Concurrent PI3K and NF-κB activation drives B-cell lymphomagenesis.

Aberrant activation of the PI3K and NF-κB pathways occurs frequently in human B-cell lymphomas.[1, 2] Recent studies suggested reciprocal molecular interactions between these two pathways in lymphomagenesis. For example, PI3K inhibition suppresses NF-κB activity in human Burkitt's lymphoma and diffuse large B-cell lymphoma,[3, 4] while blockade of NF-κB causes suppression of PI3K activity in primary effusion lymphoma cell lines.[5] Despite frequent alterations and molecular interactions of these two pathways in human lymphomas, genetic activation of anyone of these two pathways was not sufficient to initiate lymphoma development in mice.[6, 7, 8]

We recently reported that mutant mice (termed miR-17~92 TG mice) with B-cell-specific transgenic expression of miR-17~92, a cluster of six microRNAs (miRNAs) that are frequently upregulated in human cancers,[9, 10, 11] spontaneously developed B-cell lymphomas with a high incidence.[12] Subsequent molecular analyses showed that transgenic miR-17~92 expression led to constitutive activation of the PI3K and canonical NF-κB pathways by suppressing the expression of multiple negative regulators of these pathways.[12] However, it remains unclear whether functional cooperation of these two pathways is sufficient to drive lymphoma development and, thereby, to mediate the lymphomagenic effect of miR-17~92 overexpression.

To directly test this, we generated B-cell-specific double transgenic mice that concurrently activate the PI3K and NF-κB pathways (CD19cre;p110*IKK2ca mice, termed bPK mice; Figure 1a). In these mice, the PI3K pathway is activated by a p110* transgene, which encodes a constitutively active form of P110α, the catalytic subunit of PI3K,[8] while the NF-κB pathway is activated by a IKK2ca transgene, which harbors two serine-to-glutamate substitution mutations in the activation loop of the kinase domain of IKK2 and constitutively activates the canonical NF-κB pathway.[6] Both p110* and IKK2ca were knocked in at the Rosa26 locus with a loxP-flanked Neo-STOP cassette, which contains the neomycin resistance gene (Neo) and a transcriptional termination signal (STOP), inserted between the Rosa26 promoter and the transgene.[6, 8] The CD19cre transgene drives B-cell-specific expression of the Cre recombinase, which deletes the Neo-STOP cassette and turns on the expression of these two transgenes and green fluorescent protein (GFP).

Figure 1

Characterization of transgenic mice with concurrent activation of the PI3K and NF-κB pathways in B cells. (a) Scheme of B-cell-specific Rosa26-StopFL-IKK2ca and Rosa26-StopFL-p110* double transgenic mice (CD19cre;p110*IKK2ca, termed bPK mice). The endogenous Rosa26 promoter drives the expression of transgenes upon CD19cre-mediated deletion of the Neo-Stop cassette. (b) GFP expression in splenic B cells of bPK mice. (c) Immunoblot analysis of steady-state levels of IκBα, phospho-AKT (S473) and phospho-S6 (S235/236) as readouts for NF-κB and PI3K pathway activities, respectively. 2.5mo, 2.5-month old; 5mo, 5-month old. Left, representative immunoblots; right, bar graphs summarizing quantification results. (d) Total cell number and B-cell number in the spleen of 8-week-old bPK mice. Centered values and error bars indicate mean and s.e.m., respectively. (e) Increased cell size of splenic B cells in bPK mice. (f) Expansion of B1 cells, marginal zone B-like cells, and λ+ B cells in bPK mice.

We first confirmed the expression of GFP and the activation of the PI3K and NF-κB pathways in B cells of young bPK mice. As shown in Figure 1b, the vast majority (~85%) of bPK B cells were GFP-positive. Consistently with previous reports,[6, 8] both the PI3K and NF-κB pathways were active in these cells as indicated by increased phospho-AKT (S473) and phospho-S6 (S235/236) levels and IκBα degradation, respectively (Figure 1c). Similar to miR-17~92 TG mice, young bPK mice showed splenomegaly, increased splenic B-cell number and size, expanded B1 cell population (CD19+B220intCD43+CD5+), and higher percentage of λ+ B cells (Figures 1d–f).[12] A majority of splenic B cells in young bPK mice were CD19+CD1d+CD21+, a phenotype similar to marginal zone B (MZB) cells in wild-type mice, recapitulating the characteristic feature of B cells in IKK2ca single transgenic mice (Figure 1f).[6]

We next monitored a large cohort of bPK and littermate control (p110*IKK2ca but Cre-negative) mice for lymphoma development. As shown in Figure 2a, most of the 28 bPK mice died within 1 year (average lifespan: 8 months), whereas only one of the 26 littermate control mice died in the same period. We were able to analyze 15 sick bPK mice before they succumbed to diseases (Supplementary Table 1). These mice exhibited severe splenomegaly and hepatic granulomas (Figure 2b). Southern blot analysis showed that in 77% of these mice (10 out of 13 mice) B cells underwent mono- or oligoclonal expansion, a hallmark of lymphoma (Figure 2c; Supplementary Table 1). Consistently, these lymphoma cells were highly proliferative and were much bigger than B cells in littermate control mice (Supplementary Figure 1). B cells in these sick mice exhibited a surface phenotype of B1, MZB or both (Figures 2d and e). We have previously shown that most of miR-17~92-driven B-cell lymphomas were able to establish secondary tumors in immunodeficient mice.[12] We transplanted primary B cells from seven sick bPK mice exhibiting clonal B-cell expansion into Rag1-/- mice. Among them, primary B cells from four sick bPK mice were able to establish secondary tumors in the spleen or lymph nodes of Rag1-/- mice (Supplementary Table 1). Taken together, bPK mice developed lymphomas with a high incidence and the cell surface phenotypes of B cells in sick bPK mice were similar to those of B-cell-specific miR-17~92 TG mice (Supplementary Table 1).[12]

Figure 2

bPK mice developed B-cell lymphomas. (a) Kaplan–Meier survival curves of 28 bPK and 26 littermate control mice. The P-value (P<0.001) was determined by Mantel–Cox log-rank test. (b) Splenomegaly and hepatic granulomas in aged bPK mice. Centered values and error bars indicate mean and s.e.m., respectively. (c) Southern blot analysis of B-cell clonality. Genomic DNA of splenic B cells was digested with EcoRI and hybridized with a probe corresponding to the JH4 region of the IgH locus. Red arrowheads indicate clonal bands corresponding to VDJ or DJ rearrangements. Each lane represents one mouse. (d) Representative fluorescence-activated cell sorting plots of splenic B cells showing surface phenotypes of B1-like cells (CD19+B220intCD43+CD5+) or marginal zone B-like cells (CD19+CD1d+CD21+) in different bPK mice. (e) The number of sick bPK mice analyzed exhibiting expansion of B1-like cells and marginal zone B (MZB)-like cells.

In summary, we demonstrated that concurrent activation of the PI3K and NF-κB pathways is sufficient to drive lymphoma development in mice. Our previous study has shown that transgenic miR-17~92 expression activates these two pathways in B cells and leads to a high incidence of lymphomas.[12] Therefore, these results suggest that the PI3K and NF-κB pathways are two major downstream pathways mediating the lymphomagenic effect of miR-17~92 overexpression. Future investigations are warranted to explore the possibility of concurrently targeting these two pathways for the treatment of lymphomas driven by elevated miR-17~92 expression.[13]