Research ArticleInflammation

Human TH9 Cells Are Skin-Tropic and Have Autocrine and Paracrine Proinflammatory Capacity

See allHide authors and affiliations

Science Translational Medicine  15 Jan 2014:
Vol. 6, Issue 219, pp. 219ra8
DOI: 10.1126/scitranslmed.3007828


T helper type 9 (TH9) cells can mediate tumor immunity and participate in autoimmune and allergic inflammation in mice, but little is known about the TH9 cells that develop in vivo in humans. We isolated T cells from human blood and tissues and found that most memory TH9 cells were skin-tropic or skin-resident. Human TH9 cells coexpressed tumor necrosis factor–α and granzyme B and lacked coproduction of TH1/TH2/TH17 cytokines, and many were specific for Candida albicans. Interleukin-9 (IL-9) production was transient and preceded the up-regulation of other inflammatory cytokines. Blocking studies demonstrated that IL-9 was required for maximal production of interferon-γ, IL-9, IL-13, and IL-17 by skin-tropic T cells. IL-9–producing T cells were increased in the skin lesions of psoriasis, suggesting that these cells may contribute to human inflammatory skin disease. Our results indicate that human TH9 cells are a discrete T cell subset, many are tropic for the skin, and although they may function normally to protect against extracellular pathogens, aberrant activation of these cells may contribute to inflammatory diseases of the skin.


Interleukin-9 (IL-9)–producing CD4+ T helper type 9 (TH9) cells have been proposed as a newly described proinflammatory subset of TH cells. In mouse models, TH9 cells enhanced immune responses to melanoma and helminth infection and contributed to pathogenicity in autoimmune and allergic animal models of colitis, uveitis, experimental autoimmune encephalomyelitis (EAE), and asthma (111). In humans, IL-9 production was increased in the asthmatic lung and in T cells from infants with atopic dermatitis (1216). Aside from these reports, studies of human TH9 cells have focused almost exclusively on naïve or memory T cells driven to produce IL-9 in vitro by stimulation with exogenous transforming growth factor–β (TGF-β) (6, 7, 14, 17).

Several unanswered questions remain regarding the biology of TH9 cells. First, the prevalence, characteristics, and function of TH9 cells arising in vivo in humans remain unstudied. Second, TH2, TH17, and regulatory T (Treg) cells can produce IL-9 after specific in vitro manipulations, raising the question as to whether IL-9–producing TH9 cells exist as a discrete T cell subset (6, 17). Third, innate lymphoid cells (ILCs) were the major source of IL-9 in one mouse model, and the relative importance of ILCs versus T cell–derived IL-9 has not been established (3). Last, IL-9 production was no longer demonstrable at sites of established tissue inflammation in several animal models, despite the requirement of IL-9 for initial development of the inflammatory state (18, 19). The eventual fate of IL-9–producing T cells in these models remains undetermined.

We report here studies of IL-9–producing T cells isolated from human blood and peripheral tissues. We find that human TH9 cells are a discrete, identifiable T cell subset, largely tropic for the skin, that have the capacity to amplify immune responses by enhancing cytokine production from TH1, TH2, TH9, and TH17 cells.


IL-9 is transiently and selectively produced by skin-tropic TH cells after stimulation with Candida albicans

On the basis of evidence that epithelial barrier tissues are progressively colonized by memory T cells responding to pathogens encountered through that tissue (20, 21), we studied human tissue-tropic T cells from peripheral blood for their capacity to make IL-9. Skin- and gut-tropic memory TH cells were isolated based on expression of the skin-homing receptor cutaneous lymphocyte antigen (CLA) or the gut-homing integrin α4β7. To most closely mimic pathophysiologic stimulation, we stimulated T cells with autologous monocytes pulsed with a panel of viral, bacterial, and fungal pathogens and analyzed subsequent proliferation and cytokine production. We observed enhanced proliferation of tissue-tropic T cells after stimulation with pathogens commonly encountered through that respective epithelial surface (Fig. 1A and fig. S1). CLA+ skin-tropic T cells proliferated more than other populations when stimulated with the skin organism Staphylococcus epidermidis, α4β7+ gut-tropic T cells preferentially proliferated to the gut pathogens rotavirus and Salmonella typhimurium, and the non–gut-tropic/non–skin-tropic T cells (which include lung-homing T cells) proliferated maximally to the lung pathogens influenza A and Mycoplasma pneumoniae. T cell responses to herpes simplex virus-1 (HSV-1) and Candida albicans were maximal in skin-tropic T cells but also present in other tissue-homing populations.

Fig. 1. IL-9 is transiently and selectively produced by skin-tropic TH cells after stimulation with C. albicans.

(A) Tissue-tropic T cells preferentially respond to pathogens encountered through that epithelial surface. The percent maximal proliferation (% max prolif) of T cells tropic for skin (CLA+), gut (α4β7+), and other tissues (CLAα4β7) is shown. CLA+, α4β7+, and CLA4β7 memory CD4+ T cells were isolated from PBMCs of healthy donors and separately stimulated with autologous monocytes pulsed with a panel of viral, bacterial, and fungal antigen preparations. Proliferation was assayed by carboxyfluorescein diacetate succinimidyl ester (CFSE) dilution, and maximal proliferation for each antigen was used to calculate relative responses of other T cell subsets (mean + SEM of duplicates). *P = 0.0001. (B to E) IL-9 was produced selectively by skin-tropic T cells in response to C. albicans. Cytokine production of T cell subsets after stimulation with autologous antigen-pulsed monocytes was measured in cell culture supernatants using a bead-based multiplex assay (mean + SEM of duplicates). Significant IL-9 production was observed only in CLA+ skin-tropic T cells stimulated with C. albicans, whereas IFN-γ, IL-13, and IL-17 were produced in variable amounts by all tissue-tropic subsets and in response to all antigens tested. (F) All tissue-tropic T cell subsets proliferated in response to C. albicans, demonstrating that selective IL-9 production by skin-tropic T cells was not the result of poor recognition of C. albicans by other T cell subsets. The percent CFSElow T cells is shown (mean + SEM of duplicates). (G and H) IL-9 is transiently produced after pathogen stimulation. T cells were stimulated with C. albicans and S. aureus, and IL-9 and IL-17 production was assessed at different time points by flow cytometry after stimulation with phorbol 12-myristate 13-acetate and ionomycin. Cytokine production of stimulated TH cells on day 6 (G) and the kinetics of IL-9 and IL-17 production (H) are shown (mean + SD). Data are representative of independent experiments with five (A and F), three (B to E), or two (G and H) donors.

We next evaluated cytokine production in response to pathogen exposure. Surprisingly, we found substantial IL-9 production only in cultures of skin-tropic CLA+ T cells stimulated with C. albicans–pulsed monocytes (Fig. 1B); in contrast, other TH cell cytokines [for example, interferon-γ (IFN-γ) (TH1), IL-13 (TH2), and IL-17 (TH17)] were detected at variable levels in all tissue-homing subsets and with all pathogens tested (Fig. 1, C to E). IFN-γ suppresses IL-9 production in mouse T cells, and pathogen-induced production of IFN-γ could be masking IL-9 production (22). To evaluate this possibility, we tested pathogen responses in the presence of IFN-γ–neutralizing antibody. IFN-γ blockade significantly increased IL-9 production in response to C. albicans but did not affect IL-9 production in response to other pathogens (fig. S2A). The lack of IL-9 production in C. albicans–stimulated α4β7+ and CLA4β7 TH cells was not a consequence of differential stimulation; TH cell proliferation was observed in all TH cell subsets, although it was highest in CLA+ T cells (Fig. 1F). We further studied the source and expression kinetics of IL-9 produced in response to C. albicans by stimulating peripheral blood mononuclear cells (PBMCs) with C. albicans or S. aureus, organisms known to induce strong TH17 responses (23) (Fig. 1G). Both organisms induced equivalent IL-17 production. However, IL-9 was only induced by C. albicans, and IL-9 production was unique in that it was transient, peaked at day 6, and declined rapidly thereafter (Fig. 1H). These results demonstrate that transient IL-9 expression is specifically induced in skin-homing CLA+ TH cells by C. albicans.

Human TH9 cells are a distinct T cell population tropic for the skin

Our previous experiments, using pathogens to stimulate the cytokine production of circulating peripheral blood effector T cells, found IL-9–producing T cells only within the skin-homing TH population. To determine whether TH9 cells specific for other tissues existed but were not specific for the pathogens we tested, we isolated CLA+, α4β7+, and CLA4β7 CD25 effector TH cells from the blood and polyclonally stimulated these T cells with anti-CD3/anti-CD2/anti-CD28. We analyzed cytokine production by intracellular cytokine staining (Fig. 2, A and B) and multiplex bead analysis (Fig. 2C). IL-9 production was enriched in skin-tropic CLA+ TH cells, although some production was also observed in the gut-tropic and non–skin-tropic/non–gut-tropic CLA4β7 T cells, a population that included lung-homing T cells. The kinetics of IL-9 production were unique in that it was produced only transiently after activation. Multiplex bead analysis of T cell supernatants confirmed that most IL-9 was produced by CLA+ skin-homing T cells (Fig. 2C). At the time of peak IL-9 production (day 2), simultaneous staining for all four cytokines demonstrated that most IL-9+ skin-homing TH cells lacked coexpression of other TH lineage cytokines (IFN-γ, IL-17, and IL-13), consistent with their identity as a distinct T cell subset (Fig. 2, D and E). In the smaller populations of IL-9–producing T cells that did coproduce IFN-γ, IL-13, or IL-17, IL-9 was also transiently produced (fig. S4). In contrast to CLA+ cells and in accordance with previous reports, α4β7+ TH cells highly expressed IFN-γ (24), and CLA4β7 TH cells were enriched for IL-17–producing T cells (25) (Fig. 2B). The production of IL-9 by CLA+ TH cells was not affected by including anti–TGF-β–neutralizing antibody, ruling out potential autocrine induction of IL-9 by endogenous TGF-β (Fig. 2F). These results demonstrate that IL-9 is selectively and transiently produced by a distinct population of TH9 cells expressing cutaneous homing receptors, which suggests that these cells are tropic for the skin.

Fig. 2. Human TH9 cells are a distinct T cell population tropic for the skin.

(A and B) CLA+, α4β7+, and CLA4β7 CD25 memory CD4+ T cells (TEFF) were isolated from healthy donors and polyclonally stimulated with αCD3/αCD2/αCD28. The production of IL-9, IFN-γ, IL-13, and IL-17 was assessed by flow cytometry after stimulation with PMA and ionomycin at the indicated time points. Histograms from an individual donor (A) and aggregate data are shown (B) (mean + SD). Most IL-9 production was observed in CLA+ skin-tropic T cells, and IL-9 was produced only transiently after activation, in contrast to other cytokines tested simultaneously (IL-13, IL-17, and IFN-γ). (C) Cytokine production as measured by bead multiplex analysis of supernatants from tissue-tropic subsets on day 4 after stimulation (mean + SD); results confirmed flow cytometry studies, demonstrating that most IL-9 is produced by CLA+ T cells. (D and E) Most IL-9–producing T cells lacked production of other TH lineage cytokines. Two days after stimulation, CLA+ TEFF cells were analyzed by flow cytometry for coproduction of IL-9 and IL-17, and IFN-γ and IL-13. Representative histograms (D) and aggregate data (E) are shown (mean + SD). (F) IL-9 production was not dependent on the presence of TGF-β. IL-9 in supernatants from T cells stimulated for 4 days in the presence of neutralizing antibody to TGF-β or an isotype-matched control antibody is shown (mean + SD). Data are representative of independent experiments with at least four donors.

TH9 cells are selectively found in human skin and are independent of TGF-β and IL-2

Memory T cells are frequent in human peripheral tissues; there are about 20 billion antigen-experienced memory T cells in the skin of a healthy adult, twice the number of total circulating T cells (26). Analogous populations of memory T cells have been found in human lung and gut (27, 28). Studies in mice have shown that pathogen-specific T cells accumulate in epithelial barrier tissues such as the skin after infection and are protective against pathogen reexposure, even in the absence of circulating T cells (21, 2932). In both mice and humans, a subpopulation of these resident memory T (TRM) cells are sessile and nonrecirculating (20, 21, 31). It is therefore critical to study the T cells resident within peripheral tissues to gain a true understanding of effector memory T cell function. We isolated tissue TRM cells from healthy human skin, small intestine, and lung. TH cells transiently producing IL-9 after activation were found exclusively in skin and were largely absent from gut or lung (Fig. 3A). Similar findings were observed in T cells isolated from skin by collagenase digestion (Fig. 3A) and from short-term explant cultures (fig. S3). Inclusion of neutralizing antibody to TGF-β and the addition of endogenous IL-2 or neutralizing anti–IL-2 antibodies to cultures did not alter IL-9 production (Fig. 3B and fig. S2C), demonstrating that skin-resident TH9 cells were not dependent on TGF-β or IL-2 for survival or cytokine production.

Fig. 3. TH9 cells are selectively found in human skin, constitute a distinct T cell population, and are independent of TGF-β and IL-2.

(A) TH9 cells are resident in human skin but not small intestine or lung. T cells were isolated from healthy human skin, small intestine, and lung and stimulated with αCD3/αCD2/αCD28, and production of IL-9 and IFN-γ was assessed by flow cytometry at the indicated time points after stimulation with PMA and ionomycin. (B) Skin TH9 cells are independent of TGF-β and IL-2. T cells freshly isolated from healthy skin were stimulated as in (A) and cultured in the presence or absence of neutralizing antibody to TGF-β or isotype-matched control antibody (left) or exogenous IL-2 (right). Percentage of IL-9+ TH cells was assessed by flow cytometry at the indicated time points after stimulation with PMA and ionomycin (mean + SD). (C) CD3+ T cells are the major source of IL-9 in human skin. Cell suspensions from human skin were simulated with PMA and ionomycin before or after 12 hours of stimulation with αCD3/αCD2/αCD28 beads. CD3+ T cells were the major source of IL-9 both before and after T cell stimulation. (D and E) ILCs were not an appreciable source of IL-9 in healthy human skin. Cell suspensions from human skin were stimulated for 4 days (4d) with either (D) IL-2 (10 U/ml) + IL-33 (50 ng/ml) (to stimulate ILCs) or (E) αCD3/αCD2/αCD28 beads (to stimulate T cells) and then treated with PMA and ionomycin. IL-9 was not produced by CD3 cells after IL-2 + IL-33 but was produced by CD3+ cells after bead treatment. CD3+ T cells produced IFN-γ without bead stimulation, but IL-9 production was not observed unless T cells were first bead-stimulated. (F) TH9 cells coproduced granzyme B and TNF-α but lacked FoxP3 and production of other TH lineage cytokines. Four days after stimulation, skin-resident T cells were analyzed for coexpression of IL-9 with various cytokines, granzyme B, and FoxP3. (G) Frequency of skin-resident T cells producing IL-9 alone or in combination with IL-17, IFN-γ, or IL-13, 4 days after stimulation. Data are representative of independent experiments with at least three (A, D, and E) or six donors (B, C, F, and G). (C) to (E) are gated to show all viable cells as determined by forward/side scatter. (A) and (F) show viable CD3+/CD8 lymphocytes.

In mice, it has been suggested that ILCs, not T cells, are the main source of IL-9 produced in vivo (3). IL-33 and IL-2 can induce production of IL-9 from human ILCs (33). We observed no IL-9 production after treatment of healthy skin cells with IL-33 and IL-2, but CD3+ T cells were a major source of IL-9, both before and after polyclonal T cell stimulation (Fig. 3, C to E).

Most IL-9+ skin-resident TH cells lacked coexpression of other TH lineage signature cytokines, although minor subsets of IFN-γ+, IL-17+, or IL-13+ TH cells did coexpress IL-9 (Fig. 3, F and G), as revealed by combined intracellular fluorescence-activated cell sorting (FACS) staining for all four cytokines. IL-9 production was transient as observed in CLA+ TH cells from blood, regardless of coproduction of other cytokines. TH cells coproducing IL-9 with two or more TH cytokines were not observed, neither in CLA+ skin-homing nor in skin-resident TH cells (fig. S4). IL-9+ skin-resident TH cells also lacked coexpression of IL-10, IL-22, and the Treg marker FoxP3 but coexpressed tumor necrosis factor–α (TNF-α) and the cytotoxic molecule granzyme B (Fig. 3F), further supporting the notion that TH9 cells are a distinct subset with proinflammatory and cytotoxic properties (4).

Human TH9 cells have autocrine and paracrine activities

In TH9 cells isolated from human blood and skin, we observed that peak IL-9 production preceded maximal activation-induced up-regulation of IFN-γ, IL-17, and IL-13 in TH cells (Fig. 4, A and B). We next investigated if IL-9 could induce or enhance production of other effector cytokines in skin-tropic TH cells. Blocking IL-9 with neutralizing antibody at the initiation of in vitro culture strongly inhibited subsequent up-regulation of IL-9, IFN-γ, IL-13, and IL-17 in CLA+ TH cells (Fig. 4C). IL-9–neutralizing antibody also reduced T cell proliferation at early (day 4), but not late, time points (Fig. 4E) and had no significant effects on cell viability (fig. S2D). In parallel, we found that IL-9 receptor (IL-9R) expression was highly enriched in activated CLA+ TH cells and increased after activation (Fig. 4D).

Fig. 4. Human TH9 cells have autocrine and paracrine proinflammatory activity.

(A and B) IL-9 production is transient and precedes the up-regulation of other inflammatory cytokines. T cells isolated from blood (A) and skin (B) were stimulated with αCD3/αCD2/αCD28, and production of IL-9, IFN-γ, IL-17, and IL-13 was assessed at the indicated time points by flow cytometry after stimulation with PMA and ionomycin. (C) IL-9 production is required for maximal production of IL-9 itself as well for as the production of other inflammatory cytokines. CLA+ TEFF cells were stimulated for 2 days with αCD3/αCD2/αCD28 in the presence of neutralizing antibody to IL-9 or isotype-matched control antibody. Cytokine production was assessed by flow cytometry after stimulation with PMA and ionomycin. (D) Expression of IL-9R is enriched in activated skin-tropic CLA+ TEFF. IL-9R mRNA was measured by real-time quantitative polymerase chain reaction (PCR) in CLA+, α4β7+, and CLA4β7 TEFF after 2 days of stimulation with αCD3/αCD2/αCD28 (mean + SEM, three donors with triplicates). Baseline expression (day 0) versus expression after activation (day 2) of IL-9R by CLA+ TEFF cells is shown in the right panel. (E) IL-9 enhances cellular proliferation at early time points. TEFF cells were labeled with CFSE and stimulated with αCD3/2/28 with anti–IL-9 (αIL-9) or isotype-matched control antibody, and proliferative cells were identified (CFSElow). Proliferation was significantly decreased at day 4 but was unchanged at later time points. The mean and SEM of four donors are shown. Data are representative of independent experiments with six (A and B), five (C and E), and three (D) donors.

Human TH9 cells are increased in psoriatic skin lesions

Given the ability of TH9 cells to enhance the production of proinflammatory cytokines by other T cells, we carried out immunohistochemical studies on the skin lesions of TH1/TH17-mediated (psoriasis) and TH2-mediated (atopic dermatitis) human skin diseases. IL-9–producing T cells were evident within the skin lesions of both diseases (Fig. 5). The number of IL-9–producing cells was significantly higher in psoriatic skin lesions compared to healthy skin (Fig. 5B). Although atopic dermatitis skin lesions also tended to have increased numbers of IL-9–producing cells, this increase was not statistically significant. To establish that IL-9 in psoriatic skin lesions was produced by T cells, we performed double-color immunofluorescence staining of psoriatic skin lesions. A population of IL-9–producing cells expressing CD3 and CD4, but not CD8, was identified (Fig. 6, A to C). A proportion of IL-9–producing T cells were observed to be in the process of cell division (Fig. 6, D and E), consistent with our in vitro observation that IL-9 is produced transiently and only by activated cells. A subset of CD3 IL-9–producing cells were also demonstrable in psoriatic skin lesions, suggesting the possibility that ILCs or other IL-9–producing cell types may contribute to IL-9 production (Fig. 6F).

Fig. 5. IL-9–producing cells are increased in the skin lesions of psoriasis and atopic dermatitis.

(A) Healthy skin and lesional skin samples from patients with psoriasis and atopic dermatitis were immunohistochemically stained for IL-9. Positive cells appear red. (B) TH9 cells are significantly increased in the skin lesions of psoriasis but not atopic dermatitis. The number of IL-9+ cells per square millimeter in immunohistochemically stained sections of healthy, atopic dermatitis, and psoriasis skin is shown. The mean and SEM of two independent experiments with 12 donors each per condition are shown. Scale bars, 50 μm.

Fig. 6. TH9 cells are evident in human psoriatic lesional skin.

(A to C) Two-color immunofluorescence staining was performed on lesional psoriatic skin. A population of IL-9–producing cells expressed (A) CD3 and (B) CD4, and (C) lacked expression of CD8. (D and E) IL-9 production was observed in cells undergoing cell division. (F) A population of CD3 IL-9–producing cells was also present in psoriatic skin lesions. Scale bars, 10 μm. DAPI, 4′,6-diamidino-2-phenylindole.


We report here a comprehensive characterization of human IL-9–producing T cells. We found that IL-9 was primarily produced by a discrete and stable population of T cells, strongly suggesting that TH9 cells do indeed exist in humans. Unlike human and mouse T cells generated by differentiation from naïve cells in vitro, human TH9 cells isolated from human blood and tissues lacked production of other TH lineage cytokines, coproduced TNF-α and granzyme B, lacked FoxP3 expression, and were not dependent on the presence of TGF-β or exogenous IL-2. Both the naturally occurring, in vivo–differentiated TH9 cells studied in this report and TH9 cells generated by in vitro differentiation (7, 17) lack IL-10 production, a feature that distinguishes them from mouse TH9 cells.

Human TH9 cells had unique features that distinguished them from other TH lineages. In healthy adults, TH9 cells were found primarily among the CLA+ skin-homing effector T cell population and were present in healthy human skin, but were effectively absent from human small intestine and lung. A demonstrable proportion of CLA+ TH9 cells were specific for C. albicans, suggesting that under normal conditions, TH9 cells may play a role in protecting against certain extracellular pathogens. In contrast to the production of other inflammatory cytokines, production of IL-9 by TH9 cells was transient, peaking at day 2 after stimulation in circulating TH9 cells and at day 4 after stimulation in skin-resident TH9 cells. Transient production of IL-9 with subsequent down-regulation is a pattern that has also been observed in mouse in vitro–differentiated TH9 cells (34). The transient nature of IL-9 production may explain why in mouse models of TH9-mediated inflammation, IL-9–producing T cells are often not readily demonstrable at the affected sites (18, 19). What becomes of TH9 cells after they stop producing IL-9 is not clear; possibilities include cessation of cytokine production, apoptopic cell death, or differentiation into other TH cell subsets.

In both circulating and skin-resident TH9 cells, maximal IL-9 production preceded the activation-induced up-regulation of other TH inflammatory cytokines including IFN-γ, IL-17, and IL-13. We found that IL-9R expression was highly enriched in activated CLA+ TH cells, suggesting that skin-homing TH cells may be both a source and a target of IL-9. Indeed, blocking IL-9 in vitro strongly inhibited the up-regulation of not only IL-9 but also IFN-γ, IL-13, and IL-17 in CLA+ TH cells and reduced early T cell proliferation. It has been proposed previously that IL-9 functions indirectly via the induction of IL-13 and IL-5, and therefore, IL-9 has a regulatory role on cells that produce TH2 cytokines (3). Our results, however, indicate that human IL-9 induces the production of cytokines from multiple T cell subsets, at least in skin-homing T cells. Our results suggest that TH9 cell activation in tissues could initiate inflammation via TNF-α production and via their autocrine and paracrine activity, leading to enhanced activation and cytokine production by TH1, TH2, TH9, and TH17 cells. Such a role for TH9 cells in the initiation of inflammation is consistent with findings in mouse models where dermal injection of IL-9 enhanced TH17-related psoriasiform inflammation in K5.hTGF-β1 transgenic mice (35), blockade of IL-9 suppressed both production of IL-17 by autoreactive T cells and their ability to initiate disease in adoptive transfer EAE (36), and TH9 cells were found to be critical for the recruitment and activation of IFN-γ–producing anti-melanoma–specific T cells (5). Consistent with a possible role in initiating and enhancing skin inflammation, we found that IL-9–producing TH9 cells were significantly increased in the skin lesions of psoriasis.

Limitations of this study include an inability to track human TH9 cells, as is possible in transgenic mice, to determine the eventual fate of IL-9–producing T cells. Also, the presence of IL-9–producing TH9 cells in human psoriatic skin lesions, their ability to amplify T cell responses, and a role for IL-9 in one mouse model of psoriasiform dermatitis are intriguing, but our results do not constitute proof that TH9 cells play a causative pathogenic role in human psoriasis. Last, the panel of bacterial and viral antigens we used to examine T cell reactivity was diverse but not comprehensive, and it is likely that TH9 cells reactive to multiple other pathogens also exist in vivo in humans.

In summary, we find that memory TH9 cells can be readily isolated from human blood and tissue, are preferentially skin-tropic or skin-resident, and are present in both healthy and diseased human skin. TH9 cells isolated from humans were a discrete T cell subset that showed no dependence on the presence of TGF-β, lacked IL-10 production but coproduced TNF-α and granzyme B, and were essential for maximal inflammatory cytokine production of CLA+ TH1, TH2, TH9, and TH17 cells. The specificity of many memory TH9 cells for C. albicans suggests that, in addition to antitumor effects, TH9 cells may play a critical role in healthy individuals in the defense against extracellular pathogens. The ability of TH9 cells to enhance proliferation and the production of inflammatory cytokines from other T cell subsets and their increased presence in psoriasis skin lesions suggest that TH9 cells may also participate in initiating and maintaining cutaneous inflammation.


Study design/experimental design

This is an experimental laboratory study performed on human tissue samples. All studies were performed in accordance with the Declaration of Helsinki. Blood from healthy individuals was obtained after leukapheresis, skin was obtained from healthy patients undergoing cosmetic surgery procedures, and lung and intestinal specimens were obtained as discarded tissues after resection of small tumors distant from the provided specimens; all tissues were collected with previous approval from the Partners Institutional Review Board. Skin samples for immunohistochemistry and immunofluorescence studies were obtained from patients seen at the Department of Dermatology, Inselspital/University of Bern, Bern, Switzerland. The study on human patient samples was approved by the Medical Ethics Committee of the Canton of Bern, Switzerland (approval 35/08). Written informed consent was obtained from all patients. Enumeration of IL-9–producing cells in immunofluorescence and immunohistochemical studies was done in an investigator-blinded fashion. Mechanistic studies on cells derived from blood and human tissues were performed with in vitro assays without blinding or randomization. Study components were not predefined.

Antibodies and flow cytometry

All antibodies used in this study are listed in table S1. For analysis of cytokine production, T cells were stimulated with either control medium or PMA (50 ng/ml) and ionomycin (750 ng/ml) plus brefeldin A (10 μg/ml) (Calbiochem) for 4 hours. Cells were then surface-stained, fixed, permeabilized, stained with anti-cytokine antibodies, and examined by flow cytometry. Analysis of flow cytometry samples was performed on Becton Dickinson FACSCanto instruments, and data were analyzed with FACSDiva software (V5.1). To set gates, cells were stained with isotype control antibodies and gates were set such that <1% of cells were present in all three positive staining quadrants.

Culture medium

Cell culture medium consisted of X-VIVO 15 (Lonza) supplemented with 2% (v/v) human serum type AB (Sigma). Where indicated, recombinant human IL-2 was added at 50 IU/ml.

Cell sorting

PBMCs from healthy donors were isolated with Histopaque-1077 (Sigma). CD14+ monocytes were isolated by positive selection with magnetic microbeads (Miltenyi). Total CD4+ memory T cells were isolated by negative selection with the memory CD4+ T Cell Isolation kit (Miltenyi) and stained for FACS with anti-α4β7, followed by anti-mouse immunoglobulin G (IgG), then with anti-CLA and anti-CD25. Treg-depleted (CD25 TEFF) CLA+, α4β7+, and CLA4β7 memory T cell subsets were sorted on a FACSAria (BD Biosciences). Sorting purity of T cell subsets was typically more than 95% in post-sort analysis, with the exception of the CLA subset, where a purity of >85% was achieved.

T cell stimulation with pathogen-pulsed monocytes

Commercially available antigen preparations from HSV-1, rotavirus, influenza A (H3N2), and M. pneumoniae were obtained from Microbix. Antigen preparations from S. epidermidis, S. typhimurium, Staphylococcus aureus, and C. albicans were generated by heat-killing microbes at 65°C for 1 hour (all four pathogens), followed by three freeze-thaw cycles (bacteria only). Pathogen concentrations were determined by optical density measurements (Bio-Rad) according to the manufacturer’s instructions. All antigen preparations were used at the concentration that resulted in maximal proliferation in titration experiments using CFSE dilution. Monocytes (2.5 × 104) were preincubated with antigen preparations for 6 hours at 37°C. CFSE-labeled purified CD4+ memory T cell subsets (5 × 104) were then cocultured with the antigen-pulsed autologous monocytes for 7 days before analysis of CFSE dilution on a FACSCanto (BD Biosciences). CFSE labeling was performed according to standard protocols. On day 5 of coculture, supernatants were saved and cytokine concentrations were measured with a custom-made Luminex bead assay, as previously described (37). For analysis of pathogen-induced IL-9 expression kinetics, PBMCs (4 × 105) were cultured with C. albicans or S. aureus and analyzed for the expression of IL-9 and IL-17 at different time points by intracellular FACS staining after stimulation for 5 hours with PMA and ionomycin. For experiments measuring IL-9 production in the presence or absence of IFN-γ blockade (fig. S2A), CFSE-labeled PBMCs (2 × 105) were cultured in the presence of anti–IFN-γ (10 μg/ml) or mouse IgG (10 μg/ml). On day 5 of culture, supernatants were harvested and IL-9 concentrations were measured with an IL-9 ELISA kit (ELISA MAX, BioLegend) according to the manufacturer’s instructions.

Polyclonal activation of TEFF and cytokine expression kinetics

FACS-sorted TEFF subsets (2 × 104) were polyclonally activated using beads coated with antibodies against CD3, CD2, and CD28 (T cell/bead = 2:1, Miltenyi). Before activation and at different time points thereafter, T cells were additionally stimulated for 5 hours with PMA and ionomycin in the presence of brefeldin A and then intracellularly stained for IL-9, IFN-γ, IL-13, and IL-17 for FACS analysis according to standard protocols. Supernatants were collected at day 4 after polyclonal stimulation, and cytokines were measured as described above. Where indicated, the following neutralizing antibodies were added at the start of the culture (all at 10 μg/ml): anti–TGF-β (clone 1D11, R&D Systems), anti–IL-9 (MH9D1, BioLegend), IgG1 control antibody (MG1-45, BioLegend), anti–IL-2 (4 μg/ml, polyclonal goat IgG; R&D Systems), or polyclonal goat IgG (4 μg/ml, R&D Systems).

Quantitative real-time PCR for IL-9R

Total RNA was isolated from FACS-sorted T cell subsets 2 days after polyclonal activation (as described above) with the RNeasy Plus Mini Kit (Qiagen) according to the manufacturer’s instructions. Complementary DNA was generated with the SuperScript VILO cDNA Synthesis kit (Life Technologies), and quantitative real-time PCR was performed with the ABI StepOnePlus Instrument and the Fast SYBR Green Master Mix (Applied Biosystems). Expression of each ligand transcript was determined relative to the reference gene transcript (β-actin) and calculated as 2−(Ct, ligandCt, β-actin). Data are presented as arbitrary units. The primers used to detect the transcripts were purchased from Integrated DNA Technologies and were as follows: IL-9R (forward, 5′-GGGTGACAAATCACCTCCAG-3′; reverse, 5′-GCCTCACTCTCCAAGGTCC-3′) and β-actin (forward, 5′-TCACCCACACTGTGCCCATCTACGA-3′; reverse, 5′-CAGCGGAACCGCTCATTGCCAATGG-3′).

Isolation and polyclonal activation of tissue-resident T cells

Skin was obtained from healthy subjects undergoing cosmetic surgery procedures. Lung and small intestine samples were obtained from resection margins distant from the pathology of patients undergoing lung or gut surgery for various reasons. Tissue samples were extensively minced and then incubated for 2 hours at 37°C in RPMI 1640 containing 0.2% collagenase type I (Invitrogen) and deoxyribonuclease I (30 Kunitz units/ml) (Sigma-Aldrich). Thereafter, cells were collected by filtering the collagenase-treated tissue through a 40-μm cell strainer (Fisher Scientific). For experiments shown in fig. S3, T cells were isolated from 3-week explant cultures maintained in IL-2 and IL-15 as previously described (38). After being washed, tissue-derived T cells were then activated polyclonally using anti-CD3/anti-CD2/anti-CD28–coated beads, analogous to the activation of blood-derived T cells described above. Analysis of cytokine expression in TRM cells before and at different time points after polyclonal activation and experiments with neutralizing antibody to TGF-β were performed as described above for blood-derived T cells.

Isolation and activation of skin-resident ILCs

Cells isolated from healthy skin were plated at a density of 5 × 104/ml in 96-well plates and stimulated for 4 days with IL-2 [10 μg/ml; National Cancer Institute (NCI)] plus IL-33 (50 ng/ml; R&D Systems) or with anti-CD3/anti-CD2/anti-CD28–coated beads. On day 4 of culture, cells from all wells were stimulated for 5 hours with PMA and ionomycin. ILCs (Lin; CD11b CD3, CD56) and T cells (CD3+) were examined for IL-9 production by flow cytometry.

Immunohistochemical staining

Patients with chronic plaque psoriasis (3 women and 10 men) with a median age of 54 years (SD ± 11 years) and patients with atopic dermatitis (3 women and 7 men) with a median age of 45 years (SD ± 20 years) were included. Diagnosis was based on typical clinical and histopathological criteria. The patients did not receive systemic or local therapy at the site where biopsy specimens were obtained for at least 3 weeks before the investigation. Punch biopsies (5 mm) were taken from lesional skin. Healthy skin was obtained from X healthy control subjects undergoing reconstructive surgery for aesthetical reasons. Skin samples were immediately embedded in optimal cutting temperature (OCT) compound, snap-frozen, and stored at −70°C until sectioning. Immunostaining was performed with the streptavidin-biotin complex/alkaline phosphatase method, as previously described (39). Briefly, cryostat-cut tissue sections were air-dried, fixed for 10 min in 4% ice-cold acetone, and rehydrated in tris-buffered saline with 0.1% saponin. The sections were incubated with the primary antibody for 1 hour at room temperature, followed by a biotinylated secondary antibody and thereafter with streptavidin-biotin complex/alkaline phosphatase method (K0376; DakoCytomation). Finally, all sections were developed in new fuchsin-naphthol (KO624, DakoCytomation) and counterstained with hematoxylin. All skin sections were quantitatively analyzed with the digital image acquisition and analysis system NIS-Elements Software BR 2.30 (Nikon) in an investigator-blinded fashion, as previously described (39, 40).

Immunofluorescence double staining

Double immunofluorescence was performed as previously described (39). Briefly, tissue sections were serially incubated with the primary antibody (anti–IL-9) followed by incubation with Alexa Fluor 488–labeled goat anti-rabbit IgG for 1 hour each. Sections were then incubated with one of the following mouse anti-human antibodies for 1 hour: (i) anti-CD3, (ii) anti-CD4, and (iii) anti-CD8. Sections were then treated with Alexa Fluor 594–labeled goat anti-mouse IgG1 (Invitrogen). The specificity of the reaction was confirmed by omitting the primary or secondary antibodies as well as by using irrelevant isotype-matched antibodies as negative controls.

Statistical analyses

Primary methods of data analysis included descriptive statistics (means, medians, and SD). Differences between two sample groups were detected with the one-tailed Wilcoxon-Mann-Whitney test, α = 0.05. For comparisons of multiple groups, a Kruskal-Wallis one-way analysis of variance (ANOVA) with a Bonferroni-Dunn’s post test for multiple means test was used, α = 0.05.


Materials and Methods

Fig. S1. Tissue-tropic TH cells proliferate preferentially to pathogens commonly encountered through that epithelial tissue.

Fig. S2. Additional effects of IL-2, IFN-γ, and IL-9 antibody blockade on human TH9 cells.

Fig. S3. TH9 cells are enriched in human skin, and IL-9 is transiently produced.

Fig. S4. TH9 cell coproduction of other cytokines.

Table S1. Antibodies used in this study.


  1. Acknowledgments: We thank E. Butcher for anti-α4β7 (ACT-1) and B. Richards (Brigham and Women’s Hospital tissue bank), B. Pomahac, S. Talbot, and E. Eriksson (Brigham and Women’s Hospital), and T. Cochran (Boston Center) for providing access to human tissue samples. We also thank F. Sallusto and J. Lederer for technical support. Funding: Supported by a Damon Runyon Clinical Investigator Award (to R.A.C.), R01 AR056720 (to R.A.C.), R01 AR063962 (to R.A.C.), R03 MH095529 (to R.A.C.), the SPORE in Skin Cancer P50 CA9368305 NIH/NCI (to T.S.K.), R01 A1025082 NIH/National Institute of Allergy and Infectious Diseases (to T.S.K.), R01 AI097128 (to T.S.K. and R.A.C.), the Swiss National Science Foundation and the Fondation René Touraine (to C.S.), a Special Fellow Award from the Leukemia & Lymphoma Society (to R.W.), and a grant from the German Research Foundation (to E.G.). Author contributions: C.S. and A.G. carried out most of the experiments, drafted figures, and participated in writing the manuscript. C.Y., R.W., L.C., E.G., and J.E.T. assisted in carrying out experiments and helped edit the manuscript. N.Y. carried out immunohistochemical studies of psoriatic and atopic dermatitis skin. T.S.K. provided advice on approaches and edited the manuscript. R.A.C. supervised studies, drafted figures, and, together with C.S., wrote, edited, and revised the manuscript. Competing interests: R.A.C. and T.S.K. previously had an equity interest in TremRX, a start-up company that seeks as a long-term business plan to improve vaccine formulation and delivery. During the period R.A.C. and T.S.K. held the equity, the interest was deemed to create a financial conflict of interest (as defined by the specific Public Health Service regulations) with the research discussed in this article. To resolve this matter, R.A.C. and T.S.K. divested themselves of the equity interest in this company, so this financial conflict of interest no longer exists. R.A.C. has served as a consultant for Novartis, Dermira, and Stiefel. The other authors declare that they have no competing interests.

Stay Connected to Science Translational Medicine

Navigate This Article