Chemotactic Responses of IL-4-, IL-10-, and IFN-γ-Producing CD4⁺ T Cells Depend on Tissue Origin and Microbial Stimulus¹

Differentiated Th cells can be divided into Th1 and Th2 subsets based on their cytokine production upon stimulation. IFN-γ-producing Th1 cells are crucial for defense against intracellular pathogens such as viruses. Th2 cells produce IL-4, IL-5 and IL-13 and are important for defense against extracellular parasites. On the other hand, polarized Th1 and Th2 responses can harm the organism by mediating autoimmunity and transplant rejection or the development of atopic diseases (1). These deleterious inflammatory responses are limited, in part, by regulatory CD4⁺ T cells, some of which produce IL-10 (2, 3, 4). IL-10 is also important in tolerance induction and inhibits proinflammatory cytokine synthesis as well as efficient Ag presentation to T cells by APCs (3).

Chemokines and their receptors guide T cells from the blood into tissues and subsequently control the microenvironmental localization of T cells within the tissue parenchyma (5, 6). It has been demonstrated that in vitro differentiated Th1 cells express CCR5 (7, 8, 9) and CXCR3 (7, 10), whereas Th2 cells are positive for CCR3 (11), CCR4 (7, 10), and CCR8 (12, 13). In vitro polarized Th1 and Th2 cells were also found to express chemokines upon stimulation that recruit T cells of the same subset (14). Additionally, cytokines expressed by Th1 and Th2 cells, namely IFN-γ and IL-4, can differentially induce the expression of Th1- and Th2-attracting chemokines in tissues, respectively (15). These findings led to the idea that IL-4- and IFN-γ-producing CD4⁺ T cell subsets are differentially attracted by chemokines, which could lead to Th1- or Th2-dominated tissue infiltrates. This hypothesis and the association of chemokines and their receptors with certain diseases make some chemokine receptors and their ligands attractive targets in the development of therapeutics for Th1- or Th2-mediated human diseases (16, 17). Importantly, mammalian and viral IL-10 down-regulate the expression of several chemokines and their receptors (3, 18) and induce the expression of decoy receptors for inflammatory chemokines (19). The specific augmentation or inhibition of recruitment of T cells producing IL-10 into tissues could also be a useful therapeutic approach. Despite their importance, however, chemokine receptors that guide IL-10-producing T cells remain poorly defined.

In vitro Th polarizing conditions may not accurately reflect the complex conditions encountered by T cells in vivo; thus, the chemokine receptor expression and chemotactic responses of in vitro derived Th cells could differ from their in vivo primed counterparts. Indeed, ex vivo analyses of human T cells have shown that the association of cytokine subset with chemokine receptor expression may not be as strictly dichotomous as for Th cells generated in vitro (20, 21, 22). Furthermore, the ex vivo studies that systematically analyzed the correlation between cytokine production and chemokine receptor expression were performed with human cells that relied on the analysis of receptor surface or mRNA expression (20, 21, 22). It has been demonstrated, however, that chemokine receptor surface and mRNA expression do not necessarily correlate functionally with chemotaxis (19, 23, 24). Moreover, instead of promoting chemotaxis, chemokine receptors can deliver proapoptotic signals to cells (25) or act as nonsignaling scavenging receptors (19). A systematic ex vivo analysis of the functional chemotactic responses by cytokine producing T cells is currently lacking.

The lung is an important organ in the first line of defense against airborne-pathogens and a target by environmental allergens leading to a variety of Th1- and Th2-dominated inflammatory lung diseases. Numerous lung pathogens are associated with strongly polarized T cell responses providing ideal models for the study of chemotaxis by polarized T cell subsets ex vivo. More specifically, infection with influenza A virus causes Th1-dominated inflammation that is usually restricted to the respiratory tract (26). In contrast, infection of mice with the nematode, Nippostrongylus brasiliensis, leads to a strong Th2 response, characterized by the formation of eosinophilic lung granulomas (27). In both infections, the Th1 or Th2 immune response is important for pathogen clearance and protection against reinfection (26, 27, 28).

In this study, we tested the hypothesis that CD4⁺ T cells isolated from spleens of uninfected, specific-pathogen free (SPF)⁴ mice exhibit restricted patterns of chemotactic responses that correlate with their production of IL-4, IL-10, or IFN-γ. Furthermore, to assess the stability of the detected migratory pattern, we determined the chemotactic profiles of T cells isolated from the lungs and spleens of mice infected with influenza A virus or N. brasiliensis. Our analysis demonstrates that in vivo differentiated cytokine-producing CD4⁺ T cells are characterized by chemotactic profiles that are highly dependent on tissue origin as well as the polarizing microbial stimulus. Although polarized T cells show clear preferences for specific inflammatory chemokines, the differences are not absolute. Thus, the coregulation of cytokine production and chemokine receptor function is complex in vivo, revealing the fine-tuning of tissue- and pathogen-specific T cell responses by the immune system.

Male and female BALB/c mice, 10–16 wk (for infections) or 6–9 mo old, were obtained from Bundesinstitut für Risikobewertung, Charles River Laboratories, or The Jackson Laboratory and were housed under SPF conditions. For the influenza A virus infection, mice were anesthetized before intranasal (i.n.) application of a sublethal dose (360 hemagglutinating units) of the influenza A virus strain HKx31 (x31; H3N2) in 30 μl of allantoic fluid. Virus, grown in the allantoic cavities of 11-day-old embryonated chicken eggs for 2 days at 37°C and titered by hemagglutination assays, was provided by T. Wolff (Robert Koch-Institute, Berlin, Germany). Virus stocks were stored at −80°C. For infection with the nematode N. brasiliensis, 750 infectious third stage larvae (L3) in 200 μl of PBS were injected s.c. at the animals’ neck. A mouse-adapted strain of this parasite was maintained and passed in Lewis rats or mice. All animal experiments were approved by local review boards and conducted according to institutional, federal, and state guidelines.

Before lung preparation, the lung microvasculature was cleared of blood by perfusion with 10 ml of PBS through the right ventricle of the heart. Extracted spleens and lungs were passed through a wire mesh into RPMI 1640 supplemented with 5% FCS to obtain single cell suspensions. To further purify mononuclear cells from the lung, the cells were resuspended in 40% isotonic Percoll (Biochrome), layered on 70% isotonic Percoll for high-density gradient centrifugation and subsequently harvested from the interface. Mononuclear cells from spleen cell suspensions were purified by high-density gradient centrifugation using Histopaque-1083 (Sigma-Aldrich). Subsequently, cell suspensions were washed twice with RPMI 1640 containing 5% FCS.

Mice were thoroughly perfused with PBS, and total RNA was isolated from homogenized whole lung or spleen tissue using Tri-reagent (Molecular Research Center). Total RNA was treated with RNase-free DNase I and further purified using the Absolute Nanoprep RNA kit (Stratagene). The RNA was reverse transcribed with the Taqman reverse transcription reagent kit (Applied Biosystems) using random hexamer primers. Gene expression was analyzed by real-time RT-PCR using SYBRgreen (Stratagene) on an ABI Prism 5700 instrument (Applied Biosystems) following the manufacturer’s instructions. Dissociation curves of all the gene products confirmed the specificity of the amplifications. The following forward and reverse primer pairs used for the amplifications were designed using Primer Express software (Applied Biosystems): 18S rRNA, 5′-AACGGCTACCACATCCAAGG-3′ and 5′-GGGAGTGGGTAATTTGCGC-3′; CXCL12, 5′-GATTGTTGCACGGCTGAAGA-3′ and 5′-TTCGGGTCAATGCACACTTG-3′; CCL17, 5′-ATGTAGGCCGAGAGTGCTGC-3′ and 5′-TGATAGGAATGGCCCCTTTG-3′; CCL22, 5′-ATGTTTTTCCTGCTGCAGGC-3′ and 5′-AAACTTGGGAGGTGTGTGGC-3′; CXCL9, 5′-CAGTGTGGAGTTCGAGGAACC-3′ and 5′-GCAGGAGCATCGTGCATTC-3′; CXCL10, 5′- AGCTTGAAATCATCCCTGCG-3′ and 5′-CAATGATCTCAACACGTGGGC-3′; CXCL11, 5′-GACAAAGGTGCCTGGACCC-3′ and 5′-CCTGCATTATGAGGCGAGCT-3′. Sample normalizations were performed using 18S rRNA measurements, and the relative gene expressions were calculated using the formula 2^−ΔΔCt, where Ct, the cycle threshold, is the number of PCR cycles required for the first detection of the PCR product. Ct values were calculated using Applied Biosystems software. Arbitrary units were assigned so that Ct 40 = 1 arbitrary unit.

All recombinant murine chemokines were purchased from R&D Systems, except CCL17, which was purified from the supernatant of a stable transfected insect cell line (provided by I. Förster, Munich, Germany), as described (29). All chemokines were titrated to identify optimal concentrations (data not shown), and the following concentrations were used: 100 nM CCL17, 100 nM CXCL9, 10 nM CXCL12. For the experiments, mononuclear cells obtained from the spleens or lungs of 20–30 animals were pooled. Mononuclear cells from the lung were used directly after preparation (see above) without further purification. To reduce the number of required wells and to obtain statistically reliable percentages of cytokine-producing T cells, spleen cells were enriched for cytokine⁺CD4⁺ T cells by depletion of B cells, macrophages, CD8⁺ and CD62L^high cells using panning as described (30). The remaining CD4⁺ cell population was 50% CD62L⁺ (intermediate to low expression). The assay was performed as previously described (30) and all quantifications and analyses were performed with gated CD4⁺ T lymphocytes. Briefly, 5 × 10⁵ cells, suspended in 100 μl of assay medium (RPMI 1640 plus 0.5% BSA), were added per upper well of fibronectin (Invitrogen Life Technologies) coated 5-μm pore size , 24-well tissue culture inserts (Costar). Chemokines were diluted in assay medium and added to the bottom well. Assay medium alone was used as a control. Migrated cells were collected after a 90-min incubation at 37°C, and the rate of migration was determined by the combined analysis of CD4⁺ T cell number and subset frequency in the input and migrated population. Triplicates of input and migrated cells were quantified by flow cytometry using fluorescent beads (Fluoresbrite Yellow Green Microspheres (Polysciences) or TruCount (BD Biosciences)) as an internal standard and Abs (anti-CD4-CyChrome, RM4-5; BD Biosciences) without washing to set appropriate count gates for CD4⁺ lymphocytes.

The migration rates were determined as described (55) using the following equations: 1) migration rate (mr; percent migrated) = N_cm N_ci * 100 percent 2) total number of cells (N_c = (n_c/n_b) * N_b with n_c, FACS-counted number of cells of respective subset; n_b, FACS-counted number of beads, N_b, total number of beads⁵; _i, input; _m, migrated (lower well). Cytokine-producing subsets among total CD4⁺ cells of the input and migrated populations (pools of 10-36 wells, depending on efficacy of the respective chemokine) were analyzed as described below. The mr (percent cells migrated toward chemokine) for the given subset _x is calculated according to the equation: mr_x= mr_CD4 * (frequency of x in migrated CD4 cells/frequency of x in input CD4 cells).

To detect intracellular cytokines, cells were polyclonally stimulated with 10 ng/ml PMA and 500 ng/ml ionomycin (Sigma-Aldrich) for 4 h with the addition of 10 μg/ml brefeldin A (Sigma-Aldrich) during the last 2 h. To prevent nonspecific staining, all cells were preincubated with blocking anti-FcγRIII/II Ab 2.4G2/75 and purified rat IgG (manufactured by Jackson ImmunoResearch Laboratories and purchased from Dianova), before staining with anti-CD4 (GK1.5) and peridinin chlorophyll-protein-conjugated streptavidin (BD Biosciences) as a second step reagent. After fixation in 2% paraformaldehyde and permeabilization with saponin (Sigma-Aldrich), the cell samples were preincubated with purified rat IgG (Jackson ImmunoResearch Laboratories), before intracellular cytokine staining. The following Abs conjugated to FITC, PE, or allophycocyanin were used: anti-IFN-γ (AN18.17.24), anti-IL-10 (JES5-16E3), anti-IL-4 (11B11), or appropriate isotype controls.

For cell surface staining, the following Abs were used: anti-CD4 FITC (RM4-5), anti-CD62L allophycocyanin (MEL-14), anti-CXCR3 PE (220803; R&D Systems), and anti-CXCR4 biotin (2B11). Surface CCR4 expression was visualized using a CCL22-human IgG3 Fc fusion protein (produced in COS 7 cells as described (31)) (CCL22-Ig), followed by biotinylated multispecies adsorbed goat anti-human IgG (Jackson ImmunoResearch Laboratories). Specificity of CCL22-Ig was shown by staining of thymocytes from CCR4-deficient mice, blocking with anti-CCL22 Ab, or by incubation with purified whole human IgG (data not shown). Streptavidin-PE was used as a second (or third) step reagent in chemokine receptor staining procedures. Dead cells were excluded from cell surface flow cytometric analysis by propidium iodide staining.

If not otherwise indicated, all staining Abs were obtained from BD Biosciences, except for biotinylated anti-CD4 and FITC-conjugated anti-IFN-γ, which were kindly provided by H. Hecker and H. Schliemann (Deutsches Rheumaforschungszentrum, Berlin, Germany). Quadrants were set according to the isotype controls. Flow cytometric analyses were performed on a FACSCalibur using CellQuest software (BD Biosciences), and ≥30,000 CD4⁺ T cells were acquired for each sample.

If not otherwise stated, data represent the mean (±SEM) of all experiments performed; dot plots show one representative experiment. Data were considered statistically significant when p < 0.05 as determined by Student’s t test (for comparisons of 2 groups) or repeated measures one-way ANOVA combined with Bonferroni’s multiple comparisons test (for comparisons of more than or equal to three groups) using GraphPad Prism software (GraphPad).

In vitro polarized Th1 and Th2 cells differentially express chemokine receptors and respond to distinct panels of chemokines (7, 8, 9, 10, 11, 12, 13). To test whether this finding applied to murine T cells differentiated in vivo, we assessed the migratory potential of CD4⁺ IL-4-, IL-10-, and IFN-γ-producing T cells isolated from the spleens of SPF BALB/c mice in a Transwell chemotaxis assay. We tested optimal concentrations of the following chemokines that are up-regulated during inflammation: CXCL12, CCL17, and CXCL9, which are ligands for CXCR4, CCR4, and CXCR3, respectively.

When the percentages of cytokine producers in input and migrated populations were compared, IFN-γ⁺ and IL-4⁺ CD4⁺ T cells were enriched in the CXCL9- and CCL17-responsive fractions, respectively (Fig. 1,A). Quantification of chemotaxis of the distinct cytokine-producing CD4⁺ T cell subsets revealed that all of the analyzed subsets migrated strongly toward CXCL12 (means of 39–69% of input migrated; Fig. 1,B). In contrast, the Th2 chemokine, CCL17, preferentially attracted IL-4-producing T cells: 42% of the input IL-4 single-positive (IL-4⁺ and negative for IL-10 and IFN-γ) and 24% of the IFN-γ/IL-4 double-positive T cells migrated, whereas other cytokine-producing T cells migrated at significantly lower levels (p < 0.001 and p < 0.01 for IL-4 single-positive and IL-4/IFN-γ double-positive CD4⁺ T cells, respectively, when compared with all other depicted CD4⁺ T cell subsets; Fig. 1,B). The Th1 chemokine, CXCL9, recruited a high percentage of IFN-γ-producing T cells independent of IL-4 or IL-10 coexpression: >67% of the input IFN-γ single, IFN-γ/IL-4 and IFN-γ/IL-10 double-positive CD4⁺ T cells migrated to the CXCR3 ligand (Fig. 1,B). Although their migration rate was significantly lower than that of IFN-γ single-positive CD4⁺ T cells (75% migration), a relatively high frequency of the IL-4 (41%, p < 0.001) and IL-10 (35%, p < 0.001) single-positive T cells migrated in response to the CXCR3 ligand (Fig. 1 B), demonstrating that CXCL9 preferentially, but not exclusively, recruits IFN-γ-producing T cells. Thus, under homeostatic conditions, the chemotactic profiles of splenic effector/memory CD4⁺ T cells are related to, but not as strictly dichotomous as, the response profiles of in vitro derived effector T cell subsets.

We next asked whether the migration pattern observed for IL-4, IL-10-, or IFN-γ-producing T cells isolated from the spleens of uninfected SPF mice also applied to T cells isolated from animals infected with strong Th1- or Th2-polarizing pathogens. To minimize differences in tissue-specific homing patterns, we sought infection models in which robust and typical Th1 and Th2 responses are induced in the same peripheral organ. Therefore, we analyzed cytokine production of CD4⁺ T cells in two well-characterized models of Th1- and Th2-polarized lung inflammation: 1) influenza A virus infection, induced by i.n. application of a sublethal dose of the virus strain HKx31 (H3N2), and 2) infection with a mouse-adapted strain of the nematode, N. brasiliensis, by s.c. inoculation of infectious stage 3 larvae.

In the absence of infection, the lungs of SPF status mice contain few to undetectable numbers of cytokine expressing CD4⁺ T cells (32, 33). The typical cytokine responses of CD4⁺ T cells isolated from the lungs of influenza A virus- or N. brasiliensis-infected mice on day 8 or day 13 postinfection, respectively, are depicted in Fig. 2,A. The time points reflect the peak number of lung-derived cytokine-producing CD4⁺ T cells during the respective infections (data not shown); although knowledge about T cell properties at early or late phases of infection might be of additional interest, the low number of effector cells recoverable at those time points precludes a reliable analysis of their chemotactic responses. CD4⁺ T cells isolated from influenza virus-infected lungs were predominantly IFN-γ⁺ (Fig. 2,A, top row). Distinct populations of IFN-γ/IL-10 double-positive and IL-10 single-positive CD4⁺ T cells were also detectable, whereas very few IL-4-producing CD4⁺ T cells were found (Fig. 2,A, bottom row). In contrast, CD4⁺ T cells isolated from N. brasiliensis-infected mice produced little IFN-γ or IL-10, and instead, were predominantly IL-4⁺ (Fig. 2 A). In both infection models, the percentages of cytokine-producing CD4⁺ T cells recovered from the spleen displayed only marginal changes relative to uninfected mice (Ref.32 and data not shown).

Next, we determined the mRNA expression patterns of ligands for CXCR3, CXCR4, and CCR4 in the lungs and the spleens of uninfected mice, and influenza A virus-, or N. brasiliensis infected mice on day 8 and day 13, respectively. In the absence of infection, notable mRNA expression of CXCL12 was detected in the lungs and spleens of uninfected mice, whereas all other analyzed chemokine mRNA was expressed at significantly lower levels (lung, p < 0.001; spleen, p < 0.05; Fig. 2,B, top row). Upon infection with N. brasiliensis, at the analyzed time point, although still the dominant chemokine mRNA expressed in the lungs or spleens of infected animals, expression of CXCL12 mRNA was decreased (lung, p < 0.05; spleen, not significant; Fig. 2,B). Additionally, infection with N. brasiliensis led to a 3-fold induction of mRNA encoding the CXCR3 ligands CXCL9 (p < 0.05), and CXCL10 (p < 0.01) in the lung, whereas the expression of CCR4 ligands tended to be marginally elevated, although the difference did not attain statistical significance when analyzed over multiple animals (Fig. 2,B, bottom row). In contrast, infection with influenza A virus induced dramatic mRNA expression of all known CXCR3 ligands in the infected lungs. Specifically, CXCL9 mRNA was induced in virus-infected lungs by 455 ± 154, CXCL10 × 114 ± 33 and CXCL11 × 104 ± 15-fold (Fig. 2,B, bottom row). In the spleens of influenza virus-infected mice, no statistically significant increases in chemokine mRNAs were detectable (Fig. 2 B, bottom row).

To assess changes in the surface expression of CXCR4, CCR4, or CXCR3 upon infection with influenza A virus or N. brasiliensis, we stained cells isolated from spleens and lungs with mAbs against CXCR3 and CXCR4, and to detect CCR4, with a CCL22 fusion protein (CCL22-Ig). Under all conditions tested, only few splenic CD4⁺ T cells expressed detectable CXCR4 (Fig. 3,A). Interestingly, upon infection with influenza virus and more pronounced with N. brasiliensis infection an increase in CXCR4⁺ cells among CD4⁺ T cells was detectable relative to cells from the uninfected lung (Fig. 3,A). With the exception of the N. brasiliensis-infected spleen, under all conditions tested, the percentage of CCR4 expressing cells among CD4⁺ T cells was low (Fig. 3,A). In contrast, infection with N. brasiliensis, and more drastically with influenza virus, boosted the percentages of CXCR3⁺ CD4⁺ T cells in the inflamed lungs (Fig. 3 A).

To correlate the chemokine receptor surface expression profile with functional properties, we subjected lymphocytes from the spleens and lungs of uninfected, influenza A virus or N. brasiliensis -infected mice to an ex vivo chemotaxis assay. Despite the low percentage of CXCR4⁺ CD4⁺ T cells from the spleens, splenic CD4⁺ T cells showed robust migration toward CXCL12 (Fig. 3,B). CD4⁺ T cells from spleens of uninfected mice migrated at a higher rate toward CXCL12 than CD4⁺ T cells from the lungs of the same animals (p < 0.001), and titration of the chemokine between 1 and 300 nM did not alter this result (Fig. 3,B and data not shown). Migration of lung CD4⁺ T cells toward CXCL12 was dissociated from their CXCR4 surface expression: cells isolated from both the Th1 and Th2 infection model migrated at a high frequency toward CXCL12, even though only T cells from N. brasiliensis infection model contained a major population of surface CXCR4⁺ CD4⁺ T cells (Fig. 3).

Paralleling the induction of mRNA for CXCR3 ligands (Fig. 2,B) and the cell surface expression of CXCR3 by T cells upon infection with influenza virus (Fig. 3,A), CD4⁺ T cells from the spleens and lungs of influenza virus-infected animals showed a higher migration to CXCR3 ligands when compared with CD4⁺ T cells from organs of uninfected mice or N. brasiliensis-infected mice (spleen, p < 0.05; lung, p < 0.001) (Fig. 3,B). Despite the increase in CXCR3⁺ CD4⁺ T cells in N. brasiliensis-infected lungs, only marginal migration toward CXCL9 was detectable in a range of 1 to 300 nM (Fig. 3 and data not shown). These results indicate that the chemotactic properties of CD4⁺ T cells are a function both of the presence and nature of infection, and the tissue origin, and that the sole analysis of chemokine receptor surface expression is an insufficient indicator of migratory properties of CD4⁺ T cells.

Because of the dissociation of chemokine receptor expression from chemotactic properties toward the corresponding ligands, we next focused on the chemotaxis of effector T cells ex vivo. We asked whether the migration pattern observed for splenic IFN-γ-producing CD4⁺ T cells from uninfected mice (Fig. 1) also applied to T cells recovered from lymphoid and peripheral tissues during infection. To this end, we isolated T cells from the spleens and lungs of influenza A virus- or N. brasiliensis-infected mice, and assessed their chemotaxis toward CXCL12, CCL17, and CXCL9. The migration toward CXCL12 of splenic IFN-γ single-positive CD4⁺ T cells was higher when the cells were collected from uninfected mice (69% migration) than from mice infected with either influenza virus (33%, p < 0.05) or N. brasiliensis (36%, p < 0.05). However, the migration toward CXCL12 of IFN-γ single-positive T cells isolated from the inflamed lungs was high (53–70%) in both infection models (Fig. 4). Under all conditions tested, IFN-γ single-positive T cells migrated only marginally above basal level toward CCL17, but exhibited robust responses (55–82% migration) to CXCL9, the only exception being a slight drop in migration to CXCL9 (44% migration, p < 0.05) by IFN-γ single-positive T cells recovered from the spleens of N. brasiliensis-infected mice (Fig. 4). Thus, the migration of murine IFN-γ single-positive (IL-4−) T cells to CCR4 ligands is rare, whereas chemotaxis to CXCR3 ligands appears to be characteristic of this T cell subset even in a strong Th2-polarizing infection.

Because chemokine receptors expressed by IL-4-producing T cells are potential targets in the treatment of asthma and atopic diseases (16), it was important to determine the consistency of the chemotactic response profile of T cells of this phenotype. We tested the migration pattern for IL-4-producing CD4⁺ T cells isolated from the spleens and lungs after infection with influenza virus or N. brasiliensis. Similar to the IFN-γ-producing CD4⁺ T cells during influenza, migration toward CXCL12 of IL-4 single-positive T cells from the spleens of N. brasiliensis-infected animals was reduced (p < 0.05) compared with IL-4⁺ CD4⁺ T cells from normal spleen (Fig. 5). The migration toward CXCL12 of IL-4⁺ single-positive CD4⁺ T cells from the lungs of N. brasiliensis-infected mice or the spleens of influenza virus-infected mice was not significantly different relative to IL-4 single-positive CD4⁺ T cells from normal spleen (Fig. 5). Unexpectedly, the migration rate toward the CCR4 ligand, CCL17, was reduced for IL-4 single-positive CD4⁺ T cells isolated from the spleens and lungs of N. brasiliensis-infected mice relative to IL-4 single-positive T cells from normal spleen (lung, p < 0.01; spleen p < 0.05) (Fig. 5). In contrast, the migration of IL-4 single-positive CD4⁺ T cells from the spleens of influenza A virus-infected mice toward CCL17 was not different relative to uninfected mice (Fig. 5). The data demonstrate that migration toward the CCR4 ligand, although regularly detected for in vitro polarized Th2 cells (7, 10) and splenic IL-4⁺ CD4⁺ T cells from uninfected mice (Fig. 1) is not a universal feature of murine in vivo differentiated IL-4-producing CD4⁺ T cells. After N. brasiliensis infection, no chemotaxis toward CXCL9 above basal level could be detected for IL-4 single-positive T cells from the spleen (p < 0.05), although ∼20% of the IL-4 single-positive T cells from the inflamed lungs of the same animals migrated to CXCL9 (Fig. 5). This finding suggests that the recruitment of Th2 cells by typical Th1-attracting CXCR3 ligands, as well as by Th2 chemokines, is actually a differentially regulated property of these cells in vivo.

T cells coexpressing IFN-γ and IL-4 can be considered as less differentiated, uncommitted Th1 or Th2 cells (Th0 cells) (34). Because T cells of this phenotype could support Th1 and Th2 responses, we assessed the chemotactic behavior of T cells coexpressing IL-4 and IFN-γ. Very few double-positive T cells could be recovered from the lungs of infected mice (Fig. 2 and data not shown), suggesting that differentiated Th cell subsets but not Th0 cells are preferentially recruited into (or induced from recruited cells in) inflamed tissues. Because of the low percentage of Th0 cells in the lungs, the migration of IFN-γ/IL-4 double-positive T cells could only be analyzed for cells isolated from the spleens. In contrast to IL-4-single-positive T cells, the chemotaxis of IFN-γ/IL-4 double-positive CD4⁺ T cells toward CCL17 and CXCL9 were similar from uninfected mice or mice infected with influenza virus or N. brasiliensis (Fig. 1 and data not shown). Thus, IL-4-producing CD4⁺ T cells that coexpress IFN-γ have migratory properties in common with both IL-4- and IFN-γ single-positive CD4⁺ T cells, but possess relatively consistent migration pattern during infection with influenza virus or N. brasiliensis.

IL-10-expressing T cells (T regulatory-1 cells) are important in the balance of immune responses as well as the prevention of destructive tissue pathology by limiting and terminating inflammatory immune responses (3, 4, 35). Because of their importance, we determined the chemotactic responses of CD4⁺ T cells expressing IL-10. The chemotaxis of IL-10 single-positive CD4⁺ T cells toward CXCL12 and CCL17 was not significantly influenced by tissue origin or infection: 27–42% migrated toward CXCL12 and only 9–17% toward CCL17 (Fig. 6). In contrast, chemotaxis toward the CXCR3 ligand, CXCL9, was strongly dependent on the analyzed organ and type of infection. Specifically, IL-10 single-positive CD4⁺ T cells from the spleens and the lungs of N. brasiliensis-infected mice showed reduced migration (only ∼11% migrated, p < 0.05) when compared with IL-10 single-positive CD4⁺ T cells from normal spleen (35% migration, Fig. 6). The chemotaxis of IL-10⁺ CD4⁺ T cells isolated from the spleens of influenza virus-infected animals was not significantly different (32% migrated) from splenic IL-10 single-positive T cells from uninfected mice (Fig. 6). In contrast to IL-10⁺ CD4⁺ T cells from N. brasiliensis-infected mice, IL-10-producing CD4⁺ T cells from the influenza virus-infected lung exhibited an enhanced migration toward CXCL9, (67% of input migrated, compared with spleen of uninfected mice, p < 0.01; compared with lung of N. brasiliensis-infected mice, p < 0.001) (Fig. 6), suggesting that under Th1-dominated inflammatory conditions, IL-10-producing CD4⁺ T cells are efficiently recruited to effector sites by Th1 cell-attracting chemokines, such as CXCL9, in a mechanism that may serve to limit destructive inflammatory processes.

CD4⁺ T cells coexpressing IFN-γ and IL-10 are also known to exert regulatory functions at effector sites during infection (35). Therefore, we assessed the migratory behavior of IFN-γ/IL-10 double-positive CD4⁺ T cells in the two infection models. Under all conditions tested, CD4⁺ T cells coexpressing IFN-γ⁺ and IL-10 showed a similar migration pattern as IFN-γ single-positive CD4⁺ T cells (data not shown and Fig. 4), indicating that putative regulatory IFN-γ/IL-10-producing CD4⁺ T cells may be recruited to the same inflammatory sites as IFN-γ single-positive T cells.

Immunity to pathogens, but also the maintenance of immunological tolerance with absence of overt autoimmunity or allergy, relies on a fine-tuned interplay between regulatory and effector Th cell subsets that produce distinct cytokines. Chemokine-mediated recruitment of Th cells into tissues is a major determinant affecting this balance. A number of studies have described differential chemokine receptor expression on in vitro polarized Th subsets; however, in vivo veritas, studies analyzing T cell subsets differentiated in vivo are rare and have been based on examination of receptor expression without demonstrating receptor functionality (20, 21, 22). Here, we provide the first systematic analysis of the chemotactic responses of CD4⁺ T cells that express IFN-γ, IL-4, and IL-10, isolated from inflamed and normal tissues. Our analysis allows the detection of chemotactic responses by T cells expressing any combination of these cytokines on a single cell level.

When we assessed the migration patterns of different cytokine-producing CD4⁺ T cell subsets from mice infected with either N. brasiliensis or influenza A virus, we found that IFN-γ⁺ T cells exhibited the most consistent chemotactic responses. Notably, migration to the CXCR3 ligand, CXCL9, was a constant property of IFN-γ-expressing CD4⁺ T cells independent of their IL-4 or IL-10 coexpression. However, IL-4- and IL-10-single expressing T cells also migrated toward the CXCR3 ligand (Fig. 1), demonstrating that CXCL9 is not specific for IFN-γ⁺ T cells in vivo. These data are in accordance with an analysis of chemokine receptor expression and cytokine phenotype in human T cells, which revealed that subpopulations of IL-4⁺ T cells express CXCR3 (22). Consistent with this, Medoff et al. (36) demonstrated a role for CXCR3 ligands in Th2-mediated pathology by showing that the CXCR3 ligand, CXCL10, is expressed in a murine model of allergic airway inflammation and contributes to the accumulation of IL-4⁺ CD4⁺ T cells in the inflamed lung and disease onset. Although we found induction of CXCR3 ligands in the lungs of the Th2 infection model (Fig. 2,B), IL-4 single-positive CD4⁺ T cells showed only poor migratory responses toward CXCL9 when isolated from this site (Fig. 5). Whether IFN-γ-inducible CXCR3 ligands expressed at Th1-dominated inflammatory sites also recruit IL-4⁺ T cells remains to be studied. As expected, we found a dramatic induction in the mRNA expression of the IFN-γ-inducible CXCR3 ligands, CXCL9, CXCL10, and CXCL11 in the lungs of influenza virus-infected mice (Fig. 2,B). However, we did not detect significant numbers of IL-4-producing T cells in the lungs of influenza virus-infected animals (Fig. 2), consistent with a study by Wohlleben et al. showing that during influenza A virus infection, recruitment of Th2 cells to the lung is impaired (37). In fact, Th2-dominated immune responses to influenza virus exacerbate lung pathology and delay viral clearance (38). Thus, an infection that requires a Th1 response for defense, such as influenza virus infection, might decrease the responsiveness of Th2 cells to ligands that recruit IFN-γ⁺ T cells into inflamed sites (37).

The highest degree of selectivity in the recruitment of a specific cytokine subset was detected for the CCR4 ligand, CCL17, which, under homeostatic conditions, attracted ∼40% of IL-4 single-positive CD4⁺ T cells, a smaller population of IFN-γ/IL-4 double-positive CD4⁺ T cells and only few IFN-γ single-positive or IL-10⁺ T cells (Fig. 1). This is in accordance with human studies that showed that CCR4 is preferentially expressed on IL-4⁺ CD4⁺ T cells isolated from peripheral blood (20, 22). However, in our experiments, the migratory responses of T cells capable of producing only IL-4 were strongly influenced by the type of infection and tissue origin. After N. brasiliensis infection, the migration of IL-4 single-positive T cells from the spleens and lungs toward CXCL12 and CCL17 was reduced and completely abrogated toward CXCL9 in splenic IL-4⁺ T cells relative to IL-4⁺ T cells from the spleens of uninfected mice (Fig. 5). Thus, the Th2-dominated inflammation of N. brasiliensis infection, was characterized by splenic IL-4-producing T cells that were unresponsive to a CXCR3 ligand, similar to in vitro polarized Th2 cells (7, 10). It is notable, however, that IL-4⁺ T cells, isolated from the spleens or lungs of N. brasiliensis-infected mice, dissimilar to in vitro polarized Th2 cells, exhibited reduced migration toward the CCR4 ligand relative to IL-4⁺ T cells from normal spleen. This was not likely due to ligand-induced receptor desensitization, because an up-regulation of RNA for CCR4 ligands was not detected upon infection with N. brasiliensis (Fig. 2 B), and the migration rate could not be increased by incubation of the cells for 2.5h at 37°C before the chemotaxis assay (data not shown). It is possible that IL-4⁺ T cells were attracted at an earlier time point of infection to the lung by CCR4 ligands such as CCL17, but, upon arrival, they down-regulated their responsiveness to the CCR4-ligand upon further differentiation. Whether the recruitment of Th2 cells into the inflamed lung by the CCL17-CCR4 axis is as important as in T cell homing to the skin (39, 40) is an open question; Chvatchko et al. (41) demonstrated that genetic deficiency of CCR4 does not influence the outcome of disease in a mouse model of asthma. Although speculative, it is also possible that N. brasiliensis could secrete factors that interfere with chemotaxis in a similar fashion as some viruses induce the expression of chemokine receptor antagonists and decoy receptors (42). Alternatively, accumulation of Ag-specific T cells at the effector site could account for a local enrichment of cytokine-producing T cells with a certain chemotactic profile. It has been shown that T cell recruitment is independent of Ag-specificity (43), whereas predominantly Ag-specific T cells are retained at the effector site (44). Therefore, it is possible that during an immune response Ag-specific T cells enrich over time at the effector site, and due to further differentiation, change their initial chemokine-receptor profile. In support of this, we previously demonstrated that CCR7 expression is different between total IFN-γ⁺ and Ag-specific IFN-γ⁺ CD4⁺ T cells isolated from the inflamed lung during influenza A virus infection (32). Presumably, also in N. brasiliensis infection T cells at the effector site are enriched in Ag-specific T cells that might have changed their chemokine receptor profile.

It is possible that in vivo-differentiated cytokine-producing T cell subsets additionally exhibit differences in their sensitivity toward chemokines and respond to different chemokine concentrations. However, in uninfected mice as well as in influenza virus- or N. brasiliensis-infected mice, similar chemokine concentrations were optimal at attracting CD4⁺ T cells (data not shown). This suggests a stable sensitivity to the tested chemokines for CD4⁺ T cells and argues against the possibility that CD4⁺ subpopulations display major differences in responsiveness toward different chemokine concentrations.

This study provides the first systematic analysis of the chemotactic response pattern of IL-10-producing T cells ex vivo. T cells of this phenotype are important in the balance of both Th1 and Th2 immune responses (45). Under homeostatic conditions, IL-10⁺ CD4⁺ T cells migrated to ligands of CXCR3 and CXCR4 and showed negligible responses to the CCR4 ligand, CCL17. In addition, their chemotaxis toward the CXCR3 ligand, CXCL9, was highly modulated depending on tissue origin and type of inflammation. CD4⁺ IL-10 producers displayed an enhanced responsiveness to CXCL9 when isolated from the effector site during an active Th1 response to influenza virus, but a reduced migration toward CXCL9 when isolated from the lungs or spleens after N. brasiliensis infection. In several infection models, IL-10 hampers pathogen clearance, but is crucial in the prevention of pathology caused by destructive inflammatory responses against the pathogen (3, 45, 46, 47). Therefore, attracting both IL-10- and IFN-γ-producing CD4⁺ T cells by CXCR3 ligands into an acutely inflamed site could be an important regulatory mechanism. This might also explain why we detected more IL-10-producing T cells in influenza virus-infected lungs than IL-10⁺ T cells in the lungs of N. brasiliensis-infected mice (Fig. 2 A).

CXCR4 and its ligand, CXCL12, are widely expressed under homeostatic conditions (48), but have also been suggested to recruit T cells into inflamed joints of patients with rheumatoid arthritis (49, 50) as well as to contribute to lung inflammation in an allergic airway disease model (51). In addition to their role in chemotaxis, CXCR4 and CXCL12 have been shown to mediate T cell costimulation (52) and survival (53). Whether distinct cytokine-producing T cell subsets migrate preferentially in response to CXCL12 has not been thoroughly tested so far. Throughout our analysis, the CXCR4 ligand lacked the ability to specifically recruit a distinct subset of cytokine-producing T cells ex vivo; instead, all analyzed T cell subsets efficiently migrated toward CXCL12. Accordingly, CXCL12 may contribute to the development of lymphocytic infiltrates by its strong propensity to attract all types of effector/memory T cells and/or provide survival signals at the inflammatory site.

In addition to the chemotaxis data presented in this study, we also tested chemotaxis toward other chemokines that have been found to differentially attract in vitro polarized Th1 and Th2 cells. In our hands, the CCR5 ligands, CCL3 and CCL4, did not induce significant migration of CD4⁺ T cells isolated from both experimental models, despite our observation that in vitro differentiated Th1 cells migrated toward those chemokines (data not shown). Additionally, in accordance with past studies that did not identify significant CCR3 expression by human T cells (20, 54), we were unable to detect chemotaxis by T cells toward the CCR3 ligand, CCL11, although eosinophils collected from the lungs of N. brasiliensis-infected mice migrated well (data not shown). From these data, we conclude that the majority of murine in vivo polarized cytokine producing CD4⁺ T cells do not express functional receptors for CCL3/CCL4 or CCL11. Finally, the CCR8 ligand, CCL1, from three different commercial suppliers did not induce chemotaxis of CD4⁺ T cells ex vivo (data not shown).

In conclusion, our study provides the first functional analysis of the chemotactic behavior of important subsets of cytokine-producing CD4⁺ T cells differentiated in vivo. We show that the migratory responses of T cells that produce the prototypical Th1 or Th2 cytokines, namely IFN-γ and IL-4, respectively, and T cells that express the immunosuppressive cytokine IL-10 are strongly influenced by tissue origin and infection. The data of this study suggest that differential chemokine responses are not simply part of a coordinated Th1 or Th2 expression program, but instead inflammatory conditions, organ-specific environment, and microbial stimuli flexibly modulate the migratory pattern of polarized CD4⁺ T cells in vivo. These results obtained using strongly polarized immune responses, which are crucial in pathogen clearance, also raise implications for diseases mediated by pathologic Th1 and Th2 responses. Our study demonstrates that the chemotactic behavior of effector/memory T cells is complex, and should encourage further investigation to determine the factors that control this complexity in vivo.

We are indebted to Eugene Butcher, Carrie N. Arnold, Tohru Sato, and Everett Meyer for critical comments on the manuscript, and to Thorsten Wolff, Irmgard Förster, Heidi Hecker and Heidi Schliemann for providing reagents.

The authors have no financial conflict of interest.