GSK3 inhibitors enhance retinoic acid receptor activity and induce the differentiation of retinoic acid-sensitive myeloid leukemia cells

Glycogen synthase kinase3 (GSK3) is an ubiquitously expressed serine threonine kinase originally characterized (and named) as an enzyme, which phosphorylated and inactivated glycogen synthase, an important mediator of glycogen biosynthesis.1 However, GSK3 has subsequently been shown to phosphorylate a wide range of cellular proteins and to have a central role in multiple cellular functions governing cell metabolism, gene expression, cell division, apoptosis, stem cell maintenance and differentiation.2 GSK3 includes two closely related isoforms GSK3α and GSK3β, which exhibit 97% sequence identity within their catalytic domains, but which are encoded by two separate non-functionally equivalent genes. GSK3-mediated phosphorylation has been implicated in regulating the proteolytic degradation of potential oncogenic proteins including c-myc, c-jun, cyclin D1, cyclin E and Notch1(ref.3) suggesting that activated GSK3 acts as a tumor suppressor. In contrast, inhibiting GSK3 triggers the apoptosis of colorectal cancer cell lines,4 and pharmacological inhibitors of GSK3 inhibit the proliferation and/or trigger the differentiation of MLL-oncogene-associated biphenotypic leukemia.5 Thus GSK3 appears to act as a tumor suppressor in some cell types although enhancing anti-apoptotic and/or leukemogenic activity in others. Here we describe a novel role for GSK3β in regulating retinoic acid receptor (RAR) activity as well as the proliferation and differentiation of all-trans retinoic acid (ATRA)-sensitive myeloid leukemia cells. We initially assessed the relative activity of GSK3 during the ATRA-mediated granulocytic differentiation of the NB4 promyelocytic leukemia cell line. The enzymatic activity of GSK3β is markedly inhibited by phosphorylation of an N-terminal serine residue (Ser9) and this phosphorylation can be detected with a GSK3β (S9) phospho-specific antibody (Cell Signaling Technology, Danvers, MA, USA). We observe that ATRA induces this inhibitory phosphorylation of GSK3β in NB4 cells. This enhanced phosphorylation occurs rapidly (within 1 h) following ATRA exposure and progressively increases—peaking at 3–5 days following differentiation induction (Figure 1a). This ATRA-induced GSK3β N-terminal phosphorylation occurs in a dose-dependent manner (Figure 1b). In contrast the level of total GSK3β remains constant during this ATRA-induced leukemia cell differentiation (Figures 1a and b). A similar enhanced phosphorylation of GSK3β Ser9 occurs in the ATRA-responsive HL60 cell line (Figure 1c) but not in the ATRA-resistant K562 cell line (Figure 1d). The above observations indicate that the ATRA-induced granulocytic differentiation of myeloid leukemia cells is associated with inhibition of GSK3β activity. To determine whether this reduced GSK3β activity might be involved in regulating this differentiation, we incubated the ATRA-responsive NB4 cells with various small molecule inhibitors of GSK3β. Surprisingly we observed that virtually all of these GSK3 chemical inhibitors as single agents induced morphologic granulocytic differentiation of the NB4 cells to a degree comparable to that induced by ATRA (Figure 2a). As with ATRA-induced differentiation this morphologic differentiation induced by GSK3 inhibitors is accompanied by enhanced expression of the neutrophil surface antigen CD11b, a widely used marker of granulocytic differentiation (Figure 2b), as well as by growth arrest of the differentiating cells (Figure 2c). Furthermore, gene products that are known to be upregulated during ATRA-induced myeloid leukemia cell differentiation such as Id1, C/EBPε, Stat1, p21 and p27 are also upregulated by these GSK3 inhibitors (Figure 2d). Moreover, GSK3 inhibition also reduced CDK8 and c-Myc gene expression as well as the activated (phosphorylated) CaMKIIγ (Figure 2d), which acts as a negative regulator of myeloid leukemia differentiation.6 These same GSK3 chemical inhibitors also induce morphologic granulocytic differentiation of the ATRA-responsive HL60 and ML1 myeloid cell lines, and this is also accompanied by enhanced expression of Cd11b (data not shown). To determine whether there is a synergistic effect between ATRA and GSK3 inhibitors in myeloid leukemia cell differentiation, we treated NB4 cells using a combination of both drugs at concentrations that exhibit only a minimal effect on differentiation as single agents. We observed that the drug combination significantly induced morphologic granulocytic differentiation of NB4 cells with concurrent enhanced Cd11b expression (Figure 2e). Our observations that GSK3 pharmacological inhibitors trigger the differentiation of ATRA-sensitive myeloid leukemia cell lines and also synergistically enhance ATRA-induced differentiation suggest that GSK3 may be directly involved in RAR activity. Therefore we assessed the effect of GSK3β on RAR transcriptional activity. We observe in NB4 cells that the ATRA-induced activity of the ATRA-responsive luciferase reporter harboring a retinoic acid response element (RARE) is markedly diminished when co-transfected with a GSK3β expression vector (Figure 3a). This inhibition of RAR transcriptional activity is dependent upon the kinase activity of GSK3β as a ‘kinase-dead’ GSK3β construct harboring a point mutation within the ATP-binding site of GSK3β (K85R) did not exhibit any inhibition of RAR activity in the same assay (Figure 3a). We also observed that the activity of this same RARE luciferase reporter was enhanced by the GSK3 pharmacological inhibitor, BIO in both ATRA-untreated NB4 cells (Figure 3b, compare lanes 1 and 5) and in cells treated with various concentrations of ATRA (Figure 3b, compare lanes 2–4 with lanes 6–8). Thus the kinase active GSK3β downregulates RAR transcriptional activity and pharmacological inhibition of GSK3β enhances RAR activity. To determine whether RARα may be an enzymatic substrate of GSK3β, we performed in vitro kinase reactions and observed that GSK3β can directly phosphorylate RARα and that this phosphorylation was inhibited by pharmacological inhibitors (BIO and SB212763) of GSK3β (Figure 3c). This phosphorylation occurs at the N-terminus of RARα as GSKβ readily phosphorylates a Gst–RARα fusion protein harboring only aa 1-86 of RARα but does not phosphorylate a Gst–RARα fusion protein lacking this RARα N-terminus (Figure 3d, compare lanes 7 and 8). We observed two consensus GSK3β phosphorylation sites (Ser/Thr-X-X-X-Ser/Thr)2 within the N-terminus of RARα at aa51-55 and at aa63-67 (Figure 3e). Mutating these sites (S51A, T55A, T63A and S67A) markedly inhibited the in vitro GSK3β-mediated phosphorylation of RARα (Figure 3f). Thus RARα is directly phosphorylated by GSK3β at specific N-terminus sites. To determine whether this phosphorylation had any functional significance, we compared the transcriptional activity of a wild-type RARα with RARα constructs mutated at these GSK3β phosphorylation sites. RARα expression vectors harboring mutations at each of these four sites stimulated RARE-luciferase reporter activity to a significantly greater degree than the wild-type RARα expression vector (Figure 3g). This indicates that the GSK3β-mediated phosphorylation of RARα at specific N-terminus sites is associated with an inhibition of RARα transcriptional activity. At least 40 different substrates of GSK3 have been identified to date and our present observations indicate that the retinoic acid receptor (RARα) is an additional substrate of this enzyme. The observed N-terminal phosphorylation of RARα by GSK3 inhibits RARα activity, and this appears to be biologically significant as pharmacological inhibitors of GSK3 enhance RARα activity (Figure 3b) and induce the differentiation of ATRA-sensitive myeloid leukemia cells (Figure 2). We do not know precisely how this N-terminal phosphorylation of RARα by GSK3β results in this inhibition of RARα activity. The phosphorylation of certain substrates by GSK3 enhances their proteolytic degradation3 and it is possible that the GSK3-mediated phosphorylation of RARα similarly enhances its degradation. Consistent with this possibility we observe that the expression of RARα is enhanced by GSK3 pharmacological inhibitors in ATRA-sensitive leukemia cells (Figure 4a). Interestingly we have also observed that treatment of myeloid leukemia cells with GSK3 inhibitors triggers enhanced expression of a number of RAR transcriptional co-activators including p300, CBP and SRC-1 (Figure 4b). Indeed the GSK3/RARα interaction we have observed may merely represent the ‘tip of the iceberg’ and multiple other members of the RAR transcriptional complex may also be phosphorylated/regulated by GSK3. Thus the effect of GSK3 inhibitors in enhancing RAR transcriptional activity and triggering myeloid cell differentiation may involve not only their triggering enhanced RARα expression but also their inducing the enhanced expression/activity of associated RARα transcriptional co-activators. Irrespective of defining the exact mechanism of action of GSK3 in inhibiting RAR transcriptional activity, it is clear from our experimental observations that pharmacological inhibitors of GSK3 as single agents are potent inducers of the terminal differentiation of ATRA-sensitive myeloid leukemia cell lines. There has been considerable interest in the pharmaceutical industry to identify and characterize small molecule inhibitors of GSK3 given the potential role of such inhibitors in overcoming insulin resistance and treating diabetes mellitus.7 Moreover, GSK3 has been implicated in enhancing β-amyloid deposits in the brain and certain GSK3 pharmacological inhibitors have entered clinical phases of development to determine their safety and efficacy in the treatment of Alzheimer's disease.8 Our present experimental observations together with the previous observations in MLL-associated leukemias5 suggest that such specific GSK3 inhibitors may also prove beneficial in the treatment of certain human myeloid leukemias either by themselves or in combination with other therapeutic agents. The authors declare no conflict of interest. This study was supported by NIH RO1 CA118971 (to SJC) and the Samuel Waxman Cancer Research Foundation (SWCRF).