Immune aging results in diminished adaptive immunity and increased risk for autoimmunity. We previously reported a unique PD-1+ CD44highCD4+ T cell population that increases with age in normal mice. In this study, we indicate that the age-dependent PD-1+ CD44highCD4+ T cells develop as unique T follicular (TF) cells in a B cell–dependent manner and consist of two subpopulations, as follows: CD153+ cells preferentially secreting abundant osteopontin on TCR stimulation and CD153− cells that are apparently TCR anergic. These unique TF cells with essentially similar features increase much earlier and are accumulated in the spontaneous germinal centers (GCs) in lupus-prone female BWF1 (f-BWF1) mice. These TF cells showed characteristic cell-senescence features and developed in association with extensive CD4+ T cell proliferation in vivo, suggesting replicative senescence. Although the CD153+ TF cells were defective in proliferation capacity, they were quite stable and specifically responded to self GC-B cells to secret abundant osteopontin, which inhibited B cell receptor–induced GC-B cell apoptosis in f-BWF1 mice. Transfer of CD153+ PD-1+ CD4+ T cells promoted the growth of spontaneous GCs, whereas administration of anti-osteopontin Ab suppressed GC enlargement and anti-nuclear Ab production and ameliorated clinical lupus nephritis of f-BWF1 mice. Current results suggest that senescent CD153+ TF cells generated as a consequence of extensive endogenous CD4+ T cell proliferation play an essential, if not sufficient, role in lupus pathogenesis in lupus-prone genetic background and may also contribute to an increased autoimmunity risk with age.
This article is featured in In This Issue, p.5577
Germinal centers (GCs) are unique niches in secondary lymphoid tissues in which a specific Ab response occurs after Ag immunization (1). In GCs, Ag-specific B cells undergo a series of dynamic processes, including robust proliferation, somatic hypermutation, and class-switch recombination of BCR genes, affinity maturation, and differentiation into memory or plasma cells (2). The initiation of GC reaction requires Ag-specific T cells (3). After Ag immunization, some of these T cells migrate into the follicular B cell region to regulate specific GC reactions and are called T follicular (TF) cells (4–6). Accumulating evidence indicates that functionally distinct types of T cells may be recruited to GCs, including TF helper (TFH) cells, regulatory T cells, and NKT cells (7).
GC reactions may occur spontaneously in various autoimmune diseases, including lupus (8, 9). Lupus disease, which preferentially affects young females in humans, is a multifactorial disease affecting various organs, most commonly glomerulonephritis. A hallmark of the disease is production of autoantibodies against a diverse spectrum of self-Ags, including nuclear components (10, 11). Among several immune disorders in lupus is a defective checkpoint of B cell self-tolerance (12). Although potentially self-reactive B cells are normally excluded from GCs, they are included in GCs in the lupus-prone genetic background, leading to overt autoantibody production (13, 14). Recent studies highlight an important role of TF cells in the spontaneous GC reactions of lupus disease (15–17). However, it remains elusive how such spontaneous GC reactions develop and are regulated.
We previously identified a PD-1+ CD44highCD62LlowCD4+ T cell population that steadily increases age dependently in normal B6 mice (18). The CD4+ T cell population, termed senescence-associated T (SA-T) cells, shows unique genetic signature and functional features, including attenuated TCR-mediated proliferation and highly biased secretion of osteopontin (OPN), suggesting their contribution to reduced acquired immunity and increased inflammatory trait in senescence (18). Most recently, it was reported that unique TF cells with a similar genetic signature to SA-T cells were increased in B6.Sle1b congenic mice (19), which develop anti-nuclear Abs (ANAs) but no clinical disease (11). Although Slamf6 receptor coded in a Sle1b gene cluster was indicated to play a crucial role in the increase of TF cells and ANA production (19), the origin and relation of them with SA-T cells remain unknown.
In the current study, we indicate that SA-T cells develop as a unique TF cell population in normal B6 mice with age and much prematurely in bone fide lupus-prone female BWF1 (f-BWF1) mice, in association with spontaneous GC reactions. The SA-T cells in both aged B6 mice and f-BWF1 mice are developed in association with extensive endogenous CD4+ T cell proliferation in vivo and show characteristic cell-senescence features. We demonstrate that a minor cell population expressing CD153 in the SA-T cells plays a crucial role in spontaneous GC reactions and development of clinical lupus nephritis.
Materials and Methods
New Zealand White, New Zealand Black, New Zealand Black × New Zealand White F1 (BWF1), and C57BL/6 (B6) mice were purchased from Japan SLC, and μMT mice were provided by M. Reth (Albert Ludwigs Universität Freiburg, Freiburg, Germany). The enhanced GFP (EGFP)-OPN reporter mice were generated as a custom order via Unitech (Kashiwa, Chiba, Japan). In brief, the Spp1 chromosomal gene was isolated from bacterial artificial chromosome clones of C57BL/6 mice, and its exon 2 was replaced by a coding sequence for EGFP and loxp-neo-loxp genes. The targeting vector containing the diphtheria toxin A fragment upstream exon 1 was transfected into embryonic stem cells (clone: Bruce 4) derived from C57BL/6 mice (20). The embryonic stem clones carrying the Spp1-deficient allele were introduced into host embryos, and the chimeric mice were crossed with the CAG-Cre transgenic. All mice were maintained under specific pathogen-free conditions at the Centre for Experimental Animals of Kyoto University, and the animal experiments were performed in accordance with the institutional guidelines.
Cells and cultures
WEHI231 (mouse B1 cell) line was provided by K. Iwai (Kyoto University, Kyoto, Japan). Sorted CD4+ T cell subpopulations were cultured in round-bottom 96-well plates (1 × 105 cells) in the presence of coated anti-CD3 (1 μg/well) and soluble anti-CD28 (2.5 μg/ml) Abs, or with autologous B cells (1 × 106 cells) preirradiated with 15 Gy γ-ray. Anti-MHC II (BioLegend), anti–PD-1 (RMP1-14), anti–PD-L1, anti–PD-L2, anti-CD40 Ab (HM40-3; eBioscience), PD-L1-Ig (R&D Systems), or CD30-Ig fusion protein (21) (provided by N. Yoshikai, Institute of Bioregulation, Kyushu University, Fukuoka, Japan) was additionally included in the culture. Recombinant OPN (R&D Systems) was used in an immobilized form. Cell viability was assessed by Cell Titer-Glo luminescent cell viability assay (Promega).
Cytokines and autoantibodies
OPN and other cytokines in the culture supernatants were assessed with ELISA kits. Autoantibodies were also measured with ELISA kits (Shibayagi).
Multicolor flow cytometric analysis and cell sorting were performed with FACS Canto and FACS Aria II/III (BD Biosciences), respectively. The following Abs were used: PE-conjugated anti-ICOS, anti-CD200, anti-CD153, anti-CD121b, anti–IL-21, anti-CD30, anti-OPN, anti–PD-L1, anti–PD-L2, and anti-peanut agglutinin (PNA; BioLegend); anti-CXCR5, Alexa647-conjugated anti–PD-1, FITC-conjugated anti-CD44, anti-GL7, anti-B220, and anti-annexin V, Cy7-conjugated anti-CD95 (BD Biosciences); and FITC-conjugated anti-Ki67 (eBioscience). For intracellular staining of cytokines, cells were incubated with PMA (20 ng/ml) and ionomycin (1 μg/ml) in the presence of brefeldin A for 5 h before the analysis.
Tissues were snap frozen in optimum cutting temperature compound (Sakura). Serial frozen sections were fixed with 95% ethanol, followed by 100% acetone. Immunostained sections were mounted in Mowiol (Calbiochem) and examined under a microscope (Carl Zeiss). Abs included anti-B220, anti-CD4, anti-CD8, anti–PD-1 (BioLegend), anti-GL7 (eBioscience), and anti-CD3 (BD Biosciences). GC areas were assessed with the use of Axio Vision 4.8.2 SP1 software. Tile images of whole tissue sections were generated using the MosaiX tool of Zeiss Axiovision software (Carl Zeiss). For heterochromatin staining, cells were cytospun, fixed for 10 min in 4% paraformaldehyde, and permeabilized with 0.1% Triton X-100 PBS for 10 min. After blocking, the cells were incubated with anti–HP-1β Ab (Millipore), followed by fluorophore Alexa488-conjugated secondary Ab and DAPI.
Cryostat sections were affixed to frame membrane slides for the ArcturusXT LMD system (Life Technologies), fixed in acetone, and stained with biotinylated PNA and HistoGene immunofluorescence staining kit. After dehydration, the regions of interest were dissected on an ArcturusXT system (Life Technologies).
Quantitative PCR was performed with SYBR Green I Master mix (Roche) on a LightCycler 480 (Roche). Cyclophilin and U6 RNA were used as internal controls for mRNA and microRNAs (miRs), respectively. Primer sequences for genes and miR are as follows: Tnfsf8, 5′-GAGGATCTCTTCTGTACCCTGAAA-3′ and 5′-TTGGTATTGTTGAGATGCTTTGA-3′; Spp1, 5′-CCCGGTGAAAGTGACTGATT-3′ and 5′-TTCTTCAGAGGACACAGCATTC-3′; Il1r2, 5′-CCCATCCCTGTGATCATTTC-3′ and 5′-GCACGGGACTATCAGTCTTGA-3′; Ccl3, 5′-TCTGTCACCTGCTCAACATCA-3′ and 5′-CGGGGTGTCAGCTCCATA-3′; Cebpa, 5′-AAACAACGCAACGTGGAGA-3′ and 5′-GCGGTCATTGTCACTGGTC-3′; Cdkn1a, 5′-AACATCTCAGGGCCGAAA-3′ and 5′-TGCGCTTGGAGTGATAGAAA-3′; Cdkn2b, 5′-AATAACTTCCTACGCATTTTCTGC-3′ and 5′-CCCTTGGCTTCAAGGTGAG-3′; Cd30, 5′-GTCCACGGGAACACCATTT-3′ and 5′-CCAACCAGTAGCACCACCAT-3′; Bcl2a1a, 5′-TTTCCAGTTTTGTGGCAGAAT-3′ and 5′-TCAAACTTCTTTATGAAGCCATCTT-3′; and miR-181a, 5′-AGGTCAACATTCAACGCTGTC-3′ and 5′-CTCAACTGGTGTCGTGGAGTC-3′.
Cell transfer and Ab administration
Sorted T cell subpopulations were suspended in temperature-sensitive biogel (Mebiol Gel; Mebiol) at 4°C and directly injected into spleens of anesthetized mice. Mouse anti-OPN Ab (35B6) (22) or isotype-matched mouse IgG (Sigma-Aldrich) was administered i.p. at 500 μg per mouse twice per week.
Statistical analyses were performed by Student t test.
SA-T cells are developed and accumulated spontaneously with age as unique TF cells
Normal mice show a progressive increase in a PD-1+ CD44highCD4+ T cell population with age, termed SA-T cell population, which shows attenuated proliferative response, but secretes large amounts of OPN on TCR stimulation (18). Immunostaining analysis revealed that PD-1+ CD4+ T cells were mainly distributed in splenic follicular regions with a significant accumulation in PNA+ GCs, which spontaneously developed in aged B6 mice (Fig. 1A). The PD-1+ CD4+ T cells significantly expressed CXCR5 similar to TFH cells induced by SRBC immunization (Fig. 1B). Because our previous DNA microarray analysis revealed a marked increase in Cd153 expression in SA-T cells compared with the PD-1− counterpart cells (18), we examined the cell surface expression of CD153 on them. A minor portion of them expressed CD153, which was rarely expressed in SRBC-TFH cells (Fig. 1B). Both CD153+ and CD153− fractions also showed a high expression of TF-related genes such as Bcl6, Cebpa, and Cxcr5, although the levels tended to be less than those in SRBC-TFH cells (Fig. 1C). Both fractions additionally showed a markedly increased expression of proinflammatory genes, such as Spp1, Ccl3, and Il1r2, which were barely expressed in SRBC-TFH cells (Fig. 1C). Regardless of the expression of CD153, the age-dependent development of SA-T cells was remarkably reduced in μMT mice and thus required B cells (Fig. 1D). However, the two subsets of SA-T cells revealed quite distinct TCR responsiveness. Thus, CD153+ SA-T cells secreted large amounts of OPN with less proliferation than PD-1− CD4+ T cells via optimal TCR stimulation in vitro, whereas CD153− SA-T cells showed no detectable proliferation or production of cytokines at all, including OPN, despite higher basal Spp1 expression than PD-1− CD4+ T cells (Fig. 1E). As expected, SRBC-TFH cells produced IL-4 and less amounts of IFN-γ and IL-2, but no OPN (Fig. 1E). We then validated the expression of OPN in CD153+ SA-T cells in vivo with the use of EGFP-OPN knock-in mice. A significant portion of PD-1+ CD4+ cells selectively expressed quite high levels of GFP in 18-mo-old EGFP-OPN mice (Fig. 1F). Moreover, GFP+ cell fraction of the PD-1+ CD4+ cells showed a higher expression of CD153 than corresponding GFP− cell fraction, indicating that CD153+ SA-T cells actually showed a potent Spp1 activation spontaneously in vivo in aged mice (Fig. 1F). Although significant proportions of other cell types in the spleen also expressed GFP, including B220+ B cells and CD11b+ myeloid cells, the extents were by far less than that of PD-1+ CD4+ cells (Supplemental Fig. 1). The results suggest that SA-T cells develop as TF cells with unique functional features in a B cell–dependent manner.
Premature and systemic increase in TF cells highly resembling SA-T cells in bona fide lupus-prone f-BWF1 mice
Above findings prompted us to investigate the TF cells in f-BWF1 mice, which robustly develop spontaneous GCs preceding the overt clinical disease (Supplemental Fig. 2). F-BWF1 mice at as early as 6 mo of age showed a remarkable increase in PD-1+ CD44highCD4+ T cells, and significant portions of them expressed CXCR5 and CD153, similar to SA-T cells; 2-mo-old f-BWF1 mice barely developed such a population (Fig. 2A). Furthermore, the expression of TF cell–related genes, Bcl6, Cebpa, and Cxcr5, as well as genes that were uniquely increased in SA-T cells, Spp1, Ccl3, and Il1r2, was also remarkably increased in both CD153+ and CD153− fractions of the PD-1+ CD4+ T cells (Fig. 2B). Functionally, CD153+ population secreted abundant OPN and Ccl3 with much lower levels of IL-2, IL-4, IL-5, and IFN-γ via TCR stimulation, whereas the CD153− cells produced little or no detectable cytokines at all (Fig. 2C). We failed to detect the secretion of IL-21, which plays an important role in Ag-driven GC reactions (23), in either CD153+ or CD153− TF cell fraction with ELISA. However, intracellular staining indicated a significant IL-21 expression in both fractions, whereas OPN expression was mostly confined to the CD153+ fraction (Supplemental Fig. 3). Furthermore, the majority of OPN+ TF cells did not coexpress IL-21 in aged EGFP-OPN mice (Supplemental Fig. 3). Interestingly, PD-1+ CD4+ T cells were significantly more stable in culture than PD-1− CD4+ T cells (Fig. 2D). The stability of CD153+ cells was further enhanced with no actual cell increase in the presence of coated anti-CD3 Ab; the viability of CD153− cells was barely affected, whereas PD-1− CD4+ T cells were remarkably increased, as expected (Fig. 2D). The results suggested that the apparent TCR unresponsiveness of CD153− cells was unlikely due to activation-induced cell death. At later stages, when clinical nephritis became evident, PD-1+ CD4+ T cells, including CD153+ cells, were significantly detected in the circulation as well as in the tertiary lymphoid structures of the affected kidneys (Fig. 2E). The results suggest that TF cells highly resembling SA-T cells are increased in f-BWF1 mice much earlier than in B6 mice and may recirculate to lupus target organs as the disease advances.
Lupus TF cells and SA-T cells show cell-senescence features in common and increase in association with extensive CD4+ T cell proliferation in vivo
Given the similarity of TF cells in f-BWF1 mice and SA-T cells, we sought for basic cellular features underlying the two cell populations. CD153+, and to a lesser extent CD153−, PD-1+ CD4+ T cells in both f-BWF1 mice and aged B6 mice revealed a remarkably increased expression of Cdkn2b (Ink4b) and Cdkn1a (Cip1), which are typical cell-senescence biomarker genes (24), compared with corresponding PD-1− CD4+ T cells; this was not the case for SRBC-TFH cells (Fig. 3A, left). It is reported that the expression of miR-181a decreases age dependently in human CD4+ T cells (25). The expression of miR-181a was decreased with age also in mouse PD-1− CD4+ T cells; however, PD-1+ CD4+ T cells, in particular CD153+ population, showed much more profound decrease than PD-1− cells in both aged B6 mice and f-BWF1 mice, and accordingly, an increased expression of target tyrosine phosphatase genes such as Ptpn11 (Fig. 3A, right). Furthermore, PD-1+ CD4+ T cells in both groups of mice revealed a remarkable increase in senescence-associated heterochromatic foci in the nuclei (26) (Fig. 3B). Thus, TF cells in both mice show cell-senescence features in common. One of the main causes for cell senescence is extensive cell proliferation, and therefore we examined the proliferative state of CD4+ T cells in vivo. Adult B6 mice at the ages younger than 8 mo had low levels of Ki67+ cycling CD4+ T cells that minimally expressed PD-1; however, 15-mo-old B6 mice showed a significant increase in the Ki67+ CD4+ T cells, and the majority of them expressed PD-1 (Fig. 3C). F-BWF1 mice revealed a remarkable increase in Ki67+ CD4+ T cells as early as 8 mo of age, and again most of these cells showed PD-1 expression (Fig. 3C). Notably, however, CD153+ cells in the PD-1+ CD4+ T cell populations of both mice contained only minimal Ki67+ cells (Fig. 3C). The findings are in agreement with much higher expression of cdk inhibitors, Ink4b and Cip1, in CD153+ cells than in CD153− cells. These results suggest that the increase of PD-1+ CD4+ TF cells is attributable to extensive CD4+ T cell proliferation in vivo, which occurs much earlier and more robustly in f-BWF1 than in B6 mice, eventually resulting in the accumulation of cell-senescent CD153+ cell population.
Senescent CD153+ PD-1+ CD4+ TF cells in f-BWF1 mice produce OPN in response to autologous GC B cells via PD-1–resistant, CD153-costimulated TCR signal
Inasmuch as the development of SA-T cells was B cell dependent, we next investigated the functional interaction of them with B cells in f-BWF1 mice. The CD153+ PD-1+ CD4+ T cells from f-BWF1 mice secreted significant amounts of OPN in the coculture with γ-ray–irradiated autologous B cells, whereas PD-1− and CD153− PD-1+ CD4+ T cells did not and did so only minimally, respectively; among the B cells, the stimulatory activity of CD95+ (GL7+ or GL7−) GC B cells was more potent than that of CD95− B cells (Fig. 4A). The OPN production by CD153+ cells in response to B cells was inhibited significantly in the presence of anti-MHC II Ab cross-reactive with both I-Ad and I-Au and almost completely by soluble anti-CD3 Ab (Fig. 4A), confirming the TCR-mediated activation. However, although GC B cells significantly expressed PD-L1 and PD-L2, the GC B cell–induced OPN production was not affected by anti–PD-L1 or anti–PD-L2 Ab at all (Fig. 4B). Furthermore, OPN production by CD153+ cells in the presence of coated anti-CD3 Ab was not affected by either solid-phase PD-L-Ig or anti–PD-1 Ab, whereas a low level of IL-4 production was significantly inhibited by PD-L-Ig, similarly to the IL-4 production by SRBC-TFH cells (Fig. 4C). In contrast, OPN production by CD153+ cells in response to B cells was significantly enhanced in the presence of CD30-Ig (Fig. 4D). Although cell surface CD30 expression was barely detected in normal splenic cells, PD-1+ CD4+ TF cells showed significant intracellular expression of CD30 after the coculture with B cells (Fig. 4D), suggesting a possible involvement of shed CD30 in the reverse CD153 activation (27). The results suggest that senescent CD153+ PD-1+ CD4+ TF cells in f-BWF1 mice are self-reactive and are activated to produce OPN in response to GC B cells via a unique TCR signal.
OPN is expressed locally in GCs of f-BWF1 mice and inhibits BCR-mediated B cell apoptosis
Because diverse cell types can produce OPN, we then examined whether OPN expression was actually enriched in the spontaneous GCs of f-BWF1 mice in situ. Laser-capture microdissection analysis revealed that PNA+ GC regions in f-BWF1 spleen contained significantly increased Spp1 transcripts compared with PNA− regions, indicating de novo OPN activation in GCs; GCs in SRBC-immunized B6 spleen showed no such accumulation of Spp1 transcripts (Fig. 5A). CD153+ PD-1+ CD4+ T cells showed a remarkably higher expression of Spp1 than GC B cells expressing activation-induced cytidine deaminase, and therefore the transcripts were indeed attributable to the CD153+ TF cells (Fig. 5A). Stimulation of the B cells from f-BWF1 mice with anti-μ Ab in vitro resulted in decreased viability, but the effect was significantly rescued in the presence of immobilized OPN, resulting in the enhanced secretion of anti-dsDNA IgG to an extent comparable to anti-CD40 Ab (Fig. 5B). We confirmed that OPN strongly inhibited BCR-induced apoptosis of a B1 cell line, WEHI231 (Fig. 5C). Although anti-μ Ab stimulation alone resulted in the decrease of Bcl2a1a, which is crucial for follicular B cell survival (28), OPN significantly enhanced the expression of Bcl2a1a (Fig. 5D). The results suggest that OPN produced by CD153+ TF cells in the GCs protects Ag-induced apoptosis of GC B cells.
CD153+ PD-1+ CD4+ T cells secreting OPN play an essential role in lupus nephritis
Correlating with the increase in SA-T cells with spontaneous GCs, aged B6 mice developed significant anti-histone, anti-dsDNA, and anti-dsDNA IgG Abs in the sera, although the levels tended to be lower than f-BWF1 mice of 5 mo old (Supplemental Fig. 4A). We then finally investigated the involvement of CD153+ PD-1+ CD4+ T cells in the lupus pathogenesis in f-BWF1 mice. First, to directly examine the role in spontaneous GC reactions, we isolated PD-1− CD4+ T cells, CD153+ PD-1+ CD4+ T cells, and CD153− PD-1+ CD4+ T cells from 9-mo-old f-BWF1 mice and transferred them into 6-wk-old f-BWF1 mice, which showed minimal GC development. Because of the limited cell numbers, the purified cells were placed in a temperature-sensitive biogel and injected directly into the spleens. At 3 wk after transfer, the recipients of CD153+ PD-1+ CD4+ T cells showed a significant increase in GC B cells compared with those of other two populations, and this effect was attributable to an increase in individual GC sizes rather than GC numbers (Fig. 6A). Thus, CD153+ PD-1+ CD4+ T cells were capable of enhancing GC expansion in f-BWF1 environment. Although it was unfeasible to follow the fates of transferred cells in vivo due to the small cell numbers, we observed a significant expression of CD153 in a minor fraction of PD-1+ CD153− cells following anti-CD3 stimulation in vitro (Supplemental Fig. 4B), suggesting that the development of CD153+ T cells from a PD-1+ CD153− T cell population in vivo might have contributed to the slight increase in GC B cells. We next examined the possible role of OPN. To this end, we administered mouse anti-OPN Ab (35B6) or isotype-matched control mouse IgG to f-BWF1 mice at 500 μg twice per week starting from 16 wk of age. At 24 wk of age, the 35B6-treated mice showed significantly less numbers of large-sized GCs than control mice; concordantly, the development of ANAs was suppressed, with no progression of hypocomplementemia (Fig. 6B). Furthermore, the 35B6-treated mice showed much less glomerular deposition of IgG than control recipients, resulting in ameliorated nephritis scores with minimal glomerular injuries (Fig. 6C). In agreement with the histological findings, the development of proteinuria was also significantly suppressed in 35B6-treated mice at least until plasma anti-OPN activity was attenuated (Fig. 6D). Collectively, these results strongly suggest that CD153+ PD-1+ CD4+ TF cells producing OPN play a crucial role in the development of spontaneous GC reactions and provide an essential, if not sufficient, condition for the development of clinical lupus nephritis in f-BWF1 mice.
Cellular senescence is characterized by irreversible arrest of cell proliferation, grossly altered gene expression often associated with a unique signature called SASP, and relative resistance to apoptosis (24). Notably, senescent cells can be metabolically active and may become foci of tissue reactions by secreting various inflammatory factors (29). Although T cell senescence may lead to the reduced capacity of acquired immunity due to the defective Ag-driven clonal proliferation (24), other possible consequences remain to be verified.
An age-dependent PD-1+ CD44highCD4+ T cell population, termed SA-T cells, shows attenuated TCR-induced proliferation, but is capable of secreting abundant OPN (18). Our current results indicated that SA-T cells develop and increase as TF cells in a B cell–dependent manner with age in association with GCs occurring spontaneously, although the cellular features are quite different from those of Ag-driven, regular TFH cells. Robust development of spontaneous GC reactions is a hallmark of lupus disease (8, 9), and we found that the TF cells associated in lupus-prone f-BWF1 mice highly resembled the SA-T cells, except for their much earlier increase, than SA-T cells in B6 mice. First, both SA-T cells in aged B6 mice and TF cells in f-BWF1 mice consisted of two subpopulations defined by the expression of CD153, as follows: CD153+ cell population producing large amounts of OPN and other inflammatory cytokines such as Ccl3 via TCR stimulation with reduced proliferation capacity, and CD153− cells that were apparently unresponsive to TCR stimulation. Second, both populations showed characteristic signs of cellular senescence, including a remarkable increase in senescence-related genes such as Ink4b and Cip1, high basal expression levels of several inflammatory genes reminiscent of SASP, a profound decrease of SA-miR-181a, prominent senescence-associated heterochromatic foci in the nuclei, and a high stability in culture. These cell-senescence features were more extensive in CD153+ cells than in CD153− cells.
The increase of these unique TF cells was correlated with extensive proliferation of CD4+ T cells in vivo, which occurred spontaneously in aged B6 mice and much earlier in f-BWF1 mice. The actively cycling Ki67+ cells were largely confined to PD-1+ CD4+ cell fraction, but CD153+ cells in the PD-1+ CD4+ T cell population barely expressed Ki67 anymore. Considering the characteristic features of cell senescence in the CD153+ PD-1+ CD4+ T cells, it is quite conceivable that the increase of the TF cells is a consequence of extensive CD4+ T cell proliferation in vivo, namely replicative cell senescence. Notably, the senescent CD153+ PD-1+ CD4+ T cells were quite stable in culture, despite the diminished TCR-induced proliferation capacity. The mechanisms underlying such an extensive CD4+ T cell proliferation occurring spontaneously in vivo remain to be verified. A likely mechanism may be homeostatic CD4+ T cell proliferation, which requires tonic TCR signal from self MHC II (30), because the increase of SA-T cells depends on B cells. Age-dependent decline of T cell genesis in normal mice as well as persisted lymphopenia preceding clinical diseases in autoimmune disorders including lupus may endogenously promote the homeostatic CD4+ T cell proliferation (31).
It seems rather paradoxical that CD153− PD-1+ CD4+ T cells actively cycling in vivo showed no proliferative response to anti-CD3 Ab stimulation at all in vitro with no evidence of activation-induced cell death. Two possibilities may be considered. First, the CD153− PD-1+ CD4+ cells in ongoing proliferative cell cycles are refractory to an additional potent growth signal via TCR. However, the possibility may not explain the complete TCR unresponsiveness of a whole CD153− PD-1+ CD4+ T cell population, because all the cells were not in cell cycles at a given time point. Second, these T cells might have been actively imposed of TCR signal unresponsiveness, or TCR anergized, during the proliferation process. Homeostatically proliferating T cells have an intrinsic self-reactivity (32), and our current results indicated that CD153+ PD-1+ CD4+ T cells of f-BWF1 mice specifically responded to autologous B cells dependent on MHC II and TCR in vitro. Potentially self-reactive T cells may fall into a long-standing state of Ag receptor unresponsiveness after encountering specific Ags, termed immunological anergy, which is considered to play a role in peripheral self-tolerance (33, 34). Therefore, a possibility that CD153− PD-1+ CD4+ T cells are the TCR-anergized cells by the self-tolerance checkpoint is intriguing. If this is the case, CD153+ cells may represent the cells that have escaped from the checkpoint.
The nature of possible self-Ag peptides involved in the activation of CD153+ PD-1+ CD4+ T cells in f-BWF1 mice remains to be elucidated; however, the GC B cells served as the most effective APCs, suggesting that nuclear components captured via BCR and/or TLRs may be involved (35). Curiously, OPN production of CD153+ PD-1+ CD4+ T cells in response to GC B cells was unaffected by PD-1 engagement, although IL-4 production was significantly suppressed. PD-1 signaling is another important checkpoint for potentially self-reactive T cells (36). However, our current results suggest that TCR-induced OPN production of CD153+ TF cells may be mediated via an alternative signaling pathway resistant to PD-1–mediated negative signal, and as such the self-reactive response of CD153+ TF cells may be exempted from PD-1 checkpoint. Notably, the activation of CD153+ TF cells was enhanced by a reverse CD153 signaling via CD30, which is reported to rather inhibit the proliferation and IL-2 production in regular T cells (21, 26). Thus, it seems that CD153 plays a role in a plethora of OPN production by CD153+ TF cells.
Analysis using EGFP-OPN reporter mice suggested that at least a portion of CD153+ SA-T cells in normal aged mice shows spontaneous Spp1 activation at a given time in vivo. Consistently, Spp1 transcripts were accumulated in the spontaneous GCs in situ in f-BWF1 mice, conforming to the de novo activation of these TF cells in GCs. OPN is not a typical T cell cytokine; it is expressed in many other cell types, and the range of effects is very broad, including tissue remodeling, inflammation, stress responses, and autoimmunity (37, 38). We indicated that OPN prevents the BCR-induced B cell apoptosis, most likely by inducing the expression of Bcl2a1a involved in follicular B cell survival (28). Prolonged survival of the GC B cells, which otherwise have high apoptotic rates, in lupus mice may not only enhance autoantibody production, but accelerate the epitope spreading through somatic hypermutations (14). It is reported that OPN is secreted by myelin-specific pathogenic CD4+ T cells and enhances their survival in experimental autoimmune encephalitis (39). Most recently, it was reported that an intracellular form of OPN plays an important role in sustaining regular TFH cells cell autonomously as an intracellular signal adaptor via Bcl-6 stabilization (40). Although it is possible that an intracellular form of OPN also plays a role in supporting the CD153+ TF cells as a common mechanism for TF cells, the secretion of large amounts of OPN is a unique feature of the CD153+ TF cells, which is barely observed in regular Ag-driven TF cells.
We confirmed the role of CD153+ TF cells in spontaneous GC reactions and lupus pathogenesis. First, we indicated that transfer of the CD153+ PD-1+ CD4+ T cells caused significant increase in GC B cells primarily due to enlarged GCs in young f-BWF1 mice, whereas TCR-unresponsive CD153 PD-1+ CD4+ T cells showed minimal effects. Considering the extremely high proliferation and apoptosis rates of GC B cells, the results may be consistent with their enhanced survival of GC B cells via OPN. We further demonstrated that continuous administration of anti-OPN Ab for 2 mo in f-BWF1 mice significantly suppressed the development of large-sized GCs and ANA production and ameliorated the clinical lupus nephritis. Although it is conceivable that the clinical effect of anti-OPN Ab is primarily attributable to the suppression of spontaneous GC reactions, additional effects on OPN-mediated inflammation in kidney tissues remain possible, because CD153+ PD-1+ CD4+ T cells are also accumulated in the target tissue. In either case, these results strongly suggest that CD153+ PD-1+ CD4+ T cells capable of producing OPN play a crucial, if not sufficient, role in lupus pathogenesis; development of clinical lupus disease in f-BWF1 mice certainly requires other immune cell disorders, including B cells (11). The conclusion is consistent with our previous report that administration of anti–PD-1 Ab, but not anti–PD-L1 Ab, ameliorates the lethal lupus disease in f-BWF1 mice via cell-ablative effect on PD-1+ CD4+ T cells (41). Thus, CD153+ TF cells per se or OPN secreted by them may be a potential therapeutic target for controlling human SLE.
In addition to aging and lupus, we previously reported that SA-T cells also increase rapidly in systemic leukemia, causing an acute reduction of peripheral CD4+ T cell pool (18), and our results indicate that the effect was also associated with robust CD4+ T cell proliferation (Y. Fukushima, Y. Nakashima, and N. Minato, unpublished observations). We propose that an extensive CD4+ T cell proliferation in vivo results in the increase of PD-1+ CD4+ TF cells, with eventual accumulation of senescent CD153+ cells. These senescent TF cells may take a significant part in an increased risk for systemic autoimmunity and inflammation with age and overt autoimmune diseases in specific genetic predispositions. Intervening in their accumulation and/or function may provide novel strategies for controlling human immunosenescence and lupus disease.
We thank Drs. N. Yoshikai, K. Iwai, and M. Reth for providing research materials.
This work was supported by a Grant-in-Aid for Scientific Research on Innovative Areas from the Japan Society for the Promotion of Science; Special Coordination Funds for Promoting Science and Technology from the Japanese Government; the “Formation of Innovation Center for Fusion of Advanced Technologies” program of Astellas Pharma, Inc.; the Uehara Memorial Foundation; and Ono Pharmaceutical Co., Ltd.
The online version of this article contains supplemental material.
Abbreviations used in this article:
The authors have no financial conflicts of interest.