Nonmelanoma skin cancer (NMSC) is the most common cancer worldwide and the incidence continues to rise, in part due to increasing numbers in high-risk groups such as organ transplant recipients and those taking photosensitizing medications. The most significant risk factor for NMSC is ultraviolet radiation (UVR) from sunlight, specifically UVB, which is the leading cause of DNA damage, photoaging, and malignant transformation in the skin. Activation of apoptosis following UVR exposure allows the elimination of irreversibly damaged cells that may harbor oncogenic mutations. However, UVR also activates signaling cascades that promote the survival of these potentially cancerous cells, resulting in tumor initiation. Thus, the UVR-induced stress response in the skin is multifaceted and requires coordinated activation of numerous pathways controlling DNA damage repair, inflammation, and kinase-mediated signal transduction that lead to either cell survival or cell death. This review focuses on the central signaling mechanisms that respond to UVR and the subsequent cellular changes. Given the prevalence of NMSC and the resulting health care burden, many of these pathways provide promising targets for continued study aimed at both chemoprevention and chemotherapy.
There are an estimated 3.5 million people in the USA who have been diagnosed with nonmelanoma skin cancer (NMSC), making it the most common form of cancer in the country , with medical related costs reaching $650 million annually . The incidence of NMSC is also high in the UK; it has been reported that 1 of every 1000 individuals in the UK are diagnosed with the disease per year . NMSC has become a major health concern in both countries as the incidence of NMSC in the USA has increased a remarkable 300% since 1994 , while the rate of NMSC occurrence in the UK appears to be rising faster than any other European country . NMSC is classified into two major forms: basal cell carcinoma (BCC) and squamous cell carcinoma (SCC). BCCs, which arise from the basal cell layer, constitute the majority of all diagnosed skin cancers (80%) and are rarely life-threatening or metastatic. SCCs, which can arise from hair follicle stem cells , account for 16% of all skin cancers and are much more dangerous; these cancers are more likely to invade and metastasize [2,6]. The incidence of NMSC, like many cancers, is much higher in older adults, and individuals who are over 60 years of age account for nearly 80% of all NSMC cases [6,7]. More alarmingly, however, the rate of NMSC occurrence is steadily increasing in patients younger than 35 years . Taken together, these facts demonstrate the importance of understanding the molecular pathways behind NMSC development to increase our ability to diagnose, prevent, and treat NMSC.
Ultraviolet radiation (UVR) from the sun is the greatest environmental risk factor for developing NMSC . There are three distinct energetic forms of UV light that are emitted from the sun: UVA (320–400 nm), UVB (280–320 nm), and UVC (200–280 nm). However, only UVA and UVB pierce the atmosphere and reach the general populace . Although UVA can penetrate deeper into the skin, UVB is the more energetic and accounts for the majority of the biologically damaging effects from sun exposure; these effects include direct DNA damage, activation of receptor-mediated signaling pathways, and formation of reactive oxygen species (ROS) . Heightened exposure to UVR, both in intensity and in duration, directly increases the risk of NMSC. Therefore, individuals who are fair-skinned (Fitzpatrick Scale — Types I and II), spend large amounts of time in the sun or on tanning beds, and/or live near the equator are more susceptible to UVB-induced skin carcinogenesis [2,6,7,9,12]. Currently, according to the American Cancer Society (ACS), the best methods for protecting against the harmful effects of UVR when outdoors are to wear protective clothing, seek shade when possible, apply copious amounts of sunblock, and protect the eyes with sunglasses. However, previous medical conditions can also dramatically increase an individual's vulnerability to NMSC in response to UVR exposure, even when following these guidelines. Patients who have had SCCs or related diseases in the past have an increased chance to acquire additional SCCs, as well as melanoma . Similarly, it has been shown that patients who take immunosuppressive drugs following an organ transplant are 60–100× more likely to develop SCC than the general population [14,15].
The molecular mechanisms behind UVR-induced skin carcinogenesis are complex. UVR, specifically UVB, is considered a complete carcinogen as it can both initiate and promote cancer. UVR is able to penetrate through the skin to affect the keratinocyte stem cells, located in the basal layer of the epidermis. DNA damage can result from UVR exposure, which primarily generates cyclobutane pyrimidine dimers and (6–4) photoproducts . These damaged cells will undergo DNA repair, most commonly through nucleotide excision repair (NER) to correct the UVR-induced damage . The importance of the DNA repair response following UVR exposure cannot be underestimated. Individuals who are born with Xeroderma pigmentosum, an autosomal recessive genetic disease resulting in mutations in NER, have an extraordinary 2000-fold increased incidence of NMSC and melanoma [18–20]. If the irradiated keratinocyte cannot repair the UVR-induced DNA damage, it will either undergo cell cycle arrest or initiate apoptosis as a regulatory defense mechanism . Importantly, UVR irradiation can also initiate tumorigenesis by inducing pro-survival pathways in keratinocytes that oppose and counteract apoptosis, thereby allowing damaged cells to survive. UVR can activate these pathways through direct DNA damage of critical target genes , activation of cell surface receptors , and/or increased inflammation and immunosuppression . Many of these pathways can promote tumorigenesis through increased cell cycle progression and proliferation [2,24]. Thus, it is important to understand how these pathways are regulated in order to develop novel drugs and therapeutic strategies to treat NMSC, as well as act as chemopreventive agents in high-risk individuals. This review will highlight key signaling pathways that are activated in response to UVR and describe the downstream pathophysiological responses each pathway elicits.
DNA damage from direct UVR exposure results in the formation of cyclobutane pyrimidine dimers, (6–4) photoproducts, DNA cross-links, single-strand breaks (SSB) and double-strand breaks (DSB) . If the DNA damage caused by the resulting adducts is not repaired, a UVR-specific permanent mutation is introduced into the genome. Specifically, a cytosine that is adjacent to a thymine or another cytosine is mutated to a thymine . This mutation is commonly found in the tumor suppressor p53 gene of SCCs and historically was considered the primary driver of NMSC initiation . However, more recent studies have begun to show that, in addition to direct DNA damage, UVR-induced signal transduction can also play a central role in NMSC initiation. Due to the large number of molecular pathways that are activated in response to UVR, UVR exposure can elicit a response in keratinocytes that is simultaneously both oncogenic/pro-survival and tumor suppressive/apoptotic . Alterations in any of these pathways in damaged cells can shift the balance in favor of survival or death. Among the first of these signaling mechanisms to be studied were the pathways controlled by the phosphatidylinositol 3-kinase-related kinases (PIKK kinases) ataxia telangiectasia-mutated (ATM) and Rad3-related (ATR) proteins.
The PIKK family of proteins is critical in maintaining and regulating DNA damage checkpoints . Specifically, ATM and ATR are essential proteins activated in response to UVR-induced DNA damage [29,30]. In addition to their role in recruiting  or being recruited by  the DNA repair machinery to the site of damage, ATM and ATR activate a downstream signaling cascade that results in either repair or apoptosis, depending on the extent of DNA damage (Figure 1A,B). In the absence of stimulation, ATM resides as an inactive dimer. Upon UVR-induced DSB, ATM undergoes autophosphorylation (at Ser367, Ser1893, Ser1981 , and Ser2996 ) and becomes an active monomer , which subsequently phosphorylates checkpoint kinase 2 (CHK2) at Thr68 . Activated CHK2 inhibits the CDC25 phosphatases through phosphorylation at Ser216, which promotes their degradation. This reduction in CDC25 activity prevents the cell from entering mitosis . Additionally, ATM has been shown to phosphorylate p53 (at Ser9, Ser15, and Ser46), resulting in increased p53 stability and accumulation [38,39], which leads to cell cycle arrest in G1 . More recent studies have shown that in addition to direct p53 phosphorylation, ATM also phosphorylates the E3 ubiquitin-protein ligase mouse double minute 2 homolog (MDM2) at Ser394. MDM2 normally targets p53 for degradation; however, phosphorylation by ATM inhibits this action, further increasing the stability of p53 . By halting the cell cycle through either CDC25 degradation or p53 accumulation, ATM provides the damaged cell both time to repair the DSB and also prevents its replication [28,42]. The presence of DSB that cannot be repaired can trigger apoptosis as a result of prolonged activation of ATM. Both ATM and subsequent CHK2 activation can phosphorylate and activate the pro-apoptotic transcription factor E2F1 at Ser31 and Ser364, respectively [43,44]. Therefore, ATM-dependent apoptosis can occur through increased activity of both p53 and E2F1 (Figure 1A).
ATM/ATR-dependent cell cycle arrest or apoptosis in response to UVR.
In contrast, ATR is activated by generation of SSB due to UV-induced damage and replication stress/stalling [28,45]. Similar to ATM, ATR halts the cell cycle to allow sufficient time for DNA repair or, if the resulting DNA damage is too excessive, induce apoptosis. Interestingly, direct signaling cross-talk exists between the two kinases as activated ATR can directly phosphorylate ATM at Ser1981 in response to UVR exposure, resulting in the activation of ATM . UVR-induced activation of ATR leads to the phosphorylation of checkpoint kinase 1 (CHK1) at Ser345 , which subsequently phosphorylates and inhibits CDC25 activity, halting the cell cycle in a manner similar to CHK2 . CHK1 can also inhibit CDC25 activity by either the activation of never in mitosis A-related kinase 11 (NEK11), which phosphorylates and inhibits CDC25 [49,50], or via direct inhibition of the polo-like kinase 1 (PLK-1), a known activator of CDC25 . Analogous to the action of ATM, ATR increases p53 stability through phosphorylation of MDM2 at Ser407. This phosphorylation reduces the MDM2-dependent nuclear export of p53, thereby increasing p53 activity . Both ATM and ATR can also directly phosphorylate (at Ser19) and inhibit the E3 ubiquitin ligase seven in absentia homolog 1 (SIAH1), resulting in activation of p53. In the absence of stimulation, SIAH1 promotes the degradation of the homeodomain-interacting protein kinase 2 (HIPK2) , thus preventing HIPK2 from activating p53  (Figure 1B).
UVR-induced activation of ATR has also been linked to the cyclin-dependent kinase inhibitor p21WAF1/Cip1 (p21) pathway, which interacts with numerous cyclins/cyclin-dependent kinases (CDKs) to halt the cell cycle . Reports show that ATR has UVR dosage-dependent, paradoxical effects on the p21 pathway. At high doses, ATR inhibits p21, which appears counterintuitive; however, this inhibition has been shown to promote apoptosis , which may be the result of excessive DNA damage. Interestingly, low-dose UVR protects p21 from s-phase kinase-associated protein 2 (Skp2)-mediated degradation, leading to a delay in cell cycle progression to allow DNA repair . Furthering the complexity, a recent report shows that high energy UVC exposure can lead to increased p21 mRNA stability in a CHK1-dependent fashion via cytoplasmic localization of the RNA-binding protein HuR. In the absence of ATR activation, HuR is localized to the nucleus due to CDK1-dependent phosphorylation of HuR at Ser202. Activation of ATR/CHK1 inhibits CDK1, which allows HuR to translocate to the cytoplasm and stabilize p21 mRNA . While several reports show that complete inhibition of ATR can increase both tumor development and aging [59,60], another study found that the diminished expression/function of ATR in UVR-sensitive mice resulted in protection from UVR-induced skin carcinogenesis . Therefore, when developing therapeutics that target DNA repair proteins for the treatment of NMSC, it is important to consider how different levels of activity of these proteins may affect skin tumorigenesis.
Multiple studies have shown an association between inflammation and the promotion of various cancer types, and inflammation has been referred to as the ‘seventh hallmark of cancer’ [62–64]. In fact, the correlation between cancer and inflammation dates back to 1863, when Rudolf Virchow discovered leukocytes associated with tumor cells . Inflammatory pathways are classically activated upon skin injury via UVR-induced sunburn, which stimulates the dilation of blood vessels to increase blood supply to the injury site, changes in the microvascular structure that help plasma proteins enter from the blood stream, and increased migration of white blood cells to the site through the endothelium . These responses create a microenvironment that aids in wound healing. However, this environment can be hijacked to help stimulate tumorigenesis . Inflammation has been shown to increase keratinocyte survival, proliferation, and transformation in NMSC development and progression, as well as in promoting invasion and metastasis . Clinical studies have demonstrated that inflammation is involved in the progression from actinic keratosis (AK) to SCC. Additionally, chronic inflammation is consistently observed in the invasive periphery of tumor cells in patients with SCC . Since inflammation is a hallmark response of sunburned skin , it is of obvious importance to understand the molecular pathways involved and their effect on NMSC progression. Here, we will highlight two key inflammatory pathways involved in UVR-induced inflammation: the nuclear factor kappa-light-chain-enhancer of activated B-cells (NFκB) pathway and the signal transducer and activator of transcription 3 (STAT3) pathway.
The NFκB pathway
The NFκB pathway has been extensively studied and revealed to be crucial in mediating inflammation of the skin (reviewed in ref. ). Inactive NFκB exists primarily as a cytoplasmic heterodimer of the p65 and p50 subunits, which are bound to the inhibitory protein IκB. Induction of the canonical NFκB pathway can occur either via direct UVR activation or by the binding of UV-induced cytokines to plasma membrane receptors. The most commonly associated receptors with NFκB activation in NMSC are receptor tyrosine kinases (RTKs, discussed elsewhere in this review), tumor necrosis factor receptors (TNFR1, reviewed in ref. ), and Toll-like receptors (TLRs), specifically TLR2, 3, and 4  (Figure 2A). In general, TLR activation occurs following UVR-induced ligand binding, which recruits the I kappa B kinase (IKK) complex, consisting of IKKα and β, as well as the NFκB essential modulator (NEMO/IKKγ), to the TLR receptor, resulting in activation of IKK and subsequently, NFκB. Specifically, activation of the TLR4 induces a conformational change, which allows binding of the myeloid differentiation primary response 88 (MyD88) protein to the TLR. The IL-1 receptor-associated kinase (IRAK) family of kinases then binds to MyD88, inducing autophosphorylation and activation of IRAK  and recruitment of TNF receptor-associated factor 6 (TRAF6) proteins. TRAF6 proteins are E3 ubiquitin ligases that interact with ubiquitin conjugating enzymes to promote the creation of polyubiquitin chains . The recruitment of the IKK complex to the newly formed ubiquitin chains at the receptor is driven by the ubiquitin-binding protein NEMO. This NEMO-dependent binding of the IKK complex to TRAF6 allows for the proximal phosphorylation and activation of IKK (at Ser177&181 for IKKβ and at Ser176&180 for IKKα) [73–75]. The activated IKK proteins in turn phosphorylate IκB at Ser32 and Ser36, inducing ubiquitination and proteasomal degradation of the protein and releasing the inhibition of NFκB . The active NFκB heterodimer translocates to the nucleus to promote the transcription of pro-inflammatory genes , specifically genes that control apoptosis, as well as genes that lead to the production of cytokines and interferons that regulate the immune/inflammatory response. Among these targets are TNFα, IL-1, IL-6 and IL-8 . IKK can also phosphorylate the p65/50 dimer directly at Ser536 of p65. This phosphorylation site has been shown to further increase NFκB nuclear transcriptional activity [79,80]. Recent reports show that the p65 subunit of NFκB is essential for skin carcinogenesis in mice. Loss of p65 prevented both SCC tumor initiation and tumor promotion . Other reports show that dsRNA released from UVB-induced necrotic keratinocytes can activate TLR3, which results in the archetypal sunburned skin inflammatory response . In addition, there are numerous studies that link activation of NFκB to a variety of other skin diseases (reviewed in ref. ).
The inflammatory response to UVR acts through TLR4-dependent NFκB signaling and EGFR-dependent activation of STAT3.
The STAT3 pathway
The STAT family of transcription factors is induced in response to growth factors and cytokines. This family of proteins regulates factors that can affect both tumorigenesis and the tumor microenvironment . STAT3 has been shown to play an important role in survival and proliferation of keratinocytes exposed to UVB. Targeted overexpression of STAT3 in the basal epidermis (K14-STAT3) accelerates skin tumor formation in mice exposed to UVB, while targeted deletion of STAT3 in the same cells confers resistance . Interestingly, altering STAT3 activity within basal keratinocytes has profound effects on hair follicle stem cells and skin progenitor cells , which are implicated as the origin cells in NMSCs. Constitutively active STAT3 has been found in both mouse and human SCCs, and plays a role in promoting skin inflammation during tumor development . In the canonical STAT3 pathway (Figure 2B), UVR exposure activates the epidermal growth factor receptor (EGFR), a cell surface RTK, which phosphorylates the Janus-associated kinase (Jak). This allows the recruitment of STAT3 to the receptor complex through binding of the Src homology 2 (SH2) domain of STAT3 to phosphorylated Jak . Following recruitment, STAT3 is phosphorylated at Tyr705, which induces dimerization of STAT3 and allows it to bind to DNA and increase transcription of pro-inflammatory genes . Specifically, STAT3 promotes the transcription of numerous genes that are related to inflammation (i.e. interleukins), tumor promotion (c-Myc and c-Fos), cell survival [B-cell lymphoma 2 (Bcl-2), cyclin D1, and survivin], and metastasis (Twist and Zeb1) [89,90]. Additionally, studies in mice have shown that UVR-induced ROS activates protein kinase C epsilon (PKCε), which directly phosphorylates STAT3 at Ser727 [91,92]. While phosphorylation of STAT3 at Tyr705 is required to activate STAT3 transcriptional activity, studies show that dual phosphorylation of STAT3 at Tyr705 and Ser727 further increases the activity of STAT3-dependent transcription [91,93]. Noncanonical STAT3 pathways have also been linked to tumorigenesis of various cancers. It is important to note that while little is currently known of the noncanonical STAT3 pathways in UVR-induced skin carcinogenesis, these pathways are being actively investigated [94,95].
UVR exposure activates multiple stress–response pathways in keratinocytes. Oxidative stress, defined as unbalanced ROS production and antioxidant defenses, has been linked to the initiation of numerous cancers, including NMSC . Antioxidants have shown clinical efficacy to both prevent and treat cancer . However, recent studies show that prolonged increases in ROS production within cancer cells have tumor suppressive roles and initiate apoptotic pathways; therefore, administering antioxidative therapies may actually harm a patient depending on the stage of their cancer . Importantly, UVR exposure transiently increases ROS production within the skin both immediately following UVR irritation, as well as in a second response up to 3 h later . Depending on the activated molecular pathway, UVR-induced oxidative stress has been linked to multiple skin conditions, including aging, inflammation, and skin cancer. There is an inherent defense network present in keratinocytes to combat ROS accumulation; however, following UVB exposure, ROS production can be increased to such high levels that these antioxidative defenses are overwhelmed . Tumor development can result from a combination of increased ROS production and reduced antioxidant defense mechanisms in the skin, which leads to DNA damage and activating mutations [101,102]. Interestingly, differences in the accumulation of antioxidant enzymes such as the superoxide dismutases and catalase were observed in patient biopsies of both nonmelanoma and melanoma tumors. NMSC had strikingly lower levels of these enzymes than their melanoma counterparts, suggesting that chronic UVR exposure associated with NMSC development leads to weakened antioxidant defense response during tumorigenesis . This review describes two critical pathways that are activated in response to UVR-induced oxidative stress NMSC development: the p38 mitogen-activated protein kinases (p38) pathway and c-Jun N-terminal kinases (JNK) pathway.
The pathophysiological effects of UVR-induced activation of both p38 and JNK in keratinocytes is controversial, as reports show that these stress–response proteins can elicit both pro- and antisurvival mechanisms . However, given the bimodal role that oxidative stress plays in cancer, this seemingly contradictory response is not surprising. p38 is a member of the mitogen-activated protein kinase (MAPK) family of proteins and responds to a variety of cellular stress stimuli, including oxidative stress (Figure 3A). The dual specificity MAPK kinases (MKK) 3 and 6 phosphorylate p38 at Thr180 and Tyr182 within the activation loop of the protein. Once active, p38 phosphorylates both cytoplasmic and nuclear targets . ROS-dependent activation of p38 in keratinocytes has also been linked with activation of apoptosis signal-regulating kinase 1 (ASK1), a member of the MAP kinase kinase kinase (MAP3K) family of proteins that is upstream of both p38 and MKK3/6. Under oxidative stress, activation of ASK1 occurs following disassociation from the inhibitory thiol-disulphide oxidoreductase thioredoxin-1 (Trx1) protein . Other studies reveal that cells lacking specific MAPK phosphatases (MKP) have increased p38 activation following UVR exposure. Importantly, increased ROS production has been shown to inhibit MKPs, which triggers increased p38 activation in response to elevated oxidative stress [107,108]. It is also important to note that UVR can activate the p38 pathway independent of oxidative stress. Specifically, UVR-induced DNA damage can elicit a p38 response and can regulate the cell cycle through activation of MAPKAPK-2 [109,110].
UVR-induced ROS activate p38- and JNK-dependent signaling.
Recent studies have demonstrated that inhibition of p38 increases cell survival in murine SCC in vivo [111,112]. In agreement with this, activation of p38 in response to UVR promotes apoptosis in keratinocytes by phosphorylating p53 at Ser33 and Ser46, which increases the apoptotic activity of p53 . Other reports indicate that UVR-induced p38 activation induces NOXA-dependent apoptosis through increased protein expression of hypoxia-inducible factor-1-alpha (HIF-1α) in a p53-independent manner . In contrast, it has been shown that p38 activation increases UVR-induced survival of keratinocytes though up-regulation of the cancer-associated genes Bcl-extra-large (Bcl-XL) and COX-2 . The activation of p38 can also directly phosphorylate glycogen synthase kinase-3-beta (GSK3β) at Thr43 and Thr390 (Ser389 in mice), which promotes β-catenin-dependent growth and proliferation . Moreover, transgenic mice expressing a dominant negative p38 protein are resistant to skin carcinogenesis upon exposure to chronic UVR irradiation, perhaps due to a reduction in p38-dependent hyperproliferation .
JNK, like p38, is a member of the MAPK family that is activated upon phosphorylation following oxidative stress by upstream MKKs (MKK4 and 7) at Thr183 and Tyr185 (Figure 3B). MKK4 and 7, like MKK3 and 6, are activated in response to ROS production through either increased ASK-1 activity or reduced MKP activity . JNK activation occurs in human keratinocytes as quickly as 5 min following UVR exposure, with maximal activity peaking at 30 min post-UVR. This correlates perfectly with UVR-induced p38 kinase activity . A major target of JNK is the transcription factor activator protein-1 (AP-1). AP-1 exists as either a homodimer containing two c-Jun or c-Fos proteins, or a heterodimer containing one of each. AP-1 is known primarily as an oncogenic transcription factor as it is involved in promoting cell proliferation, specifically through transcription of cell cycle regulator genes (such as cyclin D1, cyclin A, cyclin E, p53, p21, p16Ink4a, and p19ARF). Additionally, AP-1 can elicit a cell survival response through cross-talk with NFκB . Once activated, JNK can phosphorylate c-Jun at Ser63 and Ser73  and this phosphorylation increases AP-1 activity in response to UVR exposure . Bowden and colleagues showed that pharmacological inhibition of JNK in human keratinocytes sensitizes cells to UVR-induced apoptosis in vitro , which correlates with a study by the same group showing that in vivo inhibition of AP-1 protected transgenic mice expressing a dominant negative c-jun from UVB-induced SCC . Additionally, pharmacological inhibition of JNK inhibited the growth of human head and neck SCCs in mouse xenografts . However, similar to p38, JNK activation can induce a pro-apoptotic response. Exposure of lung adenocarcinoma ASTC-a-1 cells to high-dose UVB results in JNK-dependent nuclear localization of the pro-apoptotic factor FOXO3a (discussed elsewhere in this review) , but this localization does not occur with lower doses of UVB. This low vs. high dosage effect may be similar to the response described above for ATR. Specifically, UVR-dependent JNK activation is initially protective; however, when the damage induced by UVR irradiation reaches a threshold level, the JNK-mediated signaling mechanisms within the cell activate apoptosis. It has also been shown in HEK293T cells that UVR-induced JNK activation can induce mitochondrial-derived intrinsic apoptosis through inhibition of Bcl-2 directly, or indirectly though activation of Bcl-2-like protein 11 (BIM) . However, this effect has not yet been studied in keratinocytes.
As mentioned above, activation of various signaling pathways in the skin can occur as quickly as 5 min following exposure to UVR with peak activity occurring between 30 min to 1 h on average [127–129]. The RTKs belong to an important family of receptors that are rapidly activated in response to UVR and have been linked to NMSC [24,130,131]. Key RTKs that are induced in response to UVR include the insulin receptor (IR), the insulin-like growth factor-1-receptor (IGF-1-R), and EGFR . Activation occurs when two transmembrane RTK subunits dimerize, typically in response to ligand binding, which induces autophosphorylation of their cytosolic kinase domains. While UVR-induced autocrine activation of RTKs has been described previously [133–135], the near immediate activation of RTKs upon exposure to UVR may be independent of ligand binding. Studies have shown that RTK activation can be propagated through UVR-induced ROS production [136,137], while other reports suggest the possibility that activation occurs through cell surface clustering of RTKs in response to UVR to promote autophosphorylation and subsequent activation [137,138]. RTK activation in keratinocytes predominantly initiates antiapoptotic and cell proliferative responses, which are critical processes for both wound repair and growth of new skin . However, aberrant UVB-induced activation of RTK pathways can lead to both NMSC initiation and promotion . It is for this reason that a variety of drugs are being developed to target and inhibit RTKs directly, as well as inhibiting the downstream components of the RTK signaling cascade (see ref. [140,141]). This review discusses the two best-studied NMSC-related RTK pathways: the phosphoinositide 3-kinase (PI3K)/mechanistic (mammalian) target of rapamycin (mTOR) pathway and the rapidly accelerated fibrosarcoma kinase (RAF)/MAPK kinase (MEK)/extracellular signal-regulated kinases (ERK) pathway.
The serine/threonine kinase mTOR is a PIKK kinase that is part of an incompletely defined signaling network that responds to diverse nutrients, growth factors and cellular stressors [142–144], including UVR [129,145,146]. mTOR exists in two distinct signaling complexes: the rapamycin-sensitive mTOR complex 1 (mTORC1) and the rapamycin-resistant mTORC2 , both of which are important in NMSC development in response to UVB exposure. mTOR was identified as a therapeutic target in several cancer types, as up-regulated mTOR signaling is frequently found in cancers [142–144]. Additionally, mutations in two critical upstream proteins of the mTOR pathway, PI3K and phosphatase and tensin homolog (PTEN), are commonly found in NMSC [147,148]. Reports show that increased levels of both mTORC1 and mTORC2 activity are present during progression from normal keratinocytes to pre-cancerous AK, and culminating in NMSC development . Activation of mTORC1-dependent pathways (Figure 4A) occurs following upstream activation of protein kinase B (AKT/PKB) via the PI3K pathway. Following autophosphorylation of the UVR-inducible RTKs (EGFR, IGF-R-1, and IR), PI3K, a heterodimer that consists of the p85 regulator subunit and the p110 kinase subunit, is recruited to the receptor through binding of the SH2 domain on p85 to the phosphorylated RTK. The p85 subunit is subsequently phosphorylated at Tyr458, activating the heterodimer. This allows PI3K to phosphorylate the phospholipid phosphatidylinositol 4,5-bisphosphate (PIP2) to generate phosphatidylinositol (3,4,5)-trisphosphate (PIP3) on the plasma membrane. PIP3 recruits both phosphoinositide-dependent kinase-1 (PDK1) and AKT to the receptor complex, where PDK1 can activate and phosphorylate AKT at Thr308. AKT is a central regulator of numerous cellular processes, including cell survival, metabolism, protein translation, and angiogenesis . Three AKT isoforms have been described, of which AKT1 and 2 have been strongly implicated in tumorigenesis; however, studies suggest that melanoma development may depend on AKT3 . The tumor suppressor PTEN negatively regulates PI3K-dependent signaling though PIP3 dephosphorylation and conversion to PIP2. Phosphorylation by PDK1 activates AKT, which subsequently phosphorylates and inhibits the mTORC1-negative regulator tuberous sclerosis complex 1/2 (TSC1/2). This inhibition allows the ras homolog enriched in brain (Rheb) GTPase to bind to and activate mTORC1 . mTORC1 activation leads to cell growth and proliferation primarily through the downstream phosphorylation (at Thr389) and activation of ribosomal protein S6 kinase beta-1 (S6K1), which controls ribosome biogenesis, as well as activation of cap-dependent translation via the phosphorylation (at Thr37, Thr46, Ser65, and Thr70) and subsequent degradation of the eukaryotic translation initiation factor 4E inhibitory 4E-binding protein 1 (4EBP1) . This mTORC1-dependent signaling cascade has been described in detail in multiple reviews [142–144,154–156].
RTK activation by UVR results in an antiapoptotic and cell proliferative response through PI3K/mTOR and Raf/MEK/ERK-dependent pathways.
Activation of mTORC1 plays a causal role in tumor promotion , and the mTORC1-specific inhibitor rapamycin (Sirolimus) is approved as a therapeutic in multiple cancer types [141,158–160]. Rapamycin allosterically inhibits mTORC1 by binding to the FK-binding protein 12 in the cytosol, which can then directly bind to the mTOR kinase subunit of mTORC1 and block its function . Notably, rapamycin has no direct inhibitory effect on the activity of mTORC2 and, depending on the length of exposure and tissue type, the indirect effects of rapamycin treatment on mTORC2 can vary . DiGiovanni and colleagues showed that mTORC1 inhibition with topical rapamycin blocks dermal inflammation that is linked with the hyperproliferative response to tumor promotion in mice . In addition, organ transplant patients who took rapamycin as an immunosuppressive drug had a significant reduction in acquired NMSC when compared with patients taking cyclosporine A . Our laboratory and others have demonstrated that mTORC1 is required for keratinocyte hyperproliferation in response to tumor promotion in murine skin carcinogenesis models [129,165]. There is controversy regarding the role of mTORC1 in NMSC tumor initiation. Studies in other cancer models demonstrate that treatment with rapamycin can either increase or decrease cell survival [166–170]. Studies from our laboratory showed that treatment with rapamycin or genetic knockdown of the essential mTORC1 scaffolding protein Raptor does not affect cell viability or apoptosis following exposure to UVB in spontaneously immortalized human keratinocytes (HaCaT cells) or mouse primary keratinocytes [129,145]. In contrast, a recent report by Bowden and colleagues showed that rapamycin treatment did increase UVR-induced keratinocyte cell death in mice . The differing results seen in these studies are likely due to the differences in both model systems and UVR-dosages used. For example, our in vitro studies used UVB dosages between 20 and 35 mJ/cm2, whereas the Bowden group used 640 mJ/cm2 UVB for acute in vivo exposure experiments.
Studies linking mTORC2 to tumor initiation have suggested that this less studied of the mTOR complexes is also a viable target for therapeutic intervention . Using mouse models with stage-specific deletion of Rictor in the basal layer of the epidermis, our recent work has established an essential role for mTORC2-controlled pathways in both skin tumor development and maintenance . These results suggest that inhibition of mTORC2 signaling may be an effective strategy for both treatment and prevention of NMSC. Though the mechanism of mTORC2 signaling is only partially defined (Figure 4A), recent reports show that the activation of mTORC2 occurs following PIP3 phosphorylation by PI3K . The pleckstrin homology (PH) domain of the mTORC2 scaffolding/inhibitory protein SIN1 is recruited to the plasma membrane, where it binds to PIP3. Binding of SIN1 to PIP3 induces a conformational change in mTORC2 and uncovers the active kinase site of mTOR . The first role described for mTORC2 was the control of actin cytoskeleton rearrangement ; however, induction of mTORC2 also activates both the serum- and glucocorticoid-induced protein kinase SGK and AKT, both of which are involved in controlling apoptosis in response to various cellular stresses and have been implicated as potential therapeutic targets in cancer . Following its initial binding of mTORC2, PIP3 recruits the PH domain of AKT to mTORC2, which permits the phosphorylation of AKT at the hydrophobic motif site Ser473 by mTORC2 . Recent studies have revealed that the DNA-dependent protein kinase catalytic subunit (DNA-PKcs) can also regulate mTORC2 activation. In response to UVR-induced DNA damage, DNA-PKcs exits the nucleus and binds to SIN1. This interaction promotes mTORC2-mediated phosphorylation of AKT at Ser473 . This phosphorylation, in addition to the PDK1-dependent Thr308 phosphorylation, increases the substrate specificity of AKT for downstream targets. While phosphorylation of AKT at Thr308 is sufficient to induce mTORC1-specific activation , studies suggest that AKT must be phosphorylated at both Thr308 and Ser473 in order to control downstream cell survival pathways, including regulation and inhibition of pro-apoptotic proteins such as Bcl-2-associated death promoter (BAD), Bcl-2-associated X (BAX), GSK-3β, ASK1, caspase-9, and FOXO3a [150,176].
The transcription factor FOXO3a has been identified as a critical component in the apoptotic pathway of several cancer types [177–185]. FOXO3a is a member of the FOX (Forkhead Box) superfamily of transcription factors that control diverse cellular functions . The subclass FOX ‘O’ proteins chiefly regulate oxidative stress, cell cycle progression, and apoptosis [183,184,186]. FOXO3a directly targets and increases transcription of both extrinsic (Trail and FasL) and intrinsic [BIM and p53 up-regulator of apoptosis (PUMA)] apoptotic genes , which are implicated in UVR-induced apoptosis [187–189]. While there are a variety of cellular elements that post-translationally modify FOXO3a, AKT and ERK are considered two of its primary negative regulators [186,190,191]. Upon phosphorylation of AKT at Thr308 and Ser473, AKT translocates into the nucleus and phosphorylates FOXO3a at three sites: Thr32, Ser256, and Ser319  (Figure 4A). Previous studies show that AKT-mediated phosphorylation of FOXO3a alone does not reduce its activity. Instead, the 14-3-3ζ chaperone binds to this evolutionally conserved motif and reduces the affinity of FOXO3a for DNA by masking the DNA-binding domain . In addition, 14-3-3ζ binding masks two nuclear localization sequences on the FOXO3a , which along with two already exposed nuclear export signals, induces FOXO3a cytoplasmic localization and sequestration . Our laboratory has recently demonstrated that knockdown of either rictor or mSIN1, two key structural components of mTORC2, sensitizes keratinocytes to UVB-induced apoptosis, and that this effect is dependent on AKT-mediated FOXO3a regulation . Upon exposure to UVR, FOXO3a is shuttled out of the nucleus and sequestered into the cytoplasm. However, disruption of mTORC2/AKT signaling inhibits this process and an increased accumulation of FOXO3a is observed in the nucleus. The increased sensitivity to UVR-induced apoptosis following inhibition of mTORC2 signaling is rescued with simultaneous knockdown of FOXO3a, suggesting that this process is dependent on the activity of FOXO3a .
Activation of RAF/MEK/ERK signaling also occurs downstream of EGFR in response to UVR exposure [196,197], and this pathway has been identified as a chemoprevention target in NMSC [2,198]. Following activation of EGFR (Figure 4B), the growth factor receptor-bound protein 2 (GRB2) is recruited to the receptor, resulting in binding to the guanine nucleotide exchange factor Son of Sevenless (SOS). This enables SOS to liberate the bound GDP on the RAS GTPase. Once GDP is released, RAS quickly binds a free GTP and induces RAF activation . There are three isoforms of RAF that are involved in signaling to the MEK/ERK pathway: A-RAF, B-RAF, and C-RAF (RAF1). Mutations in B-RAF in particular are linked with the development of melanoma and are reviewed in ref. . All three RAF isoforms can bind to and phosphorylate MEK1 and 2 (at Ser217 and Ser221). This phosphorylation activates MEK, which in turn phosphorylates (at Thr202 and Tyr204) and activates ERK (p44/42 MAPK) 1 and 2 . Once activated, ERK regulates numerous downstream targets involved in cancer that control cell growth and survival, including c-Jun, c-Myc, IKK, and FOXO3a . Interestingly, studies show that certain patients who were successfully treated for BCCs through the use of smoothened (Smo) inhibitors, which block sonic hedgehog (SHH) signaling, later develop SCCs at the same site. It was determined that following Smo inhibition, the regressed BCC cells switch from aberrant SHH activation to increased RAS/MEK/ERK signaling, which promotes the development of SCC .
Similar to AKT, ERK can phosphorylate FOXO3a (at Ser294, Ser344, and Ser425) in the nucleus, which triggers its exit into the cytoplasm (Figure 4B). However, contrary to mTORC2/AKT/14-3-3ζ regulation, FOXO3a is targeted for MDM2-mediated cytoplasmic degradation in response to the ERK-specific phosphorylation . In our recent study, we noted a UVR-induced increase in total FOXO3a abundance in HaCaT cells with intact mTORC2 signaling that was not present following inhibition of mTORC2 , suggesting a possible degradation of FOXO3a. Several studies suggest that 14-3-3ζ binding acts as a ‘protective barrier’ for FOXO3a, which guards FOXO3a from PP2A-mediated dephosphorylation and subsequent degradation triggered by other signaling pathways, including the RAF/MEK/ERK pathway [195,205]. Therefore, we simultaneously inhibited both of these pathways in HaCaT cells and showed an increase in nuclear FOXO3a protein both before and immediately after exposure to UVR. These results correlated with a further increased sensitivity to UVR-induced apoptosis compared with inhibition of either pathway alone, which was also dependent on FOXO3a . It has been demonstrated that dual targeting of the mTORC2/AKT and RAF/MEK/ERK pathways in advanced cancers produces better efficacy in patients compared to inhibition of either pathway alone . In addition, studies in colon and pancreatic cancer models show that dual inhibition of these pathways specifically increases the tumor suppressor activity of FOXO3a [185,207,208]. These results reflect the extensive cross-talk that occurs between the PI3K/mTOR and the RAF/MEK/ERK pathways (see ref.  for review). Given these results, combination therapy with a TOR kinase inhibitor and an MEK inhibitor for NMSC prevention warrants further investigation.
A common denominator relevant to each reviewed signaling pathway is the ability to inhibit and/or induce caspase-dependent apoptosis upon exposure to UVR. As discussed above, increased cell survival due to inhibition of apoptosis is a key contributor to UVR-induced initiation of NMSC; therefore, an understanding of the signaling cascades that result in apoptosis initiation will provide insights into the development of NMSC. Caspase-dependent apoptosis is described as extrinsic, death receptor (DR)-dependent apoptosis or intrinsic apoptosis initiated at the mitochondria . The initiation of extrinsic apoptosis (Figure 5) requires ligand binding and activation of plasma membrane DRs, which are members of the tumor necrosis factor superfamily of receptors (TNFR) . Three examples of DRs that have been linked to in skin carcinogenesis are TNF-R1, the Fas/APO1/CD95 receptor (FasR), and the TNF-related apoptosis-inducing ligand receptors 1 and 2 (TRAIL-R1/DR4 and TRAIL-R2/DR5) [89,203], each with a specific extracellular ligand reflected in the receptor name (i.e. TRAIL binds to TRAIL-R). Negative regulators of DRs, referred to as decoy death receptors, have parallel extracellular domains that allow them to act as ligand sinks; however, only the active DRs have the fully functional intracellular death domain (DD) needed to induce apoptosis . TNFR1 signaling is reviewed in detail in ref. . Briefly, FasR and TRAIL-R1/2 trimerize upon ligand binding, which allows the recruitment and homotypic binding of fas-associated protein with death domain (FADD) to the death domain-binding sites on the intracellular domain of the DR. This binding allows FADD to recruit the pro-apoptotic proteins pro-caspase-8 and -10 into a complex referred to as the death-inducing signaling complex (DISC) . Following recruitment by FADD, pro-caspase-8 is ubiquitinated via a complex consisting of the cullin3 ubiquitin E3 ligase (CUL3) and RING-box protein 1 (RBX1). The resulting ubiquitination permits the binding of pro-caspase-8 to ubiquitin-binding protein p62/sequestosome-1 (p62), which promotes aggregation and homodimerization of pro-caspase-8, allowing autoproteolytic cleavage of pro-caspase-8 into its active form . This process is inhibited by the cellular FLICE-like inhibitory protein (c-FLIP), which binds to FADD and prevents the dimerization and subsequent activation of caspase-8.
UVR can activate both intrinsic and extrinsic apoptosis.
Once cleaved, the caspase-8 dimer causes the downstream activation of pro-caspase-3, -6, and -7. Activation of these caspases, primarily caspase-3, induces cell death through the cleavage of DNA, as well as various proteins involved in cell maintenance and survival. In this context, cleavage of poly(ADP-ribose) polymerase is often considered an endpoint marker of caspase-dependent apoptosis . Therapeutics that target DR activation may be an effective treatment for NMSC, as in vivo studies have shown that activation of these receptors prevent skin tumor development in a mouse model . In addition, targeting of either TRAIL-R or FasR has been shown to be an effective therapy against several types of cancers , although data on NMSC are not currently available.
Classically, intrinsic apoptosis (Figure 5) is activated in response to cellular stressors such as DNA damage or hypoxia, resulting in mitochondrial outer membrane (MOM) permeabilization (MOMP) and subsequent release of cytochrome C into the cytosol [176,216]. Specifically, MOMP is generated following the formation of pores in the MOM by the proteins BAX and Bcl-2 homologous antagonist (BAK), though the specific mechanism is still the subject of investigation . In the absence of UVR-induced cellular stress, the activities of BAX and BAK are inhibited by the antiapoptotic Bcl-2 proteins, including Bcl-2, Bcl-extra-large (Bcl-xL), and Bcl-2-like protein 2 (Bcl-w) . Following UVR irradiation, the BH3-only family of proteins, which include BH3 interacting-domain death agonist (BID), BIM, BAD, PUMA, and NOXA, binds to and inhibits the antiapoptotic Bcl-2 proteins . BID and BIM can also directly activate BAX and BAK, resulting in pore formation on the MOM . UVR-induced activation of the BH3 family of proteins results from increased activity of pro-apoptotic transcription factors such as FOXO3a, HIF-1α, and p53 [218,221]. In contrast, UVR-inducible signal transduction pathways can also reduce the activity of the BH3-only proteins through direct protein inhibition (i.e. AKT phosphorylation of BAD), as well as indirect inhibition of upstream pro-apoptotic transcription factors (i.e. inhibition of p53) [24,27].
Following its release into the cytosol, cytochrome C and the apoptotic protease-activating factor 1 (APAF-1) form a complex referred to as the ‘apoptosome’. This complex recruits and activates caspase-9, which in turn activates caspase-3 and -7 to induce cell death . Cross-talk between the extrinsic and intrinsic pathway exists in keratinocytes primarily through caspase-8 [223,224]. In addition to activating caspase-3, -6, and -7, caspase-8 can cleave BID to generate a truncated form (tBID), which is able to both inhibit the Bcl-2 proteins and promote BAX/BAM activation and subsequent mitochondrial-mediated cell death  (Figure 5). Thus, because UVR irradiation can initiate intrinsic apoptosis through DR activation as well as through direct mechanisms such as DNA damage, oxidative stress, and TLR/RTK activation, this apoptotic pathway is viewed as a more viable therapeutic target in the prevention and treatment of NMSC. Therapies designed to either inhibit the antiapoptotic Bcl-2 proteins or mimic the pro-apoptotic BH3-only family are currently being tested as a means of treating both melanoma and NMSC [219,225].
Recognizing the complexity of UVR-induced signal transduction is critical to our understanding of NMSC tumorigenesis. Depending on the intensity and duration of UVR irradiation, as well as the level of cellular damage (i.e. DNA/oxidative damage) that results from exposure to UVR, each activated signaling pathway can elicit varying responses that can be tumor suppressive, oncogenic, or a mixture of both. As with most physiological mechanisms, there is a great deal of redundancy that exists within the cell, which allows extensive regulation and cross-talk between pathways. As a result, UVR exposure does not activate unique pathways in isolation, but rather elicits an intricate stress response that may become pathologic. Furthermore, genetic variation between individuals, in addition to genetic heterogeneity in tumors that develop, can play an important role in evaluating whether a targeted treatment will be beneficial, ineffective, or possibly even harmful. Thus, continued intensive study of the signaling pathways that respond to UVR will help lead to the development of more effective novel therapeutics in both the prevention and treatment of NMSC.
4E-BP1, 4E-binding protein 1; AK, actinic keratosis; AKT/PKB, protein kinase B; AP-1, activator protein-1; APAF-1, apoptotic protease-activating factor 1; ASK1, apoptosis signal-regulating kinase 1; ATM, ataxia telangiectasia-mutated protein; ATR, ataxia telangiectasia and Rad3-related protein; BAD, Bcl-2-associated death promoter; BAK, Bcl-2 homologous antagonist; BAX, Bcl-2-associated X; BCC, basal cell carcinoma; Bcl-2, B-cell lymphoma 2; Bcl-w, Bcl-2-like protein 2; Bcl-xL, Bcl-extra-large; BID, BH3 interacting-domain death agonist; BIM, Bcl-2-like protein 11; CDK1, cyclin-dependent kinase 1; c-FLIP, cellular FLICE-like inhibitory protein; CHK, checkpoint kinase; CUL3, cullin3 ubiquitin E3 ligase; DcR, decoy death receptors; DD, death domain; DISC, death-inducing signaling complex; DNA-PKcs, DNA-dependent protein kinase catalytic subunit; DR, death receptor; DSB, double-strand breaks; EGFR, epidermal growth factor receptor; ERK, extracellular signal-regulated kinases; FADD, fas-associated protein with death domain; FasR, Fas/APO1/CD95 receptor; FOXO3a, Forkhead box O3a; GRB2, growth factor receptor-bound protein 2; GSK3β, glycogen synthase kinase 3 beta; HIF-1α, hypoxia-inducible factor-1-alpha; HIPK2, homeodomain-interacting protein kinase 2; IGF-1-R, insulin-like growth factor 1 receptor; IKK, I kappa B kinase; IR, insulin receptor; IRAK, IL-1 receptor-associated kinase; Jak, Janus-associated kinase; JNK, c-Jun N-terminal kinases; MAP3K, MAP kinase kinase kinase; MAPK, mitogen-activated protein kinase; MDM2, mouse double minute 2 homolog; MEK, mitogen-activated protein kinase kinase; MKK, mitogen-activated protein kinase kinase; MKP, mitogen-activated protein kinase phosphatase; MOMP, mitochondrial outer membrane permeabilization; mTOR, mechanistic (mammalian) target of rapamycin; MyD88, myeloid differentiation primary response 88; NEK11, NIMA-related kinase 11; NEMO, NFκB essential modulator; NFκB, nuclear factor kappa-light-chain-enhancer of activated B-cells; NIMA, never in mitosis A; NMSC, nonmelanoma skin cancer; PDK1, phosphoinositide-dependent kinase-1; PH, pleckstrin homology; PI3K, phosphoinositide 3-kinase; PIKK, phosphatidylinositol-3-kinase-related kinases; PIP2, phospholipid phosphatidylinositol 4,5-bisphosphate; PIP3, phosphatidylinositol (3,4,5)-trisphosphate; PKCε, protein kinase C epsilon; PLK-1, polo-like kinase 1; PTEN, phosphatase and tensin homolog; PUMA, p53 up-regulator of apoptosis; RAF, rapidly accelerated fibrosarcoma kinase; RBX1, RING-box protein 1; ROS, reactive oxygen species; RTK, receptor tyrosine kinase; S6K1, S6 kinase beta-1; SCC, squamous cell carcinoma; SH2, Src homology 2; SHH, sonic hedgehog; SIAH1, seven in absentia homolog 1; Smo, smoothened; SOS, Son of Sevenless; SSB, single-strand breaks; STAT3, signal transducer and activator of transcription 3; tBID, truncated BID; TLR, Toll-like receptors; TNFR, tumor necrosis factor receptor; TNFR1, tumor necrosis factor receptors; TRAF6, TNF receptor-associated factor 6; TRAIL, TNF-related apoptosis-inducing ligand receptor; Trx1, thiol-disulphide oxidoreductase thioredoxin-1; UVR, ultraviolet radiation; XP, Xeroderma pigmentosum.
R.P.F. wrote the first draft of the manuscript, which was edited by L.M.S. Both authors approved the final version of the manuscript.
Work described in this manuscript from the authors' laboratory was supported by grants from the National Institute of Environmental Health Sciences, NIH [ES019242 to L.M.S. and ES026471 to R.P.F.] and the Pennsylvania Department of Health CURE Funds to LMS.
We thank Dr Amanda Nelson for critical reading of the manuscript.
The Authors declare that there are no competing interests associated with the manuscript.