Alpha-tubulin acetyltransferase/MEC-17 regulates cancer cell migration and invasion through epithelial–mesenchymal transition suppression and cell polarity disruption

MEC-17, a newly identified alpha-tubulin-N-acetyltransferase 1, serves as the major α-tubulin acetyltransferase to promote α-tubulin acetylation in vitro and in vivo. Alteration of α-tubulin acetylation may be involved in morphology regulation, cell migration, and tumour metastasis. However, MEC-17’s role in cell physiology and its effect on epithelial–mesenchymal transition (EMT) and cell polarity remain elusive. In the present study, we characterized the overexpressed or downregulated cell models through gene targeting as MEC-17 gain- or loss-of-function. Overexpression of MEC-17 enhanced the cell spreading area, suppressed pseudopods formation in a three-dimensional (3D) culture system, and inhibited cancer cell migratory and invasive ability and tumour metastasis by orthotopic lung cancer animal model. Furthermore, morphological change and migration inhibition of cancer cells were accompanied by EMT repression, Golgi reorientation, and polarity disruption caused by alteration of cdc42 activity via a decrease in Rho-GAP, ARHGAP21. By contrast, a reduction in endogenous MEC-17 accelerated the pseudopods formation and EMT, and facilitated cell migration and invasion. These results demonstrated the crucial role of MEC-17 in the modulation of intrinsic cell morphogenesis, migration, and invasive function through regulation of EMT and cell polarity.


Results
Overexpression of MEC-17 enhances α-tubulin acetylation and cell spreading area but inhibits pseudopod formation in a 3D culture. To understand the detailed function of MEC-17 in cancer cells, we determined its role in the A549 cell line through gene overexpression or knockdown methods. Western blot analyses of lysates of A549 cells stably expressing scrambled control, vector control, or MEC-17-overexpression constructs were performed. As shown in Fig. 1A, significantly increased protein expression in the MEC-17overexpressed group was noted compared with that in the parental, scrambled control, and vector control groups. In addition, the level of tubulin acetylation was higher in MEC-17-overexpressed cells in both Western blotting (Fig. 1A) and immunofluorescent examinations (Fig. 1B). Because an increase in α-tubulin acetylation can increase the cell adhesion area 27 , we examined whether MEC-17 overexpression affects cell morphology. Our results revealed that the spreading area as measured using the GFP signal profile in the MEC-17-overexpressed A549 cells was larger than that in the scrambled control and vector control groups (Fig. 1B). Besides, 3D matrices cultivation is a superior system for representing mesenchymal mode characterized by an elongated shape with protruding pseudopods 28 . Thus, we further used the 3D culture system by embedding cells into type I collagen to investigate whether the cell morphology had been altered by MEC-17 overexpression. As shown in Fig. 1C, the embedded cells in collagen were grown with pseudopod structures and the number of primary pseudopods and the extension area of pseudopods (b/a ratio, square area surrounding the cell body with (b) or without (a) pseudopods) were smaller in the MEC-17-overexpressed group than in the vector control group. These data indicated that MEC-17 played a critical role in the regulation of cell morphology.
Overexpression of MEC-17 attenuates cell migration, invasion, and tumour metastasis. Because MEC-17 regulated cell morphology, we investigated whether the increase in the cell spreading area or decrease in the pseudopod number influenced cell migration. As shown in Fig. 2A, the wound-healing assay showed that cell motility was lower in MEC-17-overexpressed cells compared with parental cells and the scrambled control or vector control groups. To examine whether the MEC-17-induced lower cell motility was due to inhibition of cell proliferation, we evaluated the natural cell growth assay of A549 parental, vector control and MEC-17-overexpressed cells to determine the doubling time (DT). The calculated DT of A549 parental, vector control and MEC-17-overexpressed cell was 24.15 ± 0.57, 23.15 ± 0 and 25.8 ± 1.63 hrs, respectively. Therefore, overexpression of MEC-17 was not affected cell proliferation. Furthermore, MEC-17-attenuated cell migration was confirmed using the Transwell migration assay. Fewer cells penetrated the membrane filter in MEC-17-overexpressed A549 cells than in the parental and vector control groups (Fig. 2B). Subsequently, we examined the invasive ability of MEC-17-overexpressed A549 cells by using the Matrigel-coated Transwell assay. As shown in Fig. 2C, the invasive ability of MEC-17-overexpressed cells was lower than that of the parental or vector control cells. Similar results were obtained in stable MEC-17-overexpressed MDA-MB-231 cells compared with parental and vector control cells ( Supplementary Fig. S1). We further determined whether the MEC-17-induced suppression of cell movement and pseudopod formation could suppress tumour metastasis in vivo. An orthotopic animal model generated by inoculating luciferase-expressing A549 cells into the right lung was evaluated by the IVIS system at 4 weeks. Our results revealed that the intensity of signaling in organs, including the liver, stomach, duodenum, kidney, pancreas, and spleen, was lower in the MEC-17-overexpressed group than in the vector control group (Fig. 2D and E). These results collectively demonstrate the inhibitory effect of MEC-17 on cell migration and invasion.  Fig. 3B, the MEC-17-knockdown A549 cells exhibited more protrusions in the 2D culture system. In the 3D culture system, the embedded cells grew multiple primary pseudopod protrusions and had a larger extension area in the MEC-17-knockdown group than in the pLVTHM vector control group (Fig. 3C). We then examined the dynamic of pseudopod protrusions by using a time-lapse microscope to monitor A549 cell growth. As shown in Fig. 3D, the pseudopod protrusions quickly extended or retracted and a greater number of protrusions, particularly elongated protrusions, were observed in MEC-17-knockdown cells than in the vector control during cell growth periods (0.5-21 h). The involvement of pseudopod protrusions in the mesenchymal movement hinted that MEC-17-knockdown cells might exhibit aggressive motility. To this end, we further evaluated the migrated and invasive ability of MEC-17-knockdown A549 cells through Transwell analysis. Our results showed that the migration and invasive ability of MEC-17-knockdown cells were higher than those of the pLVTHM vector and parental cells ( Fig. 3E and F).

MEC-17 induces EMT inhibition.
TGF-β was reported to induce EMT through HDAC6-dependent deacetylation of α-tubulin in lung adenocarcinoma A549 cells; thus, α-tubulin acetylation could serve as a marker for EMT 1 . We determined whether MEC-17 was involved in the regulation of EMT. As shown in Fig. 4A, the protein levels of vimentin and N-cadherin, markers of the mesenchymal phenotype, decreased dramatically in MEC-17-overexpressed A549 cells compared with the scrambled and vector control cells, whereas the

MEC-17 influences cell polarity through enhancement of cdc42 activity. Endogenous cdc42 was
reported to act as a negative regulatory function on intrinsic migration or invasion of some aggressive breast cancer cells 30 . We then examined the expression and activity of cdc42; the results revealed that the amount of active GTP-bound cdc42 (cdc42-GTP), an indicator of cdc42 activity, was significantly increased in MEC-17-overexpressed cells compared with scrambled control, vector control, and parental cells (Fig. 6A). By contrast, loss of MEC-17 reduced the level of cdc42-GTP in MEC-17-knockdown A549 cells compared with the parental and pLVTHM vector control groups (Fig. 6B). ARHGAP21, a Cdc42-specific GTPase-activating protein, was involved in cell migration, α-tubulin acetylation in EMT, and microtubule-and dynein-dependent Golgi motility and positioning 31,32 . Therefore, we hypothesized that cdc42 activity is regulated by ARHGAP21 after manipulation of MEC-17. We found that the level of ARHGAP21 was reduced in MEC-17-overexpressed cells compared with the scrambled control, vector control, and parental groups, whereas MEC-17 knockdown increased its expression compared with the parental and pLVTHM vector control groups ( Fig. 6C and D). These results demonstrated that MEC-17-induced inhibition of migration was associated with the dysregulation of Golgi orientation and cdc42-dependent actin dynamics disturbance.

Inhibition of cdc42 activity rescues EMT suppression, enhancement of cell spreading area, and cell migration attenuation in MEC-17-overexpressed cells. The TGF-β family signaling pathway
is essential for EMT induction and is accompanied by HDAC6-dependent deacetylation of α-tubulin, leading to decreased MT stability, and TGF-β stimulates EMT by disrupting basoapical polarity in polarized epithelial cells 1,23,[33][34][35] . Therefore, we investigated whether the MEC-17-induced EMT suppression could be rescued by treatment of TGF-β1. Against our expectations, the level of vimentin was not substantially recovered compared with the vector control group (Fig. 7A), suggesting that MEC-17 overexpression may occlude TGF-β1-induced EMT-associated gene expression and that cdc42 activity disturbance may be the major underlying mechanism downstream of MEC-17. CASIN, a selective cdc42 activity inhibitor, has been reported to reduce the elevated level of active cdc42 36 . We assessed whether CASIN application could restore the decreased expression of vimentin in MEC-17-overexpressed cells, and found that it could (Fig. 7B); this suggested that cdc42 activity is essential to MEC-17-induced EMT suppression. To this end, we further determined the importance of cdc42 activity in enhancing the MEC-17-induced spreading area, Golgi orientation, and migration ability. Our results demonstrated that MEC-17-induced enlargement of the cell spreading area and dispersal and redistribution of GM130 were blocked after pretreatment of CASIN compared with the MEC-17-overexpressed group ( Fig. 8A and B). Moreover, MEC-17-induced attenuation of cell migration was rescued after application of CASIN by the wound healing and Transwell assay ( Fig. 8C and D), suggesting that MEC-17 determined cell morphology, migration and invasion ability, and tumour metastasis through ARHGAP21-mediated cdc42 activation.
MEC-17 decreased cell migration compared with the vector control group. ***P < 0.001 in a one-way ANOVA.

Discussion
Most cancers are consequences of dysregulation in cell proliferation, morphogenesis, and migration. These cellular malfunctions have been associated with microtubule structural stability 22 . Acetylation of Lsy40 on α-tubulin was the second tubulin PTM to be discovered in early 1985 11 , and recent studies have shown that deacetylase (HDAC6/ SIRT2) and acetylase (MEC-17/ATAT1) are considered regulators for controlling the acetylation of α-tubulin [24][25][26] . HDAC6 is reportedly associated with tumorigenesis and cancer cell survival, and HDAC6 can deacetylate α-tubulin in TGF-β-induced EMT, thereby decreasing MT stability 1,23 . Therefore, downregulation of α-tubulin acetylation is involved in cancer cell growth and tumour progression. To this end, promotion of acetylation of α-tubulin by acetylases may affect the physiological regulation of morphological cell growth and motility. MEC-17, which has been identified as an α-tubulin acetyltransferase, is associated with instability of microtubule-rich structures such as cilia and axons, dentate gyrus distortion, and cortical neuron migration [37][38][39] . Boggs et al. had previously mentioned that elevated levels of α-tubulin acetylation in suspended tumour cells culture condition are a sufficient cause of metastatic potential 40 . Nevertheless, physiological research into MEC-17 and the molecular mechanism underlying the morphological and migration regulation of MEC-17-induced microtubule acetylation remain elusive.
In this study, we verified that MEC-17 inhibits cancer cell motility by increasing the spreading area for adhesion ability and reducing pseudopodial protrusions. In addition, overexpression of MEC-17 inhibited tumour metastasis, suppressed EMT, and disturbed cell polarization through cdc42 activation resulting from decreased expression of Rho-GAP, ARHGAP21. Blockage of cdc42 activity could reverse all of these changes. Notably, loss of endogenous MEC-17 led to a marked increase in pseudopodial protrusions and promoted cell movement. These results highlight the physiological importance of MEC-17 in cancer cell growth and movement correlated with EMT, cell polarity, and small GTPase activity.
Dynamic actin-dependent pseudopodial protrusions are critical for mesenchymal (tumour) cell migration and invasion, and thus also critical for cancer metastasis [41][42][43] . Our results revealed that MEC-17 overexpression reduced incidence of pseudopodial protrusions and attenuated tumour metastasis in vivo, whereas MEC-17 knockdown enhanced pseudopod formation; this suggests that MEC-17 serves as a structural modulator for actin polymerization and depolymerization. Studies have demonstrated that the expression of regulators of actin reorganization is essential for pseudopod formation and associated with tumour cell migration and invasion, such as Wiskott-Aldrich syndrome protein (WASP) family proteins, the Arp2/3 complex, Eps 8, α-actinin, fascin, filamin, LIM kinase/cofilin, and cortactin 42,44,45 . In addition, tumour cell pseudopod-specific proteins (AHNAK, Septin-9, eIF4E, and S100A11) were identified from the transcriptome/proteome analysis and closely for pseudopod formation, actin cytoskeleton dynamics, and tumour cell migration and invasion 41 . One possible means through which this effect is achieved is the influence on pseudopod formation through directly acetylated pseudopod-specific proteins or an indirectly acetylated transcription factor or repressor to promote or hinder pseudopod-specific protein production. Although we did not identify which specific proteins are acetylated by MEC-17, we believe that MEC-17 overexpression can drive or repress the expression of genes such as GM130 (Fig. 5B) and ARHGAP21 (Fig. 6C). In our unpublished data, we also found that Erk1/2 phosphorylation was reduced in MEC-17-overexpressed cells, indicating that the alteration of EMT and polarization protein expression may be regulated by the inhibition of MAPK/ERK signaling pathway. Further studies are required to address this hypothesis.
The prerequisite for cell motility is the pseudopodial protrusions formation and retraction of cell membranes driven by reorganization of the cytoskeleton 46,47 . Many studies have investigated the molecular mechanisms involved in cell migration and have demonstrated that Rho small GTPases are key regulators of actin polymerization and dynamic change. The small GTPase proteins, Rac, Rho, and Cdc42, also played roles in the initial signals, leading to the polarization of migration 29,47 and microtubule-dependent Golgi positioning 32 . Cell migration direction requires polarized alignment of the cytoskeleton; furthermore, the microtubule-organizing center (MTOC) and Golgi are fundamentally positioned facing forward at the leading edge of the cell 32 . The Golgi apparatus has been linked to polarity; studies have illustrated that GM130 is a regulator of cdc42 signaling that regulates cell polarity and provides a potential link between Golgi and cancer initiation 48 . In our study, immunofluorescence revealed GM130 distribution to be disturbed and its level to be increased in MEC-17-overexpressed cells (Fig. 5A), suggesting that high expression of GM130 protein may not limit the specific Golgi position for retaining the apical-basal polarity, eventually resulting in orientation disruption and weak migration ability. In addition, a Golgi-localized GAP for cdc42 regulated the Arp2/3 complex and F-actin dynamics by controlling cdc42 activity 49 , and endogenous cdc42 was able to inhibit intrinsic migration or invasion and alter PKCd, Erk1/2, and PKA phosphorylation in some aggressive breast cancer cells 30 . Our finding that MEC-17 can activate cdc42 followed by a reduction in ARHGAP21 to regulate cell morphology and movement is consistent with previous number and extension area compared with the pLVTHM group. (n = 14 for pLVTHM, n = 11 for sh-1830). ***P < 0.001, *P < 0.05 in a one-way ANOVA. studies. However, to investigate whether cdc42 activity is affected by α-tubulin acetylation, we used tubastatin-A, a potent selective inhibitor for HDAC6, to increase the acetylation of α-tubulin, thereby bypassing the involvement of MEC-17. Cdc42 activity was not altered (data not shown), suggesting that α-tubulin acetylation is not necessary for MEC-17-induced cdc42 activation.
In conclusion, our findings demonstrated that MEC-17-induced cdc42 activity through the reduction of GAP regulates EMT process, actin reorganization for morphological regulation, cell polarity maintenance, and cell motility.

Lentiviral particle production.
Engineered self-inactivating recombinant lentiviral particles were used to overexpress or silence the MEC-17 gene in A549 cells. All viruses were produced by cotransfection of lentiviral DNA with two helper plasmids, vesicular stomatitis virus envelope glycoprotein (VSV-G) and Δ8.9, in HEK293T cells. Media containing recombinant lentiviruses were collected twice after transfection for 48 and 96 h and were concentrated by Lenti-X according to the manufacturer's instruction to obtain a concentrated stock for stable cell production. After centrifugation, pellets were resuspended in A549 cultured media with titers of 10 8 -10 9 units/ mL. To evaluate the overexpression or knockdown of MEC-17 efficiency, the stable cell lysates were harvested for Western blotting analysis.
Wound healing assay. The monolayer confluent cells were starved for 12-16 h to suppress cell proliferation, and then scraped with a 1-mL pipette tip across a 6-well plate. After wounding, the culture media containing 10% FBS were replaced. Cells were visualized using the inverted microscope after 24 and 48 h of the migration period. The migrated cells were manually quantified by measuring the cell-covered area with image J software. For quantification of polarized cells in the scratch wounding assay, 1 × 10 5 cells were cultured on 4-well chamber slides. A scratch area for cell migratory direction was then produced by using a 100-μL pipette tip across the confluent monolayer and incubated in a fresh culture medium containing 10% FBS after overnight starvation. After 24 h, the cells were immunostained for the Golgi apparatus using anti-Golgi Matrix protein, GM130. The cell was determined as polarized when stained Golgi apparatus was compact and located primarily in the cytosolic sector facing the wound or direction of migration. Culturing and monitoring cells embedded in collagen matrices. The three-dimensional (3D) culture was performed as previously described 28 . In brief, PureCor bovine collagen solution (Advance Biomatrix) was applied for collagen matrices. To embed cells in collagen, the MEC-17-overexpressed or knockdown and vector control A549 cells were first cultured on plastic dishes. Then, trypsinizing adherent cells with 0.1% trypsin in EDTA at 37 °C for 5 min and collected in the proportion of 0.5 × 10 6 cells/mL. Subsequently, 1.7 mL of 3 mg/ mL chilled PureCor bovine collagen solution was mixed with 1.3 mL of F-12K medium for A549 cells to form a 1.7 mg/mL collagen solution (3 mL). Then, we centrifuged the cell suspensions and replaced the supernatants with the chilled 1.7 mg/mL mixing collagen solution, and thoroughly mixed the cells with the collagen solution.

Transwell migration and invasion assay.
The cell-collagen mixture was polymerized in a cell incubator at 37 °C for 1 h and then covered with a sufficient amount of the serum-containing medium. To observe the pseudopods, the live cells were cultured in 12-well plates and the GFP-positive cells were examined through immunofluorescent microscopy after 24 h or through the cell culture monitoring system (CCM-1.4/XYZ) from 0 to 24 h.
Cdc42 activation assays. Cdc42 activity was determined using a Rac1/cdc42 activation assay kit as previously described 30,50 . Briefly, cdc42-GTP from various lysates were pulled down using the GST fusion-protein, corresponding to the p21-binding domain (PBD, residues 67-150) of human PAK-1 bonded to agarose beads (PAK1-PBD agarose conjugate) for 12 h at 4 °C. The beads were washed four times with ice cold 1× MLB lysis buffer, resuspended and boiled in SDS sample buffer, and separated on 4-12% SDS-PAGE gradient gels before being subjected to immunoblot with the anti-cdc42 antibody to detect the presence of cdc42-GTP and total cdc42.
Immunofluorescence staining. Immunofluorescent staining was modified and performed as previously described 50,51 . In brief, the cells grown on CultureSlides (BD Biosciences) were fixed with 4% paraformaldehyde in PBS for 20 min at room temperature. After washing twice with PBS, the cells were permeabilized and blocked simultaneously in a solution containing 3% bovine serum albumin (BSA) and 0.2% Triton X-100 in PBS for 1 h at room temperature. Subsequently, the indicated primary antibodies, namely anti-GFP (1:1000), anti-E-cadherin (1:1000), anti-Vimentin (1:500), and anti-GM130 (1:1000), were added and incubated overnight at 4 °C. After washing with PBS, bound primary antibodies were visualized through incubation of the cells with appropriate Alexa-Fluor-488-conjugated and Alexa-Fluor-568-conjugated secondary antibodies for 1-2 h at room temperature. 4′,6-Diamidino-2-phenylindole (DAPI) was used as a counterstain to visualize the nuclei. The cells were then rinsed extensively in PBS and mounted on ProLong Anti-fade media (Molecular Probes, Eugene, OR). Confocal fluorescent images were obtained using an FV1000 confocal microscope (Olympus) with a 60× or 63× oil immersion lens, namely NA 1.35 (Uplsapo).

SDS-PAGE and Western blotting.
Cells for analysis were scraped and lysed in a lysis buffer (#C2978; Sigma), RIPA buffer solution, or 1× MLB buffer containing the following protein phosphatase and protease inhibitors: 1 mM Na 3 VO 4 , 0.5 mM phenylmethylsulfonyl fluoride (PMSF), and 1× protease inhibitor cocktail. Samples were sonicated and centrifuged at 14000 rpm at 4 °C for 15 min to remove debris. The supernatant was then assayed for total protein concentration by using a BCA protein assay kit (#23225; Pierce). An equal amount of proteins was resuspended in a 1× SDS sample buffer (2% SDS, 12.5 mM EDTA, 1% β-mercaptoethanol, 20% v/v glycerol, 0.02% bromophenol blue, and 50 mM Tris-Cl, pH 6.8) and boiled for 10 min. Each sample was subsequently separated using electrophoresis in 8% or 10% SDS-PAGE polyacrylamide gel or NuPAGE 4-12% Bis-Tris gradient protein gel. Following transfer onto PVDF or nitrocellulose membranes, the blots were blocked in TBS (150 mM NaCl and 10 mM Tris-Cl, pH 7.4) containing 3% BSA and 0.1% Tween 20 for 1 h. Subsequently, the membranes were blotted for 12-16 h at 4 °C with specific primary antibodies. Each blot was probed with horseradish peroxidase-conjugated secondary antibodies for 1 h, and the chemiluminescent signal was conducted using the Western Lightning Plus ECL immunoblotting detection system (Perkin Elmer). Immunoblots were analyzed by densitometry performed using Image J software.
Metastatic orthotopic mouse models of lung cancer. A549-Luc C8 Cells (2 × 10 6 cells in 10-μL PBS) were surgically inoculated into the right lungs of 5-6-week-old male NOD-SKID immunodeficient mice (BioLASCO Taiwan or National Laboratory Animal Center, Taipei, Taiwan) via intrathoracic injection. Mouse body weight was measured weekly and primary tumour outgrowth and metastasis were inspected using IVIS every 2 weeks. Lung and other tissue including heart, liver, stomach, duodenum, kidney, spleen, and pancreas were extracted to determine the total tumour metastatic ability at 4 weeks after inoculation. All experimental procedures were performed according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of National Health Research Institutes.
Statistical analysis. All analyzed data for each experiment were normalized relative to controls and presented as mean ± standard error of the mean (s.e.m). Differences between means was evaluated through an analysis of variance (ANOVA) with Bonferroni's post hoc analysis test for multiple comparisons or an unpaired two-tailed Student's t test. In this study, probability values (P) < 0.05 was considered statistically significant. P values are denoted with asterisks as follows: *P < 0.05, **P < 0.01, and ***P < 0.001.