Pre-clinical blocking of PD-L1 molecule, which expression is down regulated by NF-κB, JAK1/JAK2 and BTK inhibitors, induces regression of activated B-cell lymphoma

Escape from immune control must be important in the natural course of B-cell lymphomas, especially for those with activation of NF-κB. The pre-clinical LMP1/CD40-expressing transgenic mouse model is characterized by B-cell specific CD40 signaling responsible for NF-κB continuous activation with a spleen monoclonal B-cell tumor after 1 year in 60% of cases. LMP1/CD40 tumors B-cells expressed high levels of PD-L1. This expression was dependent on activation of either NF-κB, JAK1/JAK2 or BTK pathways since these pathways were activated in tumor B-cells and ex vivo treatment with the inhibitory molecules PHA-408, ruxolitinib and ibrutinib led to decrease of its expression. Treatment of LMP1/CD40-expressing lymphomatous mice with an anti-PD-L1 monoclonal antibody induced tumor regression with decreased spleen content, activation and proliferation rate of B-cells as well as a marked increase in T-cell activation, as assessed by CD62L and CD44 expression. These results highlight the interest of therapies targeting the PD-1/PD-L1 axis in activated lymphomas with PD-L1 expression, with possible synergies with tyrosine kinase inhibitors. Electronic supplementary material The online version of this article (10.1186/s12964-019-0391-x) contains supplementary material, which is available to authorized users.


Background
Aberrant expression of the programmed death-ligand 1 (PD-L1, also known as B7-H1 or CD274) checkpoint molecule has been reported in many cancers such as breast, lung and colon tumors as well as during chronic viral infections like those with Epstein-Barr virus (EBV) for example [1,2]. Efficacy of immunotherapies against the PD-1/PD-L1 axis in lung tumors or melanomal demonstrated the importance of the immune checkpoints in the control of emergence and growth of tumors [2]. As reviewed recently, various publications have indicated that disruption of immune checkpoints is also a critical step in B-cell non-Hodgkin's Lymphomas (NHL) [3]. NF-κB, one of the most cited transcription factor in B-cell lymphomas, is able to increase tumor cell expression of PD-L1 either directly or indirectly [3]. NF-κB constitutive activation is found either in aggressive diffuse large B-cell lymphomas (DLBCL) with an activated phenotype (ABC-DLBCL), or in indolent B-cell lymphomas such as chronic lymphocytic leukemia, Waldenström Macroglobulinemia, marginal zone B-cell lymphomas (MZL) [4]. Here, we wanted to explore the putative interest of PD-L1 immune therapy against Bcell lymphoma with NF-κB activation. To experimentally address this question, we used a transgenic mouse model which specifically express in B-cells a chimeric protein composed of the transmembrane moiety of the Epstein-Barr Virus latent membrane protein 1 (LMP1) and the transduction tail of CD40 (LMP1/CD40 protein), that results in continuous activation of NF-κB, responsible for a spleen monoclonal B-cell tumor (LMP1/CD40 Bcell lymphoma) after 1 year in 60% of cases [5].

Mouse models and in vivo and ex vivo treatments
LMP1/CD40-expressing mice have been already described [5]. Animals were housed at 21-23°C with a 12-h light/ dark cycle. All procedures were conducted under an approved protocol according to European guidelines for animal experimentation (French national authorization number: 87-022 and French ethics committee registration number "CREEAL": 09-07-2012). For in vivo PD-L1 treatment, LMP1/CD40-expressing mice were injected intraperitoneally every 4 days for 3 weeks with 200 μg anti-PD-L1 antibody (clone 10F.9G2; Bio X cell; US). For ex vivo treatments, splenocytes were cultured for 48 h in complete RPMI medium (Eurobio) supplemented with 10% of FBS, 2 mM of L-Glutamine, 1% of Na pyruvate, 100 U/ml of penicillin and 100 μg/ml of streptomycin (ThermoFisher Scientific) and with the following treatments: either 10 μM of PHA-408 or 1.5 μM of ruxolitinib or 1 μM of ibrutinib.

Proliferation
For in vivo proliferation, mice were injected intraperitoneally with 2 mg BrdU (Sigma-Aldrich, US), 18 h before isolating cells. Splenocytes were stained for B220 and phases of cell cycle were analyzed by measuring BrdU and Propidium Iodide (PI)-incorporation, using the FITC-BrdU Flow Kit (BD Pharmingen; US).

Real time quantitative reverse-transcription PCR
Total RNA was extracted from CD19_Cre or LMP1/ CD40 splenocytes and complementary DNAs were reverse transcribed from 1 μg of total RNA samples using the High Capacity cDNA Archive Kit (Life Technologies, Carlsbad, USA). PCR products were amplified from each cDNA using the TaqMan® Universal PCR Master Mix and TaqMan® Gene Expression Assays: IL-10/ Mm01288386_m1, PD-L1/Mm03048248_m1, GAPDH/ Mm99999915_g1. The GAPDH gene was used as a reference gene for the control of amplification. The calculated relative gene expression level was equal to 2 − DDCT, where DDCT is the delta delta cycle threshold. Gene expression fold changes were calculated as the ratio of the test condition to its control.
Immunoblot 10 × 10 6 cells were lysed and sonicated in Laemmli sample buffer. Lysates were separated on 12% SDS-PAGE gel and transferred to PVDF membrane. Revelation were done using anti-TRAF1 antibody (N-19) purchased from Santa Cruz Biotechnology; anti-STAT3 antibody (124H6) and anti-phosphoSTAT3 antibody (Tyr 705) (D3A7) from Cell Signaling Technology; and anti-ERK (p-44/42) antibody and anti-phosphoERK antibody (Thr 202/Tyr 204) from Cell Signaling Technology. Signals were visualized with Immobilon™ Western Chemiluminescent HRP Substrate (Millipore). The amount of loaded protein was standardized against GAPDH using an anti-GAPDH antibody from R&D Systems. Membranes were numerized with ChemiDoc™ Touch Imaging System (Bio-Rad) and images were analysed with the Image Lab Software (Bio-Rad).

Results
Our previous transcriptome studies from LMP1/CD40expressing mice suggested that those tumors might express high levels of CD274/PD-L1 [6]. We thus analyzed the Immune Escape Gene Signature published by C Laurent et al (Additional file 1: Table S1) [7] from the Affymetrix transcriptome (HT MG-430 PM Array, Additional file 1: Table S2 and Additional file 2: Table S3) of a series of six LMP1/CD40 B-cell lymphomas that were compared to their CD19_Cre littermate. As shown in Fig. 1a, both PD-L1 (red arrow) and PD-L2 (orange arrow) were co-clusterized with the immunosuppressive interleukin 10 (IL-10, green arrow), CD80 and MCL1, being over expressed. We confirmed overexpression of IL-10 and PD-L1 by RT-QPCR in 12-month-old LMP1/ CD40-expressing mice when compared to CD19_Cre mice (Fig. 1b). We also showed higher levels of total PD-L1 expression by immunoblotting and of PD-L1 exported to the membrane by flow cytometry on LMP1/ CD40 lymphoma cells from those mice than on B-cells from age related CD19_Cre mice ( Fig. 1c and d).
To address in vivo the role of PD-L1 in these B-cell lymphoma, LMP1/CD40-expressing mice were treated with an antibody blocking PD-1/PD-L1 signaling for 3 weeks according to a methodology already described [8,9]. We observed a reduction of spleen size and absolute number of splenocytes in anti-PD-L1 treated LMP1/ CD40-expressing mice (Fig. 2a), due to decreased B-cell numbers (Fig. 2b) without significantly affecting total numbers of T-cells and granulocytes (Additional file 3: Figure S1). With PD-L1 blockade, we noticed a decrease in numbers of activated B-cells in LMP1/CD40-expressing mice (Fig. 2c). This was associated with a reduction in the in vivo proliferation rate of spleen B-cells, as assessed by the decrease of in vivo BrdU incorporation over 18 h (Fig. 2d). Morphologically, spleen lymphocytes from anti-PD-L1 treated LMP1/CD40-expressing mice were smaller with a more condensed chromatin (Fig.  2e). We then studied the impact of PD-1/PD-L1 blockade in T-cell compartment. Expression of T-cell activation markers CD62L and CD44 was increased on both CD4 and CD8 T-cells (Fig. 3a). In parallel, an increase in PD-1 expression was observed on the surface of CD4 Tcells. This increase in PD-1 expression was more heterogeneous on CD8 T-cells (Fig. 3b).
In addition to be able to activate the classical pathway, CD40 is a strong inducer of the alternative NF-κΒ activating pathway as already shown in LMP/CD40 mice [5]. CD40 constitutive activation of B-cells increases IL-10 expression [10]. IL-10 receptor signaling is mediated by the JAK1 and Tyk2 tyrosine kinase that leads to activation of STAT3 transcription factor, STAT3 being a major inducer of PD-L1 gene expression [11]. CD40 stimulation can also augment BCR-induced B cell responses by activation of Bruton's tyrosine kinase (BTK) either directly [12] or by interacting with CD19 [13]. The BCR is also able to induce the IL-10/STAT3 signaling with increased expression of PD-L1 in diffuse large B-cell lymphoma [11]. Comparing splenocytes of LMP1/ CD40 mice to those of CD19_Cre mice, allowed to check activation of NF-κB, as assessed by increased expression of TRAF1. Increased phosphorylation of STAT3 and ERK indicated that IL-10 and BCR-related signaling pathways were also constitutively activated (Fig. 4a). LMP1/CD40 lymphoma B-cells were ex vivo treated with the PHA-408 molecule, an inhibitor of IKK2/NF-κB activation, the JAK1/JAK2 tyrosine kinase inhibitor (TKI) ruxolitinib, and the BTK inhibitor ibrutinib. As expected, treatment with PHA-408 led to decreased expression of TRAF1 protein. This treatment also resulted in decreased phosphorylation of both STAT3 and ERK. Ruxolitinib treatment only decreased STAT3 activation. In presence of ibrutinib, TRAF1 expression and STAT3 phosphorylation were moderately decreased while ERK phosphorylation was completely abolished (Fig. 4b). As shown in Fig. 4c, PD-L1 mRNA expression was decreased in presence of the three inhibitors. This result was confirmed at the protein level. At the membrane level, PD-L1 expression was strongly reduced 48 h after treatment with either PHA-408 or Ruxolitinib and moderately with ibrutinib (Fig. 4d). These inhibitors also decreased IL-10 expression (Additional file 3: Figure S2). Of note, modulation of PD-L1 expression was no longer observed on wild type CD19_Cre splenocytes treated with inhibitors (Additional file 3: Figure S3). Furthermore, CD40 plus IL-4, BCR, and IL-10R stimulation of CD19_Cre splenocytes induced over-expression of PD-L1, that induction being repressed when cells were treated with these inhibitors (Additional file 3: Figure  S4).
Altogether, these results show that CD40, IL-10 and BCR signaling pathways have complex interplays. As summarized in Fig. 5, the CD40 signaling pathway would be directly responsible for NF-κB activation and would increase the BCR signaling. Both BCR and CD40 signaling would lead to IL-10 secretion that secondary activate the JAK/STAT pathway in an autocrine manner. With NF-κB, both ERK and STAT3 activation would up-regulate expression of PD-L1. Blocking either NF-κB, BCR or STAT3 activation would led to decreased PD-L1 expression. PD-L1 expression was thus under the control of CD40 activation, either directly through NF-κB activation or indirectly through IL-10 induction of BCR sensitization and BTK activation.

Discussion
Expression of PD-1 and PD-L1 in B-cell lymphomas and effect of immune therapies against the immune checkpoint axe has been recently reviewed [3]. Expression of PD-L1 by tumor cells and effect of anti-PD-1 immune therapy has been clearly demonstrated in Hodgkin's lymphoma, a tumor which is constantly associated with both NF-κB and STAT3 activation. In DLBCL, expression of PD-L1 is variable but participates to the gene immune escape signature of ABC-DLBCLs [3,7]. ABC-DLBCL are not only associated with NF-κB activation but may exhibit a chronic active BCR [14] and are sensitive to BTK inhibitors [15]. STAT3 activation mainly found in ABC-DLBCLs and is associated with poor survival [16].
Here, our results show that NF-κB activated B-cell lymphoma of the LMP1/CD40-expressing mouse model exhibited an immune escape gene signature involving expression of PD-L1 and PD-L2, which expression was co-clusterized with IL-10. Over-expression of PD-L1 not only involved NF-κB activation by CD40 but also BTK and JAK/STAT signaling, the latter probably being indirectly regulated via an autocrine loop with participation of IL-10 for example. Indeed, PD-L1 expression could be down-regulated after treated with NF-κB, JAK1/JAK2 and BTK inhibitors. Expression of PD-L1 was very likely to be associated to tumor immune escape as demonstrated for numerous solid cancers such as melanoma of lung cancers [2]. Indeed, in vivo blockade of PD-L1 was able to rapidly repress expansion of these B-cell lymphomas, with concomitant decrease in both B-cell proliferation and B-cell expression of activation markers as well as an increase in T-cell activation. Among these markers are expression of PD-1 on both CD4 and CD8 T-cells. PD-1 is transcriptionally induced in activated T cells [17]. PD-1 expression on activated T-cells allowed subsequent exhaustion and inhibition thank to PD-1/ PD-L1 interaction to return to unactivated state [18].
Our results indicate that such T-cell inhibition no longer occurred when PD-L1 was blocked.

Conclusion
Our results indicate that therapies against the PD-L1/ PD-1 axis may work in lymphomas as long as the tumor cells express PD-L1. Our results also suggest that combination of immune therapy targeting the PD-1/PD-L1 axis and TKI specific for the JAK/STAT or the BCR/ BTK pathway could be of interest. Even if these chemical inhibitors such as ruxolitinib or ibrutinib could have their own inhibitory effect on actors of the anti-tumor response such as cytotoxic T-cells or macrophages, our results also raise the question of the effect of these molecules on the local reactivation of the immune system.