Neutrophils are the first immune cells to migrate into infected tissue sites. Therefore an important step in the initiation of an immune response is the synthesis of the neutrophil-recruiting chemokines. In this in vivo study in mice, we show that resident tissue macrophages are the source of the major neutrophil chemoattractants, KC and MIP-2. Synthesis of these chemokines is rapidly regulated at the transcriptional level by signaling through TLR2, TLR3, and TLR4 that have diverse specificities for pathogens. The major and alternative TLR signaling pathways are characterized by the adaptor proteins MyD88 or TRIF, respectively. KC and MIP-2 are both produced by signaling through MyD88. However MIP-2, but not KC, is also synthesized through the TRIF adaptor protein, identifying it as a new product of this alternative pathway. Use of both pathways by TLR4 ensures maximal levels of KC and MIP-2 that lead to robust neutrophil recruitment. However the MIP-2 generated exclusively by the TRIF pathway is still sufficient to cause an influx of neutrophils. In summary we show that TLR signaling by tissue macrophages directly controls the synthesis of neutrophil-attracting chemokines that are essential for the earliest recruitment step in the innate immune response to microbial challenge.

The major stimulus for the innate arm of the immune response is the recognition of invading microorganisms and their products by TLRs (1, 2). TLRs are a family of evolutionarily conserved transmembrane receptors, sharing a common cytosolic Toll/IL-1R domain and an extracellular leucine rich repeat region responsible for recognizing specific structural motifs of various pathogens, known as pathogen-associated microbial patterns. Lipoprotein from Gram-positive bacteria are recognized by TLR2, dsRNA from viruses, by TLR3, and the well-studied example of LPS, a component of the cell walls of Gram-negative bacteria, by TLR4. TLR4 is the most complex receptor of the family because it signals through two distinct pathways initiated by MyD88/MyD88 adaptor-like protein and Toll/IL-1R domain–containing adaptor-inducing IFN-β (TRIF)5-related adaptor molecule/TRIF. Signaling through MyD88 is shared by all the TLRs, except TLR3, and activates NF-κB and MAPK leading to the production of proinflammatory chemokines such as TNF-α, IL-6, and IL-1β. Signaling through TRIF is selectively triggered by TLR3 and TLR4 ligand binding. This pathway activates NF-κB in combination with IFN-regulated factor (IRF) transcription factors leading to the production of type I IFNs as well as chemokines such as IP-10 and RANTES (3, 4). Other products of the pathway are induced indirectly via IFN-β signaling through the IFNR (5).

Neutrophil influx into sites of infection is an early defining event in an immune response (6, 7). This recruitment is vital both for direct action against microorganisms and for attracting lymphocytes able to resolve inflammation over the longer term. The murine chemokines KC (CXCL1) and MIP-2 (CXCL2) are the major chemoattractants responsible for recruiting neutrophils and both bind to chemokine receptor, CXCR2 (8). The two chemokines are closely related and, in terms of amino acid sequence, are 65% identical and 89% similar (9). They are also homologs of the human GRO chemokines that are functionally similar to the IL-8 CXC chemokine family (9). There is conflicting evidence as to the individual importance of KC and MIP-2 in neutrophil recruitment. When injected in vivo as recombinant chemokines in models of inflammation, each is reported to cause neutrophil influx (10, 11). There is evidence that KC may be the most important chemokine in the response to Aspergillus and Klebsiella lung infections (12, 13), fibrosis (14), and atherosclerosis (15). However, other studies have highlighted MIP-2 as the major chemoattractant (10, 11).

In this in vivo study we show that resident tissue macrophages are the major source of KC and MIP-2 and that these chemokines are newly synthesized products of signaling through the TLRs. Their TLR-dependent synthesis is generally regulated through the adaptor protein MyD88. However, we find that MIP-2, but not KC, is also synthesized as a direct product of the alternative pathway that uses TRIF as the adaptor protein. MIP-2 alone can recruit neutrophils, but when both KC and MIP-2 are present, the recruitment is maximal.

The following TLR agonists were used: Ultra pure LPS from Salmonella minnesota R592 (Alexis Biochemicals); LPS from Escherichia coli strain O55:B5 (Sigma-Aldrich); polyinosine-polycytidylic acid (poly(I:C)), and Pam3Cys (InvivoGen).

C57BL/6J mice were from Charles River Breeding Laboratories; C3HeN (TLR4 wild type (WT)), C3H/HeJ (TLR4 mutant) (16), and RAG2 mice were bred at Cancer Research United Kingdom, London Research Institute Animal Unit. MyD88−/− (17), TRIF−/− (3), and MyD88−/− × TRIF−/− mice were obtained from Dr. C. Reis e Sousa (Cancer Research U.K., London Research Institute, London, U.K.) and used with the permission of Dr. S. Akira (Research Institute for Microbial Diseases, Osaka University, Osaka, Japan). The experiments made use of sex-matched 8- to 12-wk-old mice and were conducted in accordance with the regulations of the U.K. Home Office.

Peritonitis was induced by i.p. injection of 10 ng of S. minnesota LPS in 500 μl. The doses were selected following titration and the sample points represent 6–10 mice. The mice were euthanized by carbon dioxide exposure, and peritoneal cavities were washed with 5 ml PBS/5 mM EDTA. Chemokines KC and MIP-2 were functionally inhibited by i.p. injection of 10 μg of KC, MIP-2, or control mAbs (R&D Systems) 15 min before induction of peritonitis. Neutrophils were labeled using 1A8-FITC (Ly-6G) and 7/4-PE (Caltag Laboratories) as previously described (18) and analyzed on a FACSCalibur (BD Biosciences). The cells were quantified on a FACSCalibur by adding a known quantity of calibration beads (CaliBRITE; BD Biosciences) to a known sample volume and the chemokines were assessed by ELISA (R&D Systems).

Peritoneal macrophages and B cells were positively selected by labeling with mAb F4/80-biotin (Caltag Laboratories) and streptavidin-MACS beads or B220 MACS beads, respectively, followed by MACS column purification (Miltenyi Biotec). To differentiate macrophages from their progenitors, bone marrow cells were harvested by flushing both femurs and tibiae of mice and cultured for 6 days in RPMI 1640 medium containing 20% of L929 cell-conditioned medium. The resulting bone marrow macrophages were identified by F4/80 epitope expression (>95% positive).

Peritoneal or bone marrow-derived macrophages in FACSWash (PBS, 0.2% BSA) were identified using F4/80-TriColor (Caltag Laboratories). For the macrophage phenotype analysis, the F4/80-positive cells were analyzed with the following Abs: Gr-1-FITC, L-selectin-FITC, rat anti-LFA-1 mAb H68, rat anti-CD11b mAb M1/70, rat anti-mouse CCR2 mAb MC-21, which was a gift from Dr. M. Mack (Regensburg University Medical Center, Regensburg, Germany), and rabbit anti-CX3CR1 Ab (eBioscience).

Formalin-fixed, paraffin-embedded peritoneal membrane sections cut at 2-μm thickness were stained with anti-KC (1/70), anti-MIP-2 (1/40; R&D Systems), and Mac-2 (1/10,000; Cederlane Laboratories) mAbs, respectively, for 1 h at room temperature. Secondary biotinylated rabbit anti-rat Ab (Vector Laboratories) was applied for 45 min followed by incubation with ABC complex solution and diaminobenzidine (BioGenex). Images were acquired on a E1000 Nikon microscope using Eclipse net software.

Bone marrow macrophages were adhered at 1 × 106 cells/ml to 24-well plates (BD Falcon) precoated with 1% low endotoxin BSA fraction V (Sigma-Aldrich) for 1 h. Alternatively, whole peritoneal cell populations (1 × 105/well) or MACS column-purified macrophages or B cells (both at 5 × 104/well) were plated in triplicate in 200 μl of HBSS, 20 mM HEPES (Invitrogen Life Technologies) in 96-well flat-bottom Immulon 1B plates (Thermo Electron) coated with BSA for 2 h.

Following titration of TLR ligands, bone marrow macrophages were stimulated with 1 ng/ml E. coli or S. minnesota LPS (TLR4), 10 μg/ml poly(I:C) (TLR3), 1 μg/ml Pam3Cys (TLR2), or medium alone at 37°C for the indicated time periods. The cell culture supernatants were removed and analyzed by ELISA for the presence of KC, MIP-2, and RANTES (R&D Systems).

Cycloheximide (5 μg/ml; Sigma-Aldrich) and actinomycin D (5 μg/ml; Sigma-Aldrich) were added for indicated time periods to inhibit protein and mRNA synthesis, respectively. All samples were centrifuged and supernatants were removed for chemokine ELISA analysis (R&D Systems).

Total RNA was prepared using the RNeasy kit (Qiagen) and reverse transcribed into cDNA using the Superscript First-Strand Synthesis System (Invitrogen Life Technologies). The 30 ng of cDNA were added to SYBR Green DNA dye (Applied Biosystems) and the following 10 μM sense and antisense primer, respectively: murine KC 5′-CCGAAGTCATAGCCACACTCAA-3′ (sense) and 5′-GCAGTCTGTCTTCTTTCTCCGTTAC-3′ (antisense); murine MIP-2 5′-AGACAGAAGTCATAGCCACTCTCAAG-3′ (sense) and 5′-CCTCCTTTCCAGGTCAGTTAGC-3′ (antisense); and murine γ-actin 5′-ACCATTGGCAATGAGCGG-3′ (sense) and 5′-CCACAGGACTCCATGCCC-3′ (antisense).

Alternatively KC and MIP-2 mRNA levels were quantified using TaqMan technology. Briefly RNA was extracted from bone marrow macrophages using GenElute Mammalian Total RNA Miniprep kit (Sigma-Aldrich). RNA was reverse-transcribed using the First-Strand cDNA Synthesis kit (Amersham Biosciences), and purified using the QIAquick PCR Purification kit (Qiagen). A total of 20 ng of cDNA per reaction was amplified using TaqMan Gene Expression assay (Mm00433859_m1 (KC), Mm00436450_m1 (MIP-2), and control Mm99999915_g1 (GAPDH)).

The samples were analyzed on the ABI 7900HT Sequence Detection System instrument. Each sample was run in triplicate and expressed as a function of threshold cycle (ΔCt). The threshold cycle values for reactions amplifying murine KC or MIP-2 were corrected by the threshold cycle value for γ-actin or GAPDH. The difference in threshold cycle value between treated and control samples allowed the relative expression of the gene to be quantified with the following: 2^ − ((ΔCt treated − ΔCt γ-actin (or ΔCt GAPDH)) − (ΔCt control − ΔCt γ-actin (or ΔCt GAPDH)).

The upstream genomic sequences of the genes were inspected within the National Center for Biotechnology Information build 36 assemblies of the Mouse, Rat, and Human genomes using the University of California, Santa Cruz (UCSC) genome browser. The corresponding multiple species alignments for 1.5-kb upstream of the transcription start sites were extracted using the Vertebrate Multiz Alignment & Conservation track (19) within the UCSC genome browser. The alignments were then screened for selected transcription factor binding sites using MatInspector (20) and a vertebrate factor subset of Genomatix’s proprietary database.

Data are shown as mean ± SEM and represent three to six separate experiments as indicated. Data were analyzed using GraphPad Prism software version 4 for Macintosh computers. Unpaired Student’s t test analyses were performed on the sets of data. Differences were considered significant when p ≤ 0.05 (*, p < 0.05; **, p < 0.01; and ***, p < 0.001). For analysis of over two data sets (see in Fig. 6C), the experimental groups were pre-examined for equivalence to the negative (PBS) and positive (LPS) controls using Dunnett’s multiple comparison test. The unpaired Student’s t test was then applied to the samples identified as being significantly different.

Following induction of peritonitis with LPS, the CXC chemokines, KC and MIP-2, were expressed at peak levels within 1 h of stimulation and both returned to background levels by 4 h (Fig. 1 A). Neutrophil recruitment followed the increase in chemokines and reached maximal levels by 2 h.

FIGURE 1.

Kinetics of neutrophil recruitment and KC and MIP-2 synthesis in vivo by tissue macrophages. A, Time course of KC and MIP-2 production and number of recruited neutrophils in peritoneal lavages of WT mice after i.p. injection of 10 ng/mouse of LPS. Data are shown as mean ± SEM (n = 6 mice). B, Comparison of WT and RAG2 mice. KC and MIP-2 release following stimulation ex vivo of peritoneal cells from WT mice and RAG2 mice with 10 ng/ml of LPS for 2 h. Data are shown as mean ± SEM (n = 3 mice). **, p < 0.01. C, Comparison of KC and MIP-2 release from MACS-sorted peritoneal macrophages (Mph) and B cells without or with 2 h of LPS stimulation. ***, p < 0.001. Consecutive peritoneal wall sections were immunostained for macrophages and for chemokines KC (D) and MIP-2 (E) following 1 h of LPS stimulation. Coincident staining is encircled. The immunostaining is representative of n = 4 mice. Scale bar represents 25 μm.

FIGURE 1.

Kinetics of neutrophil recruitment and KC and MIP-2 synthesis in vivo by tissue macrophages. A, Time course of KC and MIP-2 production and number of recruited neutrophils in peritoneal lavages of WT mice after i.p. injection of 10 ng/mouse of LPS. Data are shown as mean ± SEM (n = 6 mice). B, Comparison of WT and RAG2 mice. KC and MIP-2 release following stimulation ex vivo of peritoneal cells from WT mice and RAG2 mice with 10 ng/ml of LPS for 2 h. Data are shown as mean ± SEM (n = 3 mice). **, p < 0.01. C, Comparison of KC and MIP-2 release from MACS-sorted peritoneal macrophages (Mph) and B cells without or with 2 h of LPS stimulation. ***, p < 0.001. Consecutive peritoneal wall sections were immunostained for macrophages and for chemokines KC (D) and MIP-2 (E) following 1 h of LPS stimulation. Coincident staining is encircled. The immunostaining is representative of n = 4 mice. Scale bar represents 25 μm.

Close modal

We next determined the type of leukocyte that was producing KC and MIP-2. Infiltrating cells were ruled out as the chemokines remained at maximal levels when influx of monocytes and neutrophils was prevented by using a combination of anti-LFA-1 and anti-α4 mAbs (18 and data not shown). We also excluded mast cells, as bone marrow-derived cells that were stimulated in culture over 16 h by cross-linking anti-FcγRII/RIII or anti-FcεRI yielded no KC and MIP-2 (data not shown). However, the two chemokines were detected following LPS stimulation of peritoneal cells from RAG2 mice that lack B and T lymphocytes, thus supporting a role for myeloid cells (Fig. 1,B). The peritoneal cells from RAG2 mice contained more macrophages implying a role for them in chemokine production. To further investigate peritoneal macrophages, we separated WT peritoneal macrophages and B cells using MACS beads and stimulated the cells with LPS for 2 h before testing for release of chemokine. The macrophages and not B cells made both chemokines (Fig. 1 C). Therefore the conclusion was that peritoneal macrophages, and not other leukocytes, produced KC and MIP-2.

As so far only the cells isolated from the peritoneal cavity had been tested for chemokine synthesis, the next step was to directly investigate peritoneal wall tissues collected before and 1, 2, and 4 h after i.p. LPS stimulation. An examination of the tissue below the muscle layer closest to the peritoneal cavity showed that KC staining in macrophages was absent at time zero, maximal at 1 h (Fig. 1,D), diminishing by 2 h and at background levels by 4 h poststimulation with LPS (data not shown). This reflected the pattern of chemokine levels detected by ELISA. When consecutive tissue sections were examined, the staining for KC was observed to substantially coincide with the tissue macrophage staining. Specifically KC positive immunostaining overlapped with ∼70% of the cells identified as macrophages (66 KC+ or 90 Mac-2+ macrophages per 1 cm peritoneal wall length). Although the immunostaining for MIP-2 was less intense, it followed the same distribution pattern as for KC (Fig. 1 D). Other cell types such as eosinophils, epithelial cells and the vasculature were negative for both chemokines (data not shown). Thus macrophages are the major tissue source of both KC and MIP-2.

To phenotypically characterize the chemokine-producing macrophage and to ask whether LPS stimulation altered the profile, we investigated the expression of receptors on macrophages that were isolated before and after LPS stimulation in vivo. We focused on markers that have been used to define the inflammatory (M1) and resident (M2) subsets of monocyte that give rise to distinct tissue macrophages (21). The pattern of epitope expression (negative for CX3CR1, low for CCR2, and LFA-1(CD11a)) is typical of monocytes of the inflammatory type (M1), whereas lack of L-selectin and the Gr-1 marker typify the constitutive subtype of monocyte (M2) (Fig. 2). The peritoneal macrophages also expressed high levels of Mac-1 (CD11b) similarly to both M1 and M2 monocyte markers. The pattern of epitope expression was unaltered following exposure to LPS injected i.p. for 2 h except for a small increase in the Gr-1 marker (Fig. 2). Therefore, the peritoneal macrophages that produce the chemokines KC and MIP-2 have both inflammatory and constitutive phenotypic characteristics, suggesting that they may potentially perform functions of both M1 and M2 macrophages.

FIGURE 2.

Phenotypic profile of peritoneal macrophages ± LPS in vivo. WT mice were injected with PBS or 10 ng/mouse of LPS for 4 h, and peritoneal cells were collected. F4/80-positive cells were analyzed for surface markers representing the M1 or M2 macrophage phenotypes. Histograms are gated on F4/80-positive pooled cells and mean fluorescence intensity ± SEM (n = 4 mice). In terms of LPS-treated compared with untreated groups, there was no significant difference except for the small increase in Gr-1 expression (p < 0.01). Data are representative of two experiments.

FIGURE 2.

Phenotypic profile of peritoneal macrophages ± LPS in vivo. WT mice were injected with PBS or 10 ng/mouse of LPS for 4 h, and peritoneal cells were collected. F4/80-positive cells were analyzed for surface markers representing the M1 or M2 macrophage phenotypes. Histograms are gated on F4/80-positive pooled cells and mean fluorescence intensity ± SEM (n = 4 mice). In terms of LPS-treated compared with untreated groups, there was no significant difference except for the small increase in Gr-1 expression (p < 0.01). Data are representative of two experiments.

Close modal

We next investigated the synthesis and release of KC and MIP-2 to discover how production of these chemokines was controlled. Cell lysates from untreated peritoneal macrophages yielded no chemokine when extracted with detergent (data not shown). Moreover blocking protein synthesis by exposure to cycloheximide before treatment with LPS prevented the expression of KC and MIP-2, further confirming that the chemokines are not stored (Fig. 3,A). Inhibiting transcription with actinomycin D retarded production of both chemokines when added during the first hour of stimulation, indicating that new mRNA is synthesized in this initial period (Fig. 3 B). These chemokines are therefore not presynthesized and stored, but are rapidly produced by macrophages in response to an inflammatory trigger.

FIGURE 3.

Analysis of macrophage KC and MIP-2 protein and mRNA. A, Blocking of protein KC and MIP-2 synthesis using cycloheximide. Cycloheximide (CHX) (5 μg/ml) was added to peritoneal macrophages before treatment with LPS (10 ng/ml). Incubation with PBS was a control in the experiments. ***, p < 0.001. B, Blocking of KC and MIP-2 transcription by actinomycin D added to peritoneal macrophages either at the same time as stimulation (0 h), 0.5, or 1 h afterward. Supernatants were tested for KC and MIP-2 release at 2 h. *, p < 0.05; **, p < 0.01. C, Quantification of macrophage chemokine KC and MIP-2 mRNA normalized to γ-actin levels in response to LPS stimulation for 1 h. **, p < 0.01. D, Stability of KC and MIP-2 mRNA followed for 0.5 or 1 h after 45 min of LPS stimulation. Data are shown as mean ± SEM (n = 3). NS, Not significant.

FIGURE 3.

Analysis of macrophage KC and MIP-2 protein and mRNA. A, Blocking of protein KC and MIP-2 synthesis using cycloheximide. Cycloheximide (CHX) (5 μg/ml) was added to peritoneal macrophages before treatment with LPS (10 ng/ml). Incubation with PBS was a control in the experiments. ***, p < 0.001. B, Blocking of KC and MIP-2 transcription by actinomycin D added to peritoneal macrophages either at the same time as stimulation (0 h), 0.5, or 1 h afterward. Supernatants were tested for KC and MIP-2 release at 2 h. *, p < 0.05; **, p < 0.01. C, Quantification of macrophage chemokine KC and MIP-2 mRNA normalized to γ-actin levels in response to LPS stimulation for 1 h. **, p < 0.01. D, Stability of KC and MIP-2 mRNA followed for 0.5 or 1 h after 45 min of LPS stimulation. Data are shown as mean ± SEM (n = 3). NS, Not significant.

Close modal

Quantification of mRNA levels in peritoneal macrophages using RT-PCR confirmed that mRNA was synthesized during the first hour of stimulation in vivo with LPS and revealed that stimulated macrophages expressed similar levels of KC and MIP-2 mRNA (Fig. 3,C). We also assessed the chemokine mRNA stability by following its decay in macrophages over 1 h in the presence of actinomycin D following 45 min of LPS stimulation. Both KC and MIP-2 mRNAs are equally stable over this time period (Fig. 3 D). Therefore KC and MIP-2 are similarly regulated by de novo synthesis at the transcriptional level following LPS stimulation.

A comparison of WT and TLR4 mutant bone marrow-derived macrophages in vitro showed that LPS signaling through TLR4 accounted for the increase in KC and MIP-2 protein (Fig. 4,A). Signaling downstream of TLR4 can be directed through adaptor proteins MAL/MyD88 or alternatively, TRAM/TRIF (1, 2). The stimulation of bone marrow macrophages from MyD88−/− × TRIF mice showed that these two pathways accounted for all the KC and MIP-2 produced following LPS stimulation (Fig. 4 B). These findings were confirmed using peritoneal macrophages from TLR4 mutant mice (data not shown).

FIGURE 4.

KC and MIP-2 production and the TLR4 signaling pathway. A, Time course of KC and MIP-2 production by WT and TLR4 mutant (TLR4 mut) bone marrow macrophages. **, p < 0.01; ***, p < 0.001. B, WT and MyD88−/− × TRIF bone marrow macrophages, all stimulated with 1 ng/ml LPS. **, p < 0.01; ***, p < 0.001. C, Transcription of KC and MIP-2 mRNA was evaluated by TaqMan analysis using WT and TLR4 mutant bone marrow macrophages stimulated for 2 h with an LPS-stimulated macrophage supernatant. Data are shown as mean ± SEM (n = 3 mice). **, p < 0.01; ***, p < 0.001.

FIGURE 4.

KC and MIP-2 production and the TLR4 signaling pathway. A, Time course of KC and MIP-2 production by WT and TLR4 mutant (TLR4 mut) bone marrow macrophages. **, p < 0.01; ***, p < 0.001. B, WT and MyD88−/− × TRIF bone marrow macrophages, all stimulated with 1 ng/ml LPS. **, p < 0.01; ***, p < 0.001. C, Transcription of KC and MIP-2 mRNA was evaluated by TaqMan analysis using WT and TLR4 mutant bone marrow macrophages stimulated for 2 h with an LPS-stimulated macrophage supernatant. Data are shown as mean ± SEM (n = 3 mice). **, p < 0.01; ***, p < 0.001.

Close modal

Products of the MyD88 or TRIF pathways can be induced indirectly via secondary signaling, potentially through IFN-β (4, 22). To ascertain whether the induction of KC and MIP-2 synthesis was direct or indirect, we exposed WT and TLR4 mutant peritoneal macrophages separately to a cytokine-containing supernatant from WT macrophages collected at 1 h poststimulation with LPS. TaqMan analysis showed that the WT, but not TLR4 mutant macrophages, were able to synthesize KC and MIP-2 mRNA demonstrating that both are direct products of TLR4 signaling (Fig. 4 C). To confirm this result we stimulated bone marrow-derived macrophages in vitro with IFN-β and this yielded no KC or MIP-2 protein (data not shown).

We next separately investigated the two TLR4 signaling pathways using MyD88 and TRIF mice. In terms of KC production, TRIF bone marrow macrophages stimulated with LPS were not significantly affected, whereas MyD88 macrophages (Fig. 5,A) were unable to make this chemokine. In contrast, MIP-2 production was partially inhibited in both TRIF and MyD88 bone marrow macrophages (Fig. 5 A). These findings were all repeated using TRIF and MyD88 peritoneal macrophages (data not shown). Therefore KC synthesis is exclusively dependent upon signaling through the MyD88 pathway, whereas MIP-2 is synthesized by signaling through both TLR4-associated pathways.

FIGURE 5.

Analysis of MyD88 and TRIF signaling pathways in KC and MIP-2 production. A, For TLR4, WT, MyD88, and TRIF bone marrow macrophages were stimulated with 1 ng/ml LPS, supernatant was collected at the indicated time periods, and KC and MIP-2 levels were measured by ELISA. *, p < 0.05; **, p < 0.01; ***, p < 0.001. WT, MyD88, and TRIF bone marrow macrophages were stimulated with 1 μg/ml Pam3Cys (TLR2) (B) or 10 μg/ml poly(I:C) (TLR3) (C), supernatants were collected at the indicated time points, and KC and MIP-2 levels were measured by ELISA. Incubation with PBS was a control in all experiments. In B, there were significant differences between WT and MyD88 but not TRIF samples points. In C, WT differed significantly from TRIF but not MyD88 sample points. Data are shown as mean ± SEM (n = 3 mice).

FIGURE 5.

Analysis of MyD88 and TRIF signaling pathways in KC and MIP-2 production. A, For TLR4, WT, MyD88, and TRIF bone marrow macrophages were stimulated with 1 ng/ml LPS, supernatant was collected at the indicated time periods, and KC and MIP-2 levels were measured by ELISA. *, p < 0.05; **, p < 0.01; ***, p < 0.001. WT, MyD88, and TRIF bone marrow macrophages were stimulated with 1 μg/ml Pam3Cys (TLR2) (B) or 10 μg/ml poly(I:C) (TLR3) (C), supernatants were collected at the indicated time points, and KC and MIP-2 levels were measured by ELISA. Incubation with PBS was a control in all experiments. In B, there were significant differences between WT and MyD88 but not TRIF samples points. In C, WT differed significantly from TRIF but not MyD88 sample points. Data are shown as mean ± SEM (n = 3 mice).

Close modal

To assess the functional importance of each pathway, we stimulated bone marrow macrophages through MyD88 alone by activating the TLR2 pathway and through TRIF alone by activating the TLR3 pathway. When the WT macrophages were stimulated in vitro with Pam3Cys, a synthetic lipopeptide ligand for TLR2 (22), both KC and MIP-2 were made in equivalent amounts (Fig. 5,B). The association with MyD88 was confirmed by showing that neither chemokine was made following TLR2 stimulation of MyD88 macrophages, whereas production of the chemokines by the TRIF macrophages was not affected. However when WT macrophages were stimulated with poly(I:C), a synthetic analog of dsRNA ligand for TLR3 (3), MIP-2, but not KC, was produced (Fig. 5 C). Furthermore poly(I:C) induced MIP-2 in MyD88 macrophages, but not in TRIF macrophages, showing that MIP-2 is definitely a product of the TRIF pathway. The synthesis of the two chemokines is therefore distinguished by the finding that MIP-2 is induced by both major TLR signaling pathways, whereas synthesis of KC is restricted to the MyD88 pathway.

To investigate the relative importance of the two chemokines in neutrophil recruitment in vivo, we focused first on MIP-2 and we examined LPS-stimulated MyD88 mice in which the TRIF pathway was intact. As expected, MyD88 mice made only MIP-2, and not KC. The overall level of chemokine was slightly reduced due to the lack of the MyD88 pathway. However, neutrophils were recruited over 4 h at ∼50% of WT level, indicating that the MIP-2 produced through TRIF pathway stimulation was sufficient to cause cell influx (Fig. 6,A). By comparison, the response of TRIF mice, that can make both KC and MIP-2 via the MyD88 pathway, was equivalent to WT mice in terms of chemokine production and neutrophil influx (Fig. 6 B).

FIGURE 6.

Neutrophil recruitment is KC- and MIP-2-dependent. A, Time course of release of chemokines KC and MIP-2 in vivo and correlation with subsequent neutrophil influx into the peritoneal cavity of WT and MyD88 mice following LPS treatment (10 ng/mouse). Data are shown as mean ± SEM (n = 6 mice). *, p < 0.05; **, p < 0.01; ***, p < 0.001. B, Time course of release of chemokines KC and MIP-2 in vivo and correlation with subsequent neutrophil influx into the peritoneal cavity of WT and TRIF mice following LPS treatment. Data are shown as mean ± SEM (n = 6 mice). The WT and TRIF data are significantly different only where indicated. *, p < 0.05. C, Neutrophil influx was quantified at 4 h after i.p. injection of anti-KC (10 μg/mouse), anti-MIP-2 (10 μg/mouse), a combination of mAbs (5 μg per mAb/mouse), or control mAb (10 μg/mouse) 15 min before LPS treatment (10 ng/mouse) of WT mice. Data are analyzed as described in Materials and Methods and are shown as mean ± SEM (n = 10 mice). *, p < 0.05. NS, Not significant. D, Schematic analysis of KC, MIP-2, and RANTES promoters for NF-κB and IRF binding sites. The sequences 1500 nt upstream of the start sites for KC, MIP-2, and RANTES transcription were analyzed for conserved IRF sites as well as NF-κB sites using Genomatix MatInspector Tool.

FIGURE 6.

Neutrophil recruitment is KC- and MIP-2-dependent. A, Time course of release of chemokines KC and MIP-2 in vivo and correlation with subsequent neutrophil influx into the peritoneal cavity of WT and MyD88 mice following LPS treatment (10 ng/mouse). Data are shown as mean ± SEM (n = 6 mice). *, p < 0.05; **, p < 0.01; ***, p < 0.001. B, Time course of release of chemokines KC and MIP-2 in vivo and correlation with subsequent neutrophil influx into the peritoneal cavity of WT and TRIF mice following LPS treatment. Data are shown as mean ± SEM (n = 6 mice). The WT and TRIF data are significantly different only where indicated. *, p < 0.05. C, Neutrophil influx was quantified at 4 h after i.p. injection of anti-KC (10 μg/mouse), anti-MIP-2 (10 μg/mouse), a combination of mAbs (5 μg per mAb/mouse), or control mAb (10 μg/mouse) 15 min before LPS treatment (10 ng/mouse) of WT mice. Data are analyzed as described in Materials and Methods and are shown as mean ± SEM (n = 10 mice). *, p < 0.05. NS, Not significant. D, Schematic analysis of KC, MIP-2, and RANTES promoters for NF-κB and IRF binding sites. The sequences 1500 nt upstream of the start sites for KC, MIP-2, and RANTES transcription were analyzed for conserved IRF sites as well as NF-κB sites using Genomatix MatInspector Tool.

Close modal

To further examine the roles of KC and MIP-2 separately, we treated WT mice with anti-KC, anti-MIP-2, or both mAbs together before LPS stimulation. Either anti-KC or anti-MIP-2, but not control Ab, blocked neutrophil recruitment to the same limited degree, and adding both mAbs together was required to reduce neutrophil entry to background levels (Fig. 6 C). Polyclonal Abs for KC and MIP-2 displayed the same pattern of neutrophil inhibition (data not shown). The fact that blocking the activity of both chemokines was required to bring neutrophil recruitment to background levels indicates that KC and MIP-2 function similarly and operate in an additive fashion.

Stimulation of the TRIF pathway through TLR3 and TLR4 to produce the chemokine MIP-2 would increase the robustness of the neutrophil recruitment to pathogens such as dsRNA viruses. An issue is why MIP-2, and not KC, is induced via TRIF stimulation. The chemokine RANTES (CCL5) is a direct product of the TRIF pathway and has both IRF and NF-κB sites in its promoter that cooperate to transcribe this chemokine (4, 23). Analysis of the sequences of the promoters of KC and MIP-2 show that MIP-2, but not KC, contains a typical IRF site as well as NF-κB sites similar to RANTES (Fig. 6 D). This provides an explanation as to why, of the two neutrophil chemokines, only MIP-2 is a product of the TRIF pathway.

In this in vivo study we identify tissue macrophages as the major source of the chemokines KC and MIP-2 that recruit neutrophils in response to TLR signaling at the initiation of an immune response. The chemokines are produced by stimulation of macrophage TLRs, but there are differences in the signaling pathways involved. Both KC and MIP-2 are synthesized through the MyD88 pathway, but MIP-2 alone is also synthesized as a direct product of the less frequently used TRIF pathway. Although past reports have promoted the functional relevance of either KC or MIP-2, we find that both chemokines have similar roles in the recruitment of neutrophils.

Following TLR4 stimulation in vitro by LPS, KC and MIP-2 were produced by peritoneal macrophages and not by T and B lymphocytes, monocytes, neutrophils, or mast cells. In addition tissue macrophages located in the peritoneal wall stained positively for KC and MIP-2 further highlighting resident peritoneal macrophages as the major source of these neutrophil-attracting chemokines. The findings are in agreement with another in vivo study using a conditional macrophage ablation transgenic mouse model in which KC and MIP-2 were substantially reduced and neutrophil entry diminished following thioglycolate stimulation (24).

The KC/MIP-2-producing macrophages have a distinctive phenotypic profile. In both mouse and human, two sets of circulating monocytes exist that give rise to tissue macrophages with individual functional characteristics. The M2 subset of monocytes matures into macrophages that are constitutively resident in tissues, whereas the M1 subset consists of monocytes that become specifically recruited to sites of inflammation (21, 25). The peritoneal macrophages that produce KC and MIP-2 display features of both resident (CD62L/Gr-1low) and inflammatory (CXCR1/CCR2low/LFA-1low) cells. They may therefore lie between the two prototype classes of macrophage in terms of function. Macrophages are regarded as having a certain degree of “plasticity” in terms of phenotype and can switch from M2 to M1 phenotype as a consequence of chronic changes of mouse diet or following activation (25, 26). However, except for a small increase in Gr-1 expression, the peritoneal macrophage phenotypic profile does not alter following exposure to LPS, suggesting that these cells have a broader functional capacity that can encompass both the role of the resident macrophage and the immediate response required when an inflammatory agent is encountered.

Chemokines are produced by leukocytes in different ways. They can be stored in granules that are released upon stimulation (27). For example, the human KC homolog Groα is maintained in endothelial cells in granule form (28). Alternatively various stimulants can induce chemokine mRNA production in vivo and in vitro (29, 30). In this study, we show that for the peritoneal macrophage there is no storage of KC or MIP-2, but new synthesis of mRNA and protein occurs following an LPS stimulus. In both in vivo and in vitro experiments, KC and MIP-2 mRNA are made during the first hour of stimulation and the subsequent production of their corresponding proteins is maximal at 1–2 h. The expression of both chemokines is transient and returns to background levels by 4 h. This timing corresponds to neutrophil recruitment that peaks at 2 h.

A key aspect of this study has been to investigate how KC and MIP-2 are produced downstream of the TLRs. We have focused chiefly on TLR4 that signals to both MyD88 and TRIF pathways, but have also investigated TLR2, which signals only through the MyD88 pathway, and TLR3, which signals exclusively to the TRIF pathway. When LPS was used as a stimulant, the production of KC/MIP-2 was observed to be completely TLR4-dependent, but there were differences in the signaling pathways leading to synthesis of the chemokines downstream of the TLR4. We used MyD88, TRIF, and MyD88 × TRIF null mice to show that KC was produced exclusively via the MyD88 pathway, but MIP-2 had a more extensive synthetic route being produced through both MyD88 and TRIF pathways.

A limited number of immune mediators, such as IFN-β, have been identified as direct products of signaling through the TRIF adaptor protein. A larger number of products have been linked to this pathway as a result of autocrine and paracrine signaling by IFN-β (4, 22). We found, however, that exposure of macrophages to cytokine-containing LPS-stimulated supernatants or to IFN-β failed to induce either MIP-2 protein or mRNA. Thus MIP-2 joins the limited number of mediators synthesized as a result of direct signaling through the TRIF pathway in macrophages.

IFN-β and the chemokine RANTES are both made by signaling via TRIF. Both also have IRF as well as NF-κB binding sites in their promoters and require a complex of the two factors to stimulate transcriptional activity in macrophages (4, 23). We found that the MIP-2, but not KC, promoter has an IRF consensus sequence as well as NF-κB sites, providing an explanation as to why the former, but not the latter, is induced through the TRIF protein. For RANTES, this cooperation between IRF and NF-κB involves a κB site adjacent to the IRF site (23). A question is why MIP-2 is still expressed when IRF activity is not involved. It is probable that the transcription induced through the MyD88 pathway in macrophages makes use of specific NF-κB sites different from the one associating with IRFs.

Reports in the literature often focus on one or other of these chemokines in terms of relevance for neutrophil recruitment (10, 11, 12, 13, 14). An issue in this study was whether KC and MIP-2 had distinctive roles or whether they functioned in an additive fashion in terms of recruiting neutrophils. There are examples of chemokine complementarity that direct monocyte transmigration and involve different roles for KC, or its human homolog GROα, in synergy with MCP-1 (15, 31). The finding that maximal neutrophil infiltration occurred when both chemokines KC and MIP-2 were expressed and that mAb inhibition of both KC and MIP-2 was required to prevent neutrophil recruitment indicates that their functional effects are similar and additive.

The fact that stimulation of various classes of TLR result in MIP-2 or KC production ensures that neutrophil recruitment is a feature of many types of pathogen infection. Although most TLRs signal through MyD88, stimulation of TLR3, which is recognized by dsRNA-containing viruses, accesses only the TRIF pathway. Both RANTES (CCL5), which is a general leukocyte recruiter, and IP-10 (CXCL10), which more specifically targets activated T cells and NK cells, are induced as a result of TLR3 signaling via TRIF (4, 32). In this study, we show that when MIP-2 alone is synthesized downstream of TRIF, the amount is sufficient to cause an influx of neutrophils, even if not at the maximal level. Thus signaling via the TRIF pathway also induces a chemokine able to recruit neutrophils that would influence a TLR3-initiated antiviral program of responses.

The role of TLR signaling in bridging innate and adaptive responses through macrophage stimulation has been well described. In particular there has been recent emphasis on the subset of chemokines made by macrophages that activate dendritic cells (32). In this study, we focus on another important role of tissue macrophages. We show that macrophage TLR signaling induces the chemokines KC and MIP-2 that are responsible for the recruitment of neutrophils constituting a first essential step in the response to a pathogen.

We thank Dr. Shizuo Akira (Osaka, Japan) for permission to use the MyD88 and TRIF null mice and Dr. Caetano Reis e Sousa (London, U.K.) for providing them. We thank Dr. Matthias Mack (Regensburg, Germany) for the CCR2 mAb. We are grateful to our colleagues Drs. Petros Takousis, Andreas Bolzer, and Stefania Segditsas for guidance with the real-time RT-PCR assay; Dr. Michael Mitchell for bioinformatics analysis; Louise Coleman, Gill Hutchinson, Clare Watkins, and Emma Murray for animal husbandry; and Emma Nye for immunohistochemistry. We are also grateful to Caetano Reis e Sousa for helpful discussion.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by the Cancer Research United Kingdom.

5

Abbreviations used in this paper: TRIF, Toll/IL-1R domain–containing adaptor-inducing IFN-β; IRF, IFN-regulated factor; poly(I:C), polyinosine-polycytidylic acid; WT, wild type.

1
Akira, S., K. Takeda.
2004
. Toll-like receptor signalling.
Nat. Rev. Immunol.
4
:
499
-511.
2
Dunne, A., L. A. O’Neill.
2005
. Adaptor usage and Toll-like receptor signaling specificity.
FEBS Lett.
579
:
3330
-3335.
3
Yamamoto, M., S. Sato, H. Hemmi, K. Hoshino, T. Kaisho, H. Sanjo, O. Takeuchi, M. Sugiyama, M. Okabe, K. Takeda, S. Akira.
2003
. Role of adaptor TRIF in the MyD88-independent toll-like receptor signaling pathway.
Science
301
:
640
-643.
4
Doyle, S., S. Vaidya, R. O’Connell, H. Dadgostar, P. Dempsey, T. Wu, G. Rao, R. Sun, M. Haberland, R. Modlin, G. Cheng.
2002
. IRF3 mediates a TLR3/TLR4-specific antiviral gene program.
Immunity
17
:
251
-263.
5
Hoshino, K., T. Kaisho, T. Iwabe, O. Takeuchi, S. Akira.
2002
. Differential involvement of IFN-β in Toll-like receptor-stimulated dendritic cell activation.
Int. Immunol.
14
:
1225
-1231.
6
Ben-Baruch, A., D. F. Michiel, J. J. Oppenheim.
1995
. Signals and receptors involved in recruitment of inflammatory cells.
J. Biol. Chem.
270
:
11703
-11706.
7
Kaplanski, G., V. Marin, F. Montero-Julian, A. Mantovani, C. Farnarier.
2003
. IL-6: a regulator of the transition from neutrophil to monocyte recruitment during inflammation.
Trends Immunol.
24
:
25
-29.
8
Lee, J., G. Cacalano, T. Camerato, K. Toy, M. W. Moore, W. I. Wood.
1995
. Chemokine binding and activities mediated by the mouse IL-8 receptor.
J. Immunol.
155
:
2158
-2164.
9
Zlotnik, A., O. Yoshie.
2000
. Chemokines: a new classification system and their role in immunity.
Immunity
12
:
121
-127.
10
McColl, S. R., I. Clark-Lewis.
1999
. Inhibition of murine neutrophil recruitment in vivo by CXC chemokine receptor antagonists.
J. Immunol.
163
:
2829
-2835.
11
Zwijnenburg, P. J., M. M. Polfliet, S. Florquin, T. K. van den Berg, C. D. Dijkstra, S. J. van Deventer, J. J. Roord, T. van der Poll, A. M. van Furth.
2003
. CXC-chemokines KC and macrophage inflammatory protein-2 (MIP-2) synergistically induce leukocyte recruitment to the central nervous system in rats.
Immunol. Lett.
85
:
1
-4.
12
Mehrad, B., R. M. Strieter, T. A. Moore, W. C. Tsai, S. A. Lira, T. J. Standiford.
1999
. CXC chemokine receptor-2 ligands are necessary components of neutrophil-mediated host defense in invasive pulmonary aspergillosis.
J. Immunol.
163
:
6086
-6094.
13
Tsai, W. C., R. M. Strieter, J. M. Wilkowski, K. A. Bucknell, M. D. Burdick, S. A. Lira, T. J. Standiford.
1998
. Lung-specific transgenic expression of KC enhances resistance to Klebsiella pneumoniae in mice.
J. Immunol.
161
:
2435
-2440.
14
Keane, M. P., J. A. Belperio, T. A. Moore, B. B. Moore, D. A. Arenberg, R. E. Smith, M. D. Burdick, S. L. Kunkel, R. M. Strieter.
1999
. Neutralization of the CXC chemokine, macrophage inflammatory protein-2, attenuates bleomycin-induced pulmonary fibrosis.
J. Immunol.
162
:
5511
-5518.
15
Huo, Y., C. Weber, S. B. Forlow, M. Sperandio, J. Thatte, M. Mack, S. Jung, D. R. Littman, K. Ley.
2001
. The chemokine KC, but not monocyte chemoattractant protein-1, triggers monocyte arrest on early atherosclerotic endothelium.
J. Clin. Invest.
108
:
1307
-1314.
16
Poltorak, A., X. He, I. Smirnova, M. Y. Liu, C. Van Huffel, X. Du, D. Birdwell, E. Alejos, M. Silva, C. Galanos, et al
1998
. Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: mutations in Tlr4 gene.
Science
282
:
2085
-2088.
17
Adachi, O., T. Kawai, K. Takeda, M. Matsumoto, H. Tsutsui, M. Sakagami, K. Nakanishi, S. Akira.
1998
. Targeted disruption of the MyD88 gene results in loss of IL-1- and IL-18-mediated function.
Immunity
9
:
143
-150.
18
Henderson, R. B., L. H. Lim, P. A. Tessier, F. N. Gavins, M. Mathies, M. Perretti, N. Hogg.
2001
. The use of lymphocyte function-associated antigen (LFA)-1-deficient mice to determine the role of LFA-1, Mac-1, and α4 integrin in the inflammatory response of neutrophils.
J. Exp. Med.
194
:
219
-226.
19
Blanchette, M., W. J. Kent, C. Riemer, L. Elnitski, A. F. Smit, K. M. Roskin, R. Baertsch, K. Rosenbloom, H. Clawson, E. D. Green, et al
2004
. Aligning multiple genomic sequences with the threaded blockset aligner.
Genome Res.
14
:
708
-715.
20
Cartharius, K., K. Frech, K. Grote, B. Klocke, M. Haltmeier, A. Klingenhoff, M. Frisch, M. Bayerlein, T. Werner.
2005
. MatInspector and beyond: promoter analysis based on transcription factor binding sites.
Bioinformatics
21
:
2933
-2942.
21
Geissmann, F., S. Jung, D. R. Littman.
2003
. Blood monocytes consist of two principal subsets with distinct migratory properties.
Immunity
19
:
71
-82.
22
Toshchakov, V., B. W. Jones, P. Y. Perera, K. Thomas, M. J. Cody, S. Zhang, B. R. Williams, J. Major, T. A. Hamilton, M. J. Fenton, S. N. Vogel.
2002
. TLR4, but not TLR2, mediates IFN-beta-induced STAT1α/β-dependent gene expression in macrophages.
Nat. Immunol.
3
:
392
-398.
23
Genin, P., M. Algarte, P. Roof, R. Lin, J. Hiscott.
2000
. Regulation of RANTES chemokine gene expression requires cooperativity between NF-κB and IFN-regulatory factor transcription factors.
J. Immunol.
164
:
5352
-5361.
24
Cailhier, J. F., M. Partolina, S. Vuthoori, S. Wu, K. Ko, S. Watson, J. Savill, J. Hughes, R. A. Lang.
2005
. Conditional macrophage ablation demonstrates that resident macrophages initiate acute peritoneal inflammation.
J. Immunol.
174
:
2336
-2342.
25
Gordon, S..
2007
. Macrophage heterogeneity and tissue lipids.
J. Clin. Invest.
117
:
89
-93.
26
Lumeng, C. N., J. L. Bodzin, A. R. Saltiel.
2007
. Obesity induces a phenotypic switch in adipose tissue macrophage polarization.
J. Clin. Invest.
117
:
175
-184.
27
Chertov, O., D. Yang, O. M. Howard, J. J. Oppenheim.
2000
. Leukocyte granule proteins mobilize innate host defenses and adaptive immune responses.
Immunol. Rev.
177
:
68
-78.
28
Oynebraten, I., N. Barois, K. Hagelsteen, F. E. Johansen, O. Bakke, G. Haraldsen.
2005
. Characterization of a novel chemokine-containing storage granule in endothelial cells: evidence for preferential exocytosis mediated by protein kinase A and diacylglycerol.
J. Immunol.
175
:
5358
-5369.
29
Introna, M., R. C. Bast, Jr, C. S. Tannenbaum, T. A. Hamilton, D. O. Adams.
1987
. The effect of LPS on expression of the early “competence” genes JE and KC in murine peritoneal macrophages.
J. Immunol.
138
:
3891
-3896.
30
Tessier, P. A., P. H. Naccache, I. Clark-Lewis, R. P. Gladue, K. S. Neote, S. R. McColl.
1997
. Chemokine networks in vivo: involvement of C-X-C and C-C chemokines in neutrophil extravasation in vivo in response to TNF-α.
J. Immunol.
159
:
3595
-3602.
31
Weber, K. S., P. von Hundelshausen, I. Clark-Lewis, P. C. Weber, C. Weber.
1999
. Differential immobilization and hierarchical involvement of chemokines in monocyte arrest and transmigration on inflamed endothelium in shear flow.
Eur. J. Immunol.
29
:
700
-712.
32
Luster, A. D..
2002
. The role of chemokines in linking innate and adaptive immunity.
Curr. Opin. Immunol.
14
:
129
-135.