Survival of Head and Neck Cancer Cells Relies upon LZK Kinase-Mediated Stabilization of Mutant p53

Head and neck squamous cell carcinoma (HNSCC) is driven by frequent copy-number alteration, the most common of which is gain of 3q, the long arm of chromosome 3 (1). Drivers identified on the 3q amplicon include PIK3CA (2, 3), TP63 (4–6), and SOX2 (7–9); however, targeting amplified PIK3CA has displayed limited success (10, 11), and small molecule inhibitors do not exist for SOX2 or p63, thus additional drivers must be identified to improve outcomes for HNSCC patients. We investigated the kinase LZK (encoded by MAP3K13, a resident gene on the 3q amplicon) as a novel target because it was previously reported to regulate the JNK (12, 13) and NFκB (14) pathways, which can be protumorigenic. There are few studies describing the function of LZK, but a previous study demonstrated that LZK and its close family member DLK play a role in synaptic growth. Specifically, the DLK/LZK Drosophila homolog, wallenda, is required for synaptic overgrowth (15), and overexpression of LZK can promote axon growth in neurons (16). In addition, DLK/LZK inhibitors have been developed and show activity in a Parkinson's disease mouse model (17).

Mutation or deletion of TP53 (encoding p53) is a frequent and early event in the development of HNSCC (1). Wild-type p53 is a critical tumor suppressor; however, many TP53 mutations produce a full-length protein, which is highly expressed in cancer cells (18). Several hotspot mutations have been identified and demonstrated to have oncogenic capabilities (reviewed in refs. 19 and 20), including the ability to transform p53-null cells and promote tumor growth in mice (21). Gain-of-function (GOF) p53 mutants are associated with antiapoptotic (22, 23), proproliferative (24–26), and proinvasive phenotypes (25, 27), and can promote resistance to chemotherapy (24, 28). GOF p53 mutants have been shown to upregulate cyclins A and B1 (26), hTERT (29), MDR1 (21), NFκB2 (30), and MYC (31) and downregulate p21, PTEN, BAX (22), and caspase-3 (28). GOF p53 has also been shown to interact with and inhibit AMPK (27), and modulate chromatin regulatory factors, including MLL1, MLL2, and MOZ (32). GOF-mutant p53 is stabilized by tumor cell–specific factors (33), the mechanisms of which are not fully understood. Currently, no therapeutics have been specifically developed to suppress the expression of stabilized mutant p53; however, HSP90 inhibitors were previously shown to reduce expression of mutant p53 (23). Here, we describe LZK as essential to maintain proliferation of HNSCC cells with 3q gain, via promoting stability of GOF-mutant p53.

LZK cDNA was prepared from RNA extracted from 293T cells, attB flanking regions were added by PCR, and the BP clonase reaction used to insert LZK into pDONR221. From here the Invitrogen Gateway system was used for cloning into destination vectors. FLAG-tagged (pReceiver-M12, GeneCopoeia) destination vector was converted into Gateway destination vector for use in transient overexpression assays. The pLenti6.3/TO/V5-DEST vector was used to generate stable overexpression. Mutations were introduced using a Site-Directed Mutagenesis Kit (Stratagene). Kinase dead (KD) construct used for LZK was a K195M mutant. For the shRNA-resistant LZK construct, mutations were as follows: g978t, t981c, a984g, t987a, c990g, t993c, and a996g.

293T, CAL33, and BEAS2B cells were transiently transfected using Attractene (Qiagen) according to manufacturer's protocol. BICR56 cells were transfected using Lipofectamine LTX & Plus reagent (Invitrogen) according to manufacturer's protocol. A pcDNA3.1(+) vector (Invitrogen) was used as an empty vector control where required. DharmaFECT1 (Thermo Scientific) was used for siRNA transfections. siRNA used for LZK was obtained from Origene and used at 10 nmol/L. Sequences were as follows: GGAACACGAUGAAUCAGAGACGGCG (siLZKA), GGCGAAUAAUUUAUACAUGGAAUTG (siLZKB), and AGAACAGAAUGAGACGGAUAUCAAG (siLZKC). siRNA to p53 was ordered from Santa Cruz (si-p53#1; sc-29435) and Cell Signaling Technology (si-p53#2; 6231) and used at 100 nmol/L. Nontargeting control siRNA was ordered from Dharmacon (D-001810-01-20) and used at the same concentration as siRNA to the appropriate target. Staurosporine was purchased from Cell Signaling Technology.

CAL33 (DMSZ, obtained October 2012) and 293T (ATCC, July 2012) cells were maintained in DMEM D6546 (Sigma) supplemented with 10% FBS (Labtech), 1% penicillin–streptomycin (pen–strep; Gibco), and 2 mmol/L GlutaMAX (Gibco). BICR6 (March 2015), BICR16 (January 2014), BICR22 (October 2013), and BICR56 (November 2012 and April 2014; all obtained from Public Health England) cells were grown in DMEM D6546 with 10% FBS, 1% pen–strep, 0.4 μg/mL hydrocortisone (Sigma), and 2 mmol/L GlutaMAX. MSK921 (MSKCC, July 2014) and BEAS2B (ATCC, October 2012) cells were maintained in RPMI 1640 + GlutaMAX (Gibco) with 10% FBS and 1% pen–strep. PE-CA/PJ15 (PJ15) and PE-CA/PJ34 (PJ34; both obtained January 2014 from Public Health England) were cultured in IDMEM with 10% FBS, 1% pen–strep, and 2 mmol/L GlutaMAX. DETROIT562 (ATCC, November 2014) cells were maintained in EMEM (ATCC) with 10% FBS and 1% pen–strep (Gibco). 293FT (Invitrogen, November 2011) were maintained in DMEM with 10% FBS, 4 mmol/L GlutaMAX, 1 mmol/L sodium pyruvate (Gibco), and 0.1 mmol/L NEAA (Gibco). OKF6/TERT2 cells (Harvard Skin Disease Research Center, June 2013) were seeded in K-sfm (Gibco) with accompanying supplements: 25 μg/mL bovine pituitary extract, 0.2 ng/mL EGF, 0.3 mmol/L CaCl₂, and 1% pen–strep. At 30% confluency, medium was changed for 1:1 K-sfm:DFK to allow growth to full confluence. DFK was composed of DMEM/F12 (Gibco) with 2 mmol/L GlutaMAX, 25 μg/mL bovine pituitary extract (Gibco), 0.2 ng/mL EGF (Gibco), 0.3 mmol/L CaCl₂ (Lonza), and 1% pen–strep. Incubation was at 37°C and 5% CO₂ for all cells. Cell lines in regular use were subject to authentication by yearly STR profiling (conducted by multiplex PCR assay by Applied Biosystems Ampflstr system). STR profiles were compared with ATCC and DMSZ databases. However, no profile was available for BICR6, BICR22, MSK921, or OKF6/TERT2 cell lines, and BICR16, PJ15, and PJ34 were only removed from freeze for a short time after receipt for collection of DNA and RNA so were not authenticated. 3q status of all HNSCC and immortalized control cell lines were verified in-house. All cell lines were used in experiments for fewer than 20 passages (10 weeks) after thawing, before a fresh vial was taken from freeze. Cell lines in use were confirmed to be mycoplasma-negative every 1 to 2 months.

CAL33, BICR56, OKF6/TERT2, BEAS2B, DETROIT562, and BICR6 inducible knockdown cells were generated by Sirion Biotech. MSK921 and BICR22 were generated in-house using lentiviral particles provided by Sirion (generated by transfection of 293TN cells with expression vectors and lentiviral packaging plasmids). Transduction was at MOI 5 with 8 μg/mL polybrene. After 24 hours, the medium was replaced with fresh medium containing puromycin (Invitrogen) to select for cells that had been effectively transduced. shRNA sequences were: CGGAATGAACCTGTCTCTGAA (sh1), GATGTAGATTCTTCAGCCATT (sh2), and AAGAGCCGATATCGAAGCAAA (sh3). The lentiviral expression plasmid was pCLVi(3G)-MCS-Puro, which expresses a doxycycline-responsive transactivator and the shRNA from the same vector. Expression of the transactivator is constitutive, while expression of shRNA is dependent on a doxycycline-inducible promoter. Binding of doxycycline to the transactivator allows it to bind the doxycycline-inducible promoter and promote expression of the shRNA. Doxycycline (Sigma) was used at 1 μg/mL to induce knockdown of LZK.

The ViraPower HiPerform T-Rex Gateway Expression System (Invitrogen) was used to generate cells with tetracycline-inducible expression of LZK. Briefly, LZK (cloned into pLenti6.3/TO/V5-DEST vector) and pLenti3.3/TR (for tetracycline repressor expression) were transfected into 293FT cells using Lipofectamine2000 to generate lentiviral stock. Parental OKF6/TERT2 and BEAS2B cells were transduced with lentiviral stocks in a stepwise fashion and cell lines generated by antibiotic selection [Blasticidin (Invitrogen) and Geneticin (Gibco)]. Tetracycline (Invitrogen) was used at 1 μg/mL to induce expression of LZK.

Forty-eight hours after treatment (transient transfection, tetracycline-induced overexpression, or doxycycline-induced knockdown), cells were lysed using RLT buffer (Qiagen) with 1% v/v 2-mercaptoethanol (BioRad). Removal of genomic DNA and preparation of RNA was conducted using an RNeasy kit (Qiagen) according to the manufacturer's protocol. The quantity of RNA was determined using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies).

RT-PCR was performed using SuperScript III One-Step RT-PCR kit (Invitrogen). Primers used were as follows: AACTGATTCGAAGGCGCAGA (LZK forward); GGGCGTTTTCCAAGAGAGGA (LZK reverse); CCATGGAGAAGGCTGGGG (GAPDH forward); GTCCACCACCCTGTTGCTGTA (GAPDH reverse). The cycling conditions for PCR were as follows: cDNA synthesis and predenaturation (1 cycle at 55°C for 30 minutes followed by 94°C for 2 minutes); PCR amplification (25 cycles of denaturing at 94°C for 15 seconds, annealing at 55°C for 30 seconds and extension at 68°C for 60 seconds) and a final extension at 68°C for 5 min. PCR products were resolved on 2% agarose gel and visualized with Nancy-520 (Sigma) DNA gel stain under UV light.

The Power SYBR Green RNA-to-C_T 1-Step Kit (Applied Biosystems) was used for qRT-PCR. Primers used were as follows: CCTTTGTCCGGAAGTCCCAAAATGTC (LZK forward); GAAGGGTATTGGGATTGAGCTTGGTG (LZK reverse); CCAACCGCGAGAAGATGA (ACTB forward); CCAGAGGCGTACAGGGATAG (ACTB reverse); TTCTGGCCTGGAGGCTATC (B2M forward); TCAGGAAATTTGACTTTCCATTC (B2M reverse). Samples were run on an ABI 7900 HT real-time PCR instrument, with the following program: reverse transcription (48°C, 30 minutes), activation of DNA polymerase (95°C, 10 minutes), PCR amplification (40 cycles of denaturing at 95°C for 15 seconds and annealing/extension at 60°C for 1 minutes). SDS software was used to compute C_T values. Graphs display relative expression of MAP3K13 compared with a reference gene and a control sample as stated in the relevant figure legend.

After appropriate treatment time, cells were lysed on ice with Triton X-100 lysis buffer (Cell Signaling Technology) supplemented with protease inhibitor tablet (Roche). Antibodies used were as follows: pJNK(T183/Y185), p-p53(S392), GAPDH, pIKBα(S32), pAMPKα(T172), AMPKα, p21, anti-rabbit, anti-mouse (Cell Signaling Technology), α-tubulin, Flag M2 (Sigma), p53 (Santa Cruz), LZK (abcam), p-p38(T180/Y182) (Genway), MDM2 (Millipore), and MDM4 (Bethyl laboratories).

Two phosphoarrays were used: a phospho-kinase array and a MAPK array (both from R&D Systems). Samples were lysed 48 hours after induction by tetracycline or doxycycline as appropriate. Protocols were as per manufacturer's instructions.

Cells were seeded into 96-well plates in triplicate. For transient knockdown, cells were transfected at the same time as seeding and analyzed 72 hours later. For inducible knockdown, doxycycline was added into the medium the following day and analyzed 72 hours later. MTT Kit (Sigma) was used to assess viability, using protocol as per manufacturer's instructions.

For MTS assays, Cell Titer 96 Aqueous One Solution cell proliferation assay (Promega) was used, following the manufacturer's protocol. Briefly, cells were reverse transfected with the corresponding p53 siRNA or negative control. Twenty-four hours later, cells were collected and seeded into 96-well plates in triplicate for MTS assay and 6-well plates for protein extraction. Doxycycline was added where appropriate and cells were incubated for 72 hours. MTS was added, incubated for 4 hours, and absorbance measured at 490 nm.

Cells were seeded into 96-well plates in triplicate, and doxycycline added into the medium the following day. Cells were treated for 48 hours before incubation with BrdUrd label for a further 24 hours prior to analysis by Calbiochem kit according to the manufacturer's instructions.

Crystal violet assays were used to assess relative cell growth and survival after knockdown. For transient knockdown, cells were transfected at the same time as seeding and analyzed 72 hours later. For inducible knockdown, cells were seeded into 6-well plates in triplicate, treated with doxycycline the following day, and plates analyzed after 72 hours of treatment. Cells were then washed with PBS and fixed in ice-cold methanol, before staining with 0.5% Crystal violet in 25% methanol. For quantification, Crystal violet stain was dissolved in 10% acetic acid, incubated for 20 minutes with shaking, and read at 595 nm.

Cells were seeded into 6-well plates in triplicate at 300/well (BEAS2B and OKF6/TERT2 controls) or 1000/well (HNSCC cell lines), and treated with doxycycline the following day. Medium was refreshed every 48 hours for 14 days. Cells were stained with Crystal violet as described above.

The “3D on-top assay” (34) was used to assess growth in matrigel. Matrigel used was Growth Factor Reduced, Phenol Red-Free Basement Membrane Matrix from mouse EHS tumor (Corning). Matrigel was thawed at 4°C overnight. Plates and tips were pre-chilled prior to experiment, and the experiment was conducted on ice, except where specified. Twelve-well plates were coated with 200 μL Matrigel and incubated at 37°C for 30 minutes. Cells were plated at 80,000/well in 0.5 mL medium and left to settle for 30 minutes at 37°C. 0.5 mL 5% Matrigel in medium was added gently to each well. The following day, cells were treated with 1 μg/mL doxycycline where required. After 72 hours of treatment, plates were observed with an Axiovert 40 CFL Inverted Microscope, and photographs taken with the aid of AxioVision software. The “Color Threshold” tool in ImageJ was used to assess average colony size. Three images per condition were used for measurement, and average colony size was calculated for each condition.

Cells were seeded into 6-well plates and treated with doxycycline the following day. After 48 hours, media with floating cells were collected to FACS tubes and plates trypsinized. Trysinized cells were added to the FACS tubes with the media that were previously collected, and the tubes spun for 3 minutes at 2,000 rpm. Supernatant was discarded and 1 ml prechilled ethanol added dropwise to each tube while vortexing. Cells were kept at −20°C until analysis. A BD Cycletest Plus DNA Reagent Kit was used for sample staining, according to the manufacturer's instructions. Samples were run on a BD 3-color FACSCalibur, and results were analyzed using FlowJo.

Animal work was approved by the CRUK-MI Ethical Review Committee and was performed within the limits of a license granted by the Home Office according to the Animals (Scientific Procedures) Act 1986.

For inducible knockdown xenografts, 2 × 10⁶ CAL33-sh1 cells were injected s.c. into the right posterior flanks of 6- to 8-week-old immunodeficient NOD-scid IL2Rgamma^null (NSG) female mice (32 mice, obtained from Charles River). Mice weighed between 19 and 24 g at the start of experiment. When tumors reached a volume of approximately 200 mm³, mice (10 per group) were randomly assigned to treatment with control or doxycycline diet (ssniff).

For parental cell xenografts, 2 × 10⁶ CAL33 cells were injected s.c. into the right posterior flanks of 8–10 week-old immunodeficient NSG female mice (8 mice; obtained from Charles River). Mice weighed between 20 and 24 g at the start of experiment. When tumors reached a volume of approximately 200 mm³, mice (4 per group) were assigned to treatment with control or doxycycline diet (ssniff).

Tumor formation was monitored every few days and tumor volume based on caliper measurements was calculated by the formula: tumor volume = (length × width × width)/2. Mice were culled when tumors reached maximum permitted volume 1,200 mm³ or after 5 weeks after injection of cells.

Error bars shown on graphs represent ± SEM. Two-tailed Welch t-test was used to assess the significance of differences between groups for assays. Analysis of variance (ANOVA) was conducted to assess the significance of differences in mRNA expression between groups based on GISTIC copy numbers.

DNA from cell lines was prepared using a Qiagen Blood & Tissue kit, according to the manufacturer's instructions. Copy-number analysis was conducted by paired-end whole genome sequencing (2 × 100 bp reads) on an Illumina HiSeq 2000. Thirty million pairs per samples were generated, which gave ∼2-fold coverage of the genome. The most common ploidy was defined as “normal” for the purposes of assessing copy-number gain and loss. Analysis was run according to previously published procedures (35). No viral genomes were detected in any sample, so cell lines were determined to be HPV-negative with 95% confidence (36).

Examination of data from The Cancer Genome Atlas (TCGA) revealed that high-level amplification of MAP3K13 was present in 20% of HNSCC tumors (Fig. 1A), while an additional 56% had copy-number gain (37). Furthermore, MAP3K13 was within the top 100 most frequently amplified genes on chromosome 3 (Fig. 1B). Further analysis of TCGA tumor samples indicated that higher MAP3K13 copy number was associated with significantly increased mRNA expression (Fig. 1C). In addition, data from the Cancer Cell Line Encyclopedia (CCLE) showed that upper aerodigestive tract (UADT) cell lines have the highest levels of MAP3K13 mRNA compared with all other available cancer cell lines (Supplementary Fig. S1A; ref. 38).

LZK expression levels were assessed in a panel of HNSCC cell lines with 3q gain based on CCLE data. Control cell lines included BEAS2B immortalized bronchial epithelial cells and OKF6/TERT2 immortalized oral keratinocytes; both cell lines were diploid with no change at 3q (Fig. 1D). Copy-number analysis confirmed the presence of the 3q amplicon in almost all HNSCC cell lines tested (Fig. 1D). MSK921 was the only HNSCC cell line tested that had no alteration at 3q (Fig. 1D). BICR22 was notable for the fact that its region of copy-number gain did not encompass MAP3K13 (Fig. 1D). Therefore, BICR22 and MSK921 were chosen as control HNSCC cell lines. qRT-PCR showed that a majority of the cell lines with copy-number gain of MAP3K13 had increased mRNA expression (Fig. 1E), confirming data observed for TCGA human primary tumor samples. Finally, Western blot analysis indicated increased LZK protein levels in cells with 3q gain compared with BEAS2B cells (Supplementary Fig. S1B).

To determine whether LZK is important for maintaining tumorigenic phenotypes, siRNA-mediated knockdown was conducted. Three individual siRNAs effectively depleted LZK protein levels in CAL33 and BICR56 HNSCC cells, and in OKF6/TERT2 and BEAS2B control lines (Fig. 2A). All three LZK siRNAs caused a highly significant reduction in viability in CAL33 and BICR56 cells, but had no effect on BEAS2B or OKF6/TERT2 cells (Fig. 2B). In addition, Crystal violet staining, which was used to assess relative cell growth and survival, showed that knockdown of LZK decreased the cell density of CAL33 and BICR56 cells (Fig. 2C and D). No effect was observed in OKF6/TERT2 controls, although a minor reduction in cell density was seen for BEAS2B cells with siLZKB and siLZKC (Fig. 2C and D).

To validate these data, HNSCC cells with doxycycline-inducible expression of shRNA targeting MAP3K13 were generated (Fig. 3A and B). Induced knockdown of LZK led to a 40% to 50% reduction in viability of CAL33 and BICR56 cells, but had no significant effect on viability of MSK921 or BICR22 control cell lines that lack MAP3K13 copy-number gain (Fig. 3C). Consistent with these results, there was a ∼50% decrease in Crystal violet staining in CAL33 and BICR56 cells after doxycycline-induced LZK knockdown, but no effect on MSK921 or BICR22 cells (Supplementary Fig. S2A and S2B). To confirm these data in a 3D setting that more closely mimics the tumor microenvironment, cells were grown in Matrigel; colony sizes of CAL33 and BICR56 cells were reduced after LZK knockdown, whereas no significant effect was seen for control cell lines (Fig. 3D and E). Clonogenicity was then assessed; a near-complete ablation of colony forming ability was observed for CAL33 and BICR56 cells after depletion of LZK (Fig. 3F; Supplementary Fig. S2C). MSK921 cells showed a smaller reduction in colony formation after doxycycline treatment; this effect occurred at least in part due to the toxicity of doxycycline, which also affected the parental cell line (Fig. 3F; Supplementary Fig. S2C). Doxycycline has been reported to have toxic effects on some cell lines (39, 40), and this toxicity is likely more pronounced when cells are seeded at a low density (41), which could explain why these effects were not seen in short-term assays. LZK depletion did not suppress growth of BICR22 cells (Fig. 3F; Supplementary Fig. S2C).

Critically, reexpression of the sh1-resistant LZK construct into CAL33-sh1 and BICR56-sh1 cells fully or partially, respectively, rescued the effect of LZK knockdown on cell density, confirming that these effects were specific to LZK depletion (Fig. 4A). LZK knockdown was then conducted in additional control and HNSCC cell lines (Fig. 4B). shRNA-mediated depletion of LZK resulted in a significant reduction in colony formation in two additional HNSCC lines with 3q gain (BICR6 and DETROIT562), but not in BEAS2B or OKF6/TERT2 immortalized control cell lines (Fig. 4C and D). To determine the importance of LZK in vivo, we established a murine xenograft model using the CAL33-sh1 cell line. There was a significant reduction in tumor growth of 39% ± 16% upon doxycycline treatment (Fig. 4E; note doxycycline diet had no significant effect on parental CAL33 xenografts (Supplementary Fig. S3A). These results highlight the importance of LZK as a regulator of viability in HNSCC.

Phenotypic assays were then carried out to investigate potential mechanisms of reduced viability. BrdUrd assays showed that depletion of LZK significantly reduced proliferation of CAL33 and BICR56 cells, while no effects were observed in control cells (Fig. 5A). The decrease seen was to a similar extent as that observed in viability and density assays, suggesting that this proliferative decrease may be the main mechanism underpinning these effects. To confirm these results, propidium iodide staining was used to assess the effect of LZK knockdown on progression through the cell cycle. An increase in the percentage of cells in G₁ phase was observed for CAL33 and BICR56 inducible knockdown cells depleted of LZK, but not for MSK921 control cells (Fig. 5B; Supplementary Table S1). This correlated with an increase in G₁–S ratio, a measure of cell-cycle arrest, of approximately 3-fold for CAL33 and 2-fold for BICR56 (Fig. 5C). These data verify that LZK regulates HNSCC viability through regulation of the cell cycle. Additionally, PARP cleavage, a marker of apoptosis, was not observed after LZK knockdown (Supplementary Fig. S 3B), suggesting that LZK does not regulate cell death in HNSCC.

To identify pathways involved in the proproliferative effect of LZK, we monitored the activation status of pathways described to be activated by LZK: JNK (12, 13), p38 (13), and NFκB (Supplementary Fig. S4A; ref. 14). In overexpression assays we observed activation of JNK, but not p38 or NFκB (Supplementary Fig. S4B–S4F). However, in BICR56 and CAL33 cells, there was no decrease in phospho-JNK after transient or inducible knockdown (Supplementary Fig. S4G and S4H; note that minor reductions were observed for MSK921 cells, but this was also seen in parental MSK921 cells, suggesting that doxycycline may be responsible for the reduction in pJNK in this cell line). This suggested that these pathways were not responsible for the reduced proliferation observed after LZK depletion.

To elucidate other pathways regulated by LZK, a phospho-kinase array was conducted using CAL33-inducible knockdown cells. The most obvious differences seen were for three p53 phosphorylation sites (S15, S46, and S392; Fig. 6A). The phospho-p53 results were validated by Western blot for the S392 phospho-site; however, it was found that total p53 levels were reduced after LZK knockdown (Fig. 6B), accounting for the observed decrease in phospho-p53 at the three phospho-sites. Effects on p53 were then investigated in BICR56 cells; p53 levels were also markedly reduced in this cell line after depletion of LZK (Fig. 6C). No reductions in total p53 were observed in MSK921 cells (Fig. 6C; note doxycycline treatment alone did not alter p53 levels in parental cells; Supplementary Fig. S5A).

Notably, CAL33 has a known GOF mutati on in p53 (R175H), so a decrease in total p53 would be expected to lead to reduced proliferation. In contrast, the MSK921 control cell line harbors wild-type p53. BICR56 has a mutation in p53 that leads to the exclusion of seven amino acids (Y126–K132) at the beginning of exon 5. To date, this mutant has not been characterized. Transient knockdown using siRNA was conducted to investigate the function of p53 in these cell lines (Supplementary Fig. S5B and S5C). Knockdown of p53 in CAL33 cells led to a 40% to 60% reduction in viability (Fig. 6D; Supplementary Fig. S5C), indicating that LZK primarily promotes viability by maintaining expression of GOF-mutant p53 in these cells. In BICR56 cells, knockdown of p53 led to a smaller but significant decrease in viability of 30% to 40% (Fig. 6D; Supplementary Fig. S5C), suggesting that these cells also require expression of stabilized mutant p53.

To determine if E3 ubiquitin ligases MDM2 and MDM4 were contributing to loss of expression of GOF-mutant p53, expression of these two proteins was monitored in cells depleted of LZK. Endogenous MDM4 was not detected in any of the HNSCC cell lines and endogenous MDM2 was only detected in the CAL33 cells (Supplementary Fig. S6A). There was no increase in MDM2 protein levels in CAL33 cells depleted of LZK (Supplementary Fig. S6B). Overall, these data indicate that regulation of these E3 ubiquitin ligases is not an underlying mechanism contributing to loss of p53 expression in HNSCC cells depleted of LZK.

To further understand the mechanism by which the LZK–GOF p53 pathway is promoting increased proliferation, we assessed activation and expression of targets downstream of GOF-mutant p53 (22, 27). Depletion of LZK resulted in increased levels of p21 and phospho-AMPKα in CAL33 and BICR56 cells (Fig. 6E), consistent with previous reports that GOF-mutant p53 suppresses these downstream targets. These results highlight a novel signaling pathway where LZK is required to maintain expression of GOF-mutant p53, which regulates activation of AMPK and expression of p21. Finally, re-expression of shRNA-resistant LZK in CAL33 and BICR56 cells restored levels of mutant p53, demonstrating that this effect was specifically due to depletion of LZK (Fig. 6F).

HNSCC accounts for approximately 600,000 cases and 350,000 deaths per year worldwide (42, 43). A majority of patients present with stage III or IV disease, and, after treatment, 50% will have disease progression within 2 years (44, 45). Five-year survival rates for patients who present with advanced disease are under 50% (46). At present, the current treatment regimens for later stages involve chemotherapy and/or radiotherapy, which carry significant toxicities (45, 47). Chemotherapy or radiotherapy can be enhanced by combination with cetuximab, which targets EGFR, in advanced stages of the disease (48, 49). EGFR is overexpressed in ∼90% of HNSCC patients; however, only a small subset (13%) of patients respond to cetuximab monotherapy (50). Additionally, the inclusion of cetuximab to a regimen of chemoradiotherapy was not found to have any additional benefit, regardless of EGFR expression levels (51). Cetuximab is the only FDA-approved targeted therapy for the treatment of HNSCC patients, and a recent review of current clinical trials for HNSCC indicated that a majority of trials are centered around EGFR inhibitors, including gefitinib and erlotinib, in combination with chemotherapy and/or radiotherapy (43). This highlights the intense need for the development of new therapies for the treatment of this disease.

The development of novel therapies has been hampered by the relative rarity of targetable mutations in HNSCC. The exception is PIK3CA, which is mutated in ∼20% of HNSCC samples (1, 10), representing a cohort of patients that may benefit from treatment with PI3K/AKT inhibitors. In addition, recent studies have indicated that inhibitors targeting the mTOR and ERK pathways hold promise for treatment of HNSCC patients, including patients positive for PIK3CA mutations (52–54).

Rather than targetable mutations, the most common genetic alterations in HNSCC are somatic copy-number alterations, the most frequent of which is copy-number gain at 3q (1). Multiple genes on the 3q amplicon can promote tumorigenic phenotypes, including PIK3CA (2, 3), TP63 (4–6), and SOX2 (7–9), and detailed studies must be undertaken to define the contribution of the various genes on this amplicon to tumor initiation and maintenance. MAP3K13 (LZK) is another resident gene of the 3q amplicon. Here, we have shown that knockdown of LZK reduced proliferation specifically in HNSCC cells with 3q gain, an effect that was rescued by reexpression of shRNA-resistant LZK. Our data identify LZK as novel genetic dependency in HNSCC tumors harboring the 3q amplicon, and suggest LZK as a potential therapeutic target in HNSCC. Future studies will explore the potential of LZK inhibitors for the treatment of head and neck cancers harboring the 3q amplicon, both alone and in combination with other inhibitors, for example, targeting PI3K, AKT, and mTOR.

Knockdown of LZK led to a reduction in protein levels of GOF-mutant p53 in cells with 3q gain; this was also rescued by reexpression of shRNA-resistant LZK. GOF-mutant p53 has been well documented to promote proliferation (24–26). In contrast to the wild-type protein, GOF-mutant p53 is found at constitutively high levels in cancer cells (18). However, the mutant proteins are not inherently more stable than the wild-type, as demonstrated by expression in normal tissue compared with tumor tissue in transgenic mice (33); this implies that GOF mutants are stabilized by tumor cell–specific factors. Our study highlights amplified LZK as one potential contributor to the maintenance of mutant p53 stability in HNSCC cells. When LZK was depleted, we also observed increases in levels of phospho-AMPK and p21, both of which are negatively regulated by GOF p53 (22, 27). Based on these observations, we propose a model whereby amplified LZK promotes stability of GOF-mutant p53, which in turn promotes proliferation of HNSCC cells via regulation of its downstream targets, including p21 and AMPK (Fig. 6G). The mechanism by which LZK stabilizes mutant p53 will be the focus of future studies.

In summary, our data indicate that LZK promotes proliferation in HNSCC with 3q copy-number gain, via maintaining expression of stabilized mutant p53. Enzymes that regulate stability of GOF-mutant p53 are attractive targets given the incidence of stabilized mutant p53 across all cancers. It was recently reported that over 11 million cancer patients live with stabilized mutant p53 (23); thus, LZK may represent a druggable genetic dependency for a subset of these patients where LZK is amplified.

No potential conflicts of interest were disclosed.

Conception and design: Z.C. Edwards, E.W. Trotter, J. Brognard

Development of methodology: Z.C. Edwards, E.W. Trotter, P. Torres-Ayuso, J. Brognard

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Z.C. Edwards, E.W. Trotter, P. Torres-Ayuso, H.M. Wood, K. Nyswaner

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Z.C. Edwards, E.W. Trotter, P. Torres-Ayuso, P. Chapman, H.M. Wood, K. Nyswaner, J. Brognard

Writing, review, and/or revision of the manuscript: Z.C. Edwards, P. Torres-Ayuso, H.M. Wood, K. Nyswaner, J. Brognard

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Brognard

Study supervision: Z.C. Edwards, J. Brognard

We thank Lisa Doar and Adam Freestone from the BRU for technical support with the xenograft mouse model, MBCF for sequencing services, cell line authentication and mycoplasma testing, and Sudhakar Sahoo for copy-number analysis. Our thanks also go to Sirion Biotech for generating the doxycycline-inducible system.

This research was primarily supported by Cancer Research UK (Z.C. Edwards, E.W. Trotter, P. Torres-Ayuso, P. Chapman and J. Brognard), including under grant number C5759/A20971 (Z.C. Edwards and J. Brognard). P. Torres-Ayuso received funding from Fundación Ramón Areces. J. Brognard, P. Torres-Ayuso, and K. Nyswaner were funded by the National Cancer Institute, grant number ZIA BC 011691.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.