Effect of the JAK2/STAT3 inhibitor SAR317461 on human glioblastoma tumorspheres

Glioblastoma (GBM) is the most common brain cancer in adults and is notorious for its diffuse invasion, heterogeneity, treatment resistance, and dismal 5 year patient survival of less than 5 % [1, 2]. The current front-line GBM therapeutic temozolomide (TMZ) combined with intensive megavoltage radiotherapy exerts some benefit, but falls far short in terms of efficacy and systemic toxicity [3, 4]. The growth of human GBM cell lines and their TMZ resistance has been inhibited by RNA based knockdown and by small molecule inhibition of STAT3, a latent transcription factor that is dysregulated in many cancers [5‐10]. Constitutive activation of STAT3 in gliomas is positively associated with tumor grade [6, 7, 11], and TCGA data reveals that elevated STAT3 in GBM is associated with reduced patient survival (Fig. 1—see “Methods”).

STAT3 is a major intracellular signaling hub and its downstream targets include Bcl-2, Bcl-X_L, c-myc, survivin, cyclin D1, vascular endothelial growth factor (VEGF), among others [12‐15]. STAT3 is activated when the signal transduction protein janus kinase 2 (JAK2) is stimulated by any of a variety of cytokines (IL-6, IL-11) and growth factors (EGF, TGF-α, PDGF and HGF) to phosphorylate the STAT3 tyrosine 705 residue [13, 14]. Constitutively activated STAT3 promotes tumor cell cycle progression and survival, elicits angiogenesis, and suppresses the immune response to tumors [16]. STAT3 is reportedly not essential to the survival of normal cells, but is critical to many types of cancer cells, making it a potentially valuable therapeutic target [16].

While positive results have been reported for JAK-STAT inhibition in preclinical GBM models, no human glioma JAK-STAT inhibitor based clinical studies have yet been performed, although one trial for a small molecule JAK-STAT inhibitor, WP1066, is pending (U.S. National Clinical Trials Database—https://​clinicaltrials.​gov/​) [8‐10]. This relative lack of activity may be a result of two considerations, (1) the preclinical results with JAK-STAT inhibition in GBM are variable, and (2) STAT3 signaling is nuanced since STAT3 can act, paradoxically, as a tumor promoter or suppressor, depending on factors that require further clarification [17, 18]. Nonetheless we were encouraged by reports of selective STAT3 blockade using small interfering RNA (siRNA) and small hairpin RNAs (shRNA) and by pharmacologic agents such as WP1066, that have inhibited GBM cell proliferation in vitro and tumor growth in vivo [5, 18, 19].

Much remains to be learned about the JAK-STAT pathway as a potential therapeutic target in GBM, and recent data motivated us to pursue the hypothesis that a potent newer generation JAK-STAT targeted inhibitor, SAR31746, could suppress multiple GBM lines representing genetic and phenotypic heterogeneity [18]. Heterogeneity is a hallmark GBM characteristic that has stymied other agents and called into question the use of JAK-STAT inhibition for this disease [18]. Structure based design was used to develop SAR317461 (formerly designated as TG101209; Sanofi-Aventis, Cambridge, MA, USA) and its analog SAR302503, both of which are orally bioavailable and have well-defined pharmacokinetics [20]. SAR302503 is approved for patients in primary myelofibrosis (PMF) and is well tolerated. SAR317461 was designed to be highly JAK2 selective, and it has exhibited marked antitumor activity in several preclinical models [20, 21]. To our knowledge neither compound has ever been tested with GBM cell lines [22, 23].

Here we report initial steps to address the following hypotheses; (1) that SAR317461 can overcome and kill immortalized, serum dependent GBM cell lines, as well as human GBM stem cell (GSC) derived tumorspheres that exhibit the genotypic and phenotypic heterogeneity of the human disease, (2) that the general mechanism of SAR317461 induced GBM cell death is apoptosis, (3) that sensitivity of GBM to SAR317461 depends on the presence of an activated JAK-STAT pathway, and (4) that inhibition of other survival mechanisms, specifically autophagy, may increase the potency of SAR317461. We also sought to determine whether SAR317461 affects the phosphorylation of STAT5 and Akt, in addition to STAT3, as is the case with WP1066 [24]. We present pilot data that for the first time chronicle the effects of SAR317461 on seven patient GSC derived tumorsphere lines and three established, immortalized GBM cell lines. SAR317461 exhibited potent activity against all lines that expressed activated STAT3, and inhibited the phosphorylation of STAT3 but not of STAT5 or Akt. The mechanism of STAT3 induced GBM tumorsphere cell death was via PARP mediated apoptosis, and we further discovered that autophagy inhibitors increased the anti-GBM effect of SAR317461. These novel data which characterize the activity of SAR317461 against a varied array of GBM lines warrant further, more comprehensive explorations of the JAK-STAT pathway as a target for GBM therapy.

STAT3 is a key intracellular effector for cytokine and growth factor-dependent transcriptional activation of genes that control proliferation, differentiation and apoptosis of a wide range of normal as well as transformed cell types [7]. Moreover, recent studies have indicated that STAT3 expression in GBM is elevated (Fig. 1), and that GBM tumors and cell lines exhibit high levels of constitutively activated STAT3 relative to normal human astrocytes, white matter, and normal tissue surrounding GBM tumors [9]. However, in preclinical models STAT3 has been reported to play either an oncogenic or suppressive role depending on the genetic background of the tumor system, and it is this variability that has in part complicated the preclinical development of JAK-STAT inhibitors [17]. For example de la Iglesia et al. [28] found that when EGFR is mutated STAT3 is oncogenic, but in tumors with PTEN mutations STAT3 can act as a tumor suppressor. The basis for this apparent functional difference is unknown. Moreover, preclinical results with previous JAK-STAT inhibitors produced variable levels of anti-GBM activity [22]. However the present study demonstrated that inhibition of STAT3 activation with the potent JAK2 inhibitor SAR317461 effectively induced apoptosis in a heterogeneous array of human GSC derived tumorspheres and established GBM cell lines. SAR317461 also triggered autophagy and importantly, the addition of an autophagy inhibitor led to markedly enhanced cell killing.

A major difficulty in targeting GBM is the heterogeneity of mutations and signaling aberrations that occur within a single tumor and between patient tumors [29]. The intra- and extratumoral heterogeneity of multiple GBM tumorsphere lines we tested did not appear to prevent SAR317461 from exerting potent anti-tumor activity. SAR317461 suppressed cell viability in several patient derived GSC lines with an IC₅₀ of 1–2 µM, as well as established GBM lines with an IC₅₀ of 2–10 µM (Fig. 3). Notably in one patient derived line, SK892, which had no constitutive expression of activated STAT3 (pSTAT3) the IC₅₀ was much higher (25 µM) than the other GSC lines. This finding is isolated but does align with the possibility that pSTAT3 expression may be associated with the ability of SAR317461 to exert a comparatively potent suppressive effect on GBM tumors.

The present study suggests that the mechanism by which SAR317461 inhibits GBM tumorspheres is apoptosis and that PARP cleavage plays a role. The relevant pathways remain to be identified, although it is known that targets of STAT3 include various prosurvival genes including those for Bcl-2, Mcl-1, and Bcl-X_L [17]. Furthermore, it has been reported for multiple myeloma that SAR317461 induced down regulation of pJAK2 and pSTAT3 levels that correlated with up-regulation of pErk and pAkt, indicating cross talk between these signaling pathways that is yet to be delineated in detail [22]. Down regulation of the antiapoptotic protein Bcl-X_L has been reported in myeloma and U251 cells with JAK2 inhibition, but curiously, in our patient derived GSC tumorspheres down-regulation of Bcl-X_L was not apparent in the Western blot analysis we performed [9, 30]. The reasons for this are unclear, although it is possible that SAR317461 induction of PARP cleavage and apoptosis in GSC tumorspheres is based on different signaling than in multiple myeloma and immortalized GBM lines. For example Epling-Burnette et al. [31] reported that induction of apoptosis was independent of Bcl-X_L regulation in large granular lymphocyte (LGL) leukemia. Instead, they found reduced protein expression of another Bcl-2-family protein, Mcl-1 which is known to be modulated by STAT3 [13].

Interestingly we also found that SAR317461 induced autophagy in tumorspheres. Autophagy can have a cytoprotective role in tumor cells and we found that cell killing was markedly enhanced by the addition of the autophagy inhibitor bifilomycin A, suggesting an opportunity for combined therapy. SAR317461 was designed to be JAK2 selective, and it would of interest to learn in the context of drug resistance, whether it affects the same targets as WP-1066, which is a JAK-STAT inhibitor that has shown promise and which may soon enter clinical trials for GBM. WP1066 has been found to variously impact multiple intracellular targets, including the inhibition of phosphorylation of STAT3 and STAT5 and AKT, and activation of immune effector cells [24, 32]. SAR317461 which was intended to be JAK2 selective, did not inhibit STAT5 and Akt phosphorylation, suggesting a differences in activity profiles between WP1066 and SAR317461 even though both induce apoptosis [18].

SAR317461 may be useful for disseminated GBM and may have broader relevance beyond GBM for several reasons. STAT3 impacts stem cell maintenance in general as the interaction between STAT3 and other pathways including EGFR, Notch, Wnt, Hedgehog, Akt, mTOR, olig2, PKC, MAPK, NF-κB and BMP4 has been shown to regulate self-renewal [33]. Hence an advantage of targeting STAT3 in GBM is that the inhibition of this pathway affects multiple downstream molecules in the GSC tumor compartment which drives GBM and is invasive [34]. In terms of other central nervous system (CNS) diseases, it is perhaps important that STAT3 plays a role in astrocyte differentiation and activity, and moreover astrocytes are immunocompetent cells and are capable of secreting inflammatory mediators that may be involved in CNS pathology such as Alzheimer’s disease, multiple sclerosis, Parkinson’s disease and brain injury [35, 36]. So, SAR317461 could potentially be therapeutically active in CNS diseases involving dysfunction of astrocytes or astrocyte precursors.

For this pilot study a limited number of GBM lines were used although all possible steps were taken to ensure that the data were internally consistent. Cell based studies utilized the same experimental methodology and analysis approach. The overall mean average deviation in percent between replicates used to construct the IC₅₀ curves for 10 separate cell lines was calculated to be less than 10 % (8.49 %). For Western blot analysis a loading control was used to ensure that the intensity of the visualized bands was real and not an artifact of different loaded concentrations of the protein of interest. For studies in which the effect of combining autophagy inhibitors with SAR317461 was evaluated, we utilized both starvation controls and control solutions (PBS).

Our data demonstrate that SAR317461 potently inhibited both STAT3 phosphorylation and induced apoptosis in multiple primary patient derived GSC based GBM tumorspheres which are genetically and phenotypically heterogeneous both internally and with respect to each other. SAR317461 also exhibited significant activity against three immortalized serum-dependent human GBM cell lines. One patient derived GBM line that did not express activated STAT3 showed a comparatively much reduced response to SAR317461, suggesting that pSTAT3 may need to be present for potent SAR317461 associated anti-GBM activity, a finding that could have relevance from a personalized treatment perspective. Future explorations with a wider range of GBM lines variously expressing activated STAT3 would bypass the size limitation of the present study and may more precisely define the possible dependence of SAR317461 GBM response on the presence of activated STAT3. Moreover, patient derived tumorsphere lines and other patient sourced cancer cell types resistant to standard therapy should be explored with SAR317461 both singly and with other agents and pathway inhibitors. Follow-on mechanistic studies in all lines may identify a possible role of other Bcl-2 family proteins in the context of SAR317461 induced apoptosis. Finally, orthotopic GBM models may help characterize tumor and brain concentrations of SAR317461 and its effects alone and in combination on GBM tumor proliferation, invasion, and angiogenesis. On the basis of the pilot results reported herein for SAR317461, we believe that the JAK-STAT pathway does merit further and comprehensive testing as a potential therapeutic target for GBM.