Gö6976 is a potent inhibitor of the JAK 2 and FLT3 tyrosine kinases with significant activity in primary acute myeloid leukaemia cells

Authors


Asim Khwaja, Department of Haematology, Royal Free and University College London Medical School, University College London, 98 Chenies Mews, London, WC1E 6HX, UK. E-mail: a.khwaja@ucl.ac.uk

Summary

Aberrant activation of Janus kinase/signal transducers and activators of transcription (JAK/STAT) signalling is implicated in a number of haematological malignancies and effective JAK inhibitors may be therapeutically useful. We found that Gö6976, an indolocarbazole inhibitor of the calcium-dependent isozymes of protein kinase C (PKC), inhibited interleukin 3/granulocyte-macrophage colony-stimulating factor-induced signalling, proliferation and survival whereas Gö6983, a broad spectrum PKC inhibitor, had no such effects. Gö6976 was found to be a direct and potent inhibitor of JAK2 in vitro. Gö6976 also inhibited signalling, survival and proliferation in cells expressing the leukaemia-associated TEL–JAK2 fusion protein and the myeloproliferative disorder (MPD)-associated JAK2 V617F mutant. In primary acute myeloid leukaemia (AML) cells, incubation with Gö6976 reduced constitutive STAT activity in all cases studied. In addition, Akt and mitogen-activated protein kinase phosphorylation were reduced in 4/5 FLT3-internal tandem duplication (ITD) positive AML cases and 7/13 FLT3-wild-type (WT) cases. Expression of FLT3-WT, ITD and D835Y in 32D cells showed that Gö6976 is also a potent inhibitor of WT and mutant FLT3. In AML cells, Gö6976 reduced the survival to 55 ± 5% of control in FLT3-ITD cases and to 69 ± 5% in FLT3-WT samples. These data may help identify clinically useful compounds based on the structure of Gö6976, which can be employed for the treatment of MPDs as well as AML.

The cytokine receptor superfamily, which includes receptors for erythropoietin (EPO), thrombopoietin (TPO) , granulocyte colony-stimulating factor, interferons and interleukins, is characterised by the lack of a catalytic domain in the cytoplasmic portion of the receptor and, as such, lacks intrinsic tyrosine kinase activity. These receptors utilise Janus kinases (JAKs), cytosolic tyrosine kinases, to couple ligand binding with downstream tyrosine phosphorylation. When a ligand binds to a receptor, the receptor dimerises and brings two JAK molecules into close proximity. The JAKs cross/autophosphorylate and become activated; once activated, they phosphorylate the intracellular domains of the receptor and create docking sites for downstream signalling proteins via their Src homology 2 (SH2) domains (Ihle, 1995). Signal transducers and activators of transcription (STATs) are latent transcription factors residing in the cytoplasm. STATs are phosphorylated by JAKs and homo- or heterodimerise by binding of a phosphorylated tyrosine activation motif on one STAT subunit to the SH2 domain on the other subunit. These dimers then translocate to the nucleus where they bind specific DNA binding motifs to bring about alterations in gene transcription governing cell proliferation and survival. In addition to STATs, other SH2 containing signalling proteins are recruited to the phosphorylated receptor including Src homologous and collagen-like protein, the p85 subunit of PI3-Kinase and other adapter molecules. Once recruited to the receptor, these proteins may become phosphorylated by JAKs, leading to activation of many downstream pathways including the phosphoinositide 3 kinase (PI3K) and Ras/mitogen-activated protein kinase (MAPK) pathways (Ihle et al, 1997; O'Shea et al, 2002).

Until recently, the most direct evidence that implicated the dysregulation of the JAK/STAT pathway in haemopoietic malignancies was the identification of the TEL–JAK2 fusion. This fusion occurs as a result of t(9;12) or t(9;15;12) and has been reported to occur in myeloid and lymphoid leukaemias (Peeters et al, 1997; Reiter et al, 2005). JAK2 is also likely to be activated by fusion of JAK2 with PCM1, resulting from a novel t(8;9) which has been described in a variety of haemopoietic malignancies (Reiter et al, 2005) and JAK2 gene amplification has been shown to be a common finding in Hodgkin and primary mediastinal B-cell lymphomas (Joos et al, 2000; Guiter et al, 2004). Recently, a mutation at position 617 resulting in a change from valine to phenylalanine (JAK2 V617F), in the JAK homology 2 (JH2) domain of JAK2, has been reported to be present in myeloproliferative disorders (MPD), including the majority of cases of polycythaemia vera as well as in essential thrombocythaemia and idiopathic myelofibrosis. This mutation in the pseudokinase domain of JAK2 is thought to lead to loss of autoregulation and increased kinase activity (James et al, 2005; Steensma et al, 2005).

In addition to these well-characterised JAK2 abnormalities, constitutive STAT activation has been reported in many malignancies including multiple myeloma, acute myeloid leukaemia (AML), chronic myeloid leukaemia (CML), non-Hodgkin lymphoma and anaplastic cell lymphoma (Bowman et al, 2000; Sternberg & Gilliland, 2004). STAT activation can occur as a result of chromosomal translocation and formation of a novel fusion product, such as BCR-ABL, by acquired mutations such as internal tandem duplications in the Flt3 gene (FLT3 ITD) or secondary to autocrine/paracrine growth factor secretion. The frequency of constitutive STAT activation reported in AML varies between studies. Biethahn et al (1999) reported constitutive STAT 3 activation in all of 25 patients examined and constitutive STAT 5 activation in 21 out of 25 patients examined. This agrees with the findings of Hayakawa et al (1998) who found STAT 5 activation in 80% and STAT 3 activation in 74% of samples. Benekli et al (2002), however, only found constitutive STAT 3 phosphorylation in 28 out of 63 (44%) patients and Xia et al (1998) found STAT 3 and STAT 5 activation in 28% and 22% of patients, respectively. The reasons for the difference in reported frequencies is not clear, but may relate to different methods of detection i.e. tyrosine phosphorylation versus DNA binding. The observation that STATs are activated in most cases of AML suggests that selective pharmacological inhibition of activation of these proteins could be a valid therapeutic strategy in this disease.

Researchers have generated inhibitors in a variety of structural classes, including tyrphostins and benzylidenemalonitriles (Lawrence & Niu, 1998; Malaviya & Uckun, 1999; Wang et al, 1999), which inhibit JAK-driven biological events at micromolar concentrations. Thompson et al (2002) reported another compound, a pyridone containing tetracycle, which inhibited JAK family members at concentrations 100-fold lower than the tyrphostins.

AG490, a tyrphostin inhibitor of JAK2 and JAK3, is the most commonly utilised JAK2 inhibitor in in vitro studies. Meydan et al (1996) found that this compound selectively blocked the acute lymphoblastic leukaemia cell growth in vitro and in vivo by inducing apoptosis (Meydan et al, 1996). Spiekermann et al (2001) found that AG490 induced a time- and dose-dependent growth arrest without overt morphological signs of differentiation in AML cell lines. However, the effects of AG490 on cell proliferation are complicated by the JAK-independent effect of this compound on cdk2 activation and cell-cycle progression (Kleinberger-Doron et al, 1998; Savell et al, 2004). To our knowledge, the effect of JAK2 inhibitors on primary AML cells has not been investigated.

In preliminary experiments looking at the role of protein kinase C (PKC) in cytokine signalling, Gö6976, an indolocarbazole kinase inhibitor, had effects that could not be explained by PKC inhibition alone. Gö6976 is derived from staurosporine (ST) and was selectively active against the calcium-dependent isozymes (α and β1) in nanomolar concentrations but even micromolar concentrations were ineffective against the calcium-independent subtypes. Gö6976 inhibits PKC in a competitive manner with respect to ATP (Martiny-Baron et al, 1993; Gschwendt et al, 1996). Gö6976 appeared to abrogate the effects of interleukin 3 (IL3) and granulocyte-macrophage colony-stimulating factor (GMCSF) on the survival of several factor-dependent cell lines, whereas other PKC inhibitors with a broader spectrum of anti-PKC activity had no such effect. Gö6976 blocked the activation of STAT 5 and the PI3-kinase target Akt, in response to GMCSF but had no effect on activation of signalling in response to stem cell factor (SCF), which acts via a receptor with intrinsic tyrosine kinase activity and not through JAKs.

The present study established that Gö6976 is a direct inhibitor of JAK2 and investigated the scope of its activity on other JAK family members. Examination of the effect of Gö6976 on constitutive STAT, Akt and MAPK activation in primary AML samples led Gö6976 to be identified as an FLT3 inhibitor with potential activity in a wide spectrum of haematological malignancies.

Materials and methods

Cell line culture

The following cell lines were used: 32D and FDCP1, murine myeloid lines, (factor-dependent on IL3); TF-1, human erythroleukaemia line (factor-dependent GMCSF/erythropoietin); Mo7E, human megakaryoblastic leukaemia line (factor dependent GMCSF/TPO/SCF); HEL, human erythroleukaemia line. All cell lines were grown in RPMI/10% FCS (RPMI 1640 medium with l-glutamine (Invitrogen, Paisley, Scotland) and 10% fetal calf serum, FCS) at 37°C, 5% CO2. TF-1 and Mo7E cells were supplemented with 20 ng/ml of recombinant human (rh)GMCSF whilst 32D cells were either supplemented with 10% WEHI 3B conditioned medium or with murine IL3 (2 ng/ml).

Growth factors

Murine IL3, Human SCF from Peprotech EC Ltd, London, UK.

rh GMCSF, Behringwerke-Hoechst, Marburg, Germany.

Human Erythropoietin, POM ‘Eprex’, Janssen-Cilag, High Wycombe, UK.

Human IL3, Sandoz, Frimley Park, UK.

AML cell culture

Samples were obtained from AML patients presenting to University College Hospital, London, at presentation or relapse. Informed consent was obtained from all patients prior to obtaining the sample. All patients had circulating blasts in the peripheral blood and these were isolated by Ficoll gradient centrifugation. All samples tested had more than 90% blasts by morphology and or immunophenotyping. Mononuclear cells were frozen in an RPMI medium, FCS and 10% dimethyl sulfoxide (DMSO) and stored in liquid nitrogen. In this study, 50% of samples analysed were from frozen cells and the remainder was analysed fresh.

Frozen samples were thawed rapidly at 37° and slowly diluted with RPMI/10% FCS. Cells were then pelleted at 350 g for 3 min and resuspended in RPMI/10% FCS. No exogenous cytokines were added.

Peripheral blood lymphocyte culture

Fifty millilitres of peripheral blood was taken and the mononuclear cell layer isolated by Ficoll centrifugation. The mononuclear cells were re-suspended in RPMI/10% FCS. Isolated lymphocytes were incubated with 10 μg/ml of phytohaemagglutinin (PHA), a plant mitogen with specificity for T cells, for 72 h prior to incubation with the indicated concentration of inhibitor and stimulated with IL2 10 IU/ml.

Inhibitors

All stock solutions were made in DMSO and kept frozen at −20° until use. Gö6976 was from LClabs, Woburn, MA, USA; Gö6983, AG490, Compound 6/JAK inhibitor 1 was from Calbiochem, San Diego, CA, USA; ST was from Sigma Aldrich, Poole, Dorset, UK.

MTS [3-(4,5-dimethyl thiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt assay

Cells were suspended at 2 × 105 per point in 200 μl RPMI/10% FCS. The inhibitor under investigation was added at the appropriate concentration and the cells incubated for 48 h at 37°C, 5% CO2. MTS activity was measured by CellTiter kit (Promega, Southampton, UK) according to the manufacturers instructions. Results are expressed as a percentage of control (cells without inhibitor).

Western blotting

Western blotting was performed as previously described by Grandage et al (2005).

Antibodies

Phospho-STAT 5 (Y694) (9351), Phospho-STAT 3 (Tyr 705) (9131), Phospho-ERK MAPK p42/44 (Thr 202/Tyr204) and Phospho-Akt (Ser 473) were from Cell Signalling Technology, Hitchin, UK; JAK 2 PY1007/1008 (44–426) from Biosource Int., CA93012, Camarillo, CA, USA; Anti-HA probe (F-7) (sc-7392),STAT 5b (C-17) and Flt-3 from Santa-Cruz Biotechnology, Santa Cruz, CA, USA; Anti-phosphotyrosine 4G10(05–321) from Upstate Biotechnology Inc., Lake Placid, NY, USA.

Annexin V

Annexin V staining was performed according to standard techniques (Grandage et al, 2005).

Stimulation experiments

Cells from the factor-dependent cell lines were washed three times in sterile phosphate-buffered saline (PBS) and resuspended in RPMI/10% FCS without growth factors overnight. These cells were then preincubated for 30 min with the indicated concentration of inhibitor and stimulated with the appropriate cytokine. Total cell lysates were made after 10 min stimulation, Annexin V staining carried out at 24 h and MTS assay at 48 h. Factor-independent cell lines and primary AML cells were cultured in RPMI/10% FCS and incubated with the given concentration of inhibitor for a minimum of 6 h before total cell lysates were made. Again, Annexin V staining was carried out at 24 h and MTS assay at 48 h.

JAK2 kinase assay

The kinase assay was carried out by using recombinant JAK 2 agarose (Upstate, Lake Placid, NY, USA), according to the manufacturer's instructions. In brief, 10–20 μl of settled agarose beads was taken per assay point and washed twice with 1 ml of kinase buffer then aspirated to dryness. The agarose beads were then suspended in 10 μl of kinase buffer. Five microlitres of diluted inhibitor was then added to the appropriate tube and 5 μl of kinase buffer to the control tube. γ-32PATP was added to a final concentration of 37kBq/μl. The tubes were then incubated at 30°C for 30 min with agitation. The reaction was stopped by washing the JAK 2 agarose three times in storage buffer, 1 ml per wash. Fifty microlitres of sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE) sample buffer was then added and the sample boiled for 5 min. Twenty-five microlitres of sample was then subjected to 7·5% SDS–PAGE. The gel was fixed with Coomassie Blue stain, dried and visualised by autoradiography.

JAK3 kinase assay

293T cells were transfected with HA-JAK3 according to a standard calcium phosphate method, one 10-cm plate of confluent cells per point. The cells were lysed in 1 ml of lysis buffer on ice. Ten microlitres of anti-HA was added and the sample placed on a mixer wheel at 4°C overnight. Twenty-five microlitres of Protein G agarose beads 50:50 slurry was added and mixed at 4°C for 90 min. The immobilised JAK 3 was then washed twice in PBS 0·1% Triton-X 2 mmol/l EDTA. One further wash was carried out in Kinase buffer (NaCl 50 mmol/l, HEPES pH 7·6 10 mmol/l, MgCl2 5 mmol/l, MnCl2 5 mmol/l, Na3 V04 0·1 mmol/l) and the agarose beads were aspirated to dryness. Forty-five microlitres of kinase buffer was then added to each tube and 5 μl of indicated inhibitor concentration for 10 min at room temperature. A quantity of 370 kBq of γ-32PATP were then added and the tubes incubated for 30 min at room temperature with regular agitation. The reaction was stopped by the addition of 1 ml of cold PBS/0·1% Triton X-100/2 mmol/l EDTA and the beads washed twice further using microbio-spin chromatography columns Bio-Rad (Hemel Hempstead, Herts, UK) according to the manufacturer's protocol. The protein was eluted from the beads by the addition of 30 μl of boiling 1× sample buffer to the column for 3 min. The column was then placed inside a clean 1·5 ml centrifugation tube and spun at 2500 g for 5 min. The eluate was then boiled for 1 min further. Twenty-five microlitre of sample was then subjected to 7·5% PAGE. The gel was then fixed with Coomassie and dried and the JAK 3 visualised by autoradiography.

Plasmid constructs and electroporation

TEL–JAK2/TEL–JAK3 in pBABE-Neo were kindly provided by Dr V Lacronique, Paris, France. HA-JAK 2 and 3 in pEF-BOS vector were kindly donated by Dr Jane McGlade, Toronto, ON, Canada. WT Flt3 and Flt-3 ITD in pMKITNeo were kindly provided by T. Naoe (Hayakawa et al, 2000) and Flt3 D835Y in pMXNeo by R. Chopra (Paterson Institute for Cancer Research, Manchester, UK).

32D cells in culture were fed the night prior to electroporation to ensure that they were in log phase growth. The following day 10–20 × 106 cells per point were washed once in PBS. They were then resuspended in 500 ml phenol red-free RPMI medium (Sigma) containing 10% FCS and 10% WEHI 3B conditioned medium and 20 mg of plasmid were added. The cells were transferred to a gene pulser cuvette (0·4 cm) Bio-Rad and electroporated at 250 V and 960 mF in a Bio-Rad Gene PulserTM. Following the transfection, the cells were left to recover for 5–10 min and then transferred to 15–20 ml of RPMI/10% FCS+10% WEHI conditioned medium for 6 h. The cells were then passed through Ficoll to remove any dead cells and suspended in RPMI/10% FCS and 10% WEHI conditioned medium. The cells were allowed to grow for 24–48 h before selecting out the transfected cells with the appropriate antibiotic.

Mutational analysis for Flt3 and N-Ras

These analyses were performed as previously described (Grandage et al, 2005).

Results

Gö6976 inhibits signalling, proliferation and survival downstream of JAK2-utilising cytokines

Growth factor-deprived 32D cells preincubated with Gö6976 and stimulated with IL3 (10 ng/ml) showed a dose-dependent reduction in IL3-induced phosphorylation of STAT 5, ERK and Akt (Fig 1A, left panel) as well as a reduction in total phosphotyrosine-containing proteins (Fig 1A, right panel). Incubation with Gö6976 blocked the proliferative effects of IL3 with a 50% inhibitory concentration (IC50) of 93 ± 50 nmol/l (n = 4) and complete abrogation at 0·5 mmol/l (Fig 1B). Withdrawal of IL3 from FDCP-1 cells, another IL3-dependent line led to a 34% ± 9% increase in apoptosis above control cells (growing with IL3) after 24 h. Incubation of the FDCP1 cells with 1 mmol/l Gö6976 plus IL3 led to a 22% ± 6% increase in apoptosis showing that Gö6976 could inhibit the protective effects of IL3 on survival.

Figure 1.

 Gö6976 inhibits signalling and proliferation downstream of interleukin 3 (IL3) in a factor-dependent cell line. (A) Signalling: 32D cells were quiesced overnight, preincubated for 30 min with the indicated concentration of Gö6976 (Go76), and then stimulated with IL3 (10 ng/ml) for 10 min. Immunoblot analysis was carried out with the indicated antibodies (phosphorylated STAT 5, Akt and ERK in left panel and phosphotyrosine in right panel). Comparison was made with a control PKC inhibitor, Gö6983 (Go83) and the JAK2 inhibitor AG490 (AG). Total STAT 5 is shown as a protein loading control. (B) Proliferation: 32D cells were incubated with IL3 plus Gö6976 or Gö6983 and cell number determined by MTS activity after 48 h. (C) Gö6983 inhibits ERK phosphorylation in response to phorbol ester stimulation. (C) Growth factor-starved Mo7E cells were preincubated with the indicated concentration of Gö6983 and stimulated with Phorbol Ester (TPA) (1 μg/ml) for 10 min. Immunoblot analysis was carried out with the indicated antibodies. (D) Gö6976 inhibits signalling downstream of JAK2 utilising cytokines. Factor-dependent cells lines were quiesced overnight, preincubated with increasing concentrations of Gö6976 for 30 min and stimulated with the indicated cytokine for 10 min. Mo7E cells were stimulated with either granulocyte-macrophage colony-stimulating factor or thrombopoietin and TF-1 cells with erythropoietin. Immunoblot analysis was carried out with the indicated antibodies. (E) Gö6976 does not inhibit the signalling downstream of stem cell factor (SCF). Quiescent Mo7E cells were preincubated with the indicated concentrations of Gö6976 (Go76) or Gö6983 (Go83), then stimulated with SCF (20 ng/ml, 10 min) and immunoblot analysis carried out with the indicated antibodies.

In order to assess the contribution of PKC inhibition to the effects of Gö6976 on IL3 signalling, we used a broad spectrum PKC inhibitor, Gö6983 as a control. The control PKC inhibitor did not have significant effects on signalling, proliferation (Fig 1) or apoptosis indicating that the effects of Gö6976 on IL3 signalling are not related to its anti-PKC activity. To confirm that Gö6983 was active against PKC at the concentrations used in these experiments, its effect on phorbol ester-induced activation of ERK was examined. Incubation with Gö6983 resulted in a dose-dependent decrease in ERK phosphorylation (Fig 1C). These results indicate that Gö6983 is an active PKC inhibitor and that its lack of effect on IL3 signalling is not due to a general lack of biological activity.

Similar effects of Gö6976 were seen on GMCSF-induced signalling and survival/proliferation. Incubation with 1 mmol/l Gö6976 in the presence of GMCSF led to a 29% increase in apoptosis of TF-1 cells (0% for Gö6983 n = 3), a 60% ± 7% reduction in proliferation of Mo7E cells (2% ± 7% for Gö6983 n = 3) and inhibition of GMCSF-induced STAT 5 phosphorylation (Fig 1D). The effect of Gö6976 on signalling downstream of other members of the cytokine receptor superfamily, TPO and EPO was examined. In both cases, Gö6976 abrogated the stimulatory effects of the cytokine on STAT 5 phosphorylation (Fig 1D). In addition, Gö6976 blocked the effects of IL6 on STAT 5 activation (data not shown).

Stem cell factor signals through KIT, which possesses intrinsic tyrosine kinase activity and does not utilise JAKs (Ward et al, 2000). SCF stimulation rapidly brought about Akt and ERK phosphorylation, which was not affected by incubation with Gö6976 (Fig 1E).

Together, these results suggest that Gö6976 exerts its effects through inhibition of a kinase other than PKC. Identical results were obtained by using Gö6976 from three separate commercial sources (LC Labs, Biomol, Calbiochem; data not shown). The fact that signalling downstream of several members of the cytokine receptor superfamily, but not that of SCF, was affected led us to postulate an effect on JAK 2.

Gö6976 has a direct inhibitory effect on JAK2 in vitro kinase activity

In order to assess the effect of Gö6976 on JAK2 activity, in vitro kinase assays were carried out. Incubation of recombinant JAK2 with 1 mmol/l Gö6976 led to a marked reduction in JAK2 kinase activity compared with the control (no inhibitor) whereas incubation with Gö6983 had no effect. (Fig 2A) Incubation with 1 μmol/l Gö6976 reduced JAK 2 activity to 18 ± 5% of control (n = 5) with an IC50 of 130 nmol/l (Fig 2B). The effect of AG490 on JAK2 phosphorylation was compared with that of Gö6976. Incubation with 100 μmol/l AG490 led to a 70% reduction in JAK2 kinase activity whereas 1 μmol/l Gö6976 by comparison led to a 74% reduction in the same experiment (Fig 2C). The IC50 for AG490-induced JAK2 inhibition was found to be 35 μmol/l (data not shown). These results showed that Gö6976 is a JAK2 inhibitor.

Figure 2.

 Gö6976 reduces the JAK2 autokinase activity in an in vitro kinase assay. (A) Recombinant JAK2 was used in a kinase assay with 32P labelled ATP in the absence or presence of Gö6976 (Go76) or Gö6983 (Go83) and phosphorylation of JAK2 (125 kD) detected by autoradiography (upper panel). Equal loading was measured by immunoblotting with anti-JAK2 (lower panel). (B) Autokinase activity was measured in the presence of increasing concentrations of Gö6976. (C) Autokinase activity was measured in the presence of Gö6976, Gö6983 and the known JAK2 inhibitor, AG490 (AG).

Gö6976 reduces the in vitro kinase activity of JAK3 and inhibits IL2 signalling

HA-tagged JAK 3 was expressed in 293T cells, immunoprecipitated and in vitro kinase assays carried out. The activity of Gö6976 was compared with AG490 and Gö6983. Incubation with 1 μmol/l Gö6976 led to a marked reduction in JAK3 in vitro kinase activity, comparable with the effects of 100 μmol/l AG490. Gö6983 had no effect on JAK3 autophosphorylation (Fig 3A).

Figure 3.

 Gö6976 has a direct effect on the in vitro kinase activity of JAK3 and inhibits interleukin 2 (IL2)-induced signalling and proliferation. (A) HA-tagged JAK 3 was transfected into 293T cells, immunoprecipitated with an anti-HA antibody and an in vitro kinase assay with 32P labelled ATP was carried out. The upper panel is an autoradiograph showing JAK3 autophosphorylation and the lower panel is an immunoblot with anti-HA antibody as a loading control. The activity of Gö6976 was compared with AG490 and Gö6983. Incubation with 1 μmol/l Gö6976 led to an 81% reduction in JAK3 in vitro kinase activity compared with a 64% reduction for 100 μmol/l AG490. Gö6983 had no effect on JAK3 phosphorylation. (B) Peripheral blood-derived lymphocytes were cultured in 10 μg/ml phytohaemaglutinin (PHA) for 72 h, incubated with Gö6976 for 30 min and stimulated with IL2 (10 IU/ml). Immunoblot analysis was carried out with the indicated antibodies. (C) The above experiment was repeated and proliferation assessed by an MTS assay carried out at 72 h after IL2 stimulation.

To examine the effect of Gö6976 on JAK3 in a whole cell system, its consequence on IL2 signalling, which is known to be JAK3 dependent (Ihle, 1995), in PHA-stimulated lymphocytes was investigated. Peripheral blood-derived lymphocytes were incubated with 10 mg/ml PHA for 72 h without IL2. They were then preincubated with Gö6976 and stimulated with IL2 (10 IU/ml). STAT 5 phosphorylation was reduced by Gö6976 in a dose-dependent manner (Fig 3B). The above experiment was repeated and an MTS assay carried out at 72 h after IL2 addition. Incubation with Gö6976 led to a dose-dependent reduction in proliferation with an IC50 of 370 nmol/l (Fig 3C). These results showed that Gö6976 is able to inhibit JAK3 in vitro kinase activity and abrogates IL2-induced STAT 5 phosphorylation in intact cells.

Gö6976 also inhibits the TEL–JAK2 fusion and the JAK2 V617F mutant proteins

The data so far show that Gö6976 can inhibit cytokine-induced changes in signalling, survival and proliferation mediated by JAK2 and 3. The leukaemia-associated TEL–JAK 2 fusion protein was stably expressed in 32D cells resulting in growth factor independence in these cells, which normally require IL3 for survival. We detected constitutive activation of PI3K/Akt, MAPK and STATs 3 and 5 in TEL–JAK2 expressing cells (Fig 4A and B). Incubation with Gö6976 for 6 h led to a dose-dependent decrease in tyrosine phosphorylated proteins, as well as phosphorylation of STAT 5, ERK and Akt (Fig 4A and B). Incubation with 1 μmol/l Gö6976 increased apoptosis by an average of 20 ± 3% at 24 h (IC50 of 242 ± 15 nmol/l n = 5) and proliferation fell to an average of 37 ± 8% of control cells (no inhibitor) at 48 h (IC50 291 ± 44 nmol/l n = 7) (Fig 4C). Gö6983 had no significant effect on signalling, apoptosis or proliferation (Fig 4). Gö6976, therefore, is able to inhibit the signalling downstream of the TEL–JAK2 fusion protein and negate its effects on survival and proliferation. The TEL–JAK3 fusion has not been described in leukaemia, but the construct is useful to investigate the signalling downstream of JAK3. 32D cells stably expressing the TEL–JAK3 fusion were rendered factor-independent and showed constitutive STAT 5 phosphorylation. Incubation with 1 μmol/l Gö6976 led to a reduction in phosphorylated STAT 5 (Fig 4D), an increase in apoptosis of 19% and a reduction in proliferation (IC50 of 144 ± 50 nmol/l n = 5) (Fig 4E).

Figure 4.

 Gö6976 inhibits TEL–JAK2 and TEL–JAK3 fusion protein-induced signalling and proliferation. (A, B) 32D cells stably expressing the leukaemia-associated TEL–JAK2 fusion were incubated with Gö6976 for 6 h and immunoblot analysis carried out with indicated antibodies. In 5b, lane 1 (marked 32D) contains lysate from parental 32D cells continuously growing in the presence of murine interleukin 3. (C) TEL–JAK2 expressing 32D cells were incubated with the indicated concentration of Gö6976 and proliferation at 48h assessed by MTS assay (n = 7). (D) TEL–JAK3 expressing 32D cells were incubated with increasing concentrations of Gö6976 for 6 h prior to immunoblot analysis with indicated antibodies. (E) TEL–JAK3 expressing 32D cells were incubated with the indicated concentration of Gö6976 and proliferation at 48 h assessed by MTS assay (n = 5).

The JAK2 V617F mutation is present in MPDs including polycythemia vera and results in its constitutive activation (Tefferi & Gilliland, 2005; Tefferi & Spivak, 2005). We utilised the HEL cell line, which has a homozygous JAK2 V617F mutation (Levine et al, 2005a; our unpublished data), to evaluate the effects of Gö6976 on JAK2 V617F-induced signalling and proliferation. The JAK inhibitor, known as Compound 6 or JAK Inhibitor 1 (JI1), previously shown to be active against JAK2 V617F (Levine et al, 2005a), was used for comparison. Figure 5A and B shows that Gö6976 and JI1 both inhibited the downstream signalling pathways in HEL cells (STAT 5, Akt, GSK3, ERK/MAPK, PIM1) and also decrease cell proliferation in a dose-dependent manner. These results indicated that Gö6976 is an effective inhibitor of JAK2 V617F.

Figure 5.

 Gö6976 inhibits the signalling and proliferation induced by JAK2 V617F. (A, B) HEL cells, which have a hemizygous JAK2 V617F mutation, were incubated with increasing concentrations of Gö6976 or JAK inhibitor 1 (Compound 6). Cell lysates were made after 6 h and analysed by immunoblot with the indicated antibodies. (B) samples were analysed by MTS assay after 48 h and are expressed as a % of control value (no inhibitor).

Gö6976 does not inhibit JAK2 phosphorylation on tyrosines 1007/1008

Several studies have shown that JAK2 is phosphorylated on a number of tyrosine residues including Y1007/1008, Y221, Y570 and Y813 in response to cytokine stimulation (Argetsinger et al, 2004; Kurzer et al, 2004). JAK2 mutant studies have shown that phosphorylation of Y1007 in the activation loop allows access of the catalytic loop to ATP in the ATP binding domain and is essential for kinase activity. We investigated JAK2 phosphorylation using an antibody specific to JAK2 that is phosphorylated at tyrosines 1007 and 1008.

Quiescent Mo7E cells were incubated with increasing concentrations of Gö6976, stimulated with GMCSF and cell lysates probed with antibodies to phospho JAK2, pAkt and pERK (Fig 6A). GMCSF induced JAK2 phosphorylation at tyrosines 1007/1008 – Gö6976 did not affect the phosphorylation at these residues although there was abrogation of downstream signalling (Fig 6A). To further investigate this finding, 32D cells stably transfected with HA-tagged JAK2 were starved, preincubated with Gö6976 and stimulated with IL3. JAK2 was immunoprecipitated by using an anti-HA antibody and samples probed with anti-phosphotyrosine and phospho-specific JAK2 (Y1007/1008) antibodies. The results are shown in Fig 6B. JAK2 was tyrosine phosphorylated after stimulation with IL3 – incubation with Gö6976 led to a reduction in total JAK2 phosphotyrosine content, but not back to quiescent levels. Similar results were seen with ST. The samples were re-run and probed with the phospho-JAK2 Y1007/1008 antibody (Fig 6B). Similar to the Mo7e experiment, JAK2 phosphorylation at Y1007/1008 was not reduced by incubation with Gö6976. Total cell lysates from the same experiment showed that downstream signalling was inhibited (data not shown). ST appeared to behave in the same way as Gö6976, whereas AG490 reduced JAK2 phosphorylation at Y1007/1008 (Fig 6B). Incubation of JAK2 V617F-expressing HEL cells with either Gö6976 or JI1 for 4 h led to increased detection of Y1007/1008 phosphorylated JAK2 without a significant change in total JAK2 levels (Fig 6C and D).

Figure 6.

 Gö6976 does not inhibit JAK2 phosphorylation on tyrosines 1007/1008. (A) Quiescent Mo7E cells were preincubated with increasing concentrations of Gö6976 or Gö6983, stimulated with GMCSF(20 ng/ml, 10 min) and cell lysates analysed by immunoblot with antibodies to phosphorylated JAK2 (tyrosines 1007/1008), p Akt and pERK. (B) Quiescent 32D cells stably expressing HA-tagged JAK2 were preincubated with 1 μmol/l Gö6976 (Go76), 100 μmol/l AG490 (AG) or 1 μmol/l staurosporine and stimulated with interleukin 3 (10 ng/ml, 10 min). JAK2 was immunoprecipitated by using an anti-HA antibody and samples probed by immunoblotting with anti-phosphotyrosine and phospho-specific JAK2 (1007/1008) antibodies. The last lane, labelled 32D, is an immunoprecipitate from parental 32D cells. (C) HEL cells expressing JAK2 V617F were incubated with Gö6976 (Go76) 1 μmol/l or JAK inhibitor 1 (JI1) 1 μmol/l for 4 h and cell lysates analysed by immunoblot with antibodies to phosphorylated JAK2 (tyrosines 1007/1008), total JAK2 and pSTAT 5. (D) HEL cells expressing JAK2 V617F were incubated with Gö6976 (Go76) or JAK inhibitor 1 (JI1) for 4 h at the indicated concentrations and cell lysates analysed by immunoblot with antibodies to phosphorylated JAK2 (tyrosines 1007/1008) and tubulin (this is a re-probe with pJAK2 of the blot in Fig 5A, upper panel, and therefore the tubulin loading from this is shown again).

The effect of Gö6976 on primary AML cells and its identification as a dual inhibitor of JAK2 and FLT3

We extensively characterised 18 cases of primary AML for the constitutive activation of signalling pathways (STAT 3/5, Akt, MAPK) and the response of these pathways to Gö6976; mutations in FLT3 and N-Ras, and for the effects of Gö6976 on cell survival/proliferation (Table I). Sixteen out of 18 patient samples were found to have constitutive STAT activation; seven had activation of both STAT 3 and STAT 5 and 9 of STAT 3 alone. Incubation of primary AML cells with Gö6976 led to decreased STAT phosphorylation in all cases and also to variable effects on phosphorylation of Akt and ERK/MAPK (Fig 7A). We noted that Gö6976 reduced the constitutive phosphorylation of both Akt and ERK/MAPK in 4/5 cases with FLT3-ITD and 7/13 cases with FLT3-WT. In the primary AML cases, those with FLT3 mutations showed a trend towards greater sensitivity to Gö6976 with regard to inhibition of proliferation and survival (Fig 7B), displaying a mean reduction in MTS activity in ITD-positive cases to 55 ± 5% of control and in FLT3-WT AML to 69 ± 5% of control (P = 0·06 for comparison between FLT3-ITD and FLT3-WT, Wilcoxon's test). The control inhibitor Gö6983 had no significant effects on cell survival (103 ± 8% for total cohort).

Table I.   Patient characteristics and presence of phosphorylated STAT3 and STAT5.
PatientSTAT 3STAT 5FAB typePresenting WCC (×109/l)CytogeneticsFLT3 statusNRAS
  1. FAB, French–American–British classification; WCC, white cell count; ITD, internal tandem duplication; WT, wild type; EX1/2, exon 1/2.

1+M23546xyWTEX1
2+M140146xxITDWT
3++M5b5246xxWTWT
4++M2117Del(9) q12q34WTEX2
5++M4233Inv16(p13q22)WTEX2
6++Trans ET146complexWTWT
7M229t(8;21)+yWTN/A
8+M569failedWTWT
9++M114046xyITDWT
10+M5a8946xxITDEX1
11N/Arelapsed3646xx,r(7)(?p?q)t(9;22)(q34;q11)WTWT
12+M247Trisomy 21ITD/D835WT
13++2°MDS38·1N/AWTWT
14+M5b18846xxWTWT
15++M13346xyITDWT
16+NKNK46xxWTWT
17+M2845xyadd(8)(q22)add(15)(p1)add(21)(p1)WTEX2
18+M4541T(10;11) Progressed with 7q-WTN/A
Figure 7.

 The effect of Gö6976 in primary acute myeloid leukaemia (AML) cells and its identification as a dual inhibitor of JAK2 and FLT3. (A) Primary AML cells were incubated ±1 μmol/l Gö6976 or Gö6983 as a control for 6 h, total cell lysates made and analysed by immunoblot with antibodies to phosphorylated STAT 3, 5, Akt and ERK. Four representative samples are shown. Examples 1 and 3 are from patients with FLT3-ITD mutation and samples 2 and 4 from those with FLT3-WT. (B) Blast samples from 18 cases of primary AML were incubated with Gö6976 (1 μmol/l) for 48 h prior to MTS assay. Points represent individual patient samples divided into two categories by the absence (ITD-NEG) or presence (ITD-POS) of a FLT3 mutation. (7) Upper panel: 32D cells expressing WT FLT3 were starved of interleukin 3 for 4 h and then incubated in 1 μmol/l Gö6976 or 25 nmol/l CEP-701, a known FLT3 inhibitor, for 30 min prior to stimulation with FLT3-ligand (FL) for 10 min. Lysates were then immunoprecipitated with anti-FLT3 antibody and immunoblots probed for phosphotyrosine (pY) and total FLT3. Lower panel: IL-3-independent 32D cells expressing a FLT3-ITD mutant were incubated with increasing concentrations of Gö6976 for 6 h and cell lysates made and analysed by immunoblot with the indicated antibodies. (D) Factor-independent 32D cells expressing FLT3-ITD or FLT3 D835Y (TKD) were incubated in the known FLT3 inhibitor CEP701 (CEP, 25 nmol/l), in Gö6976 (Go76, 0·25 μmol/l) or in the known JAK inhibitor compound 6/JAK Inhibitor 1 (JI1, 0·25 μmol/l) for 48 h and MTS assay carried out. WT-FLT3 expressing 32D cells were incubated in FLT3 ligand (FL, 20 ng/ml) and the indicated inhibitors prior to MTS assay.

These findings prompted us to examine the effects of Gö6976 on FLT3 signalling. FLT3-WT, FLT3-ITD and FLT3-D835Y (TKD) mutants were expressed in the 32D cell line. Figure 7C and D shows that Gö6976 was also an inhibitor of FLT3 with effects on WT, ITD and TKD mutants with regard to both signalling and proliferation. As a control for JAK inhibition, Compound 6/JAK inhibitor 1 did not significantly affect the proliferation induced by FLT3-ITD or TKD mutants (Fig 7D). These results indicated that Gö6976 has anti-FLT3 activity.

The effect of Gö6976 on primary normal human progenitor cells

We assessed the effect of JAK inhibition on normal progenitors in colony-forming assays as well as in liquid culture. Both Gö6976 and JAK inhibitor 1 inhibit colony formation (measured after 2 weeks) with a more potent effect on erythroid precursors (Fig 8A and B). In shorter-term liquid culture assays, Gö6976 had modest effects on cell survival (8C) and more significant activity in a proliferation assay (8D).

Figure 8.

 The effect of Gö6976 on normal human progenitor cells. (A, B) Normal human progenitors were placed in standard methylcellulose-based colony assays with or without the indicated concentration of Gö6976 or JAK inhibitor 1 and colonies scored after 2 weeks. Mean of two independent assays. (C, D) Normal human CD34+ cells were incubated in the absence of cytokines (Nil), with growth factors [interleukin 3 (IL-3), IL6, SCF, Flt3L] or with growth factors plus the indicated concentration of Gö6976. Cell survival was measured by Annexin V flow cytometric assay (8C) and proliferation by MTS assay (8D) after 48-h incubation.

Discussion

We have shown that Gö6976, an indolocarbazole inhibitor of classical PKC isoenzymes as well as PKD/PKCμ is also a potent inhibitor of JAK2 and JAK3 at nanomolar concentrations. Gö6976 is able to inhibit signalling downstream of several members of the cytokine receptor superfamily leading to an increase in apoptosis and a reduction in proliferation. Several papers have utilised Gö6976 to examine the effect of PKC inhibition in haemopoietic cells. Iankov et al (2002) found that Gö6976 inhibited cell proliferation of IL6-dependent plasmacytoma cells whereas Gö6983 did not, and concluded that this was due to an effect on PKCμ. In the light of our findings, the effects of Gö6976 could potentially have been due to JAK kinase inhibition (Iankov et al, 2002). Amin et al (2000) examined PKC inhibition in acute promyelocytic leukaemia cell lines and found that apoptosis was increased with Gö6976. The present results show that interpretation of these papers and others investigating the anti-PKC effects of Gö6976 in haemopoietic cells becomes more complex.

There is an increasing evidence for the involvement of JAK2 in haematological malignancies with the descriptions of TEL–JAK2 and PCM1-JAK2 fusion genes in leukaemia and myeloproliferative disease (Lacronique et al, 2000; Reiter et al, 2005); the amplification and overexpression of JAK2 in certain lymphomas (Joos et al, 2000; Pileri et al, 2003; Ruchatz et al, 2003; Guiter et al, 2004) and most recently with the identification of the V617F mutation at very high frequencies in polycythemia vera, essential thrombocythemia and idiopathic myelofibrosis (James et al, 2005; Levine et al, 2005a; Steensma et al, 2005). Our results show that Gö6976 can inhibit the signalling from JAK2 in its wild type (WT), fusion protein and V617F forms and may form the basis for the identification of a clinically useful inhibitor. Preliminary work shows that Gö6976 can inhibit the survival of primary erythroid progenitors from polycythaemia vera (PV) patients with the JAK2 V617F mutation (unpubl. obs.).

Several groups have investigated which of the 49 tyrosines in JAK2 are autophosphorylated (Argetsinger et al, 2004; Kurzer et al, 2004; Matsuda et al, 2004). In many kinases, regulation of catalytic activity by phosphorylation occurs on residues within the activation loop of the kinase domain. Feng et al (1997) demonstrated that Y1007 and Y1008 are sites of trans- or autophosphorylation in vivo and in in vitro kinase reactions. They found that mutation of Y1007, or both Y1007 and Y1008, to phenylalanine essentially eliminated kinase activity, whereas mutation of Y1008 to phenylalanine had no detectable effect on kinase activity (Feng et al, 1997). Other tyrosine residues known to be autophosphorylated include tyrosines 221, 570 and 813 (Argetsinger et al, 2004; Kurzer et al, 2004). Although we have demonstrated that Gö6976 is a JAK2 inhibitor, we found that JAK2 phosphorylation at Y1007/1008 in whole cells, as assessed by Western blot with a phosphospecific antibody, was not reduced by incubation with Gö6976 despite a reduction in total JAK2 phosphotyrosine content and a clear inhibition of downstream signalling. We obtained similar findings with ST and JAK inhibitor 1 (Thompson et al, 2002) but not with the tyrphostin AG490. Our findings suggest that Gö6976 preferentially binds to and inhibits JAK2, which is already phosphorylated at Y1007/1008. This inhibits kinase activity and hence phosphorylation of other tyrosine residues in JAK2 as well as various substrates, preventing activation of downstream signalling. In support of this hypothesis, Boggon et al (2005) reported on the crystal structure of JAK3 in complex with AFN941, an ST analogue. They found that the JAK3-AFN941 complex is crystallised in a catalytically active state and is tyrosine phosphorylated on both autophosphorylation sites in the activation loop. AFN 941 binds JAK3 in the catalytic cleft between the N- and C-lobes. Sequence identity between JAK2 and 3 is 62% with almost all residues in the inhibitor-binding cleft conserved and it is likely that ST and by implication Gö6976, binds to JAK2 that is phosphorylated at Y1007/1008 and is in an active conformation. More recently, Lucet et al (2006) have reported the crystal structure of the kinase domain of JAK2 and shown that compound 6/JAK inhibitor 1 also binds to JAK2 that is in the active conformation and phosphorylated at Y1007/1008. We have shown that Jak Inhibitor 1, like Gö6976, does not inhibit JAK2 phosphorylation at Y1007/1008 in intact cells. Incubation of cells expressing mutant JAK2 V617F with Go6976 or JAK inhibitor 1 resulted in increased levels of Y1007/1008 phosphorylated JAK2. This could potentially be due to a reduction in the expression of downregulatory molecules, such as suppressor of cytokine signaling family members, which rely on continued JAK/STAT activity for their expression. Other kinase inhibitors have been shown to bind distinct conformations of their target molecules – for example, imatinib binds to and inhibits an inactive conformation of Abl whereas the compound PD173955 binds to and inhibits Abl in multiple, including active, conformations (Nagar et al, 2002). Our findings also have implications for the screening of JAK2 inhibitory compounds, if phosphorylation at Y1007/1008 is used as the endpoint, potentially useful inhibitors could be scored as inactive.

Having identified Gö6976 as a potent JAK inhibitor, we were interested to examine its effects on signalling, survival and proliferation in primary AML cells. Constitutive STAT activation has previously been described in AML and we found constitutive tyrosine phosphorylation of STATs 3 and/or 5 in 16 of 18 primary AML samples examined. This was reduced by incubation with Gö6976 in all cases. All constitutively active signalling pathways present (STAT, Akt, ERK/MAPK) were inhibited in four of five cases positive for the FLT3-ITD and in 7/13 cases with FLT3-WT and that there was a trend for FLT3-ITD cases to show the increased sensitivity to Gö6976 with regard to survival/proliferation. This prompted us to examine the effects of Gö6976 on FLT3; it was found that this compound could inhibit WT and mutant forms of FLT3 in intact cells. Inhibition of more than one kinase may be advantageous in the treatment of malignancy, as has been suggested for dual inhibitors of Src and Abl kinases in CML (Martinelli et al, 2005) – indeed, it has also been postulated that the effectiveness of imatinib in CML is attributable to inhibition of both Abl and Kit kinases (Wong et al, 2004). The mechanism for activation of STAT signalling in the FLT3-WT cases is unclear, but is unlikely to be due to the JAK2 V617F mutation, as this is rarely found in de novo AML (Levine et al, 2005b). Autocrine/paracrine growth factor secretion is well described in AML and could contribute to STAT activation (Delwel et al, 1989; Murohashi et al, 1989; Schuringa et al, 2000). Further work is in progress to characterise the molecular mechanisms leading to aberrant STAT activity in AML.

In conclusion, we have shown that Gö6976 is a potent inhibitor of WT JAK2 as well as of the mutant forms (JAK2 V617F and TEL–JAK2) found in haematological malignancies. Gö6976 also has activity against mutant forms of FLT3 and is active in inhibiting the survival and proliferation of primary AML cells. These data may help identify clinically useful compounds based on the structure of Gö6976, which can be employed in the treatment of MPDs as well as AML.

Acknowledgements

VL Grandage was supported by the Kay Kendall Leukaemia Foundation and T Everington by Leukaemia Research, UK.

Ancillary