p21-Activated Kinases in Thyroid Cancer

Abstract

The family of p21-activated kinases (PAKs) are oncogenic proteins that regulate critical cellular functions. PAKs play central signaling roles in the integrin/CDC42/Rho, ERK/MAPK, PI3K/AKT, NF-κB, and Wnt/β-catenin pathways, functioning both as kinases and scaffolds to regulate cell motility, mitosis and proliferation, cytoskeletal rearrangement, and other cellular activities. PAKs have been implicated in both the development and progression of a wide range of cancers, including breast cancer, pancreatic melanoma, thyroid cancer, and others. Here we will discuss the current knowledge on the structure and biological functions of both group I and group II PAKs, as well as the roles that PAKs play in oncogenesis and progression, with a focus on thyroid cancer and emerging data regarding BRAF/PAK signaling.

Thyroid cancer is the most common malignancy of classical endocrine tissues and its incidence has risen since the 1970s, with a more dramatic rise over the past 3 decades. It is estimated that there will be 52 890 new cases and 2180 deaths in 2020 in the United States (1). The more rapid rise in incidence can be attributed mostly to an increase in use of neck ultrasound and other radiographic methods in clinical practice. Notably, while the overall incidence of thyroid cancer has stabilized in the past several years with changes in clinical guidelines, the rate at which patients succumb to thyroid cancer has continued to increase (2-4). These data suggest that current therapies for more aggressive forms of thyroid cancer have not improved survival rates for the disease. Indeed, beyond surgery, thyrotropin (TSH) suppression and iodine-131 therapy (treatments that have been used for decades), the current Food and Drug Administration-approved and experimental treatments improve progression-free survival but have not caused durable complete remissions from the disease. This emphasizes the need for improved therapeutic targeting, defining mechanisms of resistance to these compounds, and for better stratification of thyroid cancer to select patients for potential earlier targeted intervention (5, 6).

Approximately 95% of thyroid cancers derive from follicular thyroid cells and, of those, ~90% are classified as well differentiated, which are further subclassified as follicular thyroid cancer, papillary thyroid cancer (PTC), and Hürthle cell thyroid cancer (HCC). PTC is the most common histological subtype of thyroid cancer and overall has an excellent prognosis due to its typical early stage at diagnosis and slow growth pattern. However, due to its high frequency, PTC also accounts for the largest number of patients with more aggressive thyroid cancer. PTC development is dependent primarily upon activation of the mitogen activated protein kinase (MAPK) pathway via activating mutation in BRAF (most commonly resulting in BRAF^V600E), RAS mutations (mostly NRAS), and gene rearrangements such as RET/PTC that are found mostly in pediatric and/or radiation-induced thyroid cancers (7). By contrast, follicular thyroid cancers are most often characterized by activation of PI3K signaling (8), and Hürthle cell thyroid cancers are characterized by a unique pattern of chromosomal loss and activation of mTOR signaling (9). In all cases, as the cancers become more aggressive and more poorly differentiated, they typically gain mutations in the hTERT promoter, TP53, and display genomic instability and reduced expression of thyroid differentiation proteins such as the sodium–iodide symporter and the TSH receptor; in rare cases these can evolve into anaplastic cancer. Effective treatment options for these tumors become progressively more limited as they dedifferentiate.

p21-Activated kinases (PAKs) are a family of serine/threonine kinases that were first identified in mouse brain tissue in 1994 (PAK1) (10). Functional roles for these proteins have been reported in the regulation of the cytoskeleton and cell motility, proliferation, apoptosis, and gene transcription, as well as oncogenic transformation (11). It has been demonstrated that PAKs are overexpressed and/or overactivated in thyroid, pancreatic, colorectal, breast, and ovarian cancers (12). There are currently 6 known isoforms of PAK that are divided into 2 groups based on their structural homology and their ability to be activated by Rho family GTPases (13, 14). Group I consists of PAKs 1-3 and group II consists of PAKs 4-6. Current findings that have been made in the biological functions of the 2 groups of PAK will be summarized here, with particular focus on the molecular pathways involved in transformation and progression in thyroid cancer.

Group I PAK Structure and Function

Group I, and to a lesser extent group II, PAKs are homologous to the STE20 family of proteins found in yeast, where they have overlapping roles in extracellular signal transduction and cytoskeletal rearrangement, indicating an ancient origin and conserved function (15). All PAKs, regardless of group assignment, possess a N-terminal p21-binding domain (PBD), alternatively known as a CDC42 and RAC1 binding (CRIB) domain, a C-terminal kinase domain, and an isoform-dependent number of conserved proline-rich regions (Fig. 1). PAKs 1, 2, and 3 also possess an Auto-Inhibitory Domain (AID) that overlaps with the PBD. Group I PAKs share greater than 88% sequence similarity of the PBD/AID region and over 93% in the kinase domain (13). Another specific feature of group I PAKs is the presence of 3 separate SH3 (Src homology 3) binding domains, 2 of which bind to well-known adaptor proteins Nck (16) and Grb (17) and the other binds to β-PIX (PAK interacting exchange factor, which is also referred to as Rho guanine nucleotide exchange factor 7 [ARHGEF7]) (15). These adaptor proteins recruit signal transduction intermediaries to receptor tyrosine kinases, thereby coupling extracellular messages to signal transduction cascades within the cell. Thus, it is very likely that these are the primary recruiters of group I PAKs to the plasma membrane for activation (18, 19). Supporting this concept, disruption of the Nck and PAK1 interaction inactivates the kinase (20), while constitutive recruitment of PAK1 to the plasma membrane by myristoylation of the SH3 domain results in activation of PAK1 (21). PAK1 also contains 3 nuclear localization sequences in the N-terminal domain, allowing these PAKs to enter the nucleus and interact with target chromatin (22) and histone H3 (22, 23).

Figure 1.

PAK isoform structure. Linear representation of groups I and II PAK proteins, with the numbers to right indicating the total number of amino acid residues. Overlapping CRIB (CDC42/RAC interactive binding) domain, also known as the PBD, and AID (Auto-Inhibitory Domain) for group I PAKs is shown in purple. Separate CRIB domain and AID for group II PAKs are shown in blue and aquamarine, respectively. Nck and PIX binding regions are shown in green and orange, respectively. The kinase domain is shown in plum and the catalytic residue is represented by the star, with the corresponding residue number above. The vertical bar (|) indicates proline-rich segments that are potential binding regions for proteins with a SH3 domain. The asterisk indicates that the AID for these proteins, PAK5 and 6, may not actually serve as an inhibitory domain and needs further characterization. Drawing is not to scale.

As described in detail below, group I PAKs are thought to exist in 2 states: (1) as an inactive dimer in which kinase activity is autoinhibited by binding of the AID of a second PAK molecule, and (2) as an active monomer via phosphorylation of the kinase domain. Genetic, structural, and biochemical studies provide evidence for asymmetric homodimerization of group I PAKs leading to inactivation. In this model, the kinase domain of 1 PAK binds to the AID of the other monomer and vice versa, limiting the ability of each kinase to autophosphorylate, a required step for activation 24-26). Other findings also indicate that the 2 inactive monomers exist in different conformations, similar to an enzyme–product complex, and that this is important in the transition from inactive to active states via trans autophosphorylation (27). However, this traditional model of dimerization as the primary means of PAK inactivation has recently been challenged by solution-phase structural analysis, which unexpectedly showed that over 95% of the unphosphorylated PAK1 is in a monomeric state, with concordant results across multiple analytical methods (Sedimentation velocity analytical ultracentrifugation, Liquid chromatography -mass spectrometry, and High performance liquid chromatography-small angle X-ray scattering) (28). High performance liquid chromatography-small angle X-ray scattering data indicated that the unphosphorylated monomers were more compact than their phosphorylated counterparts (28). These findings suggest that group I PAK autoinhibition may also be intramolecular due to folding of PAK monomers.

The transition to an active monomer is primarily initiated by the binding of GTP-bound CDC42 or RAC1 to the PBD, destabilizing the interaction between the AID and kinase domain (29). Multiple autophosphorylation events are necessary for full kinase activity. There are serine and threonine residues between the PBD/AID region and the kinase region that when autophosphorylated are thought to act as a multistep activation switch and are necessary for autoinhibition suppression and maintenance of full activity (25, 26). This transition also results in the phosphorylation of the catalytic threonine residue in the kinase domain (Thr423 in PAK1, Thr402 in PAK2, and Thr421 in PAK3), often serving as a marker of group I PAK activation (29, 30). Alternate modes of activation that may not require GTPase binding have been described, but their physiological relevance is less understood, including autoinhibitory structure disruption via binding of sphingosine (31), direct phosphorylation by AKT (32), PDK1 (33), and JAK2 (34), caspase-3 binding during apoptosis for PAK2 (35), and via binding of PIX with subsequent recruitment to GIT1 (36). Inactivation of group I PAKs occurs through dephosphorylation by phosphatases (37, 38), direct binding and stabilization of the kinase domains (39), and via phosphorylation by PKA (40). Specifically, multiple members of the POPX family of phosphatases, PP1 and PP2, dephosphorylate PAKs both at autoinhibitory serines and at the catalytic threonine in the kinase domain, depending on the PAK isoform (37, 38). In addition, PAK activation can be blocked by the binding of proteins that interfere with access either of GTPases to the PBD/AID region, as is the case with Merlin (41), hPIP1 (42), and CRIPak (43), or that interfere with the kinase domain, such as PITSLRE (44) and Nischarin (45).

Group I PAKs have distinct tissue-specific expression patterns. PAK1 is expressed is most tissues, but to a greater extent in thyroid, brain, and muscle (13). PAK2 is ubiquitously expressed and PAK3 is the most restricted of group I, expressed primarily in brain, cartilage, liver, and spleen (13). Bone marrow, pancreas, peripheral nervous system, and thymus tissues show little to no expression of either group I or group II PAKs (13). Group I PAKs have been shown to be especially important in both neural and cardiac development and function (11).

PAKs are central nodes in a number of signaling cascades and have more than 30 identified direct substrates (13, 46). Cytoskeletal rearrangement and regulation of cell membrane extensions that are important for cell motility were among the first functions discovered for PAK1. Acting as an effector of external stimuli, PAK1 directly phosphorylates Myosin Light Chain Kinase (MLCK) (47), Tubulin cofactor B (TCoB) (48), OP18/stathmin (49), and vimentin (50) (among others), leading to dynamic cytoskeletal morphologies such as membrane ruffling and filopodia formation (51, 52). Activated PAK1 has also been shown to relocalize from the cytosol to focal adhesions and dynamic actin structures near the plasma membrane in response to external stimuli, further illustrating a role in cytoskeletal reorganization and cell motility (53, 54). Group I PAKs also play a role in proliferation promoted by cell adhesion signals through integrin-mediated CDC42 and RAC1 activation (55).

PAKs play an important activation role for MAPK signaling. PAK1 directly phosphorylates MEK1 and RAF1, both integral components of the MAPK pathway, thereby mediating growth factor–driven cell cycle progression (56, 57). RAF1 (CRAF) Ser338 is a direct phosphorylation target of PAKs 1 and 2, an event that enhances CRAF activity and activation of the MAPK pathway (57). Additionally, a kinase-independent role in MAPK signaling has also been observed for PAK1, with PAK1 serving as a scaffold for MEK1/2 and CRAF at the cell membrane, allowing CRAF to phosphorylate its primary target, MEK1/2 (58). A more detailed review of BRAF–PAK interactions and signaling effects are described in later sections of this review in the context of thyroid cancer.

PAK1 also affects cell survival and proliferation by regulating several other signaling cascades. PAK1 activates NF-κB signaling by directly phosphorylating NF-κB-inducing kinase (59) and by coordinately increasing expression of the cell cycle regulator Cyclin D1 (60). In addition, PAK1 has been shown to be necessary for PI3K-mediated activation of CRAF and cell cycle progression in some systems (61). Finally, RAC and CDC42 GEFs (ie, PIX, SOS, VAV), which are necessary for these GTPases to be activated, are stimulated by activated PI3K as a result of fibronectin or growth factors, and this in turn activates PAK1 (62, 63). Thus, group I PAK activation is important for a number of key proliferative cellular pathways.

In addition to their roles in regulating cell motility and proliferation, group I PAKs also play a role in inhibition of apoptosis. For example, PAK1 is reported to phosphorylate and inactivate the proapoptotic protein BCL2 associated agonist of cell death (BAD) by subsequently reducing the affinity of binding between BAD and antiapoptotic proteins (64). PAK1 also phosphorylates and inactivates the proapoptotic protein BimL in a similar manner (65). Finally, PAK1 phosphorylates forkhead transcription factor (FKHR), which results in cytoplasmic mislocation of the protein and inhibits its transcriptional activity, ultimately preventing the transcription of downstream targets including several proapoptotic genes (66). PAK2, on the other hand, appears to have a direct role in the later stages of caspase-3-mediated apoptosis. PAK2 is cleaved by caspase-3 and this fragment is then myristoylated and relocalized to the cell membrane. Membrane-bound PAK2 subsequently mediates apoptosis through JNK activation. Abrogation of this myristoylation greatly reduced the apoptotic effects of the caspase-3–generated PAK2 fragment (67).

Group I PAKs also play a role in the progression of mitosis and the cytoskeletal dynamics required for cell division. PAK1 interacts with histone H3 during chromosome condensation and phosphorylates it at Ser10, a requirement for the progression of mitosis during prophase (23). PAK1 is also phosphorylated by Cyclin B1/CDC2 (also called CDK1) at Thr212 during mitosis, with high levels found from metaphase to telophase (68). Additionally, PAK1 Thr212 is enriched in microtubule-organizing centers during mitosis (23, 68). PAK activation is also implicated in centrosome separation. Tiam1, a GEF for RAC1, localizes at the centrosome and is phosphorylated by CDK1, a modification that leads to subsequent activation of colocalized PAK1 and 2 (69). Cells depleted of Tiam1 or PAK1 and 2 both had defects in chromosomal alignment on the metaphase plate (69). PAKs, in complex with β-PIX and GIT1, have also been shown to phosphorylate and activate the centrosome-associated kinase Aurora-A, a critical regulator of prometaphase chromosome alignment, anaphase sister chromatid separation, and cytokinesis (70). PLK1 (Polo-like kinase 1), which is necessary for the G2/M transition as well as proper bipolar spindle formation and spindle body tension, is also a direct phosphorylation target of PAK1 (71). All this information together suggests that group I PAKs play important roles in cell cycle progression and maintaining the proper cytoskeletal mechanisms necessary for division.

Group II PAK Structure and Function

The less studied PAKs 4, 5, and 6 make up group II, and share N-terminal PBD and C-terminal kinase domains with group I, albeit with considerably less sequence similarity (Fig. 1). Group II PAKs do not appear to use the AID for autoinhibition, but rather for subcellular localization, and instead autoinhibit in cis through the Pseudosubstrate (PS) domain near the PBD (72, 73). This structure only allows for activation in the presence of a binding partner with a classical SH3 (Src homology 3) domain. Similar to group I PAKs, PAK4 (73) and PAK5 (74) have also been shown to contain at least 1 nuclear localization sequence signal in the N-terminal domain. Group II PAKs were originally thought to be constitutively active due to basal phosphorylation of the catalytic threonine in the kinase domain, but evidence of the PS/AID domain and the increased activity of mutant PAK4 lacking the PBD/AID/PS domains suggests regulation of activity (75). Another distinguishing feature of group II is the lack of a PIX binding domain and lack of binding to the Nck adaptor protein (75). CDC42, and to a lesser extent RAC1, can still interact with the PBD of these kinases; however, they are not involved in activation of the kinase domain, but rather for subcellular localization as seen with PAK4 (76).

Group II PAKs also have isoform-specific tissue distribution patterns. PAK4 is expressed mainly in testis, prostate, cervix, and colon tissue, PAK5 is found primarily in brain and eye tissue, while PAK6 is highly expressed in brain, prostate, and placenta (13). PAK4 is particularly important for neural development and cardiovascular regeneration (13).

Similar to group I PAKs, functions that have been described to group II PAKs are regulation of cytoskeletal dynamics, cell proliferation, and apoptosis. For example, PAK6 is directly stimulated by MKK6, part of the p38–MAPK pathway, possibly as a stress response (77). Apoptosis is inhibited by PAK5 via the phosphorylation of BAD (78), as described above for PAK1. Most of the knowledge about group II PAKs comes from studying PAK4. PAK4, in a similar fashion to PAK1, is involved in growth factor–stimulated signaling (79). Hepatocyte growth factor recruits PAK4 to its receptor when colocalized near lamellipodia, along with the Gab1 adaptor protein (79). The interaction between PAK4 and Gab1 is PAK kinase independent and enhances cell migration and invasion in vitro (79). PAK4 also participates in Wnt/β-catenin signaling, a key pathway in embryonic development and cell proliferation, via direct phosphorylation of β-catenin on Ser675; this helps translocate it into the nucleus, as well as by preventing β-catenin degradation overall (80). Cytoskeletal rearrangement is also regulated by PAK4, which has been shown to phosphorylate GEF-H (81, 82) and Paxillin (83). These are important in the dissolution of stress fibers and focal adhesion complex organization. PAK4 also directly phosphorylates LIMK1 which promotes actin filament stability (84).

PAKs in Cancer

The above data demonstrate that PAKs regulate a diverse set of cellular functions that suggest a potential role in promoting cancer development and progression. While PAK point mutations or indels are rarely observed in human cancers, its activation is critical to potentiate the activity of upstream oncogenic regulators, such as RAC1 and RAS in skin tumors (7). PAKs 1 and 4 are particularly overexpressed isoforms in cancer. PAK1 amplification/overexpression has been found in thyroid (85), ovarian (86), colon (87), breast (60), pancreatic (88), and brain cancers (89), where it is generally associated with a poor prognosis in association with increased invasiveness. PAKs promote malignant transformation and cancer progression in a variety of manners, including enhanced invasiveness, motility, and proliferation, promoting epithelial to mesenchymal transition (EMT), inhibiting apoptosis, and promoting angiogenic induction (90).

Augmented cell proliferation is one of the classical hallmarks of cancer (91) and PAKs stand at the intersection of multiple cell signaling pathways involved in regulating this process. For example, PAK1 and PAK4 share many of the same substrates and regulate oncogenic MAPK signaling at several levels in the cascade, as described (92). Integrin and CDC42/RAC1 signaling enable anchorage independent growth in part via PAK (11). PAK1 also influences the activation of the NF-κB pathway (93) via NF-κB-inducing kinase (59) and RAS/RAF1-mediated activation 93-95), which contributes to oncogenesis by promoting cell survival and has been shown to be upregulated in many different types of cancer (96). PAKs also promote cancer cell survival through the inhibition of proapoptotic proteins as described earlier (11).

EMT is widely considered to be a feature to enable both local invasion and metastases of solid tumors. PAKs contribute to EMT by activating the Wnt/β-catenin pathway and through direct phosphorylation of SNAIL1, resulting in loss of cell adherence via the downregulation of E-cadherin (80, 97). Additionally, PAK1 phosphorylates SHARP, which, in turn, activates Notch signaling, another embryonic development pathway implicated in EMT and cancer (98, 99). The dynamic rearrangement of the cytoskeleton and cellular polarity is critical for local tissue invasion and often is coupled with EMT. PAKs phosphorylate cytoskeletal regulators such as Merlin (100), DLC1 (65), TCoB (101), LIMK (102), OP18/stathmin (103, 104), filamin (105), and vimentin (106), which are particularly active in invasive cancers.

PAKs also have been implicated in promoting angiogenesis in vitro. Vascular endothelial growth factor secretion, a proangiogenic feature of many cancer cells, is increased in response to PAK1 activation in breast cancer epithelial cells (107). PAK2 is reported to regulate vascular development and maintenance, but as yet has not been shown to play a defined role in cancer (108).

PAKs play a role in allowing transformed cells to progress through the cell cycle, as evidenced by PAK1-mediated increase in Cyclin D1 levels and PAK4-mediated decrease in cyclin-dependent kinase inhibitor 1C levels in breast cancer cells (60, 109). Experimentally, breast (60, 66, 109), colorectal 110-112), and skin cancer (113, 114) progression and the development of schwannomas (115) are inhibited by molecular knockdown or pharmacologic inhibition of PAKs 1 or 4, in vitro and in vivo, with evidence of reduced downstream signaling, cell cycle progression, and increased apoptosis. Finally, one of the major challenges in treating aggressive cancer is the development of drug resistance. Interestingly, activation of PAKs 1, 4, 5, and 6 has been shown to be important in resistance to cytotoxic chemotherapy by reactivation of cell cycle and survival signals, and targeting PAKs has resulted in renewed sensitivity to standard chemotherapies (88, 116-122). These data together have created interest in pursuing PAKs as potential therapeutic targets for several cancer types in different contexts (12, 90).

PAKs in Thyroid Cancer

A role for PAKs in papillary thyroid cancer (PTC) was first reported when PAKs were predicted to be one of the key signaling nodes identified in the analysis of global mRNA expression in the invasive fronts versus central regions of aggressive PTCs (123). The invasive fronts of these tumors also showed high expression of genes in the transforming growth factor-β, PI3K, and integrin pathways and low expression of cell adhesion genes. These, together, implicated an EMT regardless of the initiating oncogene, although most of the cancers had MAPK-activating mutations (113). Subsequent studies confirmed not only EMT but also that PAK1 was overexpressed and functionally critical for PTC invasion and EMT pathway activation (85).

In vitro studies demonstrated that group I PAKs were necessary for thyroid cancer cell migration, invasion, and proliferation (85, 124, 125). Further siRNA studies analyzing all 6 isoforms demonstrated these functions were primarily dependent on PAK1, with modest contributions from PAK4, suggesting a broad role for PAK1 in thyroid cancer cells (85). A role for PAK1 in thyroid cancer proliferation and cell motility is further supported by studies defining that PAK1 is a functional target of microRNA 7 (miR-7), an inhibitor of thyroid cancer growth (125). PAK4 has been shown to block the proliferative response of thyroid cells to TSH in vitro (126). Thus, it appears that PAK1 is the most important isoform in cancer cells with activated MAPK, while PAK4 may be relevant to PKA signaling, which is regulated by TSH, in thyroid cells.

PTC initiation is driven primarily by MAPK activation and the BRAF^V600E mutation is particularly common (127). It is associated with more aggressive tumors and is overrepresented in clinical trial populations. These associations are particularly strong when it occurs in common with mutations in the hTERT promoter, TP53, or in PI3K regulatory genes, and similar relationships have been shown for other MAPK activating mutations, such as those in NRAS, that are common in follicular forms of thyroid cancer (7, 128, 129). Due to the known ability of PAKs to augment MAPK signaling by direct phosphorylation of CRAF (61, 130), ARAF (131), and MEK1 (132), there has been interest in determining the role of PAK in MAPK-mediated tumorigenesis. For example, in skin tumor models, PAK activity has been shown to be required for RAS-induced transformation and growth (110, 113, 133, 134). In thyroid cancer cell lines, it has been shown that inhibition of group I PAK activity, and PAK1 specifically, inhibits cellular proliferation in cancer cells with BRAF^V600E or HRAS-activating mutations (124). It, therefore, was hypothesized that in thyroid cancer cells, PAK-mediated phosphorylation of MEK1 or BRAF, at the cognate site for CRAF, would further activate MAPK signaling. Unexpectedly, inhibition of PAK activity did not alter BRAF-mediated MAPK signaling. However, PAK activation was shown to be dependent on BRAF, a feature not described previously and termed “noncanonical” (Fig. 2). Due to the known scaffold functions of PAK (58), further studies demonstrated that BRAF colocalized and coimmunoprecipitated with PAK, particularly during mitosis. In addition, the regulation of PAK by BRAF was not altered either by MEK inhibitors or MEK depletion, suggesting that BRAF regulation of PAK is independent of MEK (124). Finally, BRAF siRNA-mediated inhibition of invasion was rescued by both constitutively active MEK and PAK, further suggesting that both pathways are important in BRAF-mediated effects in thyroid cancer cells (124). These data suggested that BRAF might regulate thyroid cancer development and/or progression not only through MEK activation, but also by activation of PAKs.

Figure 2.

PAK/MAPK signaling and the downstream effectors involved in oncogenesis and progression. A restricted view of the PAK/MAPK signaling cascades, and a limited list of PAK substrates that illustrates the diverse mechanisms by which PAK dysregulation can lead to oncogenesis and progression. Overactivation of PAK, whether upstream through mutations in BRAF or RAC1, or PAK overexpression/amplification, lead to transformation exemplary of the hallmarks of cancer. The “noncanonical” activation of PAK by BRAF is noted with a red line.

In vivo studies further support a fundamental role of PAK signaling in PTC progression. Induction of BRAF^V600E in the thyroid in mice leads to PTC development associated with increased PAK isoform expression and activity. Treatment of mice harboring doxycycline-inducible thyroid-specific BRAF^V600E overexpression develop PTC with activation of both PAK and ERK. Pretreatment of these mice with the group I PAK inhibitor G-5555 inhibited BRAF^V600E-induced PTC development with reduced PAK and maintained ERK activation, consistent with an important role for PAK in BRAF^V600E-induced PTC development (135). Moreover, in thyroid cancer cells with a BRAF^V600E mutation, combination of the BRAF^V600E-specific kinase inhibitor, vemurafenib, with G-5555 was synergistic with inhibition of both PAK and MEK signaling, suggesting that inhibition of both signals may be therapeutically beneficial (135). Taken together, these data suggest that there are both overlapping and independent PAK and MEK signaling cascades and that there may be therapeutic value in inhibiting the MEK-independent cascade. Importantly, because BRAF-mediated PAK activity occurs via a nonkinase mechanism, it is not inhibited by available BRAF kinase inhibitors. Similar data have been reported in melanoma models in which PAK activation is mechanistically necessary for resistance to BRAF and/or MEK inhibitors in BRAF^V600E or mutated RAS-driven cancer models (133, 136, 137). Finally, while these data have focused on group I PAKs, PAK4 has been implicated in therapeutic resistance to MEK inhibition in RAS-mediated colorectal cancer (92) and pancreatic adenocarcinoma (138), and anaplastic thyroid cancer resistance (139) to the VEGFR inhibitor, Lenvatinib, although the mechanisms are less well defined. Group I PAKs have also been shown to be required for tumors induced by activated RAC1^P29S (113, 114, 140) and NF2 (41, 141, 142) mutants, 2 uncommon but reported genomic events in thyroid cancer. Taken together, PAKs represent a rational therapeutic target for aggressive thyroid cancer, particularly those that harbor RAS or BRAF gene mutations as drivers.

Summary and Future Directions

PAKs are a family of serine/threonine kinases that have diverse cellular functions regulating cytoskeletal dynamics, mitosis, cell survival/growth, cell cycle progression, and gene transcription. The role of PAK as a necessary signaling molecule for oncogenic transformation driven by a number key pathways implicates its potential both as a biomarker for signaling activity and as a therapeutic target. New data in thyroid cancer and melanoma models suggest a potential specific role in MAPK-driven tumors, both for combinatorial primary treatment approaches and as a key mechanistic player in therapeutic resistance to BRAF and MEK-targeted therapies that are in clinical practice and/or clinical trials. Overall, the PAKs are promising therapeutic targets warranting further translational study.

Glossary

Abbreviations

AID

Auto-Inhibitory Domain

CRIB

CDC42 and RAC1 binding

EMT

epithelial to mesenchymal transition

MAPK

mitogen activated protein kinase

PAK

p21-activated kinase

PBD

p21-binding domain

PS

Peseudosubstrate

PTC

papillary thyroid cancer

TSH

thyrotropin

Additional Information

Disclosure Summary: The authors report no conflicts of interest

Data Availability: Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.