MHC-Independent Genetic Factors Control the Magnitude of CD4⁺ T Cell Responses to Amyloid-β Peptide in Mice through Regulatory T Cell-Mediated Inhibition

Alzheimer’s disease (AD) is a severe neurodegenerative disorder characterized by progressive loss of memory and cognitive functions. Accumulation of amyloid-β peptide (Aβ), both as soluble oligomers and as compact fibrillar aggregates in neuritic plaques and diffuse deposits, seems to play a key role in initiating the pathogenic events in AD. Active immunization against Aβ has emerged as a promising therapeutic strategy that can prevent Aβ deposition and reverse cognitive decline in murine models (1–3). However, an Aβ vaccination trial (AN1792) in human AD patients was prematurely interrupted due to the occurrence of meningoencephalitides in 6% of the cases, supposedly related to inappropriate activation of Aβ-specific T cells (4, 5). In contrast, although Ab responses may control Aβ plaque burden (6), data suggest that Aβ-specific CD4⁺ T cell responses can reverse cognitive decline and synaptic loss in AD mice (7, 8). Altogether, these preclinical and clinical studies suggest that Aβ-specific T cell responses display multiple functions with complex outcomes, which could be either detrimental or beneficial, depending on the modulation of the reactivity and functionality of Aβ-specific Th cells.

T cell reactivity to Aβ varies significantly among individuals bearing different MHC haplotypes. Differing levels of Aβ-specific T cell responses have been observed in PBMCs of human subjects, and were specific for various Aβ-derived epitopes restricted to distinct HLA-DR alleles (9, 10). Similarly, distinct T cell epitopes have been identified in SJL (H-2^s haplotype) and C57BL/6 mice (H-2^b), which differ significantly in their propensity to develop Aβ-specific CD4⁺ T cell responses (11). Thus, both mouse and human studies suggest an association between MHC class II (MHC-II) alleles and Aβ immunogenicity. However, it is not known whether additional genetic parameters critically influence the magnitude of Aβ-specific T cell responses.

In this study, we examined whether non-MHC genetic factors modulate the strength of CD4⁺ T cell responses to Aβ. The comparative study of these responses in mice with different H-2 haplotypes or congenic at the H-2 locus indicated that the magnitude of Aβ-specific T cell responses induced by vaccination is genetically determined by both MHC-dependent and -independent factors. Among the latter, the intrinsic ability to develop Aβ-specific regulatory T cell (Treg) responses varies between individuals and is a crucial parameter modulating CD4⁺ T cell responses to Aβ in both nonpathological conditions and in the course of AD.

C57BL/6 (H-2^b haplotype), BALB/c (H-2^d), SJL/J (H-2^s), DBA/1 (H-2^q), and CBA/J mice (H-2^k) were purchased from Elevage Janvier (Le Genest Saint Isle, France). C57BL/6 mice congenic for the H-2^s haplotype (strain B6.SJL-H2^s C3^c/1CyJ) were obtained from the Jackson Laboratory (Bar Harbor, ME). APPPS1 transgenic mice (Thy1-APP^KM670/671NL; Thy1-PS1^L166P) on the C57BL/6 background (12) were kindly provided by Prof. Mathias Jucker (Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen, Germany). Animals were bred and maintained under strictly monitored specific pathogen-free conditions. Experiments were conducted in compliance with the French legislation and European Union recommendations.

All peptides were purchased from GeneCust (Dudelange, Luxembourg). Human Aβ1-42 peptide was dissolved at 40 mg/ml in DMSO and stored at −20°C. Overlapping Aβ-derived 15-mer peptides were dissolved at 3 mM in DMSO, before final dilution at 5 or 15 μM in RPMI 1640 medium for in vitro T cell stimulation. OVA (Sigma-Aldrich, St. Louis, MO), OVA peptides (Ova^233–248, PFASGTMSMLVLLPDE and Ova^323–339, ISQAVHAAHAEINEAGR), and proteolipid protein (PLP) peptides (PLP^139–151, HCLGKWLGHPDKF and PLP ^178–191, NTWTTCQSIAFPSK) were dissolved at 25, 1, and 2.5 mg/ml, respectively, in distilled water, and aliquots were stored at −20°C.

Before immunization, Aβ1-42 and PLP peptides were extemporaneously diluted to 2 mg/ml in 12.5 mM Tris·HCl (pH 7.4). Six- to eight-week-old mice were immunized by footpad injections with 100 μg Aβ1-42, 50 μg PLP peptides, or 1.25 mg OVA emulsified in an equal volume of CFA. In some experiments, mice were boosted 2 wk later with 60 μg Aβ1-42 emulsified in an equal volume of IFA.

At day 4 before immunization, mice received an i.p. injection of ∼150 μg anti-CD25–depleting mAb (clone PC61). To assess depletion efficiency, Tregs were quantified in blood samples the day before immunization by a combination of intracellular and surface staining using PE-conjugated anti-Foxp3 (FJK-16S; eBioscience, San Diego, CA) and biotinylated anti-CD25 (7D4; BD Biosciences, Franklin Lakes, NJ) Ab, followed by allophycocyanin-conjugated streptavidin (BD Biosciences). Fluorescence data were collected on a FACSCalibur flow cytometer (BD Biosciences) and analyzed using FlowJo software (Tree Star, Ashland, OR).

Ten days after the last immunization, popliteal and inguinal draining lymph nodes (DLNs) were harvested and mashed into single-cell suspensions. When indicated, CD4⁺ T cells were purified using the negative CD4⁺ T cell Dynal isolation kit (Invitrogen, Carlsbad, CA). For each sample, the purity of the CD4⁺ T cell population was assessed by labeling with PE-Cy5 anti-βTCR (H57-597), FITC anti-CD4 (L3T4 GK1.5), allophycocyanin anti-CD8 (53-6.7), and PE anti-CD19 Ab (1D3) (BD Biosciences) and analyzed by flow cytometry. Purified populations contained >90% CD4⁺ T cells.

For T cell-proliferation assays, splenocytes from nonimmunized syngeneic mice were used as APCs. RBCs were lysed in Tris-buffered ammonium chloride, and splenocytes were treated with mitomycin (50 μg/ml final) for 40 min at 37°C, followed by extensive washing. A total of 2 × 10⁵ DLN-derived cells or 0.8–1 × 10⁵ DLN-derived CD4⁺ T cells was stimulated with 1 × 10⁵ syngeneic APCs/well, in the presence of 5 or 15 μM peptides. For measurement of Ag-induced T cell proliferation, cells were pulsed at day 3 with 1 μCi [³H]thymidine/well, and [³H]thymidine incorporation was measured 16 h later.

Nitrocellulose-based 96-well plates (Millipore, Billerica, MA) were coated with 0.25 μg/50 μl/well of anti-mouse IFN-γ mAb (BD Biosciences) for 2 h at 37°C. Plates were washed and blocked with RPMI 1640 medium containing 10% FCS for ≥2 h at 37°C. DLN-derived cells or purified CD4⁺ T cells from individual mice were seeded at 2 × 10⁵ or 0.8–1 × 10⁵ cells/well, respectively, and stimulated with mitomycin-treated syngeneic splenocytes (2 × 10⁵ cells/well) in the presence of 5 or 15 μM antigenic peptides or with 0.4 mg/ml Con A as both a positive control and an internal reference of cell viability for normalization of data. Plates were incubated at 37°C (5% CO₂) for 24 h, washed with PBS-0.05% Tween 20, and incubated for 2 h at 37°C with 0.05 μg/50 μl/well biotinylated anti-mouse IFN-γ mAb (BD Biosciences). After washing, alkaline phosphatase-conjugated streptavidin was added and incubated for 90 min (Roche, Basel, Switzerland) (75 mU/100 μl/well) and then washed, and IFN-γ–secreting cells were visualized using NBT/bromo-chloro-indolyl phosphate substrate (Promega). Spots were counted using an automated ELISPOT plate reader. Data were expressed as the number of IFN-γ⁺ CD4⁺ T cells per 10³ Con A-reactive CD4⁺ T cells.

To measure serum anti-Aβ Ab titers in immunized mice, flat-bottom Nunc MaxiSorp ELISA plates were coated overnight at 4°C with Aβ1-42 peptide (1 μg/ml in 0.1 M NaHCO₃ [pH 8.3]). Plates were then washed twice with PBS-0.05% Tween 20 and blocked with PBS-1% BSA for 2 h at room temperature (RT). After two washes, plates were incubated for 2 h at RT with serum samples diluted at 1:1000, 1:3000, 1:9,000, 1:27,000, 1:81,000, and 1:243,000 in blocking buffer. The Aβ1-17–specific mAb 6E10 (0.01 and 1 μg/ml) (Sigma-Aldrich) was used as a positive reference sample. Plates were washed and incubated for 90 min at RT with peroxidase-conjugated goat anti-mouse IgG Ab (Amersham, Little Chalfont, U.K.) and then revealed with O-phenylenediamine/H₂O₂ substrate (Sigma-Aldrich). Results were expressed as relative OD = experimental OD/reference OD for 6E10 mAb in the same plate.

Single-cell suspensions were prepared from inguinal and popliteal lymph nodes and incubated for 15 min with FcR-blocking Ab (2.4G2; BD Biosciences) to avoid nonspecific staining. PE-Cy5–conjugated anti-TCRβ (H57-597), FITC-conjugated anti-CD4 (L3T4), biotinylated anti-CD25 (7D4), and allophycocyanin-conjugated streptavidin (all from BD Biosciences) were used for cell-surface staining. For intranuclear staining, cells were fixed, permeabilized, and incubated with PE-conjugated anti-Foxp3 (FJK-16s; eBioscience), according to the manufacturer’s protocol. Stained cells were analyzed using a FACSCalibur (Becton Dickinson) flow cytometer, and data were processed with FlowJo software (Tree Star).

Murine thymus or brain samples were homogenized in 300 or 500 μl lysis buffer (250 mM NaCl, 5 mM EDTA acid, 0.1% Nonidet P-40, 50 mM HEPES supplemented with 10 mM DTT, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 50 μg/ml phenylmethyl sulfonyl fluoride, 2 mM sodium pyrophosphate, 1 mM sodium orthovanadate), respectively, using Polytron 3100 homogenizer (Kinematica, Lucerne, Switzerland). After incubation for 30 min in lysis buffer, homogenates were centrifuged for 10 min at 4°C (14,000 rpm), supernatants containing the total protein extract were harvested, and protein concentrations were determined using the Bradford assay (BCA Protein Assay Kit; Thermo Scientific, Waltham, MA). One hundred micrograms of total protein extracts was loaded on 10% SDS-PAGE gel, separated by electrophoresis, and transferred to nitrocellulose membrane (Amersham Hybond ECL, GE Healthcare, Little Chalfont, U.K.). After blocking in PBS + 5% nonfat dry milk, membrane was incubated overnight with anti-amyloid precursor protein (APP) mouse Ab (22C11; Millipore). HRP-conjugated goat anti-mouse Ab (GE Healthcare) was used as a secondary Ab and incubated for 1 h at RT. Western blot was revealed by ECL using ultra-sensitive luminol substrate (SuperSignal West Femto Substrate; Thermo Scientific). Equal loadings were confirmed by stripping the membrane and reblotting with a specific Ab to β-actin (AC-15; Sigma-Aldrich). Densitometric quantification of bands was carried out using MultiGauge software (Fujifilm, Tokyo, Japan).

Statistical analyses were carried out with Graphpad Prism software (San Diego, CA), using the Mann–Whitney U test.

To assess CD4⁺ T cell reactivity to Aβ in different MHC contexts and genetic backgrounds, C57BL/6, BALB/c, DBA/1, CBA/J, and SJL/J mice, expressing H-2^b, H-2^d, H-2^q, H-2^k, or H-2^s haplotypes, respectively, were immunized with Aβ1-42 in CFA, and DLN-derived cells were analyzed for in vitro T cell proliferation induced by Aβ1-42. Responses were highly heterogeneous, with C57BL/6 mice (H-2^b) displaying the lowest response and SJL/J mice (H-2^s) exhibiting the strongest response (Fig. 1A). Of note, all five mouse strains raised comparable ELISA titers of anti-Aβ Abs, indicating that differences in T cell responses did not have a significant impact on the levels of Aβ-specific B cell responses (Fig. 1B).

To identify the Aβ-derived CD4⁺ T cell epitopes in each of these H-2 haplotypes, we analyzed T cell proliferation induced by nine overlapping Aβ-derived 15-mer peptides. Although responses to 15-mer peptides were relatively low in most mouse strains, class II-restricted epitopes could be identified in four of the five tested strains. C57BL/6 mice (H-2^b) displayed a weak, but reproducible, T cell proliferation to Aβ16-30 (Fig. 2), as previously described (11). In contrast, lymphocytes derived from SJL mice (H-2^s) displayed a strong proliferative response to Aβ10-24 and the flanking peptides Aβ7-21 and Aβ13-27 (Fig. 2), in accordance with a previous report (11). Lymphocytes from immunized CBA/J (H-2^k) and DBA/1 (H-2^q) mice also proliferated in response to Aβ10-24 and the flanking peptides, although the response was weaker than that of SJL/J lymphocytes. A slight T cell proliferation was also observed with CBA/J-derived lymphocytes in response to Aβ19-33, suggesting that this peptide constitutes another subdominant H-2^k–restricted epitope (Fig. 2). Although a strong proliferation was observed with BALB/c lymphocytes in response to Aβ1-42 (Figs. 1A, 2), none of the 15-mer peptides from our library induced a significant T cell proliferation, suggesting that H-2^d–restricted Aβ-derived epitopes cannot be efficiently presented to CD4⁺ T cells unless they are processed from a longer peptide. Complementary experiments indicated that at least one H-2^d–restricted epitope is located within Aβ1-30 and that Aβ7-21 may constitute one weak epitope (data not shown), in line with previous reports (13–15). Altogether, these data identified distinct H-2–restricted Aβ-derived epitopes with remarkably heterogeneous potencies for inducing CD4⁺ T cell-proliferative responses in the corresponding mouse strains.

To determine whether the weak CD4⁺ T cell responsiveness to Aβ in C57BL/6 mice is exclusively due to poor immunogenicity of I-A^b–restricted epitope(s), we compared Aβ-specific CD4⁺ T cell responses in C57BL/6 (H-2^b), SJL/J (H-2^s), and congenic B6.H-2^S mice. The B6.H-2^s mice allow the presentation, on the C57BL/6 background, of the I-A^s–restricted epitope Aβ10-24 of known high immunogenicity in SJL/J mice. Unexpectedly, CD4⁺ T cells from Aβ-immunized B6.H-2^S mice proliferated only slightly in response to Aβ1-42, much more weakly than did SJL/J-derived CD4⁺ T cells, and with a similar magnitude as C57BL/6-derived lymphocytes. In addition, very little significant proliferation could be reproducibly measured in B6.H-2^S–derived CD4⁺ T cells in response to the I-A^s–restricted epitope Aβ10-24 (Fig. 3A). Thus, C57BL/6 mice expressing an H-2 haplotype permissive for the presentation of the potentially immunogenic epitope Aβ10-24 remain unable to develop a robust CD4⁺ T cell proliferative response to Aβ.

To determine whether different in vitro secondary proliferative responses correlate with actual differences in the in vivo expansion of Aβ-specific effector T cells (Teffs), the latter were directly quantified ex vivo by ELISPOT assay. In line with proliferation data, immunized SJL/J mice displayed twice as many DLN-derived Aβ-specific IFN-γ–producing CD4⁺ T cells than did C57BL/6 mice (Fig. 3B). All of these effectors were specific for Aβ10-24 in SJL/J mice, whereas Aβ16-30–specific effectors from immunized C57BL/6 mice were barely detectable by ELISPOT. Finally, Aβ-specific IFN-γ⁺ CD4⁺ T cells from immunized C57BL/6 × SJL/J F1 mice were mostly specific for Aβ10-24, but the overall number of effectors specific for Aβ was reduced to the same level as in parental C57BL/6 mice.

Altogether, these data suggested that MHC-independent genetic factors dominantly downmodulate Aβ-specific CD4⁺ T cell responses in C57BL/6 mice.

Because Tregs play a key role in controlling immunological responsiveness to self-Ags, we analyzed their potential implication in the differential regulation of Aβ-specific CD4⁺ T cell responses in C57BL/6 and SJL/J genetic backgrounds. We compared the impact of Treg depletion on the magnitude of CD4⁺ T cell responses to Aβ. To achieve transient depletion/inactivation of CD4⁺CD25⁺ Tregs, mice were injected i.p. with anti-CD25 mAb (clone PC61) and immunized 4 d later with Aβ1-42, as described above. Flow cytometry analysis of popliteal and inguinal lymph node cells indicated that PC61 treatment induced the depletion of CD4⁺CD25⁺Foxp3⁺ Tregs and CD4⁺CD25⁺Foxp3⁻ Teffs in both C57BL/6 and SJL/J genetic backgrounds in naive animals (i.e., before vaccination) (Fig. 4A, 4B). Without PC61 treatment, the frequency of both Tregs and Teffs significantly increased in both mouse strains upon vaccination with CFA + Aβ. PC61 treatment before immunization impaired vaccination-induced expansion of Tregs, but it did not alter the expansion of Teffs in both C57BL/6 and SJL/J genetic backgrounds (Fig. 4A, 4B). The extent of Aβ-specific CD4⁺ T cell responses in DLNs was quantified by ELISPOT 10 d after immunization. Transient Treg depletion in C57BL/6 mice induced a 2.2-fold increase in the number of Aβ1-42–specific IFN-γ⁺ CD4⁺ Teffs induced by vaccination. Similarly, Aβ16-30–specific IFN-γ⁺ CD4⁺ T cells increased 2.8-fold following Treg depletion (Fig. 4C). Similarly to C57BL/6 mice, Treg depletion in B6.H-2^S mice induced 3.4- and 2.6-fold increases in the numbers of vaccine-induced IFN-γ⁺ CD4⁺ effectors responding to Aβ1-42 and Aβ10-24, respectively. Interestingly, in contrast to C57BL/6 and B6.H-2^S mice, SJL/J mice showed no increase in the numbers of Aβ1-42– or Aβ10-24–specific IFN-γ⁺ CD4⁺ effectors following Treg depletion (Fig. 4C). These data suggested that although Aβ-specific CD4⁺ T cell responses are naturally downmodulated by Tregs in the C57BL/6 background, SJL/J mice do not develop a measurable Treg-mediated spontaneous inhibition of CD4⁺ T cell responses to Aβ.

To determine whether the differential Treg-mediated inhibition in C57BL/6 and SJL/J backgrounds applies to any T cell response to self-Ags, we analyzed the impact of Treg depletion on PLP-specific CD4⁺ T cell responses in C57BL/6, SJL/J, and B6.H-2^s mice. Transient Treg depletion before immunization with either H-2^b–restricted (PLP^178–191) or H-2^s–restricted (PLP^139–151) peptides significantly increased PLP-specific CD4⁺ T cell responses in C57BL/6 and SJL/J background, respectively (Fig. 5A). To further assess the pattern of Treg-mediated inhibition in C57BL/6 and SJL/J backgrounds, we also analyzed the impact of Treg depletion on CD4⁺ T cell responses to OVA, used as a model non-self Ag, in C57BL/6, SJL/J, and B6.H-2^s mice. Similarly to PLP, transient Treg depletion before immunization with OVA increased the numbers of IFN-γ⁺ CD4⁺ Teffs detected by ELISPOT in response to full-length OVA in all three strains, the I-A^b–restricted epitope OVA^323–339 in C57BL/6 mice, and to the I-A^s–restricted epitope OVA^233–248 in SJL/J and B6.H-2^s mice (Fig. 5B). Of note, although Treg depletion enhanced OVA-specific CD4⁺ T cell responses in all three strains, the increase was less in SJL/J background than C57BL/6 background. Thus, the differential Treg-mediated inhibition of Aβ-specific T cell responses in C57BL/6 and SJL/J genetic backgrounds is not general to any self or non-self Ag and is restricted, at least in part, to certain antigenic specificities. Because Aβ is generated by proteolytic cleavage of APP, APP protein expression in the thymus of C57BL/6 and SJL/J mice was analyzed by Western blot. No difference in APP expression was observed between C57BL/6 and SJL/J mice, although the expression was very faint in the thymus (Fig. 5C, 5D). As expected, brain samples used as positive controls displayed a very strong expression of the lowest m.w. isoform of APP (Fig. 5C). Altogether, these data suggested that MHC-independent genetic factors modulating Aβ-specific CD4⁺ T cell responses include the propensity to develop Treg-mediated inhibitory responses to Aβ, which varies critically between individuals of different genetic backgrounds.

To assess the impact of Treg-mediated inhibition on CD4⁺ T cell responses to Aβ in the context of AD, we analyzed the effect of Treg depletion in 4-mo-old Aβ-depositing double-transgenic APPPS1 C57BL/6 mice. Similarly to results in C57BL/6 and SJL/J mice (Fig. 4A, 4B), flow cytometry analysis of DLN-derived cells indicated that PC61 induced the depletion of both CD4⁺CD25⁺Foxp3⁺ Tregs and CD4⁺CD25⁺Foxp3⁻ Teffs in naive APPPS1 mice and wild-type (wt) littermates. PC61 treatment before immunization impaired vaccination-induced expansion of Tregs, but did not alter the expansion of Teffs, in both wt and APPPS1 mice (Fig. 6A). Control PBS-injected APPPS1 mice and wt littermates displayed similar numbers of Aβ1-42– or Aβ16-30–specific IFN-γ⁺ CD4⁺ Teffs in response to Aβ vaccination (Fig. 6B). As expected, transient Treg depletion before immunization in wt littermates induced a 2.9-fold increase in the number of Aβ1-42–specific IFN-γ⁺ CD4⁺ T cells, but no significant difference in the number of Aβ16-30–specific IFN-γ⁺ CD4⁺ T cells was observed. Interestingly, Treg depletion in APPPS1 mice induced a stronger enhancement of Aβ-specific CD4⁺ T cell responses, with 5.9- and 3.1-fold increases in the numbers of Aβ1-42– and Aβ16-30–specific IFN-γ⁺ CD4⁺ effectors, respectively (Fig. 6B). Although the difference in Aβ1-42–specific IFN-γ–producing cells between Treg-depleted wt and APPPS1 mice did not reach statistical significance, a trend (p = 0.08) was observed toward enhanced Aβ-specific CD4⁺ T cell responses in APPPS1 mice compared with wt mice. In contrast, transient Treg depletion before immunization with either H-2^b–restricted epitope PLP^178–191 or full-length OVA increased PLP- or OVA-specific CD4⁺ T cell responses similarly in both wt and APPPS1 mice, respectively (Fig. 6C, 6D). Altogether, these data suggested that APPPS1 C57BL/6 mice spontaneously develop CD4⁺ T cell responses to Aβ, the expansion of which is prevented through Treg-mediated inhibition.

The present study demonstrated that the magnitude of CD4⁺ T cell responses to Aβ vary according to genetic factors that include non-MHC elements. Our data indicated that the heterogeneity of Aβ-specific CD4⁺ T cell responses is related, at least in part, to qualitative and quantitative differences in Aβ-derived epitopes restricted to various MHC-II alleles. Importantly, MHC-independent genetic factors that determine the overall potency of individuals to develop Aβ-specific Treg responses play a critical role in the modulation of CD4⁺ T cell responses to Aβ. Such Treg-mediated inhibitory responses completely downmodulate Aβ-specific CD4⁺ T cell responses that spontaneously arise in the course of AD in a mouse model and prevent their expansion upon Aβ vaccination.

In line with previous reports identifying Aβ-derived T cell epitopes (11, 13–15), dominant CD4⁺ T cell epitopes in SJL/J, C57BL/6, and BALB/c mice were found in Aβ10-24, Aβ16-30, and Aβ1-30 peptides, respectively. Although Aβ10-24 elicited a strong I-A^s–restricted CD4⁺ T cell response in SJL/J mice, the same epitope induced a much weaker response when presented in the H-2^q haplotype (DBA/1 mice), although the overall magnitude of CD4⁺ T cell responses to full-length Aβ1-42 was similar in both strains. Thus, similar Aβ-derived epitopes display qualitative differences in various MHC contexts. Quantitative differences in the number of Aβ-derived epitopes restricted to various MHC-II alleles were also observed. Although Aβ-specific CD4⁺ T cell responses in SJL/J, C57BL/6, and DBA/1 mice were each targeted to a unique epitope, located between residues 10 and 24, 16 and 30, and 10 and 27, respectively, CD4⁺ T cells proliferated in response to both Aβ10-24 and Aβ19-33 in CBA/J mice. In humans, differing levels of CD4⁺ T cell responses to Aβ have been observed in PBMCs and were specific for various Aβ-derived epitopes restricted to different HLA-DR alleles, with the DRB1*1501 allele being highly immunogenic (9, 10). Altogether, these studies indicated that the MHC genotype is an important parameter modulating CD4⁺ T cell reactivity to Aβ. Importantly, the impact of MHC on T cell-related immunogenicity may be more stringent for small Ags compared with large-protein Ags. Indeed, the reduced spectrum of epitopes processed from short antigenic peptides, such as Aβ1-42, likely decreases the probability of generating diverse immunodominant epitopes restricted to various MHC alleles. Hence, MHC genotype has to be taken into particular consideration when evaluating the pathophysiological impact of T cell responses induced by Aβ vaccination.

In addition to the impact of MHC, our study demonstrated that MHC-independent genes control the magnitude of Aβ-specific CD4⁺ T cell responses in mice through Treg-mediated inhibition. H-2^s–restricted CD4⁺ T cell responses of similar Aβ specificities were differentially regulated in C57BL/6 and SJL/J backgrounds. The very mild T cell response to Aβ observed in B6.H-2^s mice, which display the potentially highly Aβ-immunogenic H-2^s allele in the context of C57BL/6 genetic background, indicated that MHC-independent genetic features in C57BL/6 background prevent the expansion of Aβ-specific T cell responses. Analysis of (C57BL/6 × SJL/J) F1 mice, which display the H-2^s allele in the context of a 1/1 mix genetic background, further suggested that MHC-independent genes that downmodulate Aβ-specific CD4⁺ T cell responses in C57BL/6 mice are dominant over SJL/J-derived genetic factors. Conversely, a report by Monsonego et al. (11) showed that although NOD mice display an Aβ-immunogenic MHC context and genetic background, NOD.H-2^b congenic mice are unable to mount T cell responses to Aβ, highlighting the poor Aβ immunogenicity of the H-2^b haplotype. Hence, both MHC-independent genetic features and MHC haplotype critically control the magnitude of Aβ-specific CD4⁺ T cell responses. The MHC-independent dominant inhibitory effect in C57BL/6 mice is related to the propensity for developing Treg-mediated inhibitory responses to Aβ, which varies critically between individuals of different genetic backgrounds. Such Treg-mediated inhibitory responses may have an important implication in the context of AD, because they strongly downmodulate the expansion of Aβ-specific CD4⁺ T cell responses that spontaneously arise in the course of AD in APPPS1 mice. Interestingly, increased suppressive activity of CD4⁺CD25⁺ Tregs has been described in AD patients, but its pathophysiological significance remains to be assessed (16). Genetic background- and sex-related differences in the numbers and functional properties of Tregs were previously reported in mice. Increased frequency of Tregs in male SJL mice compared with female SJL mice, as well as their preferential secretion of IL-10, may contribute to skewed development of Th2-type immune responses (17). Of note, C57BL/6 male and female mice did not show differences in Treg numbers (17). Other studies indicated that BALB/c mice displayed increased frequency of CD4⁺CD25⁺ Tregs compared with C57BL/6 mice, although Tregs from both strains are phenotypically and functionally similar (18). However, CD4⁺CD25⁻ T cells from BALB/c and C57BL/6 mice differ in their susceptibility to Treg-mediated inhibition (18). Interestingly, transient Treg depletion in BALB/c mice induced a notable increase in the numbers of vaccine-induced Aβ1-42– or Aβ7-21–specific IFN-γ⁺ CD4⁺ effectors, suggesting that Treg-mediated inhibition of CD4⁺ T cell responses to Aβ is not restricted to C57BL/6 mice (data not shown). Whether similar quantitative and qualitative differences in Treg responses are also genetically regulated in humans remains to be determined. Functional polymorphisms in the FOXP3/Scurfin gene have been associated with altered susceptibility to autoimmune diseases, including type 1 diabetes (19), systemic lupus erythematosus (20), psoriasis (21), autoimmune thyroid diseases (22), and primary biliary cirrhosis (23). Thus, genetic polymorphisms in genes involved in Treg development and function may translate into altered potencies of developing Treg responses in humans. Finally, in contrast to its effect on Aβ-specific T cell responses, transient Treg depletion increases the number of vaccination-induced PLP-specific or OVA-specific CD4⁺ Teffs in both C57BL/6 and SJL/J genetic backgrounds. Hence, background-related differential Treg-mediated inhibition is not general to any self or non-self Ag, but instead is restricted, at least in part, to certain antigenic specificities, including Aβ. The differential effect of Tregs on Aβ-reactive T cells was not related to differences in thymic expression of APP in C57BL/6 and SJL/J mice. Further experiments will be needed to decipher the mechanistic explanation of this phenomenon.

Although severe complications observed in a limited proportion of AN1792-vaccinated patients were attributed to proinflammatory T cell responses, preclinical murine models did not show evidence of T cell-related side effects. The results of our study shed new light on the retrospective understanding of this discrepancy between preclinical and clinical settings. The critical impact of both the MHC and the dominant background-related Treg-mediated inhibition of CD4⁺ T cell responses to Aβ may explain, at least in part, the occurrence of meningoencephalitis in a fraction of AN1792 patients and underscores the need to develop appropriate preclinical murine models to accurately evaluate the potential impact of vaccine-induced Aβ-specific T cell responses. Transgenic AD mice on the SJL background, which would allow the presentation of the I-A^s–restricted strong T cell epitope Aβ10-24 without significant Aβ-specific Treg-mediated inhibition, may constitute such a relevant preclinical model. The yet unexplained complications of AN1792 led pharmaceutical companies to totally dismiss T cell responses to Aβ and design strategies focused exclusively on Aβ-specific Abs. However, experimental results in mouse models, as well as recent clinical trial data, challenge the efficiency of such approaches and highlight possible associated side effects (24–26). A follow-up study on a group of AN1792 patients suggested that the level of vaccine-induced Abs correlated with the clearance of amyloid plaques but not with clinical response (27). In our study, it is worth noting that Aβ-specific Ab responses were only mildly affected by genetic background, as opposed to T cell responses to Aβ. In parallel, several reports suggested that CD4⁺ T cells may mediate neuroprotective responses in various neurodegenerative conditions, including AD (7, 8, 28–30). Interestingly, previous reports in an optic nerve-injury model suggested that the ability to spontaneously manifest such T cell-dependent protective responses under neurodegenerative conditions may be restricted by naturally occurring CD4⁺CD25⁺ Tregs. Furthermore, this effect of Tregs was not uniform, and their expression in different individuals seemed to be genetically determined (31), reminiscent of our findings.

In conclusion, our study demonstrated that Aβ-specific CD4⁺ T cell responses are critically modulated by MHC-independent genetic factors that determine the propensity to develop Treg-mediated inhibitory responses to Aβ. These data underscore the need to reassess the roles, both pathogenic and/or beneficial, of different T cell subsets in both the pathophysiology and immunotherapy of AD. In this context, extensive analyses aimed at deciphering the impact of Tregs on the neuropathological and clinical parameters of AD in APPPS1 mice are underway.

We thank Mathias Jucker for generously providing the APPPS1 mice and for helpful discussions, Claude Carnaud for helpful comments and discussion and for critical reading of the manuscript, Thomas Chaigneau for technical support, and Delphine Muller and Tatiana Ledent for assistance with mouse husbandry.

The authors have no financial conflicts of interest.