A Two-Step Process for Cytokine Production Revealed by IL-4 Dual-Reporter Mice

Abstract

To monitor IL-4 expression at the single-cell level, we generated mice with insertions of different reporter genes into both copies of the Il4 gene that permitted the simultaneous analysis of IL-4 transcripts using GFP and IL-4 protein secretion using human CD2. Both innate and adaptive cells competent for IL-4-production were marked by GFP, while actively IL-4-secreting cells additionally displayed huCD2. After challenge with the enteric helminth, Heligmosomoides polygyrus, GFP-positive innate and adaptive cells disseminated systemically, but IL-4 secretion was predominantly mediated by CD4⁺ T cells in the intestines and draining lymphoid organs. IL-4-competent cells persisted in cured animals and memory responses reflected rapid cytokine production at the site of rechallenge. These data reveal a two-step process for cytokine production: the first generating poised cells that disseminate systemically and the second inducing the rapid production of the cytokine in response to local stimulation.

Introduction

Interleukin (IL)-4 is the signature cytokine of type 2 immunity and remains the canonical marker for Th2-polarized CD4⁺ T helper cells (Abbas et al., 1996). Other sources of IL-4 include NK T cells, basophils, mast cells and eosinophils (Brown and Hural, 1997). IL-4 is often produced in concert with IL-13 and IL-5, which are located in a contiguous cytokine gene cluster (Abbas et al., 1996; Ansel et al., 2003; Mohrs et al., 2001a). These cytokines are linked tightly to the immunopathologies of allergy and asthma but are also required for immunity to parasitic infections (Finkelman et al., 1997; Kinet, 2002; Wills-Karp, 1999).

One paradigm example of IL-4-mediated type 2 immunity is experimental infection with the strictly enteric nematode, Heligmosomoides polygyrus (Hp). Mice on a variety of inbred backgrounds are susceptible and develop persistent intestinal infection (Finkelman et al., 1997; Gause et al., 2003). After drug cure, however, rechallenged animals mount an accelerated and enhanced memory response, which results in immunity by a process dependent on both CD4⁺ T cells and IL-4 (Urban et al., 1991a; Urban et al., 1991b). Indeed, CD4⁺ T cells with a type 2 cytokine mRNA profile accumulate early at the host:parasite interface in a protective memory response (Morimoto et al., 2004). Collectively, these observations suggest that the rapid production of IL-4 by CD4⁺ Th2 memory cells at the peripheral site of infection is critical for immunity.

Despite the importance of IL-4 for type 2 immunity and immunopathology, it has been extremely difficult to identify IL-4-expressing cells in vivo, because cytokines are rapidly secreted. To overcome this limitation we previously generated bicistronic IL-4 reporter mice (4get) by the targeted addition of an IRES-GFP element to the 3’ untranslated region of the endogenous Il4 gene (Mohrs et al., 2001b). Expression of the bicistronic 4get reporter is faithfully induced under conditions that elicit IL-4 expression, such as the activation of naïve CD4⁺ T cells under Th2 polarizing conditions or the accumulation of IL-4-expressing innate and Th2 cells after infection with nematode parasites (Gessner et al., 2005; Mohrs et al., 2001b; Shinkai et al., 2002; Voehringer et al., 2004). However, it was puzzling that GFP fluorescence was maintained independently of stimulation, whereas the production of IL-4 protein, although restricted to GFP⁺ cells, was observed only upon stimulation (Gessner et al., 2005; Mohrs et al., 2001b; Voehringer et al., 2004). Other IL-4 reporter “knockin” mice that were generated by IL-4 replacement strategies (Ho et al., 1999; Hu-Li et al., 2001; Rivière et al., 1998) other than the bicistronic addition, showed that expression of these reporters were, like IL-4 protein production, observed only after stimulation. Intriguingly, the persistent GFP expression in 4get cells correlated with the presence of high IL-4 mRNA levels in Th2-polarized CD4⁺ T cells days after the initial stimulation, even at times when IL-4 protein production was not detected (Gessner et al., 2005; Grogan et al., 2001). This striking correlation suggested that the particular configuration of the bicistronic 4get reporter links GFP fluorescence to Il4 gene expression and the presence of IL-4 transcripts, whereas IL-4-substituting reporters are a surrogate of IL-4 protein production. This assumption is further supported by data showing that innate IL-4 producers such as NK T cells, basophils, eosinophils and mast cells, which are constitutively GFP⁺ in 4get mice, also contain high levels of IL-4 mRNA in the resting state, but do not produce IL-4 protein unless stimulated (Gessner et al., 2005; Stetson et al., 2003; Voehringer et al., 2004).

To investigate this further, we generated IL-4 reporter mice, designated KN2 (knockin huCD2), by replacing the first two exons of IL-4 with a huCD2-encoding sequence. The use of heterozygous 4get/KN2 mice allowed us to visualize the expression of both reporters simultaneously in single cells while IL-4 protein production is preserved on the bicistronic 4get allele (Mohrs et al., 2001b). Using these dual IL-4 reporter mice, we show that CD4⁺/GFP⁺ Th2 cells disseminate and persist widely in response to infection with the strictly enteric parasite, H. polygyrus. In contrast, IL-4 protein-producing Th2 cells are restricted to sites of antigen accumulation, such as the draining lymphoid organs and the intestinal tissue. Similar observations were made in studies of innate cells competent to produce IL-4, however our data do not suggest that these cells are a substantial source of IL-4 in response to a primary or recall infection with H. polygyrus. Our data reveal that the production of IL-4 occurs in two distinct, highly regulated steps. This process is likely applicable to other effector cytokines, and may have important implications for the understanding of chronic inflammatory processes.

Materials and Methods

Mice and Parasites

Human CD2/IL-4 reporter mice, designated KN2 (knockin CD2), were generated by replacing the first exon of the mouse Il4 locus by integration of a previously described huCD2 reporter cassette (Sawada et al., 1994). The reporter cassette was organized by using the following elements beginning at the 5’ end: the noncoding first and beginning of the second exon of the murine Cd4 gene with the first intron containing two deletions that excise the silencer element responsible for shutting off CD4 expression in CD8 cells, the human CD2 cDNA cloned inframe into the ATG start codon in exon 2 of the Cd4 gene, a SV40 polyadenylation sequence and a neomycin selection cassette flanked by loxP sites (Mohrs et al., 2001b; Sawada et al., 1994). This reporter cassette was flanked on the 5’ end with 4.8 kb of the 5’ Il4 sequence ending at the IL-4 start codon and on the 3’ end with the 700 bp of the second intron from the Il4 gene. After sequence verification, the construct was introduced by electroporation into 129/SvJ embryonic stem cells (Mohrs et al., 2001b) and G418-resistant clones were confirmed for correct integration by Southern blotting. Transient Cre transfection was used to excise the loxP-flanked neomycin resistance cassette (Mohrs et al., 1999). Following generation of chimeric mice, two individual lines of mice were generated that successfully transmitted the gene and one, designated KN2, was backcrossed to BALB/c for 10 generations. The genotype of KN2 mice was determined by multiplex PCR of genomic DNA prepared from tail biopsies using the following primers: A: 5’-AGAGAGGTGCTGATTGGCCCAG-3’, B: 5’-CTATCACAGGCATTTCTCATTCAG-3’, C: 5’-ATGGCAGGCAAAGATGAGAAGGGC-3’

4get mice, in which the Il4 gene is linked via an internal ribosomal entry site (IRES) to GFP (Mohrs et al., 2001b), were backcrossed to BALB/c mice and C57BL/6 mice for 10 and 8 generations, respectively. Homozygous 4get and KN2 mice were interbred to generate heterozygous 4get/KN2 mice. Animals were kept under specific pathogen-free conditions in filter top cages at the animal facility of the Trudeau Institute and were used at 8-12 weeks of age. Infective third-stage larvae of H. polygyrus (Hp) were prepared as described (Gessner et al., 2005). Experimental animals were inoculated by oral gavage with 200 third-stage larvae. Where indicated the animals were treated twice orally with the antihelminthic drug pyrantel pamoate (2 mg per mouse). Crude parasite extracts were prepared from extensively washed adult worms by vigorous mechanical disruption using a glass tissue grinder (Fernando Monroy et al., 1985). Hp-larvae were killed by repeated freeze-thaw cycles prior to homogenization. S. mansoni eggs were purified as published and mice were injected i.p. with 2500 eggs (MacDonald et al., 2001). All experimental procedures involving mice were approved by the Institutional Animal Care and Use Committee at Trudeau Institute and the University of California San Francisco.

Tissue Sampling and Preparation

Single cell suspensions were prepared from the mesLN, PP, and the SP by mechanical disruption. Erythrocytes were removed from SP and blood by using ammonium chloride lysis. BAL cells were collected by 5 consecutive washes of the respiratory tract with 1 ml PBS/1% BSA each. Adherent cells were depleted from BAL and PEC by incubation in complete RPMI 1640 (cRPMI, 10% heat-inactivated fetal calf serum, 50 μM 2-ME, 2 mM L-glutamine and 100 U/ml penicillin/streptomycin) at 37°C, 5% CO₂ for 2 h. Intrahepatic lymphocytes were isolated as described (Huang et al., 1994). Briefly, livers were perfused, forced through a cell strainer and digested for 40 min at 37°C with collagenase IV (100 U/ml; Sigma) and DNaseI (10 U/ml; Sigma) in RPMI 1640 medium without serum. Hepatocytes were sedimented at 30 × g. IEL and cells from the LP were isolated from the small intestine after removal of the PP as described (Laky et al., 1997). The small intestine was cut longitudinally, washed with CMF, subsequently stirred with cRPMI medium for 30 min at 37°C and finally briefly vortexed. This step was repeated and the IEL fractions were pooled. LP lymphocytes were obtained by repeated digestion with collagenase VIII (100 U/ml; Sigma) in cRPMI for 60 min at 37°C. Intrahepatic lymphocytes, IELs and LPs were enriched in the interphase of a discontinuous 60%/40% Percoll (Amersham Biosciences) gradient centrifuged at 1200 × g for 20 min at room temperature.

Flow Cytometry

Flow cytometry reagents were purchased from BD Biosciences, Caltag Laboratories or eBioscience, unless otherwise stated and clone designations are given in parentheses. All samples were first incubated with anti-CD16/32 (2.4G2) to block non-specific binding of antibodies to Fcγ III/II receptors. The following mAbs against mouse antigens were used as PE-, APC-, APC-Cy7- or biotin conjugates and are listed in alphanumerical order: α4β7 integrin (DATK32), CCR3 (83101; R&D Systems), CD3ε (145-2C11), CD4 (RM4-5), CD11a (M17/4), CD25 (PC61 5.3), CD27 (LG.3A10), CD43 (1B11), CD44 (IM7.8.1), CD45RB (16A), CD49d (9C10), CD51 (RMV-7), CD62L (MEL-14), CD69 (H1.2F3), CD95 (Jo2), CD117 (c-Kit, 2B8) CD124 (IL-4Rα, M1), CD162 (PSGL-1, 2PH1), IgE (R35-72), NK1.1 (PK136), TCRβ (H57-597). Additional reagents included anti-human CD2 (RPA-2.10), streptavidin-APC and streptavidin-PE. Annexin V staining (BD Biosciences) and the IL-4 cytokine secretion assay (Hu-Li et al., 2001) (Miltenyi Biotec) were performed according to the manufacturer's instructions. Dead cells were discriminated by the addition of propidium iodide (PI, 3 μg/ml, Sigma) and excluded from the analyses.

Samples were acquired on a FACSCalibur (Beckton Dickinson) flow cytometer or a FACScan (Beckton Dickinson) flow cytometer equipped with a Multicolor upgrade (Cytek Development) permitting the detection of APC and APC-Cy7. Data were analyzed using FlowJo (Tree Star) software. Numbers on flow cytometry plots indicate the percentage of cells in the respective quadrant and fractions of percentages were rounded to the nearest full digit. Median fluorescence intensities (MFI) were only directly compared when samples were acquired in the same experiment with the same instrument settings.

Cell culture

CD4⁺ T cells were purified by using negative selection from lymph nodes using magnetic beads (Miltenyi Biotec) according to the manufacturer's instructions. Purified CD4⁺ T cells (1 × 10⁶/ml) were stimulated as described (Mohrs et al., 2001b) in cRPMI with anti-CD3ε (145-2C11, 2 μg/ml) and anti-CD28 (37.51, 5 μg/ml) in the presence of IL-2 (5 ng/ml) and irradiated splenic APC (5 × 10⁶/ml) for 5-7 days. Th1 cultures were supplemented with IL-12 (5 ng/ml) and anti-IL-4 (11B11, 20 μg/ml), and Th2 cultures with IL-4 (50 ng/ml) and anti-IFN-γ (XMG1.2, 20 μg/ml). For restimulation, tissue culture plates were coated with anti-CD3ε (10 μg/ml) and anti-CD28 (5 μg/ml) or the cells were cultured for 4 h in the presence of PMA (50 ng/ml) and ionomycin (500 ng/ml). For intracellular cytokine staining Brefeldin A was added during the last 2 h. Stimulation ex vivo was done either with plate-bound anti-CD3ε (10 μg/ml), soluble anti-IgE (2 μg/ml), IL-3 (30 ng/ml) plus IL-18 (30 ng/ml) ionomycin with or without PMA. Larvae and worm extracts were used at 20 μg/ml to stimulate peritoneal exudate cells. Soluble S. mansoni egg antigen was used at 50 μg/ml (MacDonald et al., 2001)

In vivo activation

NK T cells were activated in vivo by the i.v. injection of anti-CD3ε (2 μg) (Min et al., 2004; Yoshimoto and Paul, 1994). FcεRI-bearing basophils were sensitized in vivo by i.p. injection of 20 μg murine anti-DNP IgE (SPE-7, Sigma). After 24 h mice were injected i.p. with PBS or 1 mg DNP_30-40-HSA (Sigma) and analyzed 2-4 h later (Dombrowicz et al., 1993). Hp-infected mice were injected i.v. either with PBS, anti-CD3ε (2 μg), anti-IgE (2 μg), with IL-3 (1 μg) plus IL-18 (1 μg), or with 50 μg of crude worm extracts and analyzed 2-4 h later.

Cell Sorting

Two weeks after Hp infection, CD4⁺ T cells were purified by negative selection from the mesenteric lymph nodes by using magnetic beads (Miltenyi Biotec). The purified samples were subsequently stained with anti-CD4-APC and anti-huCD2-PE and sorted by using a FACSVantage flow cytometer (Beckton Dickinson) equipped with DiVa electronics.

RT-PCR

Total RNA was extracted by using the RNAqueous-4PCR kit (Ambion) and reverse transcribed with the Superscript II RNase H^- kit (Invitrogen) using oligo(dT)₁₈ priming. Specific primers and probes have been described(Gessner et al., 2005). Quantitative real-time RT-PCR was performed by using an ABI Prism 7700 Sequence BioDetector (PE Biosystems) according to the manufacturer's instructions (TaqMan, Perkin Elmer). All cytokines were analyzed in triplicates. Ct values for GAPDH were routinely between 15 and 18 cycles and normalization to β2m gave similar results.

ELISA

Cytokines in culture supernatants were quantified in standard ELISA assays by using paired antibodies for IL-2 (JES6-1A12, JES6-5H4), IL-4 (11B11, BVD6-24G2), IL-5 (TRFK5, TRFK4), and IL-13 (38213, polyclonal goat anti-mouse IL-13). Recombinant mouse IL-2 (BD), IL-4, IL-5, and IL-13 (R&D Systems) were used as standards. The detection limits were typically 20 pg/ml for IL-4, 40 pg/ml for IL-2 and IL-5, and 160 pg/ml for IL-13.

Results

Generation of dual IL-4 reporter mice

The KN2 mice were generated by introducing a cDNA encoding human CD2 transmembrane reporter protein into the endogenous Il4 locus, thus disrupting IL-4 but leaving all known regulatory elements that impact IL-4 expression intact (Ansel et al., 2004; Ansel et al., 2003; Lee et al., 2003; Mohrs et al., 2001a) (See Supplementary Fig. 1). As expected from the Il4 gene replacement, cells from homozygous KN2 mice failed to produce IL-4 under conditions where IL-4 could be readily measured from heterozygous mice by using both in vitro and in vivo assays (See Supplementary Fig. 1 and data not shown). The KN2 mice are similar to previously generated animals that use huCD2 as an IL-4-replacing reporter (Rivière et al., 1998), except that the neomycin selection cassette was deleted in the KN2 line (See Supplementary Fig. 1). After backcrossing to BALB/c, KN2 homozygous mice were interbred with 4get mice to generate heterozygous 4get/KN2 mice (Fig. 1). The resulting mice have a functional copy of IL-4, as part of the bicistronic GFP transcript (Mohrs et al., 2001b), and a replacement of the other Il4 allele with the huCD2 reporter.

Figure 1. Schematic of 4get/KN2 IL-4 dual-reporter mice.

In heterozygous 4get/KN2 IL-4 dual-reporter mice one Il4 allele is marked by the addition of the bicistronic IRES-GFP reporter (4get). The first two exons of the other allele are replaced with the huCD2 reporter (KN2) cassette. 4get/KN2 mice have a functional copy of IL-4 as part of the bicistronic 4get allele. Filled boxes with numbers indicate exons. See Supplementary Fig. 1 for details.

huCD2 expression is a faithful reporter for IL-4 protein secretion

Purified CD4⁺ T cells from naïve 4get/KN2 mice were stimulated with anti-CD3 in vitro under Th1 (IL-12 plus anti-IL-4) or Th2 (IL-4 plus anti-IFN-γ) polarizing conditions in the presence of irradiated APCs. As expected (Mohrs et al., 2003; Mohrs et al., 2001b), after 6 days, the vast majority of Th2-primed cells expressed GFP, even in the absence of stimulation beyond priming on day 0 (Fig. 2a). However, neither did these cells secrete IL-4 nor did they express the huCD2 reporter. This is consistent with numerous publications showing that resting Th2 cells do not produce IL-4 protein or express an IL-4-replacing reporter (Ho et al., 1999; Min et al., 2004; Rivière et al., 1998). In contrast, when resting Th2 cells were stimulated with plate-bound anti-CD3/CD28, they secreted IL-4 protein, expressed huCD2 and increased in GFP fluorescence (Fig. 2a). Both IL-4 production and huCD2 expression were restricted to GFP⁺ cells and correlated positively with the GFP brightness. Thus, the level of huCD2 expression reflects directly the amount of IL-4 protein secretion. Th1-polarized CD4⁺ T cells did not express GFP (Mohrs et al., 2001b), huCD2 or IL-4, even after restimulation, providing further evidence for the fidelity of both reporters. Comparable results were obtained when the cultures were restimulated with PMA and ionomycin for 4 h (data not shown). To determine the half-life of the huCD2 reporter on the cell surface, the TCR-mediated stimulation was abruptly terminated by transferring the stimulated cells (Fig. 2b; 0 h) to uncoated wells. Although the cells remained GFP⁺ 24 and 48 h later (Fig. 2b), both the frequency and the MFI_huCD2 of huCD2⁺ cells decreased rapidly. The decline of both parameters was paralleled by decreased GFP fluorescence. Based on both the frequency and the expression levels we determined the half-life of surface huCD2 to be approximately 24 h. Upon restimulation of resting Th2 cells the vast majority of IL-4 secreting cells also expressed surface huCD2 at all time points analyzed (Fig. 2c). Of note, the 24 h half-life of huCD2 leaves a footprint on all cells that have been producing IL-4 at any period during the culture, while the cytokine secretion assay only marks cells that were actively secreting IL-4 during the last 45 min. These data show that surface expression of the huCD2 reporter faithfully reflects the secretion of IL-4 protein, while expression of the bicistronic IL-4-GFP reporter indicates the competence for rapid IL-4 production.

Figure 2. huCD2 expression faithfully reflects IL-4 protein secretion.

CD4⁺ T cells were purified from naïve 4get/KN2 mice and stimulated in the presence of APC under Th2 (IL-4 + anti-IFN-γ) or Th1 (IL-12 + anti-IL-4) polarizing conditions. (a) After 6 days these cultures were transferred to uncoated (resting) or anti-CD3ε-coated (αCD3) wells and analyzed the next day for huCD2 expression (top panel) or IL-4 secretion (bottom panel) using a cytokine secretion assay. (b) The plate-bound anti-CD3ε stimulation of Th2 polarized cultures was abruptly terminated by transfer to fresh, uncoated wells (0 h) and the cultures were analyzed 24 and 48 h later. Vertical numbers indicate the MFI of GFP⁺/huCD2⁺ population. (c) Th2 cultures were generated and restimulated as in (a) for the indicated periods of time (right panel). huCD2 expression and IL-4 secretion were determined as in (a). Representative FACS plot depicts the 8 h time point. Q1: IL-4^-/huCD2⁺; Q2: IL-4⁺/huCD2⁺; Q3: IL-4⁺/huCD2^-. All huCD2⁺ and/or IL-4⁺ cells were also GFP⁺ (see a, b and data not shown).

huCD2 expression reflects IL-4 protein secretion by lymphoid and nonlymphoid cells in vivo

NK T cells rapidly produce IL-4 upon injection of anti-CD3 (Min et al., 2004; Stetson et al., 2003; Yoshimoto and Paul, 1994). We have previously shown that NK T cells are constitutively GFP⁺ in untreated 4get mice, but do not spontaneously secrete IL-4 protein (Stetson et al., 2003). To analyze the expression of huCD2 on CD4⁺/NK1.1⁺ NK T cells, we crossed KN2 BALB/c mice, which do not express NK1.1, to NK1.1-bearing C57BL/6 4get mice. A substantial proportion of splenic CD4⁺/NK1.1⁺ NK T cells was GFP⁺ in naïve mice, but none of the GFP⁺ cells expressed huCD2 (Fig. 3a). However, within 45 min after i.v. injection of anti-CD3, NK T cells expressed huCD2 on their surface. Both the frequency and the MFI_huCD2 increased between 45 and 90 min after injection. In addition to NK T cells, a very low frequency of conventional CD4⁺ T cells (CD4⁺/NK1.1^-) was GFP⁺ in the spleen of untreated mice (Fig. 3a). Almost all of these cells were huCD2^- but rapidly upregulated huCD2 upon injection of anti-CD3. Similar to NK T cells, both the frequency and the MFI_huCD2 of huCD2 expression increased between 45 and 90 min after injection. The expression of huCD2 by NK T and conventional CD4⁺ T cells in 4get/KN2 mice was absolutely restricted to GFP⁺ cells. No huCD2 staining was detected in 4get/+ littermate controls despite a similar frequency of GFP⁺ cells. Similar results were observed in the liver and the PBL (data not shown).

Figure 3. huCD2 expression reflects IL-4 protein secretion by lymphoid and nonlymphoid cells in vivo.

(a) Naïve 4get/KN2 mice or 4get/+ littermate controls were injected i.v. with either PBS (0 min) or anti-CD3ε. Splenic NK T cells (NK1.1⁺/CD4⁺) or conventional (NK1.1^-/CD4⁺) CD4⁺ T cells were analyzed 45 and 90 min later for the expression of GFP and huCD2. (b) Naïve 4get/KN2 mice were either injected with PBS or were sensitized by i.p. injection of anti-DNP mouse IgE. 24 h later the sensitized animals were injected i.p. with either PBS or DNP_30-40-HSA. Basophils (FcεRI⁺/CD4^-/SSC^lo) in the livers of all groups were analyzed 2 h later for the expression of GFP and huCD2.

Basophils express the high affinity IgE receptor (FcεRI) and are a major source of IL-4 produced by non-B, non-T cells (Gessner et al., 2005; Khodoun et al., 2004; Min et al., 2004; Seder et al., 1991; Voehringer et al., 2004). To study the expression of huCD2 by basophils in vivo, we sensitized naïve 4get/KN2 mice and 4get/+ controls by injection of mouse anti-DNP IgE. FcεRI-bearing cells were then selectively activated by injection of the DNP_30-40-HSA antigen (Dombrowicz et al., 1993). Basophils (FcεRI⁺/c-kit^-/CD4^-/SSC^lo) in the livers of untreated mice were almost exclusively GFP⁺ (Gessner et al., 2005) but did not express huCD2 (Fig. 3b). Upon sensitization, basophils (gated on GFP⁺/c-kit^-/CD4^-/SSC^lo) upregulated FcεRI expression as expected (Yamaguchi et al., 1997), but remained huCD2^- (Fig. 3b and data not shown). Injection of the antigen induced the expression of huCD2 within 1 h (data not shown) and by 4 h almost half of the basophils in the liver were huCD2⁺ (Fig. 3b). Similar responses were observed in the spleen and PBL (data not shown). These data show that the IL-4-replacing huCD2 reporter is rapidly expressed on conventional CD4⁺/GFP⁺ T cells, NK T cells and nonlymphoid IL-4 producers under conditions known to induce IL-4 protein production in vivo.

GFP and huCD2 expression in response to infection with H. polygyrus

With the fidelity and exquisite sensitivity of the IL-4-replacing huCD2 reporter demonstrated in vitro and in vivo, we next analyzed the IL-4 response to infection with the strictly enteric nematode, H. polygyrus (Finkelman et al., 1997). As expected (Mohrs et al., 2001b), naïve mice had a very low frequency of GFP⁺ cells within the CD4⁺ population and huCD2 expression was minimal in all tissues analyzed (Fig. 3a and data not shown). Two weeks after infection high frequencies of CD4⁺/GFP⁺ Th2 cells were found in the draining mesLN and all other tissues analyzed (Fig. 4a). Despite the systemic dissemination of CD4⁺/GFP⁺ Th2 cells, the expression of huCD2 was largely restricted to the draining mesLN (44% of CD4⁺/GFP⁺), PP (49% of CD4⁺/GFP⁺), IEL (52% of CD4⁺/GFP⁺) and LP (16% of CD4⁺/GFP⁺). In contrast, huCD2 expression was minimal (less than 5% of CD4⁺/GFP⁺ cells expressed huCD2) in the liver, PEC and BAL, although these tissues were routinely among those with the highest frequencies of GFP⁺ cells within the CD4⁺ population (Fig. 4a).

Figure 4. GFP and huCD2 expression in response to infection with H. polygyrus.

4get/KN2 mice were infected with H. polygyrus and analyzed 2 weeks later. (a) CD4⁺ T cells in the indicated organs were analyzed directly ex vivo for the expression of GFP and huCD2. (b) Hp-infected animals were injected i.p. with PBS (PBS), anti-CD3ε (αCD3), anti-IgE (αIgE), IL-3 + IL-18 or worm extract and sacrificed 2-4 h later. CD4⁺ T cells, basophils (FcεRI⁺/CD4^-/SSC^lo), and eosinophils (CCR3⁺/CD4^-/SSC^hi) in the liver were analyzed for the expression of GFP and huCD2. (c) Peritoneal exudate cells were cultured in the absence (no extract) or presence of crude extracts prepared from adult worm (worm extract), larvae extract or S. mansoni egg antigen (SmEA) (top panel). The next day CD4⁺ T cells were analyzed for the expression of GFP and huCD2. The same procedure was performed after mice were immunized 1 week earlier i.p. with S. mansoni eggs (bottom panel).

In addition to conventional CD4⁺ T cells, a substantial fraction of non-CD4 cells contributed to the GFP⁺ population in various tissues including the liver, spleen, PBL and PEC (Gessner et al., 2005). We have previously identified these constitutively GFP⁺ cells as basophils (FcεRI⁺/c-kit^-/CD4^-/SSC^lo) and eosinophils (CCR3⁺/CD4^-/SSC^hi) (Gessner et al., 2005; Voehringer et al., 2004). By using a different strain of IL-4-replacing reporter mice (G4 mice), Paul and colleagues suggested that basophils are a major source of IL-4 in the liver of mice infected with the nematode N. brasiliensis (Min et al., 2004). As shown here, however, basophils and eosinophils in the liver (Fig. 4b), spleen, blood or the PEC (data not shown) of Hp-infected 4get/KN2 mice did not express huCD2, despite their abundance and robust GFP fluorescence.

To test whether the lack of huCD2 expression by these cells was due to their general inability to express the reporter or to cell-intrinsic non-responsiveness, blood-borne cells, which are almost exclusively huCD2^- ex vivo (Fig. 4a and Supplementary Fig. 2), were cultured ex vivo for 4 h in the absence or presence of various stimuli. The activation of basophils either with anti-IgE, IL-3 and IL-18 (Yoshimoto et al., 1999) or ionomycin induced robust huCD2 expression (See Supplementary Fig. 2). Similarly, blood-borne CD4⁺/GFP⁺/huCD2^- T cells (Fig. 4a and Supplementary Fig. 2) expressed huCD2 after selective stimulation with anti-CD3. Unexpectedly, anti-CD3 stimulation also resulted in the expression of huCD2 on basophils, presumably induced indirectly by T cell-derived factors. Although eosinophils did not respond to any of the specific stimuli, they expressed huCD2 in response to stimulation with ionomycin.

To test whether these cells were rendered non-responsive in vivo by an immunosuppressive environment, the same stimuli that induced huCD2 expression ex vivo (See Supplementary Fig. 2) were delivered to Hp-infected mice. As observed in naïve mice (Fig. 3a), the injection of anti-CD3 induced high levels of huCD2 on CD4⁺/GFP⁺ Th2 cells in the liver (Fig. 4b), the PEC and the blood (data not shown). Similar to the stimulation of T cells ex vivo (See Supplementary Fig. 2), the injection of anti-CD3 resulted in comparable huCD2 expression on basophils. Injection of either anti-IgE, IL-3 and IL-18 or worm antigens induced basophils in the liver (Fig. 4b), spleen and the blood (data not shown) to express huCD2. Although eosinophils can be induced to express huCD2 (See Supplementary Fig. 2) and produce IL-4 ex vivo (Gessner et al., 2005; Voehringer et al., 2004), neither acute infection nor any of the stimuli tested resulted in huCD2 expression in vivo (Fig. 4b).

Finally, to demonstrate that the CD4⁺/GFP⁺/huCD2^- Th2 population contains antigen-specific cells, PEC, which contain large numbers of MHC class II-bearing macrophages and B cells (data not shown), were cultured in the presence or absence of crude extracts prepared either from adult worms or larvae. As shown in Fig. 4c, stimulation with either extract induced substantial huCD2 expression within the CD4⁺/GFP⁺ population. The elicited response was antigen-specific, because neither worm nor larvae extracts induced huCD2 expression over background on peritoneal CD4⁺ T cells from mice immunized i.p. one week earlier with Th2-inducing S. maosoni eggs (Fig. 4c) (MacDonald et al., 2001). Conversely, soluble S. mansoni egg antigen (MacDonald et al., 2001) did not induce huCD2 expression over background on peritoneal CD4⁺ T cells from H. polygyrus-infected mice, but did induce expression on CD4⁺ T cells from S. mansoni egg-injected mice. In summary, these data show that IL-4 protein secretion is largely restricted to specific antigen-stimulated CD4⁺/GFP⁺ Th2 cells in certain tissues, although GFP⁺/huCD2^- cells are fully functional and can be rapidly induced to secrete IL-4 in situ.

huCD2⁺ and huCD2^- GFP⁺ Th2 cells express different patterns of surface molecules

We next compared CD4⁺ T cells with a GFP⁺/huCD2⁺ (IL-4-secreting), a GFP⁺/huCD2^- (IL-4 competent but nonsecreting) and GFP^-/huCD2^- phenotype (not competent for IL-4 production) for the expression of various surface molecules (Fig. 5). CD4⁺/GFP⁺ cells expressed high levels of CD43, CD44, CD11a and were low for CD45RB and CD62L irrespective of huCD2 expression, consistent with their activated state. In contrast, GFP^- cells displayed a naïve phenotype (Fig. 5 and data not shown and (Mohrs et al., 2001b)). Both huCD2^- and huCD2⁺ GFP⁺ T cells shared a blasting phenotype (FSC^hi), had a comparable frequency of CD27^- cells (Hendriks et al., 2000), and had down-regulated the IL-4Rα chain (CD124). Although all GFP⁺ T cells had down-regulated the IL-4Rα chain, its expression was consistently lower on huCD2⁺ cells. The huCD2⁺ population contained a higher frequency of CD69⁺ cells, consistent with acute antigen-induced activation, and expressed higher levels of CD95, although huCD2⁺ and huCD2^- cells appear to have a similar apoptotic potential as indicated by Annexin V staining. The expression of CD162 (PSGL-1) and the integrins α2 (CD49d), α_V (CD51) and α4β7, which is required for intestinal homing (Hamann et al., 1994), were lower on GFP⁺/huCD2⁺ cells than on GFP⁺/huCD2^- cells. Thus, IL-4-secreting cells display a highly activated phenotype but express lower levels of integrins associated with tissue homing.

Figure 5. Phenotype of huCD2⁺ and huCD2^- GFP⁺ Th2 cells.

4get/KN2 mice were infected with H. polygyrus and the mesLN were analyzed 2 weeks later. CD4⁺ T cells with a GFP^-/huCD2^-, GFP⁺/huCD2^-, or GFP⁺/huCD2⁺ phenotype (as indicated on the small dot plot to the left) were analyzed for the expression of the indicated surface markers (depicted in alphanumerical order).

Impaired cytokine production by huCD2⁺ T cells despite the abundance of cytokine transcripts

To analyze the transcriptional profile of CD4⁺ T cells with the GFP^-/huCD2^-, GFP⁺/huCD2^-, or GFP⁺/huCD2⁺ phenotypes in vivo, the respective populations were sorted from the mesLN of Hp-infected mice (Fig. 6a). As shown in Fig. 6b, transcripts for IL-2 were only moderately increased in both GFP⁺ populations as compared to GFP^- cells, whereas the type 2 effector cytokines IL-4, IL-5 and IL-13 were substantially more abundant, irrespective of huCD2 expression. Although GFP⁺/huCD2⁺ cells expressed higher levels of IL-4 transcripts than GFP⁺/huCD2^- cells, consistent with their enhanced GFP fluorescence intensity, they contained substantially less IL-5 and IL-13 transcripts.

Figure 6. Cytokine transcripts and protein production by huCD2⁺ and huCD2^- GFP⁺ Th2 cells.

4get/KN2 mice were infected with H. polygyrus and the mesLN were analyzed 2 weeks later. (a) CD4⁺ T cells with a GFP^-/huCD2^-, GFP⁺/huCD2^-, or GFP⁺/huCD2⁺ phenotype were sorted from the mesLN. (b) mRNA was immediately prepared from the respective populations, reversed transcribed and analyzed by real-time RT-PCR for the indicated transcripts normalized to GAPDH. Depicted is the expression relative to GFP^- cells. Shown are the mean± SD from triplicate samples. (c) The sorted populations (a) were cultured in the absence (0 h) or presence of plate-bound anti-CD3ε. Supernatants were harvested after 4 and 24 h and analyzed by ELISA for the indicated cytokines. The detection limits are indicated by the dotted line. Depicted are the mean± SD. The stimulated CD4⁺ T cells (c) were analyzed after 4 and 24 h for the expression of huCD2 (d) and GFP versus huCD2 after 24 h (e). (f) The sorted populations (a) were either directly stimulated for 4 h with PMA + ionomycin (P+I, left panel) or rested for 18 h prior to 24 h stimulation on plate-bound anti-CD3 (rested, right panel) and the supernatants were analyzed by ELISA for IL-4.

To determine the capacity of the respective populations to produce these effector cytokines, the sorted cells were cultured in the absence or presence of plate-bound anti-CD3 and the supernatants were harvested 4 and 24 h later (Fig. 6c). CD4⁺/GFP^- cells, consistent with their naïve phenotype, produced some IL-2 (116±29 pg/ml) but none of the type 2 cytokines over 24 h. In contrast, both CD4⁺/GFP⁺ populations produced IL-2 (GFP⁺/huCD2^- 1732±17 pg/ml; GFP⁺/huCD2⁺ 325±17 pg/ml), IL-4 (GFP⁺/huCD2^- 1333±82 pg/ml; GFP⁺/huCD2⁺ 377±32 pg/ml), IL-5 (GFP⁺/huCD2^- 1059±69 pg/ml; GFP⁺/huCD2⁺ <40 pg/ml)and IL-13 (GFP⁺/huCD2^- 2729±193 pg/ml; GFP⁺/huCD2⁺ 481±21 pg/ml) within 4 hours of stimulation. Unexpectedly, the GFP⁺/huCD2^- cells produced substantially larger amounts of all cytokines at 4 and 24 h than did the GFP⁺/huCD2⁺ cells. As predicted from the increased production of IL-4, cells that were GFP⁺ but did not express huCD2 in vivo (Fig. 6a) induced huCD2 within 4 h of stimulation and after 24 h almost 50% of these cells expressed very high levels of huCD2 and GFP (Fig. 6d, e). Because increased GFP fluorescence correlates with the abundance of IL-4 transcripts (Fig. 6a and 6b), these cells expressed high IL-4 transcript levels upon stimulation. In contrast, GFP⁺ cells that also expressed huCD2 in vivo (Fig. 6a) did not upregulate huCD2 within 4 h, and only 10% expressed high levels of huCD2 and GFP after 24 h, revealing a failure to induce IL-4 transcripts and produce IL-4. Cytokine production and the increase in GFP and huCD2 expression remained compromised in the GFP⁺/huCD2⁺ population even when stimulated for 4 h with PMA/ionomycin, thereby bypassing critical TCR signaling events, or after being rested for 18 h prior to plate-bound anti-CD3 stimulation (Fig 6f and data not shown). No cytokines were secreted when any of the subsets were cultured in the absence of stimulation for 4 and 24 h (Fig. 6c) and GFP⁺/huCD2⁺ cells lost huCD2 expression with similar kinetics as determined in Fig. 2b. Thus, CD4⁺/GFP⁺ Th2 cells express high transcript levels of canonical type 2 cytokines, but huCD2⁺ cells that had recently secreted IL-4 in vivo were strikingly impaired in their capacity to upregulate IL-4 mRNA and produce cytokines upon restimulation ex vivo.

huCD2 expression is highly regulated during the course of a primary and recall infection with H. polygyrus

Although primary challenge with Hp results in chronic infection (Finkelman et al., 1997; Robinson et al., 1989) it is apparent that the elicited immune response, particularly the production of IL-4, is modulated during the course of infection (Finkelman et al., 2000). Because IL-4 mRNA is found mainly in CD4⁺ T cells and the depletion of CD4⁺ T cell ablates the IL-4 response (Svetic et al., 1993), CD4⁺ T cells are the likely source of the observed modulation. To visualize the IL-4 response over time, we infected cohorts of 4get/KN2 mice and analyzed CD4⁺ T cells in the mesLN after various periods for the expression of GFP and huCD2. As shown in Fig. 7a, both GFP⁺ and huCD2⁺ cells increased dramatically in frequency and total number (approx. 100-fold) within the first week of a primary infection. Although the total number of GFP⁺ and huCD2⁺ cells increased further during the second week, the frequencies of both populations actually decreased. To follow the formation of memory, the acute infection was terminated by drug-treatment in one group after 2 weeks, while the other group remained chronically infected (Fig. 7a). Both groups were analyzed 5 weeks later (7 weeks after the primary infection). Although the frequency of GFP⁺ was comparable between both groups, the drug-cured mice contained less than 10% of the huCD2⁺ cells as compared to the untreated animals. The observation that GFP⁺ cells persisted while huCD2 expression declined was also apparent in other tissues, including the spleen, PP, and the LP (see Supplementary Fig. 3). Moreover, the reduced frequency of huCD2⁺ cells in drug-cured mice was accompanied by lower levels of huCD2 staining and reduced MFI_GFP among GFP⁺ cells (see Supplementary Fig. 3), consistent with diminished IL-4 protein and transcript levels.

Figure 7. huCD2 expression is highly regulated during the course of primary and challenge Hp infection.

(a) 4get/KN2 mice were infected with H. polygyrus (1° Hp) and CD4⁺ T cells in the mesLN were analyzed at the indicated time points for GFP (upper panels) and huCD2 (lower panels) expression. One cohort was drug-cured (Rx) after 2 weeks while the control group remained untreated. Animals from both groups were analyzed 5 weeks after drug treatment (7 weeks after 1° Hp). The remaining drug-cured animals were rechallenged (2° Hp) 5 weeks after drug treatment and analyzed at the indicated time points. Depicted are the mean± SD. (b) Naïve (primary) or drug-cured (recall) 4get/KN2 mice were infected with H. polygyrus and CD4⁺ Th2 cells (CD4⁺/FcεRI^-/GFP⁺), basophils (FcεRI⁺/CD4^-/GFP⁺), and eosinophils (CCR3⁺/GFP⁺) were analyzed at various time points in the indicated organs for huCD2 expression. (c) Naïve (primary) or drug-cured (recall) 4get/KN2 mice were infected with H. polygyrus and CD4⁺ T cells in the LP were analyzed for the expression of GFP and huCD2 at the indicated days.

As compared to naïve mice, drug-cured animals mount an enhanced memory response against a second Hp challenge (Finkelman et al., 1997; Gause et al., 2003). The protective memory response, which requires both CD4⁺ T cells and IL-4 (Urban et al., 1991a; Urban et al., 1991b), is characterized by substantially increased and accelerated IL-4 production (Finkelman et al., 2000). Indeed, when drug-cured 4get/KN2 mice were reinfected with Hp both the frequency and the number of GFP⁺ and huCD2⁺ cells in the mesLN increased within 4 days (Fig. 7a). In contrast, primary infection of naïve mice did not result in any increase in the numbers of GFP⁺ or huCD2⁺ cells over the same time period. Although the frequency of GFP⁺ and huCD2⁺ cells was comparable by one week after either a primary or recall infection, the total numbers of GFP⁺ and huCD2⁺ cells were significantly (p<0.05 for both populations in a Student's t-test) greater in the rechallenged mice. To compare the IL-4 production by basophils and eosinophils to CD4⁺/GFP⁺ Th2 cells, we analyzed the frequency of huCD2⁺ cells in the respective populations over the course of a primary and recall infection in the liver, mesLN, blood and spleen. As shown in Fig. 7b and Supplementary Figure 4, no more than 10% of the basophils or eosinophils were huCD2⁺ at any time point in any analyzed organ while up to 50% of the CD4⁺/GFP⁺ Th2 cells were huCD2⁺. The frequency of huCD2⁺ cells in was very low in the hepatic or blood-borne CD4⁺/GFP⁺ Th2 population.

Additionally, we examined reporter expression on lymphocytes taken from the site of infection at the intestinal lamina propria. In striking contrast to naïve animals, a substantial fraction of CD4⁺ T cells was GFP⁺ in the drug-cured mice, but these cells did not express huCD2 (Fig. 7c). Within 4 days after infection, however, robust huCD2 expression was apparent in the rechallenged mice, whereas neither huCD2⁺ nor GFP⁺ CD4⁺ T cells were detected at this time following primary infection. Even up to one week after infection, GFP⁺ and GFP⁺/huCD2⁺ CD4⁺ T cells were far more frequent in the intestinal walls of the rechallenged mice and huCD2 was expressed at higher levels.

Discussion

We generated dual reporter mice to analyze the relationship between effector cells competent to secrete IL-4 with effector cells actually secreting IL-4 protein. In agreement with prior findings (Gessner et al., 2005; Mohrs et al., 2001b; Voehringer et al., 2004), the latter subset contained all of the cells in the former subset, a finding that suggests that the capacity for IL-4 secretion is gained only after cells have fully committed to their discrete effector function. For T cells, and specifically Th2 cells, competence for IL-4 production was associated with the capacity to disseminate widely throughout body tissues. Using a strictly enteric helminth, we could demonstrate that active IL-4 secretion was limited to those cells at sites of presumptive antigen presentation in intestinal tissues and draining lymph nodes. Comparisons between actively secreting and nonsecreting IL-4 effector cells demonstrated additional levels of regulation affecting activation thresholds and the potential for the secretion of other cytokines. Finally, memory responses reflected the capacity of cytokine-competent effector cells, which persisted in tissues, to respond rapidly upon re-exposure to antigen. These findings extend prior observations regarding CD4⁺ and CD8⁺ IFN-γ-producing effector cells generated during systemic infections (Masopust et al., 2001; Reinhardt et al., 2003; Reinhardt et al., 2001), but add previously unsuspected detail to the dynamic nature of immune effector function.

We have previously demonstrated that IL-4-producing cells – including Th2 cells, NK T cells, eosinophils, basophils and mast cells - from 4get mice are constitutively GFP fluorescent but do not secrete IL-4 protein unless stimulated (Gessner et al., 2005; Stetson et al., 2003; Voehringer et al., 2004). In each of these cells, the presence of IL-4 transcripts is mirrored by GFP fluorescence, suggesting that the EMCV IRES relieves translation that is otherwise blocked or undetectable from the canonical 5’-cap of the bicistronic mRNA. IRES elements confer competency for translation even under conditions where 5’-cap-dependent translation is disabled (Fernandez et al., 2002; Sachs et al., 1997). Thus, the 4get mice reliably identify cells competent for IL-4 secretion (and marked by constitutive IL-4 transcripts), but actual IL-4-secretion is only observed after activation and is accompanied by increased transcript levels and GFP fluorescence (Fig. 2a) (Gessner et al., 2005; Voehringer et al., 2004). Compared to the bicistronic 4get mice, the frequency of reporter positive cells was very low in IL-4-replacing reporter mice during immune activation in vivo (Min et al., 2004; Mohrs et al., 2001b; Voehringer et al., 2004). In order to combine the advantages of the respective reporter strategies, we generated heterozygous, dual reporter mice, thus enabling us to identify IL-4-secreting cells from amongst the greater number of IL-4 competent cells. We emphasize that the IL-4-substituting reporter mice we made used knockin strategies that leave intact all of the known regulatory sites around the Il4 locus while also deleting the neomycin selection cassette (Ansel et al., 2004; Ansel et al., 2003; Guo et al., 2004; Lee et al., 2003; Mohrs et al., 2001a). As demonstrated here (Fig. 2, 3, 4, 7) and in our prior studies (Gessner et al., 2005; Mohrs et al., 2001b), monoallelic activation of the Il4 gene does not occur during initial activation, as assessed in vitro and in vivo; in contrast, stable, long-lived, clones that express IL-4 monoallelically can be identified (Bix and Locksley, 1998; Hu-Li et al., 2001). Thus, the two reporter alleles provide a unique reagent for following cytokine transcription and protein production over relevant timeframes.

The ability to identify CD4⁺/GFP⁺/huCD2^- and CD4⁺/GFP⁺/huCD2⁺ Th2 cells enabled us to compare IL-4 producing and nonproducing Th2 cells from the same animal without in vitro restimulation. Unexpectedly, both GFP⁺ populations had down-regulated the IL-4Rα chain (CD124) (Fig. 5) and were presumably no longer receptive to the Th2 promoting effects of IL-4 (Abbas et al., 1996; Mohrs et al., 2001b). Thus, Il4 gene-expressing cells have apparently committed to the Th2 lineage and do not require further signals mediated through the IL-4Rα. Further, GFP⁺/huCD2^- and GFP⁺/huCD2⁺ populations had both down-regulated CD27 (Hendriks et al., 2000), indicating a similar frequency of terminally differentiated cells as assessed by this marker, and arguing against the progressive differentiation of IL-4 protein-producing effector cells into memory cells. Both GFP⁺ populations were CD43^hi, CD44^hi, and CD62L^lo (Fig. 5) and therefore would no longer constitutively home to secondary lymphoid organs (Bradley et al., 1994; DeGrendele et al., 1997; Stockton et al., 1998). In contrast, while GFP⁺/huCD2^- cells upregulated the expression of integrins required for their emigration into tissues (CD49d, CD51), IL-4 protein-producing cells expressed strikingly lower levels of most integrins, including α4β7, which confers homing into the intestine (Hamann et al., 1994). These data suggest that IL-4 competent Th2 cells develop an increased potential to traffic into tissues whereas those cells that were additionally stimulated to produce IL-4 protein have down-modulated their migratory potential. This mechanism may serve to confine effector functions to sites of antigen exposure.

The dual knockin strategy used to generate the 4get/KN2 mice allowed us to analyze innate IL-4-producing cells as well as adaptive immunity. In agreement with prior findings, we identified basophils and eosinophils as the primary innate cells that appear in blood and move into tissues, including the intestine, in response to helminth infection (Gessner et al., 2005; Min et al., 2004; Shinkai et al., 2002; Voehringer et al., 2004). Despite the distribution of eosinophils – particularly to the liver and the peritoneal cavity – and basophils – particularly to the liver – we did not observe significant levels of huCD2 expression in vivo, and hence active IL-4 secretion, from these cells during a primary Hp infection or rechallenge. Of note, the dual-reporter system is capable of recording even the most rapid and transient IL-4 responses, because huCD2 expression can be induced in vivo within minutes (Fig. 3a) and is maintained on the cell surface with a half-life of approximately 24 h (Fig. 2b). Therefore, any cell that might have produced IL-4 only transiently within minutes of infection will still be identified as huCD2⁺ 24 h later. Although this contrasts with recent reports that basophils initiate IL-4 production during a memory T-dependent response and helminth-induced hepatic basophils may be actively secreting IL-4 (Khodoun et al., 2004; Min et al., 2004), we unequivocally demonstrate that basophils in multiple tissues can be rapidly induced to express the huCD2 reporter in naïve or Hp-infected animals in response to FcεRI cross-linkage, IL-3 and IL-18-induced activation (Yoshimoto et al., 1999), the i.v. injection of worm extracts, presumably by antigen-specific crosslinkage of the FcεRI, and in response to indirect mechanisms mediated by the activation of T cells (Fig. 3b, 4b and Supplementary Fig. 2). One explanation for this apparent discrepancy might be that we infected orally with the strictly enteric nematode parasite H. polygyrus, whereas Min et al. infected subcutaneously with N. brasiliensis which migrates systemically before eventually reaching the intestine (Finkelman et al., 1997; Min et al., 2004; Mohrs et al., 2001b; Voehringer et al., 2004). Thus, our data do not suggest that basophils or eosinophils contribute substantially to a primary or recall IL-4 response to infection with H. polygyrus.

Based on the present data, we propose a sequential two-step model for the production of IL-4, which is likely applicable to all known IL-4-producing cells (Brown and Hural, 1997). First, the Il4 gene is rendered accessible by chromatin alterations as suggested by prior studies (Ansel et al., 2003; Grogan et al., 2003). As proposed here and elsewhere (Gessner et al., 2005; Stetson et al., 2003; Voehringer et al., 2004) based on the constitutive fluorescence of IL-4 competent cells in 4get mice, these chromatin alterations correlate with the constitutive presence of IL-4 transcripts. Cells that have accomplished this initial step are poised for rapid cytokine production but do not necessarily secrete IL-4 protein. This first step is linked with lineage differentiation in the case of innate IL-4-producing cells, such as basophils, eosinophils, mast cells and NK T cells (Gessner et al., 2005; Stetson et al., 2003; Voehringer et al., 2004). In contrast, CD4⁺ T cells defer activation of the Il4 gene until stimulated under appropriate conditions in secondary lymphoid organs. In a distinct second step, IL-4 competent cells can rapidly make and secrete IL-4 protein upon activation. This step is reversible and IL-4-producing cells can revert to a nonproducing state (Fig. 2b). Under the conditions used here, however, such cells remain competent for IL-4 production and constitutively express cytokine mRNAs as revealed by their persistent fluorescence. Importantly, CD4⁺ T cells that recently had been producing IL-4 in vivo, as assessed by the presence of surface huCD2, were impaired in their capacity to upregulate IL-4 transcript levels and produce effector cytokines in response to TCR-mediated stimulation ex vivo (Fig. 6c-f). This impairment is not due to reduced CD3 or TCR expression (Fig. 5) (Corbin and Harty, 2005) and cannot be overcome by resting the cells prior to stimulation or bypassing critical TCR signaling events by stimulation with PMA+ionomycin (Fig. 7f). This impaired responsiveness might be an important mechanism to prevent immunopathology by controlling the amount of time that effector cytokines are secreted at sites of frequent antigenic stimulation, but this will require further study.

Because Th2 cells maintained constitutive IL-4 transcripts weeks after elimination of the infection (Fig. 7a, c), memory T cells become functionally equivalent to innate IL-4-producing cells, but impart exquisite antigen specificity to the repertoire of rapid immune responses. Further studies will be required to assess whether and under which conditions IL-4 competent Th2 cells can revert to a state where constitutive cytokine transcripts are no longer maintained.

The systemic dissemination of Th2 cells poised for rapid IL-4 production has important implications for type 2 immunity and immunopathology. As we show here, CD4⁺/GFP⁺ Th2 cells persist at high frequencies in the intestinal wall of drug-cured mice and are ideally positioned to respond to recall infections. Indeed, these cells rapidly produce high levels of IL-4 upon reinfection well before Th2 cells appear in the intestine during a primary infection. The accelerated IL-4 response by CD4⁺ T cells at the site of infection likely underlies the immunity of Hp-primed mice to reinfection (Urban et al., 1991a; Urban et al., 1991b). Due to the wide dissemination of CD4⁺/GFP⁺ Th2 cells, the production of IL-4 would presumably be similarly accelerated in other tissues if the appropriate reactive antigens were introduced at those sites. However, the production of type 2 cytokines by inappropriately activated Th2 cells resident in tissues, such as the lung (Fig. 4a), might contribute to immunopathologies such as asthma (Wills-Karp, 1999). An accelerated immune response is not only the hallmark of immune memory, but also the goal of successful vaccines. Further work will be needed to address the mechanisms by which effector cells disseminate widely to tissues in the apparent absence of antigen and how their longevity, turnover and effector function is regulated.

Abbreviations used in this paper

APC

antigen-presenting cell

BAL

bronchoalveolar lavage

BM

bone marrow

FSC

forward scatter

GFP

green fluorescent protein

Hp

Heligmosomoides polygyrus

IEL

intraepithelial lymphocytes

LP

lamina propria

mesLN

mesenteric lymph nodes

PEC

peritoneal exudate cells

PLC

pleural exudate cells

PP

Peyer's patches

SmEA

S. mansoni egg antigen

SP

spleen

SSC

side scatter