Histone deacetylase inhibitors suppress aggressiveness of head and neck squamous cell carcinoma via histone acetylation-independent blockade of the EGFR-Arf1 axis

HNSCC metastatic cell lines HN4, HN12, HN30 and HN31 were a gift from Dr. W. Andrew Yeudall [13]. All cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum at 37 °C in a humidified incubator supplied with 5% CO₂. Arf1 activation was determined by the glutathione resin-bound GST-GGA3-PBD fusion protein as described previously [15, 17]. Western blotting, wound closure assays, and cell proliferation assays were carried out as described previously [13, 18, 19].

TSA, PXD101 and erlotinib were purchased from Selleckchem (Houston, TX). MG132 and recombinant human EGF were purchased from Sigma-Aldrich (St Louis, MO) and ProSpecBio (East Brunswick, NJ), respectively. The Arf1 dominant negative and constitutively active constructs pcDNA3-HA-Arf1 DN-T31 N (Arf1DN) and pcDNA3-HA-Arf1-ActQ71L (Arf1CA) were purchased from Addgene (Plasmid #10833 and #10832). Antibodies that recognize acetyl-Histone H3 (Lys9/Lys14), acetyl-Histone H4 (Lys8), p-AKT (Ser473), AKT, p-ERK1/2 (Thr202/Tyr204), ERK1/2, p-STAT3 (Tyr705), STAT3, p-Src (Tyr416), Src, p-EGFR (Tyr845), EGFR, p-ErbB2 (Tyr1221/1222), ErbB2, p-ErbB3 (Tyr1289) and ErbB3, were purchased from Cell Signaling Technology (Beverly, MA). β-actin and PY20 antibodies were purchased from Sigma-Aldrich (St Louis, MO). CellTiter 96® AQueous One Solution Cell Proliferation Assay (MTS) Kit was obtained from Promega (Madison, MI).

HDAC activity was measured with the fluorometric HDAC Activity Assay kit (Abcam, Cambridge, MA) according to the manufacturer’s instruction. Briefly, the cell lysates with or without TSA treatment were sonicated, cleared, and incubated with assay buffer containing the HDAC substrate [Boc-Lys(Ac)-AMC] for 30 min at 37 °C. The reaction was terminated, and the fluorescence intensity was measured in a fluorescence plate reader (Ex/Em = 350–380/440–460 nm).

The Proteome Profiler Human Phospho-RTK Array Kit (R&D Systems, Minneapolis, MN) was used to determine phosphor-RTK profiling according to the manufacturer’s instructions. Briefly, a total of 500 μg fresh protein was diluted and incubated overnight with nitrocellulose membranes dotted with duplicate spots for 42 anti-RTK and control antibodies. Bound phospho-RTKs were detected with a pan antiphosphotyrosine antibody conjugated to horseradish peroxidase using ECL reagents from Bio-Rad (Hercules, CA).

In vitro protein-protein interactions were assessed by IP as described previously [20, 21]. Briefly, a total of 500 μg of cell lysate was used and diluted in 500 μl IP lysis buffer (containing 20 mM Tris·HCl-pH 8.0, 137 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, and a mixture of protease and phosphatase inhibitors) with the corresponding antibodies. The lysate was incubated with gentle rotation overnight at 4 °C, and reaction mixtures were incubated with Protein A/G Sepharose® (Abcam, Cambridge, MA) for another 4 h at 4 °C. The immunoprecipitated proteins were washed three times with a buffer (containing 10 mM Tris ·HCl-pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, and 1% Triton X-100) followed by SDS-PAGE analysis with the indicated antibodies.

Given that histones in mammalian cells treated with HDAC inhibitors are acetylated to an unusually high extent, we sought to determine the changes in histone acetylation levels in HNSCC cells in the presence or absence of TSA. Western blotting analysis revealed that TSA at 5 μM induced pronounced histone H3 and H4 hyperacetylation (Fig. 1a), and suppressed HDAC activity (Fig. 1b) in both HN12 and HN31 cells. MTS, wound healing and Borden chamber assays further showed that TSA has the strong potential against HNSCC cells, as evidenced by decreased proliferation, migration and invasion of HN12 and HN31 cells (Fig. 1c-1e). PXD101, another potent HDAC inhibitor, was also used in this study, which showed the similar effects observed from TSA treatment (Fig. 1a-1f). However, compared with PXD101, TSA at the same dose has superior anticancer activity in HNSCC cells (Fig. 1c-1f).

RTK-regulated pathways play key roles in various facets of cancer progression [22, 23]. To better understand the mechanisms of TSA action in HNSCC cells, we first determined the extent and duration of tyrosine phosphorylation in the presence or absence of TSA. TSA at 5 μM impaired global tyrosine phosphorylation in HN12 and HN31 cells (Fig. 2a). We then determined the phosphorylation status of RTKs following TSA treatment using Proteome Profiler Human Phospho-RTK Array Kit, which showed that TSA reduced the phosphorylation levels of most RTKs examined in this study (Fig. 2b). Among them, the phosphorylation levels of EGFR were reduced to half in HN12 cells exposed to TSA (Fig. 2b and c). Other EGFR family members, ERBB2 and ERBB3, their phosphorylation was also markedly downregulated upon TSA treatment (Fig. 2b and c).

To determine whether TSA affects total protein amount of EGFR family members, HN12 and HN31 cells were incubated in the presence or absence of TSA for 24 h, and cell lysates were collected for Western blotting. This analysis confirmed the findings obtained from Phospho-RTK Array that TSA suppressed the phospho-activation of EGFR family proteins (Fig. 3a). Most importantly, TSA not only inhibited EGFR phosphorylation, but also induced repression of EGFR total protein amount in both HN12 and HN31 cells (Fig. 3a). There were no changes in ERBB2 and ERBB3 protein levels in the presence or absence of TSA (Fig. 3a). Not surprisingly, EGFR-mediated downstream signaling molecules, including STAT3, Src, ERK1/2 and AKT, their phosphorylation levels were downregulated following TSA treatment (Fig. 3a and b). PXD101 had a similar role in inhibition of EGFR signaling, but it was less efficient compared with TSA at the same concentrations (Fig. 3b). To further elucidate the mechanism of TSA-induced EGFR inhibition, HN12 and HN31 cells were treated with different concentrations of TSA, which showed that TSA dose-dependently inhibited EGFR at the protein level (Fig. 3c). Moreover, TSA at 5 μM exerted a suppressive effect on EGFR at 30 min (Fig. 3d). A more dramatic reduction in EGFR protein amount was seen when these cells were treated with TSA for 12 h, and it reduced to less than 10% at 24 h after treatment (Fig. 3d). Real-time RT-PCR showed that no changes in EGFR mRNA levels upon TSA treatment (data not shown). We next used the proteasome inhibitor MG132 in combination with TSA to determine whether EGFR can be degraded through the proteasome pathway following TSA treatment. Consistently, treatment with TSA alone resulted in reduced EGFR protein levels (Fig. 3e). However, addition of MG132 to TSA led to a significant increase in EGFR protein amount in both HN12 and HN31 cells compared with TSA treatment alone (Fig. 3e). To determine whether EGFR undergoes polyubiquitination in TSA treatment, MG132-pretreated HN12 and HN31 cells in the presence or absence of TSA were immunoprecipitated with anti-EGFR antibody. The EGFR immunocomplex displayed an increase in total ubiquitination in the presence of TSA (Fig. 3f), suggesting that ubiquitin-proteasome-dependent degradation is the main machinery involved in TSA-induced EGFR repression in HNSCC cells.

Arf1 activation can be regulated by RTKs [24]. As TSA can induce repression of RTKs in HNSCC cells (Fig. 2), we determined whether TSA has the ability to regulate Arf1 function. TSA, at 2 or 5 μM, did not alter total Arf1 protein levels in both HN12 and HN31 cells (Fig. 4a). We thus performed pull-down experiments with a GST-GGA3-PBD fusion protein to determine Arf1 activation under the influence of TSA. Interestingly, the levels of active GTP-bound Arf1 were strongly inhibited upon TSA treatment (Fig. 4a). Arf1 has been reported to be critical for breast cancer cell invasion and metastasis [15, 25, 26]. To determine whether this is true in HNSCC cells, we examined the expression pattern of Arf1 in two cell line pairs (HN4-HN12 and HN30-HN31). HN4 and HN30 were derived from primary lesions in the base of tongue and pharynx, respectively, and HN12 and HN31 were derived from lymph-node metastatic lesions belonging to the same patients [27]. No notable difference in Arf1 protein levels was seen among these cells (Fig. 4b). The levels of active GTP-bound Arf1 were much stronger in metastatic HN12 and HN31 cells compared with their paired non-metastatic HN4 and HN30 cells (Fig. 4b), supporting the critical role of Arf1 activation in HNSCC metastasis. We then determined the importance of Arf1 activation in HNSCC cells. Arf1CA and Arf1DN were overexpressed in HN12 and HN31 cells, leading to increased or decreased Arf1 activation, respectively (Fig. 4c). These alterations in Arf1 activation levels did not affect HNSCC cell proliferation (Additional file 1: Figure S1). However, inactivation of Arf1 by transfection with Arf1DN significantly suppressed invasion of HNSCC cells and activation of Arf1 by transfection with Arf1CA counteracted this effect (Fig. 4d). These observations indicate that GTP-bound Arf1 is critical for HNSCC cell invasion. As TSA can induce repression of HNSCC cell invasion, we determined the consequences of Arf1CA transfection in the presence of TSA. Consistently, TSA at 5 μM decreased proliferation rate and invasive potential in HN12 and HN31 cells (Fig. 4e and f). No significant changes in cell proliferation were observed from TSA-treated cells transfected with Arf1CA or not (Fig. 4e). However, increased Arf1 activation attenuated TSA-induced inhibition of invasion in HN12 cells (Fig. 4f). The similar tendency was observed in HN31 cells when transfected with Arf1CA in the presence or absence of TSA (Fig. 4e and f). These observations indicate that TSA exhibits a suppressive effect on HNSCC cell invasion, at least in part, through inactivating Arf1.

EGFR is the most important RTK overexpressed in up to 90% of HNSCC compared with levels in normal mucosa, where expression levels correlate with decreased survival, independent of therapy [28‐31]. To determine the functional interaction between EGFR and Arf1 in HNSCC cells, we assessed their protein interaction. IP results from HN12 cells showed that Arf1 was one of EGFR interactors (Fig. 5a). To determine whether EGFR regulates Arf1 function in HNSCC cells, the small-molecule EGFR inhibitor erlotinib was used to treat HN12 cells. Twenty-four hours after treatment, erlotinib dose-dependently induced reduction in Arf1 activation without affecting total protein amount of Arf1 (Fig. 5b), and this reduction was associated with decreased EGFR phosphorylation levels (Fig. 5b). The total EGFR levels were only decreased at high dose of 10 μM erlotinib (Fig. 5b). IP was then used to assess the binding ability of Arf1 to EGFR in the presence of 5 μM erlotinib, which revealed decreased engagement of Arf1 in the EGFR immunocomplex compared with non-treatment (Fig. 5c). These findings suggest that phosphorylated EGFR binds to and activates Arf1 in HNSCC cells.

To better define the specific role of EGF signaling in Arf1 activation, serum-starved HNSCC cells were treated with 50 ng/ml human recombinant EGF for different times. EGF-induced Arf1 activation was accompanied by an associated increase in EGFR phosphorylation in HN12 cells within 5 min (Fig. 6a). The same was true for HN31 cells upon EGF stimulation (Fig. 6a and b). However, Arf1 activation triggered by EGF was significantly blocked when EGFR phosphorylation was impaired by 5 μM TSA at 30 min post treatment (Fig. 6a and b), indicating that TSA-induced inactivation of Arf1 is through EGFR blockade. To determine the contribution of Arf1 to EGF signaling, HNSCC cells were treated with EGF in the presence or absence of TSA or the Arf1 inhibitor Exo2. Interestingly, Exo2 markedly attenuated EGF-induced AKT activation, but it played no role in regulating ERK1/2 signaling upregulated by EGF (Fig. 6c). Moreover, either TSA or Exo2 significantly suppressed EGF-induced cell invasion (Fig. 6d). These data indicate that Arf1 activation plays an essential role in HNSCC cell invasion driven by EGFR.