*Article* **A Novel Inhibitor of STAT5 Signaling Overcomes Chemotherapy Resistance in Myeloid Leukemia Cells**

**Marie Brachet-Botineau 1, Margaux Deynoux 1, Nicolas Vallet 1,2, Marion Polomski 3, Ludovic Juen 3, Olivier Hérault 1,4, Frédéric Mazurier 1, Marie-Claude Viaud-Massuard 3, Gildas Prié <sup>3</sup> and Fabrice Gouilleux 1,\***


Received: 8 August 2019; Accepted: 14 December 2019; Published: 17 December 2019

**Abstract:** Signal transducers and activators of transcription 5A and 5B (STAT5A and STAT5B) are crucial downstream effectors of tyrosine kinase oncogenes (TKO) such as BCR-ABL in chronic myeloid leukemia (CML) and FLT3-ITD in acute myeloid leukemia (AML). Both proteins have been shown to promote the resistance of CML cells to tyrosine kinase inhibitors (TKI) such as imatinib mesylate (IM). We recently synthesized and discovered a new inhibitor (17f) with promising antileukemic activity. 17f selectively inhibits STAT5 signaling in CML and AML cells by interfering with the phosphorylation and transcriptional activity of these proteins. In this study, the effects of 17f were evaluated on CML and AML cell lines that respectively acquired resistance to IM and cytarabine (Ara-C), a conventional therapeutic agent used in AML treatment. We showed that 17f strongly inhibits the growth and survival of resistant CML and AML cells when associated with IM or Ara-C. We also obtained evidence that 17f inhibits STAT5B but not STAT5A protein expression in resistant CML and AML cells. Furthermore, we demonstrated that 17f also targets oncogenic STAT5B N642H mutant in transformed hematopoietic cells.

**Keywords:** pharmacological inhibitor; STAT5 signaling; chemotherapy resistance; myeloid leukemia

#### **1. Introduction**

STAT5A and STAT5B are two closely related signal transducers and activators of transcription family members. Both proteins are crucial downstream effectors of tyrosine kinase oncogenes (TKO) such as Fms-like receptor tyrosine kinase 3 with internal tandem duplications (Flt3-ITD), BCR-ABL and JAK2V617F which cause AML, CML and other myeloproliferative diseases (MPD), respectively [1]. STAT5 proteins are recognized as major drivers in the development and/or maintenance of CML as well as in the proliferation and survival of AML cells [2–4]. The development of tyrosine kinase inhibitors (TKI) targeting BCR-ABL such as imatinib mesylate (IM) has revolutionized the treatment of CML. Despite this success story, IM is not totally curative and approximately 50% of patients remain therapy-free after IM discontinuation. The inability of IM to completely eradicate leukemic stem cells (LSC) is probably responsible for the relapse of CML patients [5]. Moreover, the occurrence of *BCR-ABL* mutations in progressive or relapsed disease promotes IM resistance of CML cells [6]. Therefore, there is a need for complementary therapeutic strategies to cure CML. STAT5 fulfils all

the criteria of a major drug target in CML [7]. High STAT5 expression levels have been shown not only to enhance IM resistance in CML cells but also to trigger *BCR*-*ABL* mutations by inducing the production of reactive oxygen species (ROS) responsible for DNA damage [8,9]. Moreover, STAT5 was shown to play a key role in the maintenance of chemoresistant CML stem cells [10]. Thus, targeting STAT5 would also benefit relapsed CML patients who became resistant to TKI. Several approaches have been used to target STAT5 in leukemia. Among them, cell-based screening with small molecule libraries of already approved drugs allowed the identification of the psychotropic drug pimozide as a potential STAT5 inhibitor in CML cells [11]. Pimozide decreased the tyrosine phosphorylation of STAT5 and induced growth arrest and apoptosis in CML cells. In addition, pimozide was shown to target the deubiquitinating (DUB) enzyme, USP1, in leukemic cells indicating that the effects of pimozide on STAT5 activity might be indirect [12]. Indirubin derivatives were also reported to inhibit STAT5 phosphorylation in CML cells but the mechanism of inhibition is most likely suppression of upstream tyrosine kinases [13]. More recently, a number of small inhibitors that bind to the Src homology domain 2 (SH2) required for STAT5 activation and dimer formation, have been described [14]. These compounds exhibit potent and selective binding activity for STAT5 by effectively disrupting phosphopeptide interactions. Some of these inhibitors bind STAT5 proteins in a nanomolar range and inhibit the tyrosine phosphorylation of STAT5 and CML/AML cell growth in a micromolar range [15–17]. A final approach is to target STAT5 activity through the activation of peroxisome proliferator-activated receptor gamma (PPARγ) [18]. Indeed, the existence of cross-talk between PPARγ and STAT5 has been discussed. For instance, antidiabetic drugs such as glitazones, which are PPARγ agonists, were shown to have antileukemic activity [19,20]. Activation of PPARγ by pioglitazone not only decreases the phosphorylation of STAT5 in CML cells but also reduces expression of *STAT5* genes in quiescent and resistant CML stem cells [10]. Importantly, the combined use of pioglitazone and IM triggers apoptosis of these leukemic cells suggesting that besides phosphorylation, inhibition of STAT5 expression is of prime importance for resistant CML stem cell eradication. Based on these different data, we sought to identify new STAT5 inhibitors in a library of PPARα/γ ligands that were synthetized in our laboratory [21,22]. The synthesis of derivatives of a "hit" compound identified in the library screening allowed the discovery of a new inhibitor of STAT5 signaling in CML and AML cells [23]. This molecule (17f) selectively inhibits the phosphorylation and transcriptional activity of STAT5 and induces apoptosis of CML and AML cells. Herein, we showed that 17f associated with IM or Ara-C resensitizes CML and AML cells, respectively, that acquired resistance to these drugs. We demonstrated that 17f treatment reduces STAT5B protein levels in resistant CML and AML cells, suggesting that 17f overcomes chemotherapy resistance though the downregulation of this protein. We also found that 17f suppresses expression of oncogenic STAT5N642H mutant in transformed Ba/F3 cells.

#### **2. Results**

#### *2.1. E*ff*ects of 17f Compound on Growth and Viability of IM-Sensitive and IM-Resistant BCR-ABL*<sup>+</sup> *Cells*

Initial experiments were carried out to determine the effects of 17f alone (see structure in Figure S1) on K562 cells that are sensitive (K562S) or resistant (K562R) to IM treatment. These in vitro models are depicted in Figure 1A. Sensitive and resistant cells were treated with various concentrations of 17f (ranging from 1 to 10 μM). Growth and viability were determined by trypan blue exclusion (Figure 1B) and MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (Figure 1C) assays. Addition of 17f clearly blocked the growth of K562S cells while K562R cells remain insensitive to 17f treatment at the same concentration. The EC50 value was found to be two times higher in K562R cells than in K562S cells (14.5 ± 4.8 μM vs. 6.9 ± 1.7 μM). We also observed that treatment with 5 μM 17f did not affect the growth and viability of K562R cells and used this suboptimal concentration in most experiments to evaluate the combined effects of 17f and IM.

**Figure 1.** Effects of 17f molecule on K562S and K562R cell growth (**A**) Imatinib mesylate (IM)-sensitive K562 (K562S) and IM-resistant K562 cells (K562R) were treated with 1 μM IM or DMSO as control (Co) for 48 h. Cell viability was determined by MTT assays (data are presented as mean ± SD of three independent experiments (*n* = 3) in triplicates, \*\*\* *p* < 0.001; one sample *t*-test). (**B**) K562S and K562R cells were treated with 17f or DMSO as control (Co) for the indicated times. Growth kinetics were determined by trypan blue dye exclusion assays (*n* = 3 in triplicates, data are mean ± SD). (**C**) Cell viability was measured by MTT assays after treatment of K562S or K562R cells with increasing concentrations of 17f or DMSO as control (Co) during 48 h (*n* = 3 in triplicates, data are mean ± SD, \*\* *p* < 0.01, \*\*\* *p* < 0.001; one sample *t*-test).

#### *2.2. 17f Induces Apoptosis and Cell Cycle Arrest in K562R Cells and Relieves the Resistance to IM*

We then addressed whether 17f in combination with IM might directly abrogate the resistance of K562R cells to IM. K562S and K562R cells were treated with 17f in the presence of IM and cell growth and viability were determined by MTT assays (Figure 2A). As expected, IM strongly inhibited the growth of K562S cells. This inhibition was further enhanced by 17f in a dose-dependent fashion. Interestingly, we found that the addition of 1 μM 17f in the presence of IM was already enough to significantly reduce the growth and viability of K562R cells. Treatment with 5 μM and 10 μM of 17f further increased this inhibitory effect. To analyze the growth-suppressive properties of 17f in K562R cells, we determined the impact of this small molecule on apoptosis and the cell cycle. 17f induced apoptosis and changes in cell cycle phase distribution in a concentration-dependent manner (Figure 2B,C). 17f significantly increased the number of cells in the G0 phase indicating that treatment with this compound induced quiescence of K562R cells.

**Figure 2.** 17f overcomes the resistance of K562R cells to IM treatment. (**A**) K562S and K562R cells were treated with IM or not (Co) with or without 17f for 48 h. Cell viability was determined by MTT assays. (**B**) K562R cells cultured for 48 h with IM and 17f or IM vs. DMSO as control. Cells were stained with anti-annexin V coupled with FITC (fluorescein isothiocyanate) and with 7-amino-actinomycin D (7-AAD) to determine the percentages of apoptotic cells. One representative experiment is shown (left panel). (**C**) K562R cells treated for 48 h with IM and 17f or IM and DMSO as control were stained with 7-AAD and an Alexa Fluor 488-conjugated anti-Ki-67 antibody. Cell cycle phase distributions were then estimated by flow cytometry. The histogram presents the percentage of cells in the G0 phase. One representative experiment is shown (left panel) (*n* = 3 in triplicates, data are mean ± SD, \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001).

#### *2.3. 17f Inhibits STAT5-Dependent Transcriptional Activity in K562R Cells*

We previously showed that 17f inhibits the transcriptional activity of STAT5 in CML cells. We then asked whether this small molecule also affects the activity of these proteins in IM-resistant K562R cells. We first determined the impact of this compound on the transcriptional activation of a reporter gene driven by a STAT5-specific promoter. K562S and K562R cells were transfected with a construct containing six tandem copies of the STAT5 response element in front of the minimal TK promoter fused to the luciferase reporter gene (6×(STAT5)-TK-luc). As control, cells were also transfected with a TK-luciferase vector without STAT5 response elements (TK-luc). Luciferase activity was determined 48 h post-transfection in K562S and K562R cells treated with DMSO as control, 17f (5 μM) and/or IM (1 μM). As expected, constitutive STAT5 activity induced by BCR-ABL increased luciferase activity in K562S cells transfected with the STAT5-dependent promoter construct compared to cells transfected with the control TK-luc vector (Figure 3A). This enhanced luciferase activity was strongly reduced after 17f or IM treatment. In sharp contrast, the luciferase activity remained elevated after treatment with IM in K562R cells transfected with the STAT5-dependent reporter construct, although this enzymatic activity was strongly decreased after the addition of 17f and IM. qRT-PCR experiments were then conducted to determine the effects of 17f on STAT5-dependent expression of target genes such as *PIM1* and *CISH* (Figure 3B). As expected, 17f or IM reduced expression of both genes in sensitive K562S cells while this effect was observed in resistant K562R cells after treatment with both compounds. Collectively, these data strongly suggest that 17f inhibits the transcriptional activity of STAT5 to bypass IM resistance in K562R cells.

#### *2.4. 17f Inhibits STAT5B Protein Expression in IM-Resistant K562 Cells*

We then determined the impact of 17f on BCR-ABL-induced tyrosine phosphorylation of STAT5 (P-Y694/699-STAT5) by western blot and flow cytometry analysis (Figure 4A, Figures S2–S4). K562S and K562R cells were treated for 24 h instead of 48 h to analyze the early effects on STAT5 phosphorylation. IM strongly reduced P-Y-STAT5 levels in K562S, and the addition of 17f further enhanced this effect. P-Y-STAT5 levels were maintained in IM-treated K562R cells but were decreased after the addition of 17f. Interestingly, the level of STAT5 phosphorylation was strikingly enhanced in K562R cells after removal of IM and was weakly affected by the addition of 17f (Figure S5). To determine whether changes in P-Y-STAT5 levels reflect differences in protein abundance, immunoblots were performed with an anti-STAT5 antibody that recognizes STAT5A and STAT5B proteins (Figure 4B and Figure S2). As expected, IM inhibited the phosphorylation of STAT5 in sensitive and resistant cells. Interestingly, we observed that the association of 17f with IM reduces STAT5 expression in K562R cells but not in K562S cells (see Figure S3A,B for quantification). qRT-PCR experiments were then conducted to evaluate the impact of combination treatments on *STAT5A* and *STAT5B* gene expression in K562R cells. Results showed that *STAT5A*/*5B* mRNA levels were not affected by 17f when associated with IM (Figure 4B). In contrast, western blot analysis clearly evidenced that STAT5B protein expression was decreased after combination treatments suggesting that 17f sensitizes K562R cells to IM treatment by targeting STAT5B protein (Figure 4C).

**Figure 3.** 17f associated with IM inhibit STAT5 activity in resistant K562R cells. (**A**) K562S or K562R cells transfected with a 6×(STAT5)-TK-luciferase reporter construct or a control TK-luciferase vector were treated or not (Co) with 17f (5 μM), IM (1 μM) or with the combination of 17f and IM for 48 h. Luciferase activities were then determined as described in Methods. Luciferase activity (arbitrary units) in the histogram represents the relative luminescence unit (rlu) values/mg of proteins (*n* = 3 in triplicates, data are mean ± SD, \* *p* < 0.05, \*\* *p* < 0.001). (**B**) qRT-PCR analysis of *PIM1* and *CISH* expression in K562S and K562R treated or not (Co) with IM (1 μM),17f (5 μM) or with combined 17f and IM for 24 h. Results are presented as the fold change in *PIM1* and *CISH* gene expression in treated cells normalized to internal control genes (*GAPDH*, *ACTB* and *RPL13a*) and relative to control condition (normalized to 1) (*n* = 3 in triplicates, data are mean ± SD, \* *p* < 0.05; one-sample *t*-test).

**Figure 4.** 17f associated with IM inhibits STAT5B protein expression in K562R cells (**A**) Protein extracts from K562S and K562R cells treated with 17f 5 μM or DMSO with or without IM for 24 h were analyzed by western blotting to detect P-Y694/699-STAT5 and STAT5 protein expression (*n* = 2). Actin served as the loading control. (**B**) qRT-PCR analysis of *STAT5A* and *STAT5B* expression in K562R cultured with IM (1 μM) as control or treated with 17f (5μM) and IM for 24 h. Results are presented as the fold change in *STAT5A* and *STAT5B* gene expression in treated cells normalized to internal control genes (*GAPDH*, *ACTB* and *RPL13a*) and relative to control condition (normalized to 1) (*n* = 3 in triplicates, data are mean ± SD, \* *p* < 0.05; one sample *t*-test). (**C**) Expression of STAT5A and STAT5B proteins in K562R cells treated or not with 17f (5 μM) was analyzed by western blot (*n* = 2). Actin served as the loading control.

#### *2.5. E*ff*ects of 17f on Growth and Viability of Ara-C-Sensitive and Ara-C-Resistant FLT3-ITD Expressing Leukemic Cells*

STAT5 is also phosphorylated by FLT3-ITD, a major TKO in AML cells. To exclude the possibility that 17f-mediated inhibition of STAT5 and cell growth is a peculiarity of IM-resistant BCR-ABL<sup>+</sup> cells, we used MV4-11 cells expressing FLT3-ITD that acquired resistance to Ara-C, a conventional therapeutic agent that affects DNA replication. Sensitive and resistant MV4-11 cell models are depicted in Figure 5A. We first evaluated the impact of 17f alone on MV4-11S and MV4-11R cell growth and showed that MV4-11R cells were more resistant to 17f treatment than MV4-11S cells (Figure 5B). Based on these data, IC50 values were found to be three-fold higher in MV4-11R than in MV11-4S cells (10.79 ± 3.2 vs. 3.55 ± 0.47).

#### *2.6. 17f Sensitizes MV4-11R Cells to Ara-C Treatment*

We then analyzed the effects of 17f on MV4-11S and MV4-11R cell growth in the presence of Ara-C using trypan blue dye exclusion (Figure 6A) and MTT assays (Figure 6B). Addition of 17f significantly enhanced the growth inhibition and cytotoxic effect of Ara-C in MV4-11S cells. Importantly, 17f greatly reduced the growth of resistant MV4-11R cells cultured with Ara-C in a concentration-dependent fashion. This growth inhibition was already observed with 1 μM, a concentration that did not affect the growth of MV4-11R cells cultured in the absence of Ara-C. These data indicated that the addition of 17f overcomes the resistance of MV4-11R cells to Ara-C.

**Figure 5.** Effects of 17f on MV4-11S and MV4-11R cell growth (**A**) Ara-C-sensitive MV4-11 (MV4-11S) and Ara-C-resistant MV4-11 (MV4-11R) cells were treated with 1 μM Ara-C or DMSO as control (Co) for 48 h. Cell viability was then determined by MTT assays (*n* = 3 in triplicates, data are mean ± SD, \*\*\*\* *p* < 0.0001; one-sample *t*-test). (**B**) MV4-11S and MV4-11R cells were treated or not (Co) with increasing concentrations of 17f during 48 h. Cell viability was determined by MTT assays (*n* = 3 in triplicates, data are mean ± SD, \*\* *p* < 0.01, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001; one-sample *t*-test).

**Figure 6.** 17f relieves the resistance of MV4-11R cells to ARA-C treatment. (**A**) MV4-11S or MV4-11R cells were treated with Ara-C or not (Co) with or without 17f. Growth kinetics were determined by Trypan blue dye exclusion assays (*n* = 3 in triplicates, data are mean ± SD). (**B**) MV4-11S or MV4-11R cells were treated with Ara-C or not (Co) with or without 17f for 48 h. Cell viability was determined by MTT assays (n = 3 in triplicates, data are mean ± SD, \* *p* < 0.05, \*\* *p* < 0.01, \*\*\*\* *p* < 0.0001; one-sample *t*-test).

#### *2.7. 17f Triggers Apoptosis, Cell Cycle Arrest and Inhibition of STAT5B Expression in MV4-11R Cells*

We then evaluated the effects of 17f on apoptosis and the cell cycle in MV4-11R cells. A significant increase in apoptotic cells was observed (Figure 7A) only after treatment with 5 μM 17f, while the addition of 1 μM was enough to enhance the number of cells in the G0 phase of the cell cycle (Figure 7B). These results indicated that the growth-suppressive properties of 17f primarily affect the cell cycle in MV4-11R cells and apoptosis at higher concentrations. We then asked whether 17f interferes with STAT5 signaling in Ara-C-resistant AML cells and analyzed the impact of 17f on phosphorylation and expression of STAT5 in MV4-11R cells. In the absence of Ara-C, the level of STAT5 phosphorylation was slightly enhanced in MV4-11R cells (Figure S5C,D). The addition of 17f with or without Ara-C inhibited STAT5 expression in MV4-11R cells (Figure 7C and Figures S2B and S3C). Likewise, STAT5B expression was reduced after treatment with 17f alone or with Ara-C in resistant cells (Figure 7D).

**Figure 7.** 17f promotes apoptosis, cell cycle arrest and inhibition of STAT5B protein expression in MV4-11R cells. (**A**) Flow cytometry histogram of MV4-11R cells cultured for 48 h with Ara-C and 17f or Ara-C and DMSO as control. Cells were stained with anti-annexin V coupled with FITC and with 7-AAD to determine the percentages of apoptotic cells (*n* = 3 in triplicates, data are mean ± SD, \*\* *p* < 0.01). (**B**) MV4-11R cells treated for 48 h with Ara-C and 17f or Ara-C and DMSO as control were stained with 7-AAD and an Alexa Fluor 488-conjugated anti-Ki67 antibody. Cell cycle phase distributions were then estimated by flow cytometry. The histogram presents the percentage of cells in the G0 phase (*n* = 3 in triplicates, data are mean ± SD, \* *p* < 0.05, \*\*\* *p* < 0.001). (**C**) Protein extracts from MV4-11R cells treated with Ara-C and 17f or Ara-C and DMSO for 24 h were analyzed by immunoblotting to detect P-Y694/699-STAT5 and STAT5 protein expression (*n* = 2). Actin served as the loading control. (**D**) Expression of STAT5A and STAT5B proteins in MV4-11R cells treated or not with 17f (5 μM) was also analyzed by western blot (*n* = 2).

#### *2.8. 17f Inhibits Expression of Oncogenic STAT5BN642H Mutant*

Gain of function mutations of *STAT5B* have been described in hematopoietic malignancies. The recurrent hotspot mutation N642H has been identified in T cell leukemia and lymphomas and the STAT5BN642H mutant was shown to induce T cell neoplasia in transgenic mice [24–27]. We therefore tested the ability of 17f to inhibit STAT5BN642H expression and growth of hematopoietic cells transformed by this mutant. For this purpose, we used Ba/F3 cells expressing flag-tagged STAT5BN642H or flag-tagged wild-type STAT5B (wtSTAT5B) as control [27]. We found that Ba/F3-STAT5BN642H cells were more sensitive to 17f treatment than control Ba/F3-wtSTAT5B cells (Figure 8A). We then addressed whether STAT5N642H expression was impacted by 17f and showed that 17f reduces expression of this mutant in Ba/F3 cells but does not affect wtSTAT5B or endogenous STAT5A expression after 24 h treatment (Figure 8B).

**Figure 8.** 17f inhibits STAT5BN642H activity and expression in Ba/F3 cells. (**A**) Cells were treated or not with 17f (10 μM). Growth were then determined by Trypan blue dye exclusion assays at the indicated times (*n* = 5 in triplicates, data are mean ± SD). (**B**) Protein extracts from MV4-11R cells treated with 17f for 24 h were analyzed by immunoblotting to detect flag-tagged wtSTAT5B, STAT5BN642H and endogenous STAT5A/STAT5B protein expression (*n* = 2). Actin served as the loading control.

#### **3. Discussion**

The development of pharmacological inhibitors targeting the JAK/STAT pathway has been the subject of intense investigation during the last decade. Among the STAT family members, STAT5 proteins are now recognized as important therapeutic targets in hematologic malignancies and also in certain solid tumors [28]. Distinct pharmacological compounds that directly or indirectly affect STAT5 activity and leukemia cell growth have been used or developed during these last years. We recently synthesized and discovered a new compound (17f) that inhibits STAT5 phosphorylation and transcriptional activity in various CML and AML cells, without detectable effects on other signal

transduction molecules, such' as STAT3 and the protein kinases ERK1/2 and AKT [23]. We also demonstrated that 17f strongly reduces the growth of CML and AML cells with EC50 values below 10 μM close to EC50 values obtained with the STAT5 inhibitor pimozide (unpublished data) indicating that 17f as pimozide targets myeloid leukemia cells addicted to STAT5 signaling (see also Figure S1 for 17f and pimozide structures). In this study, we bring evidences that 17f also relieves the resistance of CML and AML cells to IM and Ara-C, respectively. Interestingly, we found that the concentrations of 17f required to restore the response to IM and Ara-C in resistant leukemic cells were much lower than EC50 values obtained for each resistant cell type. Indeed, inhibition of cell growth was already observed with 1 μM when combined with IM or Ara-C while EC50 values obtained for 17f compound alone were greater than 10 μM in these resistant cells. Depletion of IM or Ara-C in resistant cells might explain changes in the growth inhibitory effects of 17f. Indeed, we observed that the removal of IM strongly increases the phosphorylation of STAT5 in K562R cells. In these conditions, P-Y-STAT5 protein levels remain much higher in K562R cells after 17f treatment than in treated K562S cells, which are sensitive to lower concentrations of 17f. These data are in close agreement with a previously published study showing that high STAT5 levels mediate IM resistance in CML cells [8]. Although the removal of Ara-C results in a slight increase in STAT5 phosphorylation, the resistance of MV4-11R cells to this drug is not directly linked to overactivated STAT5. ERK1/2 and AKT kinases that also play a crucial role in cell survival, are involved in the resistance of MV4-11 cells to Ara-C [29]. It is then likely that Ara-C depletion may overexpress or overactivate these survival pathways in resistant MV4-11 cells. Whatever the resistance mechanism associated or not with STAT5 signaling, our data suggest that combination treatments with a STAT5 inhibitor might efficiently eliminate resistant CML and AML cells.

While 17f alone inhibited STAT5 phosphorylation in IM-depleted K562R cells, it decreased STAT5 expression in Ara-C-depleted MV4-11R cells. Importantly, combination treatments reduced expression of STAT5 in both resistant leukemic cells. The mechanisms involved in this downregulation remain unknown but are not associated with changes in *STAT5A* and *STAT5B* gene expression and specifically affect STAT5B protein. Importantly, we also demonstrated that 17f inhibits expression of STAT5BN642H protein expression in transformed Ba/F3 cells. STAT5BN642H is a driver mutation for T cell neoplasia and has been associated with aggressiveness, poor prognosis and an increased risk of relapse in T cell leukemia-lymphoma patients [24–27]. In addition to myeloid leukemia, 17f might be then employed to target lymphoproliferative disorders and lymphomas addicted to STAT5BN642H signaling.

It is likely that 17f inhibits STAT5B expression via the ubiquitin/proteasome-dependent degradation of this protein. Indeed, STAT5 proteins were previously shown to be ubiquitinated and several ubiquitination sites have been identified in STAT5A and STAT5B protein sequences [30,31]. Cbl, a well-known E3 ubiquitin ligase was found to interact with STAT5 and to induce its ubiquitination [30]. Moreover, cytokine-mediated STAT5 phosphorylation was enhanced in hematopoietic stem cells from *c-cbl* knockout mice [32]. 17f alone or associated with IM or Ara-C might then promote ubiquitination and proteasomal degradation of STAT5B protein in resistant leukemic cells as well as in STAT5BN642H-expressing cells. In a similar vein, pimozide was shown to target USP1, a ubiquitin specific protease involved in the deubiquitination of transcription factors such as ID-1. Pimozide-mediated inhibition of USP-1 promotes ID1 degradation and inhibition of leukemic cell growth [12]. It is therefore conceivable that 17f activity is connected to *a* proteasome regulatory network that controls STAT5B protein degradation. Alternatively, the combination of 17f and IM or Ara-C might also target chaperone molecules such as the heat shock proteins HSP90 or HSP70 proteins which were previously shown to regulate expression and/or stability of STAT5 [33,34]. The dual inhibition of BCR-ABL and HSP90 was shown to abrogate the growth of IM-resistant CML cells [35]. Furthermore, a key role of STAT5 has been demonstrated in the synergistic effects of FLT3 and HSP90 inhibitors in FLT3-ITD-expressing leukemic cells [36]. Importantly, HSP90 inhibitors not only target STAT5 but also overcome the resistance of AML cells to FLT3 inhibitors [37]. HSP70 was also found to induce STAT5

expression and drug resistance in AML and CML cells and inhibition of STAT5 activity was sufficient to resensitize resistant leukemic cells to chemotherapy [34,38].

If the downregulation of STAT5A and STAT5B expression can occur via ubiquitin/proteasome-dependent protein degradation, the selective effect of 17f on STAT5B still remains unclear. Nevertheless, using a bacterial two-hybrid screening approach, we previously identified the tumor suppressor hTid1 as a specific binding partner of STAT5B [39]. hTid1 belongs to the DnaJ chaperone protein family, which contains the J domain, a highly conserved domain that binds to Hsp70. The DnaJ-Hsp70 complexes are involved in protein folding and protein degradation and hTid1 was shown to promote the ubiquitination and degradation of various cellular proteins including transcription factors [40]. We demonstrated that overexpression of hTid1 specifically suppresses STAT5B protein expression and the transforming potential of a constitutively active STAT5B variant (STAT5B1\*6) in hematopoietic cells. 17f might then target specific effectors of STAT5B protein stability/degradation, a hypothesis that has yet to be experimentally tested.

Besides these potential mechanisms, the capacity of 17f to restore the sensitivity of resistant CML or AML cells to IM or Ara-C suggests that inhibitors targeting STAT5 expression would also benefit AML or CML patients who have developed resistance to chemotherapy. Accordingly, PPARγ agonists were shown to inhibit *STAT5A* and *STAT5B* gene expression and to synergize with IM to eradicate resistant CML stem cells [10]. Our findings suggest that targeting STAT5B protein is a promising therapeutic strategy to eradicate leukemic cells that acquired resistance to chemotherapeutic agents. This is also supported by previous works showing that STAT5B but not STAT5A plays a key role in BCR-ABL-induced leukemogenesis and in the sensitivity of CML cells to TKI treatment [41,42]. Recent studies indicated that STAT5 proteins also exert important non canonical functions in normal and cancer cells. For instance, unphosphorylated STAT5 (uSTAT5: non phosphorylated on Y694/699 residues) were shown to be transcriptionally active in self-renewing hematopoietic stem cells and to promote leukemia/lymphoma cell survival [43,44]. Selective inhibitors that only block tyrosine phosphorylation and dimer formation might then be insufficient to fully abrogate STAT5 activity and resistance to chemotherapy. Herein, we showed that that inhibition of STAT5B expression elicited by 17f might unlock drug resistance in CML and AML cells. Using these promising data as a lead, we carried out a rational search for new derivatives of 17f with enhanced antileukemic activity. Modeling work was initiated to identify a pharmacophore that could help to optimize the development of 17f derivatives working in the nanomolar range. These new compounds could represent promising drugs to overcome chemotherapy resistance in leukemia or lymphomas.

#### **4. Materials and Methods**

#### *4.1. Cell Cultures and Reagents*

IM-sensitive (K562S) and IM-resistant (K562R) BCR-ABL<sup>+</sup> cells and MV4-11 cells were obtained from American Type Culture Collection (ATCC) and Deutsche Sammlung von Mikroorganismens und Zellkulturen (DSMZ), respectively, and maintained according to the supplier's recommendations. K562R and Ara-C-resistant MV4-11 (MV4-11R) cells were obtained after cultures of K562S and sensitive MV4-11 (MV4-11S) cells with increasing concentrations of IM and Ara-C (until 1 μM). Ba/F3-STAT5BN642H and Ba/F3-wtSTAT5B cells were previously described in [27]. All cell lines were cultured in RPMI 1640, with 10% fetal bovine serum, 2 mM glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin at 37 ◦C, 5% CO2. Resistant cells were cultured with 1 μM IM or Ara-C. IM was purchased from Selleckchem (Houston, TX, USA) and Ara-C from Sandoz France (Levallois-Perret, France). Ba/F3-wtSTAT5B were cultured with IL-3. The synthesis of the 17f compound was previously described in [23].

#### *4.2. Cell Proliferation Assays*

Cell viability and proliferation were studied using a MTT cell proliferation assay (Sigma-Aldrich, St Louis, MO, USA). Briefly, 2 <sup>×</sup> 104 leukemic cells were cultured in 100 <sup>μ</sup>L of RPMI medium in 96-well plates and treated with drugs for 48 h. Cells were then incubated with 10 μL of MTT working solution (5 g/L of methylthiazolyldiphenyl-tetrazolium bromide) for 4 h. Cells were lysed overnight at 37 ◦C with 100 μL of SDS 10%, HCl 0.003%. Optical density (OD) at 570 nm was then measured using a spectrophotometer CLARIOstar® (BMG Labtech, Offenburg, Germany). Living cells were also enumerated using the trypan blue dye exclusion method.

#### *4.3. Apoptosis and Cell Cycle Analysis*

Cells were washed with PBS, then stained (10<sup>6</sup> cells) in buffer containing FITC-annexin V and 7-amino-actinomycin D (7-AAD) (Beckmann Coulter, Fullerton, CA, USA) for 15 min at 4 ◦C and analyzed by flow cytometry (Becton Dickinson Accuri™ C6 flow cytometer). For cell cycle analysis [45], cells were first incubated with fixing solution (PFA 2%, Hepes 1%, saponin 0.03%) for 15 min and then in PFS permeabilization solution (PBS 1×, SVF 10%, saponin 0.03%, Hepes 1%). Cells were next stained for 30 min at room temperature with anti-Ki67-Alexa Fluor 488 monoclonal antibody or the corresponding isotype as control (Becton-Dickinson, Franklin Lakes, NJ, USA) before analysis by flow cytometry (Becton Dickinson Accuri™ C6 flow cytometer). The FlowJo® software (V10.1, BD Biosciences, Franklin Lakes, NJ, USA) was used to analyze data.

#### *4.4. Plasmids, Transfection and Luciferase Reporter Assays*

The 6×(STAT5)-TK-luc containing six tandem copies of the STAT5 binding site linked to the minimal TK-luciferase reporter gene and control TK-luc plasmids have been described elsewhere [46]. For transient transfection assays, cells were electroporated (270 V, 950 μF) with the different constructs (50 μg). Transfected cells were expanded for 24 h in medium and then treated for 48 h. Cell extracts were then prepared in luciferase buffer according to the manufacturer's protocol (One Glo luciferase assay kit, Promega, Madison, WI, USA). Luciferase activities were measured in a luminometer CLARIOstar® (BMG Labtech, Offenburg, Germany).

#### *4.5. Western Blot*

Cells were suspended in Laemmli's 2× buffer (Bio-Rad, Hercules, CA, USA), separated on SDS/PAGE and blotted onto nitrocellulose membrane. Blots were incubated with the following antibodies (Abs): P-Y694/699-STAT5, Actin (Cell Signaling Technology, Danvers, MA, USA), STAT5 (BD Transduction Laboratories, Franklin Lakes, NJ, USA), STAT5A and STAT5B (Zymed/ThermoFisher Scientific, Waltham, MA, USA). Membranes were developed with the ECL chemiluminescence detection system (GE Healthcare, Little Chalfont Buckinghamshire, UK) using specific peroxidase (HRP) conjugated to rabbit or mouse IgG antibodies (Cell Signaling Technology).

#### *4.6. P-Y694*/*699-STAT5 Flow Cytometry Analysis*

Cells were washed in PBS and incubated with a fixing solution PFA 4% for 15 min at room temperature. The first permeabilization solution PBS/Triton X-100 0.2% was then added and incubated for 30 min at 37 ◦C. After being washed with PBS/BSA 0.5%, cells were suspended with the second permeabilization solution PBS/MeOH 50% and incubated for 10 min on ice. Cells were then stained with anti P-Y694/699-STAT5 antibodies or the corresponding isotype as control (BD Biosciences, NJ, USA) for 30 min at room temperature before analysis by flow cytometry (FACS Canto II, BD Biosciences).

#### *4.7. qRT-PCR Analysis*

RNA samples were reverse-transcribed using the SuperScript®VILO cDNA synthesis kit (Invitrogen, Carlsbad, CA, USA) as recommended by the supplier. The resulting cDNAs were used for quantitative real-time PCR (qRT-PCR). PCR primers (*PIM1*: *for 5 -*TTTCGAGCATGACGAAGAGA-3 , *rev* 5 -GGGCCAAGCACCATCTAAT-3 ; *CISH*: 5 - AGCCAAGACCTTCTCCTACCTT-3 , *rev* 5 -TGGCATCTTCTGCAGGTGT-3 ; *STAT5A*: *for* 5 -TCCCTATAACATGTACCCACA-3 , *rev* 5 -ATGGTCTCATCCAGGTCGAA-3 ; *STAT5B*: *for* 5 -TGAAGGCCACCATCATCAG-3 , *rev* 5 -TGTTCAAGATCTCGCCACTG-3 ) were designed with the ProbeFinder software (Roche Applied Sciences, Basel, Switzerland) and used to amplify the RT-generated cDNAs. qRT-PCR analyses were performed on the Light Cycler 480 thermocycler II (Roche). *GAPDH* (glyceraldehyde-3-phosphate dehydrogenase), *ACTB* (actin beta) and *RPL13A* were used as reference genes for normalization of qRT-PCR experiments. Each reaction condition was performed in triplicate. Relative gene expression was analyzed using the 2−ΔΔCt method [47].

#### **5. Conclusions**

In summary, this work shows for the first time that inhibition of STAT5B expression might be a promising targeting strategy to bypass the resistance of CML and AML cells to TKI or conventional chemotherapeutic agents. Investigations to elucidate the mechanisms involved in STAT5B downregulation induced by these combination therapies might help to design new inhibitors that specifically target cancer cells addicted to oncogenic STAT5B signaling.

**Supplementary Materials:** The following supplementary figures are available online at http://www.mdpi.com/ 2072-6694/11/12/2043/s1, Figure S1: Pimozide and 17f structures, Figure S2: Original western blot, Figure S3: Quantification of Western Blot data. Figure S4: Flow cytometry analysis of P-Y-STAT5 in K562S and K562R cells. Figure S5: Effects of 17f on P-Y-STAT5/STAT5 expression in IM-depleted K562R and Ara-C-depleted MV4-11R cells.

**Author Contributions:** Conceptualization, M.B.-B., G.P. and F.G.; Formal analysis, M.B.-B. and F.G.; Funding acquisition, G.P. and F.G.; Investigation, M.B.-B., M.D., N.V., M.P. and L.J.; Methodology, M.B.-B., M.D., N.V., M.P. and L.J.; Project administration, G.P. and F.G.; Resources, G.P. and F.G.; Supervision, G.P.; Validation, M.B.-B., M.D., N.V., M.P., F.M., G.P. and F.G.; Visualization, F.G.; Writing—original draft, M.B.-B., G.P. and F.G.; Writing—review & editing, M.D., N.V., L.J., F.M., M.-C.V.-M. and O.H.

**Funding:** This study was supported by FRM (grant number: DCM20181039564), CNRS, Ligue Contre le Cancer and University of Tours. MB-B was supported by the ARC foundation and FRM.

**Acknowledgments:** We would like to thank Emmanuel Pecnard, Farah Kouzi and Elodie Coste for technical assistance; Christina Maria Wagner and Heidi A. Neubauer for providing Ba/F3STAT5BN642H and Ba/F3wtSTAT5B cells.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **STAT3 and STAT5 Activation in Solid Cancers**

#### **Sebastian Igelmann 1,2, Heidi A. Neubauer <sup>3</sup> and Gerardo Ferbeyre 1,2,\***


Received: 1 August 2019; Accepted: 18 September 2019; Published: 25 September 2019

**Abstract:** The Signal Transducer and Activator of Transcription (STAT)3 and 5 proteins are activated by many cytokine receptors to regulate specific gene expression and mitochondrial functions. Their role in cancer is largely context-dependent as they can both act as oncogenes and tumor suppressors. We review here the role of STAT3/5 activation in solid cancers and summarize their association with survival in cancer patients. The molecular mechanisms that underpin the oncogenic activity of STAT3/5 signaling include the regulation of genes that control cell cycle and cell death. However, recent advances also highlight the critical role of STAT3/5 target genes mediating inflammation and stemness. In addition, STAT3 mitochondrial functions are required for transformation. On the other hand, several tumor suppressor pathways act on or are activated by STAT3/5 signaling, including tyrosine phosphatases, the sumo ligase Protein Inhibitor of Activated STAT3 (PIAS3), the E3 ubiquitin ligase TATA Element Modulatory Factor/Androgen Receptor-Coactivator of 160 kDa (TMF/ARA160), the miRNAs miR-124 and miR-1181, the Protein of alternative reading frame 19 (p19ARF)/p53 pathway and the Suppressor of Cytokine Signaling 1 and 3 (SOCS1/3) proteins. Cancer mutations and epigenetic alterations may alter the balance between pro-oncogenic and tumor suppressor activities associated with STAT3/5 signaling, explaining their context-dependent association with tumor progression both in human cancers and animal models.

**Keywords:** solid cancers; cell cycle; apoptosis; inflammation; mitochondria; stemness; tumor suppression

#### **1. Introduction**

Activation of Signal Transducer and Activator of Transcription (STAT) proteins has been linked to many human cancers. STATs were initially discovered as latent cytosolic transcription factors that are phosphorylated by the Janus Kinase (JAK) family upon stimulation of membrane-associated cytokine and growth factor receptors. Phosphorylation triggers STAT dimerization and translocation to the nucleus to bind specific promoters and regulate transcription [1]. Here, we review the role of STAT family members STAT3 and STAT5 in solid human malignancies, as well as the mechanisms that may explain their association with either worse or better prognosis.

#### **2. STAT3 and STAT5 in Solid Cancers**

The discovery of cancer genes has been propelled by genetic analyses and more recently by next generation DNA sequencing technologies. Combined, these studies have identified 127 significantly mutated cancer genes that cover diverse signaling pathways [2]. Mutations acting as drivers in cancer are positively selected during tumor growth and constitute solid proof of the involvement of a particular gene as a driver in the disease. Mutations in STAT3 and STAT5 have been reported in

patients with solid cancers, but unlike hyperactivation of the JAK/STAT pathway, STAT3/5 mutations in cancer are relatively infrequent and occur mostly in hematological malignancies.

An overview of reported STAT3/5 mutations in solid cancers is illustrated in Figure 1, based on data collected from the Catalogue of Somatic Mutations in Cancer (COSMIC) database. Mutations in *STAT3* are more prevalent than mutations in *STAT5A* or *STAT5B* genes. Noticeably, gastrointestinal cancers have the highest rates of STAT3/5 mutations compared with other solid cancers (Figure 1). Missense mutations tend to cluster within the SH2 domain, where gain-of-function mutations were previously characterized [3,4], as well as within the DNA binding domain and to an extent the N-terminal domain (Figure 1A). Interestingly, the *STAT3* Tyrosine 640 into Phenylalanine (Y640F) hotspot gain-of-function mutation reported in various lymphoid malignancies has also been detected in patients with liver cancer (Figure 1A). Nonsense and frameshift mutations are less frequent and more disperse, likely representing loss-of-function events (Figure 1B). Notably, a hotspot frameshift mutation at position Q368 within the DNA binding domain of STAT5B has been reported in 24 patients with various types of carcinoma; this frameshift generates a stop codon shortly after the mutation and is therefore likely to be loss-of-function, although characterization of this mutation has not been performed.

As opposed to mutation rates, STAT3/5 activation is very frequent in human cancers, perhaps reflecting increased cytokine signaling or mutations in cytokine receptors or negative regulators. STAT3/5 activation can be detected using antibodies that measure total levels or activation marks in STAT3/5 proteins (e.g. tyrosine phosphorylation). A better assessment of STAT3/5 activation can be obtained by measuring downstream signaling targets (i.e., mRNA levels of STAT3 [5] and STAT5 [6] target genes). A recent metanalysis of 63 different studies concluded that STAT3 protein overexpression was significantly associated with a worse 3-year overall survival (OS) (OR = 2.06, 95% CI = 1.57 to 2.71, *p* < 0.00001) and 5-year OS (OR = 2.00, 95% CI = 1.53 to 2.63, *p* < 0.00001) in patients with solid tumors [7]. Elevated STAT3 expression was associated with poor prognosis in gastric cancer, lung cancer, gliomas, hepatic cancer, osteosarcoma, prostate cancer and pancreatic cancer. However, high STAT3 protein expression levels predicted a better prognosis for breast cancer [7]. This study mixed data of both STAT3 and phospho-STAT3 (p-STAT3) expression limiting its ability to associate pathway activation to prognosis. Here, we summarize the data linking activation of STAT3/5 to overall survival in several major human solid cancers identifying the biomarkers used in each study (Table 1). Taken together, the results clearly show that *STAT3* and *STAT5* are important cancer genes despite their relatively low mutation frequency.

STAT3 activation is clearly a factor linked to bad prognosis in patients with lung cancer, liver cancer, renal cell carcinoma (RCC) and gliomas. In other tumors, the association is not significant. In solid tumors, STAT3 activation is more frequent than STAT5 activation although no explanation for this difference was proposed. In prostate cancer, both STAT3 and STAT5 have been associated with castration-resistant disease and proposed as therapeutic targets [8,9]. In colon cancer, the association between p-STAT3 and survival varies according to the study, but a high p-STAT3/p-STAT5 ratio indicates bad prognosis [10]. Also in breast cancer, p-STAT5 levels are clearly associated with better prognosis [11]. In liver cancer, STAT5 has ambivalent functions that were recently reviewed by Moriggl and colleagues [12]. Understanding mechanistically how STAT3/5 promote transformation and tumor suppression is important for the eventual design of new treatments. Also, survival data is highly influenced by the response of patients to their treatment and may not always reflect all mechanistic links between STAT3/5 activity and tumor biology. Of note, the effect of any gene is conditioned by the genetic context of gene action. Some genes can clearly exert a tumor suppressor effect in the initial stages of carcinogenesis that is lost when cancer mutations or epigenetic changes inactivate key effectors of these tumor suppressor pathways [13]. Human studies are usually limited to late stage tumors because it is easier to collect samples at that point. Studies in model systems, including primary cells, organoids and mouse models are thus required for a full understanding of how cancer genes work specifically at early stages in tumorigenesis.

**Figure**MapSignalTranscriptionpatientssolid cancers. Individual missense mutations found in at least two patients (**A**), as well as all reported nonsense and frameshift mutations (**B**), are depicted. Numbers in each box represent the number of cases reported for each mutation. Data were mined from the Catalogue of Somatic Mutations In Cancer (COSMIC) database. ND, N-terminal domain; CCD, Coiled coil domain; DBD, DNA binding domain; LD, Linker domain; SH2, Src homology 2 domain; TAD, Transactivation domain.


**Table 1.** STAT3/5 activity and overall survival in major human solid tumors.

ER+, estrogen receptor-positive; HCC, hepatocellular carcinoma; GBM, glioblastoma multiforme; NSCLC, non-small-cell lung carcinoma; HR, hazard ratio; RCC, renal cell carcinoma.

#### **3. Mechanisms of Transformation by STAT3**/**5 Proteins in Solid Cancers**

STAT3 and STAT5 promote tumor progression by regulating the expression of cell cycle, survival and pro-inflammatory genes. In addition, they control mitochondrial functions, metabolism and stemness, as discussed below (Figure 2).

**Figure 2.** Mechanisms of tumorigenic activity of STAT3 and STAT5 signaling in solid tumors.

#### *3.1. Cell Cycle and Apoptosis*

As transcription factors, STAT3 and STAT5 regulate many genes required for cell cycle progression and cell survival. A major target of the transcriptional control of the mammalian cell cycle is cyclin D. STAT3 regulates cyclin D expression in a complex with CD44 and the acetyltransferase p300. The latter acetylates STAT3 promoting its dimerization, nuclear translocation and binding to the cyclin D promoter [25]. Other cell cycle and survival genes regulated by STAT3 include *c-MYC (myc proto-oncogene)*, B-cell lymphoma 2 (*BCL2)*, *BCL2L1*/BCL-XL (B-cell lymphoma-extra large), *MCL1* (Myeloid Cell Leukemia Sequence 1) and *BIRC5*/survivin [26]. Recent studies combined ChIPSeq with whole transcriptome profiling in ABC DLBCL (activated B cell-like diffuse large B cell lymphoma) cell lines and revealed that STAT3 activates genes in the Phosphoinositide 3-Kinase (PI3K)/AKT/Mammalian Target of Rapamycin (mTOR) pathway, the Nuclear Factor Kappa-Light-Chain Enhancer of Activated B-Cells (NF-κB) pathway and the cell cycle regulation pathway, while repressing type I interferon signaling genes [27]. STAT5 also regulates the expression of cell cycle and cell survival genes [13] including *AKT1* [28], which encodes a pro-survival kinase.

#### *3.2. Inflammation and Innate Immunity*

Although the induction of cell proliferation and cell survival genes by STAT3/5 proteins contribute to their pro-cancer activity, in basal-like breast cancers the major genes associated with STAT3 activation control inflammation and the immune response [29]. Of note, inflammation is initially an adaptive response to pathological insults such as oncogenic stimuli, and it therefore exerts a tumor suppressive function. However, dysregulated inflammation in the long term provides a substrate for tumorigenesis [30]. STAT3 alone or in cooperation with NF-κB regulates the expression of many pro-inflammatory genes [31–33]. Starved tumor cells activate NF-κB and STAT3 via endoplasmic reticulum (ER) stress and secrete cytokines that stimulate tumor survival and clonogenic capacity [34]. The coactivation of these two transcription factors amplifies pro-inflammatory gene expression driving cancer-associated inflammation [35]. Of interest, the STAT3-NF-κB complex can repress the expression of DNA Damage Inducible Transcript 3 (DDIT3), an inhibitor of CCAAT Enhancer Binding Protein Beta (CEBPβ), another pro-inflammatory transcription factor [36].

Pharmacological agents that limit inflammation have been proposed for cancer prevention [37]. The use of metformin, a drug widely used to control diabetes, has been associated with a dramatic reduction in cancer incidence in many tissues [38]. Although the primary site of action of this drug is in mitochondria, a consequence of its effects is a potent reduction in the activation of NF-κB and STAT3, suggesting that the promising anticancer actions of metformin are related to its ability to curtail pro-inflammatory gene expression [39,40]. In contrast to STAT3, STAT5B inhibits NF-κB activity in the kidney fibroblast cell line COS by competing with coactivators of transcription [41], while it stimulates NF-κB in leukemia cells [42]. These results suggest the involvement of different regulatory mechanisms of STAT5 in hematopoietic cancers compared with solid cancers.

#### *3.3. Mitochondria*

In addition to their canonical roles in inflammation and immunity, STAT3 and STAT5 have been shown to localize to mitochondria. The mitochondrial localization of STAT3 is required for its ability to support malignant transformation in murine embryonic fibroblasts and breast cancer cells [43–46], and mito-STAT3 regulates mitochondrial metabolism and mitochondrial gene expression [45,47–51]. Several reports have suggested that STAT3 can be imported to mitochondria after phosphorylation on S727 [44,45] or upon acetylation [52,53]. Other studies have revealed that STAT3 mitochondrial translocation is mediated by interactions with Heat Shock Protein 22 (HSP22), Gene Associated with Retinoic and Interferon-Induced Mortality 19 (GRIM-19) or Translocase of Outer Mitochondrial Membrane 20 (TOM20) [54–56]. The mRNAs coding for some mitochondrial proteins are translated close to or in physical interaction with the import complex TOM [57,58]. The structural motifs mediating those interactions are located in the 3 and 5 UTRs of the mRNAs [59,60] and it will be interesting to investigate whether the mRNA of STAT3 also possesses RNA localization signals (zip codes) to localize in close proximity to mitochondria.

Whereas the role of mitochondrial STAT3 has been extensively studied, the role of STAT5 in mitochondria is less clear. The import of STAT5 to mitochondria is regulated by cytokines [43]. Once imported into the mitochondria, STAT5 binds the D-loop of mitochondrial DNA, although no increase in transcription of mitochondrial genes was observed [61]. Mito-STAT5 is also able to interact with the Pyruvate Dehydrogenase Complex (PDC) and was shown to regulate metabolism towards glycolysis, as observed in cells treated with cytokines [43,61]. In the same line, STAT3 was also shown to interact with the PDC in mitochondria [53].

#### *3.4. Reprogramming and Stemness*

The role of STAT3 in stem cell biology was initially recognized due to the requirement for the cytokine LIF to maintain pluripotency in cultures of mouse embryonic stem (ES) cells. STAT3 activation mediates the induction or repression of several genes in mouse ES cells including the pluripotency factors *Oct4*, *Klf4*, *Tfcp2l1* and polycomb proteins [62–64]. Many pluripotency factors, such as Homeobox Protein NANOG, are short-lived proteins. STAT3 controls protein stability by inducing the expression of the deubiquitinase Ubiquitin Specific Peptidase 21 (USP21), stabilizing NANOG in mouse ES cells. Induction of ES cell differentiation promotes the Extracellular Signal-Regulated Kinase (ERK)-dependent phosphorylation of USP21 and its dissociation from NANOG, leading to NANOG degradation [65]. STAT3 also plays a role in the reprogramming of somatic cells into induced pluripotent stem (IPS) cells [66] and it has been suggested that its effects depend on the demethylation of pluripotency factor promoters [67]. STAT3 also activates mitochondrial DNA transcription, promoting oxidative phosphorylation during maintenance and induction of pluripotency [68]. It is thus likely that the ability of STAT3 to stimulate stemness also plays a role in its oncogenic activity.

In many tumors, a subpopulation of cells possess a higher malignant capacity. These so-called tumor-initiating cells are suspected to regenerate the tumor after cancer chemotherapy and express many genes commonly expressed in ES cells [69]. It has been shown that STAT3 is required for the formation of tumor spheres and the viability of the cancer stem cell pool in many different tumors [39,40,70–83]. At least in breast cancer, a critical mechanism stimulated by STAT3 to regulate stemness involves genes in fatty acid oxidation [78,79] and the ability of STAT3 to adjust the levels of reactive oxygen species (ROS) produced in mitochondria [79]. In colorectal cancer cells, STAT3 forms a complex with the stem cell marker CD44 and the p300 acetyltransferase. Acetylation of STAT3 by this complex allows dimerization, nuclear translocation and binding to the promoters of genes required for stemness such as *c-MYC* and *TWIST1* [84].

The role of STAT5 in promoting cancer stemness does not affect many cell types and is mostly confined to hematopoietic cancers [85]. However, Nevalainen and colleagues reported that STAT5B induces stem cell properties in prostate cancer cells [86] in line with the increase in nuclear STAT5A/B observed in these tumors in correlation with bad prognosis [9]. Furthermore, transgenic mice with increased expression of prolactin in prostate epithelial cells displayed increases in the basal/stem cell compartment in association with activation of STAT5. This enrichment of stem cells was partially reversed by depletion of *Stat5a*/*b* [87]. The pro-stem cell oncogenic effect of STAT5 in the prostate contrasts with its effects in the mammary gland where STAT5 induces cell differentiation [88]. The ETS transcription factor Elf5 (E74-like factor 5) is a target of the prolactin-STAT5 axis and promotes mammary cell differentiation [89–91], supporting the tumor suppressive role of STAT5 in the mammary gland.

#### **4. Tumor Suppressor Functions and Negative Regulation of STAT3**/**5 Signaling**

The oncogenic activity of JAK/STAT signaling is controlled by several molecular barriers that limit the activation of this pathway. They include tyrosine phosphatases, E3 SUMO ligases of the Protein Inhibitor of Activated STAT3 (PIAS) family, E3 ubiquitin ligases and miRNAs. In addition, oncogenic STAT3/5 signaling can activate fail-safe tumor suppressors such as protein of alternative reading frame 19 (p19ARF), Suppressor of Cytokine Signaling 1 (SOCS1) and p53 that trigger apoptosis, ferroptosis and/or senescence in potentially malignant cells (Figure 3). Understanding these different responses to STAT signaling in cancer is important to further distinguish tumors that would benefit from STAT3 or STAT5 inhibitors and those that would not.

**Figure 3.** Tumor suppressor pathways acting on STAT3/5 activity (Protein Inhibitor of Activated STAT3 PIAS, miRNAs, E3 ligases, phosphatases) or activated by STAT3/5 transcriptional activity (Protein of alternative reading frame 19 (p19ARF) Suppressor of Cytokine Signaling 1 and 3 (SOCS1/3), p53). Abbreviations: (PTPN2 (Tyrosine-protein phosphatase non-receptor type 2), PTPN9/MEG2 (Tyrosine-protein phosphatase non-receptor type 9), PTPN11/SHP2 (Tyrosine-protein phosphatase non-receptor type 11), PTPN6/SHP1 (Tyrosine-protein phosphatase non-receptor type 6) and TNF receptor associated factor 6 (TRAF6)).

#### *4.1. Tyrosine Phosphatases*

Activation of STAT3 and STAT5 in tumors is often associated with tyrosine phosphorylation, a modification that can be reverted by several protein tyrosine phosphatases such as PTPN2 (Tyrosine-protein phosphatase non-receptor type 2), PTPN9/MEG2 (Tyrosine-protein phosphatase non-receptor type 9), PTPN11/SHP2 (Tyrosine-protein phosphatase non-receptor type 11) [92,93], CD45 [94] and PTPN6/SHP1 (Tyrosine-protein phosphatase non-receptor type 6) [95]. However, little is known about a possible role of these phosphatases in STAT3 activation in solid tumors. In liver cancers, SHP1 is downregulated in cells with mesenchymal features, and restoring its levels both reduced STAT3 phosphorylation and reversed the mesenchymal phenotype of liver cancer cells [95]. SHP1 and SHP2 also target STAT5 [96,97] but the significance of this regulation in solid tumors remains to be investigated.

#### *4.2. PIAS*

The Protein Inhibitor of Activated STAT3 (PIAS3) inhibits STAT3 transcriptional activity. In gliomas, PIAS3 expression is reduced [98]. Mechanistically, SMAD6 promotes PIAS3 degradation, promoting glioma cell growth and stem cell properties [76]. The PIAS proteins have SUMO E3 ligase activity acting on multiple proteins, and so their effects cannot be solely attributed to STAT3 inhibition [99]. Of interest, PIAS3 can bind NF-κB promoting its SUMOylation and inhibiting its activity [100,101], potentially targeting the expression of many pro-inflammatory genes required for tumor progression. Also, PIAS3 binds the N-terminus of p53 and prevents the interaction with its negative regulator MDM2, leading to p53 stabilization [102].

#### *4.3. E3 Ligases*

The Golgi resident and BC-box protein TATA Element Modulatory Factor/Androgen Receptor-Coactivator of 160 kDa (TMF/ARA160) was reported as an E3 ligase that catalyzes STAT3 ubiquitination leading to its proteasome-dependent degradation in myogenic C2C12 cells. The level of TMF/ARA160 was found to be significantly decreased in glioblastoma multiforme tumors, in benign meningioma and in malignant anaplastic meningioma, where STAT3 is known to play an oncogenic role [103]. TMF/ARA160 can also bind and ubiquitinate RELA/NF-κB leading to its proteasome-dependent degradation and a decrease in the expression of inflammatory genes [104]. Furthermore, the ubiquitin ligase TNF receptor associated factor 6 (TRAF6) binds and ubiquitinates STAT3 inhibiting the expression of STAT3 target genes [105]. During oncogene-induced senescence, STAT3 is degraded by the proteasome but the E3 ligase responsible has not been identified [106]. Recent results revealed that the long non-coding RNA (lncRNA) PVT1 (long non-coding RNA encoded by the human *PVT1* gene) binds STAT3 and protects it from ubiquitin-dependent degradation in gastric cancer [107]. PVT1 is upregulated in multiple cancers predicting poor prognosis for overall survival [108–110].

#### *4.4. MiRNAs*

The miRNA miR-124 regulates STAT3 signaling by targeting the mRNAs of interleukin-6 receptor (IL6R) [111] and STAT3 [112,113]. Suppression of this miRNA increases STAT3 phosphorylation and induces transformation in immortalized mouse hepatocytes. Of interest, systemic delivery of miR-124 prevented tumor growth in diethylnitrosamine (DEN)-treated mice, and miR-124 levels were found to be reduced in human hepatocellular carcinomas (HCC) [111]. In gliomas, miR-124 is poorly expressed but upregulation of its expression in glioma cancer stem cells inhibited the STAT3 pathway. In this model, STAT3 mediates immunosuppression, which was relieved upon systemic miR-124 delivery [114]. The circular RNA (circRNA) 100782 is upregulated in pancreatic cancer and its knockdown downregulates all miR-124 targets including IL6R and STAT3. This circRNA binds miR-124 suggesting that it may act as a miRNA sponge [115]. Furthermore, the miRNA miR-1181 also targets STAT3 and is downregulated in pancreatic cancer, predicting poorer overall survival. Overexpression of miR-1181 inhibited tumor formation and stem cell properties of pancreatic cancer cells [116].

#### *4.5. The Suppressor of Cytokine Signaling SOCS*

The members of the Suppressor of Cytokine Signaling (SOCS) family are major negative feedback regulators of JAK/STAT signaling and their expression is dysregulated in many human cancers [117–119]. These genes provide a barrier for cells with aberrant cytokine activation by inhibiting cytokine signaling [120]. In STAT3 driven cancers, SOCS3 seems to be the most important negative feedback regulator and mouse models of SOCS3 ablation show strong STAT3 activation [119,121–124]. On the other hand, in solid cancers where STAT5 plays a causal role such as liver and prostate cancer, in addition to SOCS3, SOCS1 is frequently inactivated and mouse models of SOCS1 ablation increase both liver and prostate tumorigenesis [125–132]. In addition to their role as JAK/STAT signaling barriers, SOCS1 and SOCS3 can bind p53 and activate tumor suppressor responses such as senescence and ferroptosis when their expression is induced by aberrant STAT5 signaling in primary cells [133–138]. In this way, SOCS1 and SOCS3 also act as fail-safe tumor suppressors in response to aberrant JAK/STAT signaling. So far, the SOCS1-p53-senescence axis has been demonstrated in primary fibroblasts and mammary epithelial cells [133,139–141]. This mechanism may explain the better prognosis of some solid cancers with high p-STAT5 [142–144] and the high frequency of SOCS1 inactivation in STAT5-driven cancers [125–132]. However, it is difficult to obtain evidence of a senescence tumor-suppression response by studying established tumors that have already circumvented this pathway. Senescence is particularly noticeable in premalignant lesions and benign tumors [40,106,145–150], and can be reactivated by cancer chemotherapy [151,152]. For this reason, evidence of STAT5-induced senescence

in human cancers is not yet available and should be studied in samples from premalignant tumors or after chemotherapy.

The mechanisms that disable SOCS1 and SOCS3 in human cancers are often epigenetic, mediated either by miRNAs, promoter methylation or protein phosphorylation [127,128,130,131,137,153–162]. The SRC family of kinases (SFK) phosphorylate SOCS1 at Y80, interfering with p53-SOCS1 interactions. SFK inhibitors can reverse this effect and could be used to restore the SOCS1-p53 axis in tumors where these two proteins remain intact [162]. It is also possible to consider treatments that re-express SOCS1/3 in tumors. Indeed, in liver cancer *SOCS3* gene expression can be re-established by drugs that activate the Farnesoid X receptor (FXR) [163,164]. Gene therapy strategies are also under development to re-express SOCS1 or SOCS3 in tumors [165–167].

#### *4.6. P19ARF-p53 Pathway*

One of the first reports demonstrating that STAT3 can act as a tumor suppressor was shown in glioblastoma multiforme (GBM) [168] where a combination of low Phosphatase and tensin homolog (PTEN) expression and loss of STAT3 in astrocytes increased their tumorigenicity. This observation is in contrast to papers cited above on the requirement for STAT3 to maintain tumor stem cells in GBM [73,75,169]. This could be explained if STAT3 acts early in tumorigenesis as a tumor suppressor but gains oncogenic functions in the context of the cancer genome and epigenome. An interesting mechanism for the tumor suppressor role of STAT3 was recently described in the prostate where STAT3 induces the expression of p19ARF [170]. The latter is a tumor suppressor that activates p53 and inhibits ribosome biogenesis inducing cellular senescence and apoptosis [171–174]. Loss of STAT3 disrupts this STAT3-ARF-p53 axis and permits tumor progression [175]. STAT3 and other STATs can also induce p21 leading to cell cycle arrest or cellular senescence [176,177]. Further evidence for STAT3 as a tumor suppressor has been reported in lung [178], colon [179,180], thyroid [181], liver [182,183], skin [184], neck [185], nasopharynx, rectum [186], salivary gland [187] and breast cancers [188] but the mechanisms remain to be investigated.

#### **5. Concluding Remarks**

Context-dependent activities of STAT3 and STAT5 in solid human cancers justify detailed molecular studies that will clarify the specific molecular mechanisms of action of these two cancer genes. The cancer genome and transcriptome are shaped and selected to favor cancer cell survival and proliferation. Although restoring mutated genes is technologically difficult, reprograming the transcriptome to restore tumor suppression may be feasible. Drugs acting on STAT3/5 and their regulators may restore the control of cell proliferation in cancer cells.

**Funding:** G.F. is supported by the CIBC chair for breast cancer research at the CRCHUM. This work was funded by a grant from the Canadian Institute of Health and Research (CIHR-MOP229774) to G.F. H.A.N. is supported by the Austrian Science Fund (FWF), under the frame of ERA PerMed (I 4218-B). H.A.N. is also generously supported by a private donation from Liechtenstein.

**Acknowledgments:** We thank V. Bourdeau for comments.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Balancing STAT Activity as a Therapeutic Strategy**

#### **Kelsey L. Polak 1,**†**, Noah M. Chernosky 1,**†**, Jacob M. Smigiel 1, Ilaria Tamagno <sup>1</sup> and Mark W. Jackson 1,2,\***


Received: 17 September 2019; Accepted: 31 October 2019; Published: 3 November 2019

**Abstract:** Driven by dysregulated IL-6 family member cytokine signaling in the tumor microenvironment (TME), aberrant signal transducer and activator of transcription (STAT3) and (STAT5) activation have been identified as key contributors to tumorigenesis. Following transformation, persistent STAT3 activation drives the emergence of mesenchymal/cancer-stem cell (CSC) properties, important determinants of metastatic potential and therapy failure. Moreover, STAT3 signaling within tumor-associated macrophages and neutrophils drives secretion of factors that facilitate metastasis and suppress immune cell function. Persistent STAT5 activation is responsible for cancer cell maintenance through suppression of apoptosis and tumor suppressor signaling. Furthermore, STAT5-mediated CD4+/CD25+ regulatory T cells (Tregs) have been implicated in suppression of immunosurveillance. We discuss these roles for STAT3 and STAT5, and weigh the attractiveness of different modes of targeting each cancer therapy. Moreover, we discuss how anti-tumorigenic STATs, including STAT1 and STAT2, may be leveraged to suppress the pro-tumorigenic functions of STAT3/STAT5 signaling.

**Keywords:** STAT3; STAT5; cancer progression; cancer-stem cell; cytokine; therapy resistance; metastasis; immunosuppression; tumor microenvironment; proliferation

#### **1. Introduction**

A complex milieu of both cellular and non-cellular components creates a heterogeneous tumor and microenvironment [1–8]. As a tumor becomes more heterogeneous, the risk of metastasis and therapy failure increases [9–12]. Cancer cells, pericytes, immune cells, fibroblasts, and endothelial cells are just some of the cellular components of the tumor [13,14]. A highly dysregulated network of cytokines and growth factors, emanating from cancer and stromal tumor microenvironment (TME) cells, contributes to the evolution of cancer cells, and often, suppression of immune cell function. Many of these cytokines and growth factors result in the phosphorylation and activation of signal transducers and activators of transcription (STAT) proteins, which can drive cell-specific changes in gene expression. STAT3 and STAT5 activity is often elevated in aggressive subtypes of cancer and serve as prognostic indicators [15–24]. Here, we discuss the impact of microenvironmental signals on STAT3/STAT5 activation during cancer development and progression. Elevated STAT3 activity promotes epithelial-mesenchymal transition (EMT) and a stem cell program in cancer cells, while also suppressing the function of immune cells within the tumor, all of which are important steps that underlie metastasis and therapy failure [25–30]. Likewise, persistent STAT5 activity induces (i) transformation, (ii) proliferation, and (iii) anti-apoptotic signals that contribute to hematological malignancies, while also suppressing anti-tumor immunity by expanding CD4+/CD25+ regulatory T cells (Tregs) [28,31–36]. We will discuss options for targeting STAT3/5 in cancer, either directly or

indirectly through the inhibition of upstream kinases, receptors and/or ligands. In addition, we will discuss how STAT1/2 activity can counter the more aggressive phenotypes induced by STAT3/5. We propose that balancing STAT-activated cytokine signaling in the TME may serve as an effective therapeutic strategy.

#### **2. Activating STAT3 and STAT5**

The STAT family of transcription factors is comprised of seven different members, STAT1, -2, -3, -4, -5a, -5b, and -6. Hereafter, when discussing the overlapping functions of STAT5A and STAT5B, we will refer to them as "STAT5". STAT proteins transduce signals from the cell membrane to the nucleus, bypassing the need for second messengers [37–39]. STATs are phosphorylated by Janus Kinases (JAK1, 2, 3) or Tyrosine Kinase 2 (Tyk2), which are recruited to ligand-activated receptors, including cytokine, growth factor, or g-protein associated receptors. Most important of the STAT3/5 activators are the IL-6 family members, which include IL-6, IL-11, ciliary neurotrophic factor (CNTF), leukemia inhibitory factor (LIF), oncostatin-m (OSM), cardiotrophin 1 (CT-1), cardiotrophin-like cytokine (CLC), and IL-27. IL-6 family cytokines form a heterodimeric complex consisting of gp130 and a cytokine specific subunit (IL-6Rα, IL-11Rα, CNTFRα, gp130, LIFR, and OSMR), recruiting and activating the JAKs or Tyk2. Once phosphorylated, STATs dimerize and translocate to the nucleus, where they bind to short stretches of DNA and act as transcription factors to induce the expression of genes implicated in cell proliferation, survival, differentiation, motility, apoptosis, and metabolism (Figure 1) [37,40–43].

The crucial role STATs play in these normal physiological processes was first demonstrated in STAT-deficient mouse models. The generation of tissue-specific STAT3 knockout models have identified STAT3 as a key component in a wide variety of processes, including, but not limited to: T-cell proliferation, suppression of apoptosis, epidermal regeneration during wound healing, macrophage and neutrophil anti-inflammatory responses, and mammary gland involution [44–46]. Deletion of STAT5 in mice demonstrated high incidence of perinatal lethality, prevented the appropriate development of B-cells and T-cells, and inhibited the function of hematopoietic stem cells [47]. Under normal conditions, JAK/STAT activation is tightly regulated by protein inhibitors of activated STATs (PIAS), tyrosine phosphatases, and suppressors of cytokine signaling (SOCS) proteins that inhibit JAK catalytic activity [48,49]. In cancer however, STAT3 and STAT5 activity becomes dysregulated, resulting in elevated STAT3/5-driven responses in tumor, stromal, and immune cells.

Advances in sequencing technologies have allowed scientists to investigate the frequency of mutations in the JAK/STAT signaling pathway, resulting in the identification of mutations that constitutively activate STAT3, STAT5, and JAK2. The majority of STAT mutations occur in the SH2 and C-terminal domains of STAT3 and STAT5B and associate with leukemias and lymphomas [50,51]. STAT3 is the most frequently mutated member of the STAT family, with high incidence of mutation in T-cell large granular lymphocytic leukemias and NK lymphoproliferative disorders [52,53]. STAT5B mutations were similarly identified in these diseases, but with lower frequency. In addition, sequencing of STAT3 exons in diffuse large B-cell lymphoma patients identified a missense point mutation (M206K) in the coiled-coil domain that affected SH2 domain function and drove robust proliferation [54,55]. Furthermore, JAK2 V617F, a constitutively active JAK2 mutant, phosphorylates both STAT3 and STAT5 proteins and has high frequency in patients with hematopoietic stem cell diseases, such as myeloproliferative diseases, essential thrombocythemia, polycythemia vera (PV), and idiopathic myelofibrosis (IMF) [56]. In addition, JAK2 V617F mutations have transformation potential in in vivo bone marrow transplantation assays and induce persistent activation of STAT5 and a PV phenotype [57]. While persistent activation of both STAT3 and STAT5 has been implicated in the transformation process, the greater impact of dysregulated STAT3/5 appears to be their influence on the induction of aggressive cancer cell properties and immunosuppression. The impacts of both cell-intrinsic reprogramming and immune dysfunction are discussed below.

**Figure 1.** STAT3/5 Signaling Cascades and Therapy Targets. Schematic representation of canonical activation of signal transducer and activator of transcription-3 (STAT3) and STAT5 by IL-6 family member cytokines. IL-6 family cytokines drive receptor heterodimerization and subsequent Janus Kinase (JAK activation). JAKs phosphorylate tyrosine residues along the cytoplasmic domain of the receptor dimer, which recruits STAT proteins and facilitates their binding to interferon-gamma activation site-like (GAS-like) elements and regulation of large sets of genes. Asterisks (\*) denote that additional info can be found in Table 1.

**Table 1.** Summary of the Roles of STAT3 and STAT5 in Cancer and Strategies for their Inhibition.



**Table 1.** *Cont.*

Therapies included in the far right column are listed by their respective name followed by their biological target in parentheses. Italicized font denotes therapies that are not currently approved by the FDA as cancer therapies.

#### **3. IL-6 Family Cytokine Dysregulation in the TME**

A number of IL-6 family members have long been recognized for their involvement in the pathogenesis of aggressive cancers [19,77–81]. For example, IL-6 levels serve as a prognostic biomarker and predictor of therapeutic response in pancreatic ductal adenocarcinoma (PDAC), bladder, gastric, lung adenocarcinoma, colorectal, cervical, liver, and breast cancers: all of which have a high incidence of metastasis and resistance [77,82–86]. This is due to the pro-tumor effects of IL-6 in tumor cells and stromal components. In tumor cells, IL-6 can drive EMT, therapy resistance, and invasive characteristics [87–89]. Concurrently, IL-6 can shift the anti-tumor immune responses towards immunosuppression via recruitment of myeloid-derived suppressor cells (MDSCs) and expansion of FoxP3+ Tregs [90–92]. Furthermore, IL-6 secretion by certain stromal components (cancer-associated fibroblasts, macrophages, and neutrophils) drives EMT in associated tumor populations, which is supported by immunohistochemical staining at the invasive edge of patient breast tumors, where levels of phosphorylated STAT3 and IL-6 is elevated [93,94].

Interestingly, IL-6 activation of STAT3 functions in a positive feed-forward loop to drive the secretion of new IL-6 into the TME, which then interacts with IL-6R/gp130 to further activate JAK/STAT signaling [93,95]. OSM also potently activates a feed-forward loop, resulting in the de novo production of additional OSM and OSMR by tumor and immune cells [96]. Interestingly, OSM was first identified to play a tumor suppressive role and inhibited the proliferation of melanoma cell models [97]. However, later studies correlated OSM-OSMR signaling with robust STAT3 and STAT5 activation and more aggressive tumor phenotypes [98–106]. Like IL-6, OSM feed-forward signaling is observed in aggressive cancers with limited-therapeutic options, including glioblastoma (GBM), non-small cell lung carcinoma (NSCLC), PDAC, and triple negative breast cancer (TNBC) [107–111]. Elevated signaling through the OSM/OSMR axis induces high levels of an 'inflammatory module', which includes IL-6, CCL2, IL-1, CXCL1, CXCL9, CXCL10, and CXCL11—all of which have been implicated in migration, invasion, therapy failure, and dedifferentiation to a cancer stem cell (CSC) program. Importantly, neutralization of OSM by treatment with an Fc-tagged soluble OSMR suppresses this inflammatory module suggesting that the OSM/OSMR feed-forward loop may be critical for the long-term maintenance of inflammatory signaling [112,113].

IL-11 similarly engages in an autocrine feed-forward mechanism to drive persistent JAK2/STAT3 activation, which can promote resistance to platinum-based therapies [114]. Furthermore, IL-11 and LIF both contribute to tumorigenesis by enhancing tumor cell survival through STAT3-mediated activation of anti-apoptotic proteins, Bcl-2 and Survivin, and the inhibition of tumor suppressor p53 [114–116]. IL-11 and LIF also contribute to cancer progression by driving EMT through STAT3 and Akt/mTOR signaling, thereby conferring an enhanced migratory capacity [81,117,118]. The impact of CT-1, CLC, CNTF, and IL-27 on cancer progression is currently understudied. While IL-6 family cytokines are important mediators of STAT3 activation, other cancer-associated receptors can also activate STAT3. STAT3-driven tumor progression can also be achieved by epidermal growth factor receptors (both wild-type EGFR and EGFRvIII), fibroblast growth factor receptors (FGFR), insulin-like growth factor receptor (IGFR), hepatocyte growth factor (HGFR, also known as MET), platelet-derived growth factor receptor (PDGFR), vascular endothelial growth factor receptor (VEGFR), v-src, and Bcr-Abl [119–127]. Interestingly, acetylation of STAT3 also induces tumor progression through enhanced pro-tumorigenic IL-17A secretion [128,129]. These examples demonstrate the abundance of mechanisms through which STAT3 can become activated.

#### **4. STAT3 Activation of a Mesenchymal**/**CSC Program in Cancer Cells**

The majority of cancers have constitutive activation of STAT3 [130–132]. STAT3 was first termed oncogenic when its persistent activation was discovered in v-src transformed mouse embryonic fibroblasts [133]. Subsequent studies demonstrated that expression of a constitutively active STAT3 isoform (STAT3-C) can drive transformation of pre-malignant human mammary epithelial cells (HMEC) and MCF-10A cells to malignant breast cancer [134]. In addition, Ras-induced transformation in bladder and breast carcinoma models exhibit mitochondrial accumulation of STAT3 and more robust cellular glycolysis, a characteristic of cancer cells [135]. In PDAC, STAT3 is required for both the development of pre-malignant pancreatic interepithelial neoplasias (PanINs), as well as the progression of PanINs to PDAC [136]. While STAT3 has been strongly implicated as a driver of oncogenesis, evidence suggests that persistent cytokine activation of STAT3 in pre-malignant lesions, engages a tumor suppressive senescence response. In non-transformed HMEC models, OSM engages senescence through a direct STAT3 interaction with mothers against decapentaplegic-3 (SMAD3). However, downstream constitutive expression of c-Myc could overcome OSM-induced senescence and drive EMT and invasion [137]. Beyond the genetic events that occur as a normal cell becomes transformed, cancer progression relies heavily on an evolving microenvironment, which impacts both the cancer cells and the immune system, ultimately influencing patient outcomes. In cancer cells, increased STAT3 activity induces EMT-driving transcription factors, such as ZEB1, SNAIL, and Twist, which initiate the repression of epithelial markers and expression of mesenchymal markers (N-cadherin, Vimentin) [138–141]. In addition, STAT3 induces the expression of matrix metalloproteinases (MMP), which can also contribute to the loss of cell-cell contacts [142–144].

While EMT is important in normal physiology to allow cells to migrate during development and wound healing, inappropriate EMT and the loss of cell-cell interactions contributes to aggressive cancer cell behaviors, including metastasis and resistance to therapy. In TNBC, the loss of E-cadherin or gain of Vimentin, N-Cadherin, or Snail expression, typically at the tumor's edge, correlates with poor clinical outcome [145–149]. Likewise, the presence and abundance of circulating tumor cells (CTC) that express mesenchymal markers have been referred to as the "silent predictors of metastasis" because they correlate with tumor cell dissemination [150]. Moreover, mesenchymal CTC can be used to track a patient's response to therapy, with increasing numbers correlating with increased risk of relapse [151–155].

Likewise, the emergence of CSC properties, which are often defined experimentally as using tumorsphere- or tumor-initiating assays, frequently occur concomitantly with a mesenchymal phenotype [156–163]. STAT3 has been identified as an essential driver of the de novo reprogramming to a CSC state through activation of Sox2, Nanog, and Oct4 [164–170]. Further evidence linking EMT and a CSC phenotype is provided by the observation that a subset of CTC, responsible for initiating metastasis termed the metastasis-initiating cells, express high levels of the STAT3-driven CSC marker CD44 [151–153,171–174]

The importance of mesenchymal/CSC reprogramming in metastasis and therapy failure continues to emerge. For example, single-cell analysis identified a predominantly mesenchymal/CSC program in early-stage TNBC micro-metastases, in contrast to late-stage metastases [154]. The findings are consistent with the idea that mesenchymal/CSC initiate metastatic outgrowth at a secondary site, followed by differentiation. Single cell RNA sequencing (scRNA-seq) confirmed that EMT in primary tumors proceeds through distinct, hybrid states, ranging from completely epithelial to completely mesenchymal [155]. The epithelial-mesenchymal hybrids, which harbor the greatest level of phenotypic plasticity, are more efficient at intravasating, surviving in circulation, extravasating to the lungs, and forming metastases [175]. Acute exposure to Adriamycin or Taxanes drives the adaptive emergence of therapy-resistant, CD44HIGH CSC in both breast tumor explants and breast cancer cell lines [176]. scRNA-seq determined that chemo-resistant cells activate an EMT program, which was not evident before treatment [177]. In mouse models, the ability to undergo EMT is important for therapeutic resistance [178,179]. These findings suggest that potent mesenchymal/CSC programming has significant consequences that allow cancer cells to adapt to chemotherapy and survive, ultimately contributing to therapy failure.

Indeed, STAT3 has also been implicated in acquired therapy resistance in many cancers including NSCLC, GBM, PDAC, melanoma, and breast cancer [107,180–185]. Elevated STAT3 phosphorylation in ovarian cancer is associated with paclitaxel resistance and increased tumor cell invasion post-therapy, consistent with the elevated expression of mesenchymal/CSC genes and increased tumor initiating potential [176,186,187]. In Trastuzumab-resistant HER2+ breast cancer, STAT3 feed-forward loops generate a TME rich with STAT3-activating cytokines that promote and maintain the mesenchymal/CSC phenotypes. Inhibition of the STAT3 feed-forward activation diminished tumor growth and metastasis and resensitized cells to therapy [188]. In cancer cells driven by diverse receptor tyrosine kinases (RTKs) (EGFR, HER2, ALK, and MET), MEK inhibitors drive feedback activation of STAT3 through FGFR and JAKs, resulting in therapy failure [189]. Likewise, in NSCLC cells resistant to molecularly-targeted therapies (EGFR-TKI, ALK inhibitor Crizotinib, and MEK inhibitor Selumetinib), OSM/JAK1/STAT3–signaling protects cells from targeted drug-induced apoptosis [190]. JAK/STAT3 signaling also interacts with numerous other growth, proliferative, and survival pathways, such as PI3K/Akt, MAPK, NF-kB, Notch, Wnt/β-Catenin, and TGF-β, among others [191–196]. Tamoxifen-resistant breast cancers have elevated STAT3 and Notch4 expression associated with metastasis and tumorigenicity. Interestingly, inhibition of Notch4 was able to reduce the phosphorylation of STAT3 and suppress metastasis, suggesting STAT3 and Notch4 cooperate to promote therapy resistance [197]. In addition, a STAT3/NF-kB complex promotes cisplatin resistance in malignant mesothelioma, and targeted inhibition of this complex inhibits the growth of refractory tumors [198]. Such evidence suggests STAT3 may not be solely responsible for therapy failure. Instead, STAT3-mediated cross-talk with additional signaling effectors may coordinate resistance [140,199–201]. Nonetheless, the common theme appears to be a STAT3-activated, mesenchymal/CSC program.

#### **5. A STAT3-Generated Pro-Metastatic Immune Microenvironment**

An important and rapidly emerging area of research is the impact of the immune microenvironment on metastasis and therapy failure. Beyond its impact on tumor cells, STAT3 has an important role in restricting immune cell functions and producing immunosuppressive factors. STAT3-induced cytokines are important mediators of the crosstalk between tumor cells, tumor-associated macrophages (TAMs), and tumor-associated neutrophils (TANs), which are responsible for generating a pro-metastatic and pro-angiogenic TME [202]. In the TME, macrophages and neutrophils are exposed to a host of STAT3-induced cytokines, such as IL-4, IL-6, IL-10, IL-13, VEGFA, and TGF-β that drive their polarization to a pro-tumorigenic M2 (macrophages) or N2 (neutrophils) state [203,204]. Conversely, robust STAT3 signaling activity in M2-TAMs and N2-TANs correlates with the production of factors able to drive cancer cell EMT (OSM, IL-6, TGF-β) and angiogenesis (VEGF, TGF-β, PDGF, and FGF) [205–208].

Importantly, because of their localization at the invasive edge of tumors, the contribution of M2-TAMs and N2-TANs to metastasis is increasingly being recognized. For example, TAMs have been demonstrated to form simultaneous physical contacts with tumor cells and endothelial cells that result in the formation of invadopodia, which assist cancer cells in transendothelial migration and escape into the circulatory system. These sites of cancer cell intravasation are called "tumor microenvironment of metastasis (TMEM)" and have been validated as prognostic markers of metastasis. Chemotherapy increases TMEM in breast cancer patients, thereby potentially facilitating metastasis [209,210]. STAT3 inhibition in TAM populations re-sensitizes breast cancer cells to paclitaxel, further suggesting tumor and TAM crosstalk is an essential component in STAT3-driven therapy resistance [205,211]. TANs, on the other hand, were found to function as circulatory escorts of CTC and promote their proliferation, survival, and seeding of secondary sites [212–214]. Importantly, OSM and IL-6 were two of the 4 most frequent cytokines secreted by neutrophils found clustered with CTC. These cytokines interact with OSMR and IL-6R, which are expressed on cancer cells [212]. The neutrophil/CTC cross-talk promotes the cell cycle progression in CTC, thereby expanding their metastatic potential.

Following the classical events of the metastatic cascade, after a tumor cell intravasates into circulation, it extravasates to establish secondary sites. An emerging concept in the metastatic cascade is the role of myeloid cells- such as basophils, neutrophils, eosinophils, monocytes, and macrophages- in the establishment of a pre-metastatic niche [215]. Although the exact TME components required for colonization of secondary sites remain a mystery, the activation of a STAT3-sphingosine 1-phosphate receptor-1 (S1PR1) axis in tumor cells, and secretion of IL-6 and IL-10, was observed to persistently activate STAT3 at distant, pre-metastatic sites. At these pre-metastatic sites, persistent STAT3 phosphorylation was associated with myeloid cell migration from the primary tumor to the secondary site. Subsequent targeting of STAT3 in myeloid cells disrupted metastatic tumor outgrowth, suggesting STAT3 plays an integral role in priming distant metastatic sites for tumor cell outgrowth [216].

#### **6. STAT5 in Cancer Cells**

As previously mentioned, cells produce two different STAT5 proteins, STAT5A and STAT5B, which share greater than 90% sequence homology [217]. However, evidence suggests STAT5A and STAT5B play different functional roles in normal and cancer cell systems. Genetic deletion of STAT5 in pure mouse backgrounds are embryonic lethal, due to the essential roles of STAT5 in erythropoiesis and iron metabolism [218–220]. While the distinct roles of STAT5A and STAT5B remain understudied, data from mammary gland-specific knockouts suggest that STAT5A is required for lactogenesis in the mammary gland, while STAT5B is imperative for mammary gland differentiation and development [221]. The function of STAT5A and STAT5B appear to be cell-specific. They can have either synergistic or opposing effects, such as in memory B-cell differentiation, which may be due to (i) the formation of STAT5A/STAT5B homo- or hetero-dimers, and/or (ii) differences in nuclear shuttling mechanisms [222–224]. Furthermore, genetic tuning models depleting STAT5A and/or STAT5B have demonstrated a critical role for STAT5 in the accumulation and development of innate lymphoid cells, such as NK cells [225–227].

Given its role in lymphoid cell development and differentiation, it is not surprising that STAT5 activity contributes to hematologic malignancies. Deletion of STAT5 prevents transformation by the Abl oncogene, thereby preventing leukemia development [228]. Genetic and pharmacologic

inhibition of STAT5 activity decreases expression of apoptosis inhibitors MCL1 and BCL2 and inhibits leukemogenesis of BCR-ABL1+ acute lymphoblastic leukemia (ALL), both in cell lines and newly diagnosed and relapsed/TKI-resistant ALL patients [229]. Likewise, a new, STAT5 inhibitor suppressed the proliferation of human acute myeloid leukemia (AML) cell lines and primary FLT3-ITD+ AML patient cells [74]. Combined inhibition of STAT3 and STAT5 by shRNAs also suppressed growth in chronic myeloid leukemia, suggesting that combinatorial suppression of STAT3 and STAT5 may be efficacious in treating hematological malignancies [230].

STAT5 activation has also been implicated in the progression of solid tumor malignancies. Deletion of STAT5 in the mammary gland, hepatocytes, and prostate cells delays the development of mammary, liver, and prostate cancer [32,231,232]. Like STAT3, experimental evidence implicates STAT5B as a driver of tumorigenesis, as it can drive EMT and increased invasiveness in hepatocellular carcinoma (HCC) [233]. Moreover, in mammary epithelial cells, thymocytes, and epithelial prostate cancer cells, persistent activation of STAT5 is sufficient to drive transformation [234–236]. Furthermore, the function of STAT5 in solid tumors extends beyond oncogenesis as evidence has emerged that STAT5 signaling can induce a metastatic cascade. For example, STAT5 inhibition in colorectal cancer induces G1 cell cycle arrest and reduces cancer cell migration, demonstrating the role of STAT5 in proliferation and metastasis [237]. Additional studies demonstrate that STAT5A/B signaling in prostate cancer and squamous cell carcinoma of the head and neck drive EMT programming, which results in enhanced cell migration, invasion, and formation of metastases [238,239].

An increasing number of reports demonstrate that STAT5 drives tumorigenesis and cancer progression through cooperation with other intracellular signaling cascades and activation of additional feed-forward loops. STAT5 activates transcription of AKT1 and PI3K, and, in turn, Akt1 phosphorylates STAT5 to induce cell survival [240,241]. Furthermore, STAT5-dependent Akt restores cyclin D expression, which promotes proliferation [241]. Once cells aberrantly proliferate, apoptosis suppressors Bcl-xL and Bcl-2 are activated via persistent STAT5 signaling, driving tumor cell survival [16,234,242–244]. BCL-XL expression is also enhanced via formation of a transcription factor complex comprised of phosphorylated STAT5 and nuclear EGFRvIII. This transcription factor complex binds to the BCL-XL promoter to induce its transcription and promote anti-apoptotic signaling [245]. Similar to STAT3, STAT5 translocates to tumor cell mitochondria, suggesting an interaction with the mitochondrial genome to promote aerobic glycolysis (the Warburg Effect), a defining characteristic of cancer cells [246]. Collectively, this data demonstrates that STAT5 mediates crosstalk between cancer cell survival, and proliferation, and metabolism signaling pathways.

Surprisingly, STAT5 in certain model systems has been demonstrated to function in a tumor suppressive manner. Human breast cancers infrequently (~7%) show signs of STAT5 activation (compared to 40% of STAT3 activation). This elevated STAT5 activity trends with more differentiated and lower grade tumors, suggesting that STAT5 does not induce the aggressive cancer cell program initiated by STAT3, at least in breast cancer [247]. STAT5 expression in these models stabilizes E-cadherin surface marker expression and reverses the undifferentiated mesenchymal phenotype [248]. In normal human fibroblasts, aberrant activation of STAT5A induces a senescence response concurrent with accumulation of p53 and DNA damaged foci. Furthermore, knockdown DNA-repair kinase, ATM, and tumor suppressor, retinoblastoma protein did not eliminate damaged foci, providing evidence for the persistence of DNA damage in pre-malignant lesions [249]. Interestingly in HCC models, liver specific STAT5 knockout results in tumor formation through the enhanced activation of TGFβ/STAT3 signaling [20]. Physiologically, STAT proteins have been identified as drivers of erythropoiesis. STAT5A/B double knockout mice in a mixed genetic background (Sv129 x C57Bl/6) results in mild hematopoietic phenotypes, due to compensatory activation and enhanced DNA-binding of STAT1/3 [218].

#### **7. STAT5 in Treg-Associated Immunosuppression**

In addition to its impact on cancer cells, activated STAT5 dampens anti-tumor immune function. This immunosuppressive function is largely driven by CD4+/CD25+ Tregs, a subset of T cells that contribute to tumor progression and metastasis [250] and correlates with poor patient prognosis [251,252]. Sustained STAT5 phosphorylation in progenitor T cells induces the differentiation to a Treg population that, in turn, significantly diminishes the function of cytotoxic and helper T cells [253–255]. Experimental depletion of Tregs from the TME results in enhanced infiltration of mature CD4+ and CD8+ T cells into the tumor, leading to tumor rejection [256,257]. Furthermore, STAT5-mediated Treg expansion increases IL-10, IL-4, and IL-13, which skew TAMs to an M2 immunosuppressive phenotype. M2 macrophages are immunosuppressive because they release elevated levels of IL-10 and transforming growth factor-β (TGF-β) and restrict secretion of immune stimulatory cytokines via NF-kB repression [258,259]. STAT5-induced Tregs and M2 macrophage populations secrete VEGF-A, which, along with TGF-β, promotes angiogenesis [255,260,261].

Tumor clearance mechanisms are also suppressed by Treg expansion through impediment of B cell development and maturation [262]. STAT5-activated Tregs result in a decrease in follicular helper T (Tfh) cell populations via Blimp-1, which severely hampers germinal center formation in lymph nodes [263]. This reduction in germinal centers diminishes the number of B cells that can be recruited and primed to aid in an anti-tumor immune response. Interestingly, the differentiation and self-renewal of memory B cells are respectively influenced by STAT5A-mediated repression and STAT5B-mediated induction of BCL-6. Immunosuppression driven by the expansion of Tregs, relies on STAT5-mediated alteration in T-cell metabolism. Mature effector T cells preferentially undergo glycolysis and require a de novo fatty acid synthesis reliant on acetyl-coA carboxylase 1 whereas Tregs undergo lipid-oxidation and readily synthesize fatty acids because of a structural reconfiguration of mitochondrial cristae [264–266]. Accumulation of intracellular lipids impairs autophagy, providing a mechanism for Treg-mediated immunosuppression [267,268]. Just as STAT5 function in the mitochondria impacts tumor cell functions, as described earlier, mitochondrial STAT5 activation drives metabolic shifts in the immune compartment, inducing an expansion of Tregs.

However, while the suppression of Tregs increases tumor immunity, provided STAT5 remains functional, the overall impact of suppressing STAT5 signaling in the remaining immune cells of the TME remains open to debate. Mouse models deficient in STAT5 have depleted CD8+ T cell, NK cell, and Treg populations, suggesting STAT5 plays an integral role in the proper development of multiple immune cell types [269]. STAT5 signaling contributes to the differentiation of naive CD4+ T cells into CD8+ T cells, Th1, Th2, Th9, ThGM, and Tregs, while inhibition of STAT5 is required for the generation of Th17 and Tfh cells [270,271]. Importantly, STAT5 is heavily involved in the development, survival, and lytic function of NK cells. Knock-out or suppression of STAT5 in NK populations sparked the secretion of VEGF-A, a growth factor that supports tumor-associated angiogenesis in melanoma and leukemia models [269,272]. Taken together, these findings suggest that targeting STAT5 may threaten the integrity of anti-tumor immune functions and drive worse outcome in patients [269,272].

#### **8. Targeting STAT Activity**

Given the roles of STAT3 and STAT5 in tumor progression and immunosuppression discussed above (Figure 2), multiple methods of inhibiting their activity are being pursued. Successful inhibition of STAT3 would prevent the acquisition of, and potentially revert, a mesenchymal/CSC program, making cancer cells less invasive and more sensitive to therapy. Moreover, STAT3 inhibition would help activate anti-tumor immunity by reducing immunosuppressive factors and increasing the infiltration of immune cells into the TME. Likewise, suppressing STAT5 in cancer cells, particularly leukemia, halts proliferation and induces apoptosis, suggesting that STAT5 may be a valuable therapeutic target [74]. However, due to STAT5 s controversial roles in tumor progression and immune cell maturation and differentiation, further studies are required to elucidate the effects of targeting STAT5 in cancer patients. [273]. Generally speaking, the majority of small molecule inhibitors are designed to have

high affinity for the catalytic domain of an enzyme; the ATP binding site for example, which STAT proteins lack. Therefore, direct inhibition of STAT3 relies on disruption of binding motifs necessary for downstream signal transduction. A series of STAT3 phospho-ester SH-2 domain inhibitors have been developed, with the intent to inhibit dimerization of activated STAT3 (PY\*L, S3I-2001, OPB-31121, etc.). While these approaches are promising, the compounds continue to be refined [274–277]. An alternative approach would involve targeting the upstream activator of STAT3, which could include neutralizing antibodies for specific IL-6 family cytokines, competitive antibodies hindering cytokine-receptor interactions, and JAK inhibitors.

**Figure 2.** Biological Impact of STAT3 and STAT5 Activation. Pre-malignant cell populations, in which apoptotic signaling and immunosurveillance are functional, exhibit low levels of phosphorylated STAT3 (pSTAT3) and STAT5 (pSTAT5). Elevated activity of pSTAT3 and/or pSTAT5 accompanies tumorigenesis, leading to the inhibition of apoptotic pathways and repression of immune cell recognition of a burgeoning tumor. pSTAT3 is utilized by malignant cell populations to drive epithelial-mesenchymal transition (EMT) and by tumor-associated macrophages (TAMs) and tumor-associated neutrophils (TANs) to drive metastasis. pSTAT5 activity in progenitor T-cells drives expansion of a Treg population that then secretes factors that inhibit the function of CD4+ and CD8+ T-cells as well as B-cells.

One of the first IL-6 monoclonal antibodies generated was Tocilizumab, which inhibits IL-6 signaling by preventing IL-6 binding to both the soluble and transmembrane forms of IL-6R [278,279]. Pre-clinical studies demonstrated strong anti-tumor cell activity in multiple myeloma and therapeutic efficacy in Castelman's disease, rheumatoid arthritis, and cytokine release syndrome, however, Tocilizumab is not currently FDA-approved for cancer treatments [280–283]. Clinical trials using Tocilizumab in combination with other monoclonal antibodies, chemotherapies, and immunotherapies to treat cancers such as HER2+ breast cancer, B-cell Non-Hodgkin Lymphoma, metastatic NSCLC, and recurrent metastatic colorectal adenocarcinoma are ongoing [284]. In addition to Tocilizumab, Siltuximab is an IL-6 specific neutralizing antibody that has emerged as another promising therapy. Pre-clinical studies support Siltuximab use in KRAS-mutant lung cancer, particularly in tumors with elevated stromal production of IL-6 [285]. Similar to Tocilizumab, Siltuximab is undergoing clinical trials for patients with malignant solid tumors in cancers with elevated stromal presence, such as ovarian, pancreatic, colorectal, head and neck, and lung neoplasms [286]. Furthermore, phase II studies of combination therapy with Siltuximab have demonstrated anti-tumor effects for patients diagnosed with metastatic prostate cancer, suggesting clinical efficacy of targeting the IL-6-STAT signaling axis with combination therapies [287].

Our lab has focused on OSM, given its potent activity at inducing numerous inflammatory cytokines that promote mesenchymal/CSC reprogramming in cancer cells and generate a pro-tumor immune microenvironment [106,288–290]. Neither OSM nor OSMR protein are abundantly expressed in normal tissues in the absence of inflammation, in contrast to JAK1, JAK2, and STAT3 (key OSM effectors) [109]. This finding is supported by the observation that knocking out OSM or OSMR from mice results in only mild phenotypes [291]. Moreover, OSMR has characteristics unique from other IL-6 family co-receptors, resulting in distinct signaling and biological effects [43,292,293]. For example, OSMR strongly recruits SHC, resulting in the hyper-activation of the MAPK signaling cascade [293]. Other gp130 co-receptors fail to do this, and rely solely on the gp130-mediated SHP-2 recruitment for MAPK activation, which is less robust than the gp130/OSMR heterodimer [43]. Therefore, suppressing the OSM signaling axis may have benefits in aggressive subtypes of cancer. Anti-OSM antibody (GSK315234) and OSMR-fusion protein (OR-FC) were first examined in pre-clinical studies in chronic inflammatory disease models, such as rheumatoid arthritis and inflammatory heart disease [294,295]. Antibodies suppressing OSM signaling ameliorate these pathological conditions, suggesting that OSM is a driver of these chronic hyper-inflamed states. More recently, OSM signaling was identified as a driver of inflammatory bowel disease (IBD), especially in patients who fail to respond to anti-tumor necrosis factor-α (TNF-α) antibodies. Neutralization of OSM in IBD models suppresses a cadre of inflammatory cytokines and reduces colitis severity, further supporting the OSM feed-forward loop as a critical mediator in the long-term maintenance of inflammatory signaling, a state common in the TME [112].

Another recent study investigating the effects of an anti-OSM antibody in a murine model of lupus nephritis demonstrated that OSM-driven EMT and extracellular matrix secretion, leading to renal fibrosis, could be suppressed, concomitant with the suppression of JAK/STAT3 activation [296]. More recently, a clinical grade OSM neutralizing antibody was used to treat pre-clinical models of squamous cell carcinoma. Again, OSM neutralizing antibodies suppressed the STAT3 feed forward signaling, resulting in reduced invasion and metastasis [96,297]. Disrupting cell surface receptor activation of STAT3 by inhibiting OSMR signaling was recently described in a study of an OSMR/gp130 antagonist (SN79), which prevents STAT3 phosphorylation in astrogliosis [298]. While these studies demonstrate the potential of OSM and OSMR inhibition as a therapeutic strategy, a number of ligand/receptors activate STAT3 and STAT5, as described above. Therefore, additional studies will be needed to define whether single ligand-receptor inhibitors can sufficiently impact STAT3 activation, thus suppressing tumor growth.

Currently, JAK inhibitors are the most promising inhibitors of STAT-driven phenotypes. Commonly used JAK 1/2 inhibitor, Ruxolitinib, has been shown to robustly block both STAT3 and STAT5 activation. Yet, while JAK inhibitors can reduce STAT3 activation, there is conflicting data on the impact of JAK inhibitors. Some studies find that JAK inhibition suppresses tumor growth [93], while other studies find that they ultimately enhance metastasis, likely because they also suppress the positive influence of JAK activity on other STAT proteins (including STAT5 and STAT1/2, as discussed below) [299]. In addition, side effects of JAK inhibitors may be more pronounced, as JAKs activate other pathways, such as MAPK and PI3K in normal cells as well as cancer [43].

#### **9. Balancing Opposing STAT-Activated Cytokine Signaling as a Therapeutic Strategy**

Though STAT proteins are grouped together based on their structural similarities and common functions as transcription factors, individual STAT proteins can have diverse functions. We have focused extensively on STAT3 and STAT5 and their described roles in cancer progression, however this is not the universal effect among all STATs. For example, Interferon-β (IFNβ) induces the phosphorylation and activation of STAT1/2, which form a complex with IRF9 to create the transcription factor complex ISGF3. IFNβ/P-ISGF3 signaling induces interferon-stimulated genes (ISG), mesenchymal-epithelial transition (MET), and the differentiation of CSC into a less aggressive epithelial, non-CSC state with reduced migratory potential and reduced tumorsphere forming capabilities [111,300,301]. Importantly, the IFNβ and OSM/STAT3 signaling pathways strongly oppose one another. OSM represses transcription of IFNβ, thereby eliminating autocrine and paracrine IFNβ-mediated activation of P-ISGF3 and repressing ISG expression in both cancer cells and immune cells [301].

In addition to the impact of IFNβ on cancer cells, increasing rationale supports developing methods for the delivery of P-ISGF3 activators (or Type I IFN-agonists more generally) directly to the TME. First, favorable responses to frontline chemotherapy correlate with robust IFN signaling in both mouse and human studies [111,302–304]. Elevated IFN signaling in the tumor correlates with immunologically "hot" tumors harboring elevated numbers of tumor infiltrating lymphocytes (TILs), activated immune surveillance, increased tumor antigen cross presentation, and diminished numbers of immunosuppressive cells including MDSCs and Tregs [111,302,305–307]. Second, loss of Type I IFN signaling correlates with metastasis and decreased survival. Restoration of Type I IFN signaling significantly decreases metastasis and improves survival outcome [303,304]. Third, in contrast to STAT3 activators (which promote a pro-tumorigenic M2 state), addition of type I IFN inhibits macrophage polarization to an M2 state [308,309]. Type I IFNs also induce the differentiation of neutrophils into anti-tumor N1s [310]. Fourth, administration of IFNβ prior to surgical resection significantly improves response rates to immunotherapies such as anti-PD1/anti-PDL-1 [311]. Therefore, we propose balancing pro-tumor STAT3 activation with anti-tumor STAT1/STAT2 activation as a novel therapeutic approach. This STAT3/STAT1 balancing would (i) reprogram mesenchymal/CSC to a non-CSC state, making them more susceptible to chemotherapy and (ii) enhance anti-tumor immunity, thereby facilitating immune cell-mediated tumor cell killing. Yet, while IFN treatment is currently approved to treat hematological malignancies and some solid tumors (melanoma), the high doses of IFNs needed to inhibit cancer cell proliferation or induce cell death result in side-effects that limit its effectiveness [311–313]. We suggest tumor targeting antibodies (or nanobodies) linked to IFNβ. Generation of an oncogenic cytokine or receptor antibody conjugated to an IFNβ first demonstrated success in limiting resistance to EGFR inhibitors in breast cancer [314]. The designed therapy sought to re-activate innate and adaptive immune components, while simultaneously targeting the oncogenic receptor EGFR [303]. IFNβ-conjugated antibodies show immense promise. They would limit toxicity by using tumor-associated receptors to target IFNs to the TME and suppress STAT3 activation, while simultaneously, activating STAT1/2 in both tumor cells and immune cells [315].

#### **10. Conclusions**

As discussed throughout this review, STAT3 and STAT5 have emerged as essential components involved in regulating tumor progression. Cytokines and cytokine receptors of the IL-6 family are some of the most widely recognized STAT activators and are abundantly expressed on cancer cells as well as tumor-infiltrating immune cells. The role of STAT3 in promoting molecular programs in cancer cells that induce tumor metastasis and therapy resistance mechanisms continue to emerge, as does the impact of STAT3 as a suppressor of immune cell function in the TME. These findings suggest that specifically suppressing STAT3 activation would be beneficial to patients. Targeting STAT5 in hematological malignancies is gaining traction, and the clinical successes of JAK and tyrosine kinase inhibitors in disrupting STAT3 and STAT5 activation, provide strong support for the development of direct STAT3 and STAT5 inhibitors, summarized in Table 1. Specific suppression of STAT5 in immune-suppressive Tregs could also prove beneficial, but STAT5 is essential for many other immune cells as well. Thus, systemic suppression of STAT5 activity could undermine tumor immunity and promote tumor progression, as recently reported for STAT5 knock-out from NK cells [272]. However, reducing the aberrant activation of STAT5 without complete ablation may have therapeutic efficacy upon combination with other vulnerabilities. We conclude that targeting STAT3, either directly by disrupting STAT3-homodimer formation or indirectly by suppressing the activation of receptors responsible for persistent STAT3 phosphorylation, would reverse the cellular programs driving metastasis and therapy failure. Furthermore, by activating STAT1/2 within TME cells, the programs that prevent metastasis and enhance therapeutic efficacy could be re-engaged. This STAT balancing would improve outcomes for patients, particularly those with aggressive cancers that may currently have limited therapeutic options.

**Funding:** This research is supported by T32 (CA059366) to Kelsey Polak.

**Acknowledgments:** Both Figures 1 and 2 were created with BioRender.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
