Introduction
A number of anti-CD20 therapeutic antibodies are now successfully used to treat B cell lymphomas and CLL [
1,
2]. The CD20 membrane-spanning 4A molecule is an unglycosylated phosphoprotein (33–37 kDa, 297 amino acids) encoded by the
MS4A1 gene and expressed by B cells from the early pre-B cell to the late B cell stages. Pro-B cells do not express CD20. CD20 disappears when B cells differentiate into plasma cells [
3‐
5]. CD20 is involved in the regulation of intracellular calcium levels and in B cell signaling, proliferation, and differentiation [
6‐
9]. It contains two extracellular loops—one small and one large—containing the epitopes bound by anti-CD20 antibodies [
10,
11].
We and others have shown in a mouse model that CD4
+ T cells play a critical role in the long-term antitumor protection elicited by anti-CD20 treatment [
12‐
14]. T cell depletion and T cell transfer experiments demonstrated that anti-CD20 treatment leads to the development of a potent and specific memory CD4
+ T cell response against CD20
+ tumor cells [
12,
14]. Another study showed that anti-CD20 mAb engages FcγRIIA expressed on dendritic cells leading to the priming of self-reactive tumor-specific CD4
+ T cells [
14]. However, the specific T cell epitopes involved in this process are unknown.
Analyses of the HLA ligandome in healthy donors or patients with B cell malignancies have allowed the identification of self-peptides derived from B cell molecules, in particular CD19 and CD20, that could be recognized by T cells [
15,
16]. Immunogenic MHC I-restricted CD20-derived peptides have also been identified in studies using an in silico approach and in vitro assays based on stimulation of CTLs with candidate peptides [
17‐
21]. Notably, one particular highly immunogenic peptide located in the CD20 transmembrane domain and recognized by CD8
+ T cells, CD20
188–196 (SLFLGILSV), induces the expansion of CTLs in healthy donors and patients. These cells efficiently kill primary tumor cells or cells from cell lines derived from B cell malignancies [
17‐
21]. A strategy developed to detect and expand allo-MHC-restricted T cells reactive to self-tumor antigens has also resulted in the characterization of 20 non-mutated HLA-A*02:01-restricted epitopes from CD20 [
22]. However, these studies have been largely focused on MHC I-restricted CD20 epitopes. Only one study has reported that a CD20 alternative splicing isoform expressed in patients with B cell lymphoma can generate immunogenic CD4
+ T cell epitopes [
23]. Thus, the identification of MHC II-restricted peptides derived from native non-mutated CD20 molecule is still needed to better understand the role of CD4
+ T cells in the long-term response to anti-CD20 treatment.
In this study, we assessed whether human CD20-derived MHC II-restricted immunogenic peptides can be identified using a combination of in vitro binding assays to recombinant human MHC II molecules and subsequent in vivo immunization experiments in human HLA-DR-transgenic mice. We could identify a number of CD20-derived MHC II-restricted long peptides (n = 21) localized in the extracellular, transmembrane and intracellular domains of CD20. These peptides induce in vitro IFN-γ responses in PBMCs from healthy donors (HD) and follicular lymphoma (FL) patients.
Discussion
Our previous studies performed in mice bearing EL4-huCD20 tumor cells have demonstrated that a protective CD4
+ T cell response directed against human CD20 molecule is induced after mAb treatment [
12,
13]. However, the relevant MHC II-restricted T cell epitopes are unknown. Thus, we investigated herein the presence of T cell epitopes in human CD20 and whether T cells directed against CD20-derived peptides can be detected in human PBMCs and splenocytes.
Based on in vitro binding assays to recombinant human MHC II molecules (frequent alleles in European populations, i.e., HLA-DR1; HLA-DR3; HLA-DR4; HLA-DR7) and on in vivo immunization of H-2 KO/HLA-A2
+-DR1
+ transgenic mice, we have identified three pools of human MHC II-restricted T cell peptides located in different domains of the CD20 protein that induce in vitro IFN-γ responses in samples from healthy donors and FL patients (Fig.
4). Of note, some differences were observed between H-2 KO/HLA-A2
+-DR1
+ transgenic mice immunization and in vitro tests of human PBMCs. In experiments using H-2 KO/HLA-A2
+-DR1
+ transgenic mice, the responses induced by peptides localized in the N-terminal intracellular domain of CD20 molecule (huMHC II_Mix 1 and huMHC II_Mix 2, position 22–56) were significantly higher as compared to the other peptides (Fig.
2). By contrast, when both human PBMCs and splenocytes were tested in vitro, the median SFU value per 10
5 cells was significantly higher for pool 133–151 (peptides located within the large extracellular loop of CD20) as compared to the other pools (Fig.
4). These differences could be due to the fact that responses achieved in H-2 KO/HLA-A2
+-DR1
+ transgenic mice result from the presentation of CD20-derived peptides solely by an HLA-DR1 molecule. By contrast, the PBMCs used in ELISPOT assays are derived from individuals with HLA-DR1 and/or HLA-DR3, HLA-DR4, HLA-DR7 haplotypes. The use of H-2 KO/HLA-A2
+-DR1
+ transgenic mice inoculated with EL4-huCD20 tumor cells enables the detection of mouse T cell responses directed against human CD20, a xenogeneic antigen in this setting, in contrast to the assays with human samples in which autologous CD20-derived peptides are used. Nevertheless, this preclinical model represents a valuable tool to establish that peptides selected in silico can be presented in vivo by human HLA-DR1.
Our results also indicate that peptides derived from the huCD20 sequence
133L-
170A (located in the large extracellular loop) are the most immunogenic. This observation is reminiscent of a previous study showing the induction of an antibody response in BALB/c mice vaccinated with a peptide from the human CD20 extracellular loop sequence (CKISHFLKMESLNFIRAHTPYINIYNCEPANPSEKNS PSTQYCY) [
32]. However, although the intensity and frequency of T cell responses against epitopes localized in the large extracellular loop appear to be higher, at least in healthy donors, IFN-γ responses to peptides derived from the intracellular and transmembrane domains (Pools 22–43 and 58–121) were also detected in some individuals. Thus, in addition to MHC I-restricted peptides derived from the extracellular CD20 loop previously described [
16‐
22], we were able herein to define 15 to 20-mer MHC II-restricted T cell epitopes derived from either intracellular, membrane, or extracellular domains of the human non-mutated CD20 protein. Of note, the data obtained in the presence of anti-HLA-DR, -DP, -DQ blocking antibody suggest that these long peptides also have the ability to stimulate CD8
+ T cells in vitro for some individuals as already reported in other studies [reviewed in
33].
We cannot exclude that the immunogenic peptides for which a CD4
+ T cell response is detected in individuals included in our study are different to the endogenously expressed CD20 polypeptide. Low-frequency mutations including SNPs or polymorphisms of the CD20-encoding
MS4A1 gene have been observed in NHL patients [
34‐
37]. It has been suggested that some CD20 tumor-associated mutations could be treatment induced [
37]. Five CD20 alternative splice variants have also been identified in human Epstein–Barr Virus (EBV)-transformed B cell lines and in primary samples of FL, CLL, mantle cell lymphoma (MCL) or diffuse large B cell lymphoma (DLBCL) patients [
38‐
40]. Interestingly, specific T cell responses against a 20-mer peptide derived from one of these CD20 splice variants (D393-CD20) were detected in both lymphoma patients and healthy individuals [
39]. However, it is important to stress that no splice variants were observed in normal B cells from healthy donors in these studies and that the different splice variants and the wild-type CD20 isoform are co-expressed in NHL B cell patients [
39,
40]. It is thus unlikely that an allogeneic T cell response is being observed rather than an autoreactive T cell response in our experimental setting.
Non-mutated self-proteins overexpressed on tumor cells are a source of universal target antigens for inducing tumor-specific T lymphocytes without the need to identify the mutanome of tumor cells. Recent results have demonstrated that thymic deletion prunes but does not eliminate self-specific CD4
+ and CD8
+ T cells, and that some self-peptide/MHC-restricted T cells can be detected at frequencies similar to those of T cells specific for non-self-antigens [
41‐
44]. While the use of such epitopes could be limited by self-tolerant T cell repertoire, therapeutic strategies have been developed to overcome the tolerance of T cells to self-peptides. For example, adjuvants, lentivectors, or inhibitory immune checkpoint blocking molecules can improve the efficacy of self-peptide-based vaccinations [
45,
46]. Moreover, anti-CA125, anti-HER2/neu, anti-MUC1, anti-EGFR mAb treatment can circumvent the tolerance to self-antigens expressed on tumor cells as shown by the increase of the frequency of CD4
+ and/or CD8
+ T cells recognizing peptides derived from the target molecule in cancer patients [
47‐
51].
In our experimental setting, priming of naive T cells in addition to the activation of memory T cells can likely occur during the 7-day expansion. Different studies have shown that T cells specific to a given antigen can be detected in the naive but not in the memory T cell compartment in non-immune donors [
52,
53]. This is consistent with the high diversity of the naive repertoire as compared to the much lower diversity of the memory repertoire, which represents a collection of clones selected during immune responses. In these studies, an amplification step has been used to detect these specific T cells due to their very low frequencies in the naive repertoire. These observations underline the importance of exploring both the naive and memory repertoires to identify anti-CD20-specific CD4
+ T cells that can be manipulated in the context of vaccination strategies. Our data suggest that both naive and memory anti-CD20 T cells can be present in healthy donors.
In conclusion, our results indicate that carefully selected CD20-derived MHC II-restricted peptides make it possible to induce CD20-specific CD4
+ T cell responses in humanized HLA-DR-transgenic mice and in human PBMCs. These peptides could serve as a therapeutic tool in B cell malignancies to improve the antitumor activity of CD4
+ T cells in the context of vaccination strategies by helping CD8
+ T cell response and eventually through direct cytotoxic effector functions [
54]. Furthermore, our results indicate that anti-CD20 T cells present in FL patients exhibit various epitope specificities (Fig.
2; Supplementary Table 2). This finding suggests that any vaccination approach based on the use of CD20-derived peptide pools should include pre-screening of patients who respond to these pools. CD20-derived peptides could also be used ex vivo to develop an adoptive T cell immunotherapy strategy. Finally, they could help in monitoring the anti-tumor T cell responses in patients treated with rituximab or other anti-CD20 antibodies.
Acknowledgements
The authors thank Patricia Rezgui (Hemato-oncology Department, Saint Louis hospital, Paris, France) for blood sample collection and all the patients who contributed to this study. Factor VIII-derived peptides were kindly provided by Sandrine Delignat and Sébastien Lacroix-Desmazes (Cordeliers Research Center, Paris, France). English writing assistance was provided by Jo Ann Cahn, Donald Zack (Johns Hopkins University School of Medicine, Baltimore, MD, USA), and Aditi Varthaman (Cordeliers Research Center, Paris, France).
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.