Introduction

Over the past 10 years, the discovery of molecular alterations susceptible to targeted inhibition has significantly improved overall survival (OS) in a small group of non-small cell lung cancer (NSCLC) patients [1–5]. However, subgroups of oncogenic drivers that are targetable are not common [6], and overall treatment outcomes of NSCLC patients remain unsatisfactory, with low long-term survival rates.

Direct inhibition of Kirsten Rat Sarcoma Viral oncogene homolog (KRAS), the most commonly mutated oncogene in lung adenocarcinoma, has proven clinically challenging [7]. The KRAS gene is located in the short arm of chromosome 12 and encodes for two splice variants, KRAS4A and KRAS4B [8]. Mutated KRAS has a putative role in NSCLC, most frequently observed in tumors arising in smokers with adenocarcinoma histology [9, 10]. RAS mediates the intracellular signaling pathway in response to activation of cell surface receptors. Once activated, a tyrosine kinase receptor binds to an adaptor protein, Grb2, which recruits the guanine nucleotide exchange factor SOS. SOS facilitates the RAS GDP-GTP exchange, leading to RAS activation, which in turns stimulates the RAF-MEK-ERK mitogen-activated protein kinase (MAPK), the phosphatidylinositol 3-kinase (PI3K)-AKT-mTOR and the RAS-like (RAL) pathways (Fig. 1). RAS signaling is switched off by GTPase activating proteins (GAPs) responsible for GTP hydrolysis [11]. KRAS mutations, which impair GTP hydrolysis, are located in the sites involved in GAP binding, including codons 12, 13 and 61, with the amino acid substitution G12C being the most common [7, 9–11].

Fig. 1
figure 1

Signal transduction pathways in KRAS mutant NSCLC and LSCC. RAS is the downstream effector of the activated tyrosine kinase receptor EGFR. SOS favors the GDP-GTP exchange, with the subsequent activation of three intracellular signaling pathways: (1) CRAF-MEK-ERK; (2) PI3K-AKT-mTOR; (3) RAL. Mutations in RAS determine its constitutive activation. ERK and AKT signaling are also stimulated by the serine/threonine kinase PAK1, which is regulated by the small GTPases RAC1 (upon PKCiota) and Cdc42. PAK1 phosphorylates CRAF at Ser338, which translocates into mitochondria and inhibits BAD, suppressing apoptosis, and MEK1 at Ser298, with the subsequent expression of cyclin D1, a key driver for cell cycle progression. PAK1 recruits AKT to plasma membrane, allowing its activation, which stimulates mTOR. mTOR is also augmented by the serine/threonine kinase LKB1, which activates AMPK, responsible for RHEB inhibition, promoting mTOR activity. LKB1 is an additional indirect regulator of the transcriptional co-activator YAP1, which induces transcription of BCL-xL, CTGF, CYR61, COX2, MMP7, IL-6, IL-1α and AXL through binding to TEAD transcription factors. LKB1 phosphorylates LATS at Thr1079, which is responsible for YAP1 phosphorylation at Ser127 and Ser381, favoring its retention into the cytoplasm and its proteasomal degradation. LATS is phosphorylated by MST1 upon MARK activation. YAP1 is also activated by the Src family kinase Yes, or Jun N terminal kinases (JUNK), and inhibited by NF2, which phosphorylates YAP1, preventing its translocation into the nucleus. NF2 additionally inhibits RAC1, blocking CRAF and MEK phosphorylation mediated by PAK1. Phosphorylation of Tyr705 of cytoplasmic STAT3 in response to activated EGFR promotes STAT3 homodimerization, which leads to nucleus translocation and DNA binding. IL-6 by JAK2 phosphorylation favors STAT3 recruitment. ERK phosphorylates STAT3 on Ser727 which subsequently interacts with GRIM-19. This interaction favors STAT3 transportation and anchorage to the inner mitochondrial membrane. Shh indirectly promotes STAT3 activation, as it induces the binding between Gli1 and IL-6 promoter. Shh is regulated by PTCH and SMO. Upon Hh ligand binding to PTCH, SMO initiates a signaling cascade that culminates in the activation of GLI transcription factors. In the absence of ligands, PATCH represses SMO activity

Full size image

Many efforts have been made to find treatment strategies able to inhibit downstream effectors of KRAS. Recently, a phase II study of combination docetaxel plus MEK inhibitor selumetinib compared with docetaxel and placebo in 87 NSCLC patients, who had previously failed with first line therapy, showed a statistically significant progression-free survival (PFS) and objective response rate (ORR) improvement in favor of selumetinib, with no overall survival (OS) advantage [12]. The heterogeneous biology of KRAS mutant NSCLC tumors may partially explain the difficulties encountered in the development of efficient therapies. It is still a matter of debate whether and how the distinct oncogenic RAS mutations may affect the biological and clinical behavior of KRAS mutant patients. The hydrophobic G12C and G12V activate the RAL pathway, while the hydrophilic G12D acts through PI3K-AKT signaling [13]. Defining whether cells are KRAS dependent or independent for tumor growth is also of great relevance [14]. Co-occuring genetic alterations in genes other than KRAS may promote cancer cell proliferation and survival and contribute to the development of escape mechanisms to MEK inhibition [15].

By contrast, progress in the field of lung squamous cell carcinoma (LSCC), representing about 30% of NSCLC cases, has lagged behind. Although molecular alterations in LSCC have been described, effective targeted therapies have not yet been developed [16]. Standard treatment for LSCC is currently based on platinum doublets in the first line, and monochemotherapy or immunotherapy in subsequent lines [17, 18]. Different potentially targetable molecular alterations have been identified in LSCC tumors, including phosphoinositide 3-kinase (PIK3CA), fibroblast growth factor receptor 1 (FGFR1), or c-MET amplification, and discoidin domain receptor tyrosine kinase 2 (DDR2) mutations, although none of these biomarkers have yet been validated in the clinical setting [19]. The epidermal growth factor receptor (EGFR) gene is commonly overexpressed in patients with LSCC [20], and two monoclonal anti-EGFR antibodies, cetuximab and necitumumab, have been tested in phase III studies. The FLEX trial demonstrated superior OS for chemotherapy plus cetuximab compared to chemotherapy alone for patients with advanced EGFR-expressing NSCLC, with the greatest survival benefit observed in the subgroup of patients with squamous histology [21]. A non-preplanned analysis, performed to build an immunohistochemistry score for EGFR expression, confirmed that high EGFR levels are more commonly detected in LSCC than in other types of NSCLC [22]. It was not possible to fully validate these results in a second phase III trial (BMS099), in which cetuximab showed no benefit in OS, progression-free survival (PFS) or response in combination with first line carboplatin and docetaxel [23]. In the SQUIRE trial, necitumumab plus cisplatin and gemcitabine was compared to cisplatin and gemcitabine alone for advanced LSCC. The study met its primary endpoint, showing a statistically significant benefit in OS with necitumumab. Moreover, a preplanned analysis showed that patients with EGFR overexpressing tumors benefited most from necitumumab [24]. Based on these results, in November 2015, the Food and Drug Administration (FDA) in the US approved necitumumab in combination with cisplatin and gemcitabine for first line treatment of patients with metastatic LSCC. However, acquired resistance eventually develops, while the lack of clearly defined predictive biomarkers to optimize patient selection remains the main limitation for the use of necitumumab.

Here, we describe the intracellular signaling pathways that may have an impact on the development of KRAS mutant NSCLC and LSCC tumors, and can represent druggable targets to be combined with selumetinib or necitumumab, and define synthetic lethal approaches. We focus on common signaling pathways activated in both KRAS mutant and LSCC tumors, which dictate distinct pathologic outputs and favor the occurrence of two distinct pathologic entities. Since tumor cells use common signaling pathways for growth, proliferation and migration, information regarding tumors other than lung is also included. Common alterations previously described in LSCC, such as those affecting members of the PI3K pathway, or amplification of fibroblast growth factor receptor 1 or mutations in the discoidin domain receptor 2 kinase gene are not the main objective of this review [25]. This article is based on previously conducted studies and does not involve any new studies of human or animal subjects performed by any of the authors.

Revisiting STAT3 Signaling in Kras Mutant NSCLC and LSCC

Deregulation of signal transducer and activator of transcription (STAT3) has been observed in different tumors [26]. Phosphorylation at tyrosine (Tyr) 705, in response to growth factors and cytokines, is required to activate STAT3. Upon Janus kinase (JAK) activation, growth factor receptors or cytokine receptors create docking sites to recruit cytoplasmic STAT3 which dimerizes, translocates into the nucleus and promotes the transcription of genes involved in cell cycle progression and apoptosis (Fig. 1) [27].

Preclinical data showed that phosphorylated STAT3 increases following MEK inhibition in KRAS mutant cells [28]. STAT3 contributes to drug resistance through activation of FGFR and JAK. Combinatorial treatment with the dual JAK1 and FGFR inhibitor ponatinib and the MEK inhibitor selumetinib suppressed STAT3 phosphorylation in vitro [28]. Similar results were observed in vivo, in which concomitant treatment with either ponatinib and GDC0973 (MEK inhibitor), or ruxolitinib (JAK1/2 inhibitor) with GDC0973, or the triple combination of ruxolitinib, ponatinib and GDC0973, significantly reduced tumor growth [28]. By exploring the pathogenic contribution of STAT3 in KRAS mutant tumors, Brooks et al. found that augmented levels of only interleukin 6 (IL-6) and its essential trans-signaling receptor subunit, sIL-6R, are observed in lung adenocarcinoma patients, and are also associated with exacerbated KRAS G12D-driven lung carcinogenesis in mice [29].

Another emerging key role of STAT3 is related to its mitochondrial activation upon serine (Ser) 727 phosphorylation, which is critical in inducing KRAS oncogene transformation [30]. Extracellular signal-regulated kinase (ERK) is responsible for Ser727 phosphorylation (Fig. 1) [31]. Once activated, mitochondrial STAT3 stimulates the mitochondrial electron transport chain and regulates tumor cell metabolism [32]. Gene-associated retinoid-IFN-induced mortality-19 (GRIM-19) interacts with Ser727 phosphorylated STAT3 and favors its anchorage to the inner mitochondrial membrane [33]. Mitochondrial membrane potential is essential to maintain KRAS-transformed cells. Loss of phosphorylated Ser727 STAT3 is associated with reduced activity of succinate oxidoreductase and ATP synthase [30]. Lack of STAT3 decreases ATP cellular production and colony formation is largely abolished, even in the presence of the RAF-MEK-ERK pathway [30].

STAT3 represents an appealing target in LSCC, particularly due to important interactions with the EGFR signaling pathway [34]. STAT3 inhibition by cucurbitacin I (JSI-124) has been shown to overcome resistance to chemotherapy and radiotherapy in vivo [35]. In tumor models with intrinsic or acquired resistance to EGFR tyrosine kinase inhibitors or cetuximab, STAT3 suppression restored the antitumor effects [36]. This suggests that targeting STAT3 may be an efficient therapeutic strategy in NSCLC patients that are not sensitive to EGFR inhibition. Similar results were observed in head and neck squamous cancer cell (HNSCC) tumors recurring after cetuximab treatment, in which increased STAT3 phosphorylation was detected compared with pretreatment biopsies [37]. Moreover, in a pilot study that enrolled patients with NSCLC receiving gefitinib before tumor surgical resection, increased STAT3 phosphorylation was observed in resected tumor tissues [38].

STAT3 blockade may abrogate therapeutic resistance to EGFR and MEK inhibitors. Several STAT3 or JAK inhibitors have been developed, but few have been entered into clinical trials due to either unfavorable chemical properties or lack of biologic activity. Table 1 summarizes some of the STAT3 inhibitors in clinical development, while an extended list of STAT3 inhibitors including repurposing drugs is reviewed by Zhao et al. [26]. OPB-51602 and OPB-31121 are oral compounds, recently investigated in phase I clinical trials, that are thought to inhibit phosphorylation of both STAT3 Tyr705 and Ser727 [39, 40]. AZD9150 is an oligonucleotide antisense molecule (ASO), designed to target the 3′-untranslated part of STAT3, preventing protein translation [41]. AZD9150 showed antitumor activity in lymphoma and lung cancer models [41]. BBI608 is an orally administered drug that blocks cancer stem cells self-renewal and inhibits STAT3, β-catenin and Nanog pathways. In a phase I study, BBI608 demonstrated tolerability as well as signs of anti-cancer activity in patients with solid tumors [42]. Finally, eriocalyxin B and evodiamine are STAT3 inhibitors which have not yet reached clinical development stages [43, 44]. Eriocalyxin B, a plant-derived diterpenoid, specifically inhibits phosphorylated Tyr705 but not Ser727. Based on computational modeling analysis, eriocalyxin B selectively interacts with the cysteine 712 located on STAT3 to form a covalent binding and inhibit STAT3 activity [43]. Evodiamine is an alkaloid compound that favors the transcription of SHP-1, a protein tyrosine phosphatase responsible for dephosphorylation of Tyr705, and reduces DNA binding activity and suppression of transcription of genes involved in cell cycle progression (cyclin D1), angiogenesis (vascular endothelial growth factor receptor), inflammation (tumor necrosis factor) and programmed cell death (survivin) [44] (Table 1).

Table 1 Drugs in clinical or preclinical development
Full size table

Pak Protein Kinases and Their Role in Cancer

The p21-activated kinase (PAK) family includes six serine/threonine protein kinases, classified into two groups, PAK1-3 (group I) and PAK 4-6 (group II), with PAK1 being the most extensively studied [45]. PAK1 may be deregulated in tumors through gene amplification of the locus 11q13, messenger RNA overexpression or activating mutations [46]. PAK1 can also be hyperactivated by molecular alterations of its upstream regulators. Inactive PAK1 forms an autoinhibited homodimer that is stabilized by interactions between its inhibitory switch domain and the kinase domain. The binding between the small GTPases cell division control protein 42 (Cdc42) or Ras-related C3 botulinum toxin substrate 1 (RAC1) with PAK1 favors the dimer dissociation, resulting in two PAK1 monomers that undergo autophosphorylation and activate downstream signaling [47]. Atypical protein kinase C isozyme iota (PKCιota) modulates RAC1 activation [48]. PAK1 regulates several important signaling pathways, including the RAF-MEK-ERK, PI3K-AKT and Wnt-β-catenin pathways (Fig. 1) [49].

PAK1 is crucial in KRAS mutant tumors. PAK1 activates ERK signaling through CRAF phosphorylation at Ser338 or MEK1 phosphorylation at Ser298 [50]. PAK1 also acts as a scaffold protein that recruits AKT to the plasma membrane allowing its activation (Fig. 1) [51]. In a transgenic mouse model of squamous skin carcinoma harboring KRAS G12D, the lack of PAK1 significantly delayed and reduced tumor development [52]. Biochemical analyses revealed a strong reduction of MEK-ERK and AKT activation in PAK1-negative tumors. However, the available data suggest that ERK has a more determining role in tumorigenesis than AKT, as observed after treatment with ERK and AKT inhibitors. In the same model, RAC1 was found to be a critical player, since the lack of RAC1 is associated with a decreased keratinocyte hyperproliferation and with MEK and AKT activation [53]. PAK1 also phosphorylates CRAF at Ser338. Phosphorylated CRAF translocates into mitochondria and phosphorylates BAD, which prevents the binding to B cell lymphoma-2 (BCL-2), and inhibits apoptosis (Fig. 1) [54].

Expression of PAK1 protein has also been analyzed on tissue microarrays of small cell lung cancer (SCLC), lung adenocarcinomas, LSCC and HNSCC. Sixty-four percent of LSCC samples were positive for PAK1 expression, with 52% of them having moderate or strong staining [55]. Nuclear localization of PAK1 was evident in a significant proportion of LSCC while lung adenocarcinomas and SCLCs expressed weak to moderate levels of PAK1 only in the cytoplasm [55]. Interestingly, epidermal growth factor has been found to stimulate PAK1 activity in HNSCC (Fig. 1) [56]. When PAK1 mRNA expression was examined in a distinct set of 54 laser-capture microdissected lung tissues, PAK1 mRNA expression was highest in LSCC compared with normal lung [55]. PAK1 inhibition in LSCC cell lines determines accumulation in the G1 phase. Levels of the E2F1 transcription factor, which regulates gene expression associated with DNA replication and mitosis and is essential for G1/S progression, were diminished after PAK1 ablation [55]. In LSCC xenograft models, PAK1 inhibition impaired tumor growth, as confirmed by a reduction of Ki-67 expression [55]. PAK1 inhibition alone did not increase apoptosis in NSCLC cells and xenograft models but displayed dramatically enhanced efficacy when combined with apoptosis proteins, EGFR, MEK1/2, or Src inhibitors [55].

Additional studies are warranted to better explore the role of PAK proteins in cancer. Several PAK inhibitors have been synthesized but are still in early stages of clinical development [57]. For instance, PF3758309 was investigated in a phase I study, but development was not continued due to unfavorable pharmacokinetic properties. FRAX-597 is a group I PAK inhibitor, able to prevent ATP binding, with in vivo antitumor activity in a KRAS model [58]. Interestingly, the racemic form of the anti-inflammatory ketorolac, R-ketorolac, beyond cyclooxygenase (COX1/2) inhibition, suppresses RAC1 and Cdc42 [59, 60]. Ivermectin, a well-tolerated antiparasitic macrocyclic lactone has also been found to block PAK1 in human ovarian cancer [61] (Table 1).

The Hippo Pathway as a New Target for Lung Cancer Treatment

The Hippo pathway is a complex signaling system, initially identified in Drosophila and responsible for organ size regulation [62]. In mammals, most components of the Hippo pathway have been implicated as tumor suppressors, including neurofibromin-2 (NF2; also known as merlin), the mammalian sterile 20-like (MST) kinases, the large tumor suppressor homologue (LATS) kinases and the adaptor proteins salvador homolog-1 (SAV1) and Mps one binder kinase activator (MOB). Downstream of LATS are two paralogous transcriptional coactivators: Yes-associated protein-1 (YAP1) and transcriptional coactivator with PDZ-binding motif (TAZ), which drive the expression of numerous effector genes [63]. YAP1 Ser127 phosphorylation by LATS 1/2 draws YAP1/TAZ in the cytoplasm, preventing it from activating TEA domain family member (TEAD)-mediated transcription of connective tissue growth factor (CTGF), AXL or other genes (Fig. 1) [64, 65]. YAP1 can also be activated upon tyrosine phosphorylation by the Src family kinase (SFK) Yes or Jun N terminal kinases (JUNK) (Fig. 1) [66].

YAP1 is a critical oncogenic KRAS effector and a promising therapeutic target for KRAS mutant tumors [67]. YAP1 induces transcription of the antiapoptotic factor BCL-xL and favors resistance to MEK inhibitors. YAP1 suppression in KRAS mutant NSCLC cell lines increases the efficacy of trametinib [67]. In a KRAS mutant pancreatic tumor mice model, YAP1 was found to promote expression of CTGF, cystein-rich angiogenic inducer 61 (CYR61), cyclooxygenase 2 (COX2), matrix metallopeptidase 7 (MMP7), IL-6 and interleukin 1α (IL-1α), and induce KRAS mutant pancreatic ductal cell proliferation [68]. ERK was found to be crucial for both phosphorylation and post-transcriptional modifications of YAP1, suggesting that KRAS may directly potentiate the transcriptional activity of YAP1, independent of the Hippo pathway (Fig. 1) [68]. The molecular mechanisms responsible for YAP1 modifications mediated by ERK are not fully characterized, even though multiple ERK phosphorylation motifs have been detected within the YAP1 sequence. To confirm the importance of YAP1 in NSCLC, YAP1 protein expression was measured in a group of NSCLC patients, and high YAP1 levels were observed in patients with KRAS mutant NSCLC [68].

The tumor suppressor liver kinase B1 (LKB1) is an indirect regulator of YAP1 [69]. LKB1 is a serine/threonine kinase, responsible for activation of adenosine monophosphate-activated protein kinase (AMPK) and AMPK related kinases (Fig. 1) [70]. LKB1 knockdown is associated with an increase of YAP1 target genes. LKB1 activates LATS through phosphorylation at threonine 1079 [69]. Indeed, LKB1 controls YAP1 through the regulation of microtubule-associated regulatory kinases (MARKs), a group of proteins implicated in cell polarity and microtubule dynamics that favor the activation of MST1/2 which phosphorylates LATS (Fig. 1) [69]. Furthermore, AMPK activation inhibits the GTP binding protein Ras homolog enriched in brain (RHEB), which suppresses the MAPK signaling and stimulates mTOR (Fig. 1) [71]. Dual inhibition of MEK and mTOR reduced proliferation in NSCLC cells harboring KRAS and LKB1 mutations. LKB1 inactivating mutations are detected in approximately 30% of patients with adenocarcinoma, and more frequently in patients with KRAS mutant NSCLC [71]. A retrospective analysis showed that the presence of concomitant KRAS and LKB1 mutations is associated with worse patient outcome [72].

NF2 is found mutated in the neurofibromatosis type 2 syndrome [73]. Inactivating mutations of NF2 increase MAPK signaling in KRAS mutant papillary thyroid cancers (PTC), through interactions with PAK1 and YAP1 (Fig. 1). NF2 inhibits RAC1, thus preventing PAK1-mediated phosphorylation of CRAF and MEK [74]. At the same time, NF2 phosphorylates YAP1, thereby blocking its nuclear translocation, and reduces transcription of the RAS genes with a TEAD binding site in their promoter [74]. Lack of NF2, due to loss of chromosome 22q, is detected in 45% of RAS mutant PTCs and is associated with marked sensitivity to selumetinib [74].

YAP1 overexpression is correlated with lymph node metastasis, pTNM stage and poor prognosis of both lung adenocarcinoma and LSCC [75]. YAP1 has been recognized as a marker of resistance to cetuximab in colorectal cancer patients [76]. YAP1 activation leads to overexpression of EGFR and its ligand amphiregulin (AREG) (Fig. 1) [77, 78]. It has also been demonstrated that extracellular matrix deprivation activates the Hippo pathway and inactivates YAP1 in lung adenocarcinoma, which eventually initiates the squamous trans-differentiation programme [79].

The YAP1-TAZ pathway may be suppressed by compounds that target the upstream Src kinase or by small molecules that prevent YAP–TEAD interaction (Table 1). Src inhibitors like saracatinib, dasatinib or bosutinib have been clinically investigated [80]. In vitro data indicate activity not only against Src kinase but also EGFR even though there are few differences in terms of mechanism of action. Saracatinib is the most efficient at inhibiting mutant EGFR, thanks to a mechanism of action independent from Src which is not yet fully understood [81, 82]. Dasatinib exhibits the highest inhibition potential for Src Kinase and wild-type EGFR [83], suggesting that combination with selumetinib or necitumumab could be a very promising treatment strategy for patients with KRAS mutated NSCLC and LSCC, respectively. Small molecules interfering with YAP–TEAD interaction hinder YAP1-induced gene transcription and are also efficient in NF2 inactivation. Verteporfin, currently used in combination with photodynamic therapy to treat macular eye degeneration, was the first investigated YAP1 inhibitor [84]. Super-TDU blocks YAP1–TEAD complexes, mimicking the action of vestigial-like family member 4 (VGLL4), the tumor suppressor that competes with YAP1 for TEAD binding (Fig. 1) [85]. Finally, ivermectin, besides its activity against PAK1 [61], has been identified as a potential YAP1 inhibitor through a mechanism that is not yet clear [86] (Table 1).

AXL: Biological Functions and Therapeutic Implications

AXL (from the Greek word ‘anexelekto,’ or uncontrolled), a direct transcriptional target of YAP1, is a tyrosine kinase receptor involved in the epithelial mesenchymal transition (EMT) process [87]. It is a potent oncogene that modulates resistance to conventional and targeted cancer therapies [88]. A gene expression and proteomics-integrated analysis identified an EMT gene signature able to differentiate the epithelial from the mesenchymal phenotype. NSCLC cell lines classified as mesenchymal were resistant to PI3K/AKT inhibitors and exhibited high AXL expression, but great sensitivity to AXL inhibitors [89]. The EMT gene signature classified 60% of the KRAS mutant cell lines tested in the mesenchymal subgroup. When the signature was applied in patients enrolled in the BATTLE-1 trial (Biomarker integrated Approaches of Targeted Therapies for Lung Cancer Elimination), KRAS mutant patients were equally distributed between the epithelial and the mesenchymal group [89]. These data appear controversial, as preclinical findings suggest that NSCLC cell lines harboring KRAS mutations or cell lines dependent on RAS signal for proliferation are associated with epithelial differentiation, while KRAS mutant cell lines independent of KRAS for survival exhibit mesenchymal behavior [14]. When a KRAS mutant NSCLC patient underwent rebiopsy after 1 week of selumetinib treatment, high expression of the ERBB2/ERBB3 receptors was found, suggesting that heterodimeric complexes involving the ERBB signaling can emerge following MEK inhibition [90]. We were able to identify AXL as the most overexpressed gene in tumors with acquired resistance to erlotinib [88]. In our in vitro and in vivo models, AXL overexpression was related to NF-κB activation while knockdown of AXL restored sensitivity to erlotinib treatment in in vivo models [88]. We additionally identified AXL overexpression as a mechanism of acquired resistance to cetuximab in NSCLC and HNSCC [91]. In fact, EGFR directly regulates expression of AXL mRNA through MAPK signaling and the transcription factor c-Jun in cetuximab-resistant cells, creating a positive feedback loop that maintains EGFR activation by AXL (Fig. 1) [91]. In addition, AXL blockade inhibits proliferation, migration and invasion in prostate cancer cell lines and in in vivo models through inhibition of the NF-κB pathway [92]. Intriguingly, AXL blockade inhibits the secretion of IL-6, which in turn reduces proliferation of androgen-insensitive prostate cancer cells via STAT3 inhibition [92].

Several AXL inhibitors are currently under development. Most are ATP competitive kinase inhibitors, with cross-reactivity against multiple tyrosine kinase receptors (Table 1). The lack of AXL kinase domain crystallographic models may explain the difficulties encountered during the development of AXL inhibitors [93]. Foretinib is a dual MET/vascular endothelial growth factor receptor 2 (VEGFR2) inhibitor, with activity against AXL [93]. Cabozantinib, already approved for the treatment of medullary thyroid cancer, inhibits MET, VEGFR2, rearranged during transfection (RET), and AXL [93]. MGCD265 is a MET/VEGFR2 inhibitor also able to suppress AXL activity [93]. BMS-777607 was initially developed as a MET inhibitor but is more potent against AXL. Finally, BGB324/R428, specifically designed to target AXL, is the only compound with two phase I studies ongoing [93].

Unraveling the Therapeutic Potential of the Hedgehog Pathway in Lung Cancer

The hedgehog (Hh) signaling pathway is highly conserved and has a crucial role in embryogenesis, adult tissue homeostasis and carcinogenesis [94]. Hh signaling is activated following binding of Sonic Hh (Shh), Desert Hh and Indian Hh ligands to their transmembrane receptor Patched (PTCH), leading to the release of Gli transcription factors by depressing Smoothened (SMO) (Fig. 1) [95].

Ligand-dependent activation of the Hh pathway has been described in epithelial tumors, but SMO inhibitors have so far failed to show any benefit in clinical trials of pancreatic, colon or ovarian cancer [96, 97]. Complex crosstalk between Hh and other pathways seems to play a significant role in resistance to Hh pathway inhibitors. LSCC tumors overexpress PKCiota, SOX2 and Hh acyltransferase (HHAT) and require PKCiota-SOX2-HHAT signaling to maintain a stem cell phenotype (Fig. 1) [98]. NSCLC cells overexpressing Gli1 can overcome the growth-inhibitory effects of Hh antagonist such as cyclopamine [99] through crosstalk between the Hh and EGFR signaling pathways. In fact, cooperation between Hedgehog and Gli-EGFR synergistically induces expression of SOX2 [100]. Targeting Hh signaling at the level of Gli may be more effective than targeting either Shh or SMO in LSCC. Gli2 is likely to be the major signaling transducer in LSCC and is the primary activator of Hh signaling, with Gli1 being a transcriptional target of Gli2 [101]. SMO has a minimal role in regulating LSCC survival via the canonical Hh pathway [101]. Loss of SMO did not reduce Gli2 mRNA level in LSCC cell lines and only Gli2 knockdown inhibited cell proliferation and survival and induced apoptosis [101]. The Hh pathway also seems to be involved in KRAS oncogene transformation [102]. Loss of Gli1 inhibits KRAS-induced pancreatic paraneoplastic lesions in mice. Shh promotes binding between Gli1 and the IL-6 promoter, thus favoring activation of STAT3 and gene transcription (Fig. 1) [102].

Vismodegib is a first-in-class small-molecule SMO inhibitor approved for treatment of advanced basal skin carcinoma [103]. GANT61 is a Gli1 and Gli2 inhibitor [97]. The isoflavone genistein also inhibits Gli1 mRNA expression and downregulates Gli reporter activity [104]. Among the catechins contained in green tea, epigallocatechin 3-gallate (EGCG) has been reported to inhibit prostate cancer and chondrosarcoma proliferation through Gli1 inhibition [105]. Finally, mebendazole, a drug approved for treatment of nematode infections, has been found to suppress the formation of the primary cilium, a microtubule-based organelle that functions as a signaling hub for Hh pathway activation [106].

Conclusion: Future Perspectives

The success of targeted agents in molecularly defined subsets of patients has radically changed treatment strategies. The emerging issue now is that KRAS mutant and LSCC are not unique entities, as observed in EGFR mutant or EML4-ALK tumors, but heterogeneous diseases in which common signaling pathways dictate distinct pathologic outputs. The characterization of KRAS mutations is not sufficient to properly classify patients. Predictive biomarkers, representative of the underlying activated signaling pathways, need to be defined to develop optimal combinations for synthetic lethal approaches.