1. Introduction
Burkitt lymphoma (BL) is an aggressive B cell lymphoma [
1] which originates from the germinal center (GC) [
2], and is characterized by oncogenic translocations of the proto-oncogene
MYC [
3]. The GC consists of two main histological and functional compartments known as dark zone (DZ) and light zone (LZ). In the DZ, B cells undergo somatic hypermutation and actively proliferate and afterwards move to the LZ where they receive survival signals via the B cell receptor (BCR) and CD40 in case of successful recombination and expression of a high affinity antibody. The DZ gene expression program depends on expression of CCND3 and the transcription factors BCL6, FOXO1, and TCF3. In contrast, the DZ program is repressed by BCR and CD40 signaling [
4] in LZ B cells. At the same time, signaling from the BCR and CD40 [
4], which activate NF-κB, JAK-STAT, ERK, and PI3K-AKT pathways, is essential for survival and further differentiation of the LZ B cells [
5,
6,
7].
Although MYC translocation under the control of immunoglobulin loci is an essential oncogenic event, it is not sufficient for BL progression. The maintenance of the main components of the DZ program [
8] including physiologically high expression of FOXO1 [
9] and TCF3 [
1,
10] is essential for BL. Moreover, activating mutations of TCF3 and FOXO1, inactivating mutations of TCF3 antagonist ID3, and protein stabilizing mutations of CCND3 belong to the most frequent oncogenic events in BL [
10,
11,
12].
The FOXO family of transcription factors regulates multiple processes, including cell cycle progression, apoptosis, glucose metabolism, differentiation, protection from oxidative stress, and stem cell maintenance [
13,
14,
15]. In some B cell malignancies, FOXO1 acts as a tumor suppressor and its activation induces growth arrest and apoptosis [
13,
16,
17,
18]. Surprisingly, FOXO1 knockdown in the MYC-PI3K driven mouse model of BL resulted in cell death and growth arrest [
9]. Moreover,
FOXO1 gene editing results in time-dependent selection of in-frame edited clones [
9] and impedes proliferation of BL cell lines [
19], indicating a role of FOXO1 in BL lymphomagenesis.
Using gene expression profiling (GEP), we found that FOXO1 knockdown, inter alia, represses DZ signature genes and this was associated with the activation of LZ signaling such as activation of PI3K-AKT and IKK-NF-κB pathways. Moreover, we demonstrated the feasibility of pharmacological inhibition of FOXO1 to mimic genetic FOXO1 down-regulation in BL as a potential therapeutic strategy.
3. Discussion
In the present work, we show that acute genetic depletion of FOXO1 inhibits the proliferation of BL cell lines. In particular, we observed that repression of FOXO1 activated signaling pathways characteristic for the GC LZ program, like the PI3K-AKT and IKK-NF-κB pathways. This ultimately downregulated the GC DZ program and DZ markers like CXCR4. We identified down-regulation of the proto-oncogene MYB as an important factor contributing to the anti-proliferative effect of FOXO1 knockdown. Importantly, pharmacological repression of FOXO1 also induced cell cycle arrest and apoptosis in BL cell lines and partially reproduced the effects of the shRNA-mediated FOXO1 knockdown on gene transcription. Finally, we demonstrated that overactivation of the FOXO1 activity also induces cell death and growth arrest, indicating the importance of a tight regulation of FOXO1 activity for the survival of BL.
We have shown that knockdown of FOXO1 in BL cell lines either interferes with cell proliferation or the apoptotic program. Recently it has been shown that knockdown of FOXO1 in a MYC/PI3K hyperactivation-driven model of BL in mice induces apoptosis. In BL cell lines, genetic editing of the FOXO1 gene resulted in positive selection of in-frame edited clones, indicating a role of FOXO1 in BL [
9]. Therefore, our data support the negative effect of FOXO1 depletion on BL proliferation, but at the same time indicate differences between BL cell lines and the mouse model, especially at the molecular level.
Interestingly, gene expression profiling of BL had revealed the virtual absence of NF-κB activity [
38] and the characteristic expression of markers of the GC DZ program [
2,
37,
44]. We show up-regulation of PI3K-AKT and NF-κB activity in BL cell lines by FOXO1 depletion. This is in line with higher PI3K-AKT activity in LZ B cells, which also express lower FOXO1 levels than DZ B cells [
5,
6]. FOXO1 has a complex effect on PI3K-AKT activity. Even though PIK3CA, the catalytic subunit of PI3K, was identified as a transcriptional target of FOXOs [
45], FOXO1 suppresses PI3K activity in NSCLC cell lines, although the mechanism is unknown [
46].
It is conceivable that activation of IKK-NF-κB contributes to PI3K-AKT activation. The NF-κB signature is repressed in BL in comparison with ABC- and even with GC-DLBCL [
44]. Moreover, NF-κB activation represses MYC-driven lymphomagenesis and is toxic for BL cell lines [
40,
47]. Since FOXO3A was shown to inhibit NF-κB activation [
48] by direct binding to RELA [
49], it is conceivable that the structurally and functionally related FOXO1 protein acts in a similar way. Being activated, AKT might also contribute to NF-κB activation by mTORC1-dependent IKK phosphorylation [
50]. NF-κB, in turn, can potentiate PI3K-AKT activity, e.g., by suppression of PTEN [
51], creating a self-amplifying circuit. Of note, the NF-κB activating kinases IKKs can inactivate FOXO proteins [
52,
53], suggesting that the efficacy of the FOXO1 depletion might be further increased by IKKs. Thus, FOXO1 acts as a regulator of NF-κB and PI3K-AKT activity in BL and its inhibition results in auto-amplification of IKK-NF-κB and PI3K-AKT pathways leading to repression of the DZ program.
In accordance with up-regulation of IKK-NF-κB and PI3K-AKT activity we observed repression of the DZ and up-regulation of the LZ signatures in BL by FOXO1 knockdown. In fact, we saw similar negative effects of
FOXO1 knockdown in BL cell lines as previously described for DZ B lymphocytes [
6]. Although downregulation of the DZ program by
FOXO1 knockdown in a MYC-driven mouse B cell lymphoma model that harbored a constitutively active version of the PI3K catalytic subunit has been recently reported [
9], the repertoire of repressed genes was different from what we obtained in human cell lines. Among others,
MYB,
CCND3, and
CXCR4 were not downregulated in the mouse model [
9]. We consider repression of the DZ program as the main cause for the cell cycle arrest induced by
FOXO1 knockdown.
Due to the complexity of the DZ program, it is difficult to filter out a single factor responsible for the growth arrest. Nevertheless, we identified repression of MYB as an important anti-proliferative event. Of note, MYB is a core BL survival factor, which is highly expressed in centroblasts [
26,
27]. Importantly, although
MYB is repressed by
FOXO1 knockout in mouse GC B cells [
6], the binding of FOXO1 to the
MYB promoter was not detected by ChIP-sequencing neither in GC B cells [
5], nor in other tissues [
30] indicating involvement of other mechanisms (e.g., protein-protein interactions) [
54]. Indeed, when we analyzed the effect of pharmacological FOXO1 inhibition we found that induction of miR-150 plays a critical role in regulating MYB levels. In contrast, CXCR4 is a direct FOXO1 transcriptional target [
5], which is downregulated by FOXO1 depletion in normal GC B cells [
36,
37]. Remarkably, we also find a strong repression of CXCR4 by FOXO1 downregulation. Since CXCR4 is also a MYB target [
55,
56,
57], MYB downregulation might potentiate FOXO1-inactivation induced CXCR4 repression.
We found that the small molecular weight FOXO1 inhibitor AS1842856 is toxic for BL. Although the specificity of small molecular weight inhibitors is a matter of concern in general, the treatment with AS1842856 reproduced most effects of the genetic FOXO1 knockdown, including repression of DZ-specific genes and growth inhibition. Interestingly, most BL cell lines were as sensitive to the inhibitor as BCP-ALL cell lines [
21], indicating the existence of common oncogenic mechanisms acting in these B cell neoplasia.
The most obvious differences between shRNA-mediated and pharmacological FOXO1 inactivation were the inability of AS1842856 to repress MYB at mRNA level. Interestingly, repression of MYB protein expression was due to induction of
miR-150 transcription. These discrepancies might be explained all by differences in the mechanisms of action. RNA interference decreases the FOXO1 levels, whereas AS1842856 does not modulate FOXO1 expression or localization, instead, AS1842856 binds to the transactivation domain and thereby
bona fide interferes with the FOXO1 transactivation activity [
42]. It is conceivable that not all interactions of FOXO1 with other proteins might be blocked by AS1842856, moreover the binding specificity of the new structures of AS1842856 might be different. Importantly, AS1842856 binds, although to a lesser extent, to other members of the FOXO family, the effects of whose are only partially redundant [
42]. Inhibition of other FOXOs might explain the generally stronger effects of pharmacological in comparison to genetic FOXO1 inhibition in regard to gene expression and PI3K-AKT and NF-κB activation. Given that NF-κB directly activates
miR-150 transcription [
58], it is conceivable that higher NF-κB activity is responsible for induction of
miR-150 by AS1842856. Moreover, since MYB is a recognized NF-κB target [
59], strong NF-κB activation by AS1842856 may also block the observed negative effect of FOXO1 depletion on MYB transcription.
Importantly, up-regulation of the
miR-150 expression in hematological malignancies is considered a promising therapeutic approach [
60,
61], warranting further investigations of antitumor effects of FOXO1 inhibitors.
Although FOXO1 inhibitors did not reach clinical trials yet, numerous preclinical
in vitro and
in vivo studies demonstrated their potential efficacy and safety for the treatment of type 2 diabetes. AS1842856, which was discovered to repress FOXO1 dependent transcription of the gluconeogenic enzymes glucose-6 phosphatase and phosphoenolpyruvate carboxykinase, normalizes blood glucose levels in diabetic but not in healthy mice even at concentrations much higher than therapeutic ones [
42]. Similarly, AS1842856 potentiated the regeneration of pancreatic β-cells and restored insulin secretion in diabetic mice but did neither increase the number of β-cells nor the insulin concentration in normal mice [
62]. Importantly, new selective FOXO1 inhibitors for the treatment of type 2 diabetes have recently been developed by AstraZeneca [
63].
The participation of FOXOs in the regulation of different biochemical processes in virtually all systems and organs predetermines high repurposing potential of the FOXO1 inhibitors. With respect to cancer therapy, it has been shown that expression of FOXO transcription factors is essential for maintaining a differential blockade in 40% of acute myeloid leukemia (AML) cases [
64]. Later, FOXO1 was identified as a critical survival factor in AML1-ETO leukemia, which appeared to be highly sensitive to AS1842856. Importantly, CD34
+ hematopoietic stem and progenitor cells were more than 10-fold less sensitive to AS1842856 than AML pre-leukemic cells and AML cell lines [
65]. Finally, in our recent publication we have shown a high sensitivity of BCP-ALL to the genetic and pharmacological depletion of FOXO1. Most important, by using a
in vivo BCP-ALL patient-derived xenograft model we demonstrated that AS1842856 induces a significant decrease of tumor load in all critical organ compartments and increases the life span of the animals when administrated at anti-diabetic concentrations [
21]. The development of new potent FOXO1 inhibitors could help to increase the efficacy and decrease the toxicity of treatment of FOXO1-dependent tumors including BL.
Paradoxically, FOXO1 overexpression was also inappropriate for BL maintenance. The antitumor effect of FOXO1 activation was also shown in different B cell lymphomas [
13,
16,
17,
18], moreover, for some of them we have provided evidence for a Goldilocks-like (too little is bad as well as too much) behavior of FOXOs [
21,
23]. Of note, the Goldilocks effect of FOXO1 on cell survival is not restricted to the B cell lineage and has in the meantime been described in other tissues and tumors [
66,
67].
Overall, we have shown that tight regulation of FOXO1 is critical for BL. In particular, our study highlights a role of FOXO1 as an essential regulator of the DZ survival and proliferation program in BL.