We reevaluated the effects of Ag dose on the polarization of CD4 effectors generated in vitro from naive pigeon cytochrome c-specific TCR transgenic T cells under conditions in which we could eliminate contaminating non-naive CD4 cells and the effects of heterogeneous Ag-presenting populations. When the possibility of contaminating non-naive T cells was reduced by using T cells from transgenic mice on a RAG-2−/− background, Ag dose did not have a significant effect in Th1 and Th2 polarization unless exogenous IL-2 was initially added to cultures. Effectors generated were uniformly Th0 but produced only IL-2 in substantial amounts. When exogenous IL-2 was added to priming cultures, T cells secreting a Th0 phenotype (large quantities of IL-2, IL-4, IL-5, and IFN-γ) developed, except at very high doses of Ag, where there was a striking reduction in IL-4 and IL-5 secretion. Our results imply that Ag dose does not have a direct effect on Th1/Th2 polarization, except under conditions that include a high level of TCR ligation and in the presence of high levels of IL-2, where production of Th2 cytokines may be down-regulated by a mechanism that is not yet clear.
CD4+ T cells secreting restricted Th1 and Th2 patterns of cytokines differentially regulate infectious diseases and autoimmunity (1). Production of the Th1-specific cytokines, IFN-γ and TNFβ, is required for clearance or control of many viral and protozoan infections, but the inflammatory reactions caused by these cytokines may mediate tissue destruction in autoimmune diseases such as rheumatoid arthritis (2) and multiple sclerosis (3). In contrast, the Th2-associated cytokines, IL-4 and IL-5, may be required for protection against helminth infections such as Nippostrongulus and Trichuris (4, 5), but may exacerbate or mediate allergic reactions. The roles of Th1 and Th2 responses in disease susceptibility have been studied extensively in parasite and in leishmanial models (6, 7). In the leishmanial model, resistant mice developed a Th1 response (mediated by IFN-γ), whereas susceptible mice developed an IL-4 dominated or Th2 response (6, 8). Importantly, susceptibility was reversed if IL-4 was blocked during the initial response (9). The generation of polarized Th1 and Th2 subsets is critical in determining the balance of cell-mediated and humoral immunoresponses and may determine the outcome of autoimmune reactions and infections by pathogens.
Understanding the factors that promote preferential polarization of naive T cells and differentiation into effectors that have the capacity to be highly polarized (10) is important for understanding how the appropriate class(es) of immune response can be elicited. It is particularly clear that cytokines IL-4 and IFN-γ/IL-12 play critical roles in achieving Th2 and Th1 polarization, respectively, but their in situ source and which signals differentially regulate their production are less clear. One clear-cut murine model indicated that the bacterium Listeria monocytogenes, can interact with host cells such as macrophages, resulting in production of high levels of IL-12, which in turn drive a Th1-polarized response (11, 12).
There have been a number of reports suggesting that the dose of Ag can determine the polarization of a CD4 response either in vivo (8, 13, 14) or in vitro (15, 16, 17, 18, 19). Low doses of Ag in vivo were reported to favor development of delayed-type hypersensitivity, presumably driven by Th1 polarization, while high doses favored Ab driven by Th2 polarization (8, 13). However, the strain of rat (13) or mouse (12, 19, 20) also had a major impact on polarization and could reverse the Ag dosage effect.
It has been suggested that the naive T cell can respond differentially to different extents of receptor ligation (15, 21), a concept fortified by the differential effects of distinct peptide analogues providing different avidity interactions on polarization (reviewed in 21 . The consensus paradigm that has emerged from the in vitro studies is that very low doses of priming Ag promote Th2 responses (15, 18), that intermediate to high doses promote Th1 responses (15, 18), and that very high doses promote Th2 polarization (18). However, several aspects of the in vitro studies reported argue against a direct effect. First, in one study, the dosage effect relied on the addition of exogenous IL-2 (15), while in the other it was shown that Th2 polarization required initial production of IL-4 during effector generation (18). Naive CD4 T cells do not produce detectable IL-4 when stimulated, unless they are stimulated at least twice over a period of several days (22, 23). These phenomena are compatible with an indirect effect mediated by cytokine production by cells that are Ag-experienced and/or respond directly to IL-2, as had been implied in some earlier studies (24, 25). Naive T cells require high Ag doses, whereas Ag-experienced effectors can be induced to secrete cytokines at much lower levels of TCR ligation (26) or peptide (10), so non-naive cells would contribute most at a low Ag dose. Finally, it is almost impossible to eliminate contaminating, non-naive T cells even in most TCR transgenic mice, because they are not of a reliably distinct phenotype (reviewed in 27 .
Therefore, we reexamined the mechanisms responsible for dose-dependent selective polarization using a well-defined fibroblast APC population and a TCR transgenic model (with a Vβ3, Vαll receptor specific for pigeon cytochrome c peptide 88–104 (PCCF)5 in which we could substantially eliminate contamination of non-naive T cells by using mice crossed to a RAG-2−/− background. We find that under conditions in which contaminating T cells are minimized because no endogenous receptor expression is present (RAG-2−/−) or likely to grow (absence of exogenous IL-2), a Th population of cells able to secrete high levels of IL-2 with only a low level of IL-4 or IFN-γ is generated over a broad range of peptide doses. The addition of exogenous IL-2 to effectors generated over a broad range of peptide doses induced the generation of a Th0 population of cells able to secrete both IL-4 and IFN-γ at substantial levels. Interestingly, at very high doses of peptide, effectors developed that no longer produced substantial levels of IL-4 and IL-5, suggesting a down-regulation by some mechanism. From these studies, we suggest that indirect mechanisms are largely responsible for the previously observed effects of Ag dose in vitro and that such mechanisms may also contribute to the strain differences in polarization and to in vivo situations in which polarization is seen.
Materials and Methods
H-2b/k and H-2k Vα11/Vβ3 AND TCR transgenic mice were bred in the animal facilities at the University of California, San Diego (UCSD) or at the Trudeau Institute (Saranac Lake, NY) and were used at 2 to 6 mo of age. Mice were obtained on a C57BL/6 × 129 background (28) and were back-crossed 6 to 9 times to C57BL/6. Transgenic males (H-2b) were bred to B10.BR females to produce transgenic H-2b/k offspring. Transgenic H-2b/k mice were then bred four or more times to B10.BR mice to obtain transgenic H-2k offspring. B10.BR and (B10.BR × C57BL/6)F1 mice were bred in our facility at UCSD. RAG-2−/− mice back-crossed on a C57BL/6 background (10 generations) were crossed to AND TCR transgenic mice (ninth generation on C57BL/6 background). Offspring (RAG-2+/− × TCR+/−) were mated, and TCR transgenic RAG-2−/− (RAG/AND) mice were used in experiments.
CD4+ T cell isolation
Spleen and lymph node cells from TCR transgenic mice were isolated over nylon wool columns, treated with a panel of depleting Abs and complement, and purified over Percoll gradients (Sigma, St. Louis, MO) to obtain small resting CD4+ T cells (29, 30). Spleen and lymph node cells were depleted with 3.155 (anti-CD8), CA4.2.12 and M5/114.15.2 (anti-class II), J11d (anti-HSA), 33D1 (anti-CD11c), and M1/70 (anti-CD11b, Mac-1). Rat Abs were cross-linked with MAR18.5 (mouse anti-rat κ) and then incubated with guinea pig complement, baby rabbit complement, and DNase I (Sigma). High density resting CD4+ T cells were isolated using discontinuous Percoll gradient centrifugation (4 layers: 40, 52, 63, and 80%). Cells at the 63/80% interface were collected and used for T cell assays and for effector generation. Cells were routinely 90 to 99% CD4+Vβ3+ and 85 to 95% L-selectin+. Naive CD4 T cells were isolated, as described (31), by magnetic separation (MACS; Miltenyi Biotech, Sunnyvale, CA) based on L-selectin (CD62L) expression. T cells were sequentially labeled with biotinylated rat anti-mouse L-selectin (MEL-14) followed by FITC-streptavidin (Zymed, San Francisco, CA) and biotinylated magnetic beads (Miltenyi Biotech). CD4+ T cells, positively selected on magnetic beads, were at least 99% L-selectin+. CD4+ T cells isolated from transgenics on the RAG-2−/− background displayed a naive phenotype (see Fig. 1) and were not sorted with L-selectin. Inclusion of anti-NK1.1 (PK136) in the depletion step or the use of T-depleted spleen APCs did not alter the differentiation (IL-4 or IFN-γ secretion) of T cells upon restimulation (data not shown).
Cells were cultured in RPMI 1640 (Irvine Scientific, Santa Ana, CA) supplemented with penicillin, streptomycin, glutamine, 2-ME, HEPES, and 7.5% FCS (HyClone Labs, Logan, UT). Cultures were set up in 2-ml volumes in 24-well plates (Costar, Cambridge, MA). CD4+ T cells (3 × 105/ml) were stimulated in the presence of mitomycin-C-treated APC (1 × 105 DCEK.ICAM APC/ml (32) (B7-1+, ICAM-1+, I-Ek-expressing mouse fibroblast line) and various doses of PCCF (0.001–100 μM) or with no PCCF, with or without exogenous IL-2 (10 ng/ml). In some experiments, APCs were pulsed with peptide for 2 h before incubation with T cells. Pulsing APCs with peptide is ∼10-fold less efficient in activating T cells compared with soluble Ag, but reduces the ability of the few possible remaining APCs in the T cell preparation from presenting Ag. We observed no significant differences in Th1 or Th2 generation in pulsed vs soluble peptide conditions. After 4 days, cells were harvested, counted, and washed three times. Cell recovery was determined by trypan blue exclusion. For restimulation, T cells were incubated at 3 × 105/ml with DCEK.ICAM APCs (1.5 × 105/ml) and 5 to 10 μM PCCF. Supernatant was collected after 24 and 48 h for cytokine analysis.
Recombinant cytokines IL-2, IL-4, IFN-γ, and IL-5 were obtained from culture supernatants of X63.Ag8-653 cells transfected with murine cDNA for the respective cytokines (33). Recombinant murine IL-12 was kindly provided by Dr. Stanley Wolf (Genetics Institute, Cambridge, MA). The following anti-cytokine Abs were purified from ascites or were prepared by Amicon (W. R. Grace, Beverly, MA) concentration of hybridoma supernatants: 11B11 (anti-IL-4), XMG1.2 and R46A2 (anti-IFN-γ), and TRFK4 and TRFK5 (anti-IL-5). IL-4, IFN-γ, and IL-5 were detected by ELISA using 11B11, R46A2, and TRFK5, respectively, as coating Abs, and biotinylated rat anti-mouse IL-4 (PharMingen, San Diego, CA), biotinylated-XMG1.2, and biotinylated-TRFK4, respectively, as second-step reagents. The data were quantitated from standard curves using recombinant cytokines and were expressed in ng/ml. One nanogram per milliliter of IFN-γ equals 9 U/ml protein, and 1 ng/ml of IL-5 equals 8.5 U/ml protein. For effector T cells, IL-2 and IL-4 were assayed from supernatants collected at 24 h, and IL-5 and IFN-γ were assayed from supernatants collected at 48 h. IL-2 was detected by bioassay as previously described (31, 34) by measuring proliferation of the NK cell line in the presence of 11B11 (anti-IL-4). The data are expressed in ng/ml and are referenced to murine rIL-2 (PharMingen). One nanogram per milliliter of IL-2 is equal to 11 U/ml protein.
Anti-L-selectin (MEL-14), anti-CD44 (IM7), and mouse anti-rat κ (RG7) were used as concentrated supernatants. Phycoerythrin (PE)-labeled and FITC-labeled anti-CD4, PE-labeled anti-Vβ3 (KJ25), and isotype controls were purchased from PharMingen.
AND and RAG/AND TCR transgenic mice were chosen to evaluate whether the effects of Ag dose were mediated directly on the naive CD4 T cell or through indirect mechanisms. The AND Vβ3/Vα11 TCR, specific for the 88–104 fragment of PCCF, complexed to I-Ek, is particularly good at excluding endogenous receptor expression (28); among the CD4 T cell population of young TCR transgenic mice, >90% of cells express the transgene and have a naive phenotype (CD44lowCD62Lhigh) (35, 36). However, there is some contamination (5–15%) with CD4 cells expressing low levels of the TCR transgene, and many of these cells are not of naive phenotype. As shown in FigureF1 1, top, ∼14% of AND CD4 T cells were Vβ3− (nontransgenic) with ∼4% nontransgenic cells remaining after purification (middle). In contrast, almost no nontransgenic CD4 cells were detected in RAG/AND mice, either before (Fig. 1, bottom left) or after (bottom middle) purification. Mice on the RAG/AND background also had few if any CD4 T cells with a memory phenotype (CD44high) (Fig. 1, bottom right). As shown in Figure 1, 5.4% of CD4 cells in AND mice were CD44high whereas <1% of CD4 cells were CD44high in RAG/AND mice. Other studies have relied on separation of naive cells with cell surface markers (CD45RB and CD62L); however, some activated or memory cells have been shown to reexpress the putative naive markers, so phenotype is not a completely reliable indicator for distinguishing or separating naive from Ag-experienced CD4 T cells (reviewed in 27 .
In aged AND mice, there are two rather clear populations, a naive TCRhigh population and a non-naive TCRlow population, shown in earlier studies to express endogenous receptors (36). The naive and memory status of the two populations is shown both by their cell surface phenotype and by their function. Since no TCR+ memory cells develop even after 2 years, it is likely that there are no common environmental Ags that are recognized by this particular receptor. When the AND mice are bred onto a RAG-2−/− background, the CD4 T cells in the resulting mice are ∼99% TCRhigh (see Fig. 1). This is in contrast to many other TCR transgenic animals in which the fraction of naive phenotype cells is less and the TCRhigh CD4 cells may get stimulated by Ag in the environment, leading to potential contamination with non-naive CD4 T cells.
Two recent reports (15, 18) have shown that the dose of Ag added in vitro to stimulate naive T cells played a role in development of Th1 and Th2 subsets. Both studies showed selective Th2 development (IL-4 secretion) at very low doses of Ag (≤0.01 μM) and Th1 development (IFN-γ secretion) at intermediate doses of Ag (0.1–10 μM). Th2 development was again seen at high doses of Ag (>10 μM) in one of the studies (18). It seemed possible that the reported effects of Ag were not mediated directly by Ag dose, but were mediated indirectly through cytokines produced by contaminating T cells or by APC populations, especially since exogenous IL-2 was used in one study (15). Therefore, we examined whether Ag dose had a direct effect on naive T cells in a transgenic model in which we could address the effects of exogenous IL-2 and contaminating non-naive (effector and memory) cells.
Effect of exogenous IL-2 on dose-dependent polarization (AND naive T cells)
We reasoned that addition of exogenous IL-2, as is routinely practiced in generation of effectors, would lead to potential Ag-independent outgrowth of contaminating non-naive CD4 T cells. Therefore, we compared the effect of IL-2 on generation of effectors from highly purified CD4 T cells derived from AND TCR transgenic mice. The cell populations obtained were 90 to 98% CD4+ and expressed high levels of Vβ3 (see Figure 1). For APC, we used a fibroblast cell line that was transfected with I-Ek and ICAM-1 and expresses B7-1 at high levels (30). This APC, called DCEK.ICAM, pulsed with optimum doses of PCCF is as efficient at stimulating naive AND CD4 T cells as the combination of plate-bound anti-CD3 and anti-CD28 (30) and is also highly efficient at stimulating effector and memory CD4 T cells (26, 37). The DCEK.ICAM line does not produce known polarizing cytokines such as IL-4, IFN-γ, or IL-12 (Ref. 38; and X. Zhang, L. Tsui, and S. L. Swain, unpublished data). This cell line also does not synthesize IL-6 (detected by RNase protection, P. Rogers and S. L. Swain, unpublished data). Moreover, we found that the addition of exogenous IL-6 (up to 10 ng/ml) had only a marginal effect on Th2 development (data not shown), increasing IL-4 secretion twofold with low dose Ag in the presence of exogenous IL-2. With higher doses of priming Ag, IL-6 had no effect on IL-4 secretion. As previously published (39), IL-6 did enhance T cell proliferation and long term cell survival, especially with high doses of priming Ag.
A wide range of doses of PCCF (0.001–100 μM plus no Ag) were added to cultures of purified AND CD4 T cells and DCEK.ICAM APCs. Parallel sets of cultures were set up with or without exogenous IL-2. After 4 days, cultures were harvested and viable CD4 cells were determined by trypan blue exclusion. Samples were readjusted to 3 × 105 T cells/ml for restimulation with a high dose of peptide (10 μM) plus DCEK.ICAM APCs. Supernatants were harvested after 24 or 48 h, and titers of IL-2, IL-4, and IFN-γ were determined. Figures 2 and 3 illustrate a set of results representative of those obtained in four such experiments. Results without IL-2 are also shown on a reduced scale (Fig. 3, top inset) so that the trends are obvious. Effector recovery (Fig. 2) and IL-2 secretion (Fig. 3) upon restimulation (at day 4) increased markedly with Ag dose (Fig. 2), but very few effectors were generated at doses below 0.01 μM PCCF. When effectors were generated in the presence of exogenous IL-2 (given on day 1), T cell recovery was dramatically increased.
Cytokine production by the recovered effectors changed with the dose of Ag added to initial culture and the pattern and quantity changed with addition of exogenous IL-2 (Fig. 3). When the recovered effectors were resuspended to a standard cell number and restimulated, the IL-2 secretion of effectors increased with dose in the absence of IL-2 (Fig. 3, top, inset). Little IL-4 or IFN-γ was produced by effectors generated without exogenous IL-2 regardless of the priming Ag dose (Fig. 3, middle and bottom), except for a little IFN-γ at high peptide dose. This pattern of cytokine production is rather like that of naive T cells (IL-2 without Th1 or Th2 characteristic cytokines). Surprisingly, when exogenous IL-2 was added, effector cytokine production increased greatly overall (up to 30-fold), and cytokines other than IL-2 were produced in substantial amounts. IFN-γ was made in greater amounts than IL-4. With IL-2 added to the priming culture, Ag doses below 0.001 μM led to effector populations that when restimulated, secreted high levels of IFN-γ, and significant levels of IL-4 (a Th0 pattern). IL-4 production decreased with increased Ag dose, whereas IFN-γ remained high, leading to a shift from a Th0 to Th1 pattern as the dose increased. High levels of IL-2 and detectable IL-4 and IFN-γ were produced even when no Ag was present. The fact that the Th0 pattern was also seen in IL-2-supplemented cultures when no PCCF was added suggested that under the conditions of low Ag dose, the few effectors present may not have been derived from resting naive CD4 T cells, but from IL-2-responsive cells. When exogenous IL-2 was present, there was also a gradual decrease in the ability of effectors to secrete IL-2 with increasing Ag doses.
In the absence of IL-2, the peak of IL-4 at lower doses of Ag was always missing (four of four experiments), and IFN-γ production was negligible (also four of four experiments), except at high doses of PCCF. It is notable that even when IL-4 was produced at low doses of Ag with exogenous IL-2, IFN-γ was also produced, so that the only “polarized” pattern was a predominant Th1 pattern seen at high dose with or without IL-2. This is consistent with many earlier reports, which found a Th1 default polarization in mice of B10 background (12, 15, 40).
Effector generation in RAG-2−/− TCR transgenic mice
The above analysis suggested that IL-2 alone could cause limited effector generation and polarization of some CD4 T cells present in cultures even without Ag, but also that generation of a Th0 or Th1 pattern from naive CD4 cells was influenced by the amount of IL-2 in the culture or the activation state of the T cell. Moreover, the level of expansion of responding cells and effector generation was also directly related to the levels of IL-2. With IL-2 playing so many roles, interpreting the mechanisms involved when it is limiting is difficult. The most likely source of non-naive CD4 cells in this and in other TCR transgenic models is cells that express endogenous TCR chains as well as the transgenic TCR chains (41). These cells probably have responded to other Ags in the environment, but may still have the capacity to respond to the test Ag (in this model, PCCF). To evaluate the dosage effects on the generation of effectors in the absence of such cells, we bred AND mice onto a RAG-2−/− background (RAG/AND) and repeated the analysis of the effects of PCCF dose on effector polarization, again analyzing the development of effectors in the presence and absence of exogenous IL-2.
CD4 T cells from RAG/AND mice were ≥98% Vβ3+ (Fig. 1) with homogeneously low expression of CD44 (Fig. 1) and high expression of CD62L (not shown), supporting a lack of contaminating non-naive cells. To minimize contributions from any contaminating APC, the purified RAG/AND CD4 T cells were cultured with DCEK.ICAM APC prepulsed with various doses of peptide, so that they were the sole APC. Two parallel sets of cultures were set up, one with and one without exogenous IL-2. As before, the cell recovery in effector cultures was highly PCCF dose dependent, with rapidly increasing cell recoveries from 0.1 to 30 μM PCCF in cultures both with and without IL-2 added (data not shown).
There were several differences between effectors generated from naive CD4 of RAG/AND vs AND mice (Fig. 4). In the absence of exogenous IL-2, effectors generated from CD4 precursors stimulated over a broad range of Ag doses of either RAG-2−/− (RAG/AND) or wild-type (AND) backgrounds developed no clear polarization to either Th1 or Th2 phenotypes (open symbols). Effectors did make substantial levels of IL-2, and there was an increase in IL-2 production and effector cell recovery with increasing dose (not shown). However, as before, we did detect small amounts of IFN-γ from AND T cells generated with very high doses of Ag. With the addition of exogenous IL-2 to the naive CD4 population, almost all doses of Ag induced a mixed (Th1 and Th2) or Th0 pattern in cells from both mice. However, effectors from CD4 T cells on the RAG-2−/− background consistently made more IL-4 and less IFN-γ (three of three experiments) than AND-derived T cells. Thus, the moderate Th1 polarization of effectors seen when AND naive CD4 were stimulated was lost in favor of a balanced IL-4 and IFN-γ production when the RAG-2−/− background was introduced.
Th2 cytokines are reduced at high doses of Ag
In both cases (RAG/AND + IL-2 and AND + IL-2), IL-4 secretion dropped in effectors generated with high doses of PCCF, whereas IFN-γ secretion remained high (Fig. 4). With AND CD4 precursors, this drop was seen when peptide doses >1 μM were used for effector generation, but with RAG/AND precursors, loss of IL-4 production was seen only at doses ≥10 μM. Thus, the shift away from IL-4 production at higher peptide dose may also depend, to some extent, on contaminating non-naive T cells.
RAG/AND precursors produce a Th0 pattern over a broad dose range
We further analyzed the effectors that could be generated in the presence and absence of IL-2 from RAG/AND precursors; results are shown in Figure 5. In the absence of exogenously added IL-2 (Fig. 5,a), the effectors generated from the naive RAG/AND cells gave results somewhat distinct from those seen in earlier experiments. First, without PCCF, the few “effectors” recovered did not secrete either IL-4 or IFN-γ. When even as little as 0.001 μM PCCF was added, the effectors generated produced small quantities of both IL-4 and IFN-γ. The dose response profiles for IL-4 and IFN-γ were remarkably similar to generation of effectors producing overall low quantities of both cytokines at all doses. Very high levels of IL-2 were produced (not shown, but see Fig. 6). Addition of IL-2 (Fig. 5 b) led to generation of effectors with much enhanced production of cytokines, but again the dose response curves were surprisingly flat, and ratios of IL-4 and IFN-γ did not vary significantly for PCCF for doses <1 μM. It should be noted again that the level of IL-4 was substantially higher that that noted previously for AND mice. As before, at the highest PCCF doses tested (1 and 10 μM), IL-4 production decreased (but only in cultures receiving exogenous IL-2).
To better understand the impact of these events and to illustrate the production of IL-5 and IL-2, Figure 6 shows normalized data in which the amount of cytokine production is multiplied by the recovery of CD4 T cells per culture, giving an estimate of the overall effect of Ag dose and addition of IL-2. It is clear that the major effect of increasing peptide density is an increase in the capacity of the cultures to produce all cytokines. A similar effect is seen for each cytokine at all doses in the absence of IL-2 and at all but the highest dose in the presence of IL-2. It is also clear that supplementation of the cultures with IL-2 results in a dramatic increase in the efficiency of effector generation. Peak IL-4 production by effectors went from 3 ng for those generated without exogenous IL-2 to nearly 100 ng with IL-2, while peak IFN-γ production rose from 15 to >1500 ng. IL-2 production by effectors is particularly high when RAG/AND precursors are used. Thus, in the absence of polarizing cytokines produced by contaminating non-naive cells or APCs, effectors with a mixed Th1 plus Th2 (Th0) phenotype develop, regardless of peptide dose, except at very high doses with IL-2, where production of Th2 characteristic cytokines, IL-4 and IL-5, drops suddenly, supporting a down-regulatory mechanism that merits further investigation. Of particular note is the lack of a peak of Th2 cytokine production at low Ag dose and the lack of a conversion to Th1 polarization except at very high peptide doses with RAG/AND precursors.
The development of populations of highly polarized Th1 and Th2 effector subsets from naive CD4 T cells can be achieved in vitro by providing high levels of TCR signaling with costimulation, the addition of IL-2, and the presence of polarizing cytokines (IL-12 or IFN-γ for Th1 and IL-4 for Th2). Recently, several groups have reported that the dose of Ag in vivo or in vitro can influence Th1 and Th2 effector polarization. Th2 polarization was favored by low Ag dose (15, 18), Th1 polarization by moderate to high doses (15, 18), and in one case, Th2 polarization at a very high dose (18). To analyze whether a similar pattern of polarization related to Ag dose would be achieved when non-naive CD4 contamination was minimized, we evaluated transgenic CD4 T cells from RAG-2−/− mice and T cells cultured without exogenous IL-2. Using TCR transgenic mice on a RAG-2−/− background (RAG/AND), we report that naive CD4 T cell populations develop into effectors with a Th0 cytokine pattern over a broad range of Ag doses. We also found that the addition of exogenous IL-2 promotes expansion of responding cells and development of effectors secreting high titers of cytokines. The pattern of cytokine production under these circumstances did not change, although at very high doses of Ag a Th1 phenotype was favored. We suggest that these data support the hypothesis that Ag dose does not selectively favor Th1 or Th2 development except in the presence of IL-2 and strong TCR signaling. In this case, a Th1 phenotype is favored.
Several factors led us to reexamine the mechanism of Ag dosage effects. First, the Th2 polarization seen at low Ag doses was found to be dependent on IL-4 when examined (18). Highly purified naive CD4 T cells stimulated with either polyclonal activators or with APC and Ag secrete IL-2 and little or no detectable IL-4 or IFN-γ (29, 30, 38), especially with low doses of Ag. Naive T cells are unable to produce significant quantities of cytokines other than IL-2 until 1 or 2 days after initial stimulation (22, 23, 42) and then only with repeated stimulation (22). Therefore, it is unlikely that at low Ag doses, responding naive T cells would be a probable source of IL-4 to induce autocrine Th2 polarization.
Second, maximal naive CD4 T cell response depends on high Ag dose and high levels of costimulation via more than one pathway (26, 30), while activated or memory CD4 can respond to lower doses of Ag and less costimulation (Ref. 26; and P.R.R. and S.L.S., manuscript in preparation). Thus at low Ag doses, the only cytokine production that would be expected is from Ag-experienced T cells or possibly other non-T cells.
Third, in the report by Constant et al., exogenous IL-2 was added to accomplish the polarization to Th2 at low Ag doses (15). IL-2 has many dramatic effects that could be required for development of effectors (the population expected to exhibit polarization), but there is also evidence that effector T cells (43), “resting” memory cells, and perhaps other non-T cells can directly proliferate in the presence of IL-2. Thus, at low Ag doses (and even with no Ag) the few non-naive T cells could potentially expand and contribute significantly either to the cytokine production upon restimulation or by secreting low levels of cytokines during culture. Indeed, the final effector recoveries at low Ag doses in the reported studies were very low. As the response of naive T cells increased with higher Ag doses, the naive response might well be expected to “swamp out” the response of contaminating cells.
In our study, the effects of IL-2 were examined over a broad range of Ag doses and with naive CD4 cells highly purified (Fig. 1) from both transgenic (AND) and RAG-2−/− transgenic (RAG/AND) mice. In the absence of exogenous IL-2, effector cell recoveries after 4 days were much lower than when IL-2 was added, especially with low doses of priming Ag (Fig. 2). Although low doses of Ag (0.001–1 μM) activate nearly all of the T cells, as assessed by early activation markers CD69 and IL-2Rα, at Ag doses <0.01 μM (.018 μg/ml), there is little IL-2 secretion, little or no T cell proliferation, and low T cell recoveries (44). Even with high doses of priming Ag, all of the IL-2 is consumed by day 3 (L. Haynes and S. L. Swain, unpublished observations). We analyzed T cell recoveries at day 4, rather than at day 6 or 7 as previously published (18), because we found that T cells consume their IL-2, especially at low doses, of Ag and start to undergo programmed cell death (apoptosis) by day 4 to 5 (10, 45). This was especially evident in cultures that received low doses of Ag and in cultures that received no exogenous IL-2, consistent with the hypothesis that IL-2 withdrawal is largely responsible for the cell death. We have also found that peak cytokine production from effector T cells generated after a single round of Ag stimulation occurs on days 3 through 5 (Ref. 43; and P. R. Rogers, X. Zhang, and S. L. Swain, unpublished observations). T cell recovery was much higher on day 4 compared with day 6, and the cells recovered were able to secrete much higher amounts of IL-2, IL-4, and IFN-γ (data not shown), suggesting that the “activation” state of the T cell or contributions (such as costimulation or cytokines) from the APC can influence cytokine secretion. In support of this theory, T cell recoveries with T-depleted spleen or activated B cell APCs were generally higher than with fibroblast APCs with peak cytokine production occurring on days 6 through 8 (data not shown).
IL-2 addition also has a second major impact, increasing cytokine secretion potential per cell or effector ( Figs. 3–6). In the absence of exogenous IL-2, low doses of Ag did not selectively induce generation of Th2 cells from either transgenic (Fig. 3) or transgenic/RAG-2−/− mice ( Figs. 4–6), but instead led to a population of recovered cells with little capacity for producing IL-4, IL-5, or IFN-γ. Thus, high levels of IL-2 seem to be required, directly or indirectly, for efficient effector development and polarization to either Th0, Th1 (IFN-γ), or Th2 (IL-4) subsets in addition to the role of IL-2 in cell recovery. It should be noted in evaluating these results that the cytokine production shown was from an equal number of cells collected at each Ag dose ( Figs. 3–5), and with the very low cell recoveries at low doses of Ag (especially in the absence of exogenous IL-2), IL-4/IFN-γ cytokine production per initial cultured cell is extremely low (as shown in Fig. 6 with normalized data).
Another important point is that in the total absence of Ag, adding IL-2 to AND transgenics led to the recovery of a population of T cells that could produce cytokines upon restimulation (see especially the results from Fig. 3). The level of non-Ag-driven effector generation was less in RAG-2−/− background experiments (Fig. 5). The induction of IFN-γ-secreting T cells in the absence of Ag but in the presence of exogenous IL-2 may be influenced by the presence of NK cells in the T cell preparation; however, we believe that this is unlikely. First, T cells, which were purified over Percoll to obtain small resting cells, contained no detectable NK1.1+ cells (data not shown). Second, depletion of NK1.1+ cells with Ab and complement or the use of T-depleted spleen APCs (which contain a small percentage of NK cells) did not significantly alter the results shown in Figure 3. Thus, we believe that it is the addition of exogenous IL-2 and the use of highly costimulatory APCs that induce a small proportion of T cells to survive and differentiate into IFN-γ (and IL-4)-secreting cells in the absence of Ag. These observations provide support for the concept that IL-2-induced outgrowth of non-naive cells may be occurring in TCR transgenic models and could be responsible for the polarization detected in some studies. However, interpretation of the requirement for exogenous IL-2 is difficult because, although it is indeed likely that it contributes to the outgrowth of cells other than naive CD4, it also strongly promotes effector development as well as expansion, and thus it cannot readily be removed from initial cultures.
In young TCR transgenic mice, a majority of CD4 T cells are naive (90–95%) (46), but with increasing age there is an increase in the fraction of T cells that express a memory phenotype (36). There is also increased expression of endogenous TCR α- and β-chains along with reduced expression of the transgenic TCR, both in the AND TCR transgenic and in other TCR transgenic mice (36, 41, 47). These non-naive CD4 T have indeed been shown to bias naive T cell responses and subset polarization (24, 25). To reduce the possibility of contaminating non-naive or nontransgenic cells, we tested the effects of Ag dose on T cells derived from TCR transgenic mice on a RAG-2−/− background. CD4 cells from RAG/AND mice have a naive phenotype, and there is a reduction from 14 to 1% in CD4+Tglow or Tg− cells (before purification) and a reduction from 5 to <1% in CD4+CD44high cells (after purification) compared with AND transgenic cells on the conventional (RAG-2+/+) background (Fig. 1).
In the absence of exogenous IL-2, T cells from mice on the RAG-2−/− background developed a weak Th0 phenotype with secretion of low amounts of both IL-4 and IFN-γ, regardless of Ag dose (Figs. 4 and 5). There was no clear effect of Ag dose on Th1 or Th2 development in T cells from RAG/AND mice and only a weak Th1 polarization at high Ag doses in T cells from AND mice. This is not a classic Th1 phenotype, which is associated with high levels of IFN-γ production, but instead may represent a partial or suboptimal Th1 polarization. When results are normalized to cell recovery, the effects of increasing Ag dose on increasing production of cytokine-secreting effectors is particularly obvious (Fig. 6).
In the presence of exogenous IL-2, low doses of priming Ag again induced a Th0 phenotype with much higher levels of IL-2, IL-4, IL-5, and IFN-γ secretion. When the amount of cytokine was adjusted for the total number of T cells recovered, the results showed that there was a dose-dependent increase in IL-2, IL-4, IL-5, and IFN-γ secretion but again no clear effect of Ag dose on Th1 or Th2 polarization (Fig. 6). T cells from mice on the RAG-2−/− background (RAG/AND) secreted considerably more IL-4 and less IFN-γ than mice on the RAG-2+/+ background (AND) (Fig. 4), especially when effectors were generated in the presence of exogenous IL-2.
As Ag dose increased >0.01 μM, effectors generated from AND mice in the presence of exogenous IL-2 gradually lost their ability to secrete IL-4 (Fig. 2) (and IL-5, data not shown). Effectors generated from RAG/AND mice produced IL-4 over a much broader range of Ag doses with the fall in ability to produce IL-4 and IL-5 observed only at high doses (>1 μM) (Figs. 5 and 6). We believe that this drop in the level of IL-4 and IL-5 secretion is not due solely to cytokine uptake but rather to a reduced ability of T cells to secrete these cytokines, thus resulting in a preferential Th1 phenotype. Although we concede that there may be some IL-4 uptake by T cells, it is not likely that such high amounts of IL-4 and IL-5 can be absorbed by the T cells or APCs, especially since IL-5R have not been described on T cells. In addition, upon extended culture (using T-depleted spleen as the initial APC), high doses of priming Ag still resulted in suppressed levels of IL-4 and IL-5 upon restimulation (data not shown). Thus, we believe that the reduced ability to secrete IL-4 and IL-5 is related to the ability of high Ag dose and IL-2 to suppress Th2 generation. Additional experiments using enzyme-linked immunospot (ELISPOT) and mRNA analysis will confirm these observations.
In contrast to our results, Hosken et al. (18) reported Th2 development with high doses of Ag (3–100 μM) (18). The difference in Th1 vs Th2 development at high Ag doses could also potentially be influenced by the strain of mice, culture conditions, purity of naive T cells, and/or the TCR affinity for the peptide/MHC. The Hosken et al. study used mice on the BALB/c background in which the percentage of transgene-expressing CD4+ cells is only ∼80%. Moreover, it is known that the BALB/c strain has a predilection for production of Th2 effectors (12). Although cells were sorted with CD4 and CD62L, contamination with non-naive T cells could be at higher levels (since CD62L expression may not be a confined to naive T cells). In addition, Hosken et al. used lightly irradiated dendritic cells or activated B cells to present an OVA peptide to CD4 and L-selectin-sorted T cells for a 6-day culture period. These culture conditions raise two other possibilities for the observations in this model. First, it is possible that APC interacting with CD4 cells can produce polarizing cytokines. Different APC preparations may thus influence the polarization of developing effectors. By using a fibroblast cell line as APC, we have minimized the (potential polarizing) cytokine contribution from APCs. Another potential difference that could affect differentiation in these systems is the differential expression of costimulatory molecules, particularly B7 (CD80 and CD86), on the APC. The differences seen in our study vs Hosken’s study, however, are not likely to stem from differential expression of B7-1 or B7-2 on the APC because T-depleted spleen and LPS-activated B cell APCs, which express high levels of B7-2 and low levels of B7-1, gave results similar (data not shown) to those using the B7-1+ fibroblast APCs (Fig. 5 a). Second, our own recent studies indicate loss of Th1 (but not Th2) effectors by programmed cell death and activation-induced cell death (45), especially between day 4 and 6 of culture. If a mixed population were present, culture after 4 days could lead to selection of Th2 cells, and in a model system in which Th1 and Th2 effectors are mixed, does indeed lead to Th2 predominance (X. Zhang et al., unpublished observations). Finally, the systems may differ by virtue of different avidities between the TCR and its ligand (48). It will be important to determine whether peptide affinity and APC effects (i.e., costimulation or cytokine secretion) are involved in other models in which Ag dose (8, 48) and peptide analogues with differing avidities (21, 49, 50) have been reported to determine CD4 effector polarization.
We thank Michael Croft for helpful discussions and DeShon Hall and Edda Roberts for technical support.
This work was supported by National Institutes of Health Grants AI26887 and AI37935.
Abbreviations used in this paper: PCCF, pigeon cytochrome c fragment 88–104; AND, Vα11/Vβ3 TCR transgenic mice; RAG/AND, AND TCR transgenic mice on a RAG-2−/− background; DCEK.ICAM, B7-1+/ICAM-1+/I-Ek-expressing mouse fibroblast line; PE, phycoerythrin.