Background
Diffuse large B cell lymphoma (DLBCL) and Burkitt lymphoma (BL) are aggressive lymphomas, which require intensive chemotherapy regimens. BL is characterised by a germinal centre B-cell phenotype and an isolated c-MYC-rearrangement placing the c-MYC gene into close proximity of one of the Ig enhancers (IgH, Igκ or Igλ) [
1]. BL depends on the activity of the transcription factor and proto-oncoprotein c-MYC for proliferation and survival [
2,
3] and can be successfully treated with high intensity chemotherapy. DLBCL is significantly more common than BL, accounting for about one third of the NHL cases and is a heterogeneous disease [
4]. DLBCL can be classified into two principle classes predictive of disease outcome: germinal-centre B cell-like (GCB), and activated B cell-like (ABC) [
5‐
7]. GCB-DLBCL cells derive from activated germinal centre B cells and about 85 % of the cases express BCL6 a key transcriptional regulator of the GC response [
8]. ABC-DLBCL cells are post-germinal centre B cells, with features indicative of arrest during plasmablastic differentiation, and frequently show evidence of NFκB pathway activation [
6,
7,
9]. Both ABC-DLBCL and GCB-DLBCL can be associated with c-MYC deregulation or evidence of c-MYC activation. C-MYC-translocations occurring in the presence of a second or third translocation usually affecting BCL2 or BCL6 identify poor risk DLBCL referred to as double or triple hit lymphoma. Deregulation of c-MYC is therefore a common feature amongst aggressive lymphoma either occurring as an isolated event or with additional rearrangements.
Amongst the genes distinguished by differential expression in DLBCL subsets are the PIM (proviral integration site for Moloney murine leukaemia virus) kinases. High
PIM1 and
PIM2 mRNA levels are characteristic of ABC-DLBCL compared to GCB-DLBCL cells [
7,
10]. Expression of PIM1 and/or PIM2 is predictive of disease-free, disease-specific and overall survival in non-GCB DLBCL [
11,
12] and predominant nuclear PIM1 staining is highly correlated with disease stage in this type [
11]. These kinases are of particular interest because of a potential role as co-regulators of c-MYC dependent oncogenesis, with c-Myc and Pim1 showing co-operation during lymphomagenesis in mouse models [
13,
14]. Furthermore c-MYC and PIM1 have also been shown to cooperate in prostate tumourigenesis, while the inhibition of PIM kinases in c-MYC-expressing cancers decreases proliferation, survival and tumourigenicity [
15,
16].
The related kinases PIM1, PIM2 and PIM3 form the PIM kinase family of constitutively active serine/threonine kinases [
17].
Pim1 and
Pim2 were initially identified as targets for the integration of Moloney murine leukaemia virus (MMLV) in murine T cell lymphoma, indicating that they function as oncogenes [
18,
19]. Early studies showed that
PIM1 is overexpressed in 30 % of human lymphoid and myeloid leukaemias, while
PIM2 is overexpressed in AML [
20]. Both
PIM1 and
PIM2 mRNAs are highly expressed in CLL, DLBCL and mantle cell lymphoma (MCL), whereas
PIM2 is also overexpressed in follicular lymphoma, MALT lymphoma, nodal marginal zone lymphoma and multiple myeloma [
12,
21]. No overexpression of
PIM3 is seen in NHL [
12]. Apart from haematopoietic malignancies,
PIM1 and/or
PIM2 are highly expressed in several solid tumours [
22‐
31].
Several mechanisms for the oncogenic potential of PIM kinases and for the cooperation between PIM kinases and c-MYC have been described. PIM1 has been shown to be recruited to the chromatin by binding to the MYC MBII (MYC box II) domain and to stimulate transcription elongation through phosphorylation of histone H3 on serine 10 (H3S10p) [
32,
33]. The presence of the H3S10p modification is proposed to promote recruitment of 14-3-3 proteins, which serve as adaptors for the acetyl transferase MOF. MOF acetylates H4K16, which is recognised by the bromodomain containing protein 4 (BRD4), an adapter for P-TEFb [
33]. PIM1 has been found to be required for the expression of at least 20 % of the c-MYC-induced genes in HEK293 cells [
32]. Further, it has been shown that PIM1 overexpression in prostate cancer cell lines enhances the expression of c-MYC target genes [
34]. These findings suggest that a central role for PIM kinases in c-MYC-dependent gene regulation may be generalizable to other cell systems. Indeed, PIM1 was described to be nuclear in BL, which would allow for interaction of PIM1 and c-MYC at the chromatin level [
35]. Cooperation has also been identified between PIM2 and c-MYC and requires the ability of PIM2 to stimulate the NFκB pathway via activation of the kinase COT. Blocking NFκB in c-MYC and PIM2 overexpressing cells induces apoptosis
in vitro and inhibits growth in a tumour graft model [
36]. Both PIM1 and PIM2 can also directly stabilise c-MYC by phosphorylating Serine 329 [
37].
In addition, PIM kinases have several other pro-survival effects. PIM1, PIM2 and PIM3 phosphorylate BAD at S112 and other sites, which leads to binding of 14-3-3 proteins and inhibits its interaction with anti-apoptotic BCL-X
L [
38,
39]. PIM1 has also been shown to phosphorylate and inhibit the apoptosis signalling kinase 1 (ASK1), which results in reduced JNK and p38 MAPK phosphorylation and protects cells from H
2O
2-induced apoptosis [
40]. PIM kinases share several substrates with AKT and PIM2 can compensate for mTORC1 inhibition during haematopoiesis and in AML [
41,
42]. Additionally, PIM kinases stimulate cell cycle progression by phosphorylating MARK3, CDC25A, CDC25C, p21
CIP1, p27
KIP1 and SKP2 [
43‐
49]. PIM1-mediated phosphorylation has also been found to promote the degradation of FOXO1a and FOXO3a, which inhibits FOXO-mediated activation of
CDKN1B transcription (encoding for p27
KIP1) [
46]. Thus several mechanisms have been identified for c-MYC-independent regulation of proliferation and survival by PIM kinases.
Taken together, these observations provide a rational for targeting PIM1 and PIM2 in c-MYC expressing aggressive lymphomas including ABC-DLBCL and BL, in which PIM kinase inhibition might reduce c-MYC-mediated cell proliferation. In this study, we have assessed the anti-proliferative potential of the pan-PIM kinase inhibitor AZD1208 and other PIM inhibitors in aggressive NHL-derived cell lines and compared it with PIM1 and/ or PIM2 knockdown. Interestingly, our experiments reveal a kinase-independent function of PIM2 in regulating cell cycle. Both AZD1208 treatment and PIM2 knockdown synergistically block cell proliferation, indicating that PIM kinase inhibitors are not sufficient to completely abolish PIM function in NHL.
Discussion
In this study, the potent and selective PIM kinase inhibitor AZD1208 did not significantly reduce proliferation of BL or ABC-DLBLC cell lines. AZD1208 has been extensively tested against a panel of 442 kinases and inhibited only 13 kinases other than PIM1/2/3 by 50 % or more, but was still at least 43-fold selective for PIM kinases [
51]. Furthermore, in the study by Keeton et al., AZD1208-sensitive cell lines showed increased apoptosis already at 1 μM AZD1208 [
51], a concentration at which none of the cell lines tested in this study displayed reduced viability. Nevertheless, AZD1208 efficiently inhibited PIM kinases, as evaluated by BAD-S112 phosphorylation, at a concentration of 1 μM in the two BL and the two ABC-DLBCL cell lines assessed in our study. In agreement with a previous study, in which pan-PIM kinase inhibition led to stabilisation of PIM3 [
52], efficient inhibition of PIM kinases also led to a stabilisation of PIM1 and PIM2. In contrast, SMI4a, inh VI and Quercetagetin had a significant impact on cell proliferation in the same cell lines. However, their anti-proliferative potential was observed at high concentration and was heterogeneous between the different cell lines suggesting an effect independent of PIM kinase inhibition. Consequently, these data argue for a limited role of PIM kinase activity in maintaining oncogenicity in NHL and are consistent with previous findings [
11]. In agreement with this conclusion, knockdown of PIM1 had no significant impact on Raji cell proliferation. In addition, while AZD1208 treatment correlated with a decrease in H3K9acS10p at the
GNL3 promoter, this histone posttranslational modification was not significantly reduced by c-MYC or PIM kinase knockdown, suggesting that another kinase may target H3S10 at the
GNL3 promoter, at least in absence of PIM1. Several kinases have been shown to target H3S10 [
53] and a certain level of redundancy between all these enzymes might exist, which would preserve a normal gene expression programme if the activity of one of them was altered.
In contrast, PIM2 knockdown significantly reduced Raji cell number, suggesting that PIM2 might have an important function in maintaining Burkitt lymphoma cell growth. This is consistent with PIM2 being more frequently overexpressed in different haematological malignancies than PIM1 and more significantly associated with the activation of oncogenic pathways [
12]. On the other hand, the knockdown of PIM2 was consistently more efficient than that of PIM1, and it cannot be excluded that PIM1 depletion in this study might not have been sufficient to alter the proliferative capacity of NHL cell lines. Because pan-PIM kinase inhibition did not significantly reduce cell proliferation, although it clearly abrogated PIM activity as assessed by BAD phosphorylation, it can be hypothesised that PIM2 might have kinase-independent functions in DLBCL and BL. A function of PIM, independent of its kinase activity, has already been described for PIM1. First, overexpression of kinase-dead PIM1 can mimic some functions of active PIM1 [
54,
55]. Furthermore, PIM1 is recurrently targeted by aberrant somatic hypermutation in DLBCL, but out of 5 mutant proteins analysed, only one showed increased kinase activity, while three mutants were significantly less active than the wildtype protein [
56]. Altogether, these previous observations and our current work suggest that PIM kinases have kinase-independent functions in lymphomagenesis.
In agreement with a role of PIM2 in cell proliferation independent of its kinase activity, a combination of AZD1208 treatment and PIM2 knockdown additively repressed proliferation of Raji cells for clones either sensitive or resistant to AZD1208 treatment only. In this context, AZD1208 was associated with a reduction in cell viability, whereas PIM2 knockdown altered cell cycle progression. Several studies have pointed to a function of c-MYC in DNA replication licensing, c-MYC being known to control DNA replication by direct interaction with the pre-replication complex [
57]. Because, none of the inhibitor or knockdown experiments conducted in this study had any impact on
GNL3 gene expression, the presence of c-MYC and PIM1 at the
GNL3 promoter might participate in c-MYC-associated DNA replication licensing. Interestingly, repression of S-phase entry after PIM2 knockdown is in agreement with a defect in DNA replication licensing and coincides with an increase in PIM1 recruitment to the
GNL3 gene. This enhanced recruitment might be a consequence of direct competition for protein-protein interaction with c-MYC, or might indicate other interactions between PIM1 and PIM2 important for PIM1 function. PIM2 could, for example, inhibit nuclear translocation of PIM1 or inhibit association of PIM1 with chromatin. Nevertheless, this inverted correlation between cell proliferation and PIM1 enrichment at the c-MYC-bound GNL3 promoter suggests a repressive role of the chromatin-associated PIM1 in cell cycle progression, independent of its kinase activity. For example, the presence of this kinase might prevent the recruitment of another complex at c-MYC-bound cis-regulatory elements. Further investigations, beyond the scope of this work will be necessary to clarify this observation.
Competing interests
The authors declared that they have no competing interests.
Authors’ contributions
SK and PL designed the experiments and interpreted the data. PL prepared the manuscript. SK and KH conducted the experiments, collected the data and helped prepare the manuscript. RT participated to the design of the experiments, the interpretation of the data and to the manuscript elaboration. All authors read and approved the final manuscript.