CCR4 versus CCR10 in human cutaneous T_H lymphocyte trafficking

Classical studies of tissue-specific homing within the human circulating T_H (CD4⁺) compartment suggest that the memory (CD45RA⁻) population can be divided into tissue-specific subsets. Each subset traffics through and engages in immunosurveillance of distinct domains of nonlymphoid tissues. For example, memory lymphocytes expressing the α4β7-integrin-heterodimer (an adhesion molecule that binds the mucosal addressin MAdCAM-1) contain antigen specificities associated with the gut1,2 and tend to traffic through intestinal sites.3,4 T_H cells that express the cutaneous lymphocyte antigen (CLA) carbohydrate (an adhesion molecule that binds E-selectin) contain specificities for cutaneous antigens5,6 and tend to traffic through cutaneous sites.7,8 CLA-expressing cells are identified by mAb HECA-452 in humans7,8 or by flow cytometry with an E-selectin–immunoglobulin (Ig) chimera in either mice or humans.9 

Differential expression of chemokine receptors also correlates with tissue-specific homing. For example, expression of chemokine receptor 9 (CCR9) is associated with a small intestine (but not colon) homing subset of T_H cells.10,11 Moreover, CCR7 function is necessary for homing of T cells to secondary lymphoid organs through high endothelial venules (HEVs).12-14 

Another chemokine receptor, CCR4, is robustly expressed by most skin-homing (CLA⁺) T_H cells in the circulation, but rarely (and at lower levels on the few positives) by intestinal-homing T_H cells.15 CCR4 is also expressed at high levels by skin-infiltrating lymphocytes (isolated directly from skin), but not by gut-infiltrating lymphocytes.10 Additionally, the CCR4 ligand TARC/CCL17 is associated with cutaneous (but not intestinal) endothelial cells.15 

CCR4 expression can easily distinguish between cutaneous (CLA⁺) and intestinal (α4β7⁺) T_H memory cells, but is also expressed by some circulating memory T_H cells that lack both CLA and α4β7.15 Some of these CLA⁻/CCR4⁺ cells may (like CLA⁺cells) have cutaneous roles, as vascular cell adhesion molecule-1/integrin interactions appear to be more important than CLA/E-selectin interactions for deep-dermal homing.16However, there also appears to be an additional, less well-understood role for CCR4 in pulmonary leukocyte homing.17-19 

The expression of CCR4 by apparently noncutaneous lymphocytes has led to the proposal that CLA and CCR4 must work together to convey skin-specific homing.15 Alternatively, it has been proposed that there may exist another chemokine receptor, with better specificity for cutaneous T_H cells than CCR4.

In fact, after the associations among CCR4, TARC, and cutaneous T_H homing were discovered, another chemokine, cutaneous T-cell–attracting chemokine (CTACK)/CCL27 (originally cloned as ALP in mice20), was found to be expressed by cutaneous keratinocytes.21 CTACK is not a ligand for CCR4, but like TARC, CTACK preferentially attracts CLA⁺ T_Hcells from peripheral blood in vitro.21 The receptor for CTACK was identified as CCR10, which has another ligand (MEC, CCL28) expressed within the colon and within secretory tissues, including salivary and mammary glands.22 CCR10 is expressed by a subset of CLA⁺ T_H cells in peripheral blood.23 

Thus, CCR4 and CCR10 are both associated with conventionally defined skin homing T_H cells from peripheral blood, but the association is imperfect in both cases: CCR4 is present on essentially all skin-homing cells, but also present on other systemic cells. In contrast, CCR10 is absent from non–skin-homing T_Hcells,23 but only present on a subset of skin-homing T_H cells.

In addition to the tissue-specific divisions among T_Hsubsets, there exist other axes of subdivision. For example, a relatively small subset of T_H cells does not express the tumor necrosis factor–receptor family member CD2724-26 and/or the chemokine receptor CCR7.24,27 Such cells are found within the CLA⁺/α4β7⁻, CLA⁻/α4β7⁺, and CLA⁻/α4β7⁻ memory compartments, but not within the naive (CD45RA⁺) compartment (virtually all naive cells express both CD27 and CCR724). The T_Hphenotypes that lack these markers have been called “effector-memory,” because they are enriched in recently seen antigen specificities, and they respond more quickly to antigen than most other memory T_H cells.25-27 Absence of CCR7 would preclude entry of such cells into lymphoid organs via HEVs through the known mechanisms,13,14,27 suggesting that such cells might home through only those nonlymphoid tissues to which they have become dedicated. Indeed, effector-memory cells are found within uninflamed or inflamed nonlymphoid tissues, often side-by-side with cells that continue to express both CD27 and CCR7.24 The relationship between effector phenotype and trafficking phenotype has not yet been thoroughly assessed.

In order to understand the interesting differences between CCR10 and CCR4 expression among human T_H subpopulations, we have generated a monoclonal antibody that recognizes functional expression of CCR10 by human lymphocytes. We have performed extensive analyses to compare and contrast CCR10 versus CCR4 expression by T_Hsubsets associated with cutaneous homing. We have examined expression (and functionality) of these receptors by several distinct subsets of peripheral blood T_H memory cells, including those classified as “effector” T_H cells. Furthermore, we have compared expression of these receptors by T_H lymphocytes directly isolated from inflamed cutaneous sites.

The murine pre-B lymphoma cell line L1/2 was transfected with an expression vector (pcDNA3.1) containing the human CCR10 DNA sequence, as previously described.28 Transfected pools were chemotaxed to CTACK/CCL27 and/or to MEC/CCL28, and clones with the best chemotactic potential were selected (L1/2 cells expressing human CCRs 1 through 9 were created in the same manner). CCR10-specific monoclonal antibody 1B5 was generated by immunizing C57Bl/6 mice intraperitoneally with 10⁷ CCR10 L1/2 cell transfectants every 2 weeks for 5 to 6 times. The final immunization was performed intravenously 4 days prior to extraction of the spleen for fusion with the SP2/O cell line as described.24CCR10-specific antibody-secreting hybridoma cell clones were identified by performing immunofluorescence staining of wild-type (WT) and CCR10 L1/2 cells with cell supernatants followed by a fluorescently-labeled antimouse secondary antibody. Fluorescent cells were detected by flow cytometry. Antibody-secreting clones were isolated by multiple rounds of subcloning. For screening the CCR10 mAb against L1/2 cells transfected with other CCRs (Figure 1), each transfectant was stained with a specific mAb to its respective receptor, to ensure high expression levels of the transfected gene (not shown).

Buffy coats from anonymous healthy donors were obtained from the Children's Hospital (Boston, MA) blood bank. Mononuclear layers were prepared as previously described.12,15,17,24,29 30 

Delayed-type hypersensitivity (DTH) responses toCandida extract, and subsequent isolation of lymphocytes from vacuum-blister fluid were performed as described previously.10,24 Experimental human chanchroid was induced by inoculation of healthy, HIV-seronegative volunteers at 3 sites withHaemophilus ducreyi 35000HP (a human passaged isolate of 35000). Sites that evolved into pustules were biopsied 7 to 14 days after inoculation. In some cases, multiple sites from one subject were biopsied and then pooled. Isolation of lymphocytes from biopsied skin using EDTA (ethylenediaminetetraacetic acid) was performed as described previously.10,24 Informed consent was obtained from the subjects in accordance with the guidelines for human experimentation of the US Department of Health and Human Services and the institutional review board of Indiana University–Purdue University at Indianapolis under grants AI31494 and AI27863 to S.M.S. and M01RR00750 to the GCRC at IU. Enrollment procedures, exclusion criteria, preparation of the bacteria, inoculation procedures, and clinical outcomes are described in detail elsewhere.31-33 

Immunostaining was performed with a 5-step method. Cells were stained first with unconjugated mAb (or its isotype-matched control); followed by polyclonal second-stage Abs; followed by blocking with 3 mg/mL mouse IgG (technical grade; Sigma) and 0.3 mg/mL mouse IgM (technical grade; Sigma); followed by directly-conjugated mAbs; followed by streptavidin conjugate. (1) The first fluorescent color used (color 1) was: CLA–fluorescein isothiocyanate (FITC; HECA-452; BD/Pharmingen, San Diego, CA) or CD27-FITC (M-T271; BD/Pharmingen); (2) color 2: CD27-phycoerythrin (PE; M-T271, BD/Pharmingen) or β7-Integrin-PE (FIB-504; BD/Pharmingen); (3) color 3: CD45RA-PE-TR (2H4; Beckman-Coulter, Brea, CA); (4) color 4: IgG2a isotype control (UPC-10; Sigma), CCR4 (2B10;15 Millennium Pharmaceuticals), or CCR10 (1B4; Millennium Pharmaceuticals); followed by goat anti–mouse IgG2a(γa)-biotin (Southern Biotechnology, Birmingham, AL); followed by Streptavidin-PE-Cy7 (Cedarlane Labs, Ontario, CA); (5) color 5: CCR7 (3D9;24 Millennium Pharmaceuticals); followed by goat anti–mouse IgM(μ)-Cy5 (Jackson Immunoresearch, West Grove, PA); and (6) color 6: CD4-allophycocyanin (APC)–Cy7 (S3.5; Caltag, Burlingame, CA).

Data on stained cells were acquired on dual-laser MoFlo cytometer (Cytomation, Ft. Collins, CO) configured for 6 colors: (1) FITC, (2) PE, (3) PE-TR, (4) PE-Cy7, (5) Cy5, (6) APC-Cy7. Raw data were compensated and analyzed using Summit 3.0 software (Cytomation).

Chemotaxis was performed as described.12 15For each chemokine for each donor, 2 chemotaxis wells of the following concentrations were set up: 1000 nM, 300 nM, 100 nM, and 30 nM; data are shown in Figure 4 for the optimal concentrations. Recombinant stromal cell–derived factor-1α (SDF-1α; PeproTech), recombinant TARC (R&D Systems, Minneapolis, MN) and synthetic CTACK (Gryphon Sciences, South San Francisco, CA) were used for all experiments shown. Input and migrated cells were stained with a 5-color protocol: CLA-FITC, CD27-PE (M-T271, BD-Pharmingen), CD45RA-PETR, CD4-biotin (13B8.2, Beckman-Coulter) + streptavidin-PE-Cy7, and CCR7 (7H12, Millennium Pharmaceuticals) + goat anti–mouse IgG (H&L)-Cy5 (Jackson Immunoresearch).

A monoclonal antibody against human CCR10 was developed by immunizing mice with L1/2 cells (a murine pre-B lymphoma) transfected with human CCR10 mRNA in a mammalian expression vector.24 The resulting mAb, 1B5, recognized CCR10-transfected (but not WT) L1/2 cells (Figure1). 1B5 did not recognize L1/2 cells transfected with any of the other known CC chemokine receptors (CCR1 through 9 plus CCR11/GusB) (Figure 1).

We first examined the CCR10 expression of naive T_H and the 3 peripheral blood memory T_H subsets classically defined by the homing receptors CLA or α4β7-integrin34 35 (Figure2A). Among the memory T_H populations, CCR10 expression was restricted to the CLA⁺/α4β7⁻ “skin homing” subpopulation (Figure 2B). Interestingly, however, CCR10 was not expressed by approximately 70% of the CLA⁺T_H population. CCR10 was absent from naive T_Hcells (Figure 2B).

Two-color plots of CCR10 or CCR4 versus CLA are shown for the memory T_H-gated population in Figure 2C. CCR4 stains the vast majority of CLA⁺ T_H cells with equal intensity and also stains a subset of CLA⁻ T_H cells (Figure 2C, right panel). In contrast, CCR10 appears to preferentially stain only those T_H cells in the brightest one half to one third of the CLA⁺ distribution (Figure 2C, left panel).

Another previously identified axis of subdivision within the peripheral blood CLA⁺ T_H memory compartment (as well as the other T_H memory compartments) involves CD27 and CCR7 expression.24-27 We therefore asked whether the observed patterns of CCR10 expression correlate with CD27 and/or CCR7 expression by using 6-color flow cytometry.

The upper panel in Figure3A shows that peripheral blood CLA⁺ T_H cells are spread among all 4 quadrants of a CD27 versus CCR7 plot. However, plots of CLA versus CD27 or CCR7 distinguish the individual subsets more clearly (Figure 3A, lower plots). The latter type of plot was used to enumerate CLA⁺ subsets displaying the 4 permutations of CD27 and CCR7 expression for 10 healthy donors (Figure 3B). Interestingly, the CD27⁻ and CCR7⁻ subsets of the CLA⁺ T_H populations were enriched for the brightest one third to one half of the CLA expressers, with striking similarity to the CLA pattern observed for CCR10 expressers in Figure2C.

For simplicity, we have chosen to focus on the 2 most extreme phenotypes identified here for more detailed analysis: the CD27⁻/CCR7⁻ “double negatives” (which meet all immunophenotypic criteria for being “effector” T_H cells), and CD27⁺/CCR7⁺“double positives” (which meet none of the criteria for being “effector” T_H cells). Figure 3C-D compares the CCR4 and CCR10 expression patterns of CD27/CCR7–double positive and –double negative subsets of CLA⁺, CLA⁻ and α4β7⁺ memory T_H populations, as well as the naive population (note that the CLA⁻ memory population also contains the α4β7⁺ population as a subset). Figure 3C shows CCR10 staining (black lines, left column) or CCR4 staining (black lines, right column) overlaid on isotype-matched controls for the same gated populations from a representative donor (gray lines, both columns). Figure 3D shows the mean CCR4 and CCR10 expression (and SD) for the same populations by 10 healthy blood donors. Although CCR4 is expressed by the same proportion of cells and (at equal levels) by both double-positive and double-negative CLA⁺ populations, CCR10 is expressed by a much greater proportion (and at much greater levels) by double-negative cells. (Note that CCR10 and CCR4 expression is enumerated in Figure 3D by counting those cells brighter than the brightest 1% of the isotype control. This results in an underestimate of CCR4 expression by the CLA⁺ populations [and CCR10 by the “effector” cutaneous population], as it is quite clear that the entire stained population is shifted with respect to the control-stained population. Also, this method does not consider the mean expression of receptor by positive cells, as exemplified by the apparently modest difference in CCR4 expression between the CLA⁻ and α4β7⁺populations in Figure 3D, which in truth have a dramatically different mean expression as seen in Figure 3C.) In fact, CCR10 is expressed by virtually all double-negative CLA⁺T_H cells. The CD27⁺/CCR7⁻ and CD27⁻/CCR7⁺ populations (not shown) expressed intermediate CCR10 levels (D.S. and J.J.C., unpublished data, May 2002).

Whenever possible, it is important to confirm chemokine receptor flow cytometry results through an independent method. We therefore set out to determine whether CCR10 expression patterns (determined by 1B5) correlate with functional responses to CTACK/CCL27 (a CCR10 ligand) in standard chemotaxis assays. Each of the populations examined express very similar amounts of the SDF-1 receptor CXCR4 (J.J.C., unpublished data, November 2000). This fact allows normalization of their CTACK-mediated migration to that of SDF-1, to correct for potential intrinsic differences in migration rates among the various populations (Figure4, upper scale). Migration is also plotted (on the same graphs) as percent of input (Figure 4, lower scale) with comparable outcome. Figure 4clearly demonstrates that the double-negative CLA⁺T_H population responds significantly better to CTACK than any of the other populations tested (including the double-positive CLA⁺ T_H population, Figure 4, left panel). There was no significant difference in TARC responsiveness between double-positive and double-negative populations (Figure 4, right panel).

The data indicate that while CCR4 is expressed by most CLA⁺ T_H cells in blood, CCR10 is expressed only by a minority subset. We therefore asked whether the CCR10⁺subset might have an advantage over other CLA⁺/CCR4⁺ T_H cells in its ability to home to inflamed cutaneous sites.

We chose to analyze the homing phenotypes of T_H cells directly isolated from 2 different types of cutaneous lesions. The first was a widely studied DTH model, in which cutaneous inflammation is induced by intradermal injection of a sterile extract ofCandida albicans.10 The second was an experimental cutaneous bacterial infection of human volunteers withH ducreyi. H ducreyi causes chancroid, a sexually transmitted genital ulcer disease that facilitates efficiency of HIV transmission by disruption of epithelial barriers and infiltration of T_H (CD4⁺) cells into the lesions.36 

Essentially all of the T_H cells isolated from DTH skin were of the CD45RA^lo/neg memory phenotype (Figure5A). T_H cells from donor-matched peripheral blood contained 40% to 50% CD45RA^hi naive cells (not shown), suggesting that the skin-derived samples were not significantly contaminated with peripheral blood lymphocytes. The number of cells expressing CLA was greatly enriched with respect to peripheral blood memory T_Hcells, but the ratio of CD27/CCR7 double positives to double negatives did not significantly differ from that of peripheral blood memory CLA⁺ T_H cells (Figure 5B-C). CCR4 was expressed by skin-infiltrating T_H cells at greatly enriched levels, but, interestingly, CCR10 was not (Figure 5D-E).

The identical flow cytometry was performed on a total of 3 DTH and 3H ducreyi lesions, and the data presented as bar graphs (Figure6). For both types of lesion, the ratio of memory to naive cells was significantly different from that of the peripheral blood T_H population from matched donors (not shown). The proportion of CLA⁺ and CCR4⁺ T_Hcells was significantly increased with respect to the peripheral blood memory population. In contrast, the ratio of CD27/CCR7 double positives to double negatives was not significantly different from that of peripheral blood memory T_H cells. Most importantly for the present study, the proportion of CCR10⁺ cells was not significantly enriched with respect to peripheral blood CLA⁺ memory T_H cells.

We have performed a detailed analysis of functional CCR4 and CCR10 expression by peripheral blood– and skin-derived human T_Hcells. These studies are part of an ongoing effort to understand the role of chemokines and their receptors in cutaneous (and other tissue-specific) lymphocyte homing.

Peripheral blood contains a heterogeneous mixture of T_H(and other) lymphocytes, many with distinct trafficking patterns.34 35 Therefore, within a given tissue, the specific enrichment of cells bearing a particular homing molecule suggests the molecule's potential importance in homing to that tissue.

We confirm that lymphocytes expressing CLA and CCR4 are greatly enriched within the cutaneous lesions studied here, which is not the case for other tissues.10 15 CCR10 is slightly enriched in skin when compared with the total circulating memory T_Hpool. However, CCR10 is not enriched in cutaneous sites when compared with CLA⁺ memory T cells from peripheral blood. Thus CLA⁺/CCR4⁺ T_H cells that express CCR10 have no apparent advantage over their CCR10⁻counterparts in homing to inflamed cutaneous sites.

One should consider the caveat that cell-surface CCR10 might indeed be necessary for skin infiltration, but becomes down-regulated upon entry. This is clearly not the case for CCR4, which is expressed at comparable levels by both CLA⁺ blood T_H cells and skin-infiltrating T_H cells.10 Further, we demonstrated that CCR10 expression is greatly enriched on cells lacking CD27 and/or CCR7 in peripheral blood (Figure 3C-D), and that the CD27/CCR7 phenotype of cutaneous-infiltrating cells is not significantly different from that of peripheral blood CLA⁺memory T_H cells. Therefore, as the CD27/CCR7-negatives are not enriched in cutaneous sites, there is no reason to propose that CCR10 expression has been underrepresented by this assay.

Both CCR10 and CCR4 have qualities that suggest a role in T_H homing to cutaneous sites. However, our data suggest that, unlike CCR4, CCR10 is not a necessary component of cutaneous homing for most T_H cells. Within the skin-homing T_H population, CCR10 is associated with the “effector” phenotype. Virtually all those cells displaying the most extreme effector phenotype (ie, CD27/CCR7 double negatives) express CCR10.

In vitro studies make it clear that effector T_H cells are functionally distinct from conventional memory T_Hcells.25-27 However, it is not clear how such cells actually contribute to in vivo immune responses. It is possible that such effector T_H cells play a regulatory role in coordinating immune responses for the tissues through which they traffic. For example, CLA⁺ cutaneous “effector” T_H cells may coordinate the responses of conventional cutaneous CLA⁺ memory T_H cells, either positively or negatively. Part of this role may involve coordinating the traffic of conventional T_H cells through the skin, perhaps by regulation of TARC/CCR4 interactions. If this scenario is proved true, effector T_H cells would require a TARC/CCR4-independent mechanism for skin entry. CTACK/CCR10 interactions could provide such a mechanism.

This hypothesis would predict that cells entering the skin in the absence of TARC/CCR4 interactions (ie, in CCR4-deficient mice, as discussed in “Murine models”37) would be greatly enriched in effector T_H cells. The potential effect that blocking CTACK/CCR10 interactions would have on cutaneous inflammation is less straightforward to predict. If effector cells tend to positively regulate the intensity of skin inflammation, such blocking might be expected to dampen the response. If, in contrast, effector cells tend to negatively regulate inflammation, blocking of CCR10 might actually exacerbate skin inflammation.

Recent immunohistochemistry studies suggest that CCR10 is expressed by many cells within allergic dermatitis and psoriatic lesions.23 However, it is not clear (from this type of data) what proportion of T_H cells express CCR10 within these lesions. Therefore, it is not clear that these findings differ significantly from our own. However, if CCR10 is indeed expressed by many T_H cells in such lesions (as suggested by Homey et al23), then CCR10 may play a more pervasive role in homing to allergic dermatitis and psoriatic sites than played in the DTH and bacterial skin lesions studied here. In such a case, one would expect the allergic dermatitis and psoriatic sites to have more CD27/CCR7 double-negative cells than DTH and bacterial skin lesions.

Two recent murine studies of dinitrofluorobenzene (DNFB)–induced DTH responses have demonstrated an inhibitory effect of anti-CTACK polyclonal antibodies on cutaneous lymphocyte homing. In the first model (on the C57Bl/6 background37), treatment with anti-CTACK inhibited cutaneous lymphocyte homing in CCR4-deficient but not WT mice. This implies that signaling through both CCR4 and CCR10 must be blocked to influence cutaneous homing through the chemokine system. In the second model (on the Balb/c background23), treatment with very large amounts of anti-CTACK polyclonal Ab alone influenced cutaneous T_Hhoming in WT mice. Such classical murine DNFB-induced models of atopic dermatitis, however, raise concerns when used to study cutaneous T_H homing in this particular case: although large numbers of T_H cells do indeed enter such lesions, skin swelling itself (as measured by changes in ear thickness after application of DNFB) is not T-cell–dependent. (We have demonstrated that the classical DNFB-induced ear thickness procedure [used in both of the murine studies under discussion] produces swelling in all T-cell–deficient mouse strains tested, including homozygous nude, scid, and Rag-1– or Rag-2–deficient strains. In such experiments, the increases in ear thickness were indistinguishable from those induced in WT mice of the same genetic background [J.J.C., unpublished data, June 2002]). Thus, antibodies to CTACK must have a not-yet-understood influence on some component of the intrinsic immune system to explain their effects on T-cell–independent swelling.

Therefore, the observed influence of anti-CTACK on cutaneous lymphocyte homing should not necessarily be interpreted as having a direct effect on T-cell trafficking. It is equally plausible that the observed reduction of T-cell infiltrates within DNFB-treated skin is secondary to anti-CTACK–mediated prevention of the inflammatory response itself, through effects on the intrinsic immune system. Further studies will be required to resolve this issue, such as reproducing the finding in strictly T-cell–dependent models of cutaneous inflammation.

Although this manuscript addresses only the putative skin-homing T_H aspects of CCR10/CTACK interactions, it should be noted that another CCR10 ligand (MEC/CCL28) is expressed by exocrine tissues.22 This suggests that CCR10 may have other roles in leukocyte trafficking yet to be discovered.

Our data support the notion that CLA and CCR4 are intimately associated with the process by which most memory T_H cells specifically enter the skin. We suggest that CCR10 may be part of an alternative cutaneous homing pathway utilized primarily by the less numerous “effector” subset of the T_H lymphocyte pool.

We thank Nasim Kassam and Meghan Wells from Millennium Pharmaceuticals for invaluable technical assistance. We thank Drs L. Silberstein, L. Jopling, M. Wurbel, and E. Baekkevold for critical reading of the manuscript and useful discussions of the data. We also thank Rob Hromas for instigating the Spinola Lab/Campbell Lab collaboration.