NFKBIE mutations are selected by the tumor microenvironment and contribute to immune escape in chronic lymphocytic leukemia

Abstract

Loss-of-function mutations in NFKBIE, which encodes for the NF-κB inhibitor IκBε, are frequent in chronic lymphocytic leukemia (CLL) and certain other B-cell malignancies and have been associated with accelerated disease progression and inferior responses to chemotherapy. Using in vitro and in vivo murine models and primary patient samples, we now show that NFKBIE-mutated CLL cells are selected by microenvironmental signals that activate the NF-κB pathway and induce alterations within the tumor microenvironment that can allow for immune escape, including expansion of CD8+ T-cells with an exhausted phenotype and increased PD-L1 expression on the malignant B-cells. Consistent with the latter observations, we find increased expression of exhaustion markers on T-cells from patients with NFKBIE-mutated CLL. In addition, we show that NFKBIE-mutated murine CLL cells display selective resistance to ibrutinib and report inferior outcomes to ibrutinib treatment in NFKBIE-mutated CLL patients. These findings suggest that NFKBIE mutations can contribute to CLL progression through multiple mechanisms, including a bidirectional crosstalk with the microenvironment and reduced sensitivity to BTK inhibitor treatment.

Fig. 1. Murine TCL1-355-TKO cells with disrupted NFKBIE gene are positively selected in vitro by microenvironmental signals that activate the NF-kB pathway.

A Generation of NFKBIE-ko TCL1-355-TKO cell lines. Left panel shows indel analysis by amplicon capillary electrophoresis of the three NFKBIE-ko TCL1-355-TKO cell lines that were generated in three independent experiments. Wild-type alleles are indicated by a red arrow; mutant alleles are indicated by a black arrow. Editing efficiency was determined by analyzing the NFKBIE mutant allele frequency (MAF). MAF was calculated by dividing the area of the mutated alleles with the total area of all amplified alleles (mutant + WT) detected in the amplicon capillary electrophoresis. Right panel shows immunoblotting analysis of IκBε protein expression in the three NFKBIE-ko TCL1-355-TKO lines. The same cells non-transfected or transfected with Cas9 without cr-RNA (mock transfected) were used as positive controls for IκBε protein expression. B TCL1-355-TKO NFKBIE-ko and NFKBIE-wt cells were mixed at a 1:2 or 1:4 ratio (n = 3 experiments per each condition) and cultured for 14 days at a cell density of 0.4–1.2 × 10⁶ cells/ml. Analysis of MAF was done at day 0 and day 14. Statistical analysis was done using t test. Error bars represent standard deviation. C Analysis of MAF at day 14 of mixed cultures of TCL1-355-TKO NFKBIE-ko and NFKBIE-wt cells stimulated with CpG-DNA (1 μM), PtC (0.5 mM), anti-IgM (20 μg/ml), TNF-α (20 μg/ml) or 3T3-msCD40L fibroblasts (1:20 ratio versus CLL cells). The ratio of NFKBIE-ko vs NFKBIE-wt cells was 1:4 in the CpG-stimulation experiment, 1:2 in the PtC, anti-IgM and TNF-α stimulation experiments and approximately 1:1 in the CD40L stimulation experiment, respectively. The cells were collected every 48 h, washed, resuspended in fresh medium and re-stimulated with the same concentration of the indicated stimuli (or the same cell ratio in the case of CD40L stimulation).

Fig. 2. Impact of IκBε loss on the proliferation, survival and NF-κB pathway activation of Eμ-TCL1 CLL cells following CpG stimulation.

A Analysis of cell proliferation by BrdU incorporation and cell viability by Annexin V/PI staining performed after 24 h culture with 1 μM CpG-ODN 1668. Cells were cultured for 6 h with BrdU prior to harvesting for flow cytometry analysis. Statistical analysis was done using One Way ANOVA with Tukey test for multiple comparisons. B, C Analysis of nuclear translocation of NF-κB transcription factors following CpG-DNA stimulation of NFKBIE-ko and NFKBIE-wt TCL1-355-TKO cells. Nuclear and cytoplasmic proteins were extracted before or after 1 h, 3 h and 5 h CpG-stimulation. One representative experiment is shown in B and a summary of 5 experiments for p65 and 6 for p52, c-Rel and p50 is shown in C. Statistical analysis was done with the paired t test.

Fig. 3. Human CLL cells with loss-of-function NFKBIE mutations are positively selected in vitro by microenvironmental signals that activate the NF-κB pathway.

A Analysis of effects of microenvironmental signals involved in NF-κB pathway activation on the growth of NFKBIE-mutated primary human CLL cells. NFKBIE mutations were introduced in primary leukemic cells from 11 CLL patients with wild type NFKBIE by CRISPR/Cas9 editing. The cells were cultured for 3 days with 3T3-CD40L fibroblasts (1:10 ratio) +IL-4/IL-21 before being split and cultured for additional 48 h alone or in the presence of 3T3-CD40L fibroblasts, CpG ODN2006 (1 μM), or anti-IgM (20 μg/ml). NFKBIE mutant allele frequency was determined at day 5 of culture. The upper left panel provides a schematic representation of the experimental approach. The graph in the upper right panel represents the summary of the results with the 11 different CLL samples. Statistical analysis was done using the Wilcoxon signed-rank test. The results of two representative samples (G398 and G392) are shown in the bottom panels. B Analysis of effects of CD40L-stimulation on the growth of CXCR4- or CD19-edited primary human CLL cells. The same experimental approach was used as in part A, except for the use of guide RNAs targeting human CXCR4 or CD19 instead of NFKBIE. Capillary plots of the targeted region of the CXCR4 and CD19 genes from one representative sample out of 3 analyzed are shown. Wild-type alleles are indicated by a red arrow; mutant alleles are indicated by a black arrow. C Flow cytometry analysis of one CLL sample transfected with GFP-labeled Cas9 protein. Analysis was done after 1 h and after 72 h from the transfection.

Fig. 4. Murine CLL cells with disrupted NFKBIE are positively selected by the tumor microenvironment in vivo.

A TCL1-355-TKO NFKBIE-ko and NFKBIE-wt cells were mixed at a 1:2 ratio and injected in 4 C57BL/6 mice (2 × 10⁷ cells/mouse). Cells were isolated from the PC, peripheral blood (PB) and spleen of the injected C57BL/6 mice 3 weeks after adoptive transfer and NFKBIE MAF was determined in purified leukemia cells by amplicon capillary electrophoresis of the targeted region. Statistical analysis was done using One-way Repeated Measures ANOVA with the Holm-Sidak test for multiple comparisons. B NFKBIE MAF in injected leukemia cells and cells isolated from the PC and spleen of NSG mice 3 weeks after transplantation. The TCL1-355-TKO NFKBIE-ko and NFKBIE-wt cells were mixed at a 1:2.5 ratio and injected in 5 NSG mice (2 × 10⁷ cells/mouse). Statistical analysis was done as above. C In vivo BrdU-incorporation analysis of TCL1-355 TKO NFKBIE-wt or NFKBIE-ko cells isolated from PC and spleen of NSG mice (n = 4/group). BrdU was injected intraperitoneally 21 days after adoptive transfer. Leukemic cells were collected after 12 h, purified by negative selection using a B cell isolation kit, and analyzed by flow cytometry. Flow charts show the analysis of one NFKBIE-wt and one NFKBIE-ko sample, summary of the data is shown in the graph. Statistical analysis was performed using the Mann–Whitney Rank Sum Test.

Fig. 5. Transcriptome analysis of NFKBIE-wild type and NFKBIE-mutated tumors.

A Principal Component Analysis of the transcript expression counts (log2 vst-method normalized counts) for the 22 samples showing the first versus the second PC. PC1 shows 47% variance and PC2 18%. Each dot represents a NFKBIE-wt sample, each triangle is a NFKBIE-ko sample. Right panel shows Venn diagram of DETs. Numbers in each circle represent the number of DETs between the different comparisons while the ones overlapping are for mutual DETs. Red numbers represent upregulated DETs while blue numbers represent downregulated DETs. B Top ten Gene Ontology terms based on the Gene Ratio (x-axis), which is the ratio between the transcripts of the set found to be enriched for that specific term and the total transcripts in the set. In each row, the dots represent the enriched biological terms. The size of each dot is proportional to the number of the transcripts enriched in the term, while the color of the dots is the p-adjusted value, spanning from red to blue. C Volcano plots of significantly upregulated transcripts (red) and significantly down-regulated transcripts (blue) based on p < 0.05 and fold change ≥ 1 or ≤−1, respectively; black dots represent non-differentially expressed transcripts. Transcripts with a P value equal to or lower than 10⁻¹⁰ were clustered together.

Fig. 6. Immune composition analysis of NFKBIE-wt and NFKBIE-ko tumors.

A Analysis of total number and percentage of different immune cells and expression of immune checkpoint molecules in spleens of mice injected with NFKBIE-wt or NFKBIE-ko leukemia cells (n = 5/group). The investigated populations included CLL cells (CD19⁺CD5⁺), T cells (CD3⁺CD4⁺ and CD3⁺CD8⁺), granulocytes (CD11b⁺F4/80⁻), and monocytes/macrophages (CD11b⁺F4/80⁺) gated on viable CD45⁺ cells. Spleen samples from the experiment in Fig. 5 were used for this analysis. B Analysis of immune cell number and composition and immune checkpoint expression in spleens from a separate experiment with mice that were injected with 3 × 10⁷ NFKBIE-wt or 1 × 10⁷ NFKBIE-ko leukemia cells. C Expression of PD-1 and TIGIT on CD4⁺ and CD8⁺ T cells from patients with NFKBIE-wt and NFKBIE-mutated CLL (MAF > 20%).

Fig. 7. Analysis of BCR inhibitor sensitivity of NFKBIE-wt and NFKBIE-ko CLL cells.

A Changes in NFKBIE MAF during ibrutinib treatment in vitro. NFKBIE-ko and NFKBIE-wt TCL1-355 TKO cells were mixed at a 1:3 ratio and cultured in the presence or absence of ibrutinib (0.2 or 1.0 µM). DMSO was present at a concentration of 0.5% in both the ibrutinib-treatment and the control condition. DNA was isolated at the indicated time points for MAF analysis. Summary of the data is shown in the left panel and analysis of one representative sample from each condition in the right panels. Statistical analysis was done using One Way ANOVA with Tukey test for multiple comparisons. B Growth curve analysis of NFKBIE-wt and NFKBIE-ko CLL cells in culture with the indicated inhibitors. Each data point represents an average of 3 independent experiments. Graph in right panel shows changes in NFKBIE MAF during R406 and idelalisib treatment in vitro. NFKBIE knockout and wild type TCL1 leukemia cells were mixed as in A and cultured in the presence or absence of 1.0 μM idelalisib or R406. DMSO was used at a concentration of 0.5% in all investigated conditions. Statistical analysis was done using One Way ANOVA with Holm-Sidak test for multiple comparisons. C Changes in NFKBIE MAF during ibrutinib treatment in vivo. NFKBIE-knockout and NFKBIE-wild type leukemia cells were mixed at a 1:3 ratio and injected in 16 wild-type mice. Treatment was started three days later (8 mice treated with ibrutinib and 8 with vehicle control) and lasted for 2 weeks. The MAF at the end of treatment for the control (Ctrl) or ibrutinib (Ibr)-treated group is shown in the graph. The PB and PC samples from 2 ibrutinib-treated mice were lost during processing and were not available for analysis. Statistical analysis was performed with the t test (PC and PB) and Mann-Whitney rank sum test (spleen).

Fig. 8. Survival and progression analyses from initiation of ibrutinib treatment in CLL patients segregated based on NFKBIE mutations.

A Overall survival calculated from initiation of ibrutinib therapy of CLL patients with wild type NFKBIE (green curve) vs mutated NFKBIE (red curve). B Cumulative risk of progression under ibrutinib treatment of CLL patients with wild type NFKBIE (green curve) vs high NFKBIE MAF (M > 10%, red curve) vs low NFKBIE MAF (M < 10%, blue curve).