Autophagy is regulated by a complex network of molecular switches, among which mammalian target of rapamycin (mTOR) is a central player (Yang et al., 2011). mTOR exists in two distinct complexes, mTORC1 and mTORC2. Unlike the relatively unexplored role of mTORC2 in cancer biology, mTORC1 is a nutrient sensitive sensor that orchestrates cell metabolism, cell cycle progression and autophagy (Yang et al., 2011). mTORC1 represses the latter at both initiation and degradation stages via inhibiting ULK1 complex and lysosomal function respectively (Elgendy et al., 2014; Puertollano, 2014). Furthermore, antiapoptotic B-cell lymphoma 2 (Bcl-2) family members such as Bcl-2, myeloid cell leukemia 1 (Mcl-1), and B-cell lymphoma-X large (Bcl-xL) - through their direct and indirect interactions with Beclin-1/Atg6 - have emerged as autophagy regulators which mediate the regulatory crosstalks between apoptosis and autophagy (Elgendy et al., 2014).

Intriguingly, in our study, cancer cells responded to subcytotoxic sunitinib doses by activating mTORC1 and increasing Mcl-1 protein levels (Elgendy et al., 2016) (Figure 1). Importantly, increased mTORC1 activity and Mcl-1 level represented a pro-survival cellular response to the mild to moderate stress triggered by these sunitinib doses since inhibiting mTORC1 or depleting Mcl-1 sensitized cancer cells to tolerated doses (Elgendy et al., 2016). Notably, mTORC1 activity and Mcl-1 levels were found to be higher in sunitinib-resistant compared to parental melanoma cells further confirming their role in mediating sunitinib resistance (Elgendy et al., 2016). Finally, analysis of pNET and RCC samples obtained from sunitinib-resistant patients showed a significant correlation between post-sunitinib increase in Mcl-1 levels and mTORC1 activity and resistance to treatment with sunitinib (Elgendy et al., 2016). In line with our observations, Makhov et al. has shown that deletion of PTEN (phosphatase and a tensin homolog deleted from chromosome 10), a negative regulator of PI3K/AKT/mTOR signaling, correlated with sunitinib resistance in renal and prostate cancer (Makhov et al., 2012). Conversely, cytotoxic doses dramatically diminished mTORC1 activity and Mcl-1 levels. In alignment with our findings, sunitinib was active against acute myeloid leukemias (AML) possessing activating PDGFR, FLT-3, and c-kit mutations through inhibiting mTOR and reducing Mcl-1 levels (Ikezoe et al., 2006; Teng et al., 2013).

Interestingly, in addition to genetic tools, our results showed that pharmacological mTOR and/or Mcl-1 inhibition using rapamycin and sorafenib respectively sensitized cancer cells to primarily tolerated doses of sunitinib (Elgendy et al., 2016). In accordance with our findings, Lin and colleagues demonstrated that while silencing Atg5 abrogated sunitinib cytotoxicity, everolimus and trehalose, mTOR-dependent and independent autophagy activators respectively, boosted the antiproliferative activity of sunitinib in medullary thyroid cancer cells (Lin et al., 2012). Furthermore, Atg5 silencing antagonized everolimus- and trehalose-triggered increase in sunitinib efficacy (Lin et al., 2012). Additionally, our data provide mechanistic rationale for the previously reported synergy between sunitinib and mTOR inhibitors identified through unbiased binary screening (Li et al., 2014).

Deeper mechanistic analysis showed that the increase in Mcl-1 protein levels upon treatment with tolerated doses of sunitinib was associated with decreased Mcl-1 proteasomal degradation and enhanced stability, rather than modulation of gene expression (Elgendy et al., 2016). One distinct feature of Mcl-1 among other antiapoptotic Bcl-2 family members is its short half-life. This has been linked to its tight regulation by complex mechanisms on several levels including phosphorylation in its unique PEST region at Thr163 and Ser159 residues by extracellular signal regulated kinase (ERK) and glycogen synthase kinase 3β (GSK3β) respectively which in turn, diminishes and enhances the rate of Mcl-1 proteasomal degradation (Domina et al., 2004; Yan et al., 2017). Enhanced Mcl-1 stability in response to lower doses of sunitinib was the net result of both ERK activation and GSK3β inhibition (Elgendy et al., 2016). Conversely, higher “cytotoxic” doses led to opposite effects where ERK was inhibited and GSK3β was activated, resulting in Mcl-1 destabilization (Elgendy et al., 2016). Furthermore, dual modulation of GSK3β activity by lower and higher doses of sunitinib also mediated the differential effect on mTORC1 activity (Figure 1).

Induced AXL and MET signaling which in turn increased ERK and GSK3β phosphorylation has been linked to sunitinib resistance in RCC (Zhou et al., 2016). Intriguingly, Qu et al. have lately identified a novel long non-coding RNA called lncARSR (lncRNA Activated in RCC with Sunitinib Resistance) which promoted sunitinib resistance via acting as a competing endogenous RNA. Indeed, IncARSR through sequestering miR-34 and miR-449 upregulated AXL/c-MET expression and activated ERK, STAT3 and AKT signaling. In a positive regulatory feedback loop, activated AKT boosted IncARSR expression. Strikingly, sunitinib-resistant RCC cells secreted IncARSR via exosomes which reached out to sunitinib sensitive cells and transformed them into resistant ones (Rna et al., 2016). Sunitinib-induced ERK activation secondary to increased EGFR phosphorylation was also linked to sunitinib resistance mediated by epithelial-to-mesenchymal transition (EMT) in mRCC cells (Mizumoto et al., 2015). In addition, sphingosine kinase-1 (SK1)-mediated ERK activation was triggered in sunitinib-resistant RCC cell lines which were resensitized to sunitinib using SK1 and ERK pharmacological inhibitors (Gao and Deng, 2014). Ras/Raf activating mutations and subsequent constitutive MEK/ERK activation has also been speculated to contribute to sunitinib resistance in thyroid carcinoma cells (Piscazzi et al., 2012). Moreover, sunitinib-induced apoptosis in medulloblastoma was associated with GSK3β and mTOR dephosphorylation (Yang et al., 2010).

mRCC resistance to sunitinib was reported to be secondary to impedance of autophagy at its termination stage. Sunitinib neutralized the acidic lysosomal pH and hence, inactivated one of the major lysosome-associated proteases, cathepsin B (Giuliano et al., 2015). In light of the observations illustrating the tightly regulated crosstalk between lysosomes and mTORC1 where active mTORC1 negatively impacted lysosomal biogenesis, function and autophagosomes-lysosomes fusion (Puertollano, 2014), we postulate that subcytotoxic sunitinib doses-induced mTOR activation could also contribute to lysosomal dysfunction associated with the “chemoresistance phenotype.” Accordingly, it has been proposed that increasing sunitinib availability at its target TKRs via preventing its lysosomal trapping using agents which increase lysosomal membrane permeability could resensitize mRCC to sunitinib.

The role of multidrug resistance proteins has been linked to chemoresistance (Giuliano et al., 2015). Indeed, sunitinib increased ABCB1 expression favoring further lysosomal accumulation of sunitinib and its cellular efflux. Inhibiting this transporter by elacridar restored sunitinib senstivity in resistant mRCC (Giuliano et al., 2015; Sato et al., 2015). Sunitinib also increased ABCG2 levels in RCC cells (Sato et al., 2015) and glioma xenografts (Shojaei et al., 2010). Such increment is thought to compensate for the inhibitory effects of sunitinib on ABCG2 function (Dai et al., 2009; Shukla et al., 2009). While targeting ABC transporters might be useful in reversing sunitinib resistance, attention should be directed toward the bioavailability of co-administered drugs (Shukla et al., 2009; Sato et al., 2015). Interestingly, ABCG2 has also been shown to modulate autophagy (Ding et al., 2016). Compared to parental cells, ABCG2 overexpression—which was associated with boosted autophagy—rendered cancer cells more rendered cancer cells more resistant to stress inducers. Consistently, knocking down ABCG2 inhibited autophagy (Ding et al., 2016). Yet, it is still unknown whether modulating ABC transporters function/expression by sunitinib mediates its regulatory effects on autophagy and cancer cell sensitivity/resistance to sunitinib.

As tumors progress, their oxygen and nutrients demands substantially outweigh their supply. To address these needs, “angiogenesis” is switched on generating tumor-associated neovasculature. Within this context, sunitinib emerged as an effective antiangiogenic agent via inhibiting VEGFR, PDGFR, and c-KIT, which was then indicated for treating tumors with high angiogenic burden as mRCC. Although an interrupted sunitinib treatment schedule of 4-weeks ON/2-weeks OFF was adopted to reduce its side effects, rapid tumor regrowth and metastasis considerably ensued during the 2-week break period (Ebos et al., 2009; Griffioen et al., 2012). While the underlying mechanisms remain elusive, microenvironmental changes are speculated to condition/acclimatize body organs to be more permissive to tumor extravasation by acting as “premetastatic niche” (Loges et al., 2009). Anti-angiogenic therapy—via triggering hypoxia—might induce autophagy in both tumor cells and microenvironment. While activating autophagy in cancer cells might retard their proliferation, stromal cancer-associated fibroblasts autodigest themselves into basic nutrients for epithelial cancer cells. This host-parasitic like paradigm is known as “Battery-Operated Tumor Growth” or “The Autophagic Tumor Stroma Model of Cancer Cell Metabolism” (Martinez-outschoorn et al., 2010). Cancer-associated cachexia is also hypothesized to start as local stromal autophagy, and then disseminates systemically. Consistently, increased metabolic rate of cachexic cancer patients is restored to their normal levels following surgical tumor excision (Martinez-outschoorn et al., 2010). Furthermore, sunitinib resistance is linked to persistence of intratumor myeloid derived suppressor cells (MDSCs) with dominating subset of neutrophilic population that produces proangiogenic MMP9, MMP8, and IL-8 (Finke et al., 2011). Activation of proangiogenic hepatocyte growth factor/c-Met signaling pathway also contributed to sunitinib resistance which was reversed using c-Met inhibitor (Shojaei et al., 2010). Notably, inhibiting c-Met signaling induced autophagic cell death in glioblastoma (Liu et al., 2013). Collectively, these divergent effects of antiangiogenics on primary tumor and its microenvironment necessitates elemental consideration of prominent issues while administering anti-VEGF therapy including; relative benefit-to-risk of “continuous vs. intermittent” treatment schedules, optimal dose, duration of treatment and tumor stage. This also emphasizes the importance of combination therapy as a possible approach to abrogate resistance to anti-VEGF therapy.

Mitophagy (or selective autophagic degradation of mitochondria) has been linked to tumorigenesis (Hjelmeland and Zhang, 2016). One of the key mitophagy mediators, Parkin, putative tumor suppressor gene, translocates to the mitochondria secondary to loss of mitochondrial membrane potential, ubiquitinating mitochondrial proteins and recruiting p62-LC3 and autophagosomes to the mitochondria (Hjelmeland and Zhang, 2016). Melatonin synergistically sensitized human heptocellular carcinoma cells to sorafenib through its pro-oxidant capacity and activating mitophagy (Prieto-domínguez et al., 2016). In fact, oxidative stress and hypoxia are fundamental autophagy inducers (Kulikov et al., 2017). Hypoxia - via hypoxia inducible factor 1α (HIF1α)- induced the expression of mitophagy receptors; BCL2/adenovirus E1B 19 kDa interacting protein 3 (BNIP3) and its homolog NIX (Kulikov et al., 2017). Though sunitinib blocked HIF1α translation and hence, repressed the expression of its downstream target genes in colorectal and renal carcinoma cells (Shin et al., 2010). Severe mitochondrial structural abnormalities was also reported in the heart of a patient with sunitinib-induced heart failure (Kerkela et al., 2008). This has been linked to its off-target inhibitory effect on AMPK (Kerkela et al., 2008). Sunitinib also triggered mitochondrial damage, cytochrome C release, caspase 9 activation and apoptotic cell death in cardiomyocytes in vitro and in vivo (Chu et al., 2007). Sunitinib increased both death receptor and mitochondrial-dependent apoptosis in AML cells (Teng et al., 2013). Nonetheless, it still remains to be elucidated whether sunitinib modulates mitophagy and therapeutic intervention with mitophagy could sensitize cancer cells to sunitinib.

Dysfunctional autophagy has been connected to increased genomic instability. Intriguingly, sunitinib-induced increased autophagic flux concurred with increased micronuclei, diagnostic marker of nuclear instability, in human RCC (Yan et al., 2017). Nuclear LC3, phosphorylated Ulk1 and p62 interacted with Rad51 or PARP-1, which are both engaged in maintaining genomic stability (Yan et al., 2017). Sunitinib was incapable of accumulating micronuclei in p62/LC3-depleted cells. Depleting Rad51/PARP-1 abolished sunitinib-induced autophagy (Yan et al., 2017). Since p62 acts as the connecting bridge between ubiquitin and autophagic machineries, both systems are speculated to coordinate genomic stability. Despite being a marker of DNA damage, γ-H2AX actively participates in DNA repair. γ-H2AX and PARP-1/Rad51 interaction was disrupted following p62 depletion (Yan et al., 2017). Although sunitinib elevated γ-H2AX level, it decreased Rad51 expression which is essential for homologous recombination repair, Accordingly, while sunitinib induced acute DNA damage may lead to cancer cell death, it might also trigger non-homologous end joint DNA repair mechanisms. Collectively, these findings proposed a mechanistic link between the modulatory effects of sunitinib on autophagy and nuclear instability.

Canonical clinicopathological evaluation is unable to predict the therapeutic response to sunitinib-treated cancer patients. Identification of novel molecular prognostic markers is therefore urgently needed. Based on the present findings, immunohistochemical assessment of ribosomal S6 phosphorylation (as readout of mTORC1 activity) and Mcl-1 protein levels could serve as markers that predict sunitinib response. Additionally, elevated IncARSR levels in pre-treatment RCC patients significantly correlated with poor sunitinib response. In contrast, low IncARSR levels conferred improved progression-free survival and favorable prognosis following sunitinib therapy (Rna et al., 2016). Notably, IncARSR levels were remarkably increased in patients who regressed and relapsed post-sunitinib therapy compared with pre-therapy levels. Hence, IncARSR is proposed as an independent predictor for sunitinib response in RCC patients (Rna et al., 2016).

Additionally, the present findings provide a rationale for the lack of cytotoxic effects of clinically relevant doses of sunitinib, and suggest novel strategies -in addition to its anti-angiogenic effects- to directly induce tumor cell death, and overcome sunitinib resistance;

(i) Ideally, though not easily achievable in clinical practice, tailoring sunitinib dose per each patient based on their response should select patients that need escalation of sunitinib dose to reach cytotoxic effects at tolerable doses. Rovithi et al. showed that an alternating schedule of high sunitinib efficiently impaired tumor growth in vivo and maintained significantly higher plasma and intratumoral sunitinib levels compared to the standard, daily regimen (Rovithi et al., 2016). Accordingly a phase I clinical trial (NCT02058901) was initiated to investigate the safety, tolerability and efficacy of intermittent, high dose sunitinib schedules (300 mg sunitinib, administered once weekly) in patients with advanced solid tumors, with preliminary indications of prolonged disease stabilization and tolerability in a heavily pretreated group of patients (Rovithi et al., 2016).

(ii) Alternatively, pharmacological targeting of Mcl-1 and mTOR presents a promising strategy in combating/reversing sunitinib resistance. It is worth mentioning that FDA has already approved sequential treatment of sunitinib-resistant mRCC patients with everolimus, mTOR inhibitor. In addition, a phase I clinical trial aiming at determining the highest tolerable dose of “sunitinib and temsirolimus (mTOR inhibitor)” combination that could be administered to mRCC patients has lately been completed. There are also parallel clinical trials testing the efficacy of sorafenib in treating sunitinib-resistant mRCC and GIST patients. Yet, the clinical outcome of combining sunitinib and Mcl-1 and/or mTOR inhibitor in sunitinib-resistant patients remains to be investigated. Consistent with the emerging role of MET and AXL signaling in mediating sunitinib resistance, combining a small molecule inhibitior of both MET and AXL, BMS777607, with sunitinib both suppressed the growth of sunitinib-resistant RCC xenografts and prevented the emergence of sunitinib resistance (Rna et al., 2016). In support with this notion, a randomized Phase III clinical trial has further confirmed the clinical value of cabozantinib, a multityrosine kinase inhibitor targeting MET, VEGFR, and AXL in RCC patients who had progressed following VEGFR inhibitor therapy including sunitinib (Géczi et al., 2015).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.