GM-CSF-neuroantigen fusion proteins reverse experimental autoimmune encephalomyelitis and mediate tolerogenic activity in adjuvant-primed environments: association with inflammation-dependent, inhibitory antigen presentation

Abstract

Single-chain fusion proteins comprised of GM-CSF and neuroantigen (NAg) are potent, NAg-specific inhibitors of experimental autoimmune encephalomyelitis (EAE). An important question was whether GMCSF-NAg tolerogenic vaccines retained inhibitory activity within inflammatory environments or were contingent upon steady-state conditions. A GMCSF-MOG fusion protein reversed established paralytic disease in both passive and active models of EAE in C57BL/6 mice. The fusion protein also reversed EAE in CD4-deficient and B cell-deficient mice. Notably, GMCSF-MOG inhibited EAE when co-injected adjacent to the MOG35-55/CFA emulsion. GMCSF-MOG also retained dominant inhibitory activity when directly emulsified with MOG35-55 in the CFA emulsion in both C57BL/6 or B cell-deficient models of EAE. Likewise, when combined with PLP139-151 in CFA, GMCSF-PLP inhibited EAE in SJL mice. When deliberately emulsified in CFA with the NAg, GMCSF-NAg inhibited EAE even though NAg was present at more than a 30-fold molar excess. In vitro studies revealed that the GMCSF domain of GMCSF-MOG stimulated growth and differentiation of inflammatory dendritic cells (DC) and simultaneously targeted the MOG35-55 domain for enhanced presentation by these DC. These inflammatory DC presented MOG35-55 to MOG-specific T cells by an inhibitory mechanism that was mediated in part by IFN-γ signaling and NO production. In conclusion, GMCSF-NAg was tolerogenic in CFA-primed pro-inflammatory environments by a mechanism associated with targeted antigen presentation by inflammatory DC and an inhibitory IFN-γ/ NO pathway. The inhibitory activity of GMCSF-NAg in CFA-primed lymphatics distinguishes GMCSF-NAg fusion proteins as a unique class of inflammation-dependent tolerogens that are mechanistically distinct from naked peptide or protein-based tolerogens.

Introduction

Experimental autoimmune encephalomyelitis (EAE) is a widely used model to advance novel therapeutics for multiple sclerosis (MS)(1–3).Many therapeutic drugs for MS in practice or on the near-term horizon represent broad-spectrum immunosuppressive drugs. These drugs typically have clinical efficacy in proportion to the severity of adverse side-effects, particularly in that many efficacious drugs for MS have serious consequences stemming from a compromised immune system and susceptibility to opportunistic infections. The quest for safe and efficacious disease-specific MS drugs has spurred development of antigen-specific vaccines to induce myelin-specific immune tolerance(4-7). Antigen-specific tolerogenic vaccines entail the use of myelin self-peptides or proteins to elicit tolerance to relevant encephalitogenic epitopes. Tolerogenic vaccines are largely based on the concept that myelin-specific tolerance is contingent upon introduction of myelin peptides into a quiescent steady-state environment where presentation of self-antigen occurs without adequate costimulatory signals(8–12). The concept is that T-helper cell recognition of myelin peptides on MHC class II (MHCII) glycoproteins without sufficient costimulatory signaling causes T cell anergy and primes tolerance to those myelin epitopes. Tolerogenic vaccination is designed to elicit an inhibitory immunological memory specific for myelin-based vaccine antigens as a consequence of depleting the myelin-specific T cell repertoire or induction of myelin-specific regulatory T cells, among other inhibitory mechanisms. Tolerogenic vaccination however has potential drawbacks. In particular, the requirement for quiescent steady-state environments will often be rife with uncertaintyin many clinical settings. Administration of tolerogenic vaccines into environments with smoldering inflammatory processes may be problematic in the context of chronic autoimmune disease, given that introduction of myelin antigens into inflammatory environments may result in immunogenic rather than tolerogenic vaccination.

Previous studies have focused on cytokine- NAg fusion proteins as a new class of myelin-specific tolerogens(13–20). Fusion proteins comprised of GM-CSF as the N-terminal domain and a myelin epitope as the C-terminal domain represent highly effective vaccine tolerogens. A rat GMCSF-NAg fusion protein that included the 69–87 epitope of myelin basic protein was tested in the Lewis rat model of EAE(16). The GM-CSF domain of the fusion protein targeted the NAg domain to myeloid APC to engender an approximate 1000-fold enhancement of antigenic potency by a mechanism that was blocked by free GM-CSF and that depended upon physicallinkage of the cytokine and NAg domains. This GMCSF-NAg fusion protein was active as a prophylactic in that pretreatment with GMCSF-NAg inhibited the subsequent induction of EAE. Likewise, GMCSF-NAg was an effective therapeutic intervention that halted progression of EAE when treatment was initiated after disease onset. As both a prophylactic and intervention, GMCSF-NAg required physical linkage of the cytokine and NAg domains for inhibitory efficacy. Murine GMCSF-NAg fusion proteins that included the 35–55 epitope of MOG or the 139–151 epitope of proteolipid protein (PLP) were effective interventions in the C57BL/6 and SJL murine models of EAE, respectively(14, 15). These observations presented a paradox in that GM-CSF is a prototypic pro-inflammatory cytokine implicated in the effect or phase of EAE(21), and the NAg represented the key instigator and encephalitogenic antigen of EAE(22). Yet when engineered as a single-chain fusion protein, the combined cytokine and NAg domains exhibited potent NAg-specific tolerogenic activity in multiple rodent models of EAE.

The purpose of this study was to assess whether these vaccines required a quiescent steady-state environment for tolerance induction or whether GMCSF-NAg vaccines retained inhibitory activities that were manifest directly within inflammatory environments. This study provided evidence that GMCSF-NAg was an effective intervention in the EAE effector phase in wild type, CD4-deficient, and B cell-deficient mice. GMCSF-MOG in saline also had potent inhibitory activity when injected side-by-side with a MOG35–55/ CFA encephalitogenic emulsion. Importantly, in C57BL/6 mice, B cell-deficient mice, and SJL mice, GMCSF-NAg inhibited disease when directly incorporated into the NAg/ CFA emulsion. GMCSF-NAg elicited differentiation of inflammatory DC in vitro and targeted the NAg domain to those DC for enhanced presentation of the NAg epitope on MHCII glycoproteins. After culture with GMCSF-MOG, inflammatory DC presented antigen to MHCII-restricted MOG-specific responders by an inhibitory mechanism that in part depended on IFN-γ signaling and NO production. In accordance, when mixed with MOG35-55 in the emulsion, GMCSF-NAg exhibited partial inhibitory activity in Ifngr1^−/− mice that were deficient in IFN-γ-induced NO production. In conclusion, GMCSF-NAg induced the development of inflammatory DC APC and elicited tolerance in the context of an inflammatory environment. These findings distinguish GMCSF-NAg fusion proteins as a unique class of antigen-specific tolerogen that inhibits encephalitogenic T cells in the context of an inflammatory environment.

Materials and Methods

Animals and reagents

C57BL/6, SJL, B cell deficient (B6.129S2-Ighm^tm1Cgn/J; stock number 002288), CD4-deficient (B6.129S2-Cd4^tm1Mak/J; stock number 002663), IFN-γR1-deficient (B6.129S7-Ifngr1^tm1Agt/J; stock number 003288), and 2D2 (C57BL/6-Tg(Tcra2D2, Tcrb2D2, 1Kuch/J), stock number 006912)mouse strains were obtained from the Jackson Laboratory (Bar Harbor, ME) and were bred and housed in the Department of Comparative Medicine at East Carolina University Brody School of Medicine. Animal care and use was performed in accordance with approved animal use protocols and guidelines of the East Carolina University Institutional Animal Care and Use Committee. Synthetic MOG35-55 and PLP139-151 peptides were obtained from University of North Carolina Microprotein Sequencing & Peptide Synthesis Facility (Chapel Hill, NC). Derivation, expression, purification, and bioassay of the murine GM-CSF fusion proteins (GM-CSF, GMCSF-MOG, and GMCSF-PLP) were described in previous studies(14, 15). Murine GM-CSF contained a C-terminal 8-histidine affinity tag without an intervening linker. GMCSF-MOG and GMCSF-PLP included the MOG35-55 amino acid sequence “M-E-V-G-W-Y-R-S-P-F-S-R-V-V-H-L-Y-R-N-G-K” or the PLP139-151 sequence “H-S-L-G-K-W-L-G-H-P-D-K-F” respectively between the N-terminal GM-CSF domain and the 8-histidine C-terminus.

Cell lines and culture conditions

Cell lines were cultured in complete RPMI medium [10% heat-inactivated fetal bovine serum, 2 mM glutamine, 100 μg/ ml streptomycin, 100 U/ ml penicillin (Whittaker Bioproducts, Walkersville, MD), 50 μM 2-ME (Fisher Scientific, Pittsburgh, PA)]. To establish lines of continuous MOG-specific 2D2 T cells, 2D2 splenocytes were activated with 1 μM MOG35-55 for 3 days. T cells were then continuously propagated in murine IL-7 (baculovirus supernatant, 1% v/v) for 1-3 weeks with passage every 3–4 days. Periodically, the 2D2 T cells were reactivated with 1 μM MOG35-55 and irradiated C57BL/6 splenocytes in complete RPMI for 3 days, followed by continuous propagation in IL-7.

Antigen presentation assays by bone marrow (BM)-derived DC

BM cells from C57BL/6 mice were cultured at 1x10⁶cells/ ml in complete RPMI with (a) 1 μM GMCSF-MOG, (b) a combination of 1 μM GM-CSF plus 1 μM MOG35-55, (c) 1 μM GM-CSF, (d) 1 μM MOG35-55, or (e) saline. These cells were cultured for 4 days and then assayed for MOG-specific presentation of antigen. At day 4, the cells were harvested, extensively washed, and added to 96-well plates at designated densities (x-axis) with 2.5x10⁴ 2D2 responder T cells with or without 1 μM MOG35-55 in the presence or absence of the anti-MHC class IIm AbY3P or an isotype control (MK-D6, ATCC). To measure cellular proliferation, cultures were pulsed with 1 μCi of [³H]thymidine (6.7 Ci/mmol, Perkin Elmer, Waltham, MA) during the last 24 hr of a 72 hr culture. Cultures were harvested onto filters by use of a Tomtec Mach III harvester (Hamden, CT), and [³H]thymidine incorporation into DNA was measured by use of a Wallac 1450 Microbeta Plus liquid scintillation counter (Perkin Elmer, Waltham, MA).

Alternatively, BM cells from C57BL/6 mice or Ifngr1^−/− mice were cultured with 5 nM GM-CSF for 8–10 days to obtain a uniform population of differentiated MHCII⁺, CD11c⁺ DC. DC were extensively washed and cultured at the designated densities (x-axis) with or without 25,000 2D2 T cells in the presence of designated concentrations of MOG35-55 and/ or aminoguanidine, a NOS2 inhibitor. On day 2, cultures were pulsed with [³H]thymidine, and 24 hr later, cells and supernatants were harvested to measure proliferation or NO production. Production of NO was determined by measuring the formation of the stable decomposition product nitrite by mixing supernatant (100 μL) with an equal volume of Griess reagent (1% sulfanilamide/ 0.1% N-[1-naphthy] ethylenediamine in 2.5% phosphoric acid)(23). After 10 min of incubation, the optical density of 540 nm was measured in a microplate reader.

For in vitro experiments, a standard concentration of 1 μM was used for antigenic stimulation, including GMCSF-MOG and MOG35-55. For GM-CSF mediated differentiation of DC in the absence of antigen, a concentration of 5 nM was fully sufficient for growth and differentiation because these GM-CSF preparations had an EC50 ≤ 10 pM in DC proliferative bioassays (data not shown). For [³H]thymidine incorporation or NO production, error bars represented standard deviations of triplicate or quadruplet sets of wells.

Induction and treatment of EAE

For active induction of EAE, CFA was prepared by mixing Incomplete Freund’s Adjuvant with heat-killed Mycobacterium tuberculosis H37Ra (4 mg/ ml) (BD Biosciences, Franklin Lakes, NJ). The CFA adjuvant was mixed 1:1 with the designated dose of antigen and/ or fusion protein in saline and emulsified by sonication. Care was taken to dissipate heat during sonication. Active induction of EAE with CFA emulsions was performed by subcutaneous injection across the lower back. Each mouse received three separate injections (~ 0.033 ml per injection) for a total injection volume of 0.1 ml per mouse. For passive induction of EAE, gender-matched donor mice were actively sensitized with MOG35-55 in CFA. After 10-14 days, splenocytes were harvested from donor mice and were cultured with MOG35-55 and IL-12 (0.2 % v/v of a baculovirus rat IL-12 supernatant). After 3 days of culture, designated numbers of activated T cells were injected into recipients in saline (0.5 ml total volume by i.p. injection). In designated protocols, sequential passive and active EAE was induced in the same mice. The first bout of EAE was induced by adoptive transfer of activated encephalitogenic T cells. After peak disease and recovery, a second bout of EAE was induced by active challenge with MOG35-55 in CFA. For both active and passive induction of EAE, C57BL/6 mice received an i.p. injection of 200 ng of Pertussis toxin (List Biological Labs, Inc., Campbell, CA) in saline on the day of immunization and again 48 hours later. Fusion proteins or controls were administered subcutaneously in saline. All immunizations were performed under isoflurane anesthesia (Abbott Laboratories, Chicago, IL).

A standard dose of 200 μg peptide was used to induce EAE in both C57BL/6 and SJL models of EAE. A dose of 200 μg MOG35-55 was equivalent to 77.5 nmoles although a slight variation of 75 nmoles was used in some experiments. A dose of 200 μg PLP139-151 was equivalent to 131 nmoles. We used twice the immunization dose (400 μg) of MOG35-55 in Cd4^−/− mice to compensate for the lack of CD4. A standard dose of 2 nmoles of the GMCSF-NAg fusion protein and controls were used throughout, except in one experiment (Figure 4AB) in which 1 nmole was used instead.

Assessment of Clinical EAE

Mice were assessed daily for clinical score and body weight. The following scale was used to score the clinical signs of classical EAE: 0, no disease; 0.5, partial paralysis of tail without ataxia; 1.0, flaccid paralysis of tail or ataxia but not both; 2.0, flaccid paralysis of tail with ataxia or impaired righting reflex; 3.0, partial hind limb paralysis marked by inability to walk upright but with ambulatory rhythm in both legs; 3.5, same as above but with full paralysis of one leg; 4.0, full hindlimb paralysis; 5.0, total hindlimb paralysis with forelimb involvement or moribund. A score of 5.0 was a humane endpoint for euthanasia. An alternative scoring system was used to measure severity of atypical EAE in Ifngr1^−/− mice:0, no disease; 0.5, partial paralysis of tail or mild torticollis (<10° head-body twist);1.0, loss of tail tonus and mild torticollis; 2.0, flaccid paralysis of the tail and ataxia or impaired righting reflex with mild torticollis; 3.0, flaccid paralysis of the tail and severe torticollis (>10° head-body twist) with flaccid paralysis of one hindlimb; 3.5, same as above plus rigid paralysis of one hindlimb but not both; 4.0, bilateral rigid paralysis of both hindlimbs without forelimb involvement; 5.0, as above but with forelimb involvement and or more severe complications. A score of 5.0 served as the endpoint for humane euthanasia.

Cumulative EAE scores were calculated by summing daily scores for each mouse across a designated time period. Maximal scores were calculated as the most severe EAE score for each mouse. Mice that did not exhibit EAE had a score of zero for the cumulative and maximal scores, and these scores were included in the group average. To calculate percent maximal weight loss, 100% body weight was assigned as the maximal body weight obtained from day 1 through day 9, and daily body weights were calculated for each day after normalization to this 100% value. The minimum body weight was defined as the lowest body weight during the entire course of EAE after normalization to the 100% value. Maximal weight loss was calculated by subtraction of the normalized minimum value from the maximum 100% value. Average daily weight loss was calculated as the average of daily body weight measurements from day 10 until the end of experiment, subtracted from the 100% maximal body weight. Cumulative and maximal EAE scores were converted to ranked scores and analyzed by nonparametric ANOVA. Weight loss was analyzed by parametric ANOVA. Nonparametric and parametric ANOVA were assessed with a Bonferroni post hoc test. Error bars representing the standard error of the mean were included in the figures to represent the variance for EAE clinical scores and weight loss.

Results

GMCSF-NAg as a therapeutic intervention in EAE

C57BL/6 mice were immunized with MOG35-55/ CFA (experiments 1 and 2 of Table 1) to induce active EAE or alternatively were injected with activated MOG-specific Th1 T cells (experiment 3 of Table 1) to induce passive EAE. Experiments 2–3 are shown in Figure 1. To induce passive EAE, splenocytes from MOG35-55 sensitized donors were cultured for 3 days with IL-12 and 1 μM MOG35-55, and activated T cells were transferred into recipients. In all three experiments, mice were matched for disease severity, and treatment was initiated after onset of EAE. In both active and passive models of EAE, mice treated with MOG35-55 or saline showed a sustained course of EAE until the end of the experiment. Mice treated with GMCSF-MOG how ever showed a marked recovery during the next 10-20 days that was sustained until the end of the experiment (Figure 1A & C) together with an accelerated gain of body weight (Figure 1B & D).The ability of GMCSF-NAg to modulate established EAE when treatment was initiated after disease onset, and particularly in passive models of EAE, revealed that this therapeutic intervention modulated pre-established effector T cells to block pathogenicity of differentiated effector/ memory T cell populations. The therapeutic effectiveness of GMCSF-MOG was not due to interference of the CFA adjuvant, because recipient mice afflicted with passive EAE had no exposure to CFA. Rather, GMCSF-MOG inhibited the action of MOG-specific effector cells after CFA-priming, T cell differentiation, and acquisition of pathogenic activity.

Table 1.

GMCSF-MOG ameliorated chronic EAE in C57BL/6 mice.

Exp. #	Induction of EAE	Treatment a	Incidence of EAE after treatmentb	Mean cumulative score	Median cumulative score	Mean maximal score	Median maximal score	% maximal weight loss
1	Active	(a) GMCSF-MOG	4 of 6	11.8 ± 11.7	11.8	1.3 ± 1.3	1.3	2.4%
1	Active	(b) MOG35-55	6 of 6	36.3 ± 24.9	33.8	2.8 ± 1.5	3.0	15.0%
1	Active	(c) Saline	9 of 9	39.2 ± 20.0	43.0	3.2 ± 1.2	4.0	13.2%
2	Active	(d) GMCSF-MOG	3 of 6	5.6 ± 8.8	2.0	0.4 ± 0.5	0.3	11.1%
2	Active	(e) MOG35-55	6 of 6	142.8 ± 87.0	150.3	2.9 ± 1.3	3.0	18.2%
2	Active	(f) Saline	6 of 6	212.8 ± 116.4	282.8	3.5 ± 0.8	4.0	19.4%
3	Passive	(g) GMCSF-MOG	2 of 6	5.8 ± 8.9	0.0	0.2 ± 0.3	0.0	3.8%
3	Passive	(h) MOG35-55	5 of 6	60.8 ± 50.6	55.5	1.7 ± 1.4	1.5	10.2%
3	Passive	(i) Saline	5 of 6	72.8 ± 62.0	52.0	2.0 ± 1.7	1.5	7.6%

^aMice were immunized with 200 μg MOG35-55 in CFA (experiments 1-2) or were injected with 5 × 10⁶ activated MOG35-55 specific Th1 T cells (experiment 3) on day 0. Mice were also given Pertussis toxin (200 ngi.p.) on days 0 and 2. Mice were matched for clinical signs of EAE on the first day of treatment (mean maximal score of 2.1, 2.0, and 2.8 for experiments 1–3 respectively). Mice were treated with GMCSF-MOG or MOG35-55 (2 nmole) or saline on days 12, 14, 16, and 18 (experiment 1), days 12, 15, 17, and 19 (experiment 2), or days 21, 24, 26, and 28 (experiment 3).

^bIncidence of EAE at the start of treatment; (a) 6 of 6, (b) 5 of 6, (c) 6 of 9, (d–i) 5 of 6. Maximal weight loss was calculated by subtraction of the normalized minimum value from the maximum 100% value for each mouse, averaged across the group.

Experiment 1, post-treatment analyses from days 21–38; (a) versus (b) and (c): cumulative and maximal scores, p≤0.05. These data were presented previously in graphical form (Figure 6A of (14)).

Experiment 2, post-treatment analysis from days 20–91; (d) versus (e) and (f): cumulative and maximal scores, p≤0.002.

Experiment 3, post-treatment analysis from days 55–91; (g) versus (h) and (i): cumulative scores, p < 0.030.

Figure 1. GMCSF-MOG was an effective therapeutic in both passive and active models of EAE.

These data are portrayed in Table 1as experiments 2–3. Mice were immunized with 200 μg (77.5 nmole) MOG35-55 in CFA (A–B) or were injected with 5 × 10⁶ activated MOG35-55 specific Th1 T cells (C–D) on day 0. Mice were also given Pertussis toxin (200 ng i.p.) on days 0 and 2. Mice were matched for clinical signs of EAE on the first day of treatment and were treated with GMCSF-MOG or MOG35-55 (2 nmole) or saline on days 12, 15, 17, and 19 (A–B) or days 21, 24, 26, and 28 (C–D).

GMCSF-MOG had tolerogenic effectiveness in CD4-deficient and B cell-deficient mice

GMCSF-MOG was also effective as a therapeutic in Cd4^−/− mice (Figure 2 and Table 2). Although T cell antigen recognition is less efficient in Cd4^−/− mice, these mice are nonetheless susceptible to EAE when immunized with high doses of MOG35-55 in CFA, and the encephalitogenic response is mediated by Cd4^−/−, MHC class II-restricted T cells (data not shown). Because suboptimal encephalitogenic responses often result in monophasic EAE, CD4^−/− mice were used as a model to study a monophasic form of EAE on the C57BL/6 background. Passive induction of EAE in Cd4^−/− mice with MOG35-55 specific Cd4^−/− T cells resulted in a monophasic course of EAE, and a second bout of EAE was induced by active challenge of recovered recipients with MOG35-55 in CFA (Figure 2A and Table 2). To induce passive EAE, splenocytes from MOG35-55 sensitized Cd4^−/− donors were cultured for 3 days in the presence of IL-12 and MOG35-55. Activated Cd4^−/− T cells were then transferred into Cd4^−/−recipients to elicit a passive monophasic bout of EAE that had an onset at day 5 and spontaneous recovery by day 25. During the recovery phase, mice were treated with saline or with 2 nmole of MOG35-55 or GMCSF-MOG on days 17, 19, 21, and 23. EAE had almost completely waned by day 25, and mice were actively re-challenged on day 29 with 400 μg MOG35-55 in CFA together with i.p. injections of Pertussis toxin on days 29 and 31 to elicit a second bout of EAE. Mice previously treated with either MOG35-55 or saline showed a second episode of monophasic EAE. However, mice treated with GMCSF-MOG during the first bout of EAE exhibited only a mild second bout of EAE despite the lack of additional vaccine treatments. Just as CD4 was not necessary for myelin basic protein-induced EAE in PL/J mice (24), CD4 was not needed for induction of EAE in C57BL/6 mice and was not required for the inhibitory action of GMCSF-MOG. Overall, these data indicate that GMCSF-MOG treatment during the recovery phase of passively-induced EAE had an enduring effect as shown by reduction in disease severity during a subsequent actively-induced bout of EAE.

Figure 2. GMCSF-MOG mediated therapeutic activity in Cd4^−/− and B cell-deficient models of EAE.

These data are portrayed in Table 2. (A) Cd4^−/− mice were injected on day 0 with activated, MOG35-55 sensitized Cd4^−/− Th1 effector T cells to induce a monophasic bout of passive EAE. Treatment with GMCSF-MOG, MOG35-55 (2 nmole each), or saline was given during the peak and recovery phase of EAE on days 17, 19, 21, and 23. On day 29, all mice were challenged by injection of 400 μg MOG35-55 in CFA together with i.p. injections of Pertussis toxin on days 29 and 31. (B & C) B cell-deficient mice were immunized on day 0 by injection of 200 μg MOG35-55 in CFA together with i.p. injections of Pertussis toxin on days 0 and 2. On days 11, 13, 15, and 17, mice were treated with 2 nmole GMCSF-MOG or MOG35-55 by subcutaneous injection.

Table 2.

GMCSF-MOG had therapeutic action in both CD4-deficient and B cell-deficient C57BL/6 mice.

Strain	Treatment	Mean cumulative score	Median cumulative score	Mean maximal score	Median maximal score
CD4-deficient	(a) GMCSF-MOG	12.9 ± 10.5	14.5	0.9 ± 0.7	1.0
CD4-deficient	(b) MOG35-55	30.1 ± 18.0	28.0	2.0 ± 0.7	2.0
CD4-deficient	(c) Saline	32.5 ± 4.4	31.0	2.5 ± 0.9	2.0
B cell-deficient	(d) GMCSF-MOG	11.8 ± 16.2	5.3	1.3 ± 1.5	0.8
B cell-deficient	(e) MOG35-55	36.6 ± 14.6	41.8	3.5 ± 1.2	4.0

(a–c) Cd4^−/− mice were injected on day 0 with 3 × 10⁶ activated Cd4^−/− Th1 effector T cells. On days 17, 19, 21, and 23, mice were treated 2 nmole GMCSF-MOG (n = 5), 2 nmole MOG35-55 (n = 5), or saline (n = 3)by subcutaneous injection. On day 29, all mice were challenged by injection of 400 μg MOG35-55 in CFA together with i.p. injections of Pertussis toxin on days 29 and 31. Data were analyzed from day 38 until the end of the experiment on day 64. (a) versus (c); cumulative score, p = 0.020, maximal score, p = 0.042.

(d–e) B cell-deficient mice (n = 6 for each group) were immunized on day 0 by injection of 200 μg MOG35-55 in CFA together with i.p. injections of Pertussis toxin on days 0 and 2. On days 11, 13, 15, and 17, mice were treated 2 nmole GMCSF-MOG or MOG35-55 by subcutaneous injection. Data were analyzed from day 18 until day 29. Cumulative score, p = 0.022, maximal score, p = 0.034, percent maximal weight loss (0% vs 21.2% respectively, p = 0.019).

GMCSF-MOG also exhibited tolerogenic activity in B cell deficient mice (Figure 2B-C and Table 2). These experiments addressed whether the inhibitory effectiveness of GMCSF-MOG might be due to the induction of anti-murine GM-CSF antibodies that could neutralize endogenous GM-CSF to block EAE. Previous studies discounted the possibility that administration of GMCSF-MOG in saline elicited anti-GM-CSF antibodies(14). Also, previous studies (14–16) showed that treatment with GM-CSF alone or with noncovalently linked NAg lacked tolerogenic activity but nonetheless had the same theoretical potential as GMCSF-NAg to generate anti-GM-CSF antibodies. The use of B cell-deficient mice provided a definitive assessment of this issue. After active induction of EAE on day 0, mice were treated with 2 nmole GMCSF-MOG or MOG35-55 on days 11, 13, 15, and 17. Mice treated with MOG35-55 exhibited a chronic course of severe EAE, whereas mice treated with GMCSF-MOG exhibited remission and had mild EAE for the remainder of the experiment (Figure 2B). Unlike mice with severe EAE, mice treated with GMCSF-MOG showed rapid recovery and maintenance of body weight (Figure 2C). These data indicated that GMCSF-MOG reversed a pro-inflammatory immune response despite the absence of B cells or antibody. These data also exclude the possibility that GMCSF-MOG elicited anti-GM-CSF Ab to inhibit the effector phase of EAE. One can also conclude that the inhibitory effect of GMCSF-MOG did not require the action of regulatory B cells.

GMCSF-NAg retained tolerogenic activity into a CFA-primed lymphatic drainage

To directly address whether GMCSF-MOG imposed tolerance in inflammatory environments, GMCSF-MOG was injected in saline simultaneously at sites overlapping the injection of the MOG35-55/ CFA emulsion. Injection of GMCSF-MOG/ saline adjacent to the MOG35-55/ CFA injection on day 0 rendered the actively-immunized mice resistant to EAE (Figure 3A and Table 3) and prevented EAE-associated loss of body weight (Figure 3B). In contrast, an injection of GMCSF-PLP adjacent to MOG35-55/CFA did not affect the course of EAE. These data provided evidence that the NAg domain of GMCSF-MOG fusion protein was necessary for disease inhibition because the GM-CSF domain of GMCSF-PLP had no inhibitory effect. This observation was notable because injection of 2 nmole GMCSF-MOG on day 0 was sufficient to inhibit EAE. All groups were also given i.p. injections of Pertussis toxin on days 0 and 2. Thus, the inhibitory activity of GMCSF-MOG superseded the adjuvant activity of both CFA and Pertussis toxin. GMCSF-MOG inhibited EAE even though the molar dose was 38.8-fold less than that of MOG35-55 (i.e., molar doses of GMCSF-MOG and MOG35-55 were 2 and 77.5 nmole, respectively). These data indicated that the tolerogenic dose of GMCSF-MOG was dominant compared to the encephalitogenic dose of MOG35-55. A separate group of mice given adjacent injections of saline/ CFA and GM-CSF-MOG in saline did not exhibit any signs of EAE. These data indicated that GMCSF-MOG, when injected in saline, did not cause EAE when introduced directly into a CFA-primed lymphatic drainage. Overall, these findings reinforce the concept that introduction of GMCSF-MOG (in saline) into an inflammatory environment was tolerogenic and did not engender disease.

Figure 3. GMCSF-MOG inhibited EAE when injected adjacent to the MOG35-55/ CFA emulsion.

These data are portrayed in Table 3. C57BL/6 mice were injected with 77.5 nmole (200 μg) MOG35-55 in CFA and an adjacent injection of GMCSF-MOG, GMCSF-PLP, or saline. Fusion proteins were injected subcutaneously in saline at a dose of 2 nmole on day 0. All groups also received Pertussis toxin (200 ng i.p.) on days 0 and 2.A separate group of mice were injected with a saline/CFA emulsion and an adjacent injection of 2 nmole GMCSF-MOG.

Table 3.

GMCSF-MOG prevented EAE in the midst of a CFA-induced pro-inflammatory environment

Treatment	Incidence of EAE	Mean cumulative score	Median cumulative score	Mean maximal score	Median maximal score	% maximal weight loss
(a) MOG/CFA + GMCSF-PLP	6 of 6	40.8 ± 15.1	43.3	3.1 ± 0.7	3.0	17.2%
(b) MOG/CFA + GMCSF-MOG	2 of 6	4.0 ± 8.4	0.0	0.8 ± 1.3	0.0	3.4%
(c) MOG/CFA alone + saline	8 of 8	56.6 ± 18.3	54.5	3.7 ± 0.5	4.0	19.0%
(d) PBS/CFA + GMCSF-MOG	0 of 6	0.0 ± 0.0	0.0	0.0 ± 0.0	0.0	1.7%

(a–d) C57BL/6 mice were injected with 77.5 nmole (200 μg) MOG35-55 in CFA with an adjacent injection of saline (c) or 2 nmole of either GMCSF-PLP (a) or GMCSF-MOG (b) in saline. A separate group of mice were injected with a saline/CFA emulsion and an adjacent injection of 2 nmole GMCSF-MOG (d). All four groups also received Pertussis toxin (200 ng i.p.) on days 0 and 2. Data analysis was based on daily assessments from day 0 to day 35. These data were previously shown as a density dot plot (Figure 6B of (14)).(b) and (d) versus (a) or (c); cumulative score, p < 0.001, maximal score, p≤0.002, maximal weight loss, p≤0.004.

GMCSF-NAg retained tolerogenic activity when directly included in a MOG35-55/CFA emulsion

To further test the hypothesis that GMCSF-NAg fusion proteins may retain activity in the midst of a CFA-primed lymphatic drainage, GMCSF-MOG was directly emulsified with MOG35-55 at a 1:75 molar ratio in the CFA, respectively. MOG35-55 in CFA elicited a full course of paralytic EAE (Figure 4 and Table 4). However, inclusion of GMCSF-MOG directly in the MOG35-55/ CFA emulsion alleviated disease such that 3 of 6 mice had mild EAE and 3 of 6 mice did not exhibit EAE (Figure 4A). This group of mice also did not exhibit EAE-associated weight loss (Figure 4B). These data indicated that GMCSF-MOG had a dominant inhibitory activity in CFA-primed inflammatory environments. Apparently, the GM-CSF domain of GMCSF-MOG was stable and largely retained a biologically active conformation despite inclusion in the lipid environment of the CFA emulsion. If GM-CSF denatured and lost biological activity, the fusion protein would default to MOG35-55 in CFA, which would add to the encephalitogenicity of the emulsion, albeit incrementally by increasing the dose of MOG35-55 from 75 to 76 nmole. The majority of mice (5 of 6 mice) immunized with GMCSF-MOG alone (without MOG35-55) in CFA did not exhibit EAE whereas one mouse showed mild EAE. The dominant inhibitory activity of GMCSF-MOG in CFA emulsions indicated that the fusion protein did not require ‘steady-state’ environments to efficiently mediate tolerance.

Figure 4. GMCSF-MOG exhibited potent inhibitory activity when mixed in the CFA emulsion with MOG35-55 in C57BL/6 and B cell-deficient mice.

These data are portrayed in Table 4. (A & B) C57BL/6 mice were immunized on day 0 with 75 nmole of MOG35-55 in CFA, with 1 nmole of GMCSF-MOG in CFA, or with a mixture of 75 nmole of MOG35-55 and 1 nmole of GMCSF-MOG in CFA. Mice were scored daily through the end of the experiment on day 41. (C & D) B cell-deficient mice were immunized on day 0 with 75 nmole of MOG35-55 in CFA, with 2 nmole of GMCSF-MOG in CFA, or with a mixture of 75 nmole of MOG35-55 and 2 nmole of GMCSF-MOG in CFA. Mice were scored daily through the end of the experiment on day 50. All mice were given i.p. injections of Pertussis toxin on days 0 and 2.

Table 4.

GMCSF-MOG exhibited inhibitory activity when included in a CFA emulsion

Included in CFA emulsiona	Incidence of EAE	Mean cumulative score	Median cumulative score	Mean maximal score	Median maximal score	% weight loss
(a) GMCSF-MOG + MOG35-55	3 of 6	15.7 ± 19.3	7.8	1.2 ± 1.5	0.5	4.4%
(b) GMCSF-MOG	1 of 6	1.1 ± 2.7	0.0	0.2 ± 0.4	0.0	4.3%
(c) MOG35-55	6 of 6	79.2 ± 36.6	83.5	3.7 ± 0.5	4.0	18.9%
(d) GMCSF-MOG + MOG	5 of 6	47.5 ± 33.7	40.5	2.8 ± 1.4	3.0	9.3%
(e) GMCSF-MOG	2 of 6	10.5 ± 16.4	0.0	0.8 ± 1.3	0.0	−1.6%
(f) MOG35-55	7 of 7	136.7 ± 10.9	142.0	4.0 ± 0.0	4.0	20.9%

(a–c) C57BL/6 mice were immunized on day 0 with 75 nmole of MOG35-55 in CFA (c), with 1 nmole of GMCSF-MOG in CFA (b), or a mix of 75 nmole of MOG35-55 and 1 nmole of GMCSF-MOG in CFA (a). All mice were given i.p. injections of Pertussis toxin on days 0 and 2. Data analysis represented the time course from day 0 through the end of the experiment on day 41. (c) versus(a) and (b); cumulative scores, p = 0.002, p < 0.001; maximal score, p = 0.002, p < 0.001; weight loss, p = 0.068, p = 0.005.

(d–f) B cell-deficient mice were immunized on day 0 with 75 nmole of MOG35-55 in CFA (c), with 2 nmole of GMCSF-MOG in CFA (b), or a mix of 75 nmole of MOG35-55 and 2 nmole of GMCSF-MOG in CFA (a). All mice were given i.p. injections of Pertussis toxin on days 0 and 2. Data analysis represented the time course from day 0 through day 50. Average daily weight loss was calculated as the average of daily body weight measurements from day 10 until the end of experiment, subtracted from the 100% maximal body weight. (f) versus(d) and (e); cumulative scores, p < 0.001, p < 0.001; maximal score, p = 0.002, p < 0.001; maximal weight loss, ns, p < 0.001; average weight loss; p = 0.018, p < 0.001;

A replicate experiment (Table 4 and Figure 4C-D) was conducted in B cell-deficient mice to address the possibility that immunization of GMCSF-MOG in CFA elicited neutralizing anti-GM-CSF Ab that in turn inhibited EAE. B cell-deficient mice were confirmed as devoid of CD19⁺s Ig⁺ splenocytes (data not shown). Unlike GM-CSF-deficient mice, immunization of B cell-deficient mice with GMCSF-MOG/ CFA (with or without MOG35-55) did not produce detectable anti-GMCSF Ab (data not shown). Immunization of B cell-deficient mice with MOG35-55 in CFA elicited a highly aggressive course of EAE in which 7 of 7 mice had sustained severe paralytic disease for the remainder of the experiment. Inclusion of 2 nmole of GMCSF-MOG in the MOG35-55/CFA emulsion attenuated the course of EAE such that 5 of 6 mice asynchronously peaked at a score of 3.0 but then exhibited spontaneous recovery and thereafter had a mild course of EAE (Figure 4C). The fusion protein also prevented much of the weight loss that was associated with full paralytic EAE (Figure 4D). Immunization with GMCSF-MOG alone in CFA resulted in a low incidence (2 of 6) of mild EAE. These data indicated that B cell-deficient mice had exacerbated EAE but that GMCSF-MOG nonetheless attenuated disease over a prolonged timeframe. These data precluded a role for anti-GM-CSF Ab as a mechanism by which GMCSF-MOG mediated inhibitory activity in the C57BL/6 model of EAE.

The same activity was observed when GMCSF-PLP was tested in the SJL model of relapsing-remitting EAE. In CFA, PLP139-151 caused chronic relapsing remitting EAE that persisted until the end of the experiment on day 69 (Figure 5 and Table 5). The GMCSF-PLP fusion protein, when included with PLP139-151 in the emulsion, had an inhibitory effect that markedly attenuated disease severity and conferred a limited, monophasic time course marked by a long-lasting convalescence. The inhibitory activity of GMCSF-PLP was dominant even though 2 nmole of GMCSF-PLP were mixed with 131 nmole of PLP139-151 in the CFA. Thus, GMCSF-PLP had a dominant inhibitory activity even though the molar dose of GMCSF-PLP was approximately 60-fold less than that of PLP139-151.

Figure 5. GMCSF-PLP exhibited potent inhibitory when mixed in the CFA emulsion with PLP139-151.

These data are portrayed in Table 5. SJL mice were immunized on day 0 with 131 nmole (200 μg) of PLP139-151 in CFA or with a mix of 131 nmole of PLP139-151 and 2 nmole of GMCSF-PLP in the same CFA emulsion.

Table 5.

GMCSF-PLP exhibited inhibitory activity when included in a CFA emulsion

Included in CFA emulsion a	Incidence of EAE	Mean cumulative score	Median cumulative score	Mean maximal score	Median maximal score
GMCSF-PLP + PLP139-151	5 of 8	0.5 ± 1.4	0.0	0.1 ± 0.4	0.0
PLP139-151	8 of 8	36.8 ± 27.0	41.8	2.6 ± 1.7	3.5

^aSJL mice were immunized on day 0 with 131 nmole (200 μg) of PLP139-151 in CFA or with a mix of 131 nmole of PLP139-151 and 2 nmole of GMCSF-PLP in the same CFA emulsion. Mice were scored daily through the end of the experiment on day 69. Shown are the data analyses for days 42-69; cumulative scores, p = 0.003; maximal scores, p = 0.002.

Mechanisms of GMCSF-NAg inhibitory action

To assess the APC activities of DC cultured in the presence of GMCSF-MOG, naïve C57BL/6 BM cells were cultured in the presence of GMCSF-MOG, a combination of GM-CSF + MOG35-55 (GMCSF+MOG), GM-CSF alone, MOG35-55 alone, or saline for 4 days. During this time, GM-CSF promoted differentiation of BM cells into CD11c⁺, MHCII⁺ DC (data not shown). Cells exposed to GM-CSF expanded during culture, whereas cultures without GM-CSF (i.e., MOG35-55 or saline) did not survive or were insufficient in number for subsequent assays. After 4 days of culture, cells were extensively washed and assayed for MOG-specific APC activity in a subsequent assay in which these APC were cultured with a continuous line of MOG35-55 specific T cells derived from 2D2 MOG35-55 TCR transgenic mice.

As shown in Figure 6A, “GMCSF-MOG”DC (i.e., BM cells cultured with GMCSF-MOG) stimulated the growth of 2D2 MOG-specific T cell sat low DC densities (1,250 – 5,000 DC/ well) but not at high DC densities. Thus, “GMCSF-MOG” DC elicited a distinct low-density stimulatory zone and a high-density inhibitory zone during presentation of MOG to 2D2 T cells. At low DC densities (Figure 6A), cell growth was due to blastogenic T cell expansion. In comparison, DC previously cultured with GMCSF + MOG35-55 as separate molecules (“GMCSF+MOG” DC) did not present MOG to 2D2 T cells at low DC densities. Rather, “GMCSF+MOG” DC optimally presented MOG35-55 to 2D2 T cells at a very high DC density (40,000 DC/ well). “GM-CSF” DC that were cultured in the initial 4-day culture without MOG35-55 did not stimulate 2D2 T cells at any density (data not shown). Overall, these data revealed an 8-fold shift in the response optima in cultures of “GMCSF-MOG” DC compared to “GMCSF+MOG” DC. This finding reflected the superior ability of GMCSF-MOG to target MOG35-55 for enhanced antigen presentation compared to the combination of GM-CSF + MOG35-55 thus resulting in optimal antigen presentation at 8-fold lower DC densities. Importantly, these findings revealed an inhibitory high-zone range of “GMCSF-MOG” DC densities in which presentation of MOG was coupled with the lack of any visible T cell response.

Figure 6. GMCSF-MOG targeted MOG35-55 for inhibitory MHCII-restricted antigen presentation by inflammatory DC.

C57BL/6 BM cells were cultured at 1 × 10⁶ cells/ ml with 1 μM additions of GMCSF-MOGor a combination of GM-CSF and MOG35-55(GMCSF+MOG) for 4 days. BM cells were extensively washed and added to wells (x-axis plot of exponent) at designated densities (cells/ well) of 80,000 (10^4.9), 40,000 (10^4.6), 20,000 (10^4.3), 10,000 (10^4.0), 5,000 (10^3.7), 2,500 (10^3.4), 1,250 (10^3.1), or no DC in the presence (B, C) or absence (A, C) of an anti-I-Ab mAb (Y3P, 2% v/v) with (C) or without (A, B) 1 μM MOG35-55. MOG-specific 2D2 T cells were added at a density of 25,000/ well. Cultures were pulsed with [³H]thymidine on day 2 of a 3-day culture as a measure of cellular proliferation. Note that the y-axis scale for A-B differs from that for C. These data are representative of three independent experiments.

The low-zone stimulatory activity of “GMCSF-MOG” DC was dramatically “right-shifted” in the presence of the anti-I-A Y3P mAb (compare Figures 6A-B). Thus, the low-zone stimulatory activity of “GMCSF-MOG” DC was dependent upon MHCII glycoproteins. Notably, cell growth at high DC densities was primarily comprised of myeloid cells rather than T cells as expected given that these cultures were seeded with a plurality of DC. At a high “GMCSF-MOG” DC density of 40,000 DC/ well, addition of Y3P dramatically augmented cell proliferation (mean of 889 cpm vs 389,338 cpm in the absence or presence of Y3P, respectively). The ability of Y3P to enable growth indicated that T cell recognition of MOG in these high density DC cultures was coupled to an inhibitory process. Overall, these data support the interpretation that “GMCSF-MOG” DC mediated an inhibitory mechanism of antigen presentation at high densities of DC. Such DC densities might be readily achieved in secondary lymphoid organs in vivo, particularly when 1–2 nmole of GMCSF-MOG is funneled into a lymphatic drainage. This inhibitory mechanism of antigen presentation might be relevant to the mechanisms by which GMCSF-NAg mediated tolerance in pro-inflammatory environments during EAE (Figures 1–5).

Addition of 1 μM MOG35-55 to cultures of 2D2 T cells with either “GMCSF-MOG”DC or “GMCSF+MOG”DC resulted in left-shifted proliferative responses that peaked at low DC densities(10^3.1 or 1250 DC/ well) (Figure 6C). Addition of MOG35-55 to these T cell/ DC cultures also resulted in a reduction in the magnitude of the proliferative response in comparison to equivalent cultures without the addition of soluble MOG35-55 (compare ***********Figure 6C to A-B, note different y-axis scales). These data show that “GMCSF-MOG”DC and “GMCSF+MOG”DChad equal capabilities for antigen presentation when saturated with equivalent concentrations of MOG35-55. These data show that the difference in presentation of MOG35-55 in Figures 6A-B was due to differential loading of MOG35-55 rather than differences in MHC class II expression or co-stimulation. In the presence of 1 μM MOG35-55 (Figure 6C), addition of the Y3P anti-I-A mAb to these cultures resulted in a right-shift in the growth curves due to abrogation of the inhibitory mechanism of antigen presentation and the outgrowth of myeloid APC. Overall, the efficiency of MOG presentation by inflammatory DC was associated with left-shifted response curves and reduced response maxima due to a broader zone of inhibitory antigen presentation. Notably, the inhibitory APC mechanism associated with MOG presentation at high DC densities blocked both T cell and myeloid growth.

Proliferative responses, when observed at low DC densities, primarily represented an outgrowth of blastogenic 2D2 T cells. This observation was consistent with the culture setup in that T cells were the dominant population in those cultures (e.g., 25,000 T cells plus 1,250 DC). In contrast, cell growth at high DC densities was primarily due to outgrowth of DC that represented the dominant starting population in these cultures (e.g., 25,000 T cells plus 80,000 DC). Thus, growth at low DC densities was T cell dominant, and growth at high DC densities was myeloid dominant.

Growth inhibition mediated by NO production

Several observations suggested a mechanism whereby presentation of MOG by proinflammatory DC to 2D2 T cells resulted in the production of the cytotoxic mediator NO to inhibit cell growth. As shown in Figure 7A, GMCSF-DC obtained from 8-day cultures with GM-CSF (e.g. without a NAg domain) exhibited potent APC activity in the presence of MOG35-55. These DC were tested across a density range in the presence of a fixed number of MOG-specific 2D2 transgenic T cells in the presence or absence of MOG35-55. Low densities of DC (up to 5,000 DC/ well) supported MOG-induced stimulation of 2D2 T cells, but higher densities of DC were inhibitory. At densities above 10,000 DC per well, MOG35-55 inhibited the proliferation of both T cells and DC. Overall, MOG35-55 stimulated T cell growth at low DC densities but markedly inhibited all growth at high DC densities. In the absence of MOG, proliferation at high DC densities involved outgrowth of DC rather than T cells (data not shown). Two lines of evidence indicated that the inverse activity of MOG at low versus high DC densities in part involved MOG-specific induction of NO. First, aminoguanidine inhibited MOG-specific induction of NO(Figure 7B) and partially restored the stimulatory action of MOG particularly at higher DC densities (10^4.0-4.6) (Figure 7A). Secondly, MOG-specific production of NO was inversely correlated with MOG-specific T cell proliferation (Figure 7C). These data support the concept that GMCSF-NAg may couple outgrowth of high density regulatory DC, antigen-targeting of a covalently-coupled myelin peptide to DC, and a NO-dependent mechanism of counter-regulation.

Figure 7. GM-CSF induced the differentiation of DC that limited T cell expansion by production of NO.

BM cells were cultured with 5 nM GM-CSF for 8 days to obtain a uniform population of differentiated CD11c⁺ DC. DC were extensively washed and cultured at the designated densities ranging from 80,000 (10^4.9) to 1,250 (10^3.1) cells/ well (x-axis) with 25,000 2D2 T cells with or without 1 μM MOG35-55 or 2 mM aminoguanidine. On day 2, cultures were pulsed with [³H]thymidine. On day 3, cells and supernatants were harvested to measure proliferation (A)or NO production (B).An independent experiment portrays MOG-stimulated proliferation and MOG-stimulated NO production in the same graph (C). These data are representative of three independent experiments.

These findings implicated the IFN-γ/ NO axis as core component of the inhibitory APC mechanism. The postulate is that myeloid cell antigen presentation consecutively evokes T cell-mediated IFN-γ production, myeloid DC-mediated production of NO, and NO-mediated growth inhibition/ cytotoxicity of both T cells and DC. To assess this possibility, wildtype DC or Ifngr1^−/− DC were passaged for 10 days in the presence of GM-CSF and were assayed for inhibitory antigen presentation in the presence of 2D2 T cells and designated concentrations of MOG35-55. In proliferative assays, the bell-shaped response curves of wildtype DC were left-shifted and exhibited lower response maxima compared to those of Ifngr1^−/− DC (compare Figures 8A-B). These data indicate that wildtype DC had more potent antigen presentation but were also associated with a substantially more robust mechanism of inhibitory antigen presentation. In accordance, wildtype DC supported MOG-stimulated NO production whereas cultures with Ifngr1^−/− DC lacked detectable NO production (compare Figures 8D-E). In cultures of wildtype DC, T cells were required for MOG-induced production of NO (Figures 8C, E). Overall, these data indicate that the inhibitory mechanism of antigen presentation was associated with NO production and that NO production by myeloid cells required T cells, MOG35-55, and IFN-γ responsiveness.

Figure 8. NO production by DC was dependent upon antigen, T cells, and IFN-γ signaling.

BM cells from C57BL/6 mice or Ifngr1^−/− mice were cultivated for 10 days in the presence of GM-CSF. DC were extensively washed and cultured at the designated densities ranging from 80,000 (10^4.9) to 1,250 (10^3.1) cells/ well (x-axis) with (A–B, D–E) or without (C, F) 25,000 2D2 T cells in the presence of 10 μM or 1 μM MOG35-55 or no antigen. On day 2, cultures were pulsed with [³H]thymidine, and 24 hr later, cells and supernatants were harvested to measure proliferation (A–C) or NO production (D–F).Note that the y-axis scale for A–B differs from that for C. These data are representative of three independent experiments.

Although these data implicate the IFN-γ/ NO axis in the mechanism of inhibitory APC activity, high zone inhibition was not abrogated in the absence of IFN-γ. Rather, inhibition of NO production by aminoguanidine (Figure 7A) or IFNGR1-deficiency (Figure 8B) shifted the range of stimulatory APC activity to higher DC densities but did not abrogate inhibitory activity at the highest DC densities. The possibility must be considered that other undefined inhibitory mechanisms acting in parallel to the IFN-γ/ NO axis may participate in this inhibitory APC mechanism.

These findings prompted the question of whether the IFN-γ/ NO axis mediated an inhibitory role in vivo and accounted, at least in part, for the inhibitory activity of GMCSF-MOG when mixed with MOG35-55 in the encephalitogenic CFA emulsion (Figure 9 and Table 6). Ifngr1^−/− mice were immunized with 75 nmole of MOG35-55 in CFA or with a mixture of 75 nmole of MOG35-55 and 2 nmole of GMCSF-MOG in CFA. Immunization with MOG35-55 in CFA resulted in a protracted and severe course of atypical EAE in 100% of the immunized mice. Severity of EAE was assessed by an alternative EAE clinical assessment scale (see Materials and Methods). Inclusion of GMCSF-MOG with MOG35-55 in the encephalitogenic emulsion resulted in a delayed disease course and a significant attenuation of EAE severity until day 30. During this period, differences in cumulative EAE score were significant and differences in maximal EAE score trended toward significance (p = 0.081). Significant differences in weight loss were noted for days 11-25, 30-33, and 35-37 (p < 0.05). After day 30, differences in EAE scores were less pronounced and lacked statistical significance as the inhibitory influence of GMCSF-MOG waned against a steadily progressive course of atypical EAE. Overall, these data are consistent with the in vitro analyses and support the conclusion that the inhibitory activity of GMCSF-MOG in CFA-induced inflammatory environments is mediated in part by an IFN-γ/ NO axis.

Figure 9. GMCSF-MOG partially inhibited the encephalitogenic activity of MOG35-55 when mixed together in the CFA emulsion.

These data are portrayed in Table 6. Ifngr1^−/−mice were immunized with 75 nmole of MOG35-55 in CFA or with a mixture of 75 nmole of MOG35-55 and 2 nmole of GMCSF-MOG in CFA. Mice were scored daily through the end of the experiment on day 52. All mice were given i.p. injections of Pertussis toxin on days 0 and 2.

Table 6.

GMCSF-MOG exhibited partial inhibitory activity in Ifngr1^−/− mice when included with MOG35-55 in a CFA emulsion

Included in CFA emulsion a	Incidence of EAE	Mean cumulative score	Median cumulative score	Mean maximal score	Median maximal score	% maximal weight loss
(a) GMCSF-MOG + MOG35-55	8 of 8	30.8 ± 25.4	40.3	2.6 ± 1.9	4.0	22.1%
(b) MOG35-55	10 of 10	68.7 ± 4.7	68.6	4.0 ± 0.0	4.0	30.6%

(a–b) Ifngr1^−/−mice were immunized with 75 nmole of MOG35-55 in CFA (b)or a mixture of 75 nmole of MOG35-55 and 2 nmole of GMCSF-MOG in CFA (a). All mice were given i.p. injections of Pertussis toxin on days 0 and 2. Shown are the data analyses through day 30. (a) versus(b); cumulative scores, p = 0.000; maximal score, ns; weight loss, p = 0.003. No consistent, significant differences were noted from day 31 until the end of the experiment on day 52.A high incidence of atypical EAE was noted in both groups; (a) 7 of 8, (b)10 of 10.

Discussion

GMCSF-NAg fusion proteins comprised of GM-CSF and dominant myelin epitopes of MBP, MOG, and PLP were potent NAg-specific tolerogens in rat and mouse models of EAE(14-16). GMCSF-NAg proteins were tolerogenic vaccines when administration was completed before immunization and were effective interventions when GMCSF-NAg treatment was initiated after onset of EAE. The current study showed that GMCSF-NAg was an effective therapeutic intervention in both active and passive models of EAE and that inhibitory activity was intact in both CD4-deficient and B cell-deficient mice (Tables 1–2, Figures 1–2). The implication of these findings was that GMCSF-NAg exerted inhibitory activity in an inflammatory environment, and this implication was confirmed by the observation that GMCSF-NAg had a dominant inhibitory activity when given at the same time in a site adjacent to the NAg/ CFA emulsion or when directly emulsified in the encephalitogenic NAg/ CFA emulsion (Tables 3–5, Figures 3–5). Both approaches resulted in the introduction of GMCSF-NAg into the same lymphatic drainage as the NAg/ CFA emulsion, and both approaches resulted in the inhibition of EAE. GMCSF-NAg had a dominant inhibitory action even though the NAg peptide was present in the emulsion at a substantial molar excess. These data indicated that GMCSF-NAg directly modified the lymphatic environment to inhibit NAg-specific encephalitogenic T cells. In vitro, GMCSF-MOG elicited differentiation and expansion of myeloid-derived DC, and GMCSF-MOG targeted MOG35-55 to these inflammatory DC to mediate an inhibitory mechanism of antigen presentation (Figures 6–8). This inhibitory mechanism of antigen presentation was mediated, at least in part, by an IFN-γ/ NO feedback pathway that inhibited MOG-specific T cells (Table 6, Figures 8–9). The main conclusion was that GMCSF-NAg was distinguished as a unique class of tolerogenic vaccine compared to other antigen-specific vaccine approaches in that GMCSF-NAg caused tolerance in CFA-primed lymphatics. In contrast, naked myelin peptides or proteins that elicit NAg-specific tolerance would predictably cause EAE if directly emulsified in CFA.

The “bait-and-trap” hypothesis

The activity of the GM-CSF domain and the coupling of the GM-CSF domain to the NAg domain were both critical for tolerogenic activity of GMCSF-NAg. The GM-CSF domain conferred several important functions to the vaccine(13–16). First, the GM-CSF domain was a potent growth and differentiation factor for myeloid-derived DC (Figures 6–8). Second, the GMCSF domain conferred competence so that those DC produced high concentrations of NO during NAg-dependent interactions with 2D2 MOG-specific T cells (Figures 7–8). Third, the GM-CSF domain targeted high concentrations of the MOG35-55 domain to those myeloid DC to confer enhanced presentation of MOG on MHCII glycoproteins (Figure 6). This mechanism of ‘antigen targeting’ was contingent upon the covalent linkage of the GM-CSF and NAg domains and was blocked by free GM-CSF(14-16). GMCSF-NAg thereby appeared to drive a negative regulatory IFN-γ/ NO axis in EAE (Table 6, Figure 9). Mouse strains deficient in IFN-γ or NO have heightened susceptibility to EAE, and these mediators have a prominent negative regulatory role in EAE(25–34). IFN-γ is the prototypic cytokine of the Th1 subset, and IFN-γ is a primary stimulus for the induction of NO by myeloid-derived phagocytes. High concentrations of NO are known to impair the growth and viability of T cells and other bystander cells. Overall, this study provides evidence that GMCSF-MOG may inhibit EAE by priming this regulatory axis via expansion and differentiation of NOS2-competent myeloid APC coupled with the targeting of NAg to these myeloid APC.

The contemporary view is that self-antigens impose tolerance preferentially in quiescent, steady-state environments(8–12). This concept however cannot be considered exclusive because mechanisms of tolerance must also function in inflamed tissues during adaptive immune responses. Otherwise, immunity to foreign antigens would routinely lead to pathogenic autoimmunity particularly in response to persistent environmental or infectious agents. Prolonged leukocyte activation elicits counter-regulatory molecules including IL-2, IL-10, TGF-β, CTLA-4, PD-1, NO, and IDO among many others. Several of these regulatory molecules are essential for maintenance of immune homeostasis and long-term tolerance because deficiency in these gene products augments autoimmune disease(28, 35–45). Although NO may be an important negative mediator downstream of GMCSF-MOG, counter-regulatory mediators aside from NO may also underlie vaccine-mediated inhibition of encephalitogenic T cells. The negative regulatory activities of GM-CSF likely depend on the growth and differentiation of myeloid-derived suppressor cells and regulatory DC. GMCSF-NAg tightly couples these myeloid counter-regulatory activities with the presentation of myelin peptides. Experimental approaches that co opt counter-regulatory pathways may have qualitative advantages of eliciting tolerance in both quiescent and inflammatory environments.

These considerations are consistent with a “bait and trap” mechanism of tolerance induction. The bait is laid when GMCSF-NAg expands inflammatory myeloid APC and simultaneously targets high concentrations of NAg for presentation by those myeloid APC. The trap is set when MOG-specific T cells recognize MOG on myeloid APC to induce high concentrations of cytotoxic mediators by those APC with the consequent demise of MOG-specific T cells docked onto those APC. Given that T cell antigen recognition was necessary for the induction of NO, the trap would be triggered only upon T cell recognition of the MHCII/ myelin peptide complex on the myeloid APC. Notably, data shown in Figures 6-8 revealed that MOG presentation at low DC densities stimulated T cell growth but MOG presentation at high DC densities inhibited all growth including both T cell and myeloid DC. A central prediction is that subcutaneous vaccination concentrated GMCSF-MOG into a lymphatic drainage to transform that leukocytic environment into an inflammatory, DC-enriched zone. The prediction therefore is that GMCSF-NAg vaccination would elicit the draining lymph nodes to exhibit the inhibitory profile noted in the high density DC cultures of Figures 6–8. This prediction is consistent with the fact that in vivo environments are maintained at substantially higher cell densities that what can be achieved in vitro. These considerations support the possibility that the high density DC cultures portrayed in Figures 6–8 represent a model to study the “trap” that underlies, at least in part, a mechanism by which GMCSF-MOG mediates inflammation-dependent tolerance.

Although a previous study indicated that GM-CSF transits the blood brain barrier(46), it is unlikely that subcutaneous administration of GMCSF-NAg at the doses used in this study would suffice to directly modulate inflammatory T cell responses within the CNS. Rather, GMCSF-MOG may act solely via a peripheral mechanism outside of the CNS. The possibility exists for example that chronic EAE in C57BL/6 mice may require a continuous low-grade infiltration of inflammatory T cells into the CNS to replace apoptotic T cells and thereby maintain threshold T cell densities necessary to sustain ongoing inflammation and disease. Thus, GMCSF-NAg may act outside the CNS in secondary lymphoid organs to compromise the peripheral MOG-specific repertoire and thereby interfere with the generation and transit of new MOG-specific T cells into the CNS and thereby subvert maintenance of EAE. Alternatively, GMCSF-NAg may promote peripheral expansion of regulatory T cells that migrate to the CNS to dampen inflammation and promote recovery.

The tolerogenic side of GM-CSF

GMCSF is considered to be a central pathogenic cytokine in EAE that may mediate complicated pro-inflammatory activities in the periphery and the autoimmune target organ(47–50). One potential locus of GM-CSF action in EAE may include the CNS whereby myelin-specific T-helper cells of the Th1-Th17 spectrum, upon re-activation by CNS-derived myelin antigens, elaborate GM-CSF to initiate activation and differentiation of myeloid phagocytes that ultimately mediate CNS inflammation, demyelination, and the neurologic deficits of EAE. Thus, the elaboration of GM-CSF directly in the target tissue upon infiltration and reactivation of self-reactive T cells (47, 48) or by target tissue-specific expression of GM-CSF as a transgene (51, 52) fosters autoimmunity. Conversely, GM-CSF also is a key cytokine involved in the differentiation of regulatory DC and myeloid-derived suppressor cells that in turn drive expansion of regulatory T cells to inhibit autoimmune disease (53–64). Administration of GM-CSF (without antigen) to the peripheral immune system is known to inhibit several autoimmune diseases including autoimmune diabetes, experimental autoimmune myasthenia gravis, and experimental autoimmune thyroiditis whereas deficiency of GM-CSF is associated with susceptibility to autoimmune diabetes and systemic lupus erythematosus (65, 66).The pathogenic and protective actions of GM-CSF however may have a common denominator. GM-CSF is known to prime the differentiation and growth of myeloid-derived phagocytic APC that mediate enhanced phagocytosis and elaboration of cytotoxic factors (i.e., NO, arginase, IDO) to damage target tissue but to also kill/ clear myelin-specific T cell subsets that recognize MHCII-restricted antigens on those inflammatory DC. A common inflammatory activity may thereby result in cell death of both tissue-specific somatic cellsto culminate in autoimmunity together with concomitant death of the invading T cellsto set the stage for remission and immunological tolerance. The idea underlying this study is that a GMCSF-NAg vaccine, unlike GM-CSF alone, represents a means to pull the GM-CSF trigger therapeutically outside of the autoimmune target tissue, such that one might achieve clearance of autoreactive T cells without inflicting damage within the autoimmune target tissue.

Utility of CD4-deficient and B cell-deficient models of EAE

GMCSF-MOG reversed EAE in CD4-deficient mice (Table 2 and Figure 2). CD4 is known to potentiate the recognition of MHCII-antigen by approximately 100-fold and is required for positive thymic selection of many T-helper cell clonotypes. Nonetheless, CD4⁻ MOG-specific T cells differentiate in the thymus and seed peripheral lymphoid tissues in sufficient frequencies to mediate EAE, but the encephalitogenic response is less potent compared to EAE in wildtype C57BL/6 mice. Whether the CD4-deficiency affects the balance of conventional and regulatory MOG-specific T cells is not known. In optimal disease models, CD4-sufficient C57BL/6 mice exhibit chronic paralytic EAE, but suboptimal priming conditions will often result in spontaneous recovery and monophasic disease. Indeed, use of Cd4^−/− mice conferred suboptimal priming and provided a model of monophasic disease on the C57BL/6 background. This model was used to show that GMCSF-MOG administration during the recovery phase attenuated a subsequent actively-induced relapse of EAE.

GMCSF-MOG was also an effective intervention in B cell-deficient mice (Tables 2&4, Figures 2B-C&4C-D). These data discount a requisite role for either B cell APC or antibody in vaccine-mediated tolerance and exclude the possibility that neutralizing anti-GMCSF Ab contributed to the inhibition of EAE, particularly when GMCSF-MOG was in corporated into the CFA emulsion. The tolerogenic activity of GMCSF-MOG was not contingent upon in vivo targeting of MOG35-55 to B cell APC and was consistent with previous reports that GMCSF-NAg primarily targeted the covalently tethered NAg domain to myeloid APC rather than B cell APC(14–16). Although these data did not address whether B cells contributed to the tolerogenic activity of GMCSF-MOG, B cells were not required for the underlying tolerogenic mechanism. If B cells did have a role in this mechanism of tolerance, other APC subsets in parallel with B cells were fully sufficient to support the regulatory action of GMCSF-MOG.

Conclusion

GMCSF-NAg vaccines represent a unique class of tolerogenic vaccine fundamentally distinct from other peptide or protein-based mechanisms of tolerance induction. The unique activity of GMCSF-NAg is that this class of vaccine retains potent tolerogenic activity when deliberately introduced into a CFA-primed lymphatic drainage. This unique activity is associated with the ability of GMCSF-NAg to prime inflammatory myeloid APC and simultaneously target NAg for presentation by those APC to facilitate inhibitory MHCII-restricted antigen presentation, based in part upon an inhibitory IFN-γ/ NO axis. These studies thereby broach the unique concept of an inflammation-dependent, NAg-specific tolerogenic vaccine.

Abbreviations

BM

bone marrow

DC

dendritic cell(s)

EAE

experimental autoimmune encephalomyelitis

GMCSF-MOG

GM-CSF fused to the myelin oligodendrocyte glycoprotein 35–55 peptide

GMCSF-PLP

GM-CSF fused to proteolipid protein 139-151 peptide

MHCII

major histocompatibility complex class II

MOG

myelin oligodendrocyte glycoprotein

MS

multiple sclerosis

nmole

nanomole(s)

NAg

neuroantigen(s)

PLP

proteolipid protein