First, we established the overall cytokine profile of the PLP: 139-151-specific CD4 cells 7 days after immunizing with this peptide. CD4 cells were purified from the dLN of the immunized mice and were tested on naive LN cells functioning as APC. As shown in Figure 1, these CD4 cells produced IFN-γ, IL-2, IL-3 and TNF-α; the frequency of the cells that secreted these cytokines was in the range 50–300 per 3 × 10⁵ CD4 cells plated per well. Confirming our data obtained with other antigen systems [35,36] we found that the number of the nominal antigen (here PLP:139-151)-induced cytokine spots was a linear function of the number of primed CD4 cells plated, and that naive or control immunized mice did not produce these cytokines when challenged with the PLP:139-151 peptide (<1:1,000,000, data not shown). In vivo-primed PLP: 139-151-specific CD4 memory cells were detected, therefore in these assays, in the absence of bystander cytokine production by other cells. The PLP:139-151-primed CD4 cells produced IL-5 and IL-13 in frequencies <1:300,000 (Figure 1) and IL-4 producing CD4 cells were in the 10 per 3 × 10⁵ range (1:30,000): that is, the IL-4 producing CD4 cells were outnumbered more than 20 fold by IL-2 and IFN-γ -producing CD4 cells (both of which occurred in frequencies ~1:1,300). Essentially identical results were obtained when unseparated dLN cells were tested, or when purified CD4 cells of PLP: 139-151-immunized mice were studied with naive spleen cells or with thymocytes functioning as APC. It is important to note, however, that the use of LN (or thymic APC) was found to be critical for revealing the cognate IL-4 signature of CD4 cells. When splenic APC were used for the same type of experiments, the induction of vigorous antigen-specific IL-4 production was seen (Table 1). However, this IL-4 was not produced by T cells: it was an IL-3-driven bystander reaction by non-T, non-B cells, most of which were basophils [37].

This cytokine profile of the CD4 memory cell pool established in single color ELISPOT assays classifies it as perfectly matching the Th1 definition. Therefore, Figure 1 sets the basis for dissecting the Type 1 cytokine profile at the single cell level while studying cytokine co-expression patterns using the dual color ELISPOT methodology.

These data show that in the inductive phase of the autoimmune response, the peripherally primed PLP: 139-151-specific CD4 memory cell pool is highly type 1 polarized. Does this overall type 1 cytokine signature reflect the secretory activity of classic Th1 cells that at the single cell level co-express these cytokines?

We performed dual color ELISPOT assays to study the cytokine co-expression by PLP: 139-151-secific CD4 cells. The data obtained for different time points after the immunization, and testing different organs are summarized in Table 1; examples of dual color assay results are shown in Figure 2. In total, we analyzed the cytokine signature of ~200,000 individual peptide-specific CD4 cells 7 days after immunization. The results showed that only 4.6% of these cells co-expressed IFN-γ and IL-2. This number also includes cells that would have switched from the production of one cytokine to the other during the 48 h assay duration, as those too appear double-positive. The vast majority (95.4%) of the peptide-specific CD4 cells were single-positive, producing either IFN-γ or IL-2.

The frequency of CD4 cells that co-expressed IFN-γ and IL-3 was in the 3% range, while essentially no cells (<0.1%) were detected to co-express IFN-γ and IL-4. The absence of IL-5 producing cells (<1:1,000,000) in the face of the relatively high frequency IFN-γ producing CD4 cells (~1:1,300), and the lack of the co-expression of IL-4 and of IFN-γ argue against the stochastic regulation of these cytokines. The co-expression of IL-2, IL-3, and of IFN-γ in the 2%–5% frequency range argues against both, coordinate expression of the type 1 cytokines in Th1 cells, or their stochastic regulation in CD4 cells 7 days after the immunization.

The first cytokine-producing CD4 memory cells became detectable on day 2 after immunization (Table 1). These cells produced either IL-2, IL-3, IFN-γ, IL-4, but no IL-5. The frequency of PLP: 139-151-specific CD4 cells that co-expressed IL-2, IL-3 and IFN-γ was between 8 and 0%. As far as IL-4 and IFN-γ producing cells are concerned, these could be detected initially only in the dLN and both occurred in low frequency (<1:30,000). Subsequently, the frequency of IFN-γ producing cells increased while those producing IL-4 stayed in the same low range. The production of IL-4 and IFN-γ at all time points was fully dissociated at the level of individual T cells. The PLP:139-151 specific T cells did not, therefore, go through an extended Th0 state, but directly acquired the single-cytokine-producing phenotype.

Immunization of SJL mice with the PLP: 139-151 peptide results in a chronic relapsing form of EAE from which the mice do not recover [32]. We tested mononuclear cells isolated from the spinal cords of such mice. While no PLP:139-151-specific ELISPOT formation was seen in the CNS isolates tested at time points prior to the clinical onset of EAE (days 2, 3 and 7 in Table 1), cytokine-producing cells became detectable at the onset of paralysis. The PLP: 139-151-induced cytokine production by the cells separated from the CNS was blocked by anti-CD4 mAbs, affirming that these cytokines were CD4 T cell derived (data not shown). Unlike the results obtained from the dLN and spleen, where the autoantigen is not presented and where no spontaneous cytokine production was seen in the absence of added peptide, “constitutive” (high background) cytokine production was seen in the cells isolated from the spinal cord (Figure 2 & Figure 3), most likely resulting from the presence of endogenous PLP. These in situ activated T cells also showed dissociated production of IFN-γ, IL-2, and IL-3 (Figure 3), and did not secrete IL-4 and IL-5 (data not shown). The addition of PLP: 139-151 induced a ~10 times higher frequency of cognate IFN-γ, IL-2, IL-3, but no IL-4 - (Table 1) and IL-5 - (not shown) - producing cells. When the frequency of cells was studied that co-expressed any combination of these cytokines during the first, second, and third paralysis, it stayed constantly <6%, for the cells isolated from the central nervous system (CNS) itself and for those in the immune periphery alike (Table 1). This outcome was independent of the type of APC used (data not shown, and see below). The data suggest that dissociated cytokine expression does not readily break down and polyfunctional T cells do not arise under these conditions of chronic autoimmune pathology. Comparing the cytokine signatures during phases of remissions and exacerbations of the disease, we did not find qualitative differences in cytokine co-expression (data not shown).

Cytokine expression has been linked to the antigen dose, that is, to the signal strength generated by the extent of TCR ligations [38,39,40]. We therefore studied the PLP:139-151 peptide-dose-dependency of cytokine production and co-expression in the dLN, spleen and the CNS. The number of IL-2, IL-3, and IFN-γ spots induced vs. the antigen concentration resulted in sigmoidal dose-response curves (Figure 4). While plateau values were reached for all the cytokines tested, the absolute number of spots at maximally stimulatory peptide doses differed for the individual cytokines suggesting different frequencies of cells capable of producing each cytokine. However, the peptide concentrations at which 50% maximal stimulation was reached (Kf value, see Experimental Section 4.4), was essentially identical for IL-2, IL-3, and for IFN-γ (Figure 4 & Table 2). The signal strength required to engage each of these cytokines was, therefore, essentially the same. No IL-4 or IL-5 was induced at any peptide concentration (data not shown). As far as the co-expression of IL-2 and IFN-γ or of IL-3 and IFN-γ was concerned, the frequency of double positive cells was found to be <8% for the full range of peptide concentration studied (Figure 4). This result reproduced when T cells were tested on days 7, 12, 21 and 55 after the immunization, and applied for dLN, spleen and the CNS. Similar Kf values were obtained when cells isolated from CNS were tested in the presence of the naive syngeneic splenocytes (Table 2) and also the numbers of cytokine co-expressing cells were unaffected by the addition of APC from different sources (LN, spleen, thymus - data not shown). The dissociated production of IL-2, IL-3, and IFN-γ was therefore not a function of the signal strength but seemed to reflect a stable feature of the respective T cells.

The antigen-specific T cell repertoire is comprised of T cell clones with different avidities for antigen. Studying the CD4 cell response to myelin basic protein (MBP) we found, for example, that T cells that matured in a mouse that did not express MBP were stimulated in average by a 10,000–fold lower MBP dose that T cells that matured in a mouse expressing MBP. The former qualify as high avidity T cells, the latter, as low avidity T cells [41]. Measuring functional avidity gives insights into conceivable repertoire selection processes in the course of EAE, such as the retention of high affinity clones in the CNS or the selective expansion/deletion of this subpopulation in the course of the disease. Moreover, if the different cytokine producing T cells represent different lineages, they might be differently affected by such selection processes. As the data in Table 2 showed remarkable stability of Kf values between anatomic sites, we found no evidence for the retention of high affinity PLP: 139-151-specific cells in the CNS vs. the immune periphery (in which case higher peptide doses would be required to stimulate the T cells in the periphery vs. the CNS). Also, we found no evidence that over time a selective expansion or deletion of the high affinity PLP: 139-151-specific cells would occur (the unchanged Kf values from day 7 to day 55 in Table 2; if this were the case, then the avidity curves of the T cells would have shifted over time toward a lower or higher peptide range). While repertoire selections have been reported for some T cell responses against foreign antigens [42,43], we found no evidence for them in the autoimmune setting of this EAE model.

Figure 5 summarizes the results of five independent experiments in which the frequency of the IFN-γ-and IL-2 (single-cytokine)-producing cells was measured in different anatomic compartments (dLN, spleen and the CNS) over the course of EAE. For each time point, and in each experiment, four mice were tested individually. IFN-γ producing cells appeared first in dLN (where the initial clonal expansion and differentiation occurs after the s.c. immunization), on day 3 where their numbers peaked on day 7 (Figure 5A, and Table 1). With a delay of 2 days, they also appeared in the spleen where their numbers peaked between days 7–12. No IFN-γ producing cells were detected in the CNS until day 12, before the onset of clinical EAE. At the onset of paralysis, ~10 fold higher frequencies were reached in the CNS compared with the peripheral lymphoid tissues. This sudden increase of numbers of IFN-γ producing PLP: 139-151 specific cells in the CNS was accompanied by an instant drop in their frequency in the dLN, but not in the spleen (day 12, Figure 5). Apparently, the IFN-γ producing memory/effector cells generated in the dLN provided the source of cells that migrated to the spleen and the CNS; the cells that migrated to the spleen might have become there temporarily sessile, not participating in the first wave of effector cell migration to the target organ. The more profound subsequent decline in numbers of IFN-γ producing cells in the CNS vs. those in the spleen might also argue for this view.

The frequencies of IL-2 producing cells followed a different pattern. On day 2, IL-2 producers were the first to appear in the dLN, where they were the prevalent cell population throughout the observation period (Figure 5). IL-2 producers appeared with a delay of 2 days in the spleen and subsequently their frequencies did not change in parallel with those CD4 cells producing IFN-γ: the IL-2 producers reached maximal-plateau numbers by day 7 in the spleen, and unlike IFN-γ producers did not undergo rapid frequency changes thereafter. IL-2 producing cells also infiltrated CNS during the onset of the first paralysis, but in contrast to the abundance of the IL-2 producers in the immune periphery, they constituted a minority in the CNS as compared with those secreting IFN-γ. We verified these frequencies by testing the CNS-derived T cells on splenic APC layers, as it was suggested previously that the induction of IL-2 production has more stringent requirements for costimulation than that of IFN-γ [40]. However, as stated before, the frequencies of cytokine producing cells were essentially unaffected by adding APC. The lower frequency of IL-2 spots detected in the CNS, therefore, resulted from the deficit of such cells within the CNS-infiltrate.

The data are consistent with IFN-γ and IL-2 single-cytokine producing cells representing independent subpopulations of memory/effector cells that each follows different migration patterns. The migration pattern of IL-3 producing cells closely followed that of the IL-2 producers (Figure 5).

SJL/J mice were purchased from The Jackson Laboratories and bred at CWRU under specific-pathogen-free conditions. Female mice were immunized at 6–8 wk of age. PLP peptide 139–151 [32] was purchased from Princeton Biomolecules (Columbus, OH, USA). IFA was purchased from Gibco BRL, (Grand Island, NY, USA) and CFA was prepared by mixing inactivated M. tuberculosis H37RA (Difco Laboratories, Detroit, MI, USA) at 1 mg/mL into IFA. PLP:139-151 peptide (2 mg/mL) and PBS were mixed 1:1 with adjuvant and was injected once in 100 μL (100 μg/animal), subcutaneously (s.c.). Pertussis toxin (0.2 μg) was injected twice, once immediately and once 24 h after immunization [32,51].

Cells from the CNS were prepared as follows. After sacrificing, the animals were perfused with PBS, the spinal cords were removed from the entire vertebral column and placed into DMEM medium. The spinal cord was disrupted with the back of a syringe. The resulting cell suspension was filtered through a Falcon Cell Strainer 2350 (Becton Dickinson, San Jose, CA, USA). The cells were washed twice with DMEM and subsequently counted. As we found in our previous work, infiltrating mononuclear cells isolated from spinal cords of mice with EAE contained 19%–43% CD4 cells. We also demonstrated that for functional T cell assays, there is no need for further ficoll/percoll purification of mononuclear cells, and for addition of APC [52,53]. Single cell suspensions from draining and non-draining LN and the spleen were prepared as previously described [6]. CD4 cell fractions were obtained with >97% purity by passing single cell suspensions over ‘Mouse CD4 Subset Columns’ (R & D Systems, Minneapolis, MN, USA) according to the manufacturer’s instructions.

Plates (MSIPN4W, Millipore, Billerica, MA) were coated overnight at 4 °C with the cytokine-specific capture antibodies specified below. The plates were then blocked with 1% BSA in PBS for 1 h at room temperature and washed 4× with PBS. Subsequently, irradiated (3,000 rad) LN or splenic APC from naive, syngeneic mice were added (5 × 10⁵ cells/well, or as specified). Freshly isolated CD4 or single-cell suspensions from spinal cords were plated in serial dilution in 2–4 replicate wells with or without the nominal antigen or control antigens. The assay medium was serum-free HL-1 (Biowhittaker, Walkersville, MD.) supplemented with 1 mM L-glutamine. After 48 h of cell culture in the incubator at 37 °C, the cells were removed by washing 3 × with PBS and then 4 × with PBS containing 0.05% Tween-20 (PBST). The two detection antibodies were added simultaneously and the plates were incubated at 4 °C overnight, after which they were washed 3 × with PBST. For the biotinylated detection mAbs, the streptavidin-ALPH (alkaline phosphatase) conjugate (Dako Corp., Carpenteria, CA, USA) was added (at 1:2000 dilution), incubated for 2 h at room temperature and removed by washing twice with PBST and twice with PBS. The Vector Blue ALPH substrate (Vector Laboratories, Burlingame, CA, USA) was added for 15–30 min. Then, after washing twice with PBS, the AEC substrate (Pierce, Rockford, IL, USA) was added and left for 20–40 min. The following coating mAbs were used for IL-2, IL-3, IL-4, IL-5, and IFN-γ: JES6-1A12 (5 μg/mL), MP2-8F8 (5 μg/mL), BVD4-1D11 (2 μg/mL), TRFK5 (5 μg/mL), and R46A2 (2.5 μg/mL). The combinations of detection antibodies for the IL-2:IFN-γ, IL-3:IFN-γ, IL-4:IL-5, IFN-γ:IL-5, IL-4:IFN-γ, and IL-2:IL-5 assays were: JES6-5H4-biotin:XMG1.2-HRP, MP2-43D11-biotin:XMG1.2-HRP, BVD4-24G2-biotin: TRFK4-HRP, XMG1.2-biotin:TRFK4-HRP, BVD4-24G2-biotin:XMG1.2-HRP, and JES6-5H4-biotin:TRFK4-HRP (all antibodies were from Pharmingen, San Diego, CA). Horse radish peroxidase (HRP)-labeling of antibodies was performed by the periodate method [54]. The combination of JES6-5H4-biotin or MP2-43D11-biotin detection mAbs with unlabeled affinity-purified goat polyclonal antibodies to mouse IFN-γ (R&D Systems, Minneapolis, MN) followed by the mixture of streptavidin-ALPH with HRP-labeled rabbit-anti-goat antibodies (Zymed, San Francisco, CA ) were also used for IL-2:IFN-γ, IL-3:IFN-γ two-color ELISPOT. This variant gives the same sensitivity for the detection of single and double-cytokine producing cells, and permits use of off-the-shelf reagents if HRP labeling cannot be performed. The detection antibody concentrations were as follows: JES6-5H4-biotin (2 μg/mL), MP2-43D11-biotin (2 μg/mL), BVD4-24G2-biotin (2.5 μg/mL), TRFK4-HRP (2 μg/mL), XMG1.2-HRP (2 μg/mL), XMG1.2-biotin (2 μg/mL), polyclonal goat-anti-mouse IFN-γ (0.5 μg/mL). Computer-Assisted Two-Color ELISPOT Image Analysis was performed as previously described [6] using a Series 6 ImmunoSpot Image Analyzer (CTL Analyzers LLC, Cleveland OH).

T cell cytokine dose-response curves obtained by ELISPOT are closely approximated by a sigmoidal function in the Log scale, or a hyperbolic function on a linear peptide scale, asymptotically reaching a plateau at high concentrations of the peptide (see Figure 4). Based on the similarities between receptor-ligand binding isotherms and dose response curves of T cell function, functional avidity (Kf) was defined as the concentration of peptide that leads to the activation of 50% maximal number of T cells [55]. As defined, functional avidity represents the integrated characteristic of the entire heterogeneous T cell population. Kf values were calculated from the ELISPOT dose-response curves in Log scale by approximation with sigmoidal function: Y = a/(1 + exp(-(X + b)/c)), where ‘X’ is a log₁₀ of the recall peptide dose and ‘Y’ is the number of spots detected at each peptide concentration. Furthermore, ‘a’,’b’, and ‘c’ are the parameters of non-linear regression, ‘b’ being equal to the Lg (Kf) and ‘a’ being the asymptotic limit for the maximal number of cells capable of producing cytokines in the T cell population tested.