CCL9 Is Secreted by the Follicle-Associated Epithelium and Recruits Dome Region Peyer’s Patch CD11b⁺ Dendritic Cells ¹

The Peyer’s patches are lymphoid organs located in the small intestine, which serve as the major sites for generation of immunity to intestinal pathogens. In previous studies, three separate populations of dendritic cells (DCs) ³ were identified with distinct anatomical distribution pattern and diverse functions (1, 2). The DCs that specifically localize in the subepithelial dome regions of the Peyer’s patches express the CD11b molecule and secrete mainly IL-10 upon in vitro stimulation with CD40L or with killed Staphylococcus aureus. The second DC subset expresses the CD8αα molecule and is localized in the T cell-rich interfollicular regions of the Peyer’s patches. The third DC subset lacks the expression of CD11b and CD8α, thus named double negative (DN) DCs, and is found in both the dome region and in the interfollicular regions. The latter two populations share similar functional characteristics, namely, that they both secrete IL-12 p70 upon bacterial stimulation and induce predominantly Th1 responses in naive TCR transgenic CD4⁺ T cells in vitro (2).

The chemokines that coordinate the localization of these DC subsets have been proposed to include CCL20 and CCL21/CCL19. The follicle-associated epithelium (FAE) overlying the dome regions of the Peyer’s patches was found to express high levels of CCL20 (1, 3). Further, the dome region CD11b⁺ DCs have a unique capacity to migrate toward CCL20 (1). In contrast, CD8α⁺ DCs and the DN DCs both express CCR7 mRNA and migrate toward CCL21 expressed in the T cell regions of the Peyer’s patches. The importance of CCL20 in CD11b⁺ DC recruitment to the dome regions of the Peyer’s patches was later reported in CCR6-deficient mice (3, 4).

In the process of analyzing other chemokines that are specifically secreted by the FAE, we have identified the chemokine CCL9. CCL9, also known as macrophage inflammatory protein (MIP)-1γ, MIP-related protein (MRP)-2 and CCF18, is a mouse CC chemokine independently cloned by three different groups (5, 6, 7). Among CC chemokine family members, the mouse CCL9 shares 45, 24, and 20% aa sequence identity with mouse CCL6 (C10), CCL3 (MIP-1α), and CCL4 (MIP-1β), respectively (7). Unlike chemokines such as CCL3 and CCL4, whose expression is induced by inflammatory stimuli, CCL9 was shown to be constitutively expressed by a wide variety of tissues, and relatively high concentrations (1 μg/ml) of CCL9 has been detected in the circulation of normal mice (7). The CCL9 protein can be secreted from Langerhans’ cells, DCs (8) and macrophage cell lines (6), and has been shown to induce chemotaxis and Ca²⁺ flux in CD4⁺ T cell clones (5) and in splenic CD4⁺ T cells (8). However, the precise expression of CCL9 in vivo has not been examined thus far.

In this study, we demonstrated the expression of CCL9 on the FAE, but minimally on villus epithelium, at the mRNA and at the protein levels. A putative receptor for CCL9, CCR1 was found to be highly expressed by the CD11b⁺ DCs. The CD11b⁺ DCs isolated from the Peyer’s patches exhibited migratory activity to CCL9. Moreover, a careful examination of the CCR6-deficient mice, generated from four separate laboratories, using an amplified immunofluorescence technique revealed the presence of CD11b⁺ DCs in the dome regions of their Peyer’s patches. These results suggested that a CCL20/CCR6-independent mechanism exists for the recruitment of CD11b⁺ DCs to the dome regions. Finally, Ab-blocking of CCL9 in either the CCR6-deficient mice or wild type (WT) mice resulted in a significant reduction in the number of the CD11b⁺ DC from the dome regions of the Peyer’s patches. Collectively, these results suggest that CCL9 is constitutively secreted by the FAE, and plays an important role in the recruitment of CD11b⁺ DCs.

Six- to 8-wk-old female BALB/c mice were obtained from the National Cancer Institute (Frederick, MD). Four independently derived CCR6-deficient strains of mice and their WT littermates were used. Their production and characterization have been reported as previously described for strains 2, (4) 3, (9) and 4 (a kind gift from Dr. D. Cook, Duke University, Durham, NC) (3), or will be described elsewhere for strain 1 (A.S. and J.M.F.). CCR1-deficient mice (10) were kindly provided by Dr. P. Murphy (National Institutes of Health, Bethesda, MD) and CCR5-deficient mice were purchased from The Jackson Laboratory (Bar Harbor, ME). All procedures used in this study complied with federal guidelines and institutional policies by the Yale Animal Care and Use Committee.

The following Abs were used for the identification of DC populations; anti-CD11c (HL-3), anti-CD8α (53-6.7), anti-CD11b (M1/70) (BD PharMingen, San Diego, CA). For immunofluorescence staining and in vivo neutralization of chemokines, goat polyclonal Abs generated to mouse CCL6, CCL9, CCL17, CCL20, CCL25, CCL27, CCL28, and normal goat IgG were purchased from R&D Systems (Minneapolis, MN).

The Peyer’s patches were dissected from the mouse small intestine, and were frozen in Tissue-Tek OCT compound (VWR Scientific, Westchester, PA). Cryosections of 7-μm thickness were cut, fixed in cold acetone, and stained with a variety of Abs in a procedure previously described (1). Briefly, fixed sections were blocked with TNB buffer (NEN Life Science Products, Boston, MA) containing 5% normal donkey serum. To block endogenous biotin, the sections were further treated with the Avidin/Biotin block (Vector Laboratories, Burlingame, CA), and endogenous peroxidase activity quenched with 1% H₂O₂. Primary Abs were applied at 5 μg/ml for 1.5 h at room temperature. Slides were washed and incubated with biotin-conjugated donkey F(ab^′)₂ specific for the species of the primary Ab (Jackson Immunoresearch Laboratories, West Grove, PA), followed by incubation with streptavidin-HRP conjugate (Zymed Laboratories, San Francisco, CA). The Ags were detected using tyramide-FITC or tyramide-Cy3 (NEN Life Science Products, Boston, MA) according to manufacturer’s instructions. For double immunofluorescence staining after the development with the first Ab, sections were blocked with Avidin/Biotin block (Vector Laboratories) followed by incubation with 2% H₂O₂. The sections were subsequently stained with the second primary Ab in a similar manner as described above with proper species-specific secondary Ab. The slides were developed with tyramide-FITC or tyramide-Cy3 (NEN Life Science Products). At the end of the staining, slides were washed and incubated with DAPI (Molecular Probes, Eugene, OR), and mounted with Fluoromount G (Southern Biotechnology Associates, Birmingham, AL). The stained slides were analyzed by fluorescence microscope (Orthoplan 2; Leitz, Wetzlar, Germany) using ×20 or ×40 objective lens.

Digoxygenin (DIG)-labeled CCL9 anti-sense and sense probes were prepared by in vitro transcription using MAXIscript T7/SP6 Kit (Ambion, Austin, TX). In brief, a region of the CCL9 coding sequence was amplified from cDNA prepared from Peyer’s patches by PCR using forward (5′-CATATGATCACACATGCAACAGAGACAA-3′) and reverse (5′-TTATTGTTTGTAGGTCCGTGGTTG-3′) primers. The PCR product was purified and cloned into pGEM-T-EASY vector (Promega, Madison, WI), and sequenced to confirm 100% match to the previously published sequence (GenBank accession number NM_011338). The restriction enzymes, NdeI and XbaI, were used to linearize the plasmid for the generation of anti-sense and sense probes, respectively. The transcript yield and integrity of the probes were determined by spectrophotometric analysis and electrophoresis, respectively. Mouse tissues were first fixed for 3 h in 4% paraformaldehyde in 0.14 M Sorenson’s phosphate buffer and then for overnight in 4% paraformaldehyde/30% sucrose in 0.14 M Sorenson’s phosphate buffer. Fixed tissues were snap-frozen in OCT compound and stored at −80°C. Sections of 7 μm thickness were cut and placed onto poly-l-lysine (Sigma-Aldrich, St. Louis, MO)-coated glass slides and in situ hybridization was performed as previously described (11). In brief, sections were treated with 0.3% Triton X-100, pretreated with Proteinase K (Roche, Indianapolis, IN) and acetic anhydride, prehybridized, hybridized overnight at 58°C with the DIG-labeled riboprobes, washed at high stringency, incubated with anti-DIG Ab conjugated to alkaline phosphatase (Roche), and developed with nitro blue tetrazolium/5-bromo-4-chloro-3-indole-phosphate (NBT-BCIP).

DCs were prepared from Peyer’s patches of naive 6- to 8-wk-old BALB/c mice as previously described (1). Cells magnetically selected on the basis of CD11c expression were then stained with DC lineage markers, FITC-conjugated anti-CD8α, PE-conjugated anti-CD11b and PE-Cy5-conjugated anti-B220 Abs and CD8α⁺CD11b⁻B220⁻, CD8α⁻CD11b⁺B220⁻, and CD8α⁻CD11b⁻B220⁻ DCs were isolated by flow cytometric sorting performed on a FACSVantage sorter. Sorted DCs were routinely 98–100% pure. These purified DC subsets were used for total RNA isolation and for chemotaxis assays.

Epithelial cells from the small intestine were collected in a procedure described previously (1). Briefly, small intestines of mice were collected and washed thoroughly with PBS. Peyer’s patches were removed from the entire length of the small intestine. The intestinal segments devoid of Peyer’s patches were cut into 1 cm pieces. The Peyer’s patches and the intestinal segments were placed separately in a solution containing 5 mM EDTA, 145 μg/ml DTT, and 10% heat-inactivated FCS, and shaken for 1h at 37°C. Supernatants were collected and washed twice in PBS. Cells were counted by hemocytometer and were determined to be of epithelial cell morphology. Equal numbers of cells were used for RNA isolation as described below.

The relative expression levels of chemokines and chemokine receptor mRNA by intestinal cells were analyzed as previously described (1). In brief, total RNA was isolated from epithelial cells or from sorted DC subsets using RNeasy Mini kit (Qiagen, Valencia, CA) and single-stranded cDNA was synthesized using Superscript II reverse transcriptase (Invitrogen, San Diego, CA). The amounts of cDNA that generate equal band intensity to 19.4 pg of the competitive plasmid pMCQ DNA (kindly provided by Dr. D. Shire (12)) (see Fig. 3,B), or to the indicated amounts of plasmid pMCQ DNA (Fig. 3 A), were thus determined for the different sorted DC subsets using the β₂-microglobulin primers, and were used for the amplification of chemokine receptors using specific primers for CCR1 (forward 5′-TGCTGTAAGAGCCTTTGGGG-3′; reverse, 5′-CTTGTAGGGGAAATGAGGGC-3′), CCR2 (forward 5′-GAAGGGGCCACCACACCG-3′, reverse 5′-GGCCACAGGTGTAATGGTG-3′), CCR5 (forward 5′-GGGTCAGTTCCGACCTATAG-3′, reverse 5′-GAGTGTGTGGAAAATGAGGAC-3′), CCR6 (forward 5′-CCATGACTGACGTCTACCTGTTGAACA-3′, reverse 5′-GAACAGCTCCAGTCCCATACCCAGCAG-3′), CCR7 (forward 5′-GCTCAACCTGGCCGTGGCAGACATCC-3′, reverse 5′-CCACTTGGATGGTGATCAAGGCCTCC-3′), CCR9 (forward 5′-CTTCCCTTCTGGGCCATTGCTGC-3′, reverse 5′-ACAGTGATGGTCACCTTGAGGGCC-3′), CCR10 (forward 5′-GGCCCTGACTTTGCCTTTTG-3′, reverse 5′-GCTGCCAGTAGATCGGCTGT-3′), CCL20 (forward 5′-GGCAAGCGTCTGCTCTTCC-3′, reverse 5′-GCCTAAGAGTCAAGAAGATGT-3′) CCL9 (forward 5′-ATGAAGCCTTTTCATACTGCCCTC-3′, reverse 5′-TTATTGTTTGTAGGTCCGTGGTTG-3′), or HPRT (forward 5′-GACACTGGTAAAACAATGCAA-3′, reverse 5′-TATCCAACACTTCGAGAGGT-3′).

Chemotactic ability of the FACS-sorted DC subsets was analyzed using the ChemoTx system (96-well ChemoTx Chamber; NeuroProbe, Gaithersburg, MD) performed as previously described (1). In brief, sorted DCs were resuspended in RPMI 1640 supplemented with 1% FCS and 25 mM HEPES. Chemokines at 0.5 μg/ml (CCL9; PeproTech, Rocky Hill, NJ) or 1.0 μg/ml (CCL21; PeproTech) were placed in the lower chamber, and a filter with 5 μm pore size was placed on top. Aliquots of 2 × 10⁴ cells/well were applied to the filter’s top surface, and the plate was incubated at 37°C in 5% CO₂ for 4h. The cells that migrated to the bottom chamber were counted using an inverted microscope. Each assay was performed in triplicates, and the results were expressed as the mean number of cells that migrate to the lower chamber ± SD.

To block CCL9 in vivo, CCR6-deficient or WT control littermate mice (strain 3) received three injections i.p. of 50 μg of purified goat Ab to CCL9 or control goat IgG on 7, 5, and 3 days before sacrifice. The Peyer’s patches were collected, frozen in OCT, and analyzed by immunofluorescence microscopy for detection of CCL9 and CD11b.

Normally distributed continuous variable comparisons were performed using the Student t test.

To determine the chemokines secreted specifically by the FAE over the dome regions of the Peyer’s patches, we first examined the expression of several candidate chemokines. The initial screening for chemokines preferentially expressed by the Peyer’s patches was conducted by hybridization of fluorescent-labeled cDNA isolated from the spleen and the Peyer’s patches of normal BALB/c mice to DNA microarray. Among other differentially expressed genes, several chemokine genes were found to be up-regulated in the Peyer’s patches compared with the spleen (data not shown). To examine the expression of these chemokines and others that have been previously shown to be expressed by epithelial cells (1, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23), cryosections of mouse Peyer’s patches from normal BALB/c mice were stained with goat polyclonal Abs to C10 (CCL6), thymus-expressed chemokine (CCL25), thymus and activation-regulated chemokine (CCL17), cutaneous T-attracting chemokine (CCL27), mucosa-associated epithelial chemokine (CCL28), MIP-3α (CCL20), and MIP-1γ (CCL9). This analysis revealed two chemokines that are specifically expressed near the FAE, CCL20, and CCL9. We have previously identified by in situ hybridization that the mRNA for chemokine CCL20 is expressed by the FAE on murine Peyer’s patches (1, 3). Both the CCL20 (Fig. 1,B) and CCL9 (Fig. 1,A) proteins were detected within the FAE as well as beneath this epithelium on structures that are suggestive of extracellular matrices in the Peyer’s patches (24). The other chemokines tested were not detected at a significant level within either the FAE or the small intestinal epithelium (data not shown), with the exception of CCL25, which was detected on both of these epithelial cell types (Fig. 1 C) as previously described (15, 25, 26, 27).

To confirm our observation of CCL9 expression from the immunofluorescence analysis, we performed in situ hybridization on normal mouse Peyer’s patches using anti-sense or sense RNA probes specific for CCL9. The CCL9 mRNA was distinctly observed within the FAE, but not in the epithelial cells of the villi (Fig. 2,A). As previously described (1, 3), the CCL20 mRNA was detected specifically over the dome region FAE (Fig. 2,C). Neither CCL9 (Fig. 1,B) nor CCL20 (Fig. 1,D) sense probes bound to the serial tissue sections examined at these sites. To evaluate the relative levels of mRNA expression by the FAE and intestinal epithelial cells of the villi, RNA from epithelial cells covering the Peyer’s patches, or those from intestinal segments without Peyer’s patches, were isolated and the relative amounts of RNA were equalized by competitive PCR (1). This analysis demonstrated that significantly higher levels of the CCL20 and CCL9 mRNA were present within the FAE compared with the villus epithelium (Fig. 3 A).

To investigate the possible role of CCL9 in DC recruitment to the dome region of the Peyer’s patches, three DC subsets were purified from the Peyer’s patches of BALB/c mice by flow cytometric cell sorting and RNA was isolated from these sorted subsets. The level of mRNA transcripts for various CC chemokine receptors were examined by competitive RT-PCR (1). The amount of cDNA prepared from DC RNA was normalized to a known concentration of competitor plasmid DNA using a pair of primers specific to a housekeeping gene β₂-microglobulin (Fig. 3,B). Using cDNA samples normalized in this manner from the three DC subsets, PCR was conducted using specific primer pairs to CCR1, CCR2, CCR5, CCR6, CCR7, CCR9, or CCR10. All three DC subsets were found to express CCR1, CCR2, CCR7, and CCR9, whereas CCR6 and CCR10 expression was confined to the CD11b⁺ and DN (CD11b⁻/CD8α⁻) subsets. These results are consistent with our previous finding for CCR6 and CCR7 (1). Previous studies showed that CCL3 and CCL9 are capable of eliciting desensitization to each other, suggesting that the putative receptors for CCL9 are shared by CCL3 (7). The major receptors for CCL3 have been identified to be CCR1 and CCR5 (28, 29). In Fig. 3 B, one of the putative receptors for CCL9, CCR1, was found to be expressed by all three Peyer’s patch DC subsets, and at particularly high levels by the CD11b⁺ DCs. Thus, these results suggest that the Peyer’s patch DCs possess the potential to migrate toward CCL9 through their expression of CCR1.

To determine the functional relevance of the CCR1 expression by the CD11b⁺ DCs in mouse Peyer’s patches, three DC subsets were isolated and purified by flow cytometric cell sorting, and were assessed for their ability to migrate toward CCL9 in a chemotaxis chamber. As previously described (1), all three DC populations migrated robustly in response to CCL21 (Fig. 4). The ability of the CD11b⁺ DCs to migrate toward CCL9 was detected, but was variable in five similar experiments that were conducted. In some of these experiments, migratory capacity of CD11b⁺ DCs toward CCL9 was determined to be statistically significant (see p values). A paired Student’s t test performed on the numbers of cells that migrated toward CCL9 vs the medium control from all five experiments revealed that the CD11b⁺ DC migration to CCL9 was statistically significant (p = 0.046). None of the other DC subsets migrated significantly toward CCL9. Therefore, CD11b⁺ Peyer’s patch DCs, but not CD8α⁺ or DN DCs, had the tendency to migrate toward CCL9.

Our observation that CCL9 was secreted specifically by the FAE and that it can mediate chemotactic migration of the CD11b⁺ DCs in vitro suggested that CCL9 may play a role in recruiting these cells to the subepithelial dome regions of the Peyer’s patches. However, absence (3) or reduction (4) in the dome region CD11b⁺ DCs have been reported in mice deficient in CCR6, suggesting that CCR6 recognition of CCL20 is necessary for the recruitment of these cells. To confirm the absolute requirement of CCR6-CCL20 interaction in CD11b⁺ DC recruitment to the dome regions of Peyer’s patches, we re-examined the tissues of CCR6-deficient mice generated by four independent laboratories that used separate knockout constructs (Fig. 5, E–H). To our surprise, using a sensitive method for double immunofluorescence labeling, we detected CD11b⁺ DCs in the dome regions of CCR6^−/− mice from all four groups. As described by Varona et al. (4), we also observed CD11b⁺ DCs in the interfollicular regions of some of the CCR6^−/− mice (Fig. 5,E). Although it is difficult to quantitatively assess the frequency of the CD11b⁺ DCs in the dome regions of CCR6^−/− mice as compared with the WT littermate controls, analysis of a total of 30 (9), 10 (30), 6 (4), or 4 (3) mice revealed no great differences in the presence of CD11b⁺ DCs in their Peyer’s patch dome regions. Although in some cases there appeared to be an increased number of CD11b⁺ DCs in the dome regions of the CCR6^−/− mice compared with their WT littermate controls (Fig. 5, E vs A, or F vs B, respectively), enumeration of the CD11b⁺ DCs in the dome regions revealed no statistically significant differences between the knockout and WT mice in all four groups (Table I). Therefore, our collective analysis of the CCR6-deficient mice demonstrate that CCR6 is not absolutely required for the recruitment of CD11b⁺ DCs to the subepithelial dome regions of the Peyer’s patch and that other chemokines such as CCL9 may indeed play a role in their recruitment in vivo.

To examine the role of CCL9 in CD11b⁺ DC recruitment, we attempted to block the chemokine in vivo using an Ab specific to the CCL9. To ensure complete blockage of CCL9, either WT or CCR6-deficient mice were treated with 50 μg of anti-CCL9 or control goat IgG Abs three times on 7, 5, and 3 days before sacrifice. These polyclonal Abs have been shown to mediate neutralization of mouse CCL9 bioactivity in vitro. Peyer’s patches from CCL9-blocked or control IgG treated mice were examined for the presence of CCL9 and the distribution of CD11b⁺ DCs by double immunofluorescence analysis. The treatment with anti-CCL9 Ab resulted in significantly reduced levels of CCL9 detected in the dome regions (Fig. 6, C and D) compared with the goat IgG treated controls (Fig. 6, A and B). Some residual CCL9 was detected just beneath the FAE in all the Peyer’s patches examined. The detection of the reduced levels of the CCL9 molecule was not due to blocking of epitopes by the in vivo administered goat-anti-CCL9 Ab since endogenously bound goat IgG was not detected in situ using anti-goat IgG detection Ab (data not shown). A concomitant decrease in the number of CD11b⁺ cells (red cells) were detected in the CCL9-blocked mice (Fig. 6, C and D), but not in the goat IgG-treated mice (Fig. 6, A, B, E–G). A closer examination of the dome area of the Peyer’s patch in goat IgG treated mice as depicted in Fig. 6,A revealed that the CCL9 staining was found on the extracellular matrix structures below the FAE and on vessel structure through the dome region (E), whereas CD11b staining was found on the dome region DCs (F). The effect of anti-CCL9 Ab treatment on the CD11b⁺ DCs was similar in the CCR6^−/− mice (Fig. 6, A and C) and the WT littermate control mice (Fig. 6, B and D). The expression of other epithelially derived chemokines examined, such as CCL20 and CCL25, were not altered as a result of anti-CCL9 treatment (data not shown). Therefore, anti-CCL9 Ab treatment of mice resulted in a significant reduction of CCL9 in vivo, and a concomitant decrease in the CD11b⁺ DC number in the dome regions.

Chemokines play an integral role in the homeostatic recruitment of various cell types in vivo. The FAE is a specialized epithelium that contains M cells and serves as the portal of entry for intestinal Ags and pathogens. Not surprisingly, professional APCs such as the DCs reside just beneath (1, 31) or even within (2, 32) the FAE, forming a fortress of sentinels that are equipped for pathogen detection and immune induction. In particular, the CD11b⁺ DCs and the CD11b⁻/CD8α⁻/CD11c⁺ DCs are localized in this area, likely in response to chemokines secreted by the FAE. Previous studies demonstrated that the FAE specifically expresses CCL20 (1, 3), and that CD11b⁺ DCs express the receptor for CCL20 (CCR6) and migrate toward CCL20 (1). Moreover, mice lacking CCR6 were reported to have no (3) or reduced (4) CD11b⁺ DCs in the dome regions, suggesting a critical role of CCR6 in mediating recruitment of CD11b⁺ DCs toward CCL20 secreted by the FAE.

In this study, we describe another chemokine, CCL9, which is specifically secreted from the FAE. We demonstrated that CD11b⁺ DCs isolated from the Peyer’s patches express CCR1, a putative receptor for CCL9, and that they possess migratory capacity toward CCL9 in vitro. Further, neutralization of CCL9 in vivo resulted in the reduction of CD11b⁺ DCs in the dome regions. To our surprise, re-evaluation of the Peyer’s patches from CCR6-deficient mice, derived independently by four separate laboratories, revealed that CD11b⁺ DCs can be found in the subepithelial dome regions of these mice. The difference in the results obtained in this study and in the previously published studies (3, 4) likely stems from the highly sensitive nature of the immunofluorescence technique employed in this study. First, the levels of CD11b expression by the DCs in the dome regions are much lower than those found on neutrophils or macrophages in the same tissue, necessitating the amplification of the signal with the Tyramide system. Secondly, the ability to detect the CD11b⁺ DCs in the dome region largely depends on having a cross section of the Peyer’s patches through the center of the dome region perpendicular to the tunica muscularis. This requires precise positioning of the tissues within the embedding medium and a survey of the Peyer’s patches from several mice to detect enough perfect dome regions to make statistically sound conclusions.

Very little is known regarding the function of the chemokine CCL9. The CCL9 molecule was originally identified by three separate laboratories (5, 6, 7). CCL9 is an unusual chemokine in that it is constitutively expressed by a variety of tissues and has been reported to circulate in the blood of normal mice at high concentrations (∼1 μg/ml) (7). Unlike CCL3, CCL9 is not induced by treatment of a DC cell line with LPS (8). Our in situ hybridization and immunofluorescence analysis of mouse small intestine and other lymphoid tissues (peripheral lymph nodes, spleen) revealed specific and high levels of expression in the FAE, but not in other epithelial cells or lymphoid tissues (data not shown). Since the immunofluorescence procedure can only detect tissue bound CCL9, and that it does not detect soluble CCL9, we have examined the amounts of CCL9 in tissues such as the spleen, mesenteric lymph nodes, Peyer’s patches, small intestine, skin and vaginal tract by ISH. This analysis revealed that CCL9 mRNA was only detectable in the FAE and not in the other organs examined (data now shown). It is interesting to note that the immunofluorescence analysis of the CCL9 protein revealed that this chemokine was found mostly attached to extracellular matrix structures in the dome region, but not within the FAE itself (Figs. 1 and 6), despite its mRNA expression to be exclusively localized within the FAE (Fig. 2). These results demonstrated that CCL9 mRNA is transcribed by the FAE and the translated product can be secreted toward the dome region of the Peyer’s patch, some of which can attach to the extracellular matrix and perhaps to endothelial cells (Fig. 6 E). Therefore, at least in the mouse Peyer’s patches, CCL9 was highly expressed by the FAE in normal healthy mice and could be involved in the homeostatic recruitment of cells.

Our current study demonstrated that in vivo Ab neutralization of CCL9 resulted in a dramatic reduction of the CD11b⁺ DCs in the dome regions of treated mice. Interestingly, the residual CD11b⁺ DC populations detected in the CCL9-depleted mice were found immediately beneath the FAE, an area that still contained remaining CCL9 protein (Fig. 6). The in vivo treatment with the Ab to CCL9 presumably neutralized CCL9 secreted from the FAE, as this Ab has been shown to effectively block the bioactivity of CCL9 for THP1 cell migration in vitro. To investigate the hypothesis that CCL9 may be necessary for CD11b⁺ DC distribution to the dome regions of Peyer’s patches, we examined mice deficient in the putative receptors for CCL9. Our preliminary analysis of the CCR1^−/− and the CCR5^−/− mice revealed no significant decrease in the dome region CD11b⁺ DC numbers (data not shown), suggesting either that both CCR5 and CCR1 can mediate the recruitment of CD11b⁺ DCs independently, or that an unidentified chemokine receptor exists for CCL9. No definitive study describing the receptor(s) for CCL9 has been reported to date. Thus, although our results are suggestive of the role of CCL9 in CD11b⁺ DC recruitment, the only definitive proof will come from the examination of CCL9 knockout mice.

Although we demonstrated that the Peyer’s patches of the CCR6-deficient mice possess CD11b⁺ DCs, and that anti-CCL9 treatment of WT mice resulted in a significant reduction of the CD11b⁺ DC in the dome region, it is too premature to exclude the importance of CCR6-CCL20 axis in DC recruitment toward the FAE. Since the CD11b⁺ DCs express CCR6 and migrate toward CCL20 expressed highly by the FAE (1), it is possible that CCL20 is one of the mediators of the recruitment of these cells to the dome region. Thus, we hypothesize two possible non-mutually exclusive mechanisms that could contribute to dome region recruitment of the CD11b⁺ DCs; 1) two subsets of CD11b⁺ DCs exist that express either CCR6 or CCL9-receptor, and 2) all CD11b⁺ DCs express both CCR6 and CCL9-receptor and migrate toward CCL20 and CCL9, secreted by the FAE, respectively.

As the FAE provides a unique site for transport of intestinal Ags, CCL9 may provide a critical migratory signal needed for the recruitment of CD11b⁺ DCs, which can readily capture Ag and present them to CD4⁺ T cells (2). Although the characterization of the precise mechanisms by which the chemokines orchestrate the distribution of cell types within the Peyer’s patches is still in its infancy, our results provide evidence that CCL9 needs to be considered as one of the key chemokines in this process.

We thank the members of the laboratory of Dr. N. Ruddle for their technical help with the ISH procedure, and Dr. B. Kelsall for critical review of the manuscript.