BCR-associated factors driving chronic lymphocytic leukemia cells proliferation ex vivo

A chronic antigenic stimulation is believed to sustain the leukemogenic development of chronic lymphocytic leukemia (CLL) and most of lymphoproliferative malignancies developed from mature B cells. Reproducing a proliferative stimulation ex vivo is critical to decipher the mechanisms of leukemogenesis in these malignancies. However, functional studies of CLL cells remains limited since current ex vivo B cell receptor (BCR) stimulation protocols are not sufficient to induce the proliferation of these cells, pointing out the need of mandatory BCR co-factors in this process. Here, we investigated benefits of several BCR co-stimulatory molecules (IL-2, IL-4, IL-15, IL-21 and CD40 ligand) in multiple culture conditions. Our results demonstrated that BCR engagement (anti-IgM ligation) concomitant to CD40 ligand, IL-4 and IL-21 stimulation allowed CLL cells proliferation ex vivo. In addition, we established a proliferative advantage for ZAP70 positive CLL cells, associated to an increased phosphorylation of ZAP70/SYK and STAT6. Moreover, the use of a tri-dimensional matrix of methylcellulose and the addition of TLR9 agonists further increased this proliferative response. This ex vivo model of BCR stimulation with T-derived cytokines is a relevant and efficient model for functional studies of CLL as well as lymphoproliferative malignancies.


A chronic antigenic stimulation is believed to sustain the leukemogenic development of chronic lymphocytic leukemia (CLL) and most of lymphoproliferative malignancies developed from mature B cells. Reproducing a proliferative stimulation ex vivo is critical to decipher the mechanisms of leukemogenesis in these malignancies. However, functional studies of CLL cells remains limited since current ex vivo B cell receptor (BCR) stimulation protocols are not sufficient to induce the proliferation of these cells, pointing out the need of mandatory BCR co-factors in this process. Here, we investigated benefits of several BCR co-stimulatory molecules (IL-2, IL-4, IL-15, IL-21 and CD40 ligand) in multiple culture conditions. Our results demonstrated that BCR engagement (anti-IgM ligation) concomitant to CD40 ligand, IL-4 and IL-21 stimulation allowed CLL cells proliferation ex vivo.
In addition, we established a proliferative advantage for ZAP70 positive CLL cells, associated to an increased phosphorylation of ZAP70/SYK and STAT6. Moreover, the use of a tri-dimensional matrix of methylcellulose and the addition of TLR9 agonists further increased this proliferative response. This ex vivo model of BCR stimulation with T-derived cytokines is a relevant and efficient model for functional studies of CLL as well as lymphoproliferative malignancies.
Like in most mature lymphoproliferative malignancies, an antigenic stimulation is believed to drive the leukemogenic process in chronic lymphocytic leukemia (CLL) [1][2][3] . A restricted use of IGHV genes and the existence of stereotypic B cell receptor (BCR) on CLL cells [4][5][6] provides evidence in favor of antigenic stimulation where different microbial antigens, as well as auto-antigens, have been suspected as actors of this chronic stimulation 7 . In addition, a chronic BCR self-activation has been shown in subtypes of CLL cells 8 . Moreover, several signaling aberrations have been described downstream of the BCR, notably in aggressive CLL with unmutated IGHV (UM-CLL), in which the expression of ZAP70 reinforces BCR responsiveness [9][10][11][12] . BCR activation, which is essential for the physiological development of lymphocytes 13 would also be indispensable for the survival and proliferation of CLL cells in vivo 2 . Accordingly, withdrawal of this stimulation is believed to be responsible for the rapid spontaneous apoptosis of CLL cells ex vivo 14 . The cellular consequences of this BCR activation has been extensively studied and ScIentIfIc REpoRTS | (2019) 9:701 | DOI: 10.1038/s41598-018-36853-8 we previously described the specific transcriptional 15,16 and proteomic programs 17 which are induced in aggressive CLL cells following BCR ligation. Nevertheless, a sustained soluble stimulation of the BCR induces apoptosis in CLL cells 9,15,[18][19][20][21] and BCR-associated factors are mandatory in inducing CLL cells proliferation. Several factors, are known for their role in CLL cells survival or proliferation, among which IL-2, IL-4, IL-10, IL-15, IL-21 and CD40L are prominent [22][23][24][25] , but an exhaustive evaluation of their role as BCR-cofactors for CLL cells proliferation is still lacking. Difficulties to achieve robust CLL cell proliferation ex vivo led to the use of stromal cells 26,27 , activated T cells 22,[28][29][30][31] or fibroblast (eventually CD40L transfected) 21,22,30,[32][33][34] as feeder cells. However, feeder cells' interactions 35 and secretion of IL-6, IL-10 or TGF-β can also participate in CLL cells survival and proliferation 26 , which makes the identification of essential leukemogenic factors difficult and prevents the specific evaluation of BCR ligation in the proliferative response in these models.
In this study, we aim to set-up culture conditions, primarily based on BCR ligation for patho-physiological relevance, inducing CLL cells proliferation. This study was conducted in two steps. We first aimed at establishing the optimal ex vivo model for CLL cells proliferation measured by carboxyfluorescein succinimidyl ester (CFSE) incorporation. For this, a selection of healthy and primary CLL cells were stimulated by anti-IgM ligation with or without co-stimulatory molecules (IL-2, IL-4, IL-10, IL-21, IL-15, sCD40L), at various concentration in different culture conditions. Next, using the optimized culture conditions, we analyzed the proliferative response of fresh negatively selected B cells isolated from a cohort of well characterized CLL patients, under informed consent, including clinical data, cell morphology, flow cytometry -including ZAP70 expression status-, FISH and IGHV mutational status, as these factors may impact the cell response to stimulation 22,28,30,31 . These culture conditions induced a proliferative response of a fraction of CLL cells, essentially ZAP70+, in soluble medium and a proliferation of nearly all CLL cells in 3D semi-solid medium, representing a valuable system for CLL functional studies.

Establishing culture conditions for CLL cells proliferation ex vivo.
To establish culture conditions for CLL cells proliferation after ex vivo activation, we first evaluated CFSE labeling in a small series of patient samples (n = 8). This approach allows calculating the percentage of dividing cells and the number of cell generations (Fig. S1). We first confirmed data from previous studies showing that ex vivo BCR activation by means of anti-IgM ligation does not induce CLL cells proliferation when these cells are cultured in soluble medium (Figs 1A and  S2A). Similarly, stimulation with IL-4, IL-21 or CD40L, used separately, in soluble medium, did not induce CLL cells proliferation either (Fig. 1A). We also confirmed that different combinations of cytokines, [CD40L + IL-4], [CD40L + IL-21] and [CD40L + IL-4 + IL-21] induced a weak (less than 40%) proliferation of CLL cells (Fig. 1A). Of note, IL-21, which has a pro-apoptotic effects on CLL cells 34 potentiates the proliferating effect of IL-4 when sequentially added after IL-4 23 and therefore IL-21 was added 24 h after all initial IL-4 stimulation. However, when we analyzed the proliferative effect of a combination of cytokines added after initial BCR stimulation (IgM ligation), we established that, even if BCR activation associated to [CD40L + IL-4] or [CD40L + IL-21] allowed a weak proliferation, the combination of anti-IgM with [CD40L + IL-4 + IL-21] induces a higher proliferation rate of CLL cells in soluble medium (Fig. 1A). Similar experiments confirmed the proliferative potential of these conditions on total B cells from healthy donors (Figs 1B and S2B). We analyzed the morphology of CLL cells submitted to these culture conditions. We observed the formation of clusters of proliferating cells in the culture medium ( Fig. S1) and cytological analysis of these cells after cytocentrifugation at day 6 revealed in all cases a monomorphic evolution consisting in large cells with a high amount of basophilic cytoplasm, prominent nucleoli and a fine chromatin that were distinct from those of control unstimulated cells (Fig. S3A). Immunophenotyic analysis of proliferating CLL cells at day 6 after ex vivo stimulation showed a lower expression of CD5, an upregulation of CD138 but not of CD38 and no IgG expression on cell surface, as compared to the expression at day 0 before stimulation ( Fig. S3B) which underlined the biological relevance of this model of ex vivo stimulation.
Response of a cohort of CLL cells submitted to the selected culture conditions. Because of the clinical and biological heterogeneity of CLL patients, we analyzed the impact of these selected culture conditions on the proliferative response of fresh CLL cells harvested from sixty-five untreated patients referred in Strasbourg Hospitals, essentially Binet stage A (58/65) ( Table 1). Among them, 25 harbored unmutated IGHV genes (UM-CLL) and expressed the ZAP70 protein (ZAP70+), 29 had mutated IGHV genes (M-CLL) and did not express the ZAP70 protein (ZAP70−), 10 were ZAP70 + M-CLL and one was ZAP70− UM-CLL. We also analyzed the proliferation of total B cells (20 healthy blood-donors) and naïve (CD19 + , CD27 − , IgM + ) B cells (16 healthy blood-donors). All of the total B-cells (20/20) and 12 out of 16 naïve B cells exhibited more than 25% of dividing cells at day 4 ( Fig. 2A)  , showing up to four generations of proliferating cells at day 6 ( Fig. 2A,B).

3D semi-solid matrix increases CLL cells proliferation ex vivo.
Then we analyzed the impact of a tri-dimensional environment using a semi-solid medium (methylcellulose) where the selected culture conditions also showed their proliferative action (Fig. S4). This culture condition enhanced the proliferation rates and number of cell generations of nearly all CLL cells (89%; 34/38 samples) (Fig. 3A), with up to six generations observed at day 6 ( Fig. 3B,C). Of importance, CLL cells stimulation on feeding cells (CD40L-transfected 3T6 cells) neither drove a higher proliferation rate (compared to the soluble stimulation alone) nor increased the number of generations of proliferating cells (Fig. S5A,B).

TLR9 agonists further increases CLL cells proliferation. Next, we tested the effects of CpG-ODN2006
(known to affect B cells proliferation by TLR9 activation) 31 in our culture conditions. While CpG-ODN2006 alone did not induce CLL cells proliferation (for 35 out of 39 CLL cells tested), we observed increased responses (p < 0.0001) when cells were stimulated with CpG-ODN2006 combined to BCR and cytokines in soluble and 3D semi-solid medium ( Fig. 4A-F). In addition, CpG/DSP30 and IL-2 cocktails being routinely used as metaphases inductors for cytogenetic diagnosis, we then compared our stimulation conditions with CpG-ODN2006 + IL-2 or commercial premix DSP30/IL-2. The combination of CpG/DSP30 and IL-2 did not increase the proliferation rate, compared to BCR with our selected cytokine stimulation conditions, in soluble or semi-solid medium ( Fig. S6A,B). Our work also confirmed previous reports 36 showing that the combination of CpG-ODN2006 and IL-15 is a modest inducer of CLL cells proliferation. Nevertheless, IL-15 addition did not increase the proliferation rate, compared to our soluble BCR stimulation (Fig. S7A,B).

Proliferative advantage of ZAP70+ CLL cells in soluble medium.
Analyzing the rate of CLL cells proliferation after BCR and cytokine stimulation according to their biological characteristics, we observed that proliferating cells, in soluble medium, exhibit the highest ZAP70 expression levels, compared to non-proliferating cells (p = 0.0043) (Fig. 5). Accordingly, 15/30 (50%) ZAP70+ CLL cells responded to the stimulation with up to four generations (median: 2 generations, with a significant (p = 0.0198) Pearson's correlation coefficient between ZAP70 expression level and the number of cell generations), whereas only 8/28 ZAP70− CLL cells proliferate with a maximum of two generations. Of note, the percentage of IGHV gene identity and CD38 expression did not associate with the cell proliferation in this soluble model (not shown). In semi-solid medium, ZAP70+ CLL cells with mutated IGHV exhibit a proliferative advantage (Fig. 5). When BCR ligation and cytokines are associated with CpG-ODN2006 stimulation, all CLL-cells respond equally in soluble or semi-solid medium, irrespective of the level of ZAP70 and the IGHV status (Fig. 5).
Increased ZAP70/SYK and STAT6 phosphorylation in proliferating ZAP70+ CLL cells. Given the heterogeneity of the proliferative response of ZAP70+ CLL-cells, we searched for signaling differences between responders (proliferating) and non-responders ZAP70+ cells. We performed western blots to analyze the main signaling pathways activated downstream of the BCR (e.g. ZAP70, pZAP/pSYK, pERK, pIkB) and the JAK/STAT pathway (e.g. pAKT and pSTAT6) in selected responders and non-responders amongst the ZAP70+ UM-CLL cells. Our results (Figs 6A and S8A,B) confirmed ZAP70 expression in all these CLL cells. They also suggested increased ZAP70 Tyr319 /SYK Tyr352 phosphorylation before stimulation (at the steady state) and a further increase upon stimulation in responding ZAP70+ CLL cells, compared to non-responding cells (p = 0.03). IkB phosphorylation was evidenced upon stimulation in both non-responding (p = 0.03) and responding (p = 0.05) CLL cells. The main signaling pathway downstream IL-21R (pSTAT3), revealed no difference between responders (proliferating) and non-responders ZAP70+ cells (Fig. S9A-C). However, we observed an increased STAT6 phosphorylation in responding CLL cells, compared to non-responders (p = 0.03). STAT6 being a major component of the IL-4 receptor signaling pathway, this result was corroborated by the significantly reduced CLL cells proliferation found in the absence of IL-4 in the stimulatory cocktails, or when a selective JAK3 inhibitor (PF-956980) was used (Fig. 6B).

Discussion
Engagement of the BCR is a crucial event in CLL leukemogenesis, but is not sufficient to induce cell proliferation ex vivo, and is even known to promote B cells apoptosis 15,18 . This caveat can be prevented with coated anti-IgM stimulation, which promotes CLL cells survival 9 but does not induce cell proliferation either. Therefore, BCR-induced CLL-cell proliferation in vivo likely requires additional co-stimulatory signals and proliferative properties of several soluble factors have been described in the literature. Here we performed an exhaustive study of the role of several co-stimulating factors, used solely or in combination, on top of anti-IgM stimulation, to identify mandatory factors sustaining CLL cells proliferation ex vivo. We evaluated CD40L (CD154), a TNF family member expressed on activated T-cells that activates the TNFRSF5 receptor on B cells and triggers several signaling pathways, including NF-kB and ERK, and participates in the survival, proliferation and differentiation of B cells 37 . IL-4, which is mainly secreted by T Follicular helper cell (TFH), activates STAT6 and participates in the activation and survival of B cells 38,39 , was also tested. Finally, we considered IL-21, which is produced by TFH, NKT and TH17 cells, for its role in the induction of JAK/STAT signaling in B cells 40 . IL-21 is known to induce apoptosis 34 , but plays also a role in CLL cells proliferation after priming by IL-4 and CD40L 22,23 , IL-21 was added at Day 1 in the culture medium. We also tested, isolated or in combination, IL-10, IL-2, IL-15, which did not show gain in BCR-induced proliferation effect (not shown) and were not considered further. Several groups have already used soluble CD40L, Il-2, IL-4, IL-10, IL15 or IL-21, isolated or in combination, to stimulate CLL cells [22][23][24]30,34,39 . However, to our knowledge, this is the first study to evaluate their role as BCR co-stimulating factors in soluble conditions without the support of feeder cells. Indeed, most ex vivo models of CLL cells proliferation described in the literature are not consistently defined, as they used co-cultures of fibroblasts expressing CD40L, sometimes in the presence of IL-21 22 , to favor CLL cells survival 32,33 . Co-cultures involving autologous activated T-cells have also been reported 22,28 , including in association with a fibroblast layer 30 . The nature of the cytokines used suggests a T cell dependent activation, possibly delivered by follicular helper T-cells found in the proliferative centers described in secondary lymphoid organs of CLL patients [41][42][43] .
Our systematic analysis of BCR ligation associated to different co-stimulations enabled us to select the optimal condition combining BCR activation and co-stimulating factors [CD40L + IL-4 + IL-21] driving CLL cells proliferation. This cocktail was used to stimulate CLL cells harvested from a cohort of patients with different biological characteristics (IGHV UM/M, ZAP70+/− and CD38+/−) and control B cells from healthy donors. In these conditions, about 1/3 of CLL cells proliferate at day 6 after stimulation, among which ZAP70+ B cells appeared particularly responsive. The same stimulation, performed on a 3D semi-solid (methylcellulose) medium, induced the proliferation of nearly all (89%) CLL cells, with a high number of cell generation, representing an efficient model of CLL cells proliferation.
A T-dependent help for CLL cells proliferation has not yet been proven in vivo. In our model, T-cells-derived cytokines (CD40L, IL-4 and IL-21) enable BCR-activated CLL proliferation which reinforces this hypothesis. Furthermore, it was shown in the literature that the proliferation of CLL cells xenografted in NOD-SCID mice required concomitant TFH graft in vivo 44 , which also sustains this model. Of note, our results showed the importance of IL-4 and IL-21 co-stimulation (in addition to anti-IgM and CD40L) for CLL cells proliferation, whereas the sole association of anti-IgM and CD40L was sufficient to induce healthy B cells proliferation. Finally, BCR and cytokine stimulation on a tri-dimensional matrix of methylcellulose allowed the proliferation of most of the CLL cells tested, irrespectively of their biological characteristics. This result, which may reflect the natural history of leukemogenesis of CLL cells within secondary lymphoid organs may further enhance the pathophysiological relevance of our ex vivo model. Furthermore, the individualization of clusters of proliferating cells, distributed in this 3D matrix, could allow studying the heterogeneity of intraclonal responsiveness to various drugs in future studies.
High proliferation rate (up to 80% with 6-8 cell generations) was only observed in a subset of CLL cells. Such variability may reflect the clinical and biological heterogeneity of CLL patients, among whom those characterized by ZAP70+ CLL cells appeared particularly responsive, as also observed by others 32,45 . However, we noted that all ZAP70+ CLL cells do not equally respond to ex vivo stimulation, which prompted us to investigate in more details the molecular features of the responders and non-responders among this subpopulation. We first confirmed by Western blots the presence of ZAP70 protein in these cells. In CLL cells, ZAP70 activates and extends SYK Tyr352 phosphorylation, independently of ZAP70 phosphorylation 10,46 . This activation induces another SYK Tyr526 phosphorylation, leading to downstream BCR signaling. Our results, performed on a limited number of CLL cell samples (4R vs 3 NR) show that CLL cells that proliferate in soluble medium could achieved a higher initial ZAP70 Tyr319 /SYK Tyr352 phosphorylation, the mechanism of which remains to be investigated. We also show a concomitant increase in pSTAT6 in the proliferating ZAP70+ CLL cells, suggesting that signaling downstream of the IL-4R could participate in CLL proliferation, which reinforces the need to evaluate therapeutic agents inhibiting this pathway 39,47 . More generally, these results highlight the need to explore the functionality of multiple signaling pathways in relation to the heterogeneity of CLL patients. Indeed, as more kinases inhibitors are now available for therapeutic use, there appears to be a rational for further personalized use of these molecules.
In conclusion, this study demonstrates the relevance of the BCR activation, combined with a defined set of cytokines, to recapitulate CLL cells proliferation ex vivo. In addition, it highlights the potential roles of T cells in this process. The soluble and 3D culture models established here represent valuable systems for further studies aimed at characterizing the initial steps of malignant evolution of the CLL, with the ultimate goal to identify novel targets for therapeutic purpose.

Materials and Methods
Subjects and B cell isolation. Peripheral 48,49 . Cytogenetic abnormalities were identified by metaphase analysis and fluorescence in situ hybridization (FISH) using a panel of probes as previously reported 50   using CD19 + or CD19 + /CD27 − staining (Beckman Coulter, Villepinte, France). CLL B cell purity was assessed after CD19 + /CD5 + staining (Beckman Coulter, Villepinte, France) and ranged from 90% to 99% (median 97%). Cell differentiation was studied after anti-CD38 and anti-CD138 stainings (Beckman Coulter, Villepinte, France) at days 0 and 6.  Apoptosis assay. Cell apoptosis was evaluated using FITC annexin-V Apoptosis detection kit and propidium iodide (PI) (both from BD Pharmingen, BD Bioscience, San Jose, CA, USA). Cells (10 6 ) were washed in phosphate-buffered saline (PBS) and re-suspended in annexin buffer before the addition of FITC annexin-V and incubated for 20 min on ice in the dark. PI was then added for 5 min before flow cytometry analysis. DAPI (Sigma-Aldrich; Missouri, USA) was also used to analyze cell viability.
Statistical analyses. Statistical analyses were performed using R 3.2.4 (R Core Team, 2016, R Foundation for Statistical Computing, Vienna, Austria). Graphics were created using Graphpad Prism 7.0 (Graphpad Software; Inc, La Jolla, CA, USA). We used permutational ANOVA for repeated measurements (lmPerm R package, https://github.com/mtorchiano/lmPerm) to compare more than two groups. Posthoc tests were done using nonparametric multiple comparisons tests carried out with the nparcomp R package 51 . Differences between two groups were assessed using either nonparametric Wilcoxon matched-pairs signed rank test of nonparametric Mann-Whitney test (unpaired, when applicable). A p value of <0.05 was considered statistically significant. *p < 0.05; **p < 0.01; ***p < 0.001.