Aurora kinases in Head and Neck Cancer

Abstract

The Aurora kinases (AURKA and AURKB) have attracted attention as therapeutic targets in head and neck squamous cell carcinomas (HNSCC). Aurora kinases were first defined as regulators of mitosis that localization to the centrosome (AURKA) and centromere (AURKB), governing formation of the mitotic spindle, chromatin condensation, activation of the core mitotic kinase CDK1, alignment of chromosomes at metaphase, and other processes. Subsequently, additional roles for Aurora kinases have been defined in other phases of cell cycle, including regulation of ciliary disassembly and DNA replication. In cancer, elevated expression and activity of Aurora kinases results in enhanced or neomorphic locations and functions that promote aggressive disease, including promotion of MYC expression, oncogenic signaling, stem cell identity, epithelial-mesenchymal transition, and drug resistance. Numerous Aurora-targeted inhibitors have been developed, and are being assessed in preclinical and clinical trials, with the goal of improving HNSCC treatment.

Introduction

There were 878,000 new cases of head and neck squamous cell carcinomas (HNSCCs) worldwide in 2020¹, and approximately 54,000 new cases in the US in 2021, representing 2.8% of all new US cancer cases². These tumors most commonly occur in the oral cavity, pharynx, and larynx. Human papillomavirus (HPV) infection and integration of HPV into the genome is a significant cause of HPV+ HNSCCs arising from the oropharynx³, where carcinogenesis occurs due to p53 and Rb degradation caused by the virally-encoded E6 and E7 oncogenic proteins^4,5. HPV- tumors more commonly develop from the oral cavity and larynx as a result of chronic exposure to carcinogens, such as alcohol and tobacco^6,7. The typical 5-year overall survival of HNSCC patients is 68% and patients are frequently diagnosed at median age of 64 years²; HPV+ tumors have a better prognosis than HPV- cancers^8,9, causing an emphasis in developing therapies on HPV- HNSCC, the primary focus of this article.

HPV- HNSCCs are marked by loss of tumor suppressor genes. TP53 is commonly mutated (72% of cases); mutation or loss of CDKN2A is also frequent (60% of cases)^10,11. During disease progression, additional common alterations include CCND1 amplification¹² and PTEN loss^13,14. Epidermal growth factor receptor (EGFR) overexpression is frequent in HNSCC and associated with poor prognosis and low response to standard therapies^15–17. Treatment depends on the tumor site and staging, but standard therapy for advanced tumors includes chemotherapy and targeted therapy¹⁸. For targeted therapy, the EGFR inhibitor cetuximab is beneficial, and sensitizes patients to chemoradiation; however, these patients often develop resistance by activation of alternative receptor tyrosine kinases (RTKs)¹⁹ and other mechanisms that compensate for EGFR inhibition. Immune checkpoint inhibitors are being actively explored for HNSCC patients with advanced disease, although response is limited^20,21.

The Aurora kinases have attracted attention as therapeutic targets in HNSCC. In humans, the Aurora kinase family has three members - AURKA, AURKB, and AURKC. In normal cells, spatially and temporally limited activation of AURKA and AURKB controls mitosis and other cell cycle related processes. However, elevated expression and activity of AURKA and AURKB often occurs in HNSCC and other forms of cancer, leading to temporally deregulated anomalous activity in multiple cellular compartments that contributes to pathogenicity, and is linked to poor prognosis. AURKC is predominantly expressed in meiotic cells²²; although it is sometimes overexpressed in cancer cells, and some data suggests overexpressed AURKC in transformed cells may enhance cancer cell migration²³, overall AURKC has attracted little study. This article will focus on summarizing biology of AURKA and AURKB kinases relevant to therapy of HNSCC and discuss recent developments in agents targeting these kinases.

Structure and cell cycle regulation of Aurora kinases

Aurora kinases have a highly conserved kinase domain located at the C-terminus (Figure 1) but diverge at the N-terminus²⁴. In contrast to other family members, AURKA has an extended N-terminal domain that is intrinsically disordered²⁵, but is the site of interaction for a number of AURKA partner proteins, such as TPX2 and NEDD9²⁶ (Figure 1). The N terminal domain of AURKB is also crucial for partner protein binding, allowing for the recruitment of the Inner centromere protein (INCENP) (Figure 1)²⁷.

Figure 1. Structure and interactions of AURKA and AURKB.

A., B. 3D structure of human AURKA in complex with VX-680 (a pan-Aurora kinase inhibitor and non-hydrolyzable analog of ATP) and TPX2 (colored blue), from Protein Data Bank (PDB) entry 3E5A. A. Structure of the nucleotide binding pocket of AURKA; the residues that participate in ATP binding are colored red. B. AURKA is rotatedabout the Y axis in A versus B to highlight TPX2 interaction. C., D. 3D structure of AURKB in complex with VX-680 and INCENP (colored blue), from PDB entry 4AF3. C. AURKB structure with binding pocket, to foreground; in D. AURKB is rotated to show INCENP interaction. E. Domain organization of AURKA and AURKB. The conserved kinase domain is colored green in both AURKA and AURKB. The domains that are necessary for APC-C mediated proteolysis (D-Box for AURKA; KEN-Box and A-Box for AURKB) are colored orange. The intrinsically disordered regions of AURKA and AURKB contains the binding site for TPX2 and INCENP, respectively. Noted phosphorylation sites are described in the text.

Aurora kinase activation is usually accompanied by auto-phosphorylation of a residue in the activation loop of the kinase domain: Thr288 for AURKA, and Thr232 for AURKB. For AURKA, this autophosphorylation is induced by complex interactions between AURKA and partner proteins²⁸ which include TPX2, BORA, NEDD9, PAK1, and Calmodulin^29–32. TPX2 also helps to enhance AURKA signaling by binding to AURKA, inducing AURKA phosphorylation and shielding the phosphorylated T288 residue, thus locking AURKA in the active conformation³³. In addition, some protein interactions that induce allosteric changes in the protein can partially activate AURKA without triggering Thr288 phosphorylation³⁴. For AURKB, autophosphorylation is induced upon interaction with INCENP³⁵, BOP1³⁶ and a small number of other proteins.

The best-studied functions of AURKA and AURKB in untransformed cells involve mitotic control (summarized in Figure 2). In non-transformed cells, AURKA predominantly concentrates at the centrosome, the mitotic spindle, and the basal body of cilia³⁷. AURKA contributes to centrosome maturation in G2 phase of the cell cycle, then interacts with a series of partner proteins to mediate formation of the mitotic spindle, and to activate the core cell cycle CDK1-cyclin B complex. AURKB localizes on chromatin and concentrates at the kinetochore, causing chromatin condensation and facilitating the alignment of sister chromatids during mitosis³⁸. The activities of both AURKA and AURKB are crucial for mitosis to initiate, and progress to cytokinesis without chromosomal segregation defects that would lead to aneuploidy. At the end of mitosis, Aurora kinases are degraded to allow cytokinesis.

Figure 2. Activity of Aurora kinases in mitosis.

In G1, AURKA localizes to the ciliary basal body (a structure derived from the centriole) and is transiently activated to cause resorption of cilia³⁹. As cell cycle progresses, the basal body undergoes changes releasing the centriole to form the centrosome; AURKA accumulates at the centrosomes and contributes to centrosome maturation in G2 phase of the cell cycle. In late G2, AURKA interacts with TACC1 and TACC3, which recruit ch-TOG/XMAP215. Interaction of AURKA with these, TPX2, and additional proteins as AURKA propagates along the spindle asters³⁰ results in the formation and appropriate function of the mitotic spindle, allowing appropriate chromatin segregation¹⁶⁶. AURKA also contributes to mitotic entry by activating core cell cycle machinery. AURKA phosphorylation activates the CDC25 phosphatase, which then removes the inhibitory Tyr15 phosphorylation from CDK1¹⁶⁷; AURKA also phosphorylates PLK1, which inhibits the cell cycle checkpoint phosphatase WEE1¹⁶⁸. Together, these changes activate CDK1 to promote the cell cycle transition from G2 to M phase and sustaining CDK1 activity through M phase.

AURKB primarily functions in association with chromatin. In S phase, AURKB contributes to DNA replication, but has low level activity. In mitosis, the chromosomal passenger complex (CPC), consisting of AURKB, INCENP, Survivin, and Borealin, is crucial for chromosomal condensation. INCENP binding to AURKB induces a conformational change in AURKB, leading to AURKB autophosphorylation on Thr232¹⁶⁹. AURKB remains active through mitosis, mediating the extensive Ser10 phosphorylation of Histone H3 to support chromosome condensation¹⁷⁰. As part of the CPC, AURKB also regulates the activity of mDia3 and Ndc80, ensuring that the chromosomes are properly arranged at the metaphase plate¹⁷¹. Briefly, the CPC accumulates at the inner centromere. When the microtubule is tightly pulled, indicating a proper attachment, mDia3 and Ndc80 are distal to the CPC, and cannot be phosphorylated by AURKB^171,172; unphosphorylated mDia3 and Ndc80 have high binding affinity for the microtubule and facilitate attachment to the kinetochore. If there is improper attachment of the centromere to the kinetochore, the CPC and AURKB are proximal to mDia3 and Ndc80, and phosphorylates these proteins, causing them to lose affinity for the microtubules. Removal of mDia3 and Ndc80 from microtubules initiates another cycle of attempted microtubule-kinetochore attachment¹⁷¹. This process is crucial for chromosomal alignment during mitosis and ensuring appropriate post-mitotic DNA segregation.

At the end of mitosis, Aurora kinases are degraded to allow cytokinesis. Aurora kinases are subject to numerous mechanisms of negative regulation, that reduce kinase activity or expression. Both AURKA and AURKB interaction with the Anaphase Promoting Complex/Cyclosome (APC/C), which acts as an E3 ubiquitin ligase for the kinases, causing their proteasomal degradation. A number of interacting partners that target AURKA for destruction have been identified, including FBXW7²⁴. The D-Box motif of AURKA is necessary for APC/C recognition and proteasomal degradation¹⁷³. In contrast, the KEN box and A box, rather than the D-box motif, is required for AURKB interaction with APC/C¹⁷⁴.

In normal cells, AURKA also has important non-mitotic roles. For example, in quiescent cells, AURKA localizes to the basal body of the primary cilium, an antenna-like structure protruding from the cell surface that serves as a receptor for soluble growth-regulating factors and mechanical cues. Transient activation of AURKA in G0/G1 cells in response to external cues leads to resorption of cilia³⁹, altering cellular response to these factors. A number of activators of AURKA during ciliary resorption have been described; these differ from AURKA activators that function during mitosis⁴⁰. Limited activation of AURKB has also been described in normal cells, with some activity in S phase that is important for subsequent accurate chromosome segregation and telomere maintenance^41,42.

Overexpression of Aurora kinases in HNSCC and other cancers

The AURKA gene is located on chromosome 20q13, a region amplified in a number of cancers^43–45, and a small number of HNSCCs^46,47. In contrast, chromosome 17p13, the site of AURKB, is not commonly amplified in tumors. Analysis of TCGA data indicates limited copy number variation (CNV) of AURKA and AURKB in HNSCC (Figure 3A, B), although low level gain occurs in a subset of tumors (Figure 3C). TCGA analysis also shows that, in contrast to other pro-oncogenic kinases such as EGFR⁴⁸, activating AURKA or AURKB mutations rarely if ever occur in cancer (Figure 3A, B). However, mRNAs for both Aurora kinases are overexpressed in many tumors^{43,44,49–52}, including HNSCC (Figure 3C), as are the proteins, allowing the overexpressed proteins to have both expanded localization, and greater activity throughout cell cycle. Interestingly, increased copy number for the AURKA and AURKB genes significantly co-occurs with alterations commonly found in head and neck neoplasms, with co-occurrence for EGFR amplification and gain (q value = <0.001 for AURKA and 0.008 for AURKB); for PIK3CA amplification and gain (q value = <0.001 for both); for PTEN loss (q value = 0.018 for AURKB); and for FBXW7 loss (q value = 0.021 for AURKB). This may reflect important functional interactions.

Figure 3. Genomic and transcriptomic alterations in AURKA and AURKB in cancer.

A., B. Mutation and amplification data for AURKA (A) and AURKB (B) in HNSCC. Data show reflect information from the MSK-IMPACT Clinical Sequencing Cohort (MSKCC)¹⁷⁵. Data analyzed were accessed through cBioportal^176,177. C. Analyses were performed with a cohort of 415 HPV− tumor samples from the HNSCC cohort reported in the TCGA PanCancer Atlas)¹⁷⁸. The cutoff for high mRNA is z score > 2, for low mRNA is z score < −2. Amplification indicates a number of copies above a limit defined by cBioportal and is often focal; gain indicates a low-level copy number increase, often extending across a significant region of a chromosome.

AURKA overexpression is associated with highly invasive tumors, as well as decreased metastasis-free survival and overall survival^53–55. HNSCC patients with overexpression of AURKA presented increased incidence of lymph node and distant metastasis, advanced disease stage, shorter disease-free survival, and overall survival^56,57,58. Increased AURKB expression is also linked to multinuclearity and increased ploidy⁵⁹. For HNSCC, increased AURKB expression and activity has been demonstrated in oral cell lines and tumor samples in comparison with normal epithelium, and correlated with increased cell proliferation, poor tumor differentiation, shorter disease-free survival, and increased disease aggressiveness^60,61.

Transcriptional upregulation of AURKA and AURKB in cancer is promoted both by oncogenic activation and tumor suppressor loss (Figure 4A). For example, MYC directly binds to AURKA gene enhancer regions to promote its transcription⁶², and downregulation of Myc alleviates the oncogenic effects mediated by AURKA⁶³. Conversely, overexpression of AURKA has been shown to mediate the tumor promoting effects caused by Myc⁶⁴, and increased Myc protein levels⁶⁵, in positive feedback. Enhanced signaling by EGFR also upregulates AURKA expression; in one study, EGFR was defined as a direct transcriptional activator of AURKA, translocating to the nucleus and interacting with STAT5 to bind the AURKA promoter and initiate transcription⁶⁶. EGFR overexpression and activation are well known characteristics of most head and neck neoplasms, where EGFR amplification occurs in approximately 30% of all cases^17,67,68. Mutational inactivation or loss of TP53 - the most mutated gene in HPV- HNSCC¹¹ – also influences AURKA expression. Intact p53 directly binds to the AURKA promoter to repress its transcription⁶⁹; p53 can also cause indirect transcription repression of AURKA through the cyclin-dependent kinase (CDK) inhibitor 1A mediated inhibition of E2F3, an AURKA transcriptional activator; in addition, p53 can directly interacts with the AURKA protein to inhibit its kinase activity⁷⁰. Loss of p53 in HNSCC impairs these processes, elevating AURKA expression. Interestingly, it has been reported that distinct TP53 genotypes can have distinct effects on AURKA levels⁷¹; this topic requires more investigation.

Figure 4. AURKA upregulation and cancer-promoting activity.

A. Mechanisms of upregulation of Aurora kinases include amplification or gain (AURKA and/or TPX2); enhanced transcriptional activation by MYC, STAT5/EGFR, and E2F3 and decreased transcriptional repression by p53; upregulation of AURKA activators (TPX2, NEDD9, etc) and decreased proteolytic degradation. Light pink icon indicates oncogenes, while light blue indicates tumor suppressor genes. Arrow next to the gene names indicates gene up or downregulated in head and neck neoplasms. B. Cancer promoting activities of upregulated AURKA include induction of p53 degradation or impairment of p53 DNA binding and transactivation functions; induction of signaling by SOX2, NF-kB, HIF-1b, hnRNPK and MYC; activation of RalA, degradation of BIM, and promotion of aneuploidy. See text for details.

At the protein level, AURKA can be upregulated by post-translational modifications and protein interactions which enhance AURKA protein stability or prevent its degradation (Figure 4A). A study in HNSCC showed phosphorylation of AURKA on Ser51 increases its protein levels by inhibiting its APC-mediated degradation⁷²; this post-translational modification is mediated by Ca2+/Calmodulin (CaM) binding to AURKA protein, which induces its activation and consequent auto-phosphorylation³². The tumor suppressor and E3 ubiquitin ligase FBXW7 is commonly downregulated in many types of cancer⁷³; as FBXW7 normally promotes proteasomal degradation of AURKA⁷⁴, loss of FBXW7 is typically associated with increased levels of AURKA⁷⁵. Importantly, FBXW7 is frequently inactivated by mutation in HNSCC, and was identified in a 2020 study as the fourth (7.7%) most commonly mutated gene in a panel of in HPV+ HNSCC tumors⁷⁶. In addition to frequent loss-of-function (LoF) mutations, transcriptional downregulation of FBXW7 in cancer also is common due to loss of p53, an activator of FBXW7 transcription^75,77.

Activation of PI3K/AKT signaling either by activating mutation of PI3K or loss of the PTEN tumor suppressor indirectly regulates AURKA levels through FBXW7 (Figure 4A). AKT-dependent inhibitory phosphorylation of GSK3β blocks its interaction and phosphorylation of AURKA (S245 and S387); this impairs the ability of FBXW7 to bind to AURKA and promote its degradation⁷⁴. In HNSCC, PTEN is commonly mutated or lost (27% based on TCGA, Figure 3C), with reports as early as 1998 demonstrating loss of heterozygosity (LOH) at the PTEN locus in 41% of HNSCC samples, as well as presence of inactivating mutations^14,78,79. Amplifications and activating mutations in PIK3CA are also frequent in HNSCC, a genomic characterization study with 279 HNSCCs samples showed that 34% of HPV- tumors had amplification and/or driver mutations on this gene¹¹. Hence, there is a common post-translational upregulation of AURKA in cancers with elevated PI3K/Akt signaling⁷⁹.

An alternative way of increasing AURKA activity in HNSCC is through elevated expression of partner proteins that interact with AURKA to stabilize the protein and support its catalytic activity (Figure 4A). The AURKA activator TPX2, is overexpressed in many cancers, and through AURKA binding, allosteric modulation, and protection of the T288 phosphorylation site from dephosphorylation, TPX2 increases AURKA levels and maintains its active state, augmenting AURKA oncogenic effects in the cell^80,81. In a study in HNSCC, TPX2 mRNA levels were upregulated in 43 tumor samples in comparison with normal paired samples⁸². Notably, the chromosomal location of TPX2 is very close to that of AURKA, and the two proteins are co-amplified in some tumor types⁸⁰. TPX2 is upregulated in 46% of HPV- HNSCCs in a cohort of 415 samples from TGCA database (Figure 3C), with most of the samples with TPX2 gain co-occurring with gain in AURKA gene, reflecting their close chromosomal location (Figure 4A). In addition, TPX2 and AURKA mRNA levels strongly correlate (Spearman correlation coefficient = 0.802, q value <0.0001). Similarly, NEDD9 overexpression increases AURKA levels. NEDD9 also stabilizes AURKA by blocking the binding of the APC complex, preventing its degradation^29,83; NEDD9 expression is commonly elevated in advanced HNSCC⁸⁴. Both TPX2 and NEDD9 levels have been implicated in sensitivity to AURKA inhibition by small molecule agents^83,85.

Cancer-promoting function of overexpressed AURKA

AURKA overexpression frequently occurs as an early event in cell transformation^44,86, and promotes tumor initiation and progression in several ways (Figure 4B). Mechanistically, AURKA amplification and overexpression in HNSCC and other tumor types causes an increase in centrosome number, and an inability to resolve M-phase, associated with failure of cytokinesis^44,87,56. Abnormal centrosome organization and number induced by AURKA upregulation causes formation of multipolar mitotic spindles, which leads to inappropriate segregation of chromosomes, chromosomal instability, and aneuploidy in HNSCC⁸⁸ and other cancers. Factors that lead to AURKA overexpression or anomalous activation, such as overexpression of the AURKA scaffolds TPX2⁸¹ and NEDD9²⁹, produce similar phenotypes.

Besides directly acting on the mitotic apparatus, overexpressed AURKA drives early cellular transformation by inducing G1 checkpoint loss through inducing downregulation of p53 (Figure 4B). AURKA phosphorylation of p53 on S215 impairs its DNA binding and transactivation function, resulting in the downregulation of some p53 target genes that act as tumor suppressors, such as PTEN and CDKN1A⁸⁹. AURKA also phosphorylates p53 on S315, which enhances p53 binding by the E3 ligase MDM2, promoting the ubiquitination-mediated degradation of p53⁹⁰. Loss of p53 and impairment of p53-mediated cell cycle checkpoints promotes cell survival in the context of increasing DNA abnormalities. In addition, AURKA-dependent loss of p53 may also contribute to the persistence of cancer stem cells in HNSCC, based on findings that downregulation of p53 by AURKA supports pluripotency of embryonic stem cells (ESCs)⁹¹. AURKA also promotes stem cell pluripotency through direct phosphorylation of SOX2, influencing its transcription factor activity⁹². As maintenance of stem cells is a major determinant of therapeutic resistance and tumor recurrence, these AURKA activities likely contribute to the poor prognosis of HNSCC with high AURKA expression.

Several important AURKA oncogenic effects arise due to the expanded localization of the overexpressed protein in cancer⁹³. Many studies of tumors have reported AURKA in the nucleus, where the degree of nuclear expression correlates with disease aggressiveness^94–96. One study found AURKA abundant in the nucleus of HNSCCs cells, and directly correlated this location with oncogenic activity⁹⁵; AURKA nuclear localization has also been associated with worse survival in HNSCC patients⁹⁴. At least some nuclear AURKA functions in a kinase-independent manner, by binding to oncogenic proteins to promote their stabilization, and acting as a transcriptional activator to enhance transcription of oncogenic genes, such as MYC and FOXM1^96,97 – another way in which AURKA increases a stem cell phenotype⁹⁷. Mechanistically, nuclear AURKA can bind to the heterogeneous nuclear ribonucleoprotein K (hnRNPK) to transactivate MYC transcription, with this function retained by a kinase dead AURKA mutant⁹⁶ (Figure 4B). In neuroblastoma, AURKA binds and stabilizes N-Myc, preventing its proteasomal degradation⁹⁸; pharmacological inhibition of AURKA triggers conformational change that disrupts its binding to N-MYC, causing a consequent decrease in N-MYC protein levels^99,100. Several studies suggest that AURKA inhibition decreases C-MYC protein levels in a similar manner^101,102. By supporting MYC levels, AURKA promotes MYC-dependent transcription that enhances cell cycle progression and cell survival in tumors. MYC also promotes transcription of many glycolytic enzymes and represses genes involved in oxidative phosphorylation and fatty acid oxidation; AURKA-mediated Myc induction was shown to contribute to converting tumor metabolism towards a glycolytic state¹⁰². Separately, AURKA has also been found to promotes glycolysis by direct phosphorylation of lactate dehydrogenase B (LDHB), which increases NAD⁺ and lactate formation¹⁰³.

Many studies have found that AURKA overexpression promotes epithelial-mesenchymal transition (EMT), a change in differentiation status linked to enhanced tumor migration and invasion in many tumor types^104–106. In a study in HNSCC, AURKA promotion of migration and invasion involves activation of focal adhesion kinase (FAK)¹⁰⁷, a partner of the AURKA activator NEDD9 (also known as HEF1)^108,109. HNSCC tumors typically contain areas with differing degrees of oxygenation, influencing response to radiation and other therapies, and tumor invasion, which is induced by hypoxia¹¹⁰. Although hypoxia-induced transcription factors such as the HIF proteins are normally inactive under conditions of oxygenation, one study has found that nuclear AURKA binds and activates HIF1B under normal oxygen conditions, inducing transcription of hypoxia-response genes that promote migration and stemness (Figure 4B). In this work, knockdown of HIF1A/B decreased the epithelial-mesenchymal transition (EMT) effect driven by nuclear AURKA, indicating some dependency of AURKA to HIF1A/B to induce EMT¹¹¹. Moreover, AURKA activates the small GTPase RALA through phosphorylation of the serine 194, increasing cell motility and migration (Figure 4B)^112,113. The role of increased AURKA in promoting EMT emphasize its involvement in disease progression, especially as a metastatic driver.

Notably, upregulation of AURKA has been implicated in resistance to chemotherapeutic agents, such as taxanes¹¹⁴. AURKA inhibition and paclitaxel have synergistic cytotoxic effects in HNSCC cell lines¹¹⁵ in vitro, and in vivo for other types of solid tumor¹⁰⁴. In EGFR-mutant lung adenocarcinoma, prior treatment with EGFR inhibitors increased levels of TPX2 and activated AURKA, driving resistance to the inhibitors. Mechanistically, TPX2/AURKA replace the inhibited activity of inhibited EGFR in phosphorylating the pro-apoptotic protein BIM to promote its degradation (Figure 4B), thereby contributing to ability of these tumors to survive treatment with EGFR inhibitors; this study showed that AURKA inhibitors synergize with EGFR inhibitors to eliminate EGFR resistant cells in vitro and in vivo⁸⁵. Aurora inhibitors have been identified as cytotoxic in patient-derived HNSCC cell lines that were both sensitive and resistant to the EGFR inhibitor gefitinib; in this study, resistant cells had wildtype EGFR with increased levels of phosphorylated AURKA¹¹⁶. Such results suggest means to apply AURKA inhibitors in combination therapies, as discussed below.

Cancer-promoting function of overexpressed AURKB

Similar to AURKA, AURKB promotes Myc stabilization through phosphorylation of serine 67 on Myc, which prevents its interaction with the E3 ubiquitin ligase FBXW7 and its consequent mediated proteasomal degradation⁶⁵. AURKB also phosphorylates p53 on residues S183, T211, and S215 to promote its degradation, repressing important p53 functions¹¹⁷ such as transcriptional induction of its downstream target genes, such as CDKN1A¹¹⁸. Knockdown of AURKB can repress AKT and mTOR activity, with co-inhibition of AURKB and use of a histone deacetylase (HDAC) inhibitor further intensifying this repression and inducing apoptosis, although the mechanism remains to be elucidated¹¹⁹. Interestingly, in cells with active AKT signaling due to mutation of PTEN or PI3K, the observed inhibition of tumor cell proliferation and in vivo tumor growth caused by the novel AKT degrader, MS21, was linked to destabilization of AURKB¹²⁰. Some studies have also implicated AURKB as a mediator of drug resistance in HNSCC. For example, increased mRNA levels of AURKB were observed in HNSCC cell lines with intrinsic EGFR resistance to cetuximab, and these cells were sensitive to the AURKB inhibitor barasertib; again, more work is required to define the resistance mechanism¹²¹.

Inhibitors of Aurora Kinases: preclinical studies.

Because of the clear contribution of elevated AURKA and AURKB activity to multiple aspects of tumorigenesis, kinase inhibitors have been developed targeting AURKA, AURKB, or both (Table 1). While the value of AURKA and AURKB inhibitors is well documented in many tumor types in preclinical studies, investigation of these inhibitors in the setting of HNSCC is limited. Overall, preclinical studies have found that cells with mutant p53 are sensitive to Aurora kinase inhibition monotherapy^122,123. AURKA-AURKB dual inhibition is useful in some cancer cells that are resistant to other anti-mitotic therapeutic agents¹²⁴. Adavosertib, a highly selective inhibitor of WEE1 (which mediates checkpoints for mitotic entry), synergizes with AURKA inhibitors, effectively reducing overall tumor size in HNSCC xenograft experiments compared to vehicle and single drug treatment¹²⁵. Disruption of signaling by Rb, which regulates G1/S checkpoints, was also shown to be synthetically lethal with AURKA inhibition¹²⁶.

Table 1.

Preclinical and clinical Aurora kinase inhibitors

Compound Name	Targets (IC50, in vitro assays)	Results in relation to HNSCC	Reference
MLN8237 (Alisertib)	AURKA (1.2nM)	Extensive preclinical trials using HNSCC xenograft models; Phase I-II in HNSCC patients	[125,147]
VIC-1911 (TAS-119)	AURKA (1.0nM) AURKB (95nM)	Limited in vitro and HNSCC xenograft model	[148]
MLN8054	AURKA (4nM)	Limited in vitro assays in the setting of EGFR mutant HNSCC	[116,149]
PF-03814735	AURKA (0.8nM) AURKB (5nM)	N/A	[150]
BPR1K653	AURKA (124nM) AURKB (45nM)	N/A	[151]
PHA-739358 (Danusertib)	AURKA (13nM) AURKB (25nM)	Extensive preclinical trials using patient derived HNSCC xenograft models	[152,153]
ENMD-2076	FLT-3 (1.86nM) RET (10.4nM) AURKA (14nM)	N/A	[154,155]
AZD1152 (Barasertib)	AURKB (0.37nM)	Limited in vitro assays in HNSCC cell lines	[156,157]
GSK1070916	AURKB (3.5nM) AURKC (6.5nM) FLT1 (42nM) Tie-2 (59nM) SIK (70nM)	N/A	[158]
CYC116	AURKA (44nM) AURKB (19nM) AURKC (65nM) VEGFR2 (69nM)	N/A	[159]
VX-680 (Tozasertib)	AURKA (0.6nM) AURKB (18nM) AURKC (4.6nM)	N/A	[160]

Importantly, pan-Aurora kinase inhibitors synergize with the EGFR inhibitor cetuximab – a critical first line agent used for HNSCC - and also overcome cetuximab resistance^121,121,12. In EGFR-mutated lung cancer cell lines, dual inhibition of EGFR and AURKB is a stronger inducer of apoptosis than EGFR inhibition alone, based on induction of BIM and PUMA caused by AURKB inhibition¹²⁷. Dual inhibition of AURKA and PI3K has been shown to greatly reduce mTOR signaling in breast cancer models, resulting in apoptosis and reduction in tumor size in vivo¹²⁸. Dual inhibition of AURKB and AURKA in early-stage lung adenocarcinoma downregulated AKT/mTOR signaling, and also reduced expression of EMT markers such as N-Cadherin and vimentin; these effects were not observed with single inhibition of either kinase¹²⁹. AURKA and AURKB inhibition may be an effective neoadjuvant for subsequent treatments such as immune checkpoint inhibition and radiation. For example, Aurora kinase inhibition in melanoma potentiated antitumor immunity by sensitizing tumor cells to T-cell mediated cytotoxicity¹³⁰. AURKA inhibition also sensitized xenografted tumors to radiation therapy¹³¹.

While most inhibitors of Aurora kinases are small molecules, PROTACs (Proteolysis Targeting Chimeras) are being developed to AURKA, such as the agents JB170 and JB158¹³². PROTACs contain two binding moieties that connect the endogenous ubiquitin ligase pathway to targeted proteins to cause their degradation, and sometimes have greater activity than agents such as kinase inhibitors, particularly if a kinase has non-catalytic roles (e.g. scaffolding complexes). Among the limited tests performed to date, one group has found that an AURKA-targeting PROTAC preferentially binds and degrades spindle associated AURKA at a higher rate than centrosomal AURKA, suggesting that PROTACs could be fine-tuned to target a specific pool of AURKA¹³³. To date, there are no studies that investigate the usage of an AURKA targeting PROTAC in a clinical trial.

Aurora Kinase Inhibitors in Clinical Trials

While Aurora inhibitors have shown great promise in preclinical studies, usage in the clinic have shown varying degrees of success (Table 2). In single agent trials, the AURKA/B inhibitor PF-03814735, one of the first inhibitors to be used in clinical trials as monotherapy for patients with advanced solid tumors (including HNSCC), showed tolerable toxicity and established a recommended phase 2 dose, but only limited antitumor activity was observed (NCT00424632)¹³⁴. The AURKA inhibitor alisertib (also called MLN8237), has been evaluated in multiple clinical trials for many types of cancer. One of the first clinical trials for alisertib included head and neck cancer in a cohort of patients with advanced solid tumors (NCT00500903 and NCT00651664)^135,136. This trial established alisertib recommended doses and safety and allowed alisertib to proceed to phase II study in a large cohort of patients with nonhematological malignancies, including 55 patients with head and neck cancer (NCT01045421). However, only 4 head and neck patients attained a partial response, with median duration of 2.6 months (95% CI 1.5 – 3.1); more promising results were observed for breast cancer and small-cell lung cancer patients¹³⁷. Overall, alisertib has tolerable side effects, but some common adverse effects including neutropenia, anemia, nausea, and fatigue, which can be problematic at higher dose levels¹³⁶. To date, alisertib is the only AURKA inhibitor that has progressed to phase III clinical trials (e.g. NCT01482962). In patients with advanced solid tumors, monotherapy with a more recently developed AURKA inhibitor with higher tolerability, TAS-119, showed limited antitumor response; no patients showed complete or partial response, with 41% of patients showing stable tumor size (NCT02448589)¹³⁸. While stronger responses are observed in liquid tumors (leukemias and lymphomas), the limited responses to single agent Aurora inhibitors in solid tumors has focused attention on combination therapies.

Table 2.

Clinical trials using Aurora kinase inhibitors in HNSCC, and other advanced solid tumors.

Clinical Trials for Head and Neck Cancer
NCT	Therapy	Target	Tumor type	Phase	Trial start year	Status	Study Publication
NCT00424632	PF-03814735	AURKA; AURKB	Advanced Solid Tumors	I	2006	Completed	[134]
NCT00560716	CYC116	Pan-aurora; VEGFR2	Advanced Solid Tumors	I	2007	Terminated	N/A
NCT00500903	MLN8237	AURKA	Advanced Solid Tumors	I	2007	Completed	[136]
NCT00651664	Alisertib	AURKA	Solid Tumors	I	2007	Completed	[135]
NCT01540682	MLN8237; Cetuximab; Radiotherapy	AURKA; EGFR	Head and neck	I	2012	Completed	N/A
NCT01613261	Alisertib; TAK-733	AURKA; MEK1/2	Nonhematological Malignancies	I	2012	Study was withdrawn before participants were enrolled.	N/A
NCT01045421	Alisertib	AURKA	Nonhematological Malignancies	I/II	2010	Completed	[137]
NCT01639911	Alisertib; Pazopanib	AURKA; VEGFR; PDGFR	Solid Tumors	I	2013	Completed	[141]
NCT04555837	Alisertib; Pembrolizumab	AURKA; PD-1	Rb-deficient HNSCC	I/II	2020	Recruiting	N/A
NCT02448589	TAS-119	AURKA	Advanced Solid Tumors	I	2014	Terminated	[138]
NCT02812056	Alisertib; TAK-228	AURKA; mTORC1/2	Human Papilloma Virus (HPV) Associated Malignancies	I	2016	Study was withdrawn before participants were enrolled.	N/A
NCT01512758	Alisertib	AURKA	East Asian Participants with Advanced Solid Tumors or Lymphomas	I	2012	Completed	[161]
Clinical trials in Selected Tumor Types
NCT00497731	AZD1152	AURKB	Advanced Solid Tumors	I	2005	Terminated	N/A
NCT00424632	PF-03814735	AURKA; AURKB	Solid Tumors	I	2006	Completed	[134]
NCT00858377	AMG 900	Pan-aurora	Advanced Solid Tumors	I	2009	Completed	[162]
NCT01118611	GSK1070916A	AURKB; AURKC	Advanced Solid Tumors	I	2010	Completed	N/A
NCT01471964	MLN8237; Erlotinib	AURKA; EGFR	Non-Small Cell Lung Cancer	I/II	2011	Terminated	N/A
NCT01923337	Alisertib; Irinotecan	AURKA; Topoisomerase I	Advanced Solid Tumors or Colorectal Cancer	I	2013	Completed	[163]
NCT01924260	Alisertib; Gemcitabine Hydrochloride	AURKA; anti-metabolite for DNA synthesis	Solid Tumors or Pancreatic Cancer	I	2013	Completed	N/A
NCT02134067	TAS-119; Paclitaxel	AURKA; microtubules	Advanced Solid Tumors	I	2014	Terminated	[164]
NCT02038647	Alisertib; Paclitaxel	AURKA; microtubules	Small Cell Lung Cancer	II	2014	Completed	[165]
NCT02719691	MLN8237; MLN0128	AURKA; mTORC1/2	Advanced Solid Tumors and Metastatic Triple-negative Breast Cancer	I	2016	Completed	[140]
NCT03898791	LY3295668; Erbumine	AURKA; ACE	Small-Cell Lung Cancer	I	2019	Completed	N/A
NCT04085315	Alisertib; Osimertinib	AURKA; EGFR	Non-Small Cell Lung Cancer	I	2019	Recruiting	N/A
NCT04479306	Alisertib; Osimertinib; Sapanisertib	AURKA; EGFR; mTORC1/2	Non-Small Cell Lung Cancer	I	2020	Active, not recruiting	N/A
NCT05017025	LY3295668; Osimertinib	AURKA; EGFR	EGFR-Mutant Non-small Cell Lung Cancer	I	2022	Recruiting	N/A

The combination of alisertib and paclitaxel in the setting of advanced breast and ovarian cancer increased progression-free survival (PFS) compared to paclitaxel treatment alone (6.7 months, compared to 4.7 months in paclitaxel monotherapy patients)¹³⁹. In patients with advanced solid tumors, combination of the mTORC1/2 inhibitor sapanisertib, also called MLN0128, with alisertib resulted in only 2 patients experiencing dose limiting toxicity at the highest dose level (NCT02719691) and caused 31% of patients to have stable disease¹⁴⁰. Alisertib had also been evaluated with pazopanib, a multi-kinase angiogenesis inhibitor, in patients with solid tumors, including head and neck cancer, showing tolerable safety and tumor inhibition (NCT01639911). Pharmacokinetic analysis indicated that alisertib plasma concentrations were 40% higher with the combination treatment than with alisertib alone, suggesting that pazopanib reduces alisertib clearance, possibly due inhibition of cytochrome P450 isozyme 3A4 (CYP3A4), but further investigation is required^141,142. Currently, alisertib is being evaluated with the immune checkpoint inhibitor pembrolizumab in a phase I/II study for patients with recurrent or metastatic HPV+ HNSCCs (NCT04555837), based on a pre-clinical study that showed synthetic lethality of the AURKA inhibitor LY3295668 with an RB1 loss of function mutation which mimics the Rb deficiency found in HPV+ HNSCCs tumors¹²⁶.

Table 2 also shows the usage of Aurora kinase inhibitors in combination with other agents in completed and ongoing clinical trials for tumor types that have some commonalities with HNSCC. Non-small cell lung carcinoma, for example, often overexpress EGFR and AURKA and have mutated TP53, similar to HNSCC. EGFR-mutated NSCLC can have increased expression of TPX2, which is associated with EGFR inhibitor resistance via upregulation of AURKA⁸⁵. Although these trials do not include HNSCC patients, some pre-clinical studies have shown anti-tumor activity for these combinations in HNSCCs, which may suggest clinical studies in this cohort. The ongoing clinical trials should provide insight into effectiveness of novel AURKA inhibitor combinations, supporting their extension to HNSCC. Drugs being tested in combination with AURKA inhibitors target proteins including mTORC1/2, ACE, and EGFR.

Summary and Future Directions.

Therapy options for patients with advanced HNSCCs are still very limited. Considering the multiple mechanisms upregulating AURKA and AURKB in HNSCC, and their extensive tumor promoting roles, it is likely that ongoing work will reveal synthetic lethal interactions involving Aurora inhibitors and other targets that improve therapeutic windows for HNSCC. Aurora kinase inhibitors have been showing promising results in combination with a variety of drugs in different types of cancers, although many of these trials do not include head and neck tumors; extension of positive results to evaluation in HNSCC patients should be prioritized. To help guide the usage of Aurora kinase inhibitors in the clinic, it is also important to identify predictive biomarkers and regulators of response to these drugs, to aid in distinguish likely responders and non-responders, and develop useful therapeutic combinations. TP53 status has been of considerable interest as a response biomarker; one interesting study in triple negative breast cancer showed that the administration of alisertib to cells lacking TP53 results in senescence, rather than the apoptosis as seen in TP53 wild-type cells, with implications for recurrence¹⁴³. TPX2 expression is associated with resistance to AURKA inhibition, with TPX2 binding to AURKA altering the kinase ATP-binding pocket, reducing binding affinity and efficacy of various inhibitors¹⁴⁴. Similarly, NEDD9 binding of AURKA reduces AURKA inhibitor binding affinity⁸³, and shRNA depletion of NEDD9, combined with AURKA inhibition, greatly reduced xenograft tumor volume⁸³. Incorporation of TPX2, NEDD9, and other biomarkers during design of clinical trials may be successful in increasing rate of therapeutic response of Aurora inhibitors. Finally, understanding of Aurora biology continues to evolve, with new functions being identified (e.g., regulation of cilia-mediated intercellular signaling^145,146). Continued investigations in these areas may help nominate ways to better exploit Aurora kinases to improve HNSCC treatment.