Approximately 30% of patients with chronic lymphocytic leukemia (CLL) can be grouped into subsets with stereotyped B-cell receptor immunoglobulin (BcR IG) displaying remarkable similarity in the heavy complementarity-determining region 3 (VH CDR3). Here, we investigated whether the consensus VH CDR3 sequences from CLL stereotyped subsets can be exploited for immunotherapy approaches.
Immunogenic epitopes from the consensus VH CDR3 sequence of the clinically aggressive subsets #1 and #2 and from Eμ-TCL1 mice, which spontaneously develop CLL with BcR IG stereotypy, were identified and used to generate specific HLA class I– and II–restricted T cells in vitro. T-cell reactivity was assayed in vitro as IFNγ production. Bone marrow–derived dendritic cells loaded with the peptides were used as vaccination strategy to restrain leukemia development in the Eμ-TCL1 mouse model.
These stereotyped epitopes were naturally processed and presented by CLL cells to the VH CDR3–specific T cells. Furthermore, we validated the efficacy of VH CDR3 peptide–based immunotherapy in the Eμ-TCL1 transplantable mouse model. Immunization of mice against defined VH CDR3 peptide epitopes, prior to the challenge with the corresponding leukemia cells, resulted in the control of CLL development in a significant fraction of mice, and increased overall survival.
Our data highlight the immunogenicity of stereotyped VH CDR3 sequences and support the feasibility and efficacy of their use for novel cancer vaccine in CLL. Such approach has the advantage to generate “off-the-shelf” therapeutic vaccines for relevant groups of patients belonging to stereotyped subsets.
See related commentary by Seiffert, p. 659
This article is featured in Highlights of This Issue, p. 657
We here report in vitro and in vivo data showing that B-cell receptor (BcR) stereotypy, a characteristic feature of chronic lymphocytic leukemia (CLL), whereby groups of patients share (quasi)identical BcR, can be exploited for immunotherapy approaches that can be quickly transferred into the clinical arena. We show that immunogenic epitopes can be isolated from the consensus heavy complementarity-determining region 3 (VH CDR3) sequence of both human and murine stereotyped CLL, efficiently processed and presented by CLL cells and effectively recognized by specific T cells. Immunization of Eμ-TCL1 CLL mouse model reduced leukemia development and increased overall survival of the animals.
Our data highlight the immunogenicity of stereotyped VH CDR3 sequences and support the feasibility and efficacy of their use for novel cancer vaccines in CLL. The “public” nature of stereotyped BcR implies that such approach has the advantage to generate “off-the-shelf” therapeutic vaccines targeting groups of patients with CLL rather than only individual cases.
Chronic lymphocytic leukemia (CLL), the most common adult leukemia in the Western world, remains incurable despite remarkable progress achieved in recent years with the approval and use of novel targeted therapies. CLL pathogenesis critically relies on stimuli from the microenvironment, including stimulation through the B-cell receptor (BcR). In line with this, CLL is uniquely characterized by the existence of quasi-identical BcRs among unrelated patients displaying restricted antigen-binding sites, including high amino acid identity within the heavy complementarity determining region 3 (VH CDR3), the main determinant of antigen specificity. This so-called “BcR stereotypy,” occurring in approximately 30% of patients, suggests that a finite number of antigenic elements may select the malignant clones (1–6).
Patients expressing particular stereotyped BcRs are categorized into subsets (7–9) with consistent clinical presentation and outcome as well as shared biological features (3, 7, 9–16). The most frequent subsets, 1 [clan I IGHV genes/IGKV1(D)-39] and 2 (IGHV3–21/IGLV3–21), are clinically aggressive (13, 14, 17) and represent approximately 8%–10% of all patients needing treatment.
Tumor vaccines are widely explored as a therapeutic strategy for cancers with different histologic origins. Limited clinical responses (15%–20%) have been registered in patients undergoing vaccination with tumor-associated antigens (TAA), although vaccine-specific T-cell responses were elicited in most patients. The suboptimal efficacy of cancer vaccines is multifactorial, including an immunosuppressive tumor microenvironment and immune evasion mechanisms by tumor cells (18, 19). Moreover, the usage of “self” TAAs that are expressed also by normal cell counterparts, for which tolerogenic immune responses are detectable in patients with cancer, may represent an additional limitation to the clinical efficacy of cancer vaccines (20, 21). Neoantigens, that are generated by cancer-associated nonsynonymous genetic mutations, display high immunogenicity and tumor specificity, thus representing optimal target molecules for immunotherapy (22–25). In CLL, stereotyped VH CDR3 sequences on the clonotypic BcR immunoglobulin (IG) are, in principle, ideal candidate neoantigens for the development of immunotherapy interventions being universal and potentially off-the-bench, at least for each patient subset, hence avoiding the need to design individual peptides as previously envisioned in B-cell malignancies (26).
Here, we developed an active immunotherapeutic approach targeting the IG VH CDR3 of CLL by defining peptide epitopes specifically binding to defined HLA alleles. Peptide-specific T-cell responses were revealed, including recognition of the autologous leukemia cells expressing the defined IG VH CDR3, indicating that these molecules play a role as target neoantigens for tumor-specific T-cell responses. In a preclinical mouse model of CLL, in vivo targeting of the murine stereotyped VH CDR3 was effective in inhibiting the progression of the malignant B-cell clone.
These results indicate that stereotyped VH CDR3 sequences may represent promising candidate neoantigens for the development of novel immunotherapy interventions for large subsets of patients with CLL, laying the basis for translation into the clinic.
Materials and Methods
Primary CLL samples
Blood samples were collected from patients with CLL after written informed consent, as approved by the institutional ethics committee of Ospedale San Raffaele (Milan, Italy) and the G. Papanikolaou Hospital (Thessaloniki, Greece), and the studies were conducted in accordance to the Declaration of Helsinki. Patient clinicobiological data are listed in Supplementary Table S1.
Peptide design and synthesis
VH CDR3 motifs from subsets 1 and 2 were identified through sequencing and bioinformatics analysis of clonotypic IGHV-IGHD-IGHJ gene rearrangements (5, 6). These motifs were used to identify in silico, through the SYFPEITHI (27) and NetMHCII 2.2 algorithms, (http://www.cbs.dtu.dk/services/NetMHCII-2.2), long peptide epitopes (15-mer) with high binding score to MHC class II molecules. Long peptides were initially selected for efficient binding to HLA-DR*03, *04, *11, and *13 molecules, representing the most frequent HLA-DRβ1 alleles expressed by patients with CLL belonging to different subsets (Supplementary Table S2). The binding of these epitopes to other HLA-DRβ1 alleles expressed by the patients was also verified (Supplementary Table S2). Importantly, these long peptides contained minimal HLA class I–binding specific epitopes (9-mer petides; Supplementary Table S2).
Murine VH CDR3 motifs from Eμ-TCL1 mice (28) were used to in silico identify peptide (9-mer) with high binding scores to MHC class I molecules expressed by C57BL/6 mice and containing minimal MHC I–binding specific epitopes (H2Db, H2Kb).
Peptides were synthesized at 90% purity by JPT Innovative Peptide Solutions, dissolved in DMSO (Merck) at 5 mg/mL and stored at −80°C. HLA typing of 14 patients with CLL belonging to stereotyped subsets was assessed by single-stranded oligonucleotide probe-PCR (Table 2).
In vitro stimulation of PBMC from patients with CLL with stereotyped BcR-derived peptides
Peripheral blood mononuclear cells (PBMC) isolated from patients with CLL (Supplementary Table S1) were enriched for T cells by depletion of the CD3− cells (EasySep Human T Cell Isolation Kit, StemCell Technologies), which were pulsed for 4 hours at 37°C with the peptides defined for each specific subset (Table 1), irradiated (50 Gy) and added at 1:1 ratio to 2 × 105 autologous T cells in 96-well flat-bottomed plates in X-VIVO 15 (Cambrex) plus 10% human serum (EuroClone). T-cell cultures were weekly restimulated with autologous irradiated CD3− cells prepulsed with VH CDR3–derived peptides. Recombinant human IL-2 (30 IU/mL; Proleukin, Novartis) and recombinant human IL15 (5 ng/mL; Biosource) were added to the cultures 2 days following peptide stimulation and fresh medium with cytokines was added to T lymphocytes every 3 days. Starting from 3 weeks of culture, the reactivity of the in vitro expanded T cells (2.5 × 104 cells) against either antigen-presenting cells (autologous CD3− PBMCs 5 × 104 cells/well) pulsed with the immunogenic epitopes or autologous CLL cells was assessed by EliSpot assay to detect the release of IFNγ (Mabtech; ref. 29). T2 cells were pulsed with 9-mer peptides and used as target cells to assess the reactivity of T cells against HLA-A2–restricted peptides. Where indicated, the anti-HLA class I W6/32 or anti-HLA class II L243 mAbs (ATCC) were added at 10 μg/mL to T-cell cultures to confirm their MHC restriction. T-cell alone or incubated with phytohemagglutinin (PHA), served as negative or positive controls, respectively. Elispot plates were analyzed using the ELISASpot Assay Video Analysis System (AELVIS).
In vitro stimulation of T lymphocytes with autologous CLL cells
CLL cells were isolated from PBMC through enrichment of B cells (EasySep Human Pan-B Cell Enrichment Kit, StemCell Technologies) and used as APCs for the in vitro stimulation of autologous CD3+ T cells at a T:B ratio of 5:1 in 24-well plates with X-VIVO15 and 5% HS. Fresh medium containing 120 IU/mL rhIL-2 and 10 ng/mL rhIL-15 (Peprotech) was replaced every 3 days. These cultures were stimulated weekly with irradiated autologous CLL cells, and their reactivity was tested starting from the third week of culture by IFNγ EliSpot assay following recognition of either autologous CLL cells, or CD3−PBMCs pulsed with BcR stereotyped-derived peptides as reported above. Target cells preincubated with either anti-HLA class I W6/32 or anti-HLA class II L243 mAbs were used to assess the MHC restriction of antigen-specific T cells.
Eμ-TCL1 CLL animal model
Eμ-TCL1 transgenic mice were provided by J. C. Byrd (Ohio State University, Columbus, OH) on a C3H/B6 background and were backcrossed for >10 generations to C57BL/6N background (30). 8 weeks old wild-type (WT) C57BL/6N mice were supplied by Charles River Laboratories. Mice were maintained in a specific pathogen-free animal facility and treated in accordance with European Union and approval of the Ospedale San Raffaele Institutional Ethical Committee.
In the transplantation experiments, the endpoints were:
(i) disease development, measured twice per month by the collection of PB from the submandibular vein and expressed as the percentage of CD5+/CD19+ double positive population on total cells in the PB.
(ii) overall survival (OS). By “overall survival,” we mean a humanitarian endpoint at which animals have been sacrificed, that is, when the percentage of CD5+/CD19+ double positive population reached 80% of total cells in the PB.
We also calculated the time to reach 50% of disease as additional information to OS.
Prophylactic CLL vaccines with bone marrow–derived dendritic cells
Dendritic cells (DC) were obtained from bone marrow (BM) cells from the femurs/tibias of C57BL6/N WT mice. Cells were cultured in Iscove's modified Dulbecco's medium with 10% FCS, 1% Pen-Strep and rmIL-4 (5 ng/mL), rmGM-CSF (25 ng/mL) (R&D System) for 7 days. On days 3 and 5, cells were seeded in new plates. On day 8, DCs (checked by flow cytometry) were pulsed for 1 hour at 37°C with the indicated IG peptide (5 μg/mL). WT C57BL/6 mice were immunized intradermally with 2.5 × 105 DC pulsed or not pulsed with the corresponding IG peptides, or with PBS. Ten days after DC immunization, WT C57BL/6 mice were challenged with an intraperitoneal inoculum of 5 × 106 Eμ-TCL1 splenic cells containing >85% CD19+CD5+ CLL cells.
Eμ-TCL1 CLL IG peptide–immunogenicity assay in WT mice
WT mice received intradermal injection of 5 × 105 BM-DC, pulsed or not pulsed with the indicated IG peptide. Nine to 11 days later, splenocytes were harvested and restimulated in vitro with the corresponding peptide. A total of 30 × 106 cells were cultured for 4 days in RPMI with 10% FCS, 1% Pen-Strep, 1% Na Pyruvate, 1% hepes, mercaptoethanol 50 mmol/L, 2 mmol/L l-glutamine, and 2.5 μg/mL of peptide. Activated T cells were separated by Lympholyte-M Cell Separation Media (Cedarlane) and plated overnight at 2 × 106/mL with rhIL-2. Purified T cells (20 × 103) from 2 vaccinated WT mice were cultured in duplicates in a 96-well plate in 1:1, 1:5, 1:10, or 1:20 ratio, unless otherwise stated, with (i) leukemic B cells expressing the corresponding VH-CDR3, (ii) a different VH-CDR3, or (iii) alone. After 24 hours, the supernatant of the co-culture was collected and the level of IFNγ was measured with Mouse IFNγ ELISA MAX Deluxe Kit (BioLegend) in duplicates. Cells were collected, stained, and analyzed by flow cytometry.
Statistical analyses were performed with GraphPad Prism 6 Software. The Shapiro–Wilk test was used to assess the normal distribution of the values for each group. For comparison between two groups, the unpaired two-tailed Student t test or the nonparametric Mann–Whitney test was used. For more than two groups, one-way ANOVA followed by Bonferroni post-test or the nonparametric Kruskal–Wallis test followed by Dunn post-test was performed. P < 0.05 was considered significant. Kaplan–Meier estimate was used to assess mice survival and statistical significance was assessed with log-rank testing.
Complete methods are included in the Supplementary Materials.
Consensus peptides from stereotyped BcR IG can elicit autologous T-cell responses against CLL cells
To validate the immunogenicity of VH CDR3 from different stereotyped subsets, multiple synthetic long peptides (N = 13, 15-mers; Table 1) encompassing the predicted epitopes were used for in vitro stimulation of patients' autologous T lymphocytes, and epitope-specific reactivity was assessed through IFNγ release assays. T cells isolated from patients in subsets #1 and #2 (N = 6 and N = 3, respectively) were weekly stimulated by autologous irradiated antigen-presenting cells pulsed with 15-mer long peptides and, starting from the third week of culture, the specific recognition of subset-specific derived epitopes and, if available, of autologous CLL cells, was determined by EliSpot assay.
T cells reactive against subset-specific peptides were isolated from 6/9 (67%) patients (Supplementary Table S2). T cells from patient p2355 showed high level of IFNγ release (135 N. spot/5 × 104 cells; P < 0.01) following incubation with autologous PBMCs preincubated with the VH CDR3.1 peptide (Fig. 1A).
Antigen-specific reactivity was demonstrated by the decrease (from 135 to 38 N. spot/5 × 104 cells, corresponding to 71.9% inhibition; P < 0.01) of IFNγ release following preincubation of target cells with anti-HLA class I (W6/32) mAb (Fig. 1A).
No inhibition of epitope recognition was observed upon preincubation of target cells with the anti-HLA-DR (L243) mAb. VH-CDR3.1 epitope-specific reactivity was detected following incubation of T cells from patients IT01–030-H1 and p7937 with autologous PBMCs pulsed with the peptide (Fig. 1B and C, respectively). 15-mer epitopes used for in vitro stimulation of T cells could be recognized specifically in association with HLA-DR presentation, as shown by the decrease of cytokine production in the presence of L243, from 172.5 to 110 (36.1% of inhibition of IFNγ release) and from 50 to 33 (34% of inhibition of IFNγ release) N. spot/5 × 104 cells for IT01–030-H1 and p7937, respectively (P < 0.01 for both inhibitions; Fig. 1B and C). HLA class I–restricted epitope from subset 1 was also processed and presented, as IFNγ release was inhibited from 172.5 to 96.5 and from 50 to 25 N. spot/5 × 104 cells (54.9% and 55.9% of inhibition), respectively (P < 0.01) through the preincubation of target cells with W6/32 (Fig. 1B and C). These T lymphocytes specifically recognized in HLA-restricted way the autologous CLL cells [from 249.5 to 110 and from 34 to 22 N. spot/5 × 104 cells in the presence of the W6/32 mAb (55.9% and 35.0% of decrease of cytokine release)]. Following preincubation of CLL cells with the L243 mAb, an inhibition of cytokine release of 41.7% and 50.0%, respectively, was observed (from 249.5 to 145.5 and from 34 to 17 N. spot/5 × 104 cells; P < 0.01 in both cases) for patients IT01–030-H1 and p7937, respectively, indicating that leukemic cells can be recognized by antigen-specific T cells upon naturally processing and presenting the epitope derived from stereotyped subset 1.
The HLA class I–restricted reactivity of VH CDR3–reacting T cells was verified through the incubation of T cells with HLA-A2+ T2 cells preincubated with 9-mer peptide derived from either VH-CDR3.1 or VH-CDR3.2 peptides (Fig. 2A–C). IFNγ release was inhibited when T2 cells were preincubated with W6/32 and not with L243 (Fig. 2A–C). Antigen-specific T cells could be isolated also from patient p6090 belonging to subset 2 (Fig. 1D). These T cells were reactive against epitopes presented on autologous PBMCs in association with either HLA class I or class II molecules. The inhibition of cytokine release was 70.9% and 58.5%, with W6/32 or L243 mAbs, respectively (from 120.5 to 35 or to 50 N. spot/5 × 104 cells; Fig. 1D). Interestingly, VH CDR3.2–specific T cells showed reactivity against autologous CLL cells only in HLA class I–restricted modality (from 48 to 15 or to 79 N. spot/5 × 104 cells; 68.7% and 0.0% of reduction of cytokine release following the preincubation of tumor cells with either W6/32 or L243 mAbs, respectively; P < 0.01). The reactivity of these T cells with the 9-mer petide presented by HLA-A2+ T2 cells was inhibited by the preincubation of target cells with the W6/32 mAb (Fig. 2C), suggesting that natural processing of this epitope by leukemia cells occurred preferentially in association with MHC class I molecules (Figs. 1D and 2C).
CLL cells can elicit T cell–mediated responses against VH CDR3–derived epitopes
CLL were found to efficiently process and present epitopes isolated from restricted antigen-binding motifs of stereotyped BcR IGs. T cells isolated from patients 7937, 6090, and IT01–030-H1 following the in vitro stimulation with autologous irradiated CLL cells displayed specific reactivity against both stereotyped BcR epitopes and autologous leukemia (Fig. 1B–D).
To further prove this evidence, CD3+ T cells isolated from patients with CLL were cultured in vitro in the presence of autologous irradiated CLL cells (mixed lymphocyte tumor cell culture, MLTC) and after 3 weeks of culture specific reactivity against both VH CDR3 epitopes presented by either autologous PBMCs or leukemia cells was observed (representative results from patient IT01–030-H1 are shown in Fig. 2D). Moreover, inhibition of cytokine release by T cells occurred following preincubation of target cells with mAbs impairing their recognition of either HLA class I or class II/peptide complexes (Fig. 2D).
Clustering analysis of VH CDR3 sequences from Eμ-TCL1 mice reveals the existence of four subsets of stereotyped BcR IG
We sequenced the IGHV-IGHD-IGHJ genes of leukemic B-cell clones from the PB of Eμ-TCL1 transgenic mice (n = 50) in search of stereotypy, as reported previously (31). Thirty-five out of 50 clones gave a productive rearrangement sequence; 15/35 clustered into four groups of sequences with stereotyped VH CDR3 (set I–IV; Supplementary Table S3; Supplementary Fig. S1). Four members of set I and 2 members of set II shared identical sequences (Supplementary Table S3).
T cells from mice immunized with VH CDR3–derived peptides show cytotoxic activity against the specific Eμ-TCL1 cells in vitro
VH CDR3 motifs were used to identify in silico peptide epitopes (9-mer) with high binding scores to the H2Db and H2Kb MHC class I molecules expressed by C57BL/6 mice (Supplementary Table S4). To assess the antigenicity of the predicted T-cell epitopes, we generated DCs from the BM of C57BL/6 wt mice. Freshly generated DCs harvested on days 7–9 coexpressed MHC II and CD11c molecules, expressed CD80 and 86 receptors and were able to produce IL12 after exposure to lipopolysaccharide for 8 hours (Supplementary Fig. S2). DCs were loaded in vitro for 1 hour with each synthetic peptide and injected intradermally in C57BL/6 wt mice. Mice were sacrificed 10 days after immunization, T cells were isolated from the spleen and seeded, in duplicates, with increasing numbers of: (i) target B cells isolated from the Eμ-TCL1 mouse carrying the corresponding VH CDR3 sequence; (ii) B cells isolated from Eμ-TCL1 carrying a different VH CDR3 sequence (negative control); (iii) RMA T lymphoma cells pulsed with the specific peptide P525 (positive control). After 1 day, supernatants of each co-culture were screened for the production of IFNγ, to assess the antigen-specific recognition of the leukemic cells in vitro. T cells from WT mice immunized with P525 peptide released more IFNγ when challenged with leukemic B cells of the specific TCL1-#525 leukemia than B cells belonging to the unrelated TCL1-#CAD (average level of IFNγ at ratio 1:5, 1,045 ± 125 pg/mL versus 191 ± 17 pg/mL, respectively; P < 0.01). IFNγ production was higher as the T:B ratio increased, but the differences were statistically significant in all four culture conditions (Fig. 3A). Peptides P445, P321, and P393 elicited IFNγ production by primed T cells, albeit not to the same extent. Peptide P445 was the most immunogenic (average level of IFNγ at ratio 1:5: 1,218 ± 189 pg/mL), while P321 and P393 induced a weaker response (535 ± 147 pg/mL and 517 ± 19.8 pg/mL, respectively; Fig. 3A).
B cells from TCL1-#525 died in vitro after 24 hours in the presence of T cells isolated from mice (n = 2) immunized with the corresponding P525 peptide in contrast to unrelated leukemic B cells (Fig. 3B).
These results suggest that the four synthetic peptides designed on the VH CDR3 sequence were efficient in inducing a leukemia-specific T-cell response upon priming in vivo through a DC-based peptide vaccine.
The in vitro response of T cells is specific for stereotyped VH CDR3 sequences
Next we assessed whether T cells primed with peptide P445 could recognize also leukemic B cells expressing a stereotyped IG sequence from an independent mouse. T cells isolated from mice (n = 2) immunized with peptide P445 were seeded with increasing numbers of either B cells from Eμ-TCL1 mouse #445 or B cells from Eμ-TCL1 mouse #115, displaying 86% VH CDR3 amino acid identity (Supplementary Table S3). The levels of IFNγ produced by T cells after 24 hours against leukemic cells from mouse #115 were comparable with those produced by T cells against leukemic cells from mouse #445, at all ratios. These data suggest that the same peptide-based preparation of DC could elicit responses against leukemic B-cell clones belonging to the same stereotyped subset (Fig. 3C).
Prophylactic vaccine with DC loaded with VH CDR3 synthetic peptides impairs the development of TCL1-derived leukemia and increases OS
We then explored the efficacy of immunization with DCs loaded with synthetic VH CDR3–derived peptides in the transplanted TCL1 mouse model. C57BL/6 WT mice received one boost of either DCs loaded with the peptide P350 or DCs alone, 10 days before the intraperitoneal challenge with the specific TCL1-#350 leukemia. Leukemic cells became 50% of total PB lymphocytes after a median time of 99 days and 50 days for mice vaccinated with DCs pulsed with the specific peptide and DCs alone, respectively (P < 0.01). Moreover, among the former group, 6 of 18 mice did not develop the disease (Fig. 4).
This was also true for leukemia TCL1-#266 with the corresponding peptide P266, where all mice benefited from the prophylactic vaccination, and for leukemia TCL1-#445 with the corresponding peptide P445 (Supplementary Fig. S3). Remarkably, when mice were challenged with a leukemic clone carrying a different VH CDR3 sequence, the vaccination had no effect on the progression of the clone, underscoring the specificity of the immune response elicited by the immunization (Fig. 5). The protection by the P350 VH CDR3 peptide loaded DCs was at least in part dependent on CD8+ T cells, as suggested by in vivo depletion experiments (Supplementary Fig. S4).
In mature B-cell lymphomas, the clonotypic BcR IG is unique and distinct from those of the remaining normal B cells, making it a tumor specific neoantigen that has been in the past exploited as target for immunotherapy (26). That said, at variance with all other B-cell lymphomas, a significant fraction (∼30%) of patients with CLL express stereotyped BcR IG with similar (>70% identity) if not identical (100% identity) VH CDR3 amino acid sequence. This feature has led to the grouping of patients with stereotyped BcR IG into homogeneous subsets with consistent profiles. In principle, the stereotyped VH CDR3 sequences, being tumor-specific, but also “shared” among patients, can represent universal neoantigens for the development of therapeutic vaccines for CLL stereotyped subsets and could be made available as off-the-shelf products with no need of individualized preparation.
On these grounds, we hereby show that it is possible to isolate antigen-specific T cells targeting the stereotyped BcR IG of subsets 1 and 2, representing clinically aggressive variants of CLL. Using subset-specific long peptides (15-mer) we were able to generate T cells reactive in HLA class I– and class II–restricted manner. The efficiency of using long peptides in generating antigen-specific T cells, including both CD8+ and CD4+ T cells, has been described in a model of cancer vaccines for patients with cervical cancer (32, 33) as well as with colorectal cancer (29). We show that the epitopes can be efficiently processed and presented in association with MHC molecules by CLL cells and be effectively recognized by VH CDR3–specific T lymphocytes.
We initially selected immunogenic long peptides (15-mers) covering the restricted VH CDR3 motif defining different stereotyped subsets based on binding score to multiple HLA class II molecules expressed by the patients. Nevertheless, these peptides contained also 9-AA HLA class I–restricted epitopes, thus making them able to induce CTL-mediated responses which indeed occurred in 67% of cases. These T-cell responses were both HLA class I– and class II–restricted, suggesting that the usage of long peptides for in vitro stimulation of PBMC can efficiently elicit antileukemia T-cell responses in patients expressing the defined stereotyped VH CDR3 motif. Importantly, VH CDR3–specific T cells recognized autologous malignant B cells in a HLA-restricted manner, indicating that indeed stereotyped VH CDR3 regions contain TAAs that can be naturally processed and presented in association with HLA molecules by CLL cells. These stereotyped regions thus represent immunogenic antigens that can be exploited for immunotherapy treatment of patients with CLL.
IG heavy or light chain variable domains have been previously shown to generate neoantigens in patients with lymphoma (34). This has been further validated in follicular lymphoma, diffuse large B-cell lymphoma and CLL, highlighting a bias for MHC class II presentation of IG-derived neoantigens (34). None of these studies investigated the immunogenic potential of stereotyped VH CDR3 regions, as in our study. In addition, here we exploited long peptides as a tool to optimize the presentation of epitopes by both HLA class I and class II and proved their natural processing and presentation by CLL cells, indicating the feasibility of this approach for immunotherapy.
To advance this concept, we tested our approach in vivo by taking advantage of the Eμ-TCL1 mouse model of CLL that recapitulates the aggressive form of the disease (35, 36), including the presence of BcR IG stereotypy (31), as also documented in our study. As vaccination with neoantigen-loaded DCs has recently been shown to promote antigen-specific T-cell responses in patients with melanoma (37), we exploited this approach, also because DCs have the unique capacity to process and present antigens in the context of both HLA class I and class II molecules and to elicit T cell–mediated responses (38–40). We were able to validate in vitro the immunogenic potency of the stereotyped epitopes from Eμ-TCL1 mice after in vivo immunization of WT mice with DCs loaded with the peptides. Indeed, T cells from immunized mice showed selective reactivity against both autologous leukemic cells and malignant B cells bearing a similar stereotyped VH CDR3.
The efficacy in vivo of our vaccine was evident in a prophylactic model when mice were vaccinated prior to the challenge with the corresponding leukemia cells and showed reduction/delay of CLL development in 50% of the mice and increased OS.
Although these results look promising, one has also to consider that immunosuppression is a hallmark of CLL due to the impairment of T-cell function, leading to increased risk of infections that are a threat to patients but also a potential hurdle for immunotherapeutic approaches. These effects are in part associated to T-cell anergy and exhaustion that can be reverted, at least in vitro, using immunomodulators (41) or the BTK inhibitor ibrutinib (42), likely through the modulation of regulatory T-cell function. Arguably, the concomitant use of these or similar drugs might elicit efficacy of this immunotherapeutic strategy based on stereotyped VH CDR3–derived epitopes. In line with this, clinical studies are now ongoing combining ibrutinib with the infusion of CAR-T cells in CLL and preliminary results show a dramatic benefit both in terms of decrease toxicity and increased efficacy (43) suggesting that the combination may effectively overcome the leukemia-induced immunosuppression (44). In general, the recent early phase success with CAR-T cell therapy in CLL together with the long-standing knowledge of the exquisite sensitivity of CLL cells to allogeneic transplant and lymphocyte donor-infusion underscore the relevance of effective and controlled immune response in the CLL treatment landscape. Besides the vaccination strategy that we hereby propose, results obtained in this study open also the possibility to design adoptive cell therapy strategy in CLL using CAR-T cells specifically targeting stereotyped VH CDR3–derived antigens. Having this unique IG epitope as target might prevent the occurrence of resistance, given the nondispensable role of the expression of the BcR for the viability of malignant B cells (45). Furthermore, CLL vaccines targeting stereotyped VH CDR3–derived antigens could be envisaged to improve the therapeutic efficacy of immune checkpoint blockade, which has not proven beneficial in CLL as monotherapy (46–50).
In conclusion, we validated the immunogenic role of stereotyped VH CDR3–derived antigens isolated from patients with CLL and their possible exploitation for immunotherapy. The advantage of VH CDR3–derived antigens is highlighted by both their high immunogenic potential, being neoantigens, and their expression in groups of individuals belonging to the same stereotyped subsets, potentially allowing for the generation of “off-the-shelf” therapeutic vaccines for patients of defined subsets. These findings open up opportunities to investigate the role of these antigens for developing effective immunotherapy interventions in CLL and warrants further validation also in the clinical arena, given the overall safety of the approach.
No disclosures were reported.
A. Rovida: Conceptualization, formal analysis, writing-original draft. C. Maccalli: Conceptualization, formal analysis, writing-original draft. L. Scarfò: Providing patient samples and clinical information and supervising specific analysis. P. Dellabona: Conceptualization, participated in experimental planning and data analysis, revised the manuscript, designed and supervised the research and wrote the manuscript. K. Stamatopoulos: Formal analysis, provided patient samples and clinical information and supervised specific analysis, revised the manuscript, designed and supervised the research and wrote the manuscript. P. Ghia: Supervision, designed and supervised the research, wrote and revised the manuscript.
This work was in part supported by Associazione Italiana per la Ricerca sul Cancro—AIRC, Milano, Italy (Investigator Grant #20246 and Special Program on Metastatic Disease—5 per mille #21198, to P. Ghia); ERA NET TRANSCAN-2 Joint Transnational Call for Proposals: JTC 2014 (project #143 GCH-CLL) and JTC 2016 (project #179 NOVEL) by the European Commission/DG Research and Innovation to P. Ghia; Bando della Ricerca Finalizzata 2018, Ministero della Salute, Roma, Italy (progetto RF-2018–12368231 to P. Ghia); the project “KRIPIS II ODYSSEUS” funded by the Operational Programme “Competitiveness, Entrepreneurship and Innovation” (NSRF 2014–2020, grant number 14510010 to K. Stamatopoulos) and co-financed by Greece and the European Union (European Regional Development Fund).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.