Indirect IL-4 Pathway in Type 1 Immunity¹

Cells of the innate immune system and lymphocytes both primarily rely on cytokines for mediating their effector and regulatory functions. However, cytokine responses by the innate immune system are fundamentally different from those of lymphocytes. The former are direct, transitory reactions to environmental stimuli that, unlike with lymphocytes, cannot be imprinted in long-term memory and are not clonally expandable. The pathways that lead to the induction of IL-12 and of IFN-γ in cells of the innate immune system have been well established, involving microbial pattern recognition by Toll-like receptors (1, 2). While NK cells, basophils, and mast cells can produce IL-4 after stimulation, the signals that induce IL-4 production in these cells of the innate immune system are less well defined (3, 4).

Naive T cells do not produce cytokines such as IL-4 and IFN-γ, but learn to express them via an instructed differentiation process (5, 6). The cytokine microenvironment during the primary Ag encounter of the naive T cell defines which cytokine the memory cells generated will express (7). The generation of IL-4-producing memory T cells (Th2 cells) requires production of IL-4, either by the cells of the innate immune system or by T cells during the primary Ag encounter (8). In contrast to cells of the innate immune system, cytokine production by memory T cells shows long-term commitment (9, 10). Memory T cells are programmed to secrete a certain cytokine whenever and wherever they reencounter Ag. This commitment enables the predictable reproduction of the cytokine environment that called the immune response into existence. Moreover, the expandable nature of the Ag-specific memory T cells permits regulation of the magnitude of the cytokine effector functions via clonal sizes.

The effective concentration of the cytokine might be fundamentally different when produced by cells of the innate immune system or by T cells. For example, IL-4 regulates Ig class switching in B cells (11, 12). When an Ag-specific Th2 cell interacts with an Ag-specific B cell, the IL-4 will be released in the immediate vicinity of the B cell, reaching very high effective concentrations for this B cell. In contrast to this cognate event, IL-4 released in the absence of direct cell-cell contact is likely to create orders of magnitude lower effective concentrations. The above occurs, first, because the concentration of the cytokine drops with the square of the distance between the secreting cell and the cell to be affected and, second, because IL-4 has a short t_1/2 in vivo (13, 14, 15). It still needs to be defined how the different IL-4-dependent mechanisms such as Ab switching to IgE and IgG1, as well as T cell differentiation, are affected when IL-4 is released by memory T cells in cognate cell-cell interactions, vs the indirect pathway production by cells of the innate immune system.

Because Th1 and Th2 cells exert fundamentally different, frequently antagonistic roles in T cell-mediated immunity, it has been a major scope of immunologic research to clearly define Th1/Th2 cell involvement in most aspects of immunobiology and immunopathology. Measurements of recall Ag-induced IFN-γ and of IL-4 have been the gold standard for assessing Th1 and Th2 immunity, respectively. The low frequency of Ag-specific T cells has made such measurements challenging in freshly isolated cell material (16), in particular using techniques that permit to directly verify the phenotype of the IL-4-producing cell. With few exceptions, in which the specific T cells reach high enough frequencies for direct flow cytometric detection (17, 18), judgment on the induction of Th1/Th2 immunity has relied primarily on measurements of recall Ag-induced IL-4 and of IFN-γ in supernatants or lysates of mixed cell populations with the cellular source of the cytokine not amenable for direct detection. However, only IL-4 and IFN-γ produced by T cells (cognate cytokine) reflect long-term commitment by the immune system, and hence the sought-after information. In contrast, IL-4 or IFN-γ production induced by the recall Ag in cells of the innate immune system reflects indirect, bystander reactions of unknown consequences as to in vivo immune processes.

In this study, we show that while recall Ag-induced production of IFN-γ is a cognate T cell response in freshly isolated lymphoid cell material, IL-4 production can reflect a cytokine-driven bystander reaction of cells of the innate immune system. Therefore, such measurements on recall Ag-induced IL-4 are unreliable indicators of Th2 T cell immunity. We also have shown that the indirect IL-4 pathway mediates in vivo immunoregulatory and effector functions distinct from those resulting from cognate T cell-derived IL-4.

BALB/cByJ, BALB/cByJ Smn-Prkdc^scid/J, C57.BL/6J; SJL/J, C57.BL/6J-Kit^W-v, WBB6F1/J-Kit^W/Kit^W-v, IL-4 ^−/−, and RAG2 ^−/− (both on C57.BL/6 background) mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and bred at Case Western Reserve University (Cleveland, OH) under specific pathogen-free conditions. Ags were mixed with the adjuvants to a final concentration of 1 mg/ml OVA and 1 mM of emulsion (OVA_323–339, proteolipid protein (PLP)³139–151, and myelin oligodendrocyte glycoprotein (MOG)35–55). A total of 100 μl of Ag in CFA or alum was injected either i.p. or s.c. Female mice were injected at 6–10 wk of age. OVA was purchased from Sigma-Aldrich (St. Louis, MO). OVA peptide 323–339, PLP peptide 139–151, and MOG peptide 35–55 were purchased from Princeton Biomolecules (Columbus, OH). IFA was purchased from Life Technologies (Grand Island, NY), and CFA was prepared by mixing inactivated Mycobacterium tuberculosis H37RA (Difco, Detroit, MI) at 1 mg/ml into IFA. Alum (Inject Alum) was purchased from Pierce (Rockford, IL). For the induction of experimental autoimmune encephalomyelitis (EAE), mice were injected with PLP139–151 (SJL) or MOG35–55 (C57.BL/6) 100 μg in CFA, and pertussis toxin was injected (0.2 μg) twice, once immediately and once 24 h after immunization (19). The care of mice was in accordance with institutional guidelines.

Plates (Nunc Immunoplate; Fisher Scientific, Pittsburgh, PA) were coated with OVA (10 μg/ml) overnight at 4°C, then blocked for 1–2 h with 0.1% gelatin, both in PBS containing 0.05% Tween 20 (PBST). The test serum was added and incubated overnight at 4°C. Plate-bound Ab was detected by alkaline-phosphatase-coupled anti-mouse Ig. Affinity-purified goat anti-mouse IgG (H and L chains) from Southern Biotechnology Associates (Birmingham, AL) was used to detect total Ig; the isotype-specific Abs used to detect IgG1 and IgG2a were also from Southern Biotechnology Associates. p-nitrophenyl phosphate disodium was used for the development of the colorimetric reaction.

Single cell suspensions from draining lymph nodes (dLN) and nondraining LN and the spleen were prepared as previously described (20). CD4 cell fractions were obtained with >97% purity by passing single cell suspensions over mouse CD4 subset columns (R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions. Cells from the CNS were prepared as follows. After sacrificing, the animals were perfused with PBS, and the spinal cords were removed from the entire vertebral column and placed into DMEM medium. The spinal cord was disrupted with the back of a syringe. The resulting cell suspension was filtered through a Falcon Cell Strainer 2350 (BD Biosciences, San Jose, CA). The cells were washed twice with DMEM and subsequently counted. The cells were resuspended in HL-1 medium (BioWhittaker, Walkersville, MD) supplemented with 1% glutamine and plated typically at 5 × 10⁴–5 × 10⁵ cells/well, or at serial dilutions. Cells from the peritoneal cavity were obtained by lavage after injecting 7 ml of DMEM medium. Mouse blood was obtained by retroorbital bleeding, using heparin as an anticoagulant. The blood was diluted four times with sterile saline, and PBMCs were obtained by density gradient centrifugation over a Ficoll Histopaque 1083 gradient (Sigma- Aldrich). Single cell suspensions from dLN and nondraining LN and the spleen were prepared as previously described (21). For flow cytometric studies of the cell populations obtained, we used the directly labeled Abs specified, all from BD PharMingen (San Diego, CA). The stained cells were analyzed with a FACScan and CellQuest software (BD Biosciences). The cells isolated from spinal cords of mice with EAE contained 19–43% CD4 cells.

Plates (ImmunoSpot M200; Cellular Technology, Cleveland, OH) were coated overnight at 4°C with the cytokine-specific capture Abs specified below. The plates were then blocked with 1% BSA in PBS for 1 h at room temperature and washed four times with PBS. Subsequently, nonirradiated or irradiated (3000 rad) LN or splenic APC from naive, syngenic mice were added (5 × 10⁵ cells/well, or as specified). Freshly isolated CD4 cells or single cell suspensions from spinal cords were plated in serial dilution in two to four replicate wells with or without the nominal Ag or control Ags. Freshly isolated spleen cells and LN cells were plated at 10⁶ and 7 × 10⁵ cells/well respectively, or as specified, in two to four replicate wells with or without the nominal Ag or control Ags. The assay medium was serum-free HL-1 (BioWhittaker) supplemented with 1 mM l-glutamine. Following 36 h of cell culture in the incubator at 37°C, the cells were removed by washing three times with PBS and then four times with PBST, and biotinylated detection Abs (as specified below) were added. After overnight incubation at 4°C, plates were washed three times with PBST, followed by a 2-h incubation at room temperature with streptavidin-alkaline phosphatase conjugate (DAKO, Carpenteria, CA) at 1/2000 dilution. Plates were then washed twice with PBST and twice with PBS. For two-color assays, two detection Abs were added simultaneously, followed by incubation with streptavidin-alkaline phosphatase and subsequent incubation with two substrates. The nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate substrate (Kirkegaard & Perry Laboratories, Gaithersburg, MD) was added first. Then, after washing twice with PBS, the 3-amino-9-ethylcarbozole substrate (Pierce) was added (15–30 min for nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate and 20–40 min for 3-amino-9-ethylcarbozole). The following coating mAbs were used for IL-2, IL-3, IL-4, IL-5, and IFN-γ: JES6-1A12 (5 μg/ml), MP2-8F8 (5 μg/ml), BVD4-1D11 (2 μg/ml), TRFK5 (5 μg/ml), and R46A2 (2.5 μg/ml) (BD PharMingen). For the IL-4:IFN-γ two-color assay, we used a combination BVD4-24G2-biotin:XMG1.2-HRP (BD PharMingen). HRP labeling of Abs was performed by the periodate method (22). The detection Ab concentrations were as follows: JES6-5H4-biotin (2 μg/ml), MP2-43D11-biotin (2 μg/ml), BVD4-24G2-biotin (2.5 μg/ml), TRFK4-HRP (2 μg/ml), XMG1.2-HRP (2 μg/ml), and XMG1.2-biotin (2 μg/ml).

The image analysis was performed as previously described (20). Briefly, we used a Series 1 ImmunoSpot Image Analyzer (Cellular Technology) that was customized for two-color analysis. Digitized images were analyzed for the presence of areas in which color density exceeded background by a factor set on the basis of the comparison of control wells (containing T cells and APC without Ag) and experimental wells. After separating spots that touch or partially overlap, additional criteria of spot size and circularity were applied to gate out speckles and noise caused by spontaneous substrate precipitation, nonspecific Ab binding, and ELISA effects. Objects that did not meet these criteria were ignored, and areas that met them were recognized as spots, counted, and highlighted. For two-color ELISPOT image analysis, the analyzer detects red, blue, and double-colored spots separately by using three different threshold settings. Each color threshold is set in red-green-blue mode and consists of three numbers reflecting the threshold in the red, blue, and green channels. The red and blue thresholds are set by using spots from single-color assays. With the blue threshold active, single-color blue and all double-positive spots (which have blue as a part of their color composition) are detected and outlined. Similarly, under the red threshold, only red single-color and double-color spots are detected. The double-color threshold is a mathematical intersection of the two single-color thresholds. In the final step of analysis, single-color and double-color spots are highlighted with artificial colors: blue, red, and green for single-blue, single-red, and double-color spots, respectively. To eliminate the assessment of partially overlapping red and blue spots as double-color spots, the image analyzer counts only double-color spots that are formed by the color mixture of concentric blue and red spots with single dense centers.

This assay was performed as previously described (23, 24). Briefly, sera were obtained from five to seven mice immunized with OVA in CFA or alum, and pooled. The abdomens of unimmunized mice were shaved, and 30 μl of immune serum was injected intradermally in serial dilutions as specified in Fig. 4. Injections were repeated 24 h later. Three hours after the second injection, mice were injected i.v. with a mixture of 100 μl of 0.5% Evans blue dye and 1 mg of OVA. PBS injection was used as a control. Thirty minutes later, the mice were sacrificed, the skin on the belly was inverted, and cutaneous reaction was evaluated by the sizes of visible blue spots. A reaction was scored according to the following scale: negative, less than 4 mm in diameter; 1, 4 mm; 2, between 4 and 6 mm; 3, between 6 and 10 mm; and 4, >10 mm.

It has been controversial to what extent highly polarized type 1 immunity can be induced by immunization with adjuvants. We revisited this question, taking advantage of cytokine ELISPOT assays that permit monitoring cytokine production by individual cells in the frequency below the detection limit of flow cytometry-based detection systems (<1:10,000), and at which Ag-specific T cells apparently invariably occur after immunizations with adjuvants (21). We studied immunity induced in C57.BL/6 mice after injecting s.c. OVA protein emulsified in CFA. dLN were isolated 10 days after the immunization and were challenged directly ex vivo with OVA, using ELISPOT assays to measure the frequency of cells induced to produce cytokine. Cells producing IFN-γ, TNF-α, IL-2, and IL-3 were detected in the 40–100 per million range, while the frequency of cells producing IL-4 or IL-5 was <1 per million (Fig. 1,A). This cytokine hierarchy was invariably seen in the dLN at all time points measured (days 3–90) at all recall Ag concentrations tested (using either OVA protein or the I-A^b-restricted peptide OVA_323–339) (data not shown). The cytokine production was seen in immunized mice only and was recall Ag specific (21). While the cytokine signature in the dLN was therefore highly type 1 polarized, measurements done simultaneously on the spleen gave a different result. In addition to the type 1 cytokines detected in the dLN, a vigorous IL-4 (but no IL-5) recall response was detected (Fig. 1 B) over a wide range of OVA or OVA_323–339 peptide concentration. This IL-4 response was not seen in naive C57.BL/6 mice, and was induced by OVA or OVA_323–339 peptide only in OVA-immunized mice (data not shown and Ref. 20). In all 53 mice studied individually, differential production of IL-4 was seen between the spleen and the dLN. Are the T cells in the spleen therefore less differentiated Th0 cells (that coexpress type 1 and type 2 cytokines) than the fully type 1 polarized cells in the dLN?

As we showed previously (20), two-color cytokine ELISPOT assays permit studying cytokine coexpression by individual cells. Because in this assay both cytokines are continuously captured around the secreting cells during the 48-h observation period, the assay accounts for possible asynchronous production, or cytokine switching. If the OVA-specific T cells were of a Th0 type, they should coexpress IL-4 and IFN-γ, appearing as double-positive cells in a two-color IL-4/IFN-γ ELISPOT assay. Performing such assays, we found that IL-4 and IFN-γ were produced by different cells (Fig. 1 C). Therefore, because the IL-4 production in the spleen cell population does not reflect Th0 cell activity, the question emerged whether the data provide evidence for polarized Th1 and Th2 subpopulations showing different organ distributions/recirculation patterns (25, 26, 27, 28), with polarized Th2 cells being present in the spleen but not in the dLN. Alternatively, the recall Ag-induced IL-4 might not be a T cell product, but might reflect an IL-4 bystander reaction that the activated Th1 cells induce in a cell type that is present in the spleen but not in the LN.

To establish unambiguously whether the recall Ag-induced IFN-γ and IL-4 detected in the spleen cell population are indeed produced by T cells, we purified CD4 cells from spleens of immunized wild-type (WT) mice and tested them on splenic APC obtained from the respective congenic cytokine knockout mice (Table I). The frequencies of IFN-γ-producing CD4 cells were in the same range when tested on WT, IFN-γ, and IL-4 knockout nonirradiated naive splenic APC. These data establish, first, that all of the recall Ag-induced IFN-γ-producing cells in primary cell populations are T cells and, second, that the IFN-γ and IL-4^−/− APC are equally functional in stimulating Th1 cells.

Different results were obtained when CD4 cells from immunized WT mice were tested for IL-4 production. Strong IL-4 responses were detected when WT or IFN-γ^−/− splenic APC were tested. However, IL-4 was not detected when the same cells were assayed on IL-4^−/− APC (Table I). Because the activation of Th2 cells might have required IL-4 feedback by the APC, we also tested these CD4 cells on APC derived from LN cells and thymocytes from WT animals. Unlike with the tests done with WT spleen cells, the LN and thymic APC of WT mice also did not yield IL-4 production. These data suggest that the IL-4 production by the WT APC might reflect bystander cell activation. To directly test this hypothesis, we isolated CD4 cells primed by OVA:CFA immunization in IL-4^−/− mice and tested them on the aforementioned panel of APC. Also with these CD4 cells, the frequency of IFN-γ-producing cells was independent of the type of APC used. No IL-4 spots were seen on IL-4^−/− APC (showing the cytokine specificity of the detection system). Yet when these CD4 cells were tested on WT APC, a relative high frequency of IL-4-producing cells was seen. These data clearly establish that the recall Ag-induced IL-4 was not CD4 memory (Th2) cell derived, but represented a bystander reaction. The notion that the recall Ag-induced production of IL-4 in CFA-immunized spleen cells is a bystander reaction was also confirmed by cell separation experiments: when spleen cells were first activated with OVA for 12 h, followed by the depletion of CD4 cells before placing the cells into an ELISPOT assay, the CD4 cell-depleted cell fraction was void of IFN-γ- and IL-2-producing cells but still contained the IL-4-producing cells (data not shown).

The CD4 cells isolated from the OVA-immunized IL-4^−/− mice also induced this bystander reaction in RAG2^−/− splenic APC (Table I). This result shows that the cell type that produces the IL-4 is not a T, or B, or NK1.1 cell (that undergoes receptor rearrangement), but a cell type of the innate immune system that is present in RAG2^−/− mice. The higher magnitude of bystander IL-4 production by RAG2^−/− APC as compared with WT APC (approximately two to three times) most likely reflects the higher frequency of the cells of the innate immune system of RAG2^−/− mice vs WT mice.

The aforementioned experiments clearly establish that within a type 1 T cell response, IL-4-producing cells can be activated in a bystander reaction in the spleen, reaching frequencies that approximate the frequencies of the Th1 cells themselves that induce this bystander reaction. However, it remained unclear what the maximal frequency of the cells capable of the bystander reaction is, and how the magnitude of the cognate response relates to the bystander IL-4 production. To address this question, we plated in serial dilution a Th1 clone (SH10, which is not capable of IL-4 production) (20) along with a constant number of congenic, unirradiated BALB/c-SCID spleen cells. The frequency of cells producing IFN-γ or IL-4 was measured by ELISPOT after adding Ag. The number of IFN-γ spots was a linear function of the number of cloned cells plated; at all cell dilutions ∼50% of the SH10 cells plated produced detectable spots (Fig. 2,A); also by intracytoplasmic staining, 50% of the clone produced IFN-γ (data not shown). The use of splenic or LN APC yielded the same results. The IL-4 assay provided different results (Fig. 2 B). No IL-4 spots were detected on LN APC. Using splenic APC, the number of IL-4 spots vs the number of cloned cells plated yielded a hyperbolic curve, reaching a plateau value at ∼200 spots/million. Similar curves were seen when freshly isolated CD4 cells purified from spleen of OVA/CFA-primed IL-4^−/− mice were titrated with constant numbers of RAG2^−/− APC (data not shown). These data show that the cell type capable of bystander IL-4 production is absent in the LN, and present in the spleen in the rather low frequency of 1:5000. While the maximal frequency of the IL-4-producing bystander cells in the spleen was in the same order of magnitude as the CD4 cells producing the cognate IFN-γ, our efforts to detect either cell type by intracytoplasmic cytokine staining failed, being in a frequency range not amenable to flow cytometric analysis.

Irradiation of splenic APC (5000 rad) did not abrogate bystander IL-4 secretion (data not shown). Therefore, even when purified T cell populations are tested on irradiated splenic APC, apparently Ag-induced IL-4 production is being detected. IL-4^−/− APC or cell types that do not have the cell population capable of the IL-4 bystander reaction (LN cells, thymocytes) can be used, however, to reveal the cognate IL-4 production by the T cell.

The induction of bystander IL-4 production by activated CD4 cells could either require direct cell-cell interaction or be triggered by soluble factors secreted by the T cell. To address this issue, we obtained 24-h culture supernatants of splenic CD4 T cells from OVA-immunized IL-4^−/− mice. Spleen cells from RAG2^−/− mice, and LN and spleen cells from naive WT C57.BL/6 mice were cultured with these supernatants for 12 h while measuring IL-4 production in an ELISPOT assay. IL-4 spots were induced in the spleen cells, but not in the LN cells in a dose-dependent manner (Fig. 3 A). Supernatants of naive spleen cells cultured with OVA did not induce IL-4 production; the supernatants from immune cells that were cultured in the absence of OVA also did not trigger IL-4 production. Therefore, secretory products of Ag-stimulated memory cells caused the bystander IL-4 reaction, explaining the apparent recall Ag specificity of the IL-4 production.

We have tested whether recombinant cytokines can induce the bystander IL-4 reaction. Of the cytokines tested, IL-3 excelled in inducing IL-4: the maximal number of ∼200 IL-4 spots/million spleen cells was induced by ∼20 ng/ml rIL-3, and 0.5 pg led to 50% maximal activation (Fig. 3,B). This activity was specific, as it could be blocked by anti-IL-3 Ab (Fig. 3,B, inset). Only at high concentrations (>500 ng/ml) did IL-2, TNF-α, and IFN-γ induce IL-4 production, and the frequencies induced were ∼10% of that triggered by IL-3 (Fig. 3,C). The type 2 cytokines IL-10 and IL-5 did not induce IL-4 production. Therefore, the recall Ag-induced production of IL-3 and of other type 1 cytokines by Th1 cells (Fig. 1) can account for the IL-4 bystander reaction seen in the spleen. IL-3 is known to be a major activation factor of basophils/mast cells. To test whether this cell lineage is the one induced to produce IL-4, we also tested spleen cells from two different strains of mast cell-deficient mice (Fig. 3 D). The frequencies of IL-4-producing cells in the spleen were about 3-fold lower in these mice after stimulation with rIL-3, as compared with the respective congenic WT control mice. The data show that mature basophils/mast cells constitute the majority of the cells engaged in the IL-4 indirect pathway, but might not be the only cell type capable of this reaction.

The frequencies of OVA-specific IFN-γ-producing T cells, and the titers of OVA-specific IFN-γ-dependent IgG2a/c Abs permit assessing the magnitude of the specific type 1 immunity induced in vivo, after immunization with CFA. Because there is essentially no cognate IL-4 production by OVA-specific T cells after immunization with OVA:CFA, differences in magnitudes of type 1 response in IL-4^−/− and WT mice can be attributed to the bystander pathway. The frequency of OVA-specific T cells was 2- to 3-fold higher in IL-4^−/− mice than in WT C57.BL/6 mice (Fig. 4,A). Similarly, the levels of IgG2a/c Abs were higher in the IL-4^−/− animals (Fig. 4 B). These data provide evidence that in vivo, the IL-4 bystander reaction can play a role regulating type 1 immunity.

Isotype switching to IgG1 and to IgE is IL-4 dependent (11, 12). Because after CFA immunization there is no detectable cognate IL-4 production by CD4 cells (Fig. 1), measuring these Ab types should yield information about the functional consequences of bystander IL-4 production on humoral immunity. After immunization with OVA:CFA, IL-4^−/− mice produced low levels of specific IgG1 Abs, as compared with WT mice (Fig. 4,C). In the latter, the specific IgG1 levels were similar or moderately higher than after immunization with alum, which induced CD4 cells that produce IL-4 (21). In the absence of cognate IL-4 production by T cells, the bystander IL-4 reaction therefore sufficed to bring about isotype switching to IgG1. In contrast, specific IgE Abs were detected only in OVA:alum-immunized mice, in whom IL-4-producing T cells were induced (Fig. 4 D); the bystander IL-4 reaction was insufficient to bring about isotype switching to IgE production. The two IL-4-dependent processes were therefore differentially sensitive to the bystander IL-4 reaction, which added features of a partial type 2 Ab phenotype to a T cell response that was essentially pure type 1.

To extend the study of indirect IL-4 production to the immune periphery, we studied PLP139–151-induced EAE in SJL mice. Cells separated directly from the CNS of animals with clinical EAE can provide insight into whether the indirect pathway of Ag-specific IL-4 production may play a role in the regulation of the immunopathology of EAE and other autoimmune diseases. Confirming our previous findings, high-frequency peptide-specific, IFN-γ-producing T cells could be detected in dLN, spleen, blood, CNS, and the peritoneal lavage (29). Fig. 5 shows that also peptide-induced IL-3-producing cells occurred in relatively high numbers in these compartments. However, the IL-4 bystander reaction was primarily seen in the spleen and in the blood only, but not in the CNS and the dLN. Similar data were obtained in C57.BL/6 mice immunized with MOG35–55 peptide (data not shown). These data suggest that the anatomic compartmentalization of the cell type(s) mediating the indirect IL-4 reaction affects different site-specific processes differently.

The characterization of immune responses such as Th1, Th2, or mixed Th0 immunity has been a major focus of immune research. With adjuvant-guided immunity, for example, the unambiguous characterization of the type of immune response induced should permit conclusions as to the prospects of successful vaccination or immune therapy. Initially, it was thought that T cells occur in well-defined subpopulations: Th0 cells (precursors to Th1 and Th2 cells that coexpress type 1 and type 2 cytokines such as IFN-γ and IL-4) and terminally differentiated Th1 and Th2 cells (that express either IFN-γ or IL-4) (30, 31, 32, 33, 34). When using the cytokine signature as of recall Ag-stimulated cell populations directly ex vivo, the immunity induced by the classic OVA:CFA immunization displayed a pure Th1 phenotype in the dLN, but a mixed Th0 cytokine signature in the spleen, apparently prompting different conclusions as to the T cell immunity induced. Similarly, the studies of the Abs that are typically used to assess Th1/Th2 immunity provided apparently inconclusive results. IFN-γ-dependent IgG2a/c Abs (12, 35, 36, 37, 38, 39) were induced, confirming the type 1 component, but as far as the IL-4-dependent Ab classes (11, 12, 40) are concerned, IgG1 was present in high titers, while IgE was not induced.

We studied the cytokine signatures of freshly isolated individual cells using ELISPOT assays. In this approach, T cells are challenged with Ag for 24 h, a time period sufficient to induce cytokine expression in differentiated T cells, but insufficient to permit proliferation and cytokine differentiation in vitro (5). The assays that we used, including those for IL-4 and IFN-γ, have established single cell resolution (20). Unlike IL-4 measurements in culture supernatants by ELISA, which are prone to underestimate IL-4 production due to receptor-mediated capture (15, 41, 42), ELISPOT assays can reliably detect even one in 1 million IL-4-producing cells (20). Using this high resolution, the OVA:CFA-induced response qualified as a highly polarized Th1 response. In addition to IFN-γ-, TNF-α-, IL-2-, and IL-3-producing CD4 cells that were detected in the ∼50–100/million frequency range, the numbers of IL-4- or IL-5-producing CD4 cells were <1/million. Therefore, when correcting for the bystander IL-4 response in the spleen, the T cell response was at least 50-fold biased toward numbers of T cells producing type 1 cytokines. Strikingly, type 1 cytokine-producing CD4 cells, in the absence of cognate IL-4-producing CD4 cells, were seen as early as 3 days after the immunization, at all Ag doses, and irrespective of the type of APC used in the assays (data not shown and Ref. 20). Therefore, this type 1 cytokine signature seems to be acquired without the T cell passing through an extended Th0 stage. Therefore, these data show that, as far as T cell responses are concerned, polarized type 1 immunity can be readily induced by the choice of adjuvant alone.

In the absence of cognate IL-4 production by CD4 cells, specific IgE was not produced, but IgG1 was. The notion about the indirect pathway of the induction of IL-4 response having an effector function on IgG1 isotope switching without affecting the IgE switch should help in interpreting partial type 2 Ab phenotype even in face of a T cell response that is essentially pure type 1. In addition to the growing complexity regarding the cellular nature of type 1 immunity itself (whether individual type 1 cytokines are expressed by the individual T cells as cassettes (32, 33, 34), stochastically (43, 44), or in a mutually exclusive manner (5, 20, 45)), this indirect IL-4 pathway adds an additional layer of complexity to type 1 immunity in vivo.

The observed differences in the effect of bystander vs cognate IL-4 on B cell differentiation most likely reflect effective cytokine concentrations depending on the way the cytokine was delivered (Fig. 6). For cytokines that have short t_1/2 in vivo, such as IL-4 (14), the targeted release of the cytokine during the direct interaction of the T cell with the target cell (e.g., the Ag-specific B cell) will reach local concentrations that are orders of magnitudes higher than present in the intercellular space after the release of IL-4 by bystander cells, due to the dilution, degradation, and capture of the IL-4 by receptors (13). Moreover, cognate T cell Ag recognition on B cells involves the transmission of signals through costimulatory molecules, which in conjunction with the cytokine concentration may have fundamentally different effects on B cell differentiation and isotype switching (46).

The IL-4 bystander reaction was also seen when immunity was induced by injecting OVA_323–339 peptide in IFA (Refs. 20 and 21 and data not shown). Therefore, the indirect pathway response does not seem to be dependent on mycobacteria activating cells of the innate immune system. It was also seen with BALB/c, C3H, SJL, B10.PL, B10.D2, and DBA mice; the frequency of the IL-4-producing bystander cells was in a fairly comparable and narrow range in all these mouse strains (150-300/million) (data not shown). In all these mice, the IL-4-bystander reaction was detected in the spleen and in the blood, but not in the LN or in the thymus. Also, using activated supernatants and recombinant cytokines to trigger it, we found that the latter compartments are deficent in this reaction. The most likely explanation is that the cytokine-inducible IL-4-producing cell type is absent in the LN and the thymus, while it is more abundant in the spleen and in the blood. Therefore, the indirect IL-4 pathway cannot be considered an ubiquitous companion to Th1 cell activity in the organism, but each organ might contribute a different level of accompanying type 2 qualtiy. To address this hypothesis, we started to extend these studies to the CNS, being the target organ of the best-studied autoimmune model, EAE. In this model, IL-4 production in the CNS has been considered to play a role in disease recovery/remissions (47). The active disease is induced by CFA immunization with neuroantigens (we have been studying PLP139–151 and MOG35–55), and all these immunizations were found to trigger the unipolar type 1 response reported in this work for OVA, including the IL-4 bystander reaction in the spleen. Studies of the inflammatory infiltrate isolated from the CNS did not show evidence of either cognate or bystander IL-4 production. Therefore, unlike in the spleen, the target organ, the CNS, was apparently void of IL-4 indirect pathway activity.

However, our studies provided evidence for an immune modulatory role of the IL-4 bystander pathway in the inductive phase of the Th1 response. After immunization with OVA:CFA, IL-4^−/− mice developed a higher frequency of OVA-specific IFN-γ-producing T cell population, as compared with the WT C57.BL/6 mice (Fig. 4,A), and they also generated higher serum levels of specific IgG2a/c Abs (Fig. 4 B). Because there are essentially no cognate IL-4-producing memory T cells present in the WT mice at any stage of the response (starting from day 2), these data suggest it has to be the indirect IL-4 pathway that suppresses the generation of specific Th1 cells in the WT mice, vs the IL-4^−/− mice. The indirect IL-4 pathway might also render the spleen a Th2-biasing organ, vs peripheral LN. For example, Th2 immunity can be induced by injecting Ags in IFA (21, 48, 49, 50, 51). The type 2 biasing effect is more pronounced when the Ag is injected i.p. (when the priming occurs in the spleen), as opposed to s.c. (when the priming occurs in LN) (21). Thus, the indirect IL-4 pathway in the spleen might explain the type 2 bias of the spleen.

The frequency of the cell type capable of the IL-4 bystander reaction was comparable with the frequency of Ag-specific Th1 cells in the spleen and with the frequency of cognate IL-4-producing Th2 cells induced by immunizations with IFA and alum (21). Therefore, in terms of creating type 1/type 2 environments in the spleen, the magnitude of the indirect pathway IL-4 response is considerable. Yet the frequency of the Ag-specific T cells and of the IL-4 bystander cells is low, in the 1:5000 range. This low frequency has made it difficult to directly phenotype the cell that produces the bystander IL-4 in the spleen. The fact that the IL-4 bystander reaction can be induced in RAG2^−/− and SCID spleen cells rules out as the cellular source B cells, T cells, and NK1.1 cells, which require the rearrangement of their Ag receptors. Among the recombinant cytokines tested, IL-3 was the most effective inducer of IL-4 production in RAG2^−/− spleen cells. Basophils and mast cells are known to be regulated by IL-3 (52, 53, 54, 55, 56); therefore, this cell type seemed to be a prime candidate for the IL-4 bystander reaction. Supporting this notion, two different strains of mast cell-deficient mice showed a 3-fold reduced IL-4 bystander response when their spleen cells were activated by rIL-3 (Fig. 3 D) or by activated T cell culture supernatants (data not shown). The residual IL-4-producing cells could either represent immature mast cells that are present in these mice, or other cell types capable of mediating the indirect IL-4 pathway.

In conclusion, we have shown that recall Ag-induced production of IL-4 in the spleen (and the blood) does not necessarily indicate the presence of Th2 cells. While the detection of recall Ag-induced IL-4 in mixed cell populations has been generally interpreted as evidence for Th2 or Th0 memory, our data show that this indirect pathway needs to be considered in immune diagnostics. The detection of such IL-4 does not reflect the long-term cytokine commitment of the T cell compartment that is of interest for providing insights, e.g., into the efficacy of vaccinations. Because this bystander reaction is radiation resistant, IL-4^−/− APC or APC from organs that are void of the bystander cells (LN and thymus) should be used to reveal T cell-derived IL-4. Our identification of a distinct pathway of recall Ag-induced IL-4 production in certain organs, but not others, sheds light on partial type 2 immune response phenotypes, including the dissociated production of IL-4-dependent IgG1 and IgE. Further dissection of cognate vs indirect pathway IL-4 production will contribute to the better understanding of the mechanisms that underlie different facets of type 1 and type 2 immunity.

We thank T. Ansari for excellent technical assistance, Earl R. Sigmund for editorial assistance, and Drs. P. Heeger, M. Tary-Lehmann, and D. Anthony for valuable discussions.