The effects of thymic stromal lymphopoietin and IL-3 on human eosinophil–basophil lineage commitment: Relevance to atopic sensitization

An important immunopathological hallmark of allergic disease is tissue eosinophilic and basophilic inflammation, a phenomenon which originates from hemopoietic progenitors (HP). The fate of HP is determined by local inflammatory cytokines that permit “in situ hemopoiesis,” which leads to the accumulation of eosinophils and basophils (Eo/B). Given that recent evidence supports a critical immunomodulatory role for thymic stromal lymphopoietin (TSLP) in allergic inflammation, as well as TSLP effects on CD34+ progenitor cytokine and chemokine secretion, we investigated the role of TSLP in mediating eosinophilo- and basophilopoiesis, the mechanisms involved, and the association of these processes with atopic sensitisation. In the studies presented herein, we demonstrate a direct role for TSLP in Eo/B differentiation from human peripheral blood CD34+ cells. In the presence of IL-3, TSLP significantly promoted the formation of Eo/B colony forming units (CFU) (including both eosinophils and basophils) from human HP (HHP), which was dependent on TSLP–TSLPR interactions. IL-3/TSLP-stimulated HHP actively secreted an array of cytokines/chemokines, key among which was TNFα, which, together with IL-3, enhanced surface expression of TSLPR. Moreover, pre-stimulation of HHP with IL-3/TNFα further promoted TSLP-dependent Eo/B CFU formation. HHP isolated from atopic individuals were functionally and phenotypically more responsive to TSLP than those from nonatopic individuals. This is the first study to demonstrate enhanced TSLP-mediated hemopoiesis ex vivo in relation to clinical atopic status. The capacity of HHP to participate in TSLP-driven allergic inflammation points to the potential importance of “in situ hemopoiesis” in allergic inflammation initiated at the epithelial surface.


Introduction
The airway epithelium is an important initiator of the allergic response; it secretes cytokines/chemokines, which regulate innate immune cells [1]. One such cytokine is thymic stromal lymphopoietin (TSLP), an IL-7-like cytokine [2], which can be elicited from human airway epithelial cells by cytokines and pathogen-associated molecular patterns [3,4]. Many effector cells involved in allergic diseases such as eosinophils [5], basophils [6,7], and mast cells [3,8] have all been shown to respond to TSLP with increased survival, differentiation, and cytokine secretion. TSLP signals through a heterodimeric receptor complex consisting of the IL-7Ra chain and a TSLP binding chain (TSLPR) [9]. TSLP is known to signal through the JAK/STAT and MAPK pathways [10,11], both of which are involved in the differentiation of hemopoietic progenitor cells into eosinophils [12,13]. However, the pathways involved in basophilopoiesis remain unclear [7]. Nonetheless, human eosinophils and basophils are closely linked during development and share a common progenitor [14][15][16], while in the mouse these differentiative pathways appear distinct [17].
Human eosinophils and basophils differentiate from a common committed CD34þ hemopoietic progenitor cell, the eosinophil-basophil (Eo/B) progenitor, found in bone marrow, cord blood, and peripheral blood (PB) [14]. We have previously provided evidence that allergic inflammation is, at least in part, a result of CD34þ progenitors homing to sites of inflammation where they differentiate, under the control of local inflammatory cytokines, into eosinophils and basophils, a process referred to as ''in situ hemopoiesis'' [18][19][20]. This overarching concept is supported of findings of many investigators: Siracusa et al. [6], demonstrated that cytokines found at sites of inflammation (IL-3 or TSLP) can differentially impact the differentiation of murine progenitors into effector cells (basophils), resulting in functional and phenotypic heterogeneity; Sergejeva et al. [21], reported that $10% of the eosinophilic cells found in murine bronchial alveolar lavage fluid postallergen exposure was derived from eosinophil-lineage committed precursor cells, or local production of eosinophils within the airway; Robinson et al. [22], Kim et al. [23], and Dorman et al. [24] collectively showed that human CD34þ progenitors are detected in the bronchial and nasal mucosa, and sputum, respectively, of patients with atopic asthma and nasal polyposis, with increased numbers of CD34þ/IL-5Raþ cells found in the airways and sputum of asthmatics following allergen challenge, suggesting that CD34þEo/B lineage committed cells are found in the tissue [22,24]; furthermore, Allakhverdi et al. [8] demonstrated that human CD34þ progenitors can be induced by TSLP to produce Th2 cytokines, principally IL-5 and IL-13, and that these double-positive CD34þ cells are present in sputum after airway allergen challenge of atopic asthmatics, suggesting that progenitors may act as proinflammatory effector cells and directly contribute to allergic inflammation.
Recent evidence supports a critical immunomodulatory role for TSLP in allergic inflammation, as well as TSLP effects on CD34þ progenitor cytokine and chemokine secretion [8], but the biological effects of TSLP on human PB CD34þ progenitor Eo/B lineage commitment have not been previously described. In this study, we examine the influence of TSLP on IL-3-dependent CD34þ progenitor differentiation via phenotypic and functional human hemopoietic progenitor (HHP)-related Eo/B lineage commitment. Additionally, we elucidate the mechanisms through which TSLP enhances IL-3-mediated eosinophilo-and basophilopoiesis, and the association of these processes with atopic sensitisation.

Subjects
This study was approved by the Hamilton Health Sciences Research Ethics Board (approval number 08-015) and all subjects provided written informed consent. Atopy-unattributable subjects were initially recruited for the study (Figs. 1-4), following which, subjects with (n ¼ 10) or without (n ¼ 10) atopy were recruited (Fig. 5). Atopy was defined as a positive skin prick test response (>2-mm wheal) to at least one of 14 common aeroallergens. Further subject characteristics are shown in Table 1.

Blood collection and processing
One hundred mL of blood were collected through direct venipuncture into heparinized vacutainer tubes (Becton-Dickinson, Franklin Lakes, NJ, USA). Peripheral blood mononuclear cells were isolated by density centrifugation and CD34þ progenitors were enriched using EasySep TM Human Progenitor Cell Enrichment Kit with Platelet Depletion (STEMCELL Technologies, Vancouver, BC, Canada) as per manufacturer's instructions.
progenitors were stained as previously described with modification [18]. Briefly, cells were washed with fluorescence-activated cell sorting buffer (PBS containing 0.1% sodium azide) and resuspended in murine block (1 Â 10 5 cells/tube) and incubated in the dark (15 min, 48C). Next, optimal amounts of isotype controls or test antibodies were added and incubated in the dark (30 min, 48C). Cells were then washed, fixed in cytofix (BD), and stored in the dark at 48C until ready for acquisition.

Acquisition and analysis
Stained cells were acquired with a LSR II flow cytometer (BD Biosciences) using the FACSDiva software (BD Biosciences). Offline analysis was performed using FlowJo software (Tree Star, Ashland, OR, USA). CD34þ cells were enumerated using a previously established multi-parameter sequential gating strategy [18]. CD34þ progenitor cells were identified as having high CD34 expression, low-intermediate CD45 expression, and low forward and side scatter (Supplementary Fig. S1). Receptor expression data were collected as the percentage of positive cells at the 98% confidence limit (i.e., relative to a quadrant marker set to include 2% of cells stained with isotype control antibody). Median fluorescence intensity (MFI) is defined as the MFI of the receptor of interest divided by the MFI of the isotype control. Due to the use of an enrichment protocol, absolute numbers of CD34þ cells were not determined.
Individual Eo/B CFU cells were picked from methylcellulose and placed into PBS. Cytospin preparations were made on glass slides using Shandon Cytocentrifuge 3 (Shandon Southern Instruments, Cambridge, UK). Eosinophils were identified using Diff-Quik (Siemens, Erlangen, Germany) and basophils identified using toluidine blue stain (Sigma).

Histamine assay
The total number of cells in each individual colony was enumerated using inverted light microscopy before being picked from methylcellulose and placed into PBS, boiled (998C, 5 min), centrifuged, and cell-free supernatant harvested and measured for histamine content using Histamine Enzyme Immunoassay Kit (Bertin Pharma, Montigny-le-Bretonneux, France) according to manufacturer's recommendation. The detection limit of this assay is 55 pg/mL.

Statistical analysis
All data are expressed as the mean AE SEM. Significance was assumed at P < 0.05. All analyses were performed with Prism version 5 (GraphPad Software, La Jolla, CA, USA) using non-parametric tests. Differences within groups were assessed by Friedman test with Dunnett post hoc test. Between-group comparisons (nonatopic vs. atopic) were made using the Mann-Whitney U-test.  One independent experiment performed per subject. Ã P < 0.05; ÃÃ P < 0.01; ÃÃÃ P < 0.001. ND; not detected.

(c)
A representative overlay histogram for expression of TSLPR on unstimulated (shaded gray), IL-3 and TNFa (gray line) and IL-3 and TSLP (black line) stimulated CD34þ cells. Normal mouse IgG was used as an isotype control (shaded-light gray). Percent expression is the percent of CD34þ cells expressing a given antigen at the 98% confidence limit (i.e., relative to a quadrant marker set to include 2% of cells stained with isotype control antibody). This experiment was performed four independent times with similar results. (d) PB CD34þ cells were pre-stimulated with IL-3 (1 ng/mL) and TNFa (50 pg/mL) for 24 h and cultured in methylcellulose colony assays in the presence of IL-3 (1 ng/mL) and increasing doses of TSLP (0.1, 1, 10 ng/mL). On day 14, Eo/B CFU were enumerated (n ¼ 5 in duplicates, ÃÃ P < 0.01; ÃÃÃ P < 0.001 compared with TSLP (0 ng/mL) within group). Results shown are mean AE SEM. One independent experiment performed per subject. Ã P < 0.05; ÃÃ P < 0.01; ÃÃÃ P < 0.001 P < 0.01; Fig. 1b). Overnight stimulation with IL-3/TSLP significantly increased percent expression of TSLPR compared to unstimulated (P < 0.001) and TSLP-stimulated HHP (P < 0.05). A similar trend was seen for IL-7Ra expression, although not significant (Fig. 1c). No significant difference in TSLPR expression was observed following IL-5/ GM-CSF and/or TSLP stimulation (Supplementary Fig. S3).

TSLP enhances IL-3-induced basophilopoiesis from PB HHP
Histochemical staining of individual colonies with DiffQuik and toluidine blue revealed the presence of cells with either eosinophilic granular cytoplasm or metachromatic granules, morphologically consistent with eosinophils and basophils, Percent expression is the percent of CD34þ cells expressing a given antigen at the 98% confidence limit (i.e., relative to a quadrant marker set to include 2% of cells stained with isotype control antibody). Results shown are mean AE SEM. One independent experiment performed per subject. Ã P < 0.05; ÃÃ P < 0.01; ÃÃÃ P < 0.001. ND; not detected. C. C. K. Hui et al.

DISCUSSION
We demonstrate for the first time that PB HHP respond directly to TSLP in vitro with enhanced Eo/B colony formation and TSLPR/IL-7Ra expression, specifically upon co-stimulation with IL-3, and not IL-5 or GM-CSF. This differentiation process appears dependent on autocrine and/ or paracrine signaling by TNFa-producing progenitors. IL-3 and TNFa, cytokines which are found at sites of allergic inflammation, may help explain the increase the sensitivity of HHP to TSLP-mediated Eo/B differentiation. Finally, we demonstrate enhanced stimulatory effects of IL-3 and TSLP on PB CD34þ progenitors derived from atopic individuals.
To the best of our knowledge, this is the first study to demonstrate such findings in humans ex vivo. These findings provide a novel mechanism underlying eosinophil and basophil accumulation in tissues during allergic inflammation, linked as it is to TSLP and its known amplification of Th2 immune responses in atopic individuals. TSLP-TSLPR interactions are crucial to the development of eosinophilia [25] and basophilia [6] in mice; however, its role in humans is unclear. We show for the first time the importance of TSLP-TSLPR in Eo/B lineage commitment of IL-3-responsive HHP. Our group and others have previously reported the presence of both eosinophils and basophils in IL-3-stimulated hemopoietic progenitor cultures [14][15][16]; our current findings therefore demonstrate that TSLP can serve the role of a key epithelial-derived factor in this process of human Eo/B differentiation. Differential counts of colony cells were not formally performed; therefore, relative proportions of eosinophils versus basophils in these colonies are unclear. However, methylcellulose colony assays have been used for many years by us and others, to both enumerate and assess progenitors and their progeny in response to hemopoietic cytokines (and other stimuli), as quantitated by Eo/B CFU [14,19,26]. We have elected to use the CFU assay, and add histamine assays, as previously extensively documented by us and others, to represent a specific, surrogate biomarker of basophil ''content'' within these Eo/B colonies, which are each derived from a single progenitor, given that histidine decarboxylase is only present within the basophil, not the eosinophil in these mixed basophil-eosinophil colonies [14,20,[27][28][29]. Since cells within these Eo/B colonies are rather immature and often possess dual phenotypes -including cells with ''hybrid'' eosinophilic-basophilic granulation by standard morphological-histochemical assessments -differentiating the cells using histochemical stains is not reliable. When the latter are routinely performed, both toluidine blue-positive granules as well as eosinophilic staining granules are found in many cells simultaneously [14][15][16]. As such, our key (and to our judgment, more robust) arbiter of basophil content remains the colony histamine content. Indeed, our group has previously reported on the close correlation between the basophil numbers and the histamine content in Eo/B CFU [14,20,30]. Additionally, in the current study, the histamine assay allowed for comparative analyses, demonstrating marked differences in histamine content between colonies grown in the presence or absence of TSLP and/or neutralizing anti-TSLP or anti-TNFa. While further analyses are required to examine precisely how TSLP may alter colony histamine content, it appears unique among the epithelialderived cytokines in its ability to promote basophilia in peripheral tissues [6]. Of note, Siracusa et al. [6] recently reported the ability of TSLP to induce IL-3-independent basophils from bone marrow-resident precursors in mice. However, we were unable to observe TSLP alone-mediated Eo/B CFU, suggesting that TSLP must work in concert with IL-3 to induce Eo/B differentiation. We speculate that this discrepancy may reflect inter-species differences in hemopoiesis due to the distinct surface phenotype of progenitors in mice and humans [31], or to the distinct differentiative pathways of eosinophil and basophil development in humans and mice [14,17], or to a combination of these factors. As such, rather than examining the relative proportions of eosinophils versus basophils in these colonies of nascent eosinophils and basophils of mixed granulation (which one also sees in liquid cultures), we have focused on PB Eo/B CFU (and thus, HHP) production after TSLP stimulation, providing novel evidence that TSLPR engagement on PB CD34þ cells has the capacity to enhance Eo/B lineage priming of myeloid progenitors, increasing the likelihood of development of allergic eosinophilic/basophilic inflammation, in addition to disease maintenance or progression.
Allakhverdi et al. recently demonstrated the ability of TSLP, together with IL-33, to promote ''Th2-like'' properties in human CD34þ progenitors, based on induction of Th2 cytokines (IL-6, IL-13, GM-CSF) and chemokines (CXCL8, CCL1, CCL17, CCL12) [8]. Likewise, we detected levels of IL-1b, IL-6, IL-13, TNFa, CXCL8, CCL2, CCL5, and CCL17 following overnight stimulation of PB HHP with TSLP; IL-4, IL-9, IL-10, GM-CSF, IFNg, and eotaxin were undetected under all conditions. However, contrary to findings by Allakhverdi et al. [8], we were unable to detect levels of GM-CSF in the supernatant and of the cytokines detected, the concentrations were comparatively low. The higher levels of cytokines reported by Allakhverdi et al. may be due to the use of stem cell factor (100 ng/mL) in their culture medium [32]. Of the Th2 cytokines and chemokines detected, of importance is the enhanced secretion of TNF-a, which, in conjunction with TSLP/IL-33, has been previously shown to enhance cytokine secretion by human cord blood and PB-derived CD34þ progenitors [8]. Furthermore, Caux et al. [33] showed enhancing effects of TNFa on IL-3-and GM-CSF-dependent proliferation of CD34þ HHP. It is plausible that IL-3/TSLP-induced TNFa is a key event in HHP autocrine secretion of Th2-like cytokines/chemokines and in Eo/B differentiation. Our observation that inhibiting TNFa reduces TSLPR expression on HHP, which is in agreement with a study that showed enhanced TSLPR expression on mature eosinophils post IL-3/TNFa-stimulation [5], may explain the reduction in Eo/B CFU formation and colony histamine levels in our cultures. The relevance of TNFa antagonists in decreasing eosinophil and/or basophil counts in vivo is unclear; however, anti-TNFa agents have been reported to decrease sputum histamine levels and improve asthma outcomes, airway hyper-responsiveness, and exacerbation rates [34,35]. Furthermore, in OVAsensitized allergic rhinitis and bleornycin-induced pulmonary fibrosis murine models, TNFa antagonists have been shown to inhibit eosinophilia in the nasal mucosa and lung, respectively [36,37].
Although we have not herein reported on which signaling pathways are involved in TSLP-induced Eo/B differentiation, we do have findings implicating the preferential dependence of p38MAPK signaling pathways in IL-3/TSLPmediated Eo/B differentiation (Supplementary Fig. S4). Indeed, Fanat et al. [38] showed that supernatants from TNFa-, IL-1b-, and IFNg-stimulated human airway smooth muscle (HASM) cells drive eosinophilic differentiation from HHP in vitro, in a p38MAPK-dependent way.
Atopic sensitization is a widely recognized risk factor for allergic diseases [40]. In the current study, we show that PB HHP from atopic individuals are more responsive to IL-3 and even more so to TSLP, resulting in elevated numbers of Eo/B CFU post IL-3/TSLP-stimulation, compared to HHP from nonatopic individuals. This novel finding related to TSLP effects is in keeping with previous observations that atopic sensitization and disease are associated with enhanced PB (and bone marrow) HHP Eo/B lineage commitment and tissue allergic inflammation related to more ''classic'' eosinophilopoietic cytokines such as IL-5 [18,22,23,26]. Furthermore, our group has demonstrated increased levels of IL-5-responsive progenitors in atopic subjects, compared with nonatopic subjects [18]. Likewise, the observation that HHP from atopic individuals respond more robustly to TSLP/IL-3 may be related to increased levels of IL-3-responsive progenitors. Alternatively, PB HHP from atopic subjects may have increased endogenous expression of IL-3, with consequent autocrine effects on Eo/B differentiation, as Kuo et al. [41] have shown with IL-5. The increased TSLP responsiveness of HHP from atopic individuals may be a reflection of increased production of TNFa compared to HHP from nonatopic individuals (Fig. 6c). We show in the current study that TNFa, together with IL-3, enhance TSLPR expression and HHP sensitivity to TSLP-mediated Eo/B differentiation. These findings are consistent with an atopic priming effect on HHP, such that they differentiate into Eo/B CFU more readily to IL-3/TSLP than HHP from nonatopic individuals, contributing to the development of tissue eosinophilia/basophilia. Of note, given the lack of any significant effects of TSLP on IL-5-and GM-CSF-Eo/B differentiation in non-attributable PB HHP (Fig. S2), we did not examine the combination of IL-5/TSLP nor GM-CSF/TSLP in the HHP of known atopic individuals, but rather concentrated on the TSLP/IL-3-mediated pathway. There is also documented increased secretion of TSLP from bronchial epithelial cells of atopic, versus nonatopic individuals [4], suggesting that TSLP may play an important tissue role, such as enhancement of ''in situ hemopoiesis'' in atopy. Although the precise mechanisms are unclear, we postulate paracrine effects of epithelial cells on tissue-resident HHP, which have been detected at mucosal sites of allergic inflammation [22,23]. We demonstrate in the current study that IL-3/TSLP induces TNFa and IL-1b secretion, previously shown to mediate TSLP secretion from airway epithelial cells [3]. Moreover, our data illustrate that pre-stimulation of PB CD34þ cells with IL-3 and TNFa upregulates TSLPR expression, which significantly increases the sensitivity of CD34þ cells to TSLP-mediated Eo/B differentiation. It is possible that epithelial-derived TSLP, along with local hemopoietic factors (i.e., IL-3), both of which are elevated in atopic individuals [4,42], may target HHP resident at mucosal sites, triggering the paracrine and/or autocrine differentiation of these recruited progenitor cells (Fig. 6). TNFa can be produced by various cell types other than CD34þ progenitors; however, in the current study, CD34þ progenitors were enriched using a protocol which virtually eliminated all T-and B-cells.
Although flow cytometry was not performed to check for residual T cells following separation, the ability of CD34þ cells themselves to produce Th2-like cytokines and TNFa following TSLP stimulation and toll-like receptor-ligation has previously been reported [3,43].
Finally, the preferential synergy of TSLP with IL-3, and not IL-5 or GM-CSF, in driving Eo/B differentiation is consistent with previous work which demonstrates that the responsiveness of progenitors to these cytokines is highly dependent on the stage of differentiation [44,45]. PB CD34þ cells consist of mainly immature progenitors, and therefore mostly respond to IL-3 and GM-CSF, early-acting cytokines [44,46], which is consistent with our findings ( Fig. 1a and Supplementary Fig. S2). Of note, IL-3 tends to drive CD34þ progenitor differentiation primarily along the basophil-lineage [47], while GM-CSF promotes the differentiation of a mixed population (Eo/B cells, macrophages, etc.) [46]. Therefore, the preferential differentiation of PB CD34þ progenitors to IL-3-responsive lineage suggests that the response of PB HHP to TSLP involves basophilmediated inflammation [6].
In summary, our study uncovers a previously unrecognized role for TSLP in allergic inflammation -as a regulator of in situ hemopoiesis, by enhancing the process of HHP (CD34þ cell) differentiation within tissues. The enhanced TSLP-mediated hemopoiesis ex vivo in relation to clinical atopic status may help explain increased eosinophils and basophils at sites of inflammation in atopic individuals.

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of this article at the publisher's web-site. Figure S1. Flow cytometric multi-gating strategy for receptor expression on PB CD34þ cells. Figure S2. TSLP has no effect on IL-5-and GM-CSFresponsive Eo/B CFU. Figure S3. IL-5 and GM-CSF has no effect on TSLPR expression. Figure S4. p38MAPK signal transduction is preferentially involved in TSLP-mediated Eo/B differentiation.