Ponatinib Activates an Inflammatory Response in Endothelial Cells via ERK5 SUMOylation

Ponatinib is a multi-targeted third generation tyrosine kinase inhibitor (TKI) used in the treatment of chronic myeloid leukemia (CML) patients harboring the Abelson (Abl)-breakpoint cluster region (Bcr) T315I mutation. In spite of having superb clinical efficacy, ponatinib triggers severe vascular adverse events (VAEs) that significantly limit its therapeutic potential. On vascular endothelial cells (ECs), ponatinib promotes EC dysfunction and apoptosis, and inhibits angiogenesis. Furthermore, ponatinib-mediated anti-angiogenic effect has been suggested to play a partial role in systemic and pulmonary hypertension via inhibition of vascular endothelial growth factor receptor 2 (VEGFR2). Even though ponatinib-associated VAEs are well documented, their etiology remains largely unknown, making it difficult to efficiently counteract treatment-related adversities. Therefore, a better understanding of the mechanisms by which ponatinib mediates VAEs is critical. In cultured human aortic ECs (HAECs) treated with ponatinib, we found an increase in nuclear factor NF-kB/p65 phosphorylation and NF-kB activity, inflammatory gene expression, cell permeability, and cell apoptosis. Mechanistically, ponatinib abolished extracellular signal-regulated kinase 5 (ERK5) transcriptional activity even under activation by its upstream kinase mitogen-activated protein kinase kinase 5α (CA-MEK5α). Ponatinib also diminished expression of ERK5 responsive genes such as Krüppel-like Factor 2/4 (klf2/4) and eNOS. Because ERK5 SUMOylation counteracts its transcriptional activity, we examined the effect of ponatinib on ERK5 SUMOylation, and found that ERK5 SUMOylation is increased by ponatinib. We also found that ponatibib-mediated increased inflammatory gene expression and decreased anti-inflammatory gene expression were reversed when ERK5 SUMOylation was inhibited endogenously or exogenously. Overall, we propose a novel mechanism by which ponatinib up-regulates endothelial ERK5 SUMOylation and shifts ECs to an inflammatory phenotype, disrupting vascular homeostasis.


INTRODUCTION
CML and Ph + ALL involve the reciprocal translocation of the Abl oncogene on chromosome 9 and the Bcr on chromosome 22 (1)(2)(3). The resulting chromosomal fusion produces a constitutively active Bcr-Abl tyrosine kinase (4) that promotes dysregulated proliferation and survival signaling, leading to leukemogenesis (1,5). Therefore, Bcr-Abl kinase is the primary therapeutic target for CML patients. Newly diagnosed CML patients commonly receive imatinib (first generation TKI), a small molecule that binds the ATP pocket on the Bcr-Abl tyrosine kinase, as a first line of treatment (6,7). Because CML patients often develop Bcr-Abl point mutations conferring resistance to imatinib (8), dasatinib, and nilotinib (second generation TKIs) were generated (9). T315I is a specific point mutation present in ∼15% of relapsed CML patients (10) that causes therapeutic resistance to all currently approved first and second generation TKIs (11)(12)(13). Ponatinib (a third generation TKI) was specifically designed to circumvent the sterical hindrance warranted by the T315I mutation (14)(15)(16)(17).
In spite of having superb clinical efficacy, ponatinib treatment comes with an array of adverse side effects attributable to the broad-spectrum inhibition of multiple kinase families in addition to Bcr-Abl (3). Common secondary effects of ponatinib treatment are xerostomia, abdominal pain, and cytopenia. Specifically in the cardiovascular system, ponatinib treatment induces substantial arterial and venous VAEs (24) including peripheral arterial occlusive disease (25), ischemic heart disease (26), cerebrovascular accident, venous thrombo-embolism (27), pulmonary hypertention (28), platelet dysfunction, and hyperglycemia (26)(27)(28)(29)(30). In a prospective analysis of 19 patients who received ponatinib therapy, 42% developed arterial cardiovascular events after 8.5 months. A phase I trial showed a significant percentage of vascular occlusive events (24,31) and a phase II trial (PACE) demonstrated a strong correlation between ponatinib administration and serious arterial thrombotic events (10,32). A randomized, opened-label phase III trial (EPIC) designed to compare efficacy between ponatinib and imatinib as first line treatments in newly diagnosed CML patients was terminated early due to serious VAEs observed in the ponatinib treated group (33). Ponatinib-associated VAEs are a serious clinical challenge in CML patients subjected to this therapeutic regime (34). A broad comparative profiling analysis of ponatinib and other TKIs showed that ponatinib inhibits VEGFRs with greater potency (26), through which it reduces viability, function, migration, and tube formation in ECs, thus causing vascular toxicity (35). Ponatinib-associated VEGFR2 inhibition has also been implicated in hypertension (26).
Even though ponatinib-mediated VAEs have been documented (36,37) the exact molecular mechanism by which this drug induces VAEs remains obscure. Interestingly, despite promoting arterial thrombotic events (26,38), ponatinib inhibits platelet activation, aggregation, spreading, and granule secretion (39). These observations suggest that ponatinibassociated thrombotic events are not due to the activation of platelets (40), but rather of other cell types. Because ponatinib treatment increases EC dysfunction and apoptosis (35), both of which are associated with a higher rate of VAEs (41), it is plausible that ponatinib-associated VAEs are related to EC inflammation, dysfunction and apoptosis.
In the current study, we tested the hypothesis that ponatinib triggers an endothelial inflammatory response by promoting ERK5 SUMOylation. Table 1.

ERK5 Transcriptional Activity Assay (Mammalian One-Hybrid Assay)
Sub-confluent ECs plated on 6-well-plate were incubated in Opti-MEM medium (Invitrogen, Carlsbad, CA, USA) containing Plus-Lipofectamine transfection reagents, the pG5 luciferase (pG5-Luc) and pBIND-ERK5 plasmids with pcDNA3.1-CA-MEK5α or control pcDNA3.1 vector, as we performed and described previously (41), for up to 4 h. Then, the transfection mixture was removed, ECs were washed, and ECM was added. Next, cells were treated with ponatinib at the concentrations indicated in the figures, for 24 h. Finally, cells were harvested, lysed, and luciferase activity was measured by a TD-20/20 Luminometer (Turner Designs, Sunnyvale, CA, USA), using The pG5-Luc plasmid has five Gal4 binding sites upstream of a minimal TATA box, which in turn, is upstream of the firefly luciferase gene. The pBIND-ERK5 plasmid has Gal4 fused with ERK5. Because pBIND vector also contains the Renilla luciferase gene, the expression and transfection efficiency were normalized to the Renilla luciferase activity. Therefore, relative ERK5 transcriptional activity was calculated by normalizing firefly luciferase activity according to Renilla luciferase activity (firefly:renilla luciferase activity ratio).

Flow Study
Confluent HAECs cultured in 100-mm dishes were exposed to laminar flow using a cone-and-plate apparatus placed in an incubator at 37 • C and 5% CO 2 for 24 h, as we previously described (52).

KLF2 Promoter Activity
Sub-confluent HAECs were transfected with Flag-ERK5, a reporter gene encoding KLF2 promoter (−924 ∼ + 14) and the luciferase control reporter vector pRL-CMV, using an OPTI-MEM/Plus-Lipofectamine mix, as we previously described (52). After incubating 3 h at 37 • C, the transfection mix was removed and replaced with complete ECM. Next day, complete ECM was replaced with low serum ECM (0.2% FBS, 1% P/S, no ECGF). After 1 h, ponatinib (150 nM) was added to the medium and cells were incubated an additional 24 h. KLF2 promoter luciferase activity was assayed using a dual-luciferase reporter system.  Graph shows the fold increase of apoptosis in ponatinib treated cells compared to control cells. Data is sourced from two independent experiments, each contains 4-5 technical replicates. Error bars represent mean ± SEM. Statistical significance was assessed using student's t-test (two-tailed). **p < 0.01. (B) Expression of phospho-p65, total p65, VCAM1, and β-actin (loading control) in HAECs treated with ponatinib was assessed by Protein simple WES system (capillary electrophoresis western analysis). Protein bands are shown as pseudoblots. (C) Graph demonstrates relative NF-kB activity, as measured by promoter-driven luciferase reporter gene assay in HAECs treated with phamacological concentrations of ponatinib for 24 h, and presented as ratio of firefly/renilla luciferase activity. A representative data set of three independent experiments is shown, contains 11-13 technical replicates. Error bars represent mean ± SEM. Statistical significance was assessed using ANOVA followed by Bonferroni post hoc testing for multiple group comparison. ***p < 0.001. (D-E) qRT-PCR analysis of relative vcam1 and icam1 expression in HAECs treated with ponatinib (150 nM, 24 h). A representative data set of three independent experiments is shown, contains 5-6 technical replicates. Error bars represent mean ± SEM. Statistical significance was assessed using student's t-test (two-tailed). ***P < 0.001; *P < 0.05 vs. vehicle control (Veh) (F) Ponatinib (150 nM) effect on transcellular electrical resistance was assessed using ECIS system as described in methods. "Scratch" Wound Closure Assay Confluent HAECs transduced with Ad-ERK5-WT, Ad-ERK5-K6/22R or Ad-LacZ were wounded using a 1000 µL microtip. Complete ECM medium was replaced with low serum ECM, and cells were treated with ponatinib. 48 h later, cells were photographed and wound closure ability was assessed by comparing the wound size of ponatinib and veh-treated cells.

Statistics
Differences between two independent groups were determined using the student's t-test (two-tailed). Differences between multiple groups were determined using one-way analysis of variance (ANOVA) followed by Bonferroni post hoc testing for multiple group comparison by GraphPad Prism (GraphPad Software, San Diego, CA, USA). P values < 0.01 were considered statistically significant and are indicated by two asterisks in the figures. P-value < 0.001 is indicated by three asterisks.

Ponatinib Triggers an Inflammatory Response in ECs
Cell viability, migration, and functionality are decreased in HUVECs treated with ponatinib (35). To examine if the observed effect is similar across different types of ECs, we treated HAECs with ponatinib to assess apoptosis. Flow cytometric analysis of Annexin V staining indicated increased cell apoptosis in ponatinib treated group (Figure 1A). Because dying cells trigger an inflammatory response (64), we asked if ponatinib-associated apoptosis triggers an endothelial inflammatory response. In HAECs treated with pharmacologically relevant concentrations of ponatinib (75 nM, 150 nM) (35,65), we noted a significant increase on NF-kB p65 phosphorylation ( Figure 1B) and NF-κB activity ( Figure 1C). Expression of inflammatory genes including vascular and intercellular cell adhesion molecule 1 (vcam1 and icam1) was also increased, both at mRNA (Figures 1D,E) and protein levels ( Figure 1B) in ponatinib treated cells. EC apoptosis FIGURE 2 | Ponatinib inhibits ERK5 transcriptional activity (A) Graph demonstrates relative ERK5 transcriptional activity, as measured by mammalian one-hybrid assay, in HAECs treated with ponatinib (150 nM, 24 h), and presented as ratio of firefly/renilla luciferase activity. A representative data set of three independent experiments is shown, contains 6 technical replicates. Error bars represent mean ± SEM. Statistical significance was assessed using student's t-test (two-tailed). *P < 0.05 vs. Veh control. (B-D) qRT-PCR analysis of relative klf2, klf4, and enos expression in HAECs treated with ponatinib (150 nM, 24 h). A representative data set is shown, contains 4-6 technical replicates. Error bars represent mean ± SEM. Statistical significance was assessed using student's t-test (two-tailed). ***P < 0.001; **P < 0.01; *P < 0.05 vs. Veh control. (E) Graph demonstrate relative ERK5 transcriptional activity, as measured by mammalian one-hybrid assay in HAECs treated with ponatinib (150 nM, 24 h). Results are presented as ratio of firefly/renilla luciferase activity. As indicated, some cells were also transfected with CA-MEK5α. A representative data set of three independent experiments is shown, contains 11-20 technical replicates. Error bars represent mean ± SEM. Statistical significant was assessed using ANOVA followed by Bonferroni post hoc testing for multiple group comparison. ***P < 0.001 vs. Veh control. (F) Graph demonstrates relative KLF2 promoter activity, as measured by promoter-driven luciferase reporter gene assay in HAECs treated with ponatinib (150 nM, 24 h), and presented as ratio of firefly/renilla luciferase activity. As indicated, some cells were also transfected with CA-MEK5α. Error bars represent mean ± SEM. Data set contains 10-12 technical replicates. Statistical significance was assessed using student's t-test (two-tailed). *P < 0.05 vs. Veh control. (G-H) qRT-PCR analysis of relative klf4 and icam1 expression in HAECs treated with ponatinib (150 nM, 24 h) in the presence of laminar flow. Data set contains 5-6 technical replicates. Error bars represent mean ± SEM. Statistical significance was assessed using student's t-test (two-tailed). *P < 0.05; **P < 0.01; vs. Veh control. (I) Expression of phospho-ERK5 at T218/Y220 (TEY), S496, and total ERK5 in HAECs treated with ponatinib (150 nM, 30 min) was assessed by Protein simple WES system (capillary electrophoresis western analysis). Protein bands are shown as pseudoblots.
can lead to disruption of the EC barrier that results in vascular leakage (64,66). Using electric cell-substrate impedance sensing (ECIS) system, we assessed the effect of ponatinib on EC barrier function by measuring transendothelial electrical resistance (TEER) of cell monolayers. TEER values revealed increased EC permeability in ponatinib treated cells (Figures 1F,G). Taken together, our results suggest that ponatinib induces apoptosis along with and inflammatory response in ECs.

Ponatinib Inhibits ERK5 Transcriptional Activity
Endothelial ERK5 plays a crucial role in vascular homeostasis, and its reduction leads to an accelerated inflammatory response in ECs (52). We asked if ponatinib triggers an endothelial inflammatory response via reducing ERK5 function. Employing a mammalian-one-hybrid assay, we noted decreased ERK5 transcriptional activity in ponatinib treated cells (Figure 2A). Similarly, expression of ERK5 responsive genes, including klf2/4 and eNOS, was also inhibited (Figures 2B-D). In HAECs over-expressing a constitutive active form of MEK5α (CAMEK5α), ERK5 transcriptional activity ( Figure 2E, bar 2 from the left) and KLF2 promoter activity ( Figure 2F, bar 2 from the left) were activated, both of which were inhibited upon ponatinib treatment (Figures 2E,F, bar 3). Ponatinib also inhibited the increase of klf4 ( Figure 2G) and decrease of icam1 expression (Figure 2H) induced by laminar flow (43). These results indicate that ponatinib reduces ERK5 transcriptional activity. Since ERK5 transcriptional activity is regulated by activation of its kinase domain, we studied the effect of ponatinib on ERK5 T218/Y220 phosphorylation (pERK5-TEY). We found that ponatinib did not affect laminar-flow induced ERK5 T218/Y220 phosphorylation (Figure 2I, first lane) suggesting that ponatinib-mediated reduced ERK5 transcriptional activity is independent of ERK5 kinase activity. Interestingly, we found that after ponatinib treatment, ERK5 S496 phosphorylation was significantly increased, even in the presence of protective laminar flow (Figure 2I, second lane). Because ERK5 S496 phosphorylation promotes an inflammatory response in EC (52), ponatinibmediated phosphorylation at this residue might play a crucial role in ponatinib-associated endothelial inflammatory response.

Ponatinib Increases ERK5 SUMOylation in ECs
ERK5 SUMOylation regulates endothelial inflammatory response via repressing laminar flow-mediated ERK5 transcriptional activation (60). We tested if ponatinib inhibits ERK5 transcriptional activity by promoting ERK5 SUMOylation. Following ponatinib treatment, we collected cell lysates for IP studies to determine the level SUMOylated ERK5 in ECs. We found that ERK5 SUMOylation was significantly increased in both HUVECs (Figure 3A,B) and HAECs (Figures 3C,D) treated with ponatinib, and that this increase was reversed in cells  overexpressing deSUMOylation enzyme Sentrin/SUMO-specific protease 2 (SENP2) (Figures 3C,D).

Ponatinib Elicits Endothelial Inflammatory Responses via Promoting ERK5 SUMOylation
To verify the involvement of ERK5 SUMOylation in ponatinibassociated endothelial inflammatory response, we transduced HAECs with either an adenovirus expressing ERK5 wild type (Ad-ERK5-WT) or ERK5 non-SUMOylatable mutant (Ad-K6/K22R). The cells were then treated with ponatinib for 24 h, and expression of various genes were determined. qRT-PCR analysis revealed that the increased tnfα, vcam1, and icam-1 as well as the decreased klf2/4 and enos expression by ponatinib seen in ECs expressing ERK5-WT was reversed in ECs expressing the ERK5 K6/22R mutant (Figure 4A-F). It is possible that ERK5-WT over-expression might skew the involvement of ERK5 SUMOylation in ponatinib-mediated endothelial inflammatory response. Thus, in an independent experiment, we used SENP2 overexpression to inhibit endogenous ERK5 SUMOylation (Figures 3C, 4G), and found that the reduced ERK5 SUMOylation by SENP2 could partially reverse ponatinib's effect on klf2 and vcam1 expression (Figures 4H,I).
A functional characteristic of ECs is their ability to migrate. During physiological processes, EC migrate during vasculogenesis and angiogenesis whereas in pathological process, such as vessel damage, EC migrate to restore vessel integrity (67). To investigate the role of ponatinib-mediated ERK5 SUMOylation in EC function, we performed an in vitro scratch wound healing assay. Of note, we minimized the contribution of cell proliferation in this process by: (1) wounding a confluent and quiescent monolayer of cells and (2) maintaining the cells in reduced-serum culture medium for the duration of the assay. In ECs expressing ERK5-WT, cell migration toward the wounded region seen in the veh-treated group was inhibited by ponatinib treatment (Figure 5, left panel). However, in ECs expressing ERK5-K6/22R mutant, ponatinib-mediated delayed cell migration was rescued (Figure 5, right panel). Taken together, our data suggests the importance of endothelial ERK5 SUMOylation in ponatinib-associated endothelial inflammatory response, and, to a greater extent, vascular adverse events.

DISCUSSION
Previously, we reported that p90RSK activation increases SENP2 T368 phosphorylation that inhibits SENP2 deSUMOylation activity, leading to increased ERK5 SUMOylation. Additionally, p90RSK activation increases ERK5 S496 phosphorylation. Both, SUMOylation and S496 phosphorylation reduce ERK5 transcriptional activity that accelerates EC inflammation, dysfunction, apoptosis, and subsequent atherosclerotic plaque formation (52,61). In the current study, we identified a novel role for ERK5 SUMOylation in ponatinib-mediated endothelial inflammatory response. Interestingly, we also detected increased p90RSK phosphorylation by ponatinib (Supplementary Figure 1A). This signaling pathway elicited by ponatinib resembles that of ECs exposed to atheroprone stimuli, such as disturbed flow, reactive oxygen species, or advanced glycation end products (52,61), suggesting that ponatinib-mediated ERK5 SUMOylation might be involved in ponatinib-associated atherosclerosis, VAEs and subsequent cardiovascular complications. Thus, inhibition of endothelial ERK5 SUMOylation can be viewed as a novel approach to mitigate VAEs resulting from ponatinib treatment.
Because we also found decreased SENP2 expression in ECs expressing SENP2 treated with ponatinib ( Figure 4G), we speculate that ponatinib elicits an endothelial inflammatory response not only by reducing SENP2 activity but also by reducing SENP2 at the protein expression level. This might be a unique feature of ponatinib, compared to other proinflammatory stimuli where effects on SENP2 are only on its enzymatic activity.
Endothelial ERK5 can be phosphorylated at multiple sites, each of which confers different biological functions (51,52,(68)(69)(70). Among them, ERK5 S496 phosphorylation plays a crucial role in EC inflammation. ERK5 S496 phosphorylation has a similar effect to that of ERK5 SUMOylation in inhibiting ERK5 transcriptional activity (52). We reported previously that ERK5 phosphorylation at S496 is induced not only by disturbed flow and reactive oxygen species (52,61) but also by radiation (IR) and that it plays a crucial role in IR-mediated EC inflammation (69). In the current study, we found that ERK5 S496 phosphorylation is increased by ponatinib (Figure 2I and Supplementary Figure 1A). Furthermore, ponatinibmediated increased tnfα expression was reversed in ECs expressing ERK5 S496A phosphorylation resistant mutant (Supplementary Figure 1B). Similarly, flow cytometric analysis of Annexin V staining revealed the contribution of ERK5 S496 phosphorylation in ponatinib-induced EC apoptosis (Supplementary Figure 1C).
It is noteworthy that the reduction of ERK5 transcriptional activity by ponatinib via ERK5 SUMOylation and, probably, S496 phosphorylation occurs independently of kinase activation, highlighting the predominance of these posttranslational modifications on ERK5 function and, subsequently, EC integrity. If and how ponatinib-induced ERK5 SUMOylation and S496 phosphorylation interact and/or coordinate to control ERK5 transcriptional activity requires further investigation. To the best of our knowledge, this is the first study to demonstrate the role of endothelial ERK5 SUMOylation in ponatinib-regulated EC inflammation.