Sex-differences in resident immune cell phenotype underlies more efficient acute inflammatory responses in female mice

Abstract

Females are protected against mortality arising from severe sepsis. The precise mechanisms that confer this survival advantage in females over males are unclear. Resident leukocytes in resting tissues have a significant influence on circulating cytokine levels and recruitment of blood leukocytes during acute inflammatory responses. Whether the phenotype of resident leukocytes is distinct in females is unknown. Herein we show that the numbers of leukocytes occupying the naive peritoneal and pleural cavities is higher in female than in male mice and rats, comprising more T- and B-lymphocytes as well as macrophages. The altered immune cell composition of the female peritoneum is controlled by elevated tissue chemokine expression. Female resident macrophages also exhibit greater Toll-like receptor expression, as well as enhanced phagocytosis and NADPHoxidase-mediated bacterial killing. However, macrophage-derived cytokine production is diminished by proportionally more resident immunomodulatory CD4+ T-lymphocytes. Ovarian hormones regulate macrophage phenotype, function, and numbers but have no significant impact on T-lymphocyte populations in females. Thus we have identified a fundamental sex-difference in phenotype of resident leukocytes. We propose that the distinct resident leukocyte population in females allows aggressive recognition and elimination of diverse infectious stimuli without recruitment of circulating neutrophils or excessive cytokine production.

Introduction

The severity and incidence of innate immune conditions such as sepsis¹ and post-surgery infections are profoundly less in women compared to age-matched men. This sex-difference is evident in multiple species such that exposure to a wide range of stimuli (including bacteria, viruses, parasites, fungi, or vascular trauma) results in reduced severity and minimal loss of tissue function in females compared to males (for review see2). The clinical consequences of this sexual dimorphism may extend beyond an increase in survival in women and have important implications for treatment of inflammatory disorders in women. Indeed, recent evidence implies a lack of efficacy of first-line anti-inflammatory drug treatments (including aspirin³ and statins⁴) in women; supporting the notion that innate immune responses may be inherently distinct in females. Therefore, understanding the nature of these differential innate immune responses in males and females is essential for identifying novel strategies to appropriately target inflammatory disorders. In addition, it will determine that human clinical trials as well as experimental animal studies are designed with sex-differences in mind such that the nature and progression of immune responses to infection and injury as well as responsiveness to anti-inflammatory drug treatment are very much sex-specific.

The detrimental effects of acute infections are mediated, in part, by the mobilization and subsequent infiltration of leukocytes into tissues together with excessive production of cytokines, such as TNFα and IL6. The mechanisms that bestow protection from infection in females are assumed to be mediated by female sex hormones, in particular 17β-estradiol, which can directly influence synthesis and signal transduction of multiple cytokines in vitro (for review see5). However, in vivo studies on estrogens have given conflicting results partly due to the multiple actions of estrogens on several different cell types and limitations of experimental models with doses of sex hormones that do not fully reflect the biological differences between the sexes. Indeed, 17β-estradiol treatment can paradoxically increase the severity of experimental sepsis⁶ as well as precipitate fatal inflammatory cardiovascular disorders^7,8 in women. Thus, whilst it is clear that estrogens can modulate several pro-inflammatory pathways, many fundamental aspects of the nature of sex-differences in acute inflammatory responses remain undefined. In particular, whether inherent differences exist in the regulation of trafficking of blood leukocytes in males and females is not known. In the current study we sought to determine the principle differences that endow females with a more efficient innate immune system. Diverging from the common approach of primarily focusing on the effects of estrogens, we have directly examined the mechanisms that regulate inflammatory cell recruitment and cytokine synthesis in age-matched females and males.

Under resting conditions tissues are populated by resident leukocytes, including macrophages, which provide basal immune surveillance necessary to mount rapid, controlled inflammatory responses to infection or injury. Pathogens, as well as components released from injured cells, are sensed by tissue macrophage using a repertoire of receptors including Toll-like receptors (TLRs)⁹ that induce the release of several cytokines (e.g. TNFα) and chemokines (e.g. CCL2). The net result is the recruitment of circulating phagocytes into inflamed tissues^10,11. Therefore resident leukocytes represent the frontline of innate/non-specific host defense against infection and injury but whether these sentinel cells are regulated in a distinct manner in females is not known. We hypothesized that the population of blood leukocytes that reside in resting tissues, such as found in the peritoneal and pleural cavities, have a distinct phenotype compared to male leukocytes and that this difference enables female tissues to mount a more robust and efficient response to subsequent inflammatory insult.

We found that the tissue resident leukocyte populations in female mice and rats are more numerous and have a greater density of pathogen/injury-sensing TLRs compared to males. Our findings demonstrate that this population of cells in females is more adept at sensing and eliminating pathogens but that cytokine synthesis is kept in check by the increased presence of immunomodulatory CD4+ T lymphocytes. The fundamental nature of this difference provides, for the first time, a unifying mechanism that accounts for why females are more efficient at responding to the multiple diverse stimuli that converge onto TLR pathways and suggests that reported sex-differences in down-stream effectors (e.g. PI3kinase, p38, NFκB) are likely to be a consequence of differential activation of TLRs in females. Importantly, our study also highlights the inherent differences in tissue and immune cell phenotype between males and females and supports the recent calls for the consideration of these sex-differences in biomedical research and drug development^12,13.

Materials and Methods

Animals

Experiments were conducted on age-matched (8-10 week) male and female C57BL/6 or Rag2-/-mice (Charles River) and Wistar rats (255-275g; Charles River). All animals were housed in pathogen-free individually ventilated cages and comparisons are made between animals treated on the same day and samples that were processed at the same time. To investigate the impact of ovarian sex hormones, female C57BL/6 mice were either ovariectomized (OVX) or sham-operated (Sham) at 4 weeks of age and allowed to recover for 4-5 weeks. Plasma 17β-estradiol was measured in Sham and OVX mice by commercially available EIA (Cayman Chemical Company). All experiments were approved under a Project Licence (Animals Scientific Procedures Act 1986) issued by the Home Office (UK) and conducted according to local guidelines. The characteristics of mice used in this study are shown in Table 1.

Table 1. Characteristics of C57BL/6 mice used in the study.

Plasma 17β-estradiol was measured by EIA (Cayman Chemical Company). Significant differences are represented as * P<0.05, ** P<0.01 and *** P<0.001 compared to male or sham-operated female by unpaired Student’s t-test. NA denotes parameters that were not measured.

	Male	Female	Sham	OVX
number	33	34	24	22
Body mass (g)	26.3 ± 0.46	19.9 ± 0.41***	20.5 ± 0.39	22.1 ± 0.37 **
Uterine mass (mg)	NA	NA	103 ± 12.7	18 ± 2.7 ***
Plasma 17β-estradiol (pg/ml)	NA	NA	95.2 ± 18.64	50.7 ± 5.50*

Collection of resident leukocytes and mesenteric tissue

Animals were sacrificed using CO₂ and resident peritoneal or pleural leukocytes were collected in sterile phenol red-free DMEM containing 10% fetal bovine serum, and counted by hemocytometer. Leukocyte pellets were either snap frozen for RNA quantification or prepared for flow cytometry (see below). The entire mesenteric vascular bed was collected by separating the mesentery from the intestinal wall and snap freezing for RNA analysis.

Flow cytometry

Flow cytometry (FACS) was carried out on a Becton Dickinson Facscalibur with data analyzed by CellQuest™. Leukocytes were incubated for 30 minutes at 4°C with antibodies to either F4/80 (eBioscience, clone BM8), murine CD3 (Serotec, clone KT3), CD19 (Serotec, clone 6C5), CD8 (Serotec, clone YTS169.4), CD4 (Serotec, clone YTS191.1), CD25 (Serotec, clone PC61.5.3), γ^δ cells (gift from Dr T. Hussell, Kennedy Institute, London, U.K), GR1 (BD PharMingen, clone RB6-8C5), TLR2 (eBioscience, clone 6C2), TLR4 (eBioscience, clone UT41), rat granulocytes (BD PharMingen, clone RP1), ED1 (BD PharMingen, clone 1C7), rat CD3 (BD PharMingen, clone 1F4) or CD45RA (BD PharMingen, clone OX-33) using respective isotype antibodies as controls and compensated as appropriate for dual labeling. Cell proliferation was determined by incorporation of Bromodioxyuridine (BrdU; 1mg, ip, BD PharMingen) injected 2h prior to collection of peritoneal cells and quantified by FACS. Apoptotic cells were identified as annexin V^+ve/propidium iodide^-ve, using an apoptosis assay (BD PharMingen).

Quantification of chemokines and cytokines

Total RNA was extracted from peritoneal cell pellets and mesenteric tissue (NucleoSpin, Macherey-Nagel, Germany), reverse transcribed (using Mouse Moloney Leukaemia Virus reverse transcriptase) and 20ng cDNA submitted to quantitative real-time PCR (Applied Biosystems 7900HT) and quantified using SYBR® green (for primer sequences see Table 2). For each sample, RNA levels of the target gene were normalized to expression of the house-keeping gene for 18S and calculated as fold expression relative to mean values of the control group, as indicated in figure legends. Cytokines in cell-free peritoneal washouts or cell culture supernatants were measured by enzyme-linked immunosorbent assay (ELISA) according to the manufacturer's instructions (eBioscience; TNFα, IL6, IL10, TGFβ and R&D systems; CCL2).

Table 2. Sequence of primers used for real-time quantitative PCR with SYBR green.

	Forward	Reverse
Chemokine Receptors
CX3CR1	GGAGACTGGAGCCAACAGAG	CCTGATCCAGGGAATGCTAA
CCR1	TGCAGGTGACTGAGGTGATTG	TGAAACAGCTGCCGAAGGTA
CCR2	GGAAGACAATAATATGTTACCTCAGTT	TGGTGGCCCCTTCATCAA
CCR5	CGAAAACACATGGTCAAACG	TTCCTACTCCCAAGCTGCAT
CXCR1	CCCGATCCGTCATGGATGTC	CACAGGGGTGTGGCCAAAAATC
CXCR2	ATGCCCTCTATTCTGCCAGAT	GTGCTCCGGTTGTATAAGATGAC
CXCR3	CTCTTTGCCCTCCCAGATTTC	GGCATAGCAGTAGGCCATGA
CXCR4	TCCAACAAGGAACCCTGCTTC	TTGCCGACTATGCCAGTCAAG
Chemokines
CX3CL1/fractalkine	TCCTGGAGACGACACAGCA	TGCCACCATTTTTAGTGAGGG
CCL2/MCP1	TTAAAAACCTGGATCGGAACCAA	GCATTAGCTTCAGATTTACGGGT
CCL5/RANTES	GCTGCTTTGCCTACCTCTCC	TCGAGTGACAAACACGACTGC
CCL7/MCP3	GGATCTCTGCCACGCTTCTGT	ACTTCCATGCCCTTCTTTGTCTTG
CXCL2/KC	TGAGCTGCGCTGTCAGTGCCT	AGAAGCCAGCGTTCACCAGA
CXCL5/LIX	GCATTTCTGTTGCTGTTCACGCTG	CCTCCTTCTGGTTTTTCAGTTTAGC
CXCL12/SDF1α	TGCATCAGTGACGGTAAACCA	TTCTTCAGCCGTGCAACAATC
Toll-like receptors
TLR2	GCAAACGCTGTTCTGCTCAG	AGGCGTCTCCCTCTATTGTATT
TLR3	GTGAGATACAACGTAGCTGACTG	TCCTGCATCCAAGATAGCAAGT
TLR4	ATGGCATGGCTTACACCACC	GAGGCCAATTTTGTCTCCACA
TLR6	TGAGCCAAGAACAGAAAACCCA	GGGACATGAGTAAGGTTCCTGTT
Myd88	AGGACAAACGCCGGAACTTTT	GCCGATAGTCTGTCTGTTCTAGT
Rat TLR2	CTCCTGTGAACTCCTGTCCTT	AGCTGTCTGGCCAGTCAAC
Rat TLR4	CTGGGTTTCTGCTGTGGACA	AGGTTAGAAGCCTCGTGCTCC
Reference gene
18S	AGCCTGCGGCTTAATTTGAC	CAACTAAGAACGGCCATGCA

Induction of experimental peritonitis and pleurisy

Peritonitis was induced by intraperitoneal injection of zymosan A (1mg) or group B streptococci (GBS, 30 x10⁶ per mouse). The clinical GBS isolate, NCTC10/84 (serotype V) was grown in Todd Hewitt Broth without agitation at 37°C to an OD₆₀₀ of 0.4, equivalent to 10⁸ cfu/mL. Bacteria were collected by centrifugation and washed with sterile PBS. Mice were inoculated intraperitoneally with 30 x10⁶ cfu in 300µl PBS. GBS-induced sepsis was scored at 3h. A score of 1 was given for ruffled fur, 2 for huddled but active, 3 for inactive, 4 for inactive when handled and 5 for moribund, as previously described⁶. Pleurisy was induced in male and female Wistar rats by injection of 0.15ml of 1% carrageenan (w/v) into the pleural cavity. Peritoneal and pleural leukocytes were collected at 3h by lavage of the cavity with PBS containing 0.3% citrate (w/v).

Bacterial Survival

Blood samples from GBS-treated mice were collected into heparin from the tail vein. To determine streptococci survival, the number of colony forming units (cfu) was determined for 3 dilutions of whole blood after overnight incubation at 37°C on agar plates. The microbicidal capacity of normal mouse plasma was assessed by determining CFU following incubation of plasma from male and female C57BL/6 mice in a 48-well plate with 10⁴ GBS/well for 1h at 37°C.

Assessment of phagocytosis and antibacterial NADPH oxidase activity

Phagocytosis of zymosan (5x10⁵ particles, 30min) by murine resident peritoneal leukocytes (5x10⁴ cells/sample) was assessed using a colorimetric assay (Cell Biolabs Inc, USA). Intracellular anti-bacterial NADPHoxidase activity of murine resident macrophage (10⁵ cells/sample) was determined by kinetic assay using Amplex® Red (Invitrogen). The rate of increase in fluorescence was calculated over 7 minutes in the absence or presence of phorbol myristate acetate (PMA; 1pg/ml) and normalized to total protein content.

In vitro stimulation of male and female peritoneal leukocytes

Peritoneal leukocytes (2x10⁵) were plated in 24-well plates and stimulated with either TLR2 ligand Pam3CysSerLys4 (Pam3CSK4, 0.1µg/ml, InvivoGen) or TLR4-specific lipopolysaccharide (LPS; Ultra Pure E. Coli LPS, 0.1µg/ml, InvivoGen) for 3h. To obtain pure macrophage, peritoneal cells were depleted of CD19+ve B-lymphocytes by Macs© (Miltenyi Biotec.) separation prior to plating and the residual lymphocytes removed by washing with PBS. This procedure yielded a population consisting of 95-98% F4/80+ cells and no CD11c+ dendritic cells. CD4+ T cells were isolated from splenocytes by negative selection using Macs© CD4 isolation kit (Miltenyi Biotec.). CD4+ T cells (1.5x10⁵) were incubated with resident peritoneal macrophage (1.5x10⁵) in a 48-well plate for 2h prior to stimulation with LPS (Salmonella typhi, 0.1μg/ml, 18h, Sigma).

Statistical Analysis

Data are expressed as mean ± sem. Comparisons between two groups were made by two-tailed unpaired t test. For comparisons between multiple groups a one-way ANOVA was performed followed by Bonferroni’s post-test. Differences between time-response curves were assessed by two-way ANOVA. Statistical analysis was performed using Prism (GraphPad Software, Inc.).

Results

Fundamental sex-differences in cellular composition of naive peritoneal and pleural cavities

In male mice, we found that total resident peritoneal leukocytes numbered 16 ±1.7x10⁵ per cavity while females had 31±2.3x10⁵ cells per cavity (n=13, P<0.001, Fig. 1A). This sex-difference was not confined to the peritoneum as total cell numbers in the female pleural cavity was also double that of males (Fig.1A). Similar sex-differences were also observed in Wistar rats, where the resting peritoneal and pleural cavities of female Wistar rats also contained more total leukocytes than in age-matched males (Fig.S1A). FACS analysis revealed that this increased number of cells in females comprised significantly greater numbers of macrophages, T-lymphocytes and B-lymphocytes compared to males (Figs.1B & S1B-C). Whilst the total number of each leukocyte subset was greater in females, the proportion of CD3+ T-lymphocytes was also greater in females of both species (Fig.1C & S1D-E). For example, in mice, the ratio of peritoneal macrophage/T-lymphocytes/B-lymphocytes was 1: 0.7: 1.5 in males and 1: 1.2: 1.5 in females. Furthermore, of this T-lymphocyte population the total number of CD4+ and CD8+ cells was significantly greater in females with little discernible sex-difference in basal levels of CD4+/CD25+ or gamma/delta T-lymphocytes in the resting peritoneal cavity (Fig.1D).

Figure 1. Distinct resident leukocyte population in the female peritoneal cavity.

(A) Increased total resident cell number in peritoneal (n=13 mice; 3 independent groups) and pleural (n=5 mice) cavities of female compared to male mice. (B) Total cell number and (C) percentage of F4/80+ macrophage, CD3+ T-lymphocytes, CD19+ B-lymphocytes and GR1+ granulocytes in peritoneal cavity of male and female mice were determined by flow cytometry (n=7 mice). (D) Increased total resident CD8+ and CD4+ T-lymphocytes but not CD4+/CD25+ T-regulatory or δγ T-lymphocytes in female peritoneal cavity (n=4 mice). All values are expressed as mean ± sem. All comparisons are relative to male. *P<0.05, **P<0.01 and ***P<0.001 by Student’s t-test.

Increased homeostatic recruitment of leukocytes into the female peritoneal cavity

To understand why the composition of the female naive peritoneum is different to that of males, levels of chemokines central to monocyte and lymphocyte trafficking were measured in unstimulated mesenteric tissues of male and female mice. Female tissues expressed significantly higher mRNA levels of CX3CL1/fractalkine (chemoattractive for CX3CR1+ monocytes, T-lymphocytes, dendritic and natural killer cells), CCL2/MCP1 (chemoattractive for CCR2+ monocytes), CXCL12/SDF1α (chemoattractive for CXCR4+ lymphocytes) and CCL5/RANTES (chemoattractive for lymphocytes, dendritic and natural killer cells expressing CCR1, CCR2 or CCR5) compared to males (Fig.2A). In addition to increased expression of tissue chemokines, female leukocytes in the peritoneal cavity also had elevated expression of chemokine receptors. These data demonstrate that on a cell-for-cell basis female resident leukocytes selectively express more chemokine receptor CCR1 (receptor for CCL5), CCR2 (receptor for CCL2) and CXCR4 (receptor for CXCL12) but relatively less chemokine receptor CX3CR1 (receptor for CX3CL1). No mRNA expression of CXCL5/LIX (chemoattractant for CXCR2+ neutrophils) was detected in mesenteric tissue of either sex (Fig. 2A). To assess whether the rate of cell turnover was responsible for the differential leukocyte numbers in males and females we measured the incorporation of BrdU or the expression of apoptotic indices (Annexin V⁺/PI^-) in resident peritoneal leukocytes. BrdU incorporation was low (<0.4%) in both sexes but was significantly less in females (0.31±0.04% and 0.15±0.04%, respectively, P<0.05, n=6) whereas the proportion of resident leukocytes undergoing spontaneous apoptosis was similar in males and females (8.8±1% and 9.0±0.8%, respectively, n=6).

Figure 2. Increased homeostatic leukocyte recruitment into female peritoneal cavity.

(A) Basal mRNA expression of chemokines CX3CL1, CCL2, CCL7, CXCL12, CXCL5 and CCL5 in mesenteric tissue (n=6 mice). (B) Chemokine receptor mRNA expression in resident peritoneal cells (n=4-6 mice). Levels of mRNA for each sample are normalized to corresponding mRNA levels of housekeeping gene for small 18S and calculated as fold expression relative to the mean value in males, except CX3CR1 (relative to female). All values are expressed as mean ± sem. *P<0.05 and **P<0.01 by Student’s t-test. ND denotes chemokine expression that was not detected within 35 PCR cycles.

Increased expression of TLRs and elevated phagocytosis by female macrophage

TLRs interact with conserved structures in pathogens and have a critical role in host defense⁹. Female peritoneal leukocytes of mice and rats expressed significantly higher mRNA levels of TLRs including TLR2 (stimulated by zymosan, gram positive bacteria, heat shock proteins), TLR3 (stimulated by viruses), TLR4 (stimulated by gram negative bacteria, fibronectin, hyaluronan, oxidized LDL, heparan sulfate, heat shock proteins) and Myd88 (activated by IL1 receptor and most TLRs except TLR3) compared to males but not TLR6 (forms heterodimer with TLR2) (Fig. 3A & S1F). Cell surface expression of murine TLR2 and TLR4 protein was predominantly expressed on F4/80+ macrophages with little TLR expression evident on lymphocytes (Fig.S2A-B). Whilst the total proportion of macrophage expressing TLRs was similar between the sexes (>90%), on a cell-for-cell basis female macrophage had significantly greater TLR2 and TLR4 expression (Fig. 3B).

Figure 3. Elevated pathogen-sensing and phagocytosis by female macrophage.

(A) Basal mRNA expression of Toll-like receptors and Myd88 in naive peritoneal cells (n=5-6 mice) and (B) Flow cytometry analysis of surface TLR2 and TLR4 protein expression on resident F4/80+ peritoneal macrophage (n=6-8; 2 independent experiments). (C) Phagocytosis of zymosan A (5x10⁶ particles/10⁵ cells, 30min) by equivalent numbers of resident peritoneal leukocytes (macrophage and lymphocytes), measured in vitro by a colorimetric assay (n=5 mice). Basal levels of TLR mRNA in (D) mesenteric tissue and (E) aortae of male and female mice (n=6 mice). Levels of mRNA for each sample are normalized to corresponding mRNA levels of housekeeping gene for small 18S and calculated as fold expression relative to the mean value in females. All results are shown as mean ± sem. *P<0.05, **P<0.01 and ***P<0.001 compared to male by Student’s t-test.

As macrophage TLRs promote phagocytosis through a Myd88-dependent pathway^14,15, in addition to increased cytokine synthesis and inflammatory signaling, one biological consequence of greater TLR expression on female leukocytes is efficient phagocytosis. Testing this hypothesis, we found that uptake of zymosan was significantly greater (Fig.3C) in female compared to male peritoneal leukocytes. Differential TLR expression was only evident on leukocytes as assessment of tissue TLRs in the mesentery and aorta revealed similar TLR levels in both sexes (Fig.3D-E).

Blunted acute inflammatory responses in females

To determine the impact of the sex-specific basal resident leukocyte composition of the peritoneal cavity on subsequent pathogen-stimulated inflammatory cell recruitment, we examined peritonitis severity and duration in male and female mice. Intraperitoneal injection of group B streptococcus (GBS) resulted in the accumulation of 9.5 ± 1.19 x10⁶ cells (n=7) in the male peritoneal cavity at 3h, with only 3.8 ± 0.78 x10⁶ cells (n=7) recovered from females (P<0.01, Fig. 4A). Together with dampened leukocyte influx, the sepsis severity score was also significantly less in females (Fig.4B) with fewer recoverable live bacteria in their peripheral blood compared to males (Fig.4C). Consistent with increased macrophage-dependent bacterial killing in females, intracellular anti-bacterial NADPH oxidase activity was significantly elevated in female resident peritoneal macrophage (Fig.4D) whereas the microbicidal activity of female plasma was not greater than that of males (Fig.4E). Whilst levels of IL6 in the peritoneal washout of GBS-treated mice were lower in females (Fig.4F), production of other cytokines (including CCL2/MCP1, IL10, and TGFβ) were not directly correlated with severity of sepsis or neutrophil recruitment as there were no discernible differences in peritoneal inflammatory cytokine levels between both groups (Fig.4F).

Figure 4. Reduced severity and neutrophil recruitment in peritonitis in females.

(A-C & F) Male and female mice were treated with Group B Streptococcus (GBS; 30x10⁶ bacteria per mouse, ip; n=7 mice) for 3h. (A) Total cell number recovered from the peritoneal cavity, (B) sepsis severity score, and (C) whole blood bacterial count. (D) Phorbol ester PMA (1pg/ml)-induced NADPHoxidase activity in male and female resident peritoneal macrophage (10⁵ cells/sample), measured in vitro by Amplex® Red over 7 mins (n=3 mice). (E) GBS levels following incubation in vitro (10⁴ bacteria/sample) for 1h at 37°C with normal mouse plasma (n=3 samples from 6 mice in each group). (F) Concentration of GBS-induced cytokines in cell-free peritoneal lavage (n=7 mice). (G&H) Male and female mice were injected with zymosan A (1mg, ip). (G) Total peritoneal cell number (n=5-10; 2 independent experiments) and (H) number of F4/80+ macrophage, CD3+ T-lymphocytes, CD19+ B-lymphocytes and GR1+ granulocytes in peritoneal cavity of male and female mice 3h after injection of zymosan A (n=6 mice). All values (A-H) are expressed as mean ± sem. All comparisons are relative to male. *P<0.05, **P<0.01 and ***P<0.001 by Student’s t-test. § P<0.05 by two-way ANOVA followed by Bonferroni post-test; #P<0.001.

Due to the lytic nature of GBS we used 1mg zymosan (which triggers a resolving peritonitis¹⁶) in mice and carrageenan-induced rat pleurisy to discern precise leukocyte subtypes recruited to inflamed female tissues. Similar to GBS, zymosan- and carrageenan-stimulated inflammatory cell recruitment was significantly dampened in females compared to males (Fig.4G & S1G). This difference in cell number in both peritonitis and pleurisy in females was accounted for by reduced trafficking of neutrophils into the cavity at the onset phase (3h) of the response (Fig.4H & S1H).

Resident lymphocytes control the severity of innate inflammatory responses in females

Resident macrophages are an important source of the pro-inflammatory cytokines (e.g. TNFα, IL6) whose overproduction is responsible for detrimental effects in sepsis. Despite finding twice as many total resident peritoneal macrophage in females compared to males as well as elevated basal TLR expression on these cells, total cytokine levels in cell-free peritoneal exudates following GBS administration in vivo was similar in both sexes (Fig.4E). Similarly, equivalent numbers of total peritoneal washouts (comprising macrophages and lymphocytes) from males and females incubated ex vivo with either TLR2 ligand Pam3CysSerLys4 or TLR4-specific LPS released quantitatively equivalent amounts of pro-inflammatory TNFα, IL6, CCL2/MCP1 and anti-inflammatory IL10 and TGFβ (Fig.5A). Recent evidence indicates that T-lymphocytes can suppress splenocyte cytokine production and modulate neutrophil trafficking in vivo^17,18. Questioning whether the increased proportion of resident T-lymphocytes curb pro-inflammatory cytokine synthesis by resident macrophage, isolated macrophage were incubated with CD4+ T-lymphocytes and their ability to generate cytokine TNFα was compared to macrophages alone (Fig.5B). Adding CD4+ lymphocytes to isolated macrophage population in 1:1 ratio (as found in female peritoneal cavity, Fig.1C) significantly suppressed LPS-stimulated TNFα synthesis (Fig.5B). Similarly, GR1+ neutrophil recruitment in vivo was significantly elevated in zymosan-induced peritonitis in the absence of T-lymphocytes in lymphocyte-deficient Rag2 knockout mice (Fig.5C).

Figure 5. T-lymphocytes control the severity of innate inflammatory responses.

(A) Cytokine production in vitro by male and female resident peritoneal cells (2x10⁵ cells/sample, n=6 mice) following 3h stimulation by TLR4-specific LPS (0.1µg/ml) or TLR2 agonist Pam3CSK4 (Pam3, 0.1µg/ml). (B) TNFα production by isolated resident male peritoneal macrophage (1.5x10⁵ cells/sample, n=3 mice) treated with LPS (0.1µg/ml, 18h) in the absence or presence of CD4+ve T lymphocytes (1.5x10⁵ cells). (C) Zymosan-induced (1mg, ip, 3h) recruitment of GR1+ granulocytes into peritoneal cavity of C57BL/6 (wild type) and T-lymphocyte deficient Rag2 knockout (KO) mice (n=5 mice). All values are expressed as mean ± sem. #P<0.05 by one-way ANOVA compared to male and § P<0.05 by one-way ANOVA relative to macrophage alone. *P<0.05 by Students t-test relative to wild type.

Ovarian sex hormones contribute to sex-differences in resident immune cell population

To determine the relative impact of female ovarian sex hormones on the naive resident peritoneal cell population in female mice, ovaries from 4-week-old female mice were removed. This procedure significantly decreased plasma 17β-estradiol and prevented increases in uterine mass (Table 1). Compared to sham-operation, ovariectomy (OVX) caused a significant reduction in total resident immune cells in the peritoneal cavity due to reduced numbers of F4/80+ macrophages and CD19+ B cells but not CD3+ or CD8+ T-lymphocytes (Fig.6A). Tissue expression of chemokines CCL2/MCP1, CX3CL1/fractalkine, and CXCL12/SDF1α were also significantly suppressed in OVX females (Fig.6B) with the greatest impact on monocyte-attracting CCL2. In contrast, OVX had no effect on tissue expression of lymphocyte-attracting CCL5/RANTES (Fig.6B). The elevated expression of CCR1, CCR2, and CXCR4 in female leukocytes was also significantly suppressed by OVX (Fig.6C). Surprisingly, whilst CX3CR1 expression was found to be low in female resident leukocytes its expression was lower in OVX (Fig.6C). OVX also significantly reduced leukocyte mRNA expression of TLR2, TLR3, TLR4 and Myd88 (Fig.6D) as well as protein expression of TLR2 and TLR4 on female resident macrophage (Fig.6E) and phagocytosis of zymosan (Fig.6F). These OVX-induced changes in peritoneal leukocyte composition and phenotype resulted in increased leukocyte recruitment into the peritoneal cavity by GBS (Fig.6G).

Figure 6. Ovarian sex-hormones contribute to sex-differences in resident immune cell population.

Ovariectomy (OVX) or sham-operation was performed on female mice at 4 weeks of age and allowed to recover for 4-5 weeks. (A) Total number of resident F4/80+ macrophage, CD3+ or CD8+ T-lymphocytes, CD19+ B-lymphocytes in peritoneal cavity (n=5 mice). Basal mRNA expression of (B) mesenteric tissue chemokines (n=6-7 mice), (C) chemokine receptors on resident peritoneal leukocytes (n=6 mice), and (D) peritoneal leukocyte TLR expression (n=6 mice). (E) Surface expression of TLR2 or TLR4 on F4/80+ resident peritoneal macrophage (n=3-5 mice). (F) Phagocytosis of zymosan A (5x10⁶ particles/10⁵ cells, 30min, n=4 mice) by resident peritoneal cells in vitro. (G) Group B Streptococcus-induced (30x10⁶ bacteria/mouse, ip, 3h, n=7 mice) accumulation of leukocytes in peritoneal cavity. All values (A-G) are expressed as mean ± sem. Expression of mRNA for each sample is normalized to corresponding levels of the housekeeping gene for small 18S and calculated as fold expression relative to ovariectomized (OVX) females. All comparisons are relative to sham-operated females. *P<0.05, **P<0.01 and ***P<0.001 by Student’s t-test. NS denotes P>0.05.

Discussion

Substantial sex-differences exist in the incidence of several inflammatory disorders. Women overwhelmingly account for the majority of cases of autoimmune disease¹⁹ but are relatively protected from diseases that involve excessive or uncontrolled activation of innate immune responses including inflammatory cardiovascular diseases, and severe infections such as sepsis¹. However, women have predominantly been under-represented or excluded from clinical trials and male and female animals are often used interchangeably in experimental studies with little consideration given to differences in innate immune responses to infection/injury between the sexes. Thus despite changes in FDA regulations²⁰ and numerous reports outlining the importance of consideration of sex-differences in biomedical research^12,13, little specific information is available on mechanisms of innate immune responses in females and how they differ from responses elicited in males. In this study we demonstrate a sex-difference in the phenotype and quantity of leukocytes resident within the un-stimulated peritoneal and pleural cavities of mice and rats. Compared to males, female resident macrophages express higher levels of pathogen/injury-sensing TLRs and are more efficient at phagocytosis and bacterial killing. This increased capacity to detect and eliminate infectious stimuli is restrained by proportionally more CD4+ T-lymphocytes that limit excessive cytokine production and recruitment of tissue-damaging neutrophils.

The naive peritoneal and pleural cavities are populated by resident CD3+ T lymphocytes, B1 and B2 lymphocytes as well as macrophages, with few monocytes and neutrophils. In females, we consistently found greater numbers of total leukocytes in both the peritoneal and pleural cavities. In mice, total numbers of macrophage and B-lymphocytes in females was approximately twice that found in males. However, it transpires that T-lymphocytes (mainly CD4+ T-helper and CD8+ cytotoxic T cells) are proportionally higher in females such that the total number of this cell type in females is more than double the number in males. Our data implies that this differential cellular composition in females is governed by increased activity of tissue chemokines.

Mobilization and recruitment of blood leukocytes into tissues under homeostatic or inflammatory conditions is governed by the activity of chemokines (for review see21). Female mesenteric tissues expressed greater levels of specific chemokines that are typically chemoattractive for both monocytes/macrophages and lymphocytes i.e. CCL2/MCP1, CX3CL1/fractalkine, CXCL12/SDF1α and CCL5/RANTES. Chemokine CXCL5/LIX is a potent chemoattractant for CXCR2+ neutrophils but was not detectable in un-stimulated mesenteric tissue of either sex, consistent with a paucity of GR1+ neutrophils in the un-stimulated peritoneal cavity of both sexes. In parallel with elevated production of tissue chemokines, female resident leukocytes also had higher expression of the corresponding target chemokine receptors. Thus we suspect that the increased expression of these specific chemokine pathways by both naive mesenteric tissue and resident peritoneal immune cells underpins the fundamental difference in leukocyte composition of peritoneal cavities of female versus male mice. Moreover, our conclusion is further supported by the findings that the rate of proliferation of peritoneal leukocytes is not greater in females and that the extent of spontaneous apoptosis is similar in both sexes.

In addition to a quantitative difference in resident tissue leukocyte numbers in females, we also found differential expression of pathogen- and injury-sensing TLRs and the TLR/IL1-receptor adapter signaling molecule Myd88 on resident macrophage. TLRs are activated by numerous pathogens that harbor “pathogen-associated molecular patterns” (PAMPs) but also by host-derived “danger signals” (DAMPs) released from stressed tissue^22,23, to cause de novo cytokine/chemokine synthesis and phagocytosis^14,15. In the context of acute infection, detection and elimination of invading pathogens by TLRs is the front-line of host defense. However, prolonged or excessive TLR activity is also responsible for over-exuberant pro-inflammatory cytokine production and neutrophil recruitment²⁴ in systemic inflammatory disorders such as bacterial sepsis^25,26 where the influx of neutrophils contributes to vascular damage through the production of destructive reactive oxygen intermediates²⁷. In the current study, we show that female resident peritoneal macrophage have significantly higher total expression of TLR2, TLR3 and TLR4 that collectively recognize several diverse pathogens including various bacteria, yeast, viruses as well as injury-elicited DAMPs. The elevated expression of TLRs on tissue macrophage in females implies that these cells have a greater capacity to detect and eliminate pathogens, as confirmed by their increased capacity to engulf zymosan particles (TLR2 agonist). Indeed, exposure of peritoneal macrophage in males and females in vivo to equivalent amounts of live TLR-activating bacteria (by intraperitoneal administration of GBS) resulted in a less severe sepsis and lower blood bacteria in females. The reduced bacterial load arose from elevated intracellular anti-bacterial NADPH oxidase activity of female resident peritoneal macrophage coupled with enhanced phagocytosis and was not a consequence of increased influx of phagocytic neutrophils into the peritoneum or enhanced microbicidal factors in female plasma. In fact, in females, zymosan-induced peritonitis was accompanied by substantially less neutrophil recruitment than in males; thereby sparing female tissues from the deleterious effects of neutrophil-derived mediators.

TLRs are not restricted to cells of the immune system but have been described in the vasculature. The function of these receptors in the vasculature is not well established but one study suggests that TLRs located on the vascular endothelium can directly bind bacteria and contribute to vascular dysfunction and granulocyte recruitment during sepsis²⁸. To determine whether female blood vessels also exhibit high TLR expression, we measured TLR mRNA expression in mesenteric tissue and the aorta but no sex-difference in expression was evident in either vascular bed. Thus sex-specific up-regulation of TLR expression in females is restricted to expression on leukocytes, an effect that permits heightened sensitivity to diverse infectious agents whilst protecting the vasculature from TLR-mediated vascular dysfunction.

In contrast to phagocytosis, elevated expression of macrophage TLR in females did not correlate with elevated TLR-induced cytokine production in vivo or in vitro. Indeed, the amount of IL6 released into the peritoneal cavity following GBS challenge was suppressed in females; a finding that concurs with other reports of direct modulation of this cytokine by endogenous estrogens²⁹. As noted earlier, the proportion of resident T-lymphocytes is significantly greater in female than male peritoneal cavities of mice. Similarly, circulating CD4+ T-lymphocytes are also significantly greater in women³⁰ and other female primates³¹, indicating that this is a common feature across female species. The presence of this additional population of CD4+ T-lymphocytes in females is likely to modulate macrophage function. We found that co-incubation of murine resident peritoneal macrophage with CD4+ T-lymphocytes significantly suppressed macrophage-derived LPS-induced release of cytokine TNFα in vitro. However, this dampening of TLR-induced macrophage function does not appear to affect TLR-induced phagocytosis since zymosan uptake by female macrophage was greater than in males despite the presence of resident T-lymphocytes. Previous studies have demonstrated that T-lymphocytes can modulate neutrophil recruitment in innate immune responses¹⁷. Similarly, neutrophil influx in zymosan-peritonitis was enhanced in lymphocyte deficient mice; supporting a role for resident peritoneal T-lymphocytes in suppression of TLR-induced neutrophil recruitment. Thus the increased proportion of resident peritoneal T-lymphocytes in female mice may act as an endogenous brake on resident macrophage to control the severity of cytokine production and granulocyte recruitment in the face of elevated TLR activation and phagocytosis.

In women, menopause initiates complex biological changes that are also thought to be associated with the loss of survival benefit over men with respect to inflammatory conditions. The precise mechanisms affected by reduced ovarian function that lead to menopause-induced changes in immune responses are not clear. Therefore, we investigated whether ovarian hormones influence the differential resident leukocyte population in females and how these changes impact the subsequent response to infection. Tissue expression of monocyte-attracting chemokines CCL2 and CX3CL1, as well as CXCL12 were significantly suppressed by OVX. However, these differences in tissue chemokine levels (particularly lymphocyte/monocyte-attracting CXCL12) compared to sham-operated females were substantially less than the difference between males and females; indicating only a partial transition towards the male phenotype in OVX females. This partly reflects the fact that whilst OVX suppresses circulating levels of ovarian hormones, such as 17β-estradiol, it does not eliminate all sources of sex steroid hormones (i.e. the adrenal cortex and adipocytes). The reduction of expression of chemokines that are central to trafficking of leukocytes was accompanied by a reduction in resident macrophage and B-lymphocytes in OVX females. However, OVX had no effect on tissue expression of lymphocyte-attracting CCL5 and consequently did not affect the number of resident peritoneal T-lymphocytes. Thus, reduction of functional ovarian hormone activity in females alters basal chemokine function and thereby trafficking of monocytes and B-lymphocytes but not T-lymphocytes into healthy tissues. The elevated expression of chemokine receptors CCR1, CCR2, and CXCR4 in female leukocytes was also significantly suppressed by OVX, consistent with previous studies implicating estrogen-dependent upregulation of CCRs on CD4+ splenocytes³². However, whilst CX3CR1 expression was found to be low in female resident leukocytes its expression was further suppressed by removal of ovarian sex hormones, indicating that CX3CR1 is regulated in a distinct manner from other chemokine receptors. It is possible that endogenous testosterone levels that are high in males but reduced by OVX in females influence CX3CR1 expression. Overall, our study shows that ovarian hormones affect the specific chemokine pathways that are differentially expressed between males and females in both tissue and resident immune cells such that the composition of peritoneal cavity of OVX females was similar but not equivalent to that in males.

In addition to controlling homeostatic recruitment of leukocytes, ovarian hormones also influenced the expression and activity of macrophage TLRs. Whilst ovariectomy affects physiological levels of several steroid hormones and gonadotropins, it is likely that the hormone responsible for altered TLR expression in females is 17β-estradiol since estrogen response elements have been reported in the promoters of murine TLR2 and TLR4 genes³³. Similarly, other recent studies demonstrate elevation of TLR4 on OVX macrophage following chronic exposure to 17β-estradiol but not progesterone⁶. In line with a reduced expression of TLRs, OVX also diminished the phagocytic capacity of resident macrophage. Thus ovarian hormones have a profound impact on female macrophage population with respect to their homeostatic recruitment, TLR expression and phagocytosis but have no significant effect on T-lymphocyte populations. The change in macrophage phenotype in OVX females also affected neutrophil influx into the peritoneal cavity in bacterial peritonitis. The total number of leukocytes recruited in GBS-induced peritonitis was significantly greater in OVX females. However, consistent with a partial transition to the male phenotype, elevations in leukocyte influx were substantially less in OVX females than in males. The modest effect of OVX on GBS-peritonitis is likely to be due to the unaltered T-lymphocyte population in the peritoneal cavity that suppress cytokine production and granulocyte recruitment. Therefore, the abundant resident T-lymphocytes in females appear to have a dominant inhibitory role in modulating the magnitude of recruitment of blood leukocytes into inflamed tissues.

In summary, our results explain why females exhibit a dampened inflammatory response and suffer less tissue injury to a broad range of noxious stimuli since they possess the combined capacity of (1) heightened sensitivity to infectious and injurious stimuli (in the form of increased number of tissue macrophage with a greater density of pathogen/injury-sensing TLRs); and (2) more efficient phagocytosis and NADPH oxidase-mediated killing by resident macrophage that eliminate pathogens faster than in males; and (3) increased population of resident anti-inflammatory T-lymphocytes that selectively prevent excessive macrophage-derived cytokine production without affecting phagocytosis. Thus the mechanisms that regulate leukocyte function in females are more efficient to that in males as rapid detection and elimination of pathogens increases the threshold for pathogen-induced tissue injury in females. Ultimately, this robust response in females circumvents the need to recruit substantial numbers of neutrophils from the circulation and thus protects tissues from collateral damage incurred by neutrophil-derived mediators that contribute to tissue injury and loss of function.