Presence of T cells directed against CD20-derived peptides in healthy individuals and lymphoma patients

Preclinical and clinical studies have suggested that cancer treatment with antitumor antibodies induces a specific adaptive T cell response. A central role in this process has been attributed to CD4+ T cells, but the relevant T cell epitopes, mostly derived from non-mutated self-antigens, are largely unknown. In this study, we have characterized human CD20-derived epitopes restricted by HLA-DR1, HLA-DR3, HLA-DR4, and HLA-DR7, and investigated whether T cell responses directed against CD20-derived peptides can be elicited in human HLA-DR-transgenic mice and human samples. Based on in vitro binding assays to recombinant human MHC II molecules and on in vivo immunization assays in H-2 KO/HLA-A2+-DR1+ transgenic mice, we have identified 21 MHC II-restricted long peptides derived from intracellular, membrane, or extracellular domains of the human non-mutated CD20 protein that trigger in vitro IFN-γ production by PBMCs and splenocytes from healthy individuals and by PBMCs from follicular lymphoma patients. These CD20-derived MHC II-restricted peptides could serve as a therapeutic tool for improving and/or monitoring anti-CD20 T cell activity in patients treated with rituximab or other anti-CD20 antibodies. Electronic supplementary material The online version of this article (10.1007/s00262-019-02389-7) contains supplementary material, which is available to authorized users.


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][4][5]. CD20 is involved in the regulation of intracellular calcium levels and in B cell signaling, proliferation, and differentiation [6][7][8][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 Benoit Milcent and Nathalie Josseaume are first co-authors.
* Sophie Sibéril sophie.siberil@upmc.fr Extended author information available on the last page of the article protection elicited by anti-CD20 treatment [12][13][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 CD20derived 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][18][19][20][21]. Notably, one particular highly immunogenic peptide located in the CD20 transmembrane domain and recognized by CD8 + T cells, CD20 [188][189][190][191][192][193][194][195][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][18][19][20][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*02:01restricted 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 CD20derived 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-DRtransgenic mice. We could identify a number of CD20derived 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.

Definition of MHC II-restricted human CD20-derived candidate peptides
Bioinformatics tools for epitope prediction (SYFPEITHI; IEDB; BIMAS) applied to the human CD20 sequence deposited in the NCBI database (NP_068769.2) were used to identify MHC II-restricted CD20-derived candidate peptides. The candidate peptides were then screened at ProImmune for high MHC-peptide binding scores with the high-throughput ProImmune REVEAL ® MHC-peptide binding assay (https ://www.proim mune.com/ecomm erce/ page.php?page = revea l_class 2). This binding assay quantifies the ability of test peptides to bind to the human MHC II molecules HLA-DR1 (allele HLA-DRB1*01:01), HLA-DR3 (allele HLA-DRB1*03:01), HLA-DR4 (allele HLA-DRB1*04:01), and HLA-DR7 (allele HLA-DRB1*07:01), and also measures the ability of the bound peptide to stabilize the resulting MHC II-peptide complex. The assay is based on determining the presence or absence of the native conformation of the MHC II-peptide complex, as recognized 1 3 by a specific antibody. Each peptide is given a score relative to a positive control peptide, which is known to bind MHC II molecules with high affinity.

ELISPOT assays
Human PBMCs (n = 26 for HD and n = 9 for FL patients) or spleen cells (n = 7) (2 × 10 6 cells/ml) were cultured for 7 days in RPMI1640 medium supplemented with 10% heatinactivated human serum, 1% l-glutamine, 1% penicillin, and streptomycin (Life Technologies) in the presence of CD20-derived peptide mixture (10 µg of each peptide). A pool of two human MHC II-restricted Factor VIII (FVIII)derived peptides (peptide 1972, KMALYNLYPGVFETV, and peptide 2145, IARYIRLHPTHYSIR) was used as a control of self-antigen-derived peptides that bind to human HLA-DR molecules [27][28][29]. In some experiments, 10 μg/ ml of blocking anti-HLA-DR, DP, DQ monoclonal antibody (clone Tu39; BD Biosciences) were added to the culture. On day 1, 50 ng/ml IL-7 (Peprotech) and on day 3, 50 UI/ml IL-2 (Peprotech) were added to the culture. On day 7, 10 5 PBMCs or splenocytes/well were incubated in IFN-γ ELIS-POT plates (CTL-Europe) for 36 h in serum-free medium with FVIII-or CD20-derived peptides, in the presence or absence of anti-HLA-DR, -DP, -DQ blocking antibody. Each sample was tested in triplicate and the mean value of each triplicate was reported. The positive threshold was set at ≥ 10 SFU per 10 5 cells after subtracting the background noise, as described previously [30]. In other experiments, 5 × 10 5 PBMCs/well from HD (n = 10) were tested ex vivo in the absence of cytokines with either the peptide pool (containing 10 µg of each peptide) or with each of the 21 peptides (10 µg) (1 well/peptide or peptide pool). Human and mouse IFN-γ ELISPOT assays were performed with the Single Color ELISPOT kit according to the manufacturer's recommendations (CTL-Europe). Following completion of the ELISPOT protocol, the plates were air dried in a laminar flow hood prior to analysis. The resulting spots were counted using a computer-assisted ELISPOT image analyzer (S6 Ultra-V Analyzer, CTL-Europe) customized for analyzing ELISPOT assays to meet the objective criteria for size, chromatic density, shape, and color. Spot forming units (SFU) were automatically calculated by the Immunospot SC Suite Software (CTL-Europe) using the SmartCount™ and Autogate™ functions.

Statistical analyses
Statistical evaluation of mouse and human ELISPOT data was performed using non-parametric paired (Wilcoxon) or unpaired (Mann-Whitney) tests, and multiple t tests with Bonferroni correction (indicated in each figure legend). Prism software (version 5, Graphpad, San Diego, CA, USA) was used for statistical analyses. For all statistical tests performed, p values were considered significant if ≤ 0.05.

CD20-derived peptides that bind strongly to human MHC II are immunogenic in HLA-DR transgenic mice
Using the ProImmune REVEAL ® MHC-peptide binding assay, we assessed the binding of 95 overlapping 15-mer human CD20-derived peptides with an offset of 3 amino acids to recombinant human MHC II molecules frequently found in European populations (HLA-DRB1*01:01; HLA-DRB1*03:01; HLA-DRB1*04:01; HLA-DRB1*07:01). Six of these peptides failed in synthesis, and therefore could not be tested. The binding assays revealed frames of densely packed high-scoring peptides (Fig. 1a), and thus clusters of potentially immunogenic epitopes within the intracellular, transmembrane, and extracellular domains of the human CD20 molecule (Fig. 1b).

MHC II-restricted CD20-derived peptides induce in vitro IFN-γ production by T cells from healthy individuals and follicular lymphoma patients
We then investigated whether HLA-DR-restricted CD20derived peptides could stimulate the production of IFN-γ by PBMCs from healthy donors. Since a role for T cells bearing low-avidity TCRs for self-antigens in the immune surveillance of spontaneous spleen B cell lymphoma has been reported [31], we also assessed the IFN-γ production in response to these peptides using human spleen cells.
Three pools comprising all the peptides that activated CD4 + T cells in H-2 KO/HLA-A2 + -DR1 + transgenic mice were designed for human ELISPOT assays (Fig. 3). The first one (termed pool  contained all the peptides of huMHC II_Mix. 1, huMHC II_Mix. 2, and huMHC II_Mix. 3. The pool 58-121 included huMHCII_Mix. 4 and huMH-CII_Mix. 5 peptides, and the last one, termed 133-151, was obtained by pooling huMHC II_Mix. 6, huMHC II_Mix. 7, and huMHCII_Mix. 8 peptides ( Fig. 3; Supplementary   Fig. 1 Screening of immunogenic HLA-DR-restricted CD20-derived peptides. a Cumulative scores of the binding of human CD20-derived peptides to recombinant HLA-DRB1*01:01 (blue), *03:01 (red), *04:01 (green), and *07:01 (purple) molecules as calculated with the ProImmune REVEAL ® MHC-peptide binding assay. High scoring peptides within intracellular, transmembrane, and extracellular domains of the human CD20 molecule were pooled into 9 different mixtures of 18 to 20-mer MHC II-restricted peptides (huMHC II_ Mix 1 to 9) (see also Supplementary Table 1). b Localization of the different MHC II-restricted CD20-derived peptide mixtures (huMHC II_Mix 1 to 3 in red; huMHC II_Mix 4 in green; huMHC II_Mix 5 in dark blue; huMHC II_Mix 6 to 8 in light blue; huMHC II_Mix 9 in pink) Table 1). All three pools of MHC II-restricted CD20-derived peptides induced T cell responses by PBMCs or spleen cells from a number of healthy individuals (Fig. 4a, b; Supplementary Table 2). 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 pool 22-43 (peptides located within the intracellular domain) (Fig. 4a, b). This was observed both with PBMCs and splenocytes. Various response profiles were observed among the HD PBMC samples. 23% exhibited IFN-γ production in response to each of the three pools of peptides. Responses to one or two out of the three pools were detected in 19% and 15% of HD PBMC samples, respectively. IFN-γ production in response to one or two of the three pools was detected in 28% or 57% of splenocytes, respectively (Supplementary Fig. 1). Of note, FVIII peptides used as a control for self-antigen-derived peptides (see "Materials and methods") did not induce IFN-γ production in PBMCs from healthy donors (data not shown).
Finally, we assessed whether CD20-specific T cell responses could be detected in PBMCs from patients diagnosed with high tumor-burden FL and treated with R-CHOP (Fig. 4c). Again, the IFN-γ ELISPOT assays revealed T cell responses against all the peptide pools (Supplementary  Table 2). No significant differences in median SFU per 10 5 cells were detected between the different pools of peptides (Fig. 4c), and about the same percentages of responses to one or three pools were observed (Supplementary Fig. 1). The intensity of IFN-γ response detected in ELISPOT assays with the three pools of peptides was similar between the different types of samples (PBMCs and splenocytes from healthy individuals, and PBMCs from FL patients) (Supplementary Fig. 2).
To further analyze the involvement of CD4 + T cells in IFN-γ responses detected in ELISPOT assays, a monoclonal anti-HLA-DR, -DP, -DQ blocking antibody was added to culture of PBMCs from several healthy donors and from FL patients in the presence of the different pools of CD20derived peptides (Fig. 4d-f). The IFN-γ production could be blocked by the anti-HLA-DR, -DP, -DQ antibody for some, but not all the samples tested ( Fig. 4d-f and Supplementary  Fig. 3). Thus, these results indicate that CD4 + T cells are implicated in IFN-γ response and also suggest that the presence of CD20-derived long peptides could stimulate CD8 + T cells.
All the ELISPOT assays were performed after expanding cells in vitro for 7 days in presence of IL-2, IL-7, and peptide pools. Both naive T cell priming and memory-specific T cell expansion can occur in this setting. To analyze the memory T cell pool specifically, we performed ex vivo ELISPOT assays with the pools of peptides or with individual peptide incubated with 5 × 10 5 PBMCs for 48 h in absence of cytokines. A low background was observed in 6/10 of the HD samples tested when cells were cultured without peptide. HD31 exhibited a marked response to pool 22-43 whereas responses to the other peptide pools were barely detected in these 6 donors (Fig. 5). Interestingly, when single peptides were tested, responses above the baseline could be observed in HD27 ( 58 M-75 G; 142 K-161 Y), HD31 ( 25 S-44 F; 28 K-47 R; 34 M-53 G; 40 P-59 N; 43 S-60 G; 91 G-108 S; 112 L-131 M; 115 G-134 S; 118 I-137 D; 121 S-138 I; 133 L-152 L; 145 H-164 I; 151 S-170 A), and HD33 ( 40 P-59 N; 148 K-167 C). Thus, memory T cells against CD20-derived peptides can be detected in some healthy donors using a short-term in vitro incubation.
Taken together, these results show that T cell responses against MHC II-restricted CD20-derived peptides are detected in samples from healthy donors and FL patients. While these T cells recognize epitopes located in different domains of the CD20 protein (extracellular, transmembrane, and intracellular domains), the intensity and frequency of T cell responses against epitopes in the large extracellular loop appear to be higher, at least in healthy individuals. The frequency of IFN-γ producing CD4 + T cells directed against CD20-derived peptides (huMHC II_Mix 1 to huMHC II_Mix 9) from HLA-A2.1/HLA-DR1-transgenic mice inoculated with EL4-huCD20 tumor cells (+ EL4-huCD20) or from their naive counterparts (non-injected) was evaluated by ELISPOT assays. Results were expressed as SFU per 10 5 CD4 + T cells. Bars represent the mean values from two (huMHC II_Mix 3, huMHC II_Mix 5, huMHC II_Mix 6, huMHC II_Mix 7, huMHC II_Mix 8) or three (huMHC II_Mix 1, huMHC II_Mix 2, huMHC II_Mix 4, huMHC II_Mix 9) independent experiments. The positive threshold (horizontal dotted line) was set at ≥ 10 SFU per 10 5 cells as previously described [30]. *Indicates that IFN-γ responses obtained with CD4 + T cells stimulated with huMHC II_Mix 1 or huMHC II_Mix 2 were significantly higher than those obtained in all other conditions (Multiple t tests followed by Bonferroni correction)

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 CD20derived 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 133 L-170 A (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][17][18][19][20][21][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 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][35][36][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  [38][39][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 coexpressed 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 selfpeptide/MHC-restricted T cells can be detected at frequencies similar to those of T cells specific for non-self-antigens [41][42][43][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 selfpeptides. 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][48][49][50][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-CD20specific 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. l'Informatique et des Libertés » , Paris, France) (CNIL N°DR-2015-237). PBMCs from HD were obtained from CTL-Europe or from the EFS. Spleens were obtained from organ transplant donors at the Hôpital Pitié-Salpêtrière (Paris, France), in accordance with national ethics guidelines governing the use of human tissues (Scientific collecting authorization, government institution "Agence de la biomédecine", Saint Denis, France) (N°PFS14-009). All animal studies were performed in compliance with guidelines from the European Union (EU guideline on animal experiments, European Directive #2010/63/ EU) and the national charter on ethics in animal experiments and were approved by the local Charles Darwin Ethics Committee in Animal Experiments, Paris, France (Authorization Number 01530.02).
Informed consent All FL patients provided informed written consent to participate in this research study (CCTIRS N°14.626 and CNIL N°DR-2015-237). Written informed consents were obtained by the French state agency EFS (www.efs.sante .fr) for the use of blood of anonymous healthy donors for research purposes. Human PBMCs from CTL-Europe were collected from sources that have confirmed that they were obtained from healthy donors with informed consent in accordance with applicable laws. Spleens were collected from organ transplant donors at the Hôpital Pitié-Salpétrière (Paris, France) (written protocol N°PFS14-009, Agence de la biomédecine), following national ethical guidelines (L1232-1, L1232-3; www.legif rance .gouv.fr) regulating the use of human tissues for research purposes.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.