T-helper (Th)1 and Th17 cells are hypothesized to be pathogenic T cells in RA joints; however, lines of evidence do not fully support this hypothesis, showing polyfunctionality of the T cells. Recent progress in single-cell analysis technology has led to the discovery of a novel helper T-cell subset, peripheral helper T cells, and attracted attention to the previously unappreciated T-cell subsets, such as cytotoxic CD4 and CD8 T cells, in RA joints.
1. A Novel CD4 T-Cell Subset Identified by Emerging High-Dimensional Single-Cell Analysis: Tph Cells
Although the “classical” flow cytometric analysis provides much information on molecular expression at single-cell levels, it depends on a hypothesis-based selection of fluorochrome antibodies for detection. However, recent progress in comprehensive high-dimensional analyses, such as mass cytometry (cytometry by time of flight, CyTOF) and single-cell RNA sequence (scRNAseq), has enabled examination of the expression of huge numbers of molecules or genes in each cell without prior information, which has drastically changed many fields of biological research, including RA. By using CyTOF, Rao et al. identified a novel CD4 T-cell population expressing high levels of programed death-1 (PD-1), an inhibitory receptor induced in T cells continually stimulated with T-cell receptors (TCRs), in the joints of RA patients
[1]. An increase in the PD-1 expression of T cells in RA joints was demonstrated even before the discovery of Tph cells
[2][3], but these studies did not recognize PD-1 as a marker of distinct CD4 T-cell populations. PD-1
high CD4 T cells secrete IL-21 and C-X-C motif chemokine ligand (CXCL)13, a chemoattractant for B cells, and the ligand for C-X-C motif chemokine receptor (CXCR) 5, and are able to induce antibody production from B cells in vitro, similar to follicular helper T (Tfh) cells, which also express PD-1, as well as IL-21 and CXCL13, and help germinal center (GC) B cells in secondary lymphoid organs. However, the PD-1
high CD4 T cells in RA joint do not express CXCR5, unlike Tfh cells. Hence, the PD-1
high CXCR5
- CD4 T cells are called peripheral helper T (Tph) cells. A cluster of T cells corresponding to Tph cells was also detected by scRNAseq analysis of RA synovial cells
[4]. A more recently developed analysis detects two populations of Tph cells, namely PD-1
high CXCL13
low cells and CXCL13
high cells in RA SF
[5], but these are clonally related and, therefore, might reflect different activation statuses of the same subset of T cells.
There are also differences in the expression of transcription factors between Tfh and Tph cells. While Tfh cells express B-cell lymphoma 6 (BCL6) but not B-lymphocyte-induced maturation protein 1 (BLIMP1), the opposite trend is seen in Tph cells, although both express Maf (
Table 1). In relation to this, Tfh and Tph cells are different in their ability to help naive B cells in vitro
[6]. The development of human Tfh cells is experimentally induced by IL-12, Activin A, and transforming growth factor (TGF)-β
[7]. As for Tph cells, Yoshitomi et al. have investigated their differentiation mechanism
[8]. They first reported the presence of a T-cell population that spontaneously produces CXCL13 but is distinct from Tfh cell populations in RA joints, and named it Th
CXCL13
[9], which is likely identical to the later discovered Tph. They subsequently found that in vitro culture of naive T cells in the presence of TCR stimulation and TGF-β induced CXCL13-producing CD4 T cells, and Sox4 was identified as the key transcription factor
[8][10]. However, Sox4 failed to induce CXCL13 production in murine CD4 T cells in vitro
[8], and there has been no study showing the presence of Tph cells in the joints of mouse models of RA. On the other hand, another group showed that in mice, CD4 T cells that had undergone lymphopenia-induced proliferation in vivo showed the PD-1
+ CXCR5
- phenotype. These T cells helped B cells to produce antibodies via IL-21 secretion, similar to human Tph cells
[11]. This might be an alternative mechanism of human Tph cell differentiation, although its relevance needs to be addressed.
Table 1. Similarities and differences between Tfh and Tph cells.
|
Tfh |
Tph |
Similarity |
|
Surface molecules |
PD-1high, ICOS+ |
Cytokines |
IL-21, CXCL13 |
Transcription factors |
Maf |
Difference |
|
|
Location |
Lymphoid organs |
Inflamed tissues |
Chemokine receptors |
CXCR5+, CCR2− |
CXCR5−, CCR2+ |
Transcription factors |
BCL6 |
BLIMP1, Sox4 |
As CXCL13 plays critical roles in GC formation, Tph cells are suggested to be involved in ectopic lymphoid neogenesis (ELN) in RA synovium. In line with this, Tph cells are abundant in the joints of ACPA-positive RA patients
[1], and ACPA production by synovial B cells has been demonstrated
[12]. However, it should be noted that ELNs are detected in the synovium of seronegative RA patients and those with even psoriatic arthritis
[13]. Taken together with the notion that ACPA is detected before the onset of RA
[14] and that the formation of ELNs correlates with local inflammation but not with autoantibody production
[15], ELNs, as well as Tph cells, do not contribute critically to the production of ACPA in the serum of RA patients.
Aside from their helper functions for antibody production, Tph cells are equipped with proinflammatory functions. Rao et al. reported the expression of IFN-γ mRNA in sorted Tph cells
[1]. Others reported, before the discovery of Tph, the presence of CD4 T cells that produce IL-21 and TNF-α in RA SF
[16]. As described above, researchers also noticed a portion of CD4 T cells in RA joints that co-produce IFN-γ, IL-21, and GM-CSF
[17]. Researchers extend this finding by showing that PD-1
high CD4 T cells, namely Tph cells, in RA joints can produce TNF-α, IFN-γ, and GM-CSF, in addition to IL-21 and CXCL13, although there is a heterogeneity in the profile of cytokine production
[18]. Intriguingly, blocking PD-1 signaling in Tph cells enhanced cytokine production and induced self-MHC-dependent spontaneous proliferation in vitro
[18]. Taken together, this suggests a role for Tph cells as proinflammatory effectors in RA synovitis, not only helping B cells in local antibody production but also aiding ELN formation. It should be noted that Tph cells are not specifically detected in RA joints but are also present in other inflammatory tissues, such as the salivary grand of patients with Sjogren syndrome
[19]. An expansion of Tph cells was also observed in the PB of patients with systemic lupus erythematosus
[20]. These findings suggest the involvement of Tph cells in a variety of immune pathogenies and the presence of a common mechanism for their development. Thus, the discovery of Tph cells, which was achieved by using the latest technology, provides novel insights into the role of T cells in the joints of RA patients. It might also change the understanding of immune systems and treatment strategies for various immune-mediated disorders.
2. The Emerging Importance of Cytotoxic CD4 and CD8 T Cells in RA Joints
The high-dimensional single-cell analysis identified additional T-cell subsets in RA joints (
Table 2). Fonseka et al. found an increase in the CD27
-HLA-DR
+ CD4 T cell population in the joints compared to the PB of RA patients
[21]. Notably, these cells express higher levels of cytotoxicity-associated genes, including
PRF1,
GZMB,
GZMA, and
GNLY (genes for perforin-1, granzyme B, granzyme A, and granulysin, respectively) in addition to Th1-related genes, including
IFNG and
TBX21 [21]. Another group showed an increase in eomesodermin-expressing CD4 T cells in RASF
[22]. They later showed, in flow cytometric analysis, an increased expression of cytotoxic molecules, including GZMB, PRF1, Hobit, NKG7, and GPR56 proteins, on CD4 T cells in SF of ACPA
+ RA patients
[5]. A scRNAseq analysis of SF cells identified a cytotoxic CD4 T-cell cluster expressing higher levels of
NKG,
GZMH,
PRF1, and
ZNF683 (Hobit) genes, while
GPR56 was highly expressed in Tph cells. An earlier study on the histology of RA synovium showed the presence of perforin-expressing CD4 T cells
[23]. Thus, CD4 T cells equipped with cytotoxic functions can be pathogenic in RA joints, although the roles of cytotoxic activity in the inflammatory process as well as the types of target cells, which likely express MHC class II, remain to be determined.
Table 2. CD4 T-cell clusters in RA joints reported by Argyriou et al.
[5].
Cluster Name |
Differentially Expressed Genes |
Frequency in SF |
Naive CD4 |
TCF7, CCR7, LEF1 |
− 1 |
CXCL13high Tph |
TNFRSF18, LAG3, CXCL13 |
+++ |
Central memory CD4 |
LTB, ZFP36L2, KLF2 |
± |
Effector CD4 |
CXCR3, TGFB1, KLRB1 |
++ |
Treg |
FOXP3, IL2RA, TIGIT |
++ |
Cytotoxic CD4 |
NKG7, GNLY, GZMH |
+ |
SESN3 CD4 |
TNFAIP3, SLC2A3, CDC14A |
± |
CXCL13low Tph |
PTPN13, PRDM1, NEAT1 |
+ |
Humanin CD4 |
MT-ATP6, MT-ND4, MTRNR2L12 |
− 1 |
EGR1 CD4 |
EGR1, IER2, NR4A1 |
− 1 |
Proliferating CD4 |
STMN1, MKI67, TUBA1B |
+ |
Activated CD4 |
CST3, HLA-DRA, HLA-DPA1 |
± |
CD8 T cells have largely been ignored in the field of RA research, mainly due to a lack of evidence supporting their involvement, in contrast to the case of CD4 T cells as described at the beginning. However, CD8 T cells not only exist in the joints of RA patients; their frequency among T cells is actually higher in the joints than PB
[24]. Most CD8 T cells in the joint show an activated phonotype
[25][26][27]. Here, the results of high-dimensional analysis of RA synovial cells further attract attention to CD8 T cells. Three CD8 T cell subsets with different expression patterns of cytotoxic molecules,
GZMK,
GZMB, and
GNLY, were identified by scRNAseq
[4]. On the other hand, CyTOF analysis identified four subsets of CD8 T cells based on the expression patterns of HLA-DR, PD-1, and PD-1
-HLA-DR
+ populations, including
GZMK+GZMB+ effector T cells and
GNLY+GZMB+ cytotoxic T lymphocytes (CTLs) (
Table 3)
[4]. Notably, the expression levels of IFN-γ of all these CD8 T-cell subsets were higher than those of CD4 T cells in RA joints. More recent analysis additionally demonstrated clonal expansion of the GZMK
+GZMB
+ CD8 T-cell population
[24]. GZMK
+GZMB
+ CD8 T cells have lower cytotoxic potential than GZMK
- GZMB
+ CD8 T-cell populations. However, recombinant enzymatically active GZMK can activate synovial fibroblast to produce IL-6, C-C motif ligand (CCL)2 and reactive oxygen species in vitro
[24]. Consistently, the culture supernatants of synovial CD8 T cells exerted similar effects on synovial fibroblasts, suggesting their potential as pathogenic T cells in RA. Interestingly, such CD8 T cells can be activated to produce IFN-γ in an antigen-independent manner, namely stimulation with IL-12 + IL-15 in vitro, and the profile of gene expression is similar to that of innate T cells, such as mucosal-associated invariant T (MAIT) and invariant natural killer T (iNKT) cells
[24]. This remind researchers of the earlier observation in a mouse experiment that self-specific, innate-like CD8 T cells can be activated by IL-12 + IL-15 independently of TCR signaling
[28], although the antigen specificity of CD8 T cells is unclear in the case of human RA. The high levels of IFN-γ production by CD8 T cells shed light on a role for IFN-γ as an important proinflammatory mediator of RA synovium, which was once thought not to be the case due to the emergence of Th17 cells. Notably, scRNAseq analysis of the synovial macrophages and fibroblasts of RA patients identified cell clusters with a signature of IFN stimulation in both cell lineages
[4]. IFN-γ is also implicated in the generation of RANKL+ effector B cells in RA joints
[29].
Table 3. CD8 T-cell clusters in RA joints reported by Zhang et al.
[4] and Jonsson et al.
[24].
scRNAseq Clusters |
CyTOF Clusters |
DEGs |
Abundance |
GZMK+GZMB- |
PD-1+HLA-DR+/− |
GZMK, NKG7 |
Joint = PB |
GZMK+GZMBlow (effector) |
PD-1−HLA-DR++ |
IFNG, HLA-DRB1 |
Joint > PB |
GZMK−GZMBhigh (cytotoxic) |
PD-1−HLA-DR+ |
PRF1, GNLY |
Joint < PB |
GZMK−GZMB− (naive) |
PD-1−HLA-DR− (?) 1 |
CCR7, IL7R |
Joint < PB |
Regarding the PD-1
+ CD8 T cells identified in the above study
[4], researchers found them capable of producing IL-21 but CXCR5-negative, like Tph cells
[30]. In line with this, an earlier study showed an association between the formation of ELN and the presence of CD40L-expressing CD8 T cells in the synovium
[31]. In vivo depletion of human CD8 T cells in RA synovial explants that were implanted in severe combined immunodeficiency (SCID) mice resulted in disintegration of the GC structure in ELN
[32]. Therefore, although less attention has been paid to them than CD4 T cells, CD8 T cells might also be a candidate for the pathogenic T cells in the joints of RA patients.