MKK7 and ARF

Abstract

Sensing, integrating, and processing of stressogenic signals must be followed by accurate differential response(s) for a cell to survive and avoid malignant transformation. The DNA damage response (DDR) pathway is vital in this process, as it deals with genotoxic/oncogenic insults, having p53 as a nodal effector that performs most of the above tasks. Accumulating data reveal that other pathways are also involved in the same or similar processes, conveying also to p53. Emerging questions are if, how, and when these additional pathways communicate with the DDR axis. Two such stress response pathways, involving the MKK7 stress-activated protein kinase (SAPK) and ARF, have been shown to be interlocked with the ATM/ATR-regulated DDR axis in a highly ordered manner. This creates a new landscape in the DDR orchestrated response to genotoxic/oncogenic insults that is currently discussed.

Introduction

To protect their genome and avoid genomic instability, which is a hallmark of cancer,¹ cells have evolved signaling checkpoints that include DNA damage sensors, transducers and effectors, as well as repair proteins.² This signaling process comprises a complex network called the DNA damage response (DDR). Among the key apical kinases that have central roles in sensing and responding to chromosomal insults in mammalian cells are the ataxia telangiectasia mutated (ATM), ATM and Rad3-related (ATR), and DNA-dependent protein kinases (DNA-PK).²

ATM is mainly activated by DNA double-strand breaks (DSBs) and acts as a master regulator of the DDR network. It is generally accepted that a wide range of oncogenes activate the DDR due to forced entry into S phase, replication stress, and the resultant generation of DSBs.^3-6 To prevent accumulation of DNA damage and potential malignant transformation, ATM imposes p53-mediated anti-tumor barriers. These include oncogene-induced senescence (OIS), which is telomere attrition-independent, and apoptosis.⁷

Αn increasing amount of data delineates a series of interplays and time-coordinated actions of the apical DDR kinases, ATM and ATR, with other factors, like the alternative reading frame (ARF) tumor suppressor and stress-activated protein kinases (SAPKs), in response to activated oncogenes.^8-10 This information suggests that the DDR network is an oncogenic stress response “axis” that does not function alone. It is, rather, intertwined with other pathways responding to such stimuli through intricate loops and back-up mechanisms in a finely orchestrated DDR landscape (Fig. 1A). Many of these signaling routes converge mainly to p53 as the end DDR effector.^11-15 The current work provides a new perspective into the DDR scenery, taking into consideration the additional roles of the ARF tumor suppressor and of the MKK7 and p38 MAPK SAPKs, including unpublished data from our lab, in response to oncogene activation.

Figure 1. ATM, ATR, p38 MAPK, MKK7, and ARF are possibly interconnected in the activation of p53-mediated anti-tumor responses. (A) Oncogenes induce replication stress (RS), which, in turn, causes DNA damage including double-strand breaks (DSBs). The DNA damage response (DDR) cascade is stimulated culminating in the activation of the p53 tumor suppressor. Oncogenic stress is also known to induce ARF, which is another p53 stabilizer. The p38 MAPK plays a role as a complementary DNA damage response transducer. Apart from activating anti-tumor responses, p53 also upregulates Wip-1 phosphatase, which creates a feedback loop through targeting p38 MAPK for dephosphorylation and deactivation. Notably, p38 MAPK suppresses ARF. Among the best-studied negative feedback loops is that between p53 and MDM2.¹⁵ In this loop MDM2 is phosphorylated by ATM⁷⁵ and dephosphorylated by Wip-1.⁵⁹ (B) Within this network, MKK7 plays a role as a complementary DDR transducer, resembling that of p38 MAPK. It is also targeted by Wip-1 for dephosphorylation and deactivation. Overall, Wip-1 participates in self-regulatory circuits affecting multiple DDR molecules at any signaling level. (C) Activated ATM, by the DNA damage, counteracts the NPM/B23-mediated stabilization of ARF, so that ARF can be engaged at latter stages of cancer progression to halt tumorigenesis. ATR stimulated by RS negatively influences ARF levels possibly through interfering with nucleophosmin (NPM/B23) phosphorylation at Ser125 (see text). ATM/ATR and downstream pathways are activated earlier than ARF (as indicated by clock cartoons).¹⁰ −(p-), dephosphorylation; +(p-), phosphorylation.

MKK7 and ARF as New Partners of the DDR Pathway in p53 Stabilization

Post-translational modifications (PTMs) are regarded as a crucial step in p53 stabilization and activation in response to diverse stress signals.^12-15 These include phosphorylations, acetylations, methylations, sumoylations, ubiquitinations, neddylations, ADP-ribosylations, and glycosylations at various sites (Fig. S1). The existence of such a vast spectrum of PTMs has been proposed to reflect the ability of p53 to function as a hub that receives and integrates a wide array of signals. In turn, proper differential response(s) are elicited according to the type and extent of the exerted stimulus.^13-15 Among the best-studied PTMs are those affecting the N-terminal (transactivation) domain of p53 in response to DNA damage/UV light/radiation.^13,14 For example, a broad range of DDR kinases, including ATM, ATR, DNA-PK, Chk1 and Chk2, and p38 MAPK, a SAPK, phosphorylate this domain, stabilizing p53 (Fig. S1).^13,16

Interestingly, many of the PTMs modulating p53 activity function via several auto-regulatory loops, either by self-sustained or self-limiting feedback circuits.^11,12,14 Some of these controls are mediated by E3 ubiquitin ligases and include mouse double minute-2 (Mdm2), constitutively photomorphogenic 1 (Cop-1), ARF-binding protein 1 (ARF-BP1), and p53-induced RING-H2 protein (Pirh-2), which target p53 for ubiquitination and proteasomal degradation.^11,14,15 Another well-established negative feedback circuit is mediated by the type 2C Ser/Thr phosphatase named wild-type p53-induced phosphatase 1 (Wip-1), which is also a p53-regulated gene. Notably, Wip-1 acts to reverse the activity of p38 MAPK, and also of several components of the DDR pathway (γH2AX, ATM, Chk2) (Fig. 1A).^17,18

Recently, evidence was provided for the critical role of another SAPK, namely the MKK7, as a “molecular bridge” conveying the oncogenic stress stimuli––at least those associated with the expression of KRas^G12D and NeuT––to p53.⁸ As discussed in the following section, this new link is separate from the canonical DDR axis. Given the ability of Wip-1 to deactivate the DDR pathway,^18-20 an important question is whether MKK7 is dephosphorylated in a similar manner. If this holds true, it implies that the MKK7-mediated signaling branch responding to oncogenic stress is timely orchestrated with the DDR. Interestingly, Wip-1 was reported to suppress the ARF key tumor suppressor via inactivation of the p38 MAPK.²¹

The INK4/ARF locus encodes 2 different tumor-suppressive products, the cyclin-dependent kinase inhibitor (CDKI) p16^INK4a and ARF (p14^ARF in humans and p19^ARF in mice).²² Interestingly, ARF is short lived in normal human cells but stable in cancer cells.²³ It predominantly localizes in the nucleolus in complex with nucleophosmin/B23 (NPM/B23), which stabilizes it.^24,25

One way by which ARF accomplishes its tumor-suppressive tasks is through activating the p53 protein.²⁵ This involves several E3 ubiquitin ligase-mediated mechanisms that either control p53 levels or its own stability. Mdm2 and ARF-BP1 are E3 ubiquitin ligases that are blocked by ARF, thus protecting p53 from degradation.^26-28 On the other hand, ULF (ubiquitin-protein ligase for ARF), Siva 1, and MKRN1 (Makorin RING finger protein 1) indirectly regulate p53 levels by targeting ARF for degradation.²⁹ Nevertheless, ARF can also exert p53-independent tumor-suppressive tasks, mainly through inhibition of ribosome biogenesis.^30,31 Notably, ARF expression is negatively regulated by p53,³² highlighting a cross-talk between the 2 tumor suppressors.²⁵

ARF is known to respond to various oncogenes, like Ras, E2F1, and Myc, leading to activation of p53 as described above.²⁵ The predominant view until now was that the DDR and ARF act independently of each other. Interestingly, recent evidence indicates that ARF expression must attain specific threshold levels in order to exert its tumor suppressor function. This is finely balanced by the challenging oncogenic load and ATM (unpublished results; see further in the text).^9,10 As a result, in human and animal tumor samples a time order is established between DDR and ARF activation,¹⁰ leading to an orchestrated ARF back-up mechanism counteracting tumor progression.

MKK7 as a novel oncogene-induced DDR player: evidence for auto-regulation

MKK7 is a dual-specificity mitogen-activated protein kinase kinase (MAPKK). It physically associates with c-Jun NH2-terminal protein kinase (JNK) and preferentially targets Thr amino acid residues within its activation loop for phosphorylation, thereby stimulating its activity.^33-35

A tissue-specific conditional inactivation strategy enabled the delineation of the oncosuppressive role of MKK7/JNK/c-Jun in murine lung and mammary cancer.⁸ Specifically, MKK7 was demonstrated to significantly delay the onset of early-stage tumors, as well as to impede the progression to malignancy in mouse models of lung and mammary cancer driven by KRas^G12D and NeuT overexpression, respectively. Regardless of the initial oncogenic stimulus and the origin of the tissue, the oncogenic challenge could activate the DDR, as evidenced by the detection of phosphorylated Chk2 or the presence of γH2AX and 53BP1 foci. Importantly, a marked reduction in p53 protein levels was observed upon MKK7 deletion. This was not due to impaired TP53 transcription but to the lack of stabilizing phosphorylations at serines 6 and 33 at the N-terminal end of p53 that were known to be mediated by JNKs. Notably, these are different residues from the typically phosphorylated Ser 15 and Ser 20 targeted by the classical DDR axis (Fig. S1). In the absence of MKK7, p53 was found to be unable to fulfill its tumor-suppressing tasks—that is the activation of the G₂/M cell cycle checkpoint and the induction of OIS. In this way, tumor burden as well as malignant progression was increased in animals with an MKK7-deficient background, thereby decreasing their mean survival.⁸ A question that remains to be answered concerns which MKK7 isoform(s) is/are participating in this pathway. Notably, the Map2k7 locus, which encodes MKK7, is alternatively spliced to produce 6 different kinase isoforms.³⁴ Also, the components upstream of MKK7 that are responsive to oncogenic stress are still unknown (Fig. 1B).

The fact that RAS harbors oncogenic mutations in ~30% of human malignancies,³⁶ while Her2 (Neu) is estimated to be overexpressed in 30% of breast tumors of human origin,³⁷ indicates that the above described MKK7/JNK/p53 axis is relevant in a wide range of tumors (Fig. 1B). Moreover, it suggests that MKK7 could be a candidate target for pharmacological intervention (activation) in tumors carrying wild-type p53. In contrast, the potential usefulness of JNK inhibitors, which are thought to be of therapeutic importance in many diseases,³⁸ should be reassessed. This is because in cancer both tumor-promoting and tumor-suppressive JNK functions have been reported.³⁹

As the DDR and MKK7 signaling routes can be activated in parallel by oncogenic stress, and considering the known DDR/p53/Wip-1 and p38/p53/Wip-1 negative feedback loops, an emerging question is whether MKK7/p53/Wip-1 function in a similar manner. Data from human non-small cell lung carcinomas (NSCLCs) harboring wild-type p53 showed that the detected phospho-MKK7 levels were minimal, as compared with their normal counterparts.⁸ On the other hand, in NSCLCs samples with mutant p53, the phospho-MKK7 levels were increased relative to the corresponding histologically normal counterparts. This finding suggested a disruption in the MKK7/JNK/p53 signaling axis that could be explained by the inability of mutant p53 to activate a potential MKK7/p53/Wip-1 negative feedback loop. Unpublished data from the same NSCLCs cases imply that Wip-1 may indeed be the missing piece of the puzzle, since the increase in phospho-MKK7 coincided with loss of Wip-1 expression in the majority of these samples (Fig. 2A).

Figure 2. Evidence for an MKK7/p53/Wip-1 negative feedback loop. (A) Immunoblot (IB) analysis of MKK7 (rabbit polyclonal antibody, 1:1000 dilution, Cell Signaling Technology Inc, #4172), phosphorylated (P) MKK7 (rabbit polyclonal antibody, 1:1000 dilution, EMD Millipore, #36-013) and Wip1 (rabbit polyclonal antibody, 1:1000 dilution, kind gift from Dr E Apella) in human non-small cell lung carcinomas (NSCLC) stratified according to p53 status and in adjacent normal tissues. Secondary anti-rabbit (R&D Systems) horseradish peroxidase (HRP) conjugated antibody was used. Signal development was performed by employing enhanced chemiluminescence (ECL, Thermo Fisher Scientific). An amount of 50 μg of total protein/sample was loaded for IB analysis. (B) IB analysis of MKK7, phosphorylated (P) MKK7, and p53 (mouse monoclonal DO-7 antibody, Santa Cruz Biotechnology, sc-47698) in A549 cells treated with nutlin-3a (Cayman Chemichal Company, 18585), followed by administration of the the highly selective Wip-1 phosphatase inhibitor CCT007093⁴⁰ (Calbiochem) for the time points indicated. An amount of 20 μg of total protein/sample was loaded for IB analysis that was performed as described above. (C) Wip-1 siRNA silencing in A549 cells. A549 cells were grown on DMEM at a density of 1.2 × 10⁵ cells per plate. Twenty-four hours later, cells were treated with short interfering RNA (siRNA) against Wip-1 (Wip-1 stealth RNAi-set of 3 oligos), negative control (Stealth^™ RNAi Negative Control Med GC) using Lipofectamine 2000 (all reagents obtained from Invitrogen) and following the manufacturer’s instructions. IB analysis was performed as described in (B). Actin (rabbit polyclonal antibody, 1:1000 dilution, Cell Signaling Technology Inc, #4967) serves as loading control. (ctrsi: control siRNA) (Nutlin: nutlin-3a).

To functionally recapitulate and validate the in vivo findings, the A549 human lung carcinoma cells were treated with the highly selective Wip-1 phosphatase inhibitor CCT007093⁴⁰ after administration of nutlin-3a, an Mdm2 antagonist that leads to p53 stabilization. The concurrent treatment showed that the expression of phospho-MKK7 undergoes an increase from 2 h post-treatment (Fig. 2B). Similar results were obtained by silencing of Wip-1, which also led to increased phospho-MKK7 levels (Fig. 2C). Hence, these data pinpoint the existence of a novel MKK7/p53/Wip-1 feedback loop that adds further complexity to the regulation of p53-dependent responses in tumorigenesis. Even more, this suggests that the MKK7 signaling branch may act synchronized with the DDR pathway at both activation and deactivation levels.

Why cells have evolved the MKK7-mediated signaling as a complementary pathway to the DDR response? Notably, and as already mentioned, p38 MAPK responds in an analogous way to a variety of stress stimuli, forming also a p38 MAPK/p53/Wip-1 negative auto-regulatory feedback loop.^11,16,17 A further level of complexity arises from how and when the MKK7 and/or p38 MAPK signaling branches are selected to function along with the DDR axis (Fig. 1B). A potential explanation for the existence of these additional circuits overlaying with the DDR response may stem from the fact that these pathways share p53 as a nodal end effector.^7,12,15 As previously described, p53 is known to be an integrator of a variety of upstream stress signals followed by a differential response that should be the most proper for the fate of a cell.^11-15 In this context, it is plausible to hypothesize that the oncogenic/genotoxic stimuli exert multiple stressing effects.⁴¹ To elicit the appropriate response, p53 should integrate exactly this spectrum.¹⁵ Thus, the DDR pathway should be “assisted” by complementary ones, enabling p53 to “sense” the variety of these stresses and to achieve maximal signaling coordination through its exact PTM stabilization (Fig. 1B).

Interconnections among DDR and ARF

Although ARF has long been assigned as an “oncogenic sensor,”^42,43 its role in tumor suppression was initially thought to be independent from that of the DDR pathway.^43-45 However, emerging evidence supports an interplay among ARF and the key DDR checkpoint kinases ATM and ATR.

Since oncogenic stimuli can activate the DDR pathway as well as ARF,^7,43 an important issue is whether these pathways are triggered concomitantly or at different stages during cancer development. To answer this question, we examined their status in various types of human and mouse models of epithelial tumors covering the entire histopathological spectrum of cancer development.¹⁰ The findings demonstrated that the γH2AX, phospho-ATM, or phospho-Chk2 DDR markers were positive from the earliest preneoplastic lesions. In contrast, ARF protein was detected in more advanced stages of cancer development, considerably later than DDR and even p16^INK4a, the tumor-suppressor gene encoded also by the INK4/ARF locus.²² These in vivo observations were reinforced upon challenging human bronchial epithelial cells and normal fibroblasts with various oncogenes, alone or in combination.¹⁰ Whereas a single oncogene could efficiently trigger a DDR response, ARF induction required an escalating oncogenic load, and this was transcriptionally based. Notably, escalating oncogenic stimuli activated both pathways in the employed cellular models.¹⁰ Similarly in another oncogenic setting, low levels of Myc were sufficient to induce proliferation and promote transformation, but did not activate ARF.⁴⁶ Only very high levels of Myc expression could induce ARF and, consequently, apoptosis.⁴⁶

What is the physiological significance of these findings? Possibly ARF may need a high threshold for accumulation in response to oncogenes in order to allow their low levels of proliferative activity to fulfill normal physiological cellular processes. If ARF would be induced by low mitogenic signals, then tissue homeostasis would be affected by a massive and unnecessary loss of functional cells. On the other hand, if such a low mitogenic activity spontaneously exerted an oncogenic effect that would probably be limited in terms of both cell numbers and extent of DNA damage, in this case the higher sensitivity of the DDR pathway would render it functional, allowing repair or restricted elimination of these cells without affecting the whole tissue. While this process supports and preserves normal tissues, at the same time it controls premalignant cells.⁷ Nevertheless, during cancer development, the progression to overt tumors is accompanied by an increased oncogenic burden.⁴⁷ This would require a more effective eradication of abnormally proliferating cells harboring such active oncogenes to prevent their accumulation. Thus, the increased oncogenic load taking place during tumor progression may essentially signal the entrance of ARF in the battlefield for tumor annihilation. Therefore, could ARF represent a “second barrier” that is activated when the upstream components of the DDR pathway are compromised?

Until now the predominant view was that the DDR and ARF pathways were functionally unlinked. Recent data demonstrated that the DDR pathway is responsible for maintaining the threshold for ARF protein at high levels.⁹ Central to this mechanism is the DDR kinase ATM. Upon DNA damage induction, ATM activates the serine–threonine protein phosphatase 1 (PP1) through phosphorylation of its inhibitory subunit I-2.⁴⁸ In turn, PP1 dephosphorylates NPM/B23, while at the same time suppresses Nek2A kinase. Through this combined action, PP1 weakens the biochemical interaction between ARF and NPM/B23, releasing the former into the nucleoplasm. There, unprotected ARF is downregulated. Conversely, ATM inhibition or silencing reverses this effect. Specifically, PP1 is not activated, allowing Nek2A to phosphorylate NPM/B23. As a result, ARF is strongly bound by NPM/B23 and stabilized in the nucleolus. NPM/B23 is a highly negatively charged protein containing many phosphorylation sites. To define the amino acid residues responsible for ARF binding to NPM/B23 that are regulated by ATM/PP1/Nek2A, a mass spectrometry-based comparative phosphoproteome analysis was employed. The results demonstrated that Ser 70 and Ser 88 were the critical ones. We also interrogated additional NPM/B23 amino acid residues with phosphospecific antibodies against serines 4 and 125 along with threonines 95, 199, 234 and 237 (Fig. S2). Although these sites are targeted by various kinases, including cyclin-dependent kinases, none of these sites showed increased phosphorylation.

A major question was how ARF is downregulated into the nucleoplasm upon release from NPM/B23. Searching a database of proteins phosphorylated by ATM,⁴⁹ TBP-1-IP (Tat binding protein-1 interacting protein) and TRIP12 (thyroid hormone receptor interactor protein 12) were found as a potential link between ATM and ARF stabilization. TBP-1-IP (Tat binding protein-1 interacting protein) proved to have no effect on ARF levels (Fig. S3). Investigating TRIP12, also known as ULF, which is a nucleoplasmic E3-ubiquitin ligase that mediates lysine-independent ubiquitylation of ARF,⁵⁰ revealed that ATM regulated the access of ARF to ULF.⁹ Hence, ULF was responsible for ARF degradation in the nucleoplasm upon ATM-dependent release from NPM/B23.^9,50

Thus, if ATM is lost or inhibited, ARF is NPM/B23 stabilized and exceeds the threshold required for its accumulation. Consequently, it can exert its anti-tumor activities in both p53-dependent and -independent manner. Of note, in p53-null cellular and xenograft models, the ATM-dependent ARF upregulation indeed restrained ribogenesis and cell cycle progression, supporting the biological significance of the deciphered ATM/ARF link.^9,30,31 Also, in human clinical samples, loss of ATM expression correlated with increased ARF levels, further delineating the importance of this mechanism.⁹ Even more, in p53 mutant tumors this ATM-dependent mechanistic trigger of ARF activity as a second tumor barrier could be exploited at therapeutic level.⁹

In another compatible scenario of escalating oncogenic load, recent data have shown that Myc binds to ULF, thereby preventing an ULF/ARF interaction.²³ Once Myc attains high levels it protects ARF from ULF degradation, despite the fact that DDR activation shuttles it out from the shelter of the nucleolus.

Although both the DDR machinery and ARF are well-established oncogenic sensors, they do not always respond in the same manner upon such challenges. In particular, overexpression of the replication licensing factor Cdc6 is considered, on the one hand, as an oncogenic stimulus that triggers replication stress promoting genomic instability,^5,7,51,52 but on the other it has been reported to repress the INK4/ARF locus.⁵³ Given that oncogenic Cdc6 activates the DDR checkpoint^5,51 and ATM downregulates ARF⁹ we were intrigued to assess which Cdc6-dependent ARF-suppressive pathway is more important. Challenging Saos2 cells with Cdc6 we noticed that downregulation of endogenous ARF was restored to almost initial levels upon ATM silencing, denoting the significance of the ATM–ARF interrelation (Fig. 3).

Figure 3. ATM activation is more effective than Cdc6 overexpression in repressing ARF. Saos2 cells were transiently transfected with Cdc6 and siATM. Western blot analysis of Cdc6, ATM and p14^ARF. Actin (rabbit polyclonal antibody, 1:1000 dilution, Cell Signaling Technology Inc., #4967) serves as loading control. p14^ARF and actin were probed as described above. Reagents and immunoblotting for Cdc6 and ATM as well as siRNA knockdown of Cdc6 and ATM have been previously described.^9,52 (ctrsi, control siRNA).

Emerging evidence suggests that ARF is also linked with the replication stress associated ATR kinase. A first demonstrated cross-talk was that between ARF and an ATR-depedent suppression of the Rel-A subunit of NF-κB.⁵⁴ Recently, replication stress (RS), caused either by ATR silencing or treatment with hydroxyurea (HU), a ribonucleotide reductase inhibitor, was shown to upregulate both p16^INK4a and ARF in a transcriptional-dependent manner.⁵⁵ Another study showed that, in cells depleted of BRCA2, ATR activation induced ARF expression at the post-translational level.⁵⁶ Based on a report demonstrating that ATR phosphorylates NPM/B23 at Ser125 (p-125) we asked whether this PTM could play a role in ATR-mediated ARF induction by increasing ARF’s protein stability.^57,58 Surprisingly, however, HU treatment of H1299 lung cancer cells showed decreased phosphorylation of NPM/B23 at Ser 125 followed by a clear reduction in ARF expression (Fig. 4), questioning the precise action of ATR on ARF regulation. Further studies are warranted to clarify this issue. The scenario that an unknown phosphatase targeting p-Ser125 NPM/B23 leading to ARF reduction, resembling the interplay between ATM and ARF,⁹ should be considered.

Figure 4. Evidence for an ATR-dependent ARF suppression that is mediated by reduced NPM/B23 Ser 125 phosphorylation. (A) Hydroxyurea (HU) (#127-07-1, Sigma-Aldrich) induced replication stress in H1299 cells was assessed by immunoflourescence (IF) analysis of the single-stranded DNA binding RPA protein subunit RPA32. For IF analysis of RPA32 a mouse monoclonal anti-RPA32 antibody was used (1:1000 dilution, Santa Cruz Biotechnology, Inc, sc166886). 4′,6-Diamidino-2-phenylindole (DAPI) was obtained from Sigma-Aldrich. IF was performed as was previously described.⁹ (B) Immunoblot analysis of Ser125-NPM/B23 phosphorylation (rabbit monoclonal antibody, 1:10000 dilution, Abcam, ab109546) and p14^ARF (mouse monoclonal antibody, 1:100 dilution, Abcam, ab49166). An amount of 20 μg of total protein/sample was loaded. Secondary antibodies were from R&D Systems and signal development was performed by employing enhanced chemiluminescence (ECL, Thermo Fisher Scientific). Actin (rabbit polyclonal antibody, 1:1000 dilution, Cell Signaling Technology Inc, #4967) serves as loading control. Scale: 10 μm.

In the BRCA2 depleted context, Carlos et al. also displayed that ARF mediated p53-dependent activation of the dual specific phosphatases, DUSP4 and DUSP7, that suppressed the MARK-ERK1 signaling cascade.⁵⁶ The later finding resembles the previously described negative regulatory loop effect of Wip-1 on the MAPK kinases MKK7 and p38 (Fig. 1).¹⁷ Interestingly, Wip-1 appears to be another connective link between DDR and ARF, since, apart from de-activating (dephosphorylating) multiple DDR molecules,^18,59 it also suppresses the INK4/ARF locus via p38 MAPK.^2,21,58,59 In various human tumors, Wip1 is either aberrantly expressed or mutated, exhibiting an oncogenic behavior.⁵⁹ The unique physiological positioning of Wip-1 in these failsafe mechanisms seems to make it the “Achilles’ heel” for both of them.

Two additional publications suggest that ARF is involved in activating the ATM/ATR pathway.^60,61 Nevertheless, an important caveat in these works is the strong exogenous overexpression of ARF in the employed cellular systems. The high isoelectric point (pI = 14) of ARF makes it a very “sticky” protein⁶² that when overexpressed artificially could mask the true functional interplays that take place. Furthermore, the in vivo study showing a correlation between ARF and p-Chk2 (Thr68) expression in human lung tumors⁶¹ was not performed either by serial section analysis or by double immunohistochemistry. Since ATM was shown to suppress ARF expression, such an in situ experimental approach is essential to avoid misleading results.⁹

Recently, transcriptional activation of ARF in response to unrepaired DNA strand break (SB) accumulation was reported.^59-63 The authors suggest that poly(ADP-ribose) synthesis catalyzed by PARP1 at sites of unrepaired DNA activates ARF. This is achieved through a signaling cascade involving decreased activity of the NAD+-dependent deacetylase SIRT1 (silent information regulator 1) followed by activation of the transcription factor E2F1, a known ARF inducer. Intriguingly, many of the experimental evidence provided depict very small-scale amplitude changes in ARF’s mRNA and protein levels, as well as SIRT1’s activity. Even more, there is a lack of information regarding any biological outcome, like senescence or apoptosis, that could support the potential significance of the proposed model. As the DDR checkpoint is activated from the earliest stages of cancer,^3,4,7 and ATM, according to a recent report, inhibits SIRT1,⁶⁴ a prediction of the signaling cascade proposed by Orlando and colleagues would be the permanent upregulation of ARF throughout cancer development. Nevertheless, and in contrast to this prediction, a study employing mice models and human clinical samples covering the whole spectrum of carcinogenesis demonstrate that ARF is robustly activated only at advanced stages of cancer.¹⁰

In general, the aforementioned studies shed light on a novel cross-talk between the DDR- and ARF-dependent anti-tumor barriers, showing that cells evolve parallel and stratified defense lines to deal with potential cancer-risk threats. The questions arose especially regarding the ATR-ARF branch seek answers to elucidate the mechanistic details that preside over this vital functional link.

Conclusions and Prospects

It is evident that in response to various genotoxic insults, like the oncogene-induced replication stress, the DDR is not a simply linear reacting pathway (Fig. 1C). While providing a main response axis for a wide range of insults, it is often complemented and assisted by many other pathways at multiple levels. A first level is in sensing and relaying through the proper signaling routes the full spectrum of stressogenic stimuli. Subsequently, these signals must be integrated, with p53 being a prominent end effector performing this role due to its ability to receive a vast range of PTMs.^12-15 Finally, transduction of these responses must be accompanied by various feedback loops that either functionally release the engaged network or further activate it. This also creates stratified cellular responses, like in the case of ATM and ARF.^9,10

In the case of MKK7, as evidenced by a bioinformatic analysis (Fig. S4), this pathway may possibly be modulated by factors that apart from their oncogenic ability can exert additional effects. Specifically, they are involved in the formation of reactive oxygen species (ROS), such as H-Ras and Rac1,^65,66 or in the case of glutathione S-transferase P1 (GSTP1) and IGF-1, provide protection against oxidative stress.^67-69 This is consistent both with the fact that MKK7 is responsive to oxidants⁷⁰ as well as with the role of p53 as an integrator of diverse cellular stress stimuli.^71,72 Hence, the implication of MKK7 in pathways depending on pro- or anti-oxidants and in redox control may also contribute to p53 stabilization, complementary to the DDR machinery.

Importantly, data stemming from this in silico analysis (Fig. S4) corroborate the notion for a putative signaling relationship among MKK7 and ARF, since they both share these upstream activators. Furthermore, the bioinformatic analysis also revealed potential commonly downstream-regulated protein targets (Fig. S4; Table S1). The existence of potential common upstream inducers for MKK7 and ARF underlines the functional importance of ARF as a type of “back-up” tumor suppressor, in case the DDR pathway, possibly including the interplay with MKK7, collapses during cancer progression (Fig. 1C).

Response to the most lethal genotoxic insult, i.e., DSBs, is through local accrual of proteins in the damaged region, which takes place very fast at the defined chromatin region by means of a cascade of PTMs.^73-75 The termination of DNA damage checkpoint and recovery of cell division is known to rely on molecules carrying out the reversal of these PTMs. Among these molecules, Wip-1, which targets multiple DDR components, has been suggested to be one of the rate-limiting factors during this process.¹⁸ On the other hand, Wip-1 might serve for fine-tuning of DDR, resulting in a well-organized genome-protective signaling network. In whatever way the observed Wip-1-mediated dephosphorylation affects DDR and MKK7 activity, experimental data highlight Wip-1 as a homeostatic rheostat and a pivotal intervention point in the field of cancer therapeutics having a wider role in carcinogenesis than that reported until now.⁵⁹ Together, DDR kinases, ARF, MKK7, and Wip-1 are all key players in the coordination of p53-dependent tumor suppression, functionally and temporally interconnected via relationships whose complexity may be much higher than it is currently known. Characterization and expansion of these potent signaling interconnections as a future task will enable the better understanding of the physiological basis of DDR as well as its therapeutic (either pharmacological or biotechnological/genetical) manipulation in the battle against cancer.

Glossary

Abbreviations:

DDR

DNA damage response

SAPK

stress-activated protein kinase

ATM

ataxia telangiectasia mutated

ATR

ATM and Rad3-related

DSBs

double-strand breaks

OIS

oncogene-induced senescence

JNK

c-Jun NH2-terminal protein kinase

RS

replication stress

PTMs

post-translational modifications

Mdm2

mouse double minute-2

MAPKK

mitogen-activated protein kinase kinase

NSCLC

non-small cell lung cancer

siRNA

short interference RNA

HU

hydroxyurea

MEFs

mouse embryonic fibroblasts

NPM/B23

nucleophosmin

PP1

protein phosphatase 1

TBP-1-IP

tat binding protein-1 interacting protein

TRIP12

thyroid hormone receptor interactor protein 12

ROS

reactive oxygen species

siRNA

short interfering RNA

DNA-PK

DNA-dependent protein kinase

CSN

COP9 signalosome

TAF1

TBP-associated factor 1

HIPK2

homeodomain-interacting protein kinase 2

CK1 and CK2

casein kinase 1 and 2

PKC

protein kinase C

GSTP1

glutathione S-transferase pi 1

IGF-1

insulin-like growth factor 1

Tip60

tat-interactive protein 60