Expression of the chemokine binding protein M3 promotes marked changes in the accumulation of specific leukocytes subsets within the intestine

Abstract

Background & Aims

Chemokines are small proteins that direct leukocyte trafficking under homeostatic and inflammatory conditions. We analyzed the differential expression of chemokines in distinct segments of the intestine and investigated the importance of chemokines for the distribution of leukocytes in the intestine during homeostatic and inflammatory conditions.

Methods

We analyzed mRNA for all known chemokines in different segments of the gut by quantitative PCR. To study the effect of multiple-chemokine blockade in the gut, we generated transgenic mice that expressed the chemokine binding protein, M3 in the intestine (V-M3 mice). We used flow cytometry to evaluate the changes in the numbers of leukocytes.

Results

We observed distinct chemokine expression profiles in the six segments of the gut. Some chemokines were expressed throughout the intestine (CCL28, CCL6, CXCL16, and CX3CL1), whereas others were expressed preferentially in the small (CCL25 and CCL5) or large intestine (CCL19, CCL21 and CXCL5). Expression of the chemokine blocker M3 in intestinal epithelial cells resulted in reduced numbers of B and T cells in Peyer’s patches, reduced numbers of intraepithelial CD8αβ⁺/TCRαβ⁺ and CD8αα⁺/TCRαβ⁺ T cells, and reduced numbers of lamina propria CD8⁺ T cells. Strikingly, M3 expression markedly reduced the number of eosinophils and macrophages in the small and large intestines. DSS treatment of control mice led to marked changes in the expression of chemokines and in the number of myeloid cell in the colon. These cellular changes were significantly attenuated in the presence of M3.

Conclusion

Our study reveals a complex pattern of chemokine expression in the intestine and indicates that chemokines are critical for leukocyte accumulation in the intestine during homeostasis and inflammation.

Introduction

The intestine plays an important role in host protection and immune system modulation. To detect pathogenic organisms a large number of immune cells circulate through the gastrointestinal tract, but the mechanisms underlying the trafficking of immune cells is still poorly understood^{1, 2}. It is well established that α4β7 integrin and its ligand MAdCAM-1 play an important role in imprinting lymphocytes for gut homing^3–6. However, the integrin imprinting system alone is not sufficient to explain the large diversity of cell types in the small and large intestine^{1, 7}.

A role for chemokines in immune cell trafficking to the intestine has been gradually appreciated in the last decade^{1, 8–10}. Chemokines comprise a family of small basically charged chemoattractant proteins that regulate trafficking of leukocytes under both homeostatic and inflammatory conditions. It is well established that chemokines and their receptors control influx of T cells and B cells into Peyer’s patches. T cell homing to the Peyer’s patches (PP) depends on CCR7 and its ligands, CCL19 and CCL21^11–13, whereas B cell homing to PP depends on the coordinated signaling of CCR7, CXCR4, CXCR5, and CCR6¹⁴.

Lymphocyte homing to lamina propria (LP) and intestinal epithelium is less clearly defined. CCL25 is highly expressed in small intestinal epithelium^15–17 and most lymphocytes in the small intestine express CCR9¹⁸, the only known receptor for CCL25. Upon activation in gut associated lymphoid tissues, T cells express CCR9 and α4β7 integrin, the addressins that target T cell homing to the small intestine¹⁹. Either blocking CCL25 function or deleting CCR9 impairs the homing to small intestinal mucosa of recently activated CD8αβ⁺ T cells, recent thymic emigrant CD8αβ⁺ T cells, and activated CD4⁺ T cells^20–23. CCL25 also promotes the generation of CD8αα⁺ T cells in intestinal epithelium and enhances their adhesion to intestinal epithelium^{24, 25}. However, mice deficient in CCR9 or CCL25 do not show a dramatic decrease of T cells in either intestinal epithelium or LP. The changes are surprisingly subtle and restricted to reduction in the intraepithelial population of TCRγδ⁺ T cells^26–28 and a mild reduction in LP CD8⁺ T cells²⁷. The lack of dramatic change in T cell homeostasis in CCR9 and CCL25 deficient mice suggests that other chemokines may compensate for CCL25 and CCR9 deficiency to promote T cell recruitment into the small intestine. Indeed, CXCR3 deficient mice have reduced numbers of CD8αβ⁺ T cells in the intestinal epithelium²⁹. Another chemokine highly expressed in the intestinal epithelium is CCL28 (mucosa-associated epithelial chemokine, MEC). In contrast to CCL25, which is expressed only in the small intestine, CCL28 is expressed in both small and large intestine^30–32. CCL28 specifically recruits IgA⁺ plasmablast cells by interacting with CCR10^33–35. The combined action of CCL25 and CCL28 to recruit IgA⁺ plasmablast cells to the small intestine may be essential for IgA mediated immunity³³.

Mechanisms governing homeostatic leukocyte recruitment to the colon are even less clearly defined than the small intestine. Trafficking of T cells into the murine colon appears to be independent of CCR9⁸. Blocking antibodies to CXCL12 or its receptor CXCR4 significantly inhibit leukocyte adhesion to microvessels in the ileum and colon under both normal and inflammatory states³⁶, but a requirement for CXCL12 in homeostasis has not been established. Thus, the role of chemokines in the recruitment of lymphocytes to the colon or ileum during homeostasis remains unclear.

A clear picture of the role of chemokines in homeostatic trafficking of leukocytes in the intestine has not yet emerged. To date, there is no detailed mapping of chemokine expression in different areas of the intestine, and the role of many chemokines has not yet been studied. Furthermore, many studies utilize knockout animals in which the deficiency of chemokines or chemokine receptors is general, rather than restricted to the intestine. Thus, the interpretation of some of the studies is marred by the fact that the results may reflect generalized defective trafficking of leukocytes to other tissues or defective leukocyte development and differentiation. In addition, because of the redundancy in mechanisms regulating cell recruitment and trafficking, genetic deletion of single chemokines or chemokine receptors may result in subtle phenotypes. In this particular context, tools that could promote simultaneous blockade of several chemokines would clearly be useful to elucidate the roles of chemokines in cell recruitment to the intestine.

To begin evaluating the contribution of the chemokine system to homeostatic and inflammatory leukocyte trafficking in the intestine, we mapped the expression of all murine chemokine ligands in six intestinal segments. We then devised a strategy to block chemokines locally in the intestine, by generating transgenic mice expressing M3, a chemokine blocker, in the intestine. M3 is a murine gamma herpesvirus 68 (MHV-68) encoded chemokine binding protein that binds to and inhibits multiple chemokines from all subfamilies (namely CC, CXC, C, and CX3C)^37–40, by interacting with the chemokine N-loop^{41, 42}, thus blocking the binding of chemokines to their receptors. M3 also inhibits chemokine-GAG interactions and disrupts pre-formed chemokine gradients⁴³. Our laboratory has shown that M3 blocks chemokine function in vivo, and that its expression attenuates inflammatory and autoimmune diseases^44–47.

We demonstrate here that chemokines have three distinct expression patterns in the intestine: some chemokines are expressed throughout the intestine, whereas others are expressed primarily in the small intestine or in the large intestine. Blockade of chemokine function by M3 resulted in changes in the numbers of lymphocytes in PP, IEL and LP, and affected specific myeloid populations in the LP. Chemokine blockade by M3 also markedly attenuated the changes in intestinal cellularity subsequent to colitis induced by dextran sodium sulfate (DSS) treatment. Together, our results strongly implicate the chemokine system as a central determinant of leukocyte accumulation in the gut.

Materials and Methods

Mice

The plasmid containing the mouse villin promoter (pBS-Villin) has been previously described⁴⁸. The M3 gene was amplified by PCR with a BsiWI site at the 5′ end and MluI at the 3′ end. The PCR product was cleaved with BsiWI and MluI and cloned into pBS-Villin and sequences were verified. The transgene was released with SalI, microinjected into BDF1 eggs (The Jackson Laboratory, Bar Harbor, Maine), which were transferred into oviducts of ICR foster mothers (Charles River Laboratories, MA). Identification of the transgenic mice was accomplished by PCR amplification of mouse tail DNA using specific primer sets as previously described⁴⁴. For the DSS experiments, we used V-M3 mice backcrossed to C57BL/6 mice for 10 generations (referred to as 6V-M3 mice). All experiments involving animals were performed following guidelines of the Animal Care and Use Committee of Mount Sinai School of Medicine.

Determination of M3 binding affinities to chemokine ligands

Histidine-tagged M3 protein was generated in Hi5 insect cells infected with a recombinant baculovirus expressing M3³⁹. The protein was purified from cell culture supernatant in a Ni-NTA resin (Qiagen, CA). The chemokine binding to M3 was determined in Ni²⁺-coated 96-well FlashPlates (PerkinElmer, MA) incubated overnight with M3 protein (4 ng/well) as described⁴². To calculate the binding affinities, the M3-coated FlashPlates were incubated for 4 h at room temperature with 200 pM of iodinated CCL3 (PerkinElmer, MA) in the presence of various concentrations of unlabelled chemokines (0.025 to 250 nM) purchased from PeproTech or R & D Systems. Bound iodinated CCL3 was determined after addition of scintillant on a Packard TopCount Microplate Counter. The affinity constants were calculated using the data analysis package PRISM3 from Graphpad. The results are representative of at least two experiments and triplicate samples in each experiment.

Surface plasmon resonance (SPR)

The interaction between M3 and the murine chemokines CCL5, CCL6, CCL11, CCL22, CCL24, CCL25, CCL28, and CXCL12α was analysed by SPR using a Biacore 2000 biosensor. Briefly, 500 ng of affinity-purified recombinant M3 was amine-coupled to the surface of a CM5 chip (GE Healthcare, NJ). Coupling was monitored in real time and stopped when it reached approximately 2,000 response units (RU). Recombinant chemokines were injected at 100 nM in HBS-EP buffer [10 mM Hepes, 150 mM, NaCl, 3 mM EDTA, 0.005% (vol/vol) surfactant P20, pH 7.4] at a flow rate of 10 μl/min, and association and dissociation were monitored. The chip surface was regenerated after the dissociation phase by injecting 10 mM glycine-HCl, pH 2.0. All Biacore sensorgrams were analyzed with the software BIAevaluation 3.2. Bulk refractive index changes were removed by subtracting the reference flow cell responses. The average response of a blank injection was subtracted from all sensorgrams to remove systematic artefacts.

Immunostaining

Immunostaining of M3 in sections of fresh tissue frozen in OCT was performed as previously described⁴⁶. Briefly, M3 primary antibodies were incubated for 1 h at room temperature, followed by incubation with the appropriate fluorescent-labeled secondary antibodies for 30 min.

Isolation of lamina propria leukocytes (LPL) and intraepithelial lymphocytes (IEL)

Lamina propria leukocytes (LPL) were isolated as described previously⁴⁹. Briefly, the small intestine was removed, flushed with ice-cold calcium- and magnesium-free Hank’s balanced salt solution (HBSS) supplemented with 2% FBS, and freed of fat, mesentery, and PP. The tissue was then cut into small pieces about 0.5 cm in length and incubated at 37°C with shaking for 20 min in HBSS containing 1 mM dithiothreitol (DTT), followed by PBS with 1.3 mM EDTA for 1 h at 37°C. Fragments of intestine were further incubated in 7 ml of collagenase (1.6 mg/ml) in RPMI for 1 h at 37°C and LPL were isolated at the interface of a 44/66% Percoll density gradient. To prepare IEL, the tissues were cut into small pieces about 0.5 cm in length and incubated at 37°C with shaking for 20 min in HBSS containing 1 mM DTT. The tissues were removed and put into another tube with HBSS containing 1 mM DTT and incubated at 37°C with shaking for another 20 min. The supernatants from the above two incubations were combined and the cells were collected by centrifugation at 1400 rpm for 5 min at 4°C. The cells were washed twice in RPMI and purified on a 44/66% Percoll density gradient.

RNA extraction and quantitative PCR (Q-PCR)

We collected five to seven samples (0.5 cm in length) from each segment of the intestine. The segments were sampled as follows: duodenum, 7 consecutive segments; jejunum, starting 6 cm from proximal end of the small intestine, samples every 2 cm (n=7); ileum, starting 0.5 cm from the distal end, sampled every 2 cm (n=5); cecum, divided into 5 even pieces, all 5 samples used for RNA preparation; proximal colon, starting from the proximal end of the colon, 5 consecutive samples; distal colon, starting 0.5 cm from the rectum, 5 consecutive samples (Fig. 1A). Total RNA from segments of the intestine was extracted using the RNeasy Mini Kit (Qiagen, CA) according to the manufacturer’s instructions. The cDNA from these samples were pooled for Q-PCR analysis. Q-PCR was performed as previously described⁴⁶. The data in figures are representative of three independent experiments.

Figure 1. Expression pattern of chemokines in intestinal segments.

A) Sampling of intestinal segments for expression analysis is schematically depicted. Total RNA was isolated from 5 to 7 pieces of duodenum, jejunum, ileum, cecum, proximal colon, and distal colon for each mouse (n=3). B) Differential expression of chemokines in intestinal segments. The cDNA samples from 5 to 7 pieces of each intestinal segment were pooled for analysis of mRNA. The mRNA expression values were normalized to ubiquitin in each sample and the data shown is the mean of three independent samples isolated from three BDF1 mice.

Western blotting

Proteins were extracted from different segments of the intestine and run on 4–20% pre-cast gradient gel (Biorad, CA). After transfer to a PVDF membrane (Biorad, CA), M3 protein was detected using a rabbit anti-M3 antibody and HRP anti rabbit IgG⁴⁶. An antibody against βactin (Abcam, MA) was used to verify the amount of loaded protein.

FACS analysis

Cells from Peyer’s patches or LP were incubated with Fc block (eBioscience Cat. #14-0161-85) for 15 min at room temperature, and were then stained with different antibodies to detect different subsets of leukocytes. The antibodies used were CD45-APC-Cy7 (BD Biosciences, Cat. #557659), CD3-FITC (eBioscience, Cat. #11-0031-85), CD19-PE (BD Biosciences, Cat. #557399), CD4-PE-cy7 (BD Biosciences, Cat. #552775), CD8a-APC (BD Biosciences, Cat. #553035), CD8b-FITC (eBioscience, Cat. #11-0083), SIGLEC-F-PE (BD Biosciences, Cat. #552126), CD11c-FITC (eBioscience, Cat. #11-0114-82), CD11b-PE-Cy7 (BD Biosciences, Cat. #552850), F4/80-APC (eBioscience, Cat. #17-4801-82), TCRβ-APC-Cy7 (eBioscience, Cat. #275961), TCRγδ-PE (eBioscience Cat. #12-5711-82) and CCR3-APC (R&D systems, Cat. #83101). Events were acquired on a Becton Dickinson FACScan and analyzed using FlowJo software.

Statistical analysis

Statistical analysis of the data was performed using Prism 2.0c (GraphPad Software, CA). Data in the text are given as means ± standard errors unless otherwise stated. Unpaired Student’s t test was used to determine statistical significance. Differences were considered significant when p < 0.05.

Results

Chemokines are differentially expressed within intestinal segments

To obtain a more complete picture of the chemokine network in the intestine during homeostasis, we analyzed the expression of chemokines by quantitative PCR (Q-PCR) in RNA isolated from duodenum, jejunum, ileum, cecum, proximal colon, and distal colon of BDF1 mice (Figure 1). The expression level of all chemokines in intestinal segments is shown in Suppl. Figure 1. Most chemokines were either not expressed or were expressed at low levels. Among the chemokines expressed at medium to high levels there were three different patterns of expression (Figure 1B): 1) High level expression throughout the intestine (CCL28, CCL6, CXCL16, and CX3CL1); 2) Predominant expression in the small intestine (CCL25 and CCL5); and 3) Predominant expression in the large intestine and ileum (CCL19, CCL21, and CXCL5). These results indicate that the expression pattern for chemokines varies throughout the length of the intestine and suggest that these different patterns may be responsible for accumulation of different leukocyte subsets in these segments.

The chemokine binding protein M3 binds to many chemokines expressed in the intestine

To investigate the contribution of chemokines to leukocyte homeostasis in the intestine, we used the chemokine binding protein M3, which interacts with many members of all chemokine subfamilies^{39, 40} and blocks their activity. To explore the potential of M3 to interfere with the intestinal chemokine network, we analyzed the binding of M3 to the chemokines CCL5, CCL6, CCL11, CCL22, CCL24, CCL25, CCL28, and CXCL12α by surface plasmon resonance (SPR), using tumor necrosis factor α (TNFα) as a negative control. M3 interacts with all chemokines tested with the exception of CXCL12α (Suppl. Figure 2)^{39, 40}. The affinities of the interactions were quantified by a scintillation proximity-binding assay (Suppl. Table 1). M3 binds with high affinity (in the pM-nM range) to CCL5, CCL6, CCL11, CCL22, CCL24, and CCL25. The affinity of M3 for CCL28 was below the detectable level for this method. Affinities of M3 to CXC chemokines were lower than those of M3 to CC chemokines in general (Suppl. Table 1).

Generation of transgenic mice expressing M3 in IEC

To probe the roles of the chemokine network in the intestine in vivo, we generated transgenic mice expressing M3 in IEC by placing M3 under the control of the mouse villin promoter (Figure 2A), which has been previously shown to target transgene expression to IEC of both the small and large intestine⁴⁸. Eight founders were generated from microinjection of the transgene into fertilized mouse eggs and seven transgenic lines were established (referred to as V-M3 mice). All seven lines expressed the transgene at different levels by Q-PCR (data not shown). We selected the line with the highest expression (Line #12) for further study. The transgene was expressed predominantly in the small intestine (Figure 2B and data not shown), with low levels of expression in the kidney, in agreement with previous reports⁴⁸. M3 protein was present in all intestinal segments including duodenum, ileum, and colon (Figure 2C). Immunostaining with an M3 antibody confirmed M3 expression in IEC of V-M3 mice (Figures 2D, WT and 2E, TG).

Figure 2. Expression of M3 in the intestine of V-M3 mice.

A) Diagram of the V-M3 transgene. p(A) represents SV40 poly A sequences. B) M3 mRNA expression in different tissues of V-M3 mice. The values were normalized to ubiquitin in each sample. H, Heart; Lu, Lung; K, Kidney; Br, Brain; LI, Large intestine; SI, Small intestine. C) Western blot analysis of M3 expression in different segments of intestine of V-M3 mice. D & E) Representative immunostaining for M3 in small intestine of WT (D) and V-M3 (E) mice.

M3 expression in IEC decreases the size of PP in V-M3 mice

V-M3 mice were healthy and grew normally. Upon necropsy, we observed that, although V-M3 mice had a normal number of PP, the sizes of the PP were much reduced compared to controls (Figure 3A & 3B). The sizes of the V-M3 PP were smaller because they had fewer domes than WT PP but the size of individual domes of PP from V-M3 and WT mice were comparable. Flow cytometric (FACS) analysis showed that although the relative percentages of B and T cells were not changed by M3 expression (Figure 3C), the absolute B and T cell numbers were dramatically reduced in PP of V-M3 mice (Figure 3D). The B and T cell numbers in the spleen, inguinal LN and MLN were not affected by M3 expression in the intestine (Suppl. Figure 3).

Figure 3. Peyer’s patches of V-M3 mice are smaller and have fewer cells than WT Peyer’s patches.

A & B) V-M3 mice had much smaller PP (B, arrows) than WT littermates (A). C) The percentage of T and B cells in PP of WT and V-M3 mice. D) Absolute numbers of B and T cells in PP of WT and V-M3 mice. (For C & D, WT n=3, V-M3 n=4, ** p<0.01).

M3 expression alters CD8⁺ T cell homeostasis in the epithelium and LP

To investigate if M3 expression affected IEL composition, we analyzed IEL by FACS. IELs from small intestine of control mice consist primarily of CD8⁺ T cells with a low percentage of CD4⁺ T cells (Figure 4A). CD8⁺ T cells in the IEL include conventional CD8αβ⁺ T cells and unconventional CD8αα⁺ T cells^{7, 8}. The CD8αβ⁺ T cells are all TCRαβ⁺ and the CD8αα⁺ T cells include both TCRαβ⁺ and TCRγδ⁺ T cells^{7, 8}. We found that the proportions of CD4⁺ and CD8⁺ cells were altered by M3 expression. Notably, CD8⁺ T cells were reduced from 85% in the WT to 70 % in V-M3 mice, while the percentage of CD4+ T cells increased from 6% to 13% (Fig. 4A). In addition, both the proportion and absolute number of CD8αβ⁺/TCRαβ⁺ T cells and CD8αα⁺/TCRαβ⁺ cells were reduced in small intestinal epithelium of V-M3 mice (Figure 4B). However, the number of CD8αα⁺/TCRγδ⁺ T cells was not significantly altered (Figure 4B). No change was detected in any subset of colon IEL (data not shown).

Figure 4. The effects of M3 expression on lymphocyte subsets in the intestinal epithelium of V-M3 mice.

A) Representative FACS analysis of intraepithelial T cell subsets in V-M3 mice. The upper panel was gated on CD3⁺ T cells. The middle panel was gated on CD8⁺ T cells. The bottom panel was gated on CD8αα⁺ T cells. B) Quantitative analysis of intraepithelial T cell subsets in V-M3 mice. The upper panel represents the percentage of CD3⁺ cells for each T cell subset. The bottom panel shows the absolute cell number of each subset (n=4 for each group, * p<0.05, ** p<0.01).

We also analyzed the LPL populations. In WT mice, the LP contains B cells, T cells, IgA secreting plasma cells, and myeloid cells¹. M3 expression did not affect either the relative or absolute numbers of small intestine LP B or T cells (Figure 5A), or the number of IgA⁺ cells or fecal IgA levels (data not shown). However, the percentage of CD8⁺ cells in the T cell compartment of V-M3 mice was decreased 50% relative to WT mice (Figure 5B), and the absolute number of CD8⁺ T cells was also significantly reduced (Figure 5B). In the large intestine, no differences were found in either B or T cell numbers (Figure 5C) or in T cell subsets (Figure 5D). In summary, we found that M3 expression reduced the number of CD8⁺ T cells in both small intestinal epithelium and LP. However, the number of CD4⁺ T cells was increased in the small intestinal epithelium of V-M3 mice.

Figure 5. Reduction of CD8⁺ T cells in the small intestine but not large intestine lamina propria (LP) of V-M3 mice.

A) FACS analysis of B and T cells in the small intestine lamina propria of WT and V-M3 mice (n=4 for each group). B) FACS analysis of CD4 and CD8 T cells in the small intestine lamina propria of WT and V-M3 mice The analysis was gated on CD3⁺ cells (n=4 for each group, * p<0.05). C) FACS analysis of B and T cells in the large intestine of WT and V-M3 mice (n=4 for each group). D) FACS analysis of CD4 and CD8 T cells in the large intestine of WT and V-M3 mice. The analysis was performed on CD3⁺ cells.

M3 expression results in a marked reduction of eosinophils and a subset of CD11b⁺/CD11c^low/MHC class II^high myeloid cells in the LP

We analyzed the effect of M3 expression on myeloid cell homeostasis in the LP by FACS. In the small intestine of WT mice we found two distinct populations of CD11b⁺ cells, one population with high levels of CD11c (Figure 6A, CD11b⁺/CD11c^high), and a second population with low levels of CD11c (Figure 6A, CD11b⁺/CD11c^low). The CD11b⁺/CD11c^high population expressed high levels of MHC class II (MHCII), demonstrated typical dendritic cell morphology (data not shown), and was unchanged in the V-M3 mice (Figure 6A). However, the CD11b⁺/CD11c^low population was dramatically reduced to less than 1% in the small intestine LP of V-M3 mice, compared to 8–12% for WT mice (Figure 6A).

Figure 6. Effect of M3 expression on myeloid cell subsets in the LP of WT and V-M3 mice.

A & B) FACS analysis of lamina propria leukocytes in the small intestine (A) and large intestine (B) of WT and V-M3 mice. The lower FACS panel was gated on the CD11b⁺/CD11c^low subset of the upper panel. The R1, R2 and R3 populations in the right panel correspond to the gates in the lower FACS panel (n=4 for each group, * P<0.05, ** P<0.01, *** p<0.001). C) Cytospin morphology of cells from R1, R2 and R3 gates of CD11b⁺/CD11c^low cells (as shown in (A)) from the LP of WT mice. D) Analysis of CD11b⁺/CD11c^low cells from LP of WT mice for eosinophil markers SIGLEC-F and CCR3. The analysis was performed on the CD11b⁺/CD11c^low cells from LP of WT mice.

To analyze the CD11b⁺/CD11c^low population further, we stained for MHC class II and Gr-1. The CD11b⁺/CD11c^low population could be further divided into three subsets (Figure 6A, lower panel), the MHCII^{low or null}/Gr-1^low subset (gate R1), the MHCII^low/Gr-1^high subset (gate R2), and the MHCII^high/Gr-1^low subset (gate R3). In WT mice, the majority (>80%) of the cells from CD11b⁺/CD11c^low population were MHCII^{low or null}/Gr-1^low. Cells in this subset had a high side scatter, the morphology of eosinophils (Figure 6C, left panel and data not shown), and expressed two eosinophil markers, CCR3 and SIGLEC-F (Figure 6D), confirming that they were eosinophils. Cells in the MHCII^low/Gr-1^high subset, which also showed high side scatter (data not shown), were neutrophils (Figure 6C, middle panel). Cells of the MHCII^high/Gr-1^low subset had mononuclear morphology (Figure 6C, right panel) and were most likely to be macrophages. In V-M3 mice, the absolute number of eosinophils in LP of small intestine decreased more than 20 fold compared to WT mice (Figure 6A). A milder but still significant decrease (2 fold) of this population was also observed in the colon LP of V-M3 mice (Figure 6B). The absolute number of the macrophage subset was also reduced in LP of both small intestine (Figure 6A) and large intestine (Figure 6B), with a more striking reduction in the small intestine (10 fold) and a milder reduction in the large intestine (4 fold). In contrast, the absolute number of neutrophils was not altered by M3 expression in the LP of either small or large intestine (Figure 6A & 6B).

CD8 T cells, eosinophils and CD11b⁺/CD11c^low/MHCII^high/Gr-1^low myeloid cells express chemokine receptors for ligands blocked by M3

To understand how M3 blocks the homeostatic accumulation of CD8αβ⁺/TCRαβ⁺ and CD8αβ⁺/TCRαα⁺ T cells, eosinophils and the CD11b⁺/CD11c^low/MHCII^high/Gr-1^low macrophages, we sorted these populations and the CD4⁺/TCRαβ⁺ T cells from the intestine of WT mice and analyzed their chemokine receptor expression profile by Q-PCR. The CD8αβ⁺/TCRαβ⁺ subset expressed CCR9, CCR5, CXCR3, CXCR6, and low levels of CCR7 and CXCR4 (Figure 7A). The CD8αβ⁺/TCRαα⁺ subset expressed CCR9 and low levels of CCR5, CXCR3, and CXCR6 (Figure 7B). Intestinal eosinophils expressed only three chemokine receptors, CCR1, CCR3, and CXCR4 (Figure 7C), whereas the macrophages expressed high levels of mRNA for CCR1 and CXCR4 and expressed CCR2, CCR5, CCR7, CCR10, and CX3CR1 at lower levels (Figure 7D). CD4+/TCRαβ+ T cells expressed the receptors expressed by TCRαβ+/CD8αβ+ T cells, but also expressed CCR2, CCR4, CCR6 and low levels of CCR8. Furthermore, CD4+/TCRαβ+ T cells expressed much higher levels of CCR7 and CXCR4 than TCRαβ+/CD8αβ+ T cells (Figure 7A). These results demonstrate selective expression of chemokine receptors by leukocyte subsets in the intestine.

Figure 7. Chemokine receptor profiles of CD8⁺ IEL T cells and of CD11b⁺/CD11c^low LPL myeloid cells that were affected by M3 expression.

A & B). Q-PCR analysis of chemokine receptor profiles of the CD8αβ/TCRαβ⁺ subset, CD4⁺/TCRαβ⁺ subset (A) and the CD8αα/TCRαβ⁺ subset (B) of IELs in WT mice. (C & D) Chemokine receptor profiles of the eosinophils (C) and CD11b⁺/CD11c^low/MHC class II⁺/Gr-1^low macrophages from the LP of WT mice (D).

M3 expression attenuates the DSS-induced accumulation of myeloid cells in the colon

Expression of chemokines is markedly changed in inflamed tissues. To evaluate whether chemokine expression would be altered in the intestine during inflammation we induced colitis using DSS and compared the chemokine expression profiles between treated and non-treated mice. DSS treatment promoted marked changes in the expression of chemokines in the colon of control mice. The expression of several CC and CXC chemokines was elevated compared to that of non-treated mice. These changes were more dramatic for the CXC chemokines, some of which (CXCL1, CXCL2, CXCL9 and CXCL10) were upregulated 10–100 fold (Figure 8A & 8B). To test whether expression of M3 in this setting would alter development of colitis we treated control and V-M3 mice with 3% DSS. Both WT and transgenic mice lost weight after DSS treatment, but the transgenic mice lost less weight than the control mice (Figure 8C). To determine whether the expression of M3 altered the nature of the cellular infiltrates subsequent to DSS-induced colitis, we used flow cytometry. DSS treatment caused a marked accumulation of myeloid cells including dendritic cells (Figure 8D, both CD11b+ and CD11b− dendritic cells), neutrophils, macrophages and eosinophils in the colon of WT mice (Figure 8E). Expression of M3 in the gut markedly affected this response. The accumulation of dendritic cells, macrophages and eosinophils was greatly reduced by M3 expression (Figure 8D & 8E); no significant differences were observed regarding neutrophils (Figure 8E). Together these results indicate that expression of M3 affects leukocyte accumulation in the intestine during inflammation.

Figure 8. Effects of M3 expression on DSS-induced myeloid cell accumulation in the colon.

A & B). DSS treatment of C57BL/6 WT mice altered expression of many CC (A), CXC and CX3C (B) chemokines in the colon. Chemokines shown here were those significantly altered by DSS treatment (p<0.05, n=5 for WT and n=3 for WT+DSS). C). 6V-M3 transgenic mice (dot line) lost less body weight than WT littermates (solid line) in response to DSS treatment (* p<0.05). D & E). M3 expression reduced DSS-induced dentritic cell (D), eosinophil and macrophage (E) accumulation in the colon of 6V-M3 mice (n=4 per group, * p<0.05, ** p<0.01, *** p<0.001).

Discussion

In this study, we examined the chemokine expression profile in six different segments of the mouse intestine, and assessed the role of the chemokine network in homeostasis and inflammation. We found three major patterns of chemokine expression in the intestine. CCL6, CCL28, CXCL16, and CX3CL1 were expressed in both small and large intestine. CCL25 and CCL5 were expressed mainly in the small intestine, and CCL19, CCL21, and CXCL5 were expressed mainly in the ileum and large intestine. The different patterns of expression of these chemokines suggest that they have different functions.

To probe if these different patterns contribute to the overall distribution of cells within the intestine, we generated transgenic mice expressing the chemokine binding protein M3 in IEC and investigated whether blockade of chemokine function would alter the leukocyte populations in the intestine. V-M3 animals showed deficits in specific cell populations, indicating a clear role for chemokines in cell recruitment and/or accumulation in the intestine. Blockade of chemokines by M3 affected the accumulation of cells in both the lamina propria, intraepithelial compartments and in PP. Our previous studies have shown that M3 blocks chemokines in a dose dependent manner⁴⁸. In agreement with these observations, the changes were more evident in the small intestine and were likely due to higher expression of M3 in this part of the bowel.

Expression of M3 did not disrupt the organogenesis of PP, but significantly reduced the number of lymphocytes within these structures. The number of PP in V-M3 mice was normal, in contrast to CXCR5 or CXCL13 deficient mice, which have a reduced number of PP^{50, 51}. However, the sizes of the PP from V-M3 mice were greatly reduced with the majority containing only one dome, similar to CCR6 deficient mice^{52, 53}. As the ligand for CCR6, CCL20, is avidly bound by M3, it is possible that these changes reflect inactivation of CCL20 function.

CD8⁺ T cells are the most abundant lymphocytes present within the intestinal epithelium, whereas equal numbers of CD4⁺ and CD8⁺ cells are present within the underlying lamina propria⁸. Several subsets of CD8⁺ T cells are present in the intestinal epithelium: CD8αβ⁺/TCRαβ⁺ T cells, CD8αα⁺/TCRαβ⁺, and CD8αα⁺/TCRγδ⁺ T cells ^{8, 54}. V-M3 mice had reduced numbers of all subsets, except CD8αα⁺/TCRγδ⁺. The most striking and unexpected finding was that V-M3 mice had a substantial decrease in the number of the CD8αα⁺/TCRαβ⁺ T cells. Although the CD8αα⁺ T cells of the IEL do not circulate, the generation of this subset requires the neonatal thymus^{55, 56}. The precursors migrate from thymus to the intestine early in life and develop into mature CD8αα in either the cryptopatches or in the IEL compartment^{57, 58}. The observation that the CD8αα/TCRαβ⁺ subset is reduced in V-M3 mice strongly suggests that the recruitment of this population is dependent on the activity of particular chemokines. Chemokine profile analysis by Q-PCR demonstrated that the CD8αα^+/TCRαβ⁺ subset expressed CCR9 and low levels of CCR5, CXCR3, and CXCR6 (Suppl. Figure 3B). M3 may affect the accumulation of this subset by blocking ligands for these chemokine receptors. Other findings were consistent with previous reports in the literature. For instance, the reduction of CD8αβ⁺/TCRαβ⁺ T cells is comparable to that seen in CXCR3 deficient mice^27–29. As shown here, CXCR3 is expressed by CD8αβ⁺/TCRαβ⁺ T cells of IEL. Therefore, M3 may affect homeostasis of this cell population by blocking the function of CXCR3 ligands.

V-M3 mice did not show any change in the CD8αα⁺/TCRγδ⁺ IEL subset, which is found to be reduced in both CCR9 and CCL25 deficient mice^26–28. Since CCL25 is highly expressed in the small intestine, it is possible that the levels of M3 in the intestine were not high enough to efficiently block CCL25 function. Supporting this argument is the observation that other defects seen in the CCR9 mice such as reduction in number of IgA⁺ plasma cells, are not apparent in the V-M3 mice^{28, 59}.

Interestingly, we did not observe changes in the numbers of CD4⁺/TCRαβ⁺ T cells in the V-M3 mice. CD4⁺/TCRαβ⁺ T cells express many of the same receptors expressed by CD8αβ^+/TCRαβ⁺ cells, which are reduced in the V-M3 mice. However, CD4⁺/TCRαβ⁺ T cells express other receptors, including CCR2, CCR4, CCR6 and higher levels of CCR7 and CXCR4. This suggests that the recruitment/accumulation of CD4⁺/TCRαβ⁺ T cells in steady state may be dependent on many more chemokine ligands than that of CD8αβ^+/TCRαβ⁺ T cells and that M3 expression may not be sufficient to interfere with it.

Changes were also noted in the myeloid compartment. A cell population found to be decreased in the V-M3 mice was the CD11b⁺/CD11c^low/MHCII^high/Gr-1^low macrophages. These cells expressed high levels of mRNA for CCR1 and CXCR4 and lower levels of CCR2, CCR5, CCR7, CCR10, and CX3CR1 (Figure 6F). Of these receptors, only the ligands of CCR1/CCR5 (CCL5), CCR10 (CCL28), and CX3CR1 (CX3CL1) were expressed at high levels in the small intestine. M3 may reduce the accumulation of the macrophages by blocking CCL5 and/or CX3CL1, since it does not bind CCL28.

Changes were also seen in the number of eosinophils. Eosinophils, which comprise less than 1% of circulating leukocytes60, reside under normal conditions in both the thymus and intestine61, 62. Expression of M3 in the intestine reduced the accumulation of eosinophils in the intestine (Figure 6A & 6B), but not in the thymus (data not shown). Eosinophils sorted from control animals express three chemokine receptors: CCR1, CCR3, and CXCR4 (Figure 6E). Since M3 does not bind CXCL12 (the ligand for CXCR4) with high affinity, it may block the recruitment/accumulation of eosinophils by binding ligands for CCR1 and CCR3 (CCL11 and CCL24). These ligands are expressed in the intestine (Suppl. Figure 1) and bind M3 with high affinity (Suppl. Table 1). The number of eosinophils in the intestine and thymus of CCR3 and CCL11 deficient mice61, 63 are sharply reduced, suggesting that these molecules are critical for their recruitment into these tissues. However, in contrast to the global defect in eosinophil accumulation in CCL11 deficient mice61, the deficits observed in V-M3 mice were restricted to the intestine, suggesting that intestinal blockade of chemokines results in profound and localized deficits in eosinophils. Also arguing for local effects of M3 is the observation that M3 expression in the intestine drastically reduced the cellularity of B and T cells in the Peyer’s patches and of T cells in the IEL and LPL compartments, but did not alter the number of B and T cells in spleen, inguinal LN or MLN. Together the data support the concept that expression of M3 in the intestine results in a local inhibition of leukocyte accumulation.

The homeostatic numbers of neutrophils, distinguished from eosinophils and macrophages by higher expression of Gr-1 and lower levels of MHC class II, were not affected by M3 expression. We also did not detect any change in CD11c⁺ dendritic cell numbers in the LP of V-M3 mice. These findings suggest that either the recruitment of DC and neutrophils is unaffected due to inefficient blockade of particular chemokines, or sets of chemokines, or that they are not dependent upon the chemokine system for trafficking into the intestine during homeostasis. Several factors other than chemokines have been implicated in immune cell trafficking⁶⁴. However, we should note that the paucity of changes for some cell populations in the V-M3 mice does not necessarily reflect a lack of function for the chemokine system. As discussed previously, M3 does not bind all chemokines with the same affinity in vitro (Suppl. Table 1) and M3 blocking ability is dependent on the level of expression in vivo⁴⁶. Furthermore, M3 does not bind all known chemokines. In fact, two chemokines highly expressed within the intestine (CCL25 and CCL28) bind poorly to M3, which also limits the interpretation of the results. These limitations highlight a need for the generation of better reagents (in particular mutant lacking CCL28 and its receptor CCR10) to study leukocyte trafficking into the intestine. Notwithstanding these caveats, a clear role for chemokines emerges from the analysis of the results.

Another question that we asked in our study was whether chemokine blockade could affect the accumulation of leukocytes after an inflammatory challenge. DSS treatment dramatically altered the expression of chemokines in the colon. These changes were associated with increased influx of neutrophils, eosinophils, macrophages and dendritic cells into the colon. M3 expression dramatically attenuated infiltration by macrophages, eosinophils and dendritic cells, but did not significantly affect neutrophil numbers. We suggest that this may be due to inefficient blockade by M3 of the neutrophil chemoattractants chemokines CXCL1, CXCL2 and CXCL5, which were upregulated by the DSS treatment, and bind poorly to M3 (Supplemental Table 1). Other experiments involving M3 provide additional support to the concept that chemokines are important mediators of gut inflammation. Experiments described in Viejo-Borbolla et al (submitted) show that intestinal expression of M3 markedly changes the composition of the inflammatory infiltrates observed in the TNFDARE mice⁶⁵. In addition, M3 expression significantly reduces B cell recruitment to the small intestine of animals expressing a constitutively active form of TLR4 in intestinal epithelial cells ⁶⁶. Together these findings support the hypothesis that chemokines are critical for leukocyte accumulation in the gut during inflammation.

In conclusion, our study shows that chemokines are differentially expressed in the intestine during homeostasis and inflammation and that chemokine blockade affects the accumulation of leukocytes during these conditions. These results strongly implicate the chemokine system as an important determinant of leukocyte accumulation in the gut.

Nonstandard abbreviations

LP

Lamina propria

LPL

Lamina propria leukocytes

IEL

Intraepithelial lymphocytes

IEC

Intestinal epithelial cells

PP

Peyer’s patches