and J774 (genetically male) were investigated. Similar to primary AM, estradiol inhibited TiO₂ particle uptake in a dose-dependent manner (Figure 2D and E). In a range of doses from 1 ng/ml–10 µg/ml (36.7 µM), estradiol provided a dose-dependent inhibition, with ≈25% less cells engaged in phagocytosis after the maximum estradiol dose. Because the percent values fluctuated slightly from experiment to experiment, to assure comparability the data were normalized by dividing each percent value by naive control sample percent value. In Figure 2(D), the average normalized data from three experiments (total n in each group = 21) showed 0.76 [± 0.04] of the control value for a dose of 10 µg estrogen/ml (p < 0.05). Doses of 3 and 5 µg/ml were less effective but statistically significant; doses of ≤1 µg/ml were inconclusive. For J774 cells, the effect was similar but less prominent at 0.80 [± 0.06] of control. Estradiol did not compromise cell viability in this range of doses (Figure 2F).

Progesterone is another hormone of pregnancy produced in increasing concentrations during gestation; we therefore sought to determine its effect on particle uptake. We tested a range of doses from 1 ng/ml–1 µg/ml (∼3.1 µM) and found that progesterone alone did not produce a significant effect. However, progesterone provided a mild tendency to potentiate the inhibitory effect of estradiol when used in combination (Figure 3). Figure 3(A) shows a representative experiment aimed to test the effect of progesterone on particle uptake; only the cells treated with estradiol in addition to progesterone show significant decrease of phagocytic activity to the value of 0.85 of that seen in controls (n = 4, p < 0.05), which is consistent with other estradiol data. Another experiment in Figure 3(B) demonstrates that the effect of 5 µg estradiol/ml treatment was downward enhanced from 0.9 [± 0.03] (estradiol alone) to 0.87 [± 0.01] (with 0.1 µg/ml progesterone) to 0.8 [± 0.4] (with 1 µg/ml progesterone) (n = 4 each); however, this trend was not statistically significant. Similarly, Figure 3(C) shows a summary (mean values) from three experiments and suggests that progesterone provided a reproducible trend towards potentiating the effect of estradiol. Thus, estradiol was necessary and sufficient for the effect; the potentiating effect of progesterone was minor.

Figure 3. Progesterone potentiates the effect of estradiol. While progesterone alone had minimal or no effect, it potentiated the action of estradiol. (A) RAW cells were pre-incubated with 0.1 or 1 µg/ml progesterone, with the positive control of 5 µg/ml estradiol +0.1 µg/ml progesterone, which was the only exposure that inhibited the uptake. (B) Cells were incubated with 5 µg/ml of estradiol with or without progesterone. (C) Average values from three experiments—comprising data in (A), (B), and data not shown elsewhere—using various combinations of estradiol (5 or 10 µg/ml) and progesterone (0.1 or 1 µg/ml). *p < 0.05 versus vehicle control. Mean ± SEM.

Microarray profiling

Estradiol induced significant transcriptional change in RAW cells: analysis of genomic data in TIGR MeV via ANOVA at a p value threshold of 0.05 and with Bonferroni adjustment identified 1141 transcripts as significant. In contrast, the changes in pregnancy versus control AM were relatively small and, to obtain a similar number of genes with significantly different expression, the criteria were relaxed by removing the Bonferroni adjustment and Welch approximation; this produced a list of 1765 significantly changed transcripts. These lists had 78 overlapping transcripts; of these, only 43 had concordant direction of change (i.e. up-regulated by estradiol in RAW and by pregnancy in the AM, or down-regulated in both) (Figure 4A). In addition to ANOVA, SAM analysis was performed; this typically is more stringent but may produce slightly different results; SAM identified five other transcripts significantly up-regulated by pregnancy in the AM: they are of interest even though their change in RAW cells was not significant, hence we added them to the list (resultant cluster is shown in Figure 4B). Pathway analysis of this list indicated that, with the exception of 10 genes, the others have known interactions in a Dijkstra network, suggesting they can affect each other via ≤2 intermediaries (Figure 4C).

Figure 4. Estradiol effect on gene expression profiles of RAW cells had overlapping effects with effect of pregnancy on AM expression profiles. RNA samples from RAW cells treated with 1 µg estradiol/ml and AM from pregnant and control mice were profiled using Affymetrix 430A 2.0 gene chip microarrays. Robust Multichip Average (RMA) values were compared pair-wise (control versus estrogen and control versus pregnancy), and the lists of statistically significant changes were overlapped. After removal of discordantly-changed genes (elevated by estradiol but inhibited in pregnancy and v/v) the (A) gene list was subjected to (B) hierarchical cluster analysis (HCL). (C) Pathway analysis of the list in (A) using GeneGo MetaCore indicates that many but not all genes interact in a “shortest paths” network over <2 intermediaries. Gene symbols circled with a large blue ring come from the original dataset; others are intermediaries known to interact with these but were not represented in the original gene list. Thin and thick lines demonstrate various types of interactions. The network is stratified by cellular localization.

Discussion

This study investigated the mechanism of aberrant innate responses to TiO₂ particles in pregnancy using a single exposure mouse model. We have earlier observed that normally “inert” environmental particulates are no longer innocuous in the pregnant airway. Instead, they lead to enhanced airway inflammation associated with increased asthma risk in the offspring (Fedulov et al., 2008). Hence, while pregnancy is a normal condition for the female organism, it is related to enhanced sensitivity to disease and changes in immune or inflammatory responsiveness. We hypothesized a previously unknown role of pregnancy hormones in altering macrophageal responses to innocuous particulates, specifically their uptake and clearance. Estrogen and progesterone are essential for implantation and successful maintenance of pregnancy, and are produced in increasing concentrations (Albrecht et al., 2000; Arredondo & Noble, 2006; Cahill, 1995). These hormones are, therefore, the first most likely candidates for pregnancy-specific immune regulation issues. Prior studies of the effects of sex steroids on phagocytosis demonstrated that, depending on the dose, tissue, context, and cellular target, the influence may be both activating (e.g. enhanced uptake of opsonized sheep erythrocytes (Chao et al., 1994, 1996) and cytokine production (Chao et al., 1995, 2000) and inhibitory (e.g. stimulated phagocytosis of carbon particles, but impaired protection against bacteria (Salem et al., 1999)). In general, while estrogen plays controversial roles in inflammation, its effects in a specific model cannot be readily predicted (Straub, 2007).

This report demonstrated for the first time the inhibitory effect of pregnancy on AM uptake of fine-size TiO₂ particles; we were able to narrow down the effect to 17β-estradiol which led to inhibition of uptake in primary AM and in RAW264.7 and J774 cell lines. We note that in vitro doses of hormones needed to observe the effect seen in ex vivo ‘pregnant’ AM and RAW cells were relatively high. In normal mice, the concentration of circulating estradiol in serum is measured as pg/ml (Wood et al., 2007); however, during pregnancy, this level can elevate into the ≥1–2 ng/ml range (Zhang et al., 1999). Moreover, we presume that 24 h incubation cannot as easily mimic an E14 gestation status where cells resident in tissues are constantly exposed to elevating hormone doses. Moreover, the exact concentration of estradiol in airway mucosa and within the lung structures is unknown; however, historical data suggest that tissue levels of steroid hormones may exceed plasma levels by 20–30-fold (Akerlund et al., 1981; Batra, 1976; Straub, 2007).

It is also worth noting that in vitro bioavailability of the hormone from a single administration cannot be directly interpolated dose-wise to the in vivo tissue exposure of the cells. Interestingly, AM from pregnant mice untreated showed similar inhibition of uptake versus normal control AM, as that caused by estradiol treatments. Even though estradiol levels used in vitro here may be higher than those that reach AM in vivo, such exaggeration of stimulant is not uncommon in experimental models and should not detract from the mechanistic data reported here. For instance, Drew & Chavis (2000) reported inhibitory effects of estradiol and progesterone on microglial macrophages at concentrations of up to 100 µM, a level 3-times the highest levels used in the current study. Of note also is the potentiating effect of progesterone, which was not effective alone; thus, in vivo in tissues, relatively lower doses may cause the effect due to such interactions with other factors. In this interaction, a potential role may be played by a known ability of estrogen to upregulate progesterone receptor expression (Ing & Tornesi, 1997), explaining why estradiol was needed for progesterone to be effective. While high doses of estrogen are known to induce cellular toxicity, in this study no increase in levels of necrotic or apoptotic markers were noted, even though this was not a comprehensive toxicity analysis.

The fine-sized TiO₂ particles used here have minimal reactivity, do not have soluble components, and have historically been used as a negative control agent in studies of toxic particulates of environmental contaminants; however, they are not completely innocuous as shown by our laboratories (Fedulov et al., 2008) and by others. Specifically, evidence of inflammation in response to some forms of TiO₂ has been seen in a small number of studies; thus, they are not completely innocuous (Gilmour et al., 1989). Moreover, in rare reports, TiO₂ particles could promote pulmonary inflammation, e.g. when specially coated (Warheit et al., 2005). To add, TiO₂ particles were shown to induce pulmonary inflammation with activation of antigen-presenting cells (APC) and production of certain chemokines (Drumm et al., 1999; Renwick et al., 2004) and cytokines (Ahn et al., 2005; Kang et al., 2005).

TiO₂ is produced and used in the workplace in varying particle size fractions, including fine and ultrafine. Moreover, outside of the production aerosol exposure, TiO₂ is present in many food, personal care, and other consumer products; thus, virtually the entire population in the US and other developed countries is exposed to the substance, albeit not in an airborne way (Weir et al., 2012; Windler et al., 2012) but often in the form of suspension (e.g. when released from textiles during washing (Windler et al., 2012)). This emphasizes the importance of studying potential deleterious effects of environmental substances we consider inert – more so because not only can such environmental pollutants enhance the risk of disease (Gilmour et al., 2001), but they can become a single cause of it.

The AM is primarily responsible for the binding, ingestion, and, ultimately, clearance of inhaled macromolecules, particles, and pathogens that reach the lower respiratory tract (Brain, 1985; Bowden, 1987; Lohmann-Matthes et al., 1994; Thepen et al., 1994). The receptors on AM that mediate binding of unopsonized particles were, until recently, not known. A few studies suggested a role for the scavenger receptor (SR) with collagenous structure MARCO (macrophage receptor with collagenous structure) and SR-A in this process (Arredouani et al., 2004, 2005). Hence, we hypothesized that scavenger receptor expression could have been impaired by the hormones via an unknown mechanism. However, expression of MARCO and SR-A remained unchanged by the estrogen (data not shown due to space constraints).

We hypothesized, on one hand, that, under the influence of pregnancy hormones, the AM produce and release inhibitory signaling molecules that suppress normal particle clearance in the recipients. This thought is based on two main findings, that (a) in the adoptive transfer experiments the amount of AM per recipient is relatively small (compared to the numbers of AM in the lung), and (b) estrogen exposure leads to ‘defective’ particle uptake. We postulate that the effect is probably not due to the functional deficiency of the transferred AM per se, since 1 million cells cannot replace the AM in a recipient lung, but rather to the release of an inhibitory factor of factors, e.g. a cytokine or another signaling molecule with the ability to suppress phagocytosis. To identify the likely candidates, we performed microarray analysis of both the primary AM from pregnant and control mice and the RAW cells, control, or stimulated with estradiol. Bioinformatic analysis was aimed to identify the genes upregulated after estrogen similarly in both the AM and RAW cells, based on the premise that the estradiol effect seen in vitro is also essential to the effect in vivo. A small group of genes presented here that satisfies these criteria is likely to include this inhibitor(s). While we identified a list of prospective ‘candidate’ inhibitors, the experimental evidence for their role remains beyond the scope of this manuscript (e.g. direct testing cannot yet be performed due to the absence of recombinant proteins) and will become the subject of future studies.

In combination with our prior findings (Fedulov et al., 2008) demonstrating prolonged inflammatory response to PMN in the lung of a pregnant host, we postulate that the deficient particle uptake leads to prolonged persistence of particles in the airways of ‘pregnant’ mice that, in turn, sustains the recruitment of PMN and other inflammatory events. Macrophage production of the main PMN chemoattractant KC was evaluated and found to not change with pregnancy or estrogen in either AM or RAW cells (data not shown), an outcome consistent with this paradigm and suggesting that KC cannot be a potential therapeutic target. The expression of scavenger receptors MARCO and SRA was also tested; it was seen that estradiol exposure did not affect these molecules, suggesting that particle binding was unimpaired and could not be manipulated for therapeutic purposes. As evident from the unchanged expression of markers Fizz1, Arg1, and YM1 in the RAW cell model, estrogen did not induce the phenomenon of alternative activation of macrophages (data not shown).

In summary, this study demonstrated that a macrophage response to TiO₂ particles was impaired by estrogen and expressed as a less efficient uptake. This, we postulate, would prolong the persistence of the particles in the airways after an exposure and lead to enhanced prolonged inflammation during pregnancy. The ability of these deficient AM to carry the phenotype to recipients suggested to us that they alter the recipients’ local milieu via as yet unidentified inhibitory signaling factors.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper. This research was supported by NIEHS grant ES015425 to AVF.

Acknowledgements

We thank Dr Zhiping Yang from Harvard School of Public Health for sharing helpful methodological suggestions and research insights.

YZ and LM performed cell culture experiments including phagocytosis assays and flow cytometry. LK provided expertise on phagocytosis assay and microarray analysis and co-authored the manuscript. AF is the principle investigator, designed experiments, performed adoptive transfers and microarray data analysis.

Appendix

Real-time PCR validation of microarray results. Five genes were randomly chosen to be validated by real-time PCR from the samples of RNA from RAW cells treated with 1 µg/ml estradiol or control cells. The data are expressed as fold-change in estradiol over control (Figure A1).

Figure A1. Microarray validation. Three samples per group of RAW cell RNA were tested for each of five genes to validate the results of the microarray analysis. Data are expressed as fold-change in 1 µg/ml estradiol-stimulated cells over control untreated cells (same for microarray and for PCR). All data are presented as mean ± SEM.