Activation of Human Basophils by A549 Lung Epithelial Cells Reveals a Novel IgE-dependent Response Independent of Allergen

Abstract

Evidence for epithelial cell (EC)-derived cytokines (e.g. TSLP) activating human basophils remains controversial. We therefore hypothesize that ECs can directly activate basophils via cell-to-cell interaction. Basophils in medium alone or with IL-3±anti-IgE, were co-incubated with TSLP, IL-33, or IL-25. Analogous experiments co-cultured basophils (1–72h) directly with EC lines. Supernatants were tested for mediators and cytokines. Antibodies targeting receptors were tested for neutralizing effects. Lactic acid (pH 3.9) treatment combined with passive sensitization tested the role of IgE. Overall, IL-33 augmented IL-13 secretion from basophils co-treated with IL-3, with minimal effects on histamine and IL-4. Conversely, basophils (but not mast cells) released histamine and marked levels of IL-4/IL-13 (10-fold) when co-cultured with A549 EC and IL-3, without exogenous allergen or IgE cross-linking stimuli. The inability to detect IL-33/TSLP, or to neutralize their activity, suggested a unique mode of basophil activation by A549 EC. Half-maximal rates for histamine (4h) and IL-4 (5h) secretion were slower than observed with standard IgE-dependent activation. Immunoglobulin stripping combined with passive sensitization±omalizumab showed a dependency for basophil-bound IgE, substantiated by requirement for cell-to-cell contact, aggregation, and FcεRI-dependent signaling. A yet unidentified IgE-binding lectin associated with A549 EC is implicated after discovering that n-acetyllactosamine suppressed basophil activation in co-cultures. These findings point to a lectin-dependent activation of basophil requiring IgE but independent of allergen or secreted cytokine. Pending further investigation, we predict this unique mode of activation is linked to inflammatory conditions whereby IgE-dependent activation of basophils occurs despite absence of any known allergen.

Introduction

Human basophils secrete inflammatory mediators and cytokines central to the pathogenesis of allergic disease (1). In particular, pre-formed histamine and newly generated leukotriene C4 (LTC4) are released within minutes upon cross-linking of the high affinity IgE receptor, FcεRI. This mode of activation also results in IL-4 production de novo, but with secretion first evident after 1h and peaking by ~4h. Basophils also secrete IL-13 in response to IgE-dependent activation, but are generally more proficient in making this cytokine in response to IL-3.

The notion of basophils being a significant source of IL-4 has been substantiated with development of IL-4 reporter mouse strains, which clearly demonstrate the capacity of these cells to secrete this cytokine in vivo (2, 3). In fact, several studies have shown that basophils, by producing IL-4, facilitate the pro-Th2 activities of other immune cells, including T cells (4), B cells (5), monocytes (6, 7), eosinophils (8) and, most recently, innate lymphoid cells type 2 (ILC2) (9).

Although studies clearly indicate that human and mouse basophils are prolific producers of IL-4, differences persists between the two species regarding the stimuli that induce this cytokine. In vitro production of IL-4 by human basophils is tightly coupled to IgE-mediated activation and such responses are greatly augmented by IL-3 (1). In contrast, studies done in mice show that epithelial cell (EC)-derived cytokines, particularly thymic stromal lymphopoeitin (TSLP), are essential components in eliciting IL-4 from basophils (10). Indeed, several studies support the existence of a so-called “basophil-TSLP axis” whereby the latter plays an important role in conditioning basophils for IL-4 production (10, 11). While evidence for this axis in humans has been less forthcoming, one group has recently reported on TSLP’s capacity to activate basophils from asthmatic subjects –a response dependent on IL-3 (12). These authors additionally report that other EC-derived cytokines (IL-25 & IL-33) activate basophils from asthmatics, particularly post allergen challenge (13).

Therefore, we probed herein for additional evidence of whether EC-derived cytokines activate human basophils. Among those investigated (e.g. TSLP, IL-33, and IL-25), we found that only IL-33 mediates activity on human basophils, thus confirming earlier reports (14, 15). We also explored the hypothesis that basophils might be activated by EC via other mechanisms, either through direct interaction and/or by unique cytokines/factors. As a result, our investigations revealed an unexpected finding –that basophils co-cultured with the lung-derived EC line, A549, are activated to release histamine but in a delayed manner (2–5h) unlike classic mediator release via IgE-dependent activation. Basophils also generated large quantities of IL-4 and IL-13 in these co-cultures. These responses were dependent on cell-to-cell interaction and aggregation, a requirement for basophil-bound IgE, and of signaling through FcεRI. Additional experiments revealed that n-acetyllactosamine (LacNAc) suppressed basophil activation in the co-cultures, thus indicating the involvement of a yet unidentified IgE-binding lectin associated with A549 cells. Overall, these data provide evidence for a unique mechanism whereby basophils are potentially activated by IgE-binding lectins expressed on EC, but with secretion parameters very different from those seen with conventional IgE-dependent activation. Such a phenomenon could have mechanistic relevance to basophil participation in allergic conditions, including those not typically driven by allergen (e.g. urticaria and atopic dermatitis), but also in non-allergic conditions recently implicating basophil involvement (e.g. lupus & cancer).

Materials and Methods

Special Reagents, buffers, and media

The following reagents were purchased: crystallized human serum albumin (Calbiochem-Behring Corp, La Jolla, CA); PIPES, FCS, crystallized BSA, and n-acetyllactosamine (LacNAc) (Sigma-Aldrich, Allentown, PA); gentamicin, IMDM and nonessential amino acids (Life Technologies, Inc, Grand Island, NY); Percoll (Pharmacia Biotec, Inc, Piscataway, NJ); rhIL-3 (Biosource, Inc. Camarillo, CA); rhTSLP, rhGM-CSF, rhIL-5, Anti-galectin (Gal)-9 antibody and ELISAs (R&D Systems, Minneapolis, MN); anti-Gal-3 and IgG₁ isotype control (Santa Cruz, Dallas, TX); Gal-3 ELISAs (e-Bioscience, San Diego, CA). Polyclonal goat anti-human IgE (provided by Dr. Robert Hamilton, JHU). The Btk inhibitor, Ibrutinib (PCI-32765), was purchased from APExBio Technologies, Houston, TX. The selective syk inhibitor, 161y,(16) was provided by Dr. Donald W. MacGlashan, Jr., JHU. The specificity of these inhibitors for syk and Btk activity induced through FcεRI signaling is documented (17, 18).

ECs & Culture Conditions

The A549 and BEAS-2B epithelial cell lines were from the American Type Culture Collection, ATCC, Manassas, VA. A549 (bronchial origin) were maintained in medium consisting of F-12K nutrient mixture –Kaigh’s Modification, with 10% heat-inactivated FBS and 1% penicillin streptomycin. BEAS-2Bs (lung origin) were maintained in DMEM/Ham’s F-12 medium, with 5% heat-inactivated FBS and 1% penicillin streptomycin. Cultures were split twice weekly during the duration of these experiments, with no detectable changes in their capacity to activate basophils. To facilitate the co-culture experiments described below, 5×10³ ECs were plated into wells of 96-well plates in 0.100mL of the medium specific for each. These were incubated up to 48h in a humidified incubator (37°C, 5% CO₂) to allow cells to adhere and achieve 50–80% confluency.

Basophil Purification

Venipuncture was performed on consenting adults (age range, 21–65 years) using a protocol approved by the Johns Hopkins University Institutional Review Board. Subjects were selected regardless of allergic status. An additional source of basophils included residual TRIMA cassettes from subjects undergoing platelet pheresis within the Hemaphersis Unit at Johns Hopkins University. Buffy-coats from both specimen sources were subjected to double-Percoll density centrifugation, which produced both basophil-depleted cell (BDC) and basophil-enriched cell (BEC) suspensions, as described (19, 20). Basophils were purified from the BEC suspensions using negative selection reagents (StemCell Technologies, Vancouver, Canada) and immunomagnetic LS columns (Miltenyi Biotec, Gaithersburg, MD). Cell suspensions were routinely ≥99% basophils, as assessed by Alcian blue staining (21).

In some experiments, basophil surface IgE was depleted (i.e. “stripped”) using lactic acid buffer, pH 3.9 (22). In brief, basophils were washed with 0.9% NaCl before resuspending in cold lactic acid (0.5 ml) for 1 min. on ice. Cell suspensions were immediately flooded with cold PAG (~10 volumes), equally portioned between two tubes and centrifuged. After removing supernatant, both pellets were resuspended in PAG containing heparin (10 μg/ml) and EDTA (4 nM), both added from a 50× stock. Those prepared for passive sensitization additionally received polyclonal JK IgE (500 ng/ml, provided by Dr. Robert Hamilton, JHU) with or without omalizumab (10μg/ml). A third tube containing an equal number of basophils not treated with lactic acid was prepared but did not receive JK IgE. All tubes were incubated at 37°C for 30 minutes. Cells were washed twice with PAG before counting and adjusting cell densities in C-IMDM.

Culture-derived basophils (CDBA) and mast cells (MC)

BDC suspensions produced during density centrifugation (see above) were used as a source of PBMC for isolating CD34⁺ cells by positive selection (Miltenyi Biotec). When starting with 2–4×10⁹ BDC obtained from the TRIMA cassettes, 2–5 ×10⁶ enriched CD34⁺ cells were typically prepared. These CD34⁺ cells were differentiated into basophil-like cells by culturing 2 weeks in IL-3 (10 ng/ml) alone or into MC after 8–10 weeks in SCF/IL-6 (IL-3 during 1^st week), as described elsewhere (23, 24).

Flow Cytometry

In some instances, EC were investigated for Gal-3 expression using flow cytometry. In brief, adherent cells were removed from culture flasks using EDTA without trypsin and then washed with PBS. Cells were prepared for surface and intracellular staining using 4% paraformaldehyde or by the Fix & Perm^™ kit (Invitrogen)(25). Blocking was done using 1% human IgG. Anti-Gal-3 antibody (clone B2C10, Santa Cruz Biotechnology) and IgG₁ isotype control (clone MOPC-21) were each used at 10μg/ml in 20 minute incubations at RT. Cells were then washed before staining 30 min. on ice with anti-IgG₁-Alex647 (2.5μg/ml). Flow cytometry was performed using a FACSCalibur machine.

Real-time RT-PCR

Total RNA was isolated using the RNA-Bee protocol (Tel-test, Inc, Friendswood, Tex). After isopropanol precipitation, RNA was washed with ethanol before resuspending in DNase-free water. It was then subjected to “RNA cleanup” (Qiagen Corp., Germantown, MD) and quantified using Nanodrop. Quantitative RT-PCR was performed, as described (26), using validated primer/probes combinations for Gal-1, -3, -9, (Applied Biosystems, Foster City, CA).

Co-Culture Conditions

Cultures were conducted in IMDM supplemented with 5% FCS, non-essential amino acids, L-glutamine, 10 μg/ml gentamicin, pH 7.2–7.4 (C-IMDM). Depending on the set of experiments, basophils or MC (1.0×10⁵) were added in 0.100 ml volumes to wells of 96-well plates [plated with EC or without –both at 0.100 ml volumes in the F-12K nutrient mixture. Immediately after adding basophils or MC, 0.050 ml of 5× stimulus (e.g. medium, IL-3/SCF, anti-IgE or the combination) was added and the cultures incubated as indicated at 37°C, 5% CO₂. In some experiments, LacNAc or inhibitors were added to EC before the addition of basophils and stimuli. Several experiments were done using Corning polycarbonate 0.4μm trans-well plates (Sigma-Aldrich, St. Louis, MO). Combinations were set-up whereby bottom or insert wells were seeded with EC. Basophils were then added the next day to the opposite well to prevent direct contact with EC, or added directly with EC. Stimuli were added to bottom wells, the inserts, or both before culturing at 37°C, 5% CO₂. Supernatants were harvested after 20h unless otherwise indicated and tested for cytokine secretion and basophil histamine release (BHR).

Histamine and cytokine measurements

Supernatants were analyzed for IL-4/IL-13 protein by ELISA (e-Bioscience, San Diego, CA), and histamine by automated flourimetry (19, 27).

Statistical Analysis

Statistical analyses were performed with Prism 7.0 software (GraphPad, Software, LaJolla, Calif.) Analyses were performed using Wilcoxon non-parametric and paired t-test analyses unless specified otherwise. Differences were considered statistically significant at a P value <0.05.

Results

We first investigated several recombinant EC-derived cytokines to test their capacity to activate human basophils and to establish a pattern of responsiveness to compare in the subsequent EC-basophil co-cultures. Included in our analyses were TSLP, IL-25 and IL-33. Both TSLP and IL-25 showed no capacity to induce IL-4/IL-13 secretion when incubated 20h, and IL-33 alone was only marginal in inducing detectable levels of these cytokines (Fig. S1, supplemental data). IL-3 induces IL-13 (>IL-4), from human basophils (28, 29), and this was once again demonstrated herein (Fig. S1). However, adding TSLP or IL-25, along with IL-3, did not further augment IL-13 or IL-4 secretion. In contrast, IL-33 markedly increased IL-13 secretion in response to IL-3, and did so over several doses, consistent with findings published elsewhere (14, 15). A similar pattern was seen for IL-4 secretion, although levels of this cytokine, as predicted, were ~8- to 10-fold less than those for IL-13. None of the cytokines directly induced histamine release, and only IL-33 showed a marginal capacity to enhance this mediator when combined with IgE-dependent activation (data not shown).

The inability of TSLP to stimulate human basophils was not from a lack of biological activity, since it proved capable of inducing OX40L on myeloid dendritic cells (data not shown). In addition, TSLPR was inducible on basophils following IgE-dependent activation, as reported elsewhere (30), with similar levels induced following incubation with IL-3 (data not shown). Thus, despite induction of TSLPR by these stimuli, adding TSLP did not yield a detectable response.

A549 EC-Dependent Activation of Human Basophils

In vitro co-culture approaches have demonstrated that EC can directly activate mast cells, in part, via TSLP (31, 32). In addition, the activity of some cytokines, such as IL-15, is mediated through a phenomenon known as transpresentation, whereby they are presented in trans to other cells expressing the required receptors (33). Since TSLP is classified in the same family of cytokines as IL-15, we considered the possibility that it too may function through transpresentation, in this instance, by EC to basophils. We therefore conducted experiments whereby basophils were co-cultured with lung-derived EC lines to allow direct cell-to-cell contact. By focusing on IL-4 secretion, we made several surprising observations with A549 EC, as shown in Fig. 1. First, the amount of IL-4 secreted after 20h incubation was significantly greater in the co-cultures compared to those with basophils alone. While this was true regardless of the mode of stimulation, IL-4 was unusually high in co-cultures receiving IL-3 where levels exceeded those without EC nearly 10-fold (332±74 vs. 37±7 pg/10⁶ basophils, P=0.0005, n=12). For comparison, relatively low levels of IL-4 were detected when basophils were cultured in medium alone, but these increased ~7-fold if co-cultured with EC (4±1 vs. 27±10 pg/10⁶ basophils, P=0.0078). Even the high levels of IL-4 produced in response to anti-IgE stimulation were consistently increased ~35% when basophils were co-cultured with EC (225±42 vs. 304±51, P<0.001). In contrast, co-culturing basophils with BEAS-2B EC did not show any effects on cytokine secretion, compared to basophils stimulated alone (Fig. S2, supplemental data). Experimental controls indicated that A549 EC did not contribute to the production of IL-4, even when treated with supernatants from activated basophils (data not shown).

Figure 1. IL-4 production in basophil/EC Co-cultures.

A, Basophils were cultured alone and B, with A549 EC in the presence of medium, IL-3 (10 ng/mL) or activating anti-IgE (10 ng/mL). After 20h incubation, supernatants were harvested and assayed for IL-4 protein by ELISA. Box-Whisker plots shown (n=12). * P<0.008, ** P≤0.001 vs. corresponding conditions with basophils alone.

IL-3 Priming of Basophils Augments their Activation by EC

IL-3’s capacity to augment IL-4 (and IL-13) in the basophil/A549 co-cultures was further investigated by titering its effects and comparing it to other growth factors. As shown in Fig. 2A, IL-4 levels secreted by basophils after 20h incubation in IL-3 (1–100ng/ml) averaged less than ~30 pg/10⁶ cells. However, these increased 10-fold when conducting the same experiment in the presence of A549 EC. Even IL-3 concentrations (0.1 ng/ml) not inducing detectable IL-4 from basophils when used alone, synergized with EC to result in levels approaching 100 pg/10⁶ basophils. Marked levels of IL-13 were detected in the co-cultures, with levels exceeding ~10-fold those produced by basophils stimulated with IL-3 alone (Fig. 2B). Other cytokines, including GM-CSF, IL-5, and TSLP did not synergize with A549 EC to activate basophils (data not shown).

Figure 2. IL-3 parameters affecting IL-4/IL-13 production by basophil/A549 vs. basophil alone.

A, B, effect of IL-3 dosing on IL-4 and IL-13, respectively. C, D, effect of IL-3 (10 ng/ml) priming overnight followed by washing on IL-4 and IL-13 secretion, respectively, after re-challenging with IL-3, IL-33, anti-IgE (each at 10 ng/ml.) or with medium alone. Supernatants were harvested (20h) and assayed for cytokine by ELISA. Mean±SEM, n=3–5. *P<0.05, **P<0.009, ***P<0.0008.

With the possibility that IL-3 additionally acted on A549 ECs, experiments were done whereby basophils were first primed with IL-3 (10 ng/ml) overnight, washed extensively, and then co-cultured 20h with EC. As shown in Fig. 2C, no detectable IL-4 was produced by the primed basophils cultured without EC. In contrast, IL-3-primed basophils co-cultured with EC secreted IL-4 in excess of 100 pg/10⁶ basophils –a response not significantly increased by adding IL-3 (or IL-33). As expected, primed basophils readily secreted IL-4 when stimulated by anti-IgE, with a trend for greater production if the cells were co-cultured with EC. A similar pattern occurred for IL-13 production, with overall levels 2–3 times greater than those for IL-4 (Fig. 2D). IL-13 secreted by IL-3-primed basophils was not augmented by IL-33, whether co-cultured alone or with EC, despite the fact that IL-3 and IL-33 synergized to induce IL-13 from basophils when both were added together (Fig. S1).

Delayed Time Course for BHR and Cytokine Secretion in Basophil/A549 EC co-cultures

Although human basophils secrete IL-4 in response to IL-3, the levels produced are generally far less than those seen with IgE-dependent activation (28). This latter mode of stimulation is also typically preceded by histamine release, which occurs within 30 minutes upon activation, whereas IL-4 is secreted over several hours. Thus, the relatively high levels of IL-4 produced in the co-cultures prompted testing to show its time course and whether BHR becomes evident with IL-3 –an IgE-independent stimulus. The time course for maximal BHR (~30 min.) and IL-4 secretion (~4h) was as expected in control cultures with basophils alone stimulated with an activating anti-IgE antibody (Fig. 3A and B). Likewise, little to no BHR and IL-4 (above the medium control) was released by basophils alone in response to IL-3. In sharp contrast, both mediators were secreted in basophil/A549 co-cultures, especially in response to IL-3 –the major difference being their time course, with nearly twice the time for IL-3-induced release relative to the kinetics seen with IgE-dependent activation. For comparison, basophils, which typically secrete IL-13 whether stimulated with anti-IgE or IL-3 (as shown in Fig. 3C), generated on average ~25-fold greater levels of this cytokine when co-cultured with EC in the presence of IL-3 (Fig. 3F). As a result of these findings, all subsequent experiments continued to look at cytokine secretion after 20h, since this time allowed for peak or near peak secretion. Those experiments investigating histamine release focused on times of up to 8h, since peak responses for this preformed mediator were unlikely to extend beyond this time.

Figure 3. Delayed Time Course for BHR and Cytokine Secretion in Basophil/A549 co-cultures.

Basophils cultured alone (upper panels) and with A549 EC (lower panels) were stimulated with medium, IL-3 (10 ng/mL), or anti-IgE (10 ng/mL). Supernatants were harvested at the indicated times and assayed for: A, D, histamine, B, E, IL-4 and C, F, IL-13. Results are the Mean±SEM, n=3.

Cell-to-cell contact is required for A549 EC to activate basophils

Additional clues regarding the requirements for A549-dependent basophil activation were provided in experiments that produced negative results. First, supernatants from A549 EC did not support increased IL-4/IL-13 secretion from basophils, whether in the absence or presence of IL-3 (data not shown). Second, attempts to detect TSLP and IL-33 in A549 supernatants were also negative, signifying that neither cytokine was secreted by A549 EC. Third, attempts to neutralize activity with antibodies targeting TSLPR and IL-33R (ST2) did not suppress cytokine secretion in A549/basophil co-cultures (data not shown). Collectively, these results argue against the likelihood that A549-derived cytokines (e.g. TSLP, IL-33) accounted for the marked induction of IL-4/IL-13 by basophils co-cultured with the EC line. Of note, evidence of cell clumping in the IL-3 treated co-cultures was seen (Fig. 4A), prompting the hypothesis that cell-to-cell contact is necessary for A549 EC to activate basophils –a belief substantiated by conducting transwell experiments. As expected, co-culturing basophils and A549 EC with IL-3 for 20h resulted in high levels of IL-4 (Fig. 4B) and IL-13 (Fig. 4C). Under the same conditions, but separating the two cell types by transwells, the levels of both cytokines were no different than culturing basophils alone.

Figure 4. Cell-to-cell contact required for A549 EC-dependent activation of basophils.

A, Cell-clumping evident in basophil/A549 co-cultures receiving IL-3 (middle panel). Original magnification = 40x. B, C, IL-4 and IL-13, respectively, detected in culture supernatants (20h) whereby basophils (BA) and A549 EC were co-cultured together or separated in transwells. Cytokine levels compared to those from basophils cultured alone. (Mean±SEM, n=3). **P<0.01.

A549 EC-dependent activation of basophils is IgE-dependent and requires FcεRI signaling

The combined observations of BHR and IL-4 secretion along with the need for cell-to-cell contact, suggested that basophil-bound IgE is somehow involved. To address this hypothesis, we tested the effect of “stripping” IgE from basophils using pH 3.9 lactic acid and restoring its expression via passive sensitization. Both IL-4 (Fig. 5C) and IL-13 (Fig. 5D) were significantly reduced in co-cultures that included “stripped” basophils, compared to A549 co-cultured with untreated basophils. Most remarkably, this was evident with IL-3 stimulation. Yet, the production of these cytokines was restored under the same conditions if stripped basophils were re-sensitized with IgE. And, the addition of omalizumab to neutralize IgE during passive sensitization prevented this from happening (Fig. 5E). For comparison, the results for these same parameters were as expected when tested in control cultures with basophils alone (Fig. 5A,B).

Figure 5. A549 EC activate basophils via IgE/FcεRI.

A,B, IL-4/IL-13, respectively, by basophils depleted of IgE (stripped), stripped and sensitized with IgE (stripped+IgE), or left untreated (non-stripped) –without A549 EC. C, D, IL-4/IL-13 using basophils identically treated as in A&B but co-cultured with EC. E, IL-4 from co-cultures that includes stripped basophils sensitized with IgE in presence of omalizumab. F, IL-4 secretion in basophil+A549 co-cultures using inhibitors of syk and BtK tyrosine kinase. All cytokines assayed after 20h incubation. (Mean±SEM, n=3) *P<0.05.

Given the requirement for basophil IgE expression, we next addressed whether FcεRI signaling is also involved by testing the effects of pharmacologically inhibiting syk and Btk tyrosine kinase activity. As shown in Fig. 5F, selective inhibitors of syk (161Y) and Btk (ibrutinib) used at IC_50s (50nM and 5nM, respectively) targeting standard IgE-dependent activation, likewise inhibited those induced by IL-3 in the co-cultures. This latter response was inhibited nearly 90% by the Btk inhibitor, implying that this mode of activation is more sensitive to Btk inhibition than is standard IgE-dependent activation.

Basophils and MC derived from CD34⁺ precursors cultured in serum-free medium express FcεRI but are devoid of IgE (23). These cells therefore enabled further confirmation regarding the role of IgE in A549-dependent activation of basophils and to additionally test whether these EC activate MC±IgE. As shown in Fig. 6, CDBA passively sensitized with IgE responded by releasing histamine, IL-4 and IL-13 when co-cultured with A549 cells, especially in the presence of IL-3, and at peak times consistent with those by blood basophils. In contrast, the same CDBA, if not sensitized with IgE, failed to release these mediators when co-cultured with A549. Interestingly, MC derived from progenitors cultured for 8–10 weeks and passively sensitized with IgE released slightly more histamine to anti-IgE when co-cultured with A549. However, this increased responsiveness was absent in co-cultures stimulated with IL-3 or SCF (Fig. 7). Unlike basophils, MC (alone or co-cultured for up to 20h with A549) did not release detectable IL-4 and no to little amounts of IL-13 (data not shown).

Figure 6. CDBA are activated by A549 EC only after sensitization with IgE.

CDBA were grown from CD34⁺ precursors cultured 14d in serum-free medium containing IL-3 (10 ng/ml). Cells were then sensitized ~16h with IgE (500 ng/ml) or left unsensitized, washed, and cultured with or without A549 EC. Histamine (panels A,B,C), IL-4 (panels D,E,F), and IL-13 (panels G,H,I) were then measured in culture supernatants at the times indicated. (Mean±SEM, n=2–3). *, P<0.02; **, P=0.05.

Figure 7. Mast cells sensitized with IgE are not activated by A549.

Mast cells were grown from CD34⁺ precursors cultured 8 weeks in serum-free medium containing IL-6 and SCF, as described in Materials & Methods. Cells were then sensitized ~16h with IgE (500 ng/ml), washed, and cultured with or without A549 EC, as indicated. Histamine was measured at the times indicated. (Mean±SEM, n=3).

Potential Role for an IgE-binding Lectin Expressed by A549 EC

Since the capacity of A549 cells to promote mediator release and cytokine secretion was dependent on basophils expressing IgE, we considered the hypothesis that a protein expressed by these EC interacts with this immunoglobulin to promote these basophil responses. After finding no evidence for FcεRIα or CD23 using flow cytometry (data not shown), we considered involvement of a lectin possibly binding and crosslinking IgE, thus triggering basophil activation. We therefore conducted a final set of experiments testing two sugars, lactose and LacNAc, for their capacity to act as competitive antagonists and inhibit cytokine secretion in basophil/A549 co-cultures exposed to IL-3 for 20h. In testing lactose, we found no significant effect of this sugar on basophil cytokine responses, even at concentrations of up to 25 mM (data not shown). At concentrations exceeding 25 mM, lactose began showing non-specific inhibitory activity, as it began suppressing responses in control cultures of basophils alone (data not shown). In contrast, LacNAc showed a dose response suppression with ~20% inhibition of the IL-4 response mediated by this sugar at 1 mg/ml (~2.6 mM), which culminated in ~60% inhibition when used at 10 mg/ml (~26 mM) (Fig. 8). The same concentrations of LacNAc showed no suppressive activity on the IL-4 produced by basophils alone stimulated with IL-3, and actually mediated a 10–25% increase in this response. The IL-4 secreted in response to anti-IgE was also suppressed by the sugar in basophil/EC co-cultures, but this became indistinctive from the control IL-4 at the higher concentration.

Figure 8. Putative role for an IgE-binding lectin.

Effect of n-acetyllactosamine (LacNAc) on IL-4 secretion in basophil (BA)/A549 EC co-cultures vs. basophils alone after stimulating as indicated for 20h. Shown are fractional responses (Mean±SEM, n=4–5) of baseline IL-4 levels (dotted line) by basophils cultured alone with IL-3 (21±6 pg/10⁶ basophils) or anti-IgE (115±17 pg/10⁶ basophils) vs. levels in co-cultures with IL-3 (338±82 pg/10⁶ basophils) or anti-IgE (190±36 pg/10⁶ basophils). *P<0.02.

While results using LacNAc strongly indicated the involvement of a lectin, particularly a galectin, preliminary experiments proved inconclusive in linking A549-dependent activation of basophils to this family of lectins. First, there are some fifteen family members of galectins, but only Gal-3 and Gal-9 reportedly bind IgE (34–36). We therefore assessed the probability of a Gal-3/Gal-9 involvement by investigating A549 EC for expression of these galectins. In doing so, we confirmed evidence that A549 EC expresses Gal-3 (Fig. 9A) (37). However, BEAS-2B cells, which lacked the capacity to activate basophils in co-culture (Fig. S2, online repository), also showed Gal-3 expression at levels equal to or greater than those detected on A549. Both cell lines expressed Gal-1 equally (data not shown). BEAS-2B did differed slightly from A549 cells in expressing nearly 50-fold greater levels of Gal-9 mRNA (Fig. 9B). Confirmation of this was also evident at the protein level in that BEAS-2B lysates showed detectable amounts of Gal-9 protein whereas A549 lysates did not (Fig. 9C). Overall, there was no clear pattern of Gal-3/Gal-9 found in/on A549 EC that readily linked the expression of these lectins to the activation of basophils, although further investigation is required.

Figure 9. Gal-3, -9 expression by A549 and BEAS-2B cell lines.

A, Flow cytometry demonstrating Gal-3 expression by A549 and BEAS-2B cell lines. B, Relative expression of Gal-3 and Gal-9 mRNA and C, protein for these galectins in the two epithelial cell lines.

Attempts were also made to test whether antibodies specific for these galectins inhibit basophil activation mediated by A549 EC, even though none of the commercially available reagents (most containing azide) report neutralizing activity. At best, just ~20% inhibition was seen using the B2C10 antibody specific for Gal-3 (as compared to isotype control) and no effect was seen using anti-Gal-9 (data not shown). Both reagents do not appear to target the carbohydrate recognition domain (CRD) of either lectin, which is presumably needed to neutralize activity.

Discussion

Our rationale for conducting this study initially sought translational evidence that human basophils respond to EC-derived cytokines (e.g. TSLP, IL-33, IL-25). When unable to detect any discernable in vitro effect on basophil function using two of these cytokines (TSLP & IL-25), we proceeded to co-culture basophils with EC lines. This approach stems from the concept that some cytokines (e.g. TSLP) may exert its function, in part, though a phenomenon known as transpresentation (33), whereby a cytokine is physically presented to effector cells (i.e. basophils) by EC in a manner requiring cell-to cell contact. While our findings have not supported a role for transpresentation, they did expose a potentially unique mode of EC-dependent basophil activation –one that may ultimately have significance in multiple diseases in addition to allergy. Indeed, the results herein support a concept whereby a yet unidentified IgE-binding lectin expressed by A549 EC, is capable of crosslinking IgE/FcεRI complexes on basophils to induce mediator release and cytokine secretion. Certainly, our findings do not exclude a role for TSLP (or any other EC cytokine) in activating human basophils, but every attempt thus far to test this hypothesis have proved negative. In this respect, it is important to note that Salter, et al. recently provided evidence of TSLP activating human basophils for mediator release and cytokine production (12). This group investigated allergic asthmatics whose basophils were somewhat hyper-responsive, spontaneously releasing ~40% of total histamine in buffer alone. These basophils were not only responsive to TSLP but were also shown to secrete histamine and cytokine in response to IL-3. In contrast, the basophils we investigated were predominately from normal subjects, which may account for why they did not respond to TSLP (or IL-25). They did respond to recombinant IL-33, as previously reported by others (14, 15), but in a manner quite different than what was ultimately observed when co-cultured with A549 EC.

Several experimental observations led to the hypothesis that an IgE-binding protein is a key player in the mechanism by which A549 EC activate basophils. The first came from the relatively high levels of IL-4 generated in the co-cultures, which were comparable (and generally higher) to those we typically measure following IgE-dependent activation. In fact, it’s our experience that IL-4 secretion does not typically exceed ~50 pg/10⁶ basophils unless the cells are activated via an IgE-dependent stimulus or by calcium ionophore. Certainly, IL-3 alone can induce IL-4 secretion but typically in the range of ~10–50 pg/10⁶ basophil, and only after prolong (~20h) incubation (28). Yet, the IL-4 levels generated herein among the basophil-A549 EC co-cultures stimulated with IL-3, often exceeded 500 pg/10⁶ basophils after 20h. Upon seeing these robust responses, we immediately predicted two additional parameters: 1) that basophils are likely secreting histamine under the same conditions, and 2) that the mode of activation somehow involves IgE/FcεRI crosslinking, despite no exogenous activator of this pathway (i.e. allergen or anti-IgE antibody). Again, the capacity of human basophils to secrete IL-4 has long been linked to IgE-dependent activation (38–40). In addition, this response is typically preceded by rapid degranulation (i.e. BHR) occurring within the first 30 min. In fact, the two mediators correlate quite well, despite having different kinetics and dose response curves. As predicted, histamine was released in basophil/A549 EC co-cultures stimulated with IL-3. Remarkably, the half-maximal time of 4h for this response far exceeded the 30 min. necessary for maximal BHR following standard IgE-dependent activation. IL-4 production in these co-cultures was also slower (half-maximal at 5h), requiring nearly twice the time to be secreted compared to standard IgE-dependent activation. The exact cause(s) for these delayed time courses (from standard IgE-dependent activation) remains poorly understood but may point to the time necessary for optimal aggregation mediated by the proposed lectin. Nonetheless, the importance of IgE was clearly demonstrated two-fold. First, by depleting this basophil-bound immunoglobulin (with pH 3.9 lactic acid buffer), which significantly diminished IL-4 production, and then restoring this response by re-sensitizing basophils with IgE. Second, by using CDBA devoid of IgE, which were completely unresponsive when co-cultured with A549. Only after sensitization with IgE did these cells become responsive to co-culture with A549 EC. Then, the observation that inhibitors of syk and Btk kinases also blocked cytokine secretion in the co-cultures, further linked the response to FcεRI-dependent activation. Most significantly, these findings were evident only in the co-cultures and not with basophils alone. Collectively, they add to the notion that A549 ECs provide some sort of protein/factor that interacts with IgE on the surface of basophils, effectively crosslinking FcεRI, which results in activation.

The mechanism by which IL-3 (but not IL-5 or GM-CSF) facilitated the overall response also remains unclear, yet further points to multiple factors contributing to the capacity of A549 EC to activate basophils. For instance, basophils co-cultured with A549 EC in medium alone or with anti-IgE, showed greater responsiveness than if cultured alone with these stimuli. However, the greatest magnitude in response was with those in which co-cultures received IL-3. Even basophils primed overnight with IL-3 and then washed free of this cytokine, showed significantly greater responses, while also indicating that the effect of IL-3 was on basophils rather than the A549 EC. Of course, IL-3 has long been known to amplify just about every biological response attributed to basophils, so the underlying mechanism(s) herein are likely linked to the signaling mediated by this cytokine. It remains equally plausible that IL-3 augments the expression of a third component –one that facilitates interaction between the putative lectin (on EC) with IgE (on basophils), which then amplifies EC-dependent basophil activation. Obviously, this hypothesis requires further investigation and can only be tested upon identification of the lectin and/or co-factor augmenting basophil activation.

The observation that culture-derived mast cells were unresponsiveness when co-cultured with A549 further supports the importance of IL-3. Like basophils, mast cells express FcεRI and bind IgE with high affinity, which should enable them responsive to A549 if IgE was the only requirement. However, human mast cells don’t generally express IL-3 receptors and are unresponsive to this cytokine. We therefore considered that SCF might do for mast cells what IL-3 does for basophils, but the addition of this factor (or IL-3) also failed to make mast cells responsive to A549. CDMC are generally thought to possess a MC_T phenotype. It therefore remains possible that skin mast cells (bearing the MC_TC phenotype) may respond to A549 cell, but this remains to be tested. Overall, the capacity of A549 EC to activate IgE-bearing cells in general might be unique to basophils, since they express IL-3 receptors (CD123) and bind IgE.

In observing that LacNAc significantly suppressed IL-4 production in the basophil-A549 co-cultures but not in cultures with basophils alone, our attention quickly turned towards possible involvement of an IgE-binding lectin expressed on A549 cells. In particular, the galectin family consists of 15 fifteen members, but only Gal-3 and Gal-9 reportedly bind IgE. (34, 35). Moreover, Gal-3 is widely expressed by EC and fibroblasts especially those having a cancer origin (41, 42). In fact, its been reported in mice that EC actually bind IgE via Gal-3 (43). Gal-3 is also the only chimera galectin, meaning it can exist as a monomer or form multivalent structures (e.g. pentamers) that facilitate “lattice” formation between glycoproteins on different cell types (44). This characteristic alone provided good rationale as to how Gal-3 might facilitate IgE/FcεRI crosslinking to activate basophils. Gal-9 is more recently reported to bind IgE, and apparently does so with greater affinity than does Gal-3. Interestingly, Gal-9 reportedly inhibits IgE-dependent activation, both in vitro and in vivo (36). Exactly how this galectin mediates this effect is currently unclear, but seemingly does so by binding free IgE and preventing sensitization. Whether and how it might interact with cell-bound IgE has not been fully addressed, let alone whether it mediates FcεRI-crosslinking. Nonetheless, LacNAc can target all galectins and thus lacks specificity for any given family member, although it is recently shown to inhibit Gal-3 binding to IgE (45).

We therefore proceeded to investigate A549 cells for galectin expression, confirming what is well cited in the literature -that A549 cells express Gal-3. However, BEAS-2B cells, which did not activate basophils for increased IL-4/IL-13 secretion, also expressed Gal-3 at nearly equal levels, thus confounding the hypothesis that this IgE-binding lectin mediates basophil activation. As it stands, these preliminary data challenge the hypothesis that Gal-3 is the sole component mediating A540-dependent basophil activation. Obviously, further investigation is required to fully address the role of Gal-3, Gal-9, and possibly other galectins –work that extends beyond the scope of the present study. Antibodies specific to these galectins did not reveal neutralizing activity, but are unlikely to do so being that none target the carbohydrate recognition domain (CRD) of the lectin. Instead, we are currently exploring the use of CRISPR, with the intent of systematically knocking-out galectins (e.g. Gal-3 & Gal-9) in A549 EC to directly address their importance. Of course, even this approach may prove negative, if a yet unidentified lectin is mediating A549-dependent activation of basophils. Our hypothesis regarding the role of galectins, particularly Gal-3 and Gal-9, rests solely on the fact that these bind IgE, are expressed by A549 EC, and are targeted by LacNAc. More direct approaches are required to fully address the role of these lectins.

The proposed mechanism herein of how an A549 EC-associated lectin activates basophils raises several important questions requiring further investigation. First, it remains unknown whether other EC lines and/or normal/disease primary EC possess a similar capacity to activate basophils. And, with the putative IgE-binding lectin currently unidentified, such investigations will require the same rigorous testing done herein to determine the dependency on IgE. Thus, it seems possible that some EC will activate basophils for mediator release and/or cytokine production, but the potential role of EC-derived cytokines (IL-33, TSLP, and IL-25) will have to be ruled-out before concluding a mechanism identical to that seen using A549 EC. And, with BEAS2B cells not possessing the same capacity to activate basophils, then another important parameter requiring investigation pertains to how expression of this putative lectin is regulated on EC. If not expressed on all EC, then what are the conditions regulating its expression? We are also hesitant to conclude that basophils are the only IgE-bearing cell activated in this manner, despite the augmentative effects seen using IL-3 –a cytokine known to prime basophils. The conditions tested herein for MC may not have been fully optimal. Likewise, IgE-bearing DC could also be activated in a similar manner, even though the mediators and cytokines produced by these cells will differ from basophils (and MC). With MC already in organ tissue, they could be the primary targets responding to this mode of activation. In contrast, basophils and DC would respond only after infiltrating lesion sites expressing the lectin.

Finally, there is also the question of whether this mode of basophil activation is even relevant to human disease. Again, this issue may only be fully addressed upon identification of the supposed lectin. Nonetheless, there is emerging evidence that lectins, in particular IgE-binding galectins, are linked to a variety of conditions, including allergic disease. For example, Mauri, et al. recently identified Gal-3 as a predictive biomarker of airway remodeling in allergic asthmatics –a finding that was unveiled in subjects responsive to anti-IgE therapy using omalizumab (46). Knowing that omalizumab administration rapidly decreases IgE/FcεRI expression on circulating basophils, DC, and eventually on MC, these cells could become unresponsive to tissue expressing Gal-3. Chronic idiopathic urticaria (CIU) and atopic dermatitis (AD) are other clinical conditions whereby lectin/IgE interactions could theoretically contribute to their pathogenesis. Again, at least one report has shown that using a mouse model of AD, mice deficient for Gal-3 do not develop active disease (47). IL-4 levels in these mice were greatly reduced whereas IFN-γ remained high. While there are currently no reports of whether galectins are expressed in CIU tissue, it’s reasonable to hypothesize the possibility and these having a role in activating basophils, MC and DC. Histamine, IL-4, and IL-13 are all inflammatory mediators identified in CIU (48) and basophils are known to infiltrate CIU lesions (49). As with allergic asthma, given the recent efficacy using omalizumab (50), it’s apparent that an IgE component contributes to the pathogenesis of CIU, despite these patients not typically being sensitized to any known allergen. It’s therefore intriguing to think that basophils infiltrating this skin disease become activated via a lectin/IgE pathway. Anti-IgE therapy potentially reduces the chance of this occurring by neutralizing IgE and depleting its expression on basophils. Finally, galectin-dependent activation of basophils is conceivably linked even to diseases for which basophils have only recently been implicated in, such as lupus (51) and cancer (52). This is because other reports show that galectins are widely expressed in the same tissues (53, 54). Thus, a synthesis of these findings, along with our proposed mechanism, might further explain why IL-4-producing basophils are seen in both diseases –they are triggered through an IgE/galectin interaction. Obviously, this hypothesis and those above, remain speculation, yet the findings herein provide feasibility and thus warrant further investigation.

Abbreviations

LTC4

Leukotriene C4

ILC2

innate lymphoid cell type 2

TSLP

thymic stromal lymphopointin

Gal

Galectin

C-IMDM

conditioned Iscove’s Modified Dulbecco’s Medium

PAG

PIPES, albumin, glucose

BDC

basophil-depleted cell

BEC

basophil-enriched cell

BHR

basophil histamine release

EC

epithelial cell

LacNAc

n- acetyllactosamine

CDBA

culture-derived basophils

MC

mast cells