1
CAMSAP2-mediated noncentrosomal microtubule acetylation drives 1
hepatocellular carcinoma metastasis 2
3
Dongxiao Li1,2,#
, Xiangming Ding1,2,#
, Meng Xie1,2
, Zheng Huang1,2
, Ping Han1, Dean 4
Tian1,2
, and Limin Xia1,2,*
5
6
1Department of Gastroenterology, Tongji Hospital of Tongji Medical College, 7
Huazhong University of Science and Technology, Wuhan 430030, Hubei Province, 8
China 9
2Institute of Liver and Gastrointestinal Diseases, Tongji Hospital of Tongji Medical 10
College, Huazhong University of Science and Technology, Wuhan 430030, Hubei 11
Province, China 12
#These authors contributed equally to this work. 13
14
*Corresponding author 15
Limin Xia, Department of Gastroenterology, Institute of Liver and Gastrointestinal 16
Diseases, Tongji Hospital of Tongji Medical College, Huazhong University of Science 17
and Technology, Wuhan, Hubei, 430030, China. Phone: 8627-83662688; Fax: 18
8627-83662640; Email: [email protected] 19
20
21
22
2
Abstract 23
Rationale: Emerging evidence suggests that noncentrosomal microtubules play an 24
essential role in intracellular transport, cell polarity and cell motility. Whether these 25
noncentrosomal microtubules exist or function in cancer cells remains unclear. 26
Methods: The expression and prognostic values of CAMSAP2 and its functional 27
targets were analyzed by immunohistochemistry in two independent HCC cohorts. 28
Immunofluorescence and co-immunoprecipitation were used for detection of 29
CAMSAP2-decorated noncentrosomal microtubule. Chromatin immunoprecipitation 30
and luciferase report assays were used to determine the c-Jun binding sites in HDAC6 31
promoter region. In vitro migration and invasion assays and in vivo orthotopic 32
metastatic models were utilized to investigate invasion and metastasis. 33
Results: We reported a microtubule minus-end-targeting protein, CAMSAP2, is 34
significantly upregulated in hepatocellular carcinoma (HCC) and correlated with poor 35
prognosis. CAMSAP2 was specifically deposited on microtubule minus ends to serve 36
as a ―seed‖ for noncentrosomal microtubule outgrowth in HCC cells. Upon depletion 37
of CAMSAP2, the noncentrosomal microtubule array was transformed into a 38
completely radial centrosomal pattern, thereby impairing HCC cell migration and 39
invasion. We further demonstrated that CAMSAP2 cooperates with EB1 to regulate 40
microtubule dynamics and invasive cell migration via Trio/Rac1 signaling. Strikingly, 41
both immunofluorescence staining and western blotting showed that CAMSAP2 42
depletion strongly reduced the abundance of acetylated microtubules in HCC cells. 43
Our results revealed that HDAC6, a promising target for cancer therapy, was inversely 44
3
downregulated in HCC and uniquely endowed with tumor-suppressive activity by 45
regulation CAMSAP2-mediated microtubule acetylation. Mechanistically, CAMSAP2 46
activates c-Jun to induce transrepression of HDAC6 through Trio-dependent 47
Rac1/JNK pathway. Furthermore, NSC23766, a Rac1-specific inhibitor significantly 48
inhibited CAMSAP2-mediated HCC invasion and metastasis. 49
Conclusions: CAMSAP2 is functionally, mechanistically, and clinically oncogenic in 50
HCC. Targeting CAMSAP2-mediated noncentrosomal microtubule acetylation may 51
provide new therapeutic strategies for HCC metastasis. 52
53
Keywords: Hepatocellular carcinoma; Noncentrosomal microtubule; CAMSAP2; 54
Acetylation; HDAC6 55
56
Graphical Abstract: 57
58
CAMSAP2, a microtubule minus-end-targeting protein, promotes HCC invasion and 59
metastasis by regulating noncentrosomal microtubule remodeling and acetylation. 60
NSC23766, a Rac1-specific inhibitor significantly inhibited CAMSAP2-mediated 61
HCC metastasis. 62
4
Introduction 63
Hepatocellular carcinoma (HCC) is the sixth most common neoplasm and the third 64
leading cause of cancer-related death worldwide [1]. The poor prognosis of patients 65
with HCC mainly results from distant metastasis and tumor recurrence after curative 66
resection, which occur at high frequencies [2]. Diverse microtubule-targeting agents 67
have been part of the pharmacopoeia of anticancer therapy for decades; however, they 68
have little effect on HCC metastasis [3]. 69
70
It is generally accepted that the centrosome is the dominant microtubule-organizing 71
center. Microtubules in interphase cells are typically arrayed in a starburst pattern, 72
with minus ends anchored at the centrosome and plus ends extending radially toward 73
the plasma membrane [4]. However, in many cell types—epithelial and neuronal 74
cells—a unique subset of microtubules is untethered from the centrosome and instead 75
distributed throughout the cytoplasm with minus ends free; this subset is required for 76
intracellular transport, cell polarity, and cell motility [5,6]. Whether these 77
noncentrosomal microtubules exist or function in cancer cells remains unclear. 78
Therefore, exploring microtubule nucleation and distribution patterns in HCC cells 79
may provide new insights into the suppression of HCC metastasis. 80
81
Microtubules are polarized cytoskeletal filaments with two structurally and 82
functionally distinct ends. While the function and dysfunction of microtubule 83
plus-end dynamics and plus-end-tracking proteins (+TIPs)—EB1, CLIP170, and 84
5
CLASP2—are quite well understood [7], studies on microtubule minus-end-targeting 85
proteins—calmodulin-regulated spectrin-associated proteins (CAMSAPs) in 86
mammals and Patronin in invertebrates [5,6]—have only just begun to emerge. 87
CAMSAP2, a member of the CAMSAP/Nezha/Patronin family, is specifically 88
deposited on microtubule minus ends and serves as a ―seed‖ for noncentrosomal 89
microtubule outgrowth. Depletion of CAMSAP2 results in the transformation of the 90
partially noncentrosomal microtubule network into a completely radial centrosomal 91
array, impeding cell polarization, intracellular transport, and organelle assembly [8,9]. 92
Notably, CAMSAP2 depletion impairs directional cell migration [10,11], in line with 93
observations that microtubules are released from the centrosome in migrating cells 94
[12]. These findings raise the question of whether migration and invasion in cancer 95
cells are promoted through maintenance of noncentrosomal microtubules and 96
suppression of the microtubule-organizing ability of the centrosome. 97
98
Tubulin undergoes several highly conserved posttranslational modifications (PTMs) 99
including detyrosination and further cleavage to Δ2-tubulin, acetylation, 100
polyglutamylation, and polyglycylation [13,14]. Tubulin PTMs provide a mechanism 101
for adapting to specific cellular functions by regulating microtubule properties and, 102
therefore, extended the ―tubulin code‖ [15]. Acetylated microtubules accumulate at 103
the leading edge, thus promoting directional cell locomotion and chemotaxis [16,17]. 104
Therefore, we questioned whether CAMSAP2-mediated noncentrosomal 105
microtubules are subjected to acetylation for promoting HCC cell migration and 106
6
invasion. 107
Here, we report a microtubule minus-end-targeting protein, CAMSAP2, promotes 108
HCC invasion and metastasis by regulating noncentrosomal microtubule remodeling 109
and acetylation. Investigating the function and mechanism of CAMSAP2-decorated 110
noncentrosomal microtubule in HCC may provide new clues to suppress HCC 111
metastasis. 112
113
Materials and Methods 114
Cell lines and culture 115
Huh7 and MHCC97H cell lines were purchased from the Stem Cell Bank, Chinese 116
Academy of Sciences. HepG2, PLC/PRF/5, SW1990 and SW620 cell lines were 117
obtained from the American Type Culture Collection. All the cell lines were 118
authenticated by short tandem repeat analysis and routinely checked for Mycoplasma 119
contamination using the MycoAlert Mycoplasma detection kit. Cells were cultured in 120
Dulbecco’s Modified Eagle’s Medium (HyClone, UT, USA) containing 10% fetal 121
bovine serum (Gibco, CA, USA), maintained at 37°C in a 5% CO2 incubator. 122
123
RNA interference 124
Cells were transfected with small interfering (si)RNAs (Ribo Bio, Guangzhou, China; 125
siCAMSAP2#1: 5’-GAAACAGTTTAGCCACATA-3’ and siCAMSAP2#2: 126
5’-GAACAACAGTCATGTATCT-3’) using Lipofectamine 3000 (Invitrogen, CA, 127
USA) per the manufacturer’s instructions. The interference efficacy was verified by 128
7
western blotting. 129
130
Immunofluorescence (IF) and imaging 131
Cells were fixed with 4% paraformaldehyde at room temperature for 15 min and, then, 132
permeabilized with phosphate-buffered saline containing 0.2% Triton X-100 for 10 133
min. For tissue IF, human HCC and corresponding adjacent noncancerous tissues 134
were fixed with 4% paraformaldehyde, paraffin-embedded, and cut into 4-µm-thick 135
sections. After routine dewaxing, rehydration, and antigen retrieval, the cells were 136
permeabilized, blocked with 5% goat serum, and incubated with primary antibodies at 137
4°C overnight. The cells or tissues were washed with PBS and, then, incubated with 138
the appropriate secondary antibodies. Antibodies are listed in Table S9. 139
140
Fluorescence was detected using an Olympus fluorescence microscope equipped with 141
oil-immersion lenses with 100×1.40, 40×0.9, 20×0.75, 10×0.40, or 4×0.16 numerical 142
aperture, and an Olympus laser-scanning confocal microscope equipped with a Plan 143
Apo 60×1.40 numerical aperture oil-immersion lens. Images were processed using 144
Photoshop CS5 (Adobe Systems) and Imaris (Bitplane) software. IF signal intensity 145
was quantified as described previously [11], with slight modifications. IF signal 146
intensity distribution was measured using the ImageJ Radial Profile plugin: a circle 147
with the indicated radius was drawn at the center of gamma-tubulin, the Golgi 148
complex, or the nucleus, and the signal intensity along the radius was measured 149
Fluorescence intensities were normalized to the maximum intensity of each cell. 150
8
151
Other protocols used in this study are described in the Supplementary Materials. 152
153
Results 154
CAMSAP2 is significantly upregulated in HCC tissues and indicates a poor 155
prognosis 156
We first investigated the expression of CAMSAP family members in different 157
publicly available liver cancer datasets. The Cancer Genome Atlas dataset (TCGA) 158
revealed that mRNA levels of the CAMSAP family were significantly increased in 159
liver cancer specimens when compared to the levels in normal liver tissues (Figure 160
S1A). Immunohistochemistry (IHC) tissue microarray data from Human Protein Atlas 161
program database revealed high or medium CAMSAP2 staining intensity in 10 out of 162
12 liver cancer samples, whereas only 3 out of 12 cases showed medium staining of 163
CAMSAP1 and CAMSAP3 (Figure S1B). Kaplan–Meier analysis based on TCGA 164
data revealed that liver cancer patients with high CAMSAP2 mRNA levels had a 165
significantly shorter overall survival (OS) and disease-free survival (DFS) than those 166
who with low CAMSAP2 mRNA levels (Figure S1C). There was no obvious 167
correlation between poor patient outcome and high expression of CAMSAP1 or 168
CAMSAP3 (Figure S1C). Moreover, the increased mRNA expression of CAMSAP2 169
also observed in pancreatic adenocarcinoma (PAAD), stomach adenocarcinoma 170
(STAD) and colon adenocarcinoma (COAD) tissues based on TCGA data (Figure 171
S1D). Kaplan-Meier analysis based on TCGA dataset revealed that PAAD, STAD and 172
9
COAD patients with high levels of CAMSAP2 mRNA had a shorter OS and DFS than 173
those with low mRNA expression of CAMSAP2 (Figure S1E). Collectively, these 174
findings suggested that CAMSAP2 may serve as a candidate biomarker for HCC 175
prognosis. 176
177
We quantified CAMSAP2 expression in 90 pairs of HCC and adjacent nontumorous 178
tissue samples and 20 normal liver tissues using quantitative reverse-transcription 179
polymerase chain reaction (RT-q)PCR. HCC tissues displayed marked upregulation of 180
CAMSAP2 mRNA, compared with adjacent nontumorous and normal liver tissues 181
(Figure 1A). CAMSAP2 mRNA expression was higher in HCC tissues from patients 182
with recurrence than in those from patients without recurrence (Figure 1A). 183
CAMSAP2 mRNA expression was significantly upregulated in metastatic HCC 184
tissues, compared to matched primary HCC tissues (Figure 1A). IHC, 185
immunofluorescence (IF), and western blot analyses revealed that CAMSAP2 protein 186
expression was upregulated in HCC compared with adjacent noncancerous tissues 187
(Figure 1B-E). Intriguingly, IF staining revealed symmetrical staining of CAMSAP2 188
and α-tubulin throughout the cytoplasm in nontumor tissues, whereas highly 189
asymmetrical signals were detected in HCC tissues (Figure 1C). 190
191
To investigate the potential role of CAMSAP2 in determining clinical outcomes for 192
HCC, we analyzed its expression in two independent cohorts of human HCC (cohort I, 193
n = 360; cohort II, n = 178) by IHC. CAMSAP2 was dramatically upregulated in 194
10
HCC compared with adjacent nontumor tissues (Figure 1F) (Table S1). Positive and 195
negative controls are shown in Figure S3B. In addition, IHC staining confirmed that 196
CAMSAP2 was also upregulated in PAAD, STAD and COAD cancer specimens 197
compared with adjacent nontumor tissues (Figure S1F). Analysis of the 198
clinicopathological characteristics revealed that CAMSAP2 overexpression was 199
strongly associated with multiple tumors, increased tumor size, microvascular 200
invasion, poor tumor differentiation, and a higher tumor-nodule-metastasis stage 201
(Table S1). HCC patients with high CAMSAP2 expression had a higher chance of 202
recurrence and a shorter OS than those with low CAMSAP2 expression (Figure 1G). 203
Multivariate analysis identified CAMSAP2 expression as an independent predictor of 204
postoperative recurrence and OS (Table S2 and S3). Together, these data indicated 205
that CAMSAP2 is a potential prognostic biomarker for malignant progression and 206
metastasis of HCC. 207
208
CAMSAP2 promotes HCC invasion and metastasis in vitro and in vivo 209
For loss- and gain-of-function analyses, we knocked down and overexpressed 210
CAMSAP2 in MHCC97H and Huh7 cells, denoted as 211
―MHCC97H/Huh7-siCAMSAP2‖ and ―MHCC97H/Huh7-Lv-CAMSAP2,‖ and we 212
evaluated CAMSAP2 expression by western blotting (Figure 2A). CAMSAP2 213
depletion markedly reduced HCC cell migration and invasion in MHCC97H and 214
Huh7 cells, whereas CAMSAP2 upregulation had the opposite effect (Figure 2B–C). 215
Further, the decreased migration and invasion capabilities also confirmed in 216
11
CAMSAP2-knockout MHCC97H cell line (Figure S2A-C). 217
218
To mimic the in-vivo invasion phenotype, we used three-dimensional cultures to 219
visualize tumor sphere invasion. Because of its transient effects, small interfering 220
(si)RNA-mediated knockdown cannot be used in three-dimensional culture. Instead, 221
we used a lentiviral vector to establish two CAMSAP2-knockdown cell lines, denoted 222
as ―MHCC97H-shCAMSAP2‖ and ―Huh7-shCAMSAP2.‖ IF staining indicated that 223
MHCC97H cells with high metastatic potential showed an obvious invasion 224
phenotype, whereas MHCC97H-shCAMSAP2 cells formed a spheroidal tumor nest 225
with well-defined borders (Figure 2D). Opposite results were found in 226
CAMSAP2-overexpressing Huh7 compared with control cells (Figure 2D). These 227
results suggested that CAMSAP2 promotes HCC cell invasion in an in-vivo-like 228
environment. 229
230
To assess the effect of CAMSAP2 on HCC metastasis in vivo, 231
CAMSAP2-overexpressing, -knockdown, and control HCC cells were injected into 232
the livers of nude mice to establish an orthotopic metastatic model. Bioluminescence 233
imaging of the HCC cells prior to orthotopic implantation is shown in Figure S3A. 234
Consistent with the in-vitro results, CAMSAP2 upregulation significantly increased 235
intrahepatic and lung metastases, resulting in a shorter OS time of the 236
Huh7-Lv-CAMSAP2 compared with the control group (Figure 2E–F). Hematoxylin 237
and eosin (H&E) staining confirmed that CAMSAP2 overexpression significantly 238
12
increased the incidence of lung metastasis and the number of metastatic pulmonary 239
nodules (Figure 2G-H). In contrast, CAMSAP2 knockdown reduced the incidence of 240
lung metastasis and the number of metastatic pulmonary nodules, resulting in a longer 241
OS time (Figure 2G-H). Collectively, these findings suggested that CAMSAP2 242
significantly promotes HCC metastasis. 243
244
CAMSAP2-mediated noncentrosomal microtubule remodeling contributes to 245
directional cell migration in HCC cells 246
CAMSAP2-decorated noncentrosomal microtubules have been observed in various 247
cell types, including fibroblasts, epithelial cells, and neurons [5,6]. Therefore, we 248
questioned whether CAMSAP-decorated noncentrosomal microtubule stretches exist 249
and function in HCC cells. Double IF staining of CAMSAP2 and α-tubulin in 250
MHCC97H and Huh7 cells yielded stretched or punctate CAMSAP2 signals scattered 251
throughout the cytoplasm (Figure 3A). Closer observation revealed that each 252
CAMSAP2 cluster capped a microtubule minus end (Figure 3A). Similar observations 253
were made in HepG2 and PLC/PRF/5 HCC cell lines (Figure S4A). During 254
wound-induced migration, CAMSAP2 reorganized towards the direction of migration 255
along the tubulin-based protrusion (Figure 3B). Endogenous co-immunoprecipitation 256
assays revealed that CAMSAP2 could be immunoprecipitated with α-tubulin antibody 257
in MHCC97H cells. Similarly, α-tubulin was detected in CAMSAP2 258
immunoprecipitates (Figure 3C). Additionally, EB1, a +TIP, was also detected in the 259
immunoprecipitates (Figure 3C). These results indicated that CAMSAP2 caps 260
13
microtubule minus ends and forms CAMSAP-decorated noncentrosomal microtubule 261
stretches in HCC cells. 262
263
CAMSAP2 depletion alters the microtubule assembly pattern and, thus, the 264
distribution of microtubule networks. Therefore, we explored the effect of CAMSAP2 265
depletion on microtubule assembly in HCC cells. Immunostaining revealed that 266
CAMSAP2 depletion transformed the noncentrosomal microtubule network into a 267
symmetrical radial pattern wherein the majority of microtubules nucleated from the 268
centrosome in MHCC97H cells (Figure 3D). Similar observations were made in Huh7 269
cells (Figure S4B). Next, we treated subconfluent HCC cells with nocodazole for 30 270
min to completely depolymerize microtubules and, then, observed microtubule 271
repolymerization after nocodazole washout. IF staining revealed that the initial 272
repolymerization microtubules radiated from the centrosome 15 min after drug 273
washout in CAMSAP2-depleted cells, whereas noncentrosomal microtubules were 274
not observed (Figure 3E). These observations implied that CAMSAP2 works to 275
maintain noncentrosomal microtubules, while suppressing the microtubule-nucleating 276
ability of the centrosome in HCC cells. 277
278
Next, we investigated the consequences of the altered microtubule pattern on 279
directional cell migration. Maintenance of the Golgi ribbon is crucial for polarized 280
cell migration [10,11]. Intriguingly, while protein expression of the Golgi marker 281
GM130 was not altered (Figure S4C), the Golgi apparatus lost its typical ribbon-like 282
14
morphology in CAMSAP2-depleted MHCC97H cells. The microtubule system 283
became more radial and failed to colocalize with the Golgi stacks (Figure 3F-G). 284
During wound-induced migration, CAMSAP2-depleted HCC cells failed to polarize 285
properly because of defective Golgi reorientation towards the wound edge (Figure 286
3H-I). Collectively, these results suggested that CAMSAP2 is essential for 287
noncentrosomal microtubule organization, which is required for HCC cell polarization 288
and migration. 289
290
CAMSAP2 cooperates with EB1 to regulate microtubule dynamics and invasive 291
cell migration via Trio/Rac1 signaling 292
We further investigated the regulatory roles of CAMSAP2 in microtubule dynamics 293
by monitoring the behavior of EB1, which autonomously tracks the growing 294
microtubule end and recruits multiple +TIPs to couple the microtubule dynamics to 295
specific cellular events [7,18]. IF staining for EB1 in HCC cells produced stretched or 296
punctate signals scattered through the cytoplasm. Closer observation revealed that 297
EB1 cluster capped a microtubule end (Figure S4F). Microtubule fractionation assays 298
revealed that microtubule-associated EB1 was reduced after CAMSAP2 knockdown 299
(Figure 4A). The polymerized-to-total a-tubulin ratio was also decreased in 300
CAMSAP2-knockdown HCC cells (Figure 4A). These findings implied that loss of 301
CAMSAP2 led to a reduction in the amount of polymerized a-tubulin as well as 302
growing microtubule ends. 303
304
15
Microtubule plus ends function as concentration devices for signaling 305
molecules—Rho GTPase guanine nucleotide exchange factors (GEFs) and kinases [7]. 306
EB1 at growing microtubule ends promotes the formation of a complex with the GEF 307
triple functional domain protein (Trio), leading to Trio-dependent activation of Rac1 308
[19]. Co-immunoprecipitation assays indicated that CAMSAP2 and a-tubulin directly 309
interacted with EB1 and Trio in HCC cells (Figure 4B). Then, we examined Rac1 310
activity after CAMSAP2 depletion. Western blot analysis revealed that active Rac1 311
was markedly reduced in CAMSAP2-knockdown HCC cells (Figure 4C). EB1 and 312
Trio knockdown or treatment with the specific Rac1 inhibitor NSC23766 significantly 313
ameliorated the increased expression of GTP-Rac1 (Figure 4D-E) and markedly 314
suppressed the increased migration and invasion capabilities (Figure 4F) induced by 315
CAMSAP2 upregulation. 316
317
To determine the role of the EB1/Trio complex in CAMSAP2-mediated HCC 318
metastasis in vivo, we established an orthotopic metastatic model. Bioluminescence 319
imaging of the indicated HCC cells prior to orthotopic implantation is shown in 320
Figure S3A. Consistent with the in-vitro results, EB1 or Trio knockdown markedly 321
lowered the incidence of lung metastasis and prolonged the OS time in the 322
Huh7-Lv-CAMSAP2 group (Figure 4G-H). Histological analysis confirmed that 323
depletion of EB1 or Trio abrogated lung metastasis induced by CAMSAP2 324
overexpression (Figure 4I-J). Together, these results suggested that CAMSAP2 325
cooperates with the EB1/Trio complex to direct microtubule dynamics to invasive cell 326
16
migration by promoting Rac1 signaling. 327
328
Prognostic significance of the correlation between CAMSAP2 and EB1 or Trio 329
expression in HCC tissues 330
Clinical associations between CAMSAP2 and EB1 or Trio expression were evaluated 331
in two independent cohorts of HCC patients. IHC revealed that CAMSAP2 332
overexpression was positively correlated with EB1 and Trio expression (Figure 5A-B). 333
Positive and negative controls are shown in Figure S3B. Analysis of the 334
clinicopathological characteristics in paired HCC tissues showed that upregulation of 335
either EB1 or Trio was significantly correlated with malignant tumor progression and 336
poor prognosis (Figure 5C) (Tables S4 and S5). Further, coexpression of either 337
CAMSAP2/EB1 or CAMSAP2/Trio was associated with the highest recurrence rate 338
and lowest OS in both HCC cohorts (Figure 5D). 339
340
CAMSAP2 promotes microtubule acetylation to control invasive cell migration 341
via an inhibitory interaction with HDAC6 342
Tubulin PTMs provide a mechanism for microtubules to adapt to specific cellular 343
functions [20]. Acetylated microtubules accumulate at the leading edge, thus 344
promoting directional cell locomotion and chemotaxis [16,17]. Therefore, we 345
questioned whether CAMSAP2-mediated noncentrosomal microtubules are subjected 346
to acetylation for promoting HCC cell migration and invasion. Western blot analysis 347
revealed that microtubule acetylation was markedly reduced in 348
17
CAMSAP2-knockdown HCC cells (Figure 6A, Figure S4D). However, there were no 349
obvious changes in the expression of detyrosinated or tyrosinated α-tubulin after 350
downregulation of CAMSAP2 (Figure S5A). Conversely, CAMSAP2 overexpression 351
dramatically elevated α-tubulin acetylation in MHCC97H and Huh7 cells (Figure 352
S4E). The decreased expression of tubulin acetylation and the suppressed migration 353
and invasion capabilities also confirmed in CAMSAP2 knockout HCC cell line as 354
well as SW1990 and SW620 CAMSAP2 knockdown cells lines (Figure S2A-C, 355
Figure S6A-B). Intriguingly, in CAMSAP2-knockdown cells, the polarized acetylated 356
microtubule network was no longer asymmetrical, but showed a symmetrical radial 357
pattern, as revealed by co-immunostaining (Figure 6B). IF staining revealed that the 358
acetylated microtubule network, which typically extends towards the direction of cell 359
migration, was disrupted in migrating CAMSAP2-knockdown HCC cells (Figure 6C). 360
Acetylation of a canonical α-tubulin site has been identified on lysine 40, and 361
α-tubulin point mutations at lysine 40 to arginine (K40R) abrogate tubulin acetylation. 362
To further confirm the role of tubulin acetylation in CAMSAP2-mediated migration 363
and invasion of HCC cells, Huh7-Lv-CAMSAP2 cells were transfected with 364
tubulin-K40R mutant or WT plasmid. Western blot analysis revealed that 365
tubulin-K40R markedly ameliorated the increased expression of microtubule 366
acetylation induced by CAMSAP2 overexpression (Figure S6C). The elevated 367
migration and invasion of CAMSAP2-overexpressing HCC cells were significantly 368
abrogated after K40R mutation. (Figure S6D). 369
370
18
Microtubule acetylation is tightly controlled by the enzymatic activities of multiple 371
acetyltransferases and is negatively regulated through the activities of the deacetylases 372
HDAC6 and to a lesser extent, sirtuin 2 [16,17]. Therefore, we hypothesized that 373
CAMSAP2 may promote microtubule acetylation by either activating 374
acetyltransferases or inhibiting deacetylases. Inhibition of HDAC6 activity by the 375
specific inhibitor tubacin or HDAC6 depletion enhanced microtubule acetylation in 376
HCC cells (Figure 6D). Acetylated microtubule extended towards the leading edge 377
after HDAC6 inactivation or depletion compared with that in control cells (Figure 6E). 378
However, suppression of sirtuin 2, the only other microtubule deacetylase identified, 379
by treatment with the specific inhibitor thiomyristoyl had no significant effect on 380
α-tubulin acetylation (Figure 6D-E). Inactivation or downregulation of HDAC6 381
dramatically increased the migration and invasion capabilities of HCC cells (Figure 382
6F). Thus, HDAC6 was identified as the predominant α-tubulin deacetylase in HCC 383
cells. In addition, suppression of tubulin detyrosination by treatment with the inhibitor 384
PTL had no significant effect on HCC cell migration and invasion (Figure S5B-C). 385
386
To validate the deacetylation role of HDAC6 in CAMSAP2-mediated noncentrosomal 387
microtubule acetylation, we examined HDAC6 expression in transfected HCC cells. 388
Both western blotting and real-time PCR analysis showed that CAMSAP2 389
knockdown in MHCC97H cells significantly increased HDAC6 expression, and that 390
HDAC6 was markedly downregulated in CAMSAP2-overexpressing Huh7 cells with 391
no apparent change in the expression of αTAT1 in the transfected HCC cells (Figure 392
19
6G, Figure S7A). The increased expression of HDAC6 also confirmed in CAMSAP2 393
knockout HCC cell line (Figure S2B). Overexpression of HDAC6 abrogated the 394
increase in microtubule acetylation induced by CAMSAP2 overexpression, 395
suggesting that CAMSAP2-mediated microtubule acetylation is likely due to 396
dysfunctional HDAC6 rather than the activation of acetyltransferases (Figure 6H-I). 397
Upregulation of HDAC6 significantly ameliorated the enhanced migration and 398
invasion capabilities of CAMSAP2-overexpressing Huh7 cells (Figure 6J). 399
400
To determine the effect of HDAC6 on CAMSAP2-mediated metastasis in vivo, we 401
established an orthotopic metastatic model. HDAC6 down-regulation markedly 402
increased the incidence of lung metastasis and decreased the OS time of the 403
MHCC97H-shCAMSAP2 group compared with that in the control group (Figure 404
S7B-C). Histological analysis further confirmed that the incidence of lung metastasis 405
and the number of metastatic lung nodules in the 406
MHCC97H-shCAMSAP2-shHDAC6 group were significantly increased (Figure 407
S7D-E). Collectively, these results suggested that CAMSAP2 promotes microtubule 408
acetylation to control invasive cell migration via an inhibitory interaction with 409
HDAC6. 410
411
Prognostic significance of the correlation between CAMSAP2 and acetylated 412
α-tubulin or HDAC6 expression in HCC tissues 413
We evaluated the putative relationship between CAMSAP2 and acetylated α-tubulin 414
20
or HDAC6 expression in the two independent HCC cohorts. IHC revealed that 415
CAMSAP2 overexpression was positively correlated with acetylated α-tubulin, but 416
inversely correlated with HDAC6 expression (Figure 7A-B). Positive and negative 417
controls are shown in Figure S3B. Both western blotting and IF further confirmed that 418
HDAC6 dramatically down-regulated in HCC tissues (Figure S6E-F). Surprisingly, 419
unlike acetylated microtubule s, detyrosinated α-tubulin scarcely is expressed in HCC 420
tissue but is expressed specifically in endothelial tissue (Figure S5D). Further analysis 421
of the clinicopathological characteristics in paired HCC and adjacent nontumor tissues 422
indicated that upregulation of microtubule acetylation or downregulation of HDAC6 423
was significantly correlated with malignant tumor progression (Tables S6 and S7) and 424
poor prognosis (Figure 7C). Kaplan–Meier analysis revealed that in both cohorts, 425
patients with CAMSAP2(+)/acetylated α-tubulin(+) and CAMSAP2(+)/HDAC6(–) 426
expression patterns showed the highest recurrence rate and lowest OS (Figure 7D). 427
428
CAMSAP2 activates c-Jun to induce transrepression of HDAC6 through 429
Trio-dependent activation of the Rac1/JNK pathway 430
Recent research findings have revealed that the HDAC6 expression is inhibited by 431
c-Jun [21], a downstream transcription factor in Rac1/JNK signaling [22]. Thus, we 432
hypothesized that endogenous HDAC6 expression may be suppressed by CAMSAP2 433
through Trio-dependent activation of the Rac1/JNK/c-Jun pathway. CAMSAP2 434
knockdown decreased JNK and c-Jun phosphorylation and microtubule acetylation, 435
whereas it increased HDAC6 protein expression in HCC cells (Figure 8A). 436
21
Suppression of Rac1 activation upon treatment with the specific inhibitor NSC23766 437
or EB1 and Trio depletion abrogated JNK/c-Jun cascade activation induced by 438
CAMSAP2 overexpression (Figure 8B). Pretreatment with the specific JNK inhibitor 439
SP600125 or c-Jun knockdown to inactivate c-Jun in CAMSAP2-overexpressing 440
Huh7 cells significantly rescued the decrease in HDAC6 expression induced by 441
CAMSAP2 overexpression; α-tubulin acetylation was markedly decreased compared 442
with the level in the controls (Figure 8C). Further, the elevated migration and invasion 443
due to CAMSAP2 overexpression were significantly abrogated after SP600125 444
treatment or c-Jun knockdown in Huh7 cells (Figure 8D). The similar results also 445
confirmed in MHCC97H cells (Figure S8A-B). 446
447
As activated c-Jun transrepresses target genes by directly interacting with their 448
promoters, we examined whether JNK/c-Jun inactivates HDAC6 via a similar 449
mechanism. RT-qPCR and luciferase reporter assays revealed that overexpression of 450
CAMSAP2 inhibited HDAC6 mRNA expression and promoter activity (Figure 8E-F). 451
Pretreatment with SP600125 or c-Jun knockdown significantly rescued the decreases 452
in HDAC6 expression and promoter activity induced by CAMSAP2 overexpression 453
(Figure 8E-F). Sequence analysis revealed three putative c-Jun-binding sites in the 454
HDAC6 promoter. Serial truncation experiments indicated that the first c-Jun-binding 455
sites are critical for c-Jun-induced HDAC6 transrepression (Figure 8G). A luciferase 456
reporter assay revealed that mutation of the first c-Jun-binding site significantly 457
ameliorated c-Jun-induced transrepression of the HDAC6 promoter (Figure 8G). 458
22
Chromatin immunoprecipitation assay results confirmed that c-Jun binds directly to 459
the HDAC6 promoter in HCC cells (Figure 8H). These data suggested that 460
CAMSAP2 activates c-Jun to induce transrepression of HDAC6 through 461
Trio-dependent activation of the Rac1/JNK pathway. 462
463
The Rac inhibitor NSC23766 suppresses CAMSAP2-mediated HCC invasion and 464
metastasis 465
NSC23766, a Rac-specific inhibitor that does not affect Cdc42 or RhoA, has shown 466
remarkable antitumor effects in multiple cancers [23-25]. Therefore, we determined 467
whether treatment with NSC23766 could reverse CAMSAP2-mediated HCC invasion 468
and metastasis. NSC23766 effectively ameliorated the increase in GTP-Rac1 469
expression in Huh7-Lv-CAMSAP2 cells (Figure 4E) and abrogated JNK-c-Jun 470
pathway activation induced by CAMSAP2 overexpression (Figure 8B). The elevated 471
migration and invasion of CAMSAP2-overexpressing HCC cells were significantly 472
abrogated after NSC23766 treatment (Figure 4F). We next evaluated an intrahepatic 473
tumor implantation mouse model. Daily administration of NSC23766 in a single 474
intraperitoneal dose of 3 mg/kg significantly reduced lung metastasis and prolonged 475
the OS time in Huh7-Lv-CAMSAP2 cells (Figure 8I-K). H&E staining confirmed that 476
the incidence of lung metastasis and the number of metastatic pulmonary nodules 477
were significantly decreased upon NSC23766 treatment (Figure 8L-M). These 478
findings suggested that NSC23766 inhibits CAMSAP2-mediated HCC invasion and 479
metastasis by interrupting Rac1/JNK/c-Jun signaling and may provide a novel 480
23
therapeutic option for HCC. 481
482
Discussion 483
This study revealed that CAMSAP2, a microtubule minus-end-targeting protein, 484
functions as an oncoprotein in HCC metastasis. CAMSAP2 upregulation promoted 485
HCC cell invasion and metastasis in vitro and in vivo, whereas loss of CAMSAP2 486
suppressed HCC cell growth. Upregulated CAMSAP2 expression in HCC tissues was 487
closely correlated with poor clinicopathological characteristics and prognosis, and 488
clinical evidenced indicated that CAMSAP2 may serve as a prognostic factor for 489
HCC metastasis. Moreover, the increased mRNA expression of CAMSAP2 was also 490
observed in PAAD, STAD and COAD tissues based on TCGA data. Kaplan-Meier 491
analysis revealed that PAAD, STAD and COAD patients with high levels of 492
CAMSAP2 mRNA had a shorter OS and DFS. IHC staining confirmed that 493
CAMSAP2 was upregulated in PAAD, STAD and COAD cancer specimens. The 494
decreased migration and invasion capabilities also confirmed in SW1990 and SW620 495
CAMSAP2 knockdown cells. These results suggest that CAMSAP2 may play a 496
carcinogenic role in a variety of tumors. 497
498
Our data suggested that CAMSAP2-mediated noncentrosomal microtubule 499
rearrangement is essential for directional cell migration in HCC. Upon CAMSAP2 500
depletion, the partially noncentrosomal microtubule array was transformed into a 501
completely radial centrosomal pattern and directional cell migration was impaired and 502
24
characterized by defective Golgi reorienting towards the wound edge. 503
CAMSAP2-decorated microtubule stretches function during interphase, but disappear 504
during mitosis [5,10]; however, we found that CAMSAP2, along with α-tubulin, 505
exhibited a specific distribution pattern in HCC cells during mitosis. The molecular 506
effects and mechanism of CAMSAP2 in mitosis remain to be further investigated 507
(Figure S9A-B). 508
509
Microtubules are polarized cytoskeletal filaments with two structurally and 510
functionally distinct ends. It is generally accepted that the plus end is the dynamic end, 511
which recruits multiple +TIPs to couple the microtubule dynamics to specific cellular 512
events [7]. We investigated the regulatory roles of CAMSAP2 in microtubule 513
dynamics by monitoring the behavior of EB1, which is considered to be the core of 514
the +TIPs network as it autonomously tracks the growing microtubule plus ends [7, 515
18]. Loss of CAMSAP2 led to reductions in the amount of polymerized a-tubulin as 516
well as growing microtubule plus ends. Further, microtubule plus ends function as 517
concentration devices for signaling molecules—Rho GTPase GEFs and kinases [6,19]. 518
We found that CAMSAP2 cooperates with the EB1/Trio complex to direct 519
microtubule dynamics to invasive cell migration by promoting Rac1 signaling. 520
521
PTMs are emerging as important regulators of the microtubule cytoskeleton. Tubulin 522
PTMs provide a mechanism for microtubules to adapt to specific cellular functions by 523
regulating microtubule properties [13,14]. α-Tubulin acetylated at lysine 40 is a 524
25
hallmark of poor prognosis in patients with breast, head-and-neck, and pancreatic 525
cancer [26-28]. Acetylated microtubules accumulate toward the leading edge and 526
regulate cell polarization, migration, and invasion [16,17]. Therefore, we questioned 527
whether CAMSAP2-mediated noncentrosomal microtubules are acetylated to promote 528
HCC cell invasive migration. CAMSAP2 depletion decreased microtubule acetylation, 529
whereas CAMSAP2 overexpression increased α-tubulin acetylation in HCC cells. The 530
acetylated microtubule network, which typically extends protrusions towards the 531
direction of migration, was disrupted in CAMSAP2-knockdown cells. However, there 532
were no obvious changes in the expression of detyrosinated or tyrosinated α-tubulin 533
after downregulation of CAMSAP2. Further, suppression of tubulin detyrosination by 534
treatment with the inhibitor parthenolide had no significant effect on HCC cell 535
migration and invasion. These results suggested that microtubule detyrosination may 536
not be involved in HCC cells migration and invasion induced by CAMSAP2. 537
Surprisingly, unlike acetylated microtubules, detyrosinated α-tubulin scarcely is 538
expressed in HCC tissue but is expressed specifically in endothelial tissue, which 539
suggests that detyrosinated microtubules may be related to angiogenesis in HCC. 540
541
Acetylation of α-tubulin at lysine 40 is tightly controlled by multiple 542
acetyltransferases and deacetylases. Numerous enzymes, including but not limited to 543
α-TAT1, ELP3, ARD1–NAT1, and GCN5, acetylate tubulin [29]. In contrast, HDAC6 544
and sirtuin 2 are the only α-tubulin deacetylases described. Here, we identified 545
HDAC6 as the predominant α-tubulin deacetylase in HCC cells: HDAC6 inhibition or 546
26
depletion enhanced microtubule acetylation and promoted migration and invasion in 547
HCC cells, whereas the protein level and distribution of acetylated microtubules were 548
not obviously affected by sirtuin 2 suppression. We found that knockdown of 549
CAMSAP2 in MHCC97H cells significantly increased HDAC6 expression, whereas 550
HDAC6 was markedly downregulated in Huh7-Lv-CAMSAP2 cells with no apparent 551
change in the expression of αTAT1 in the transfected HCC cells. Upregulation of 552
HDAC6 abrogated the increases in migration and invasion as well as the expression 553
and morphological alteration of acetylated α-tubulin induced by CAMSAP2 554
overexpression, suggesting that CAMSAP2 promotes microtubule acetylation to 555
control invasive cell migration via an inhibitory interaction with HDAC6. 556
557
Notably, there are conflicting studies showing that HDAC6 has been implicated as an 558
oncogene in several human cancers, including HCC [30, 31]. However, recent studies 559
from two other groups and our study showed that HDAC6 was uniquely 560
down-regulated and endowed with tumor suppressive activity in HCC [32, 33]. 561
HDACs are widely considered to be critical epigenetic regulators involved in the 562
regulation of malignancy and tumor microenvironment due to the importance of 563
histone modifications [34]. Unlike other HDACs, HDAC6 contains two catalytic 564
domains and a conserved nuclear export signal, and thus may function predominately 565
in the cytoplasm where it associates with modifications of nonhistones, such as α566
-tubulin [33, 35]. Therefore, to better clarify the role and mechanism of HDAC6 in 567
HCC, it would be too much of demand to use HDAC6 knockout cell lines in 568
27
transplantation experiments or construct liver-specific HDAC6 conditional knockout 569
mice. 570
571
Recent research findings have revealed that the HDAC6 expression is inhibited by 572
c-Jun [21], a downstream transcription factor in Rac1/JNK signaling [22]. Thus, we 573
hypothesized that endogenous HDAC6 expression may be suppressed by CAMSAP2 574
through Trio-dependent activation of the Rac1/JNK/c-Jun pathway. Our results 575
indicated that CAMSAP2 represses endogenous HDAC6 expression through 576
Trio-dependent activation of the Rac1/JNK /c-Jun pathway. Ample experimental 577
evidence indicated that HDAC6 is a direct transcriptional target of c-Jun. 578
579
In an attempt to develop a new pharmacological strategy against HCC metastasis 580
based on our findings, we focused on inhibitors targeting Rac1 because CAMSAP2 581
inhibitors are currently not available. NSC23766 has shown remarkable antitumor 582
effects in multiple cancers, including chronic myelogenous leukemia, glioblastoma, 583
and colorectal cancer [23-25]. Thus, we determined whether treatment with 584
NSC23766 could reverse CAMSAP2-mediated HCC invasion and metastasis. We 585
found that NSC23766 inhibited CAMSAP2-mediated HCC invasion and metastasis 586
by interrupting Rac1/JNK/c-Jun signaling; thus, it may serve as a novel therapeutic 587
option for HCC. 588
589
Thus, CAMSAP2 is functionally, mechanistically, and clinically oncogenic in HCC. 590
28
Targeting CAMSAP2-mediated noncentrosomal microtubule acetylation may provide 591
new therapeutic strategies for HCC metastasis (Figure S9C). 592
593
Abbreviations 594
CAMSAP: calmodulin-regulated spectrin-associated protein; COAD: colon 595
adenocarcinoma; DFS: disease-free survival; GEFs: guanine nucleotide exchange 596
factors; HCC: hepatocellular carcinoma; HDAC6: histone deacetylase 6; H&E: 597
hematoxylin and eosin; IF: immunofluorescence; IHC: immunohistochemistry; K40R: 598
lysine 40 to arginine; OS: overall survival; PAAD: pancreatic adenocarcinoma; PBS: 599
phosphate-buffered saline; PCR: polymerase chain reaction; PTMs: post-translational 600
modifications; siRNA: small interfering RNA; STAD: stomach adenocarcinoma; 601
TCGA: The Cancer Genome Atlas dataset; +TIPs, plus-end-tracking proteins; Trio: 602
triple functional domain protein; 603
604
Supplementary Material 605
Supplementary materials and methods, nine supplementary figures, and nine 606
supplementary tables, can be found with this article online. 607
608
Acknowledgements 609
This research was supported by grants from the National Key Research and 610
Development Program of China No. 2018YFC1312103 (L.X.), National Natural 611
Science Foundation of China No. 81772610 (D.T.), No. 81974071 (D.T.), No. 612
29
81972237 (L.X.), and No. 81772623 (L.X.), and the Fundamental Research Funds for 613
the Central Universities (to X.D.: No. 2018JYCXJJ005). 614
615
Author Contributions 616
Dongxiao Li and Xiangming Ding conceived, designed and performed the 617
experiments, analyzed data and wrote the manuscript. Meng Xie and Zheng Huang 618
performed animal experiments. Ping Han performed 3D cell invasion model. Dean 619
Tian and Limin Xia designed, supervised the study and provided critical review. 620
621
Competing Interests 622
No potential conflicts of interest were disclosed. 623
624
30
References 625
1. Forner A, Reig M, Bruix J. Hepatocellular carcinoma. Lancet. 2018; 391: 626
1301-14. 627
2. El-Serag HB. Hepatocellular carcinoma. N Engl J Med. 2011; 365: 1118-27. 628
3. Dumontet C, Jordan MA. Microtubule-binding agents: a dynamic field of cancer 629
therapeutics. Nat Rev Drug Discov. 2010; 9: 790-803. 630
4. Silva VC, Cassimeris L. CAMSAPs add to the growing microtubule minus-end 631
story. Dev Cell. 2014; 28: 221-2. 632
5. Akhmanova A, Hoogenraad CC. Microtubule minus-end-targeting proteins. Curr 633
Biol. 2015; 25: R162-71. 634
6. Akhmanova A, Steinmetz MO. Control of microtubule organization and 635
dynamics: two ends in the limelight. Nat Rev Mol Cell Biol. 2015; 16: 711-26. 636
7. Akhmanova A, Steinmetz MO. Tracking the ends: a dynamic protein network 637
controls the fate of microtubule tips. Nat Rev Mol Cell Biol. 2008; 9: 309-22. 638
8. Pongrakhananon V, Saito H, Hiver S, Abe T, Shioi G, Meng W, et al. CAMSAP3 639
maintains neuronal polarity through regulation of microtubule stability. 640
Proc Natl Acad Sci U S A. 2018; 115: 9750-5. 641
9. Tanaka N, Meng W, Nagae S, Takeichi M. Nezha/CAMSAP3 and CAMSAP2 642
cooperate in epithelial-specific organization of noncentrosomal microtubules. 643
Proc Natl Acad Sci U S A. 2012; 109: 20029-34. 644
10. Jiang K, Hua S, Mohan R, Grigoriev I, Yau KW, Liu Q, et al. Microtubule 645
minus-end stabilization by polymerization-driven CAMSAP deposition. Dev Cell. 646
2014; 28: 295-309. 647
11. Wu J, de Heus C, Liu Q, Bouchet BP, Noordstra I, Jiang K, et al. Molecular 648
Pathway of Microtubule Organization at the Golgi Apparatus. Dev Cell. 2016; 39: 649
44-60. 650
12. Abal M, Piel M, Bouckson-Castaing V, Mogensen M, Sibarita JB, Bornens M. 651
Microtubule release from the centrosome in migrating cells. J Cell Biol. 2002; 159: 652
731-7. 653
13. Janke C, Bulinski JC. Post-translational regulation of the microtubule 654
31
cytoskeleton: mechanisms and functions. Nat Rev Mol Cell Biol. 2011; 12: 773-86. 655
14. Magiera MM, Singh P, Gadadhar S, Janke C. Tubulin Posttranslational 656
Modifications and Emerging Links to Human Disease. Cell. 2018; 173: 1323-7. 657
15. Park IY, Powell RT, Tripathi DN, Dere R, Ho TH, Blasius TL, et al. Dual 658
Chromatin and Cytoskeletal Remodeling by SETD2. Cell. 2016; 166: 950-62. 659
16. Deakin NO, Turner CE. Paxillin inhibits HDAC6 to regulate microtubule 660
acetylation, Golgi structure, and polarized migration. J Cell Biol. 2014; 206: 395-413. 661
17. Montagnac G, Meas-Yedid V, Irondelle M, Castro-Castro A, Franco M, Shida T, 662
et al. alphaTAT1 catalyses microtubule acetylation at clathrin-coated pits. Nature. 663
2013; 502: 567-70. 664
18. Zhang Y, Luo Y, Lyu R, Chen J, Liu R, Li D, et al. Proto-Oncogenic Src 665
Phosphorylates EB1 to Regulate the Microtubule-Focal Adhesion Crosstalk and 666
Stimulate Cell Migration. Theranostics. 2016; 6: 2129-40. 667
19. van Haren J, Boudeau J, Schmidt S, Basu S, Liu Z, Lammers D, et al. Dynamic 668
microtubules catalyze formation of navigator-TRIO complexes to regulate neurite 669
extension. Curr Biol. 2014; 24: 1778-85. 670
20. Magiera MM, Singh P, Janke C. SnapShot: Functions of Tubulin Posttranslational 671
Modifications. Cell. 2018; 173: 1552- e1. 672
21. Bae HJ, Jung KH, Eun JW, Shen Q, Kim HS, Park SJ, et al. MicroRNA-221 673
governs tumor suppressor HDAC6 to potentiate malignant progression of liver cancer. 674
J Hepatol. 2015; 63: 408-19. 675
22. Coso OA, Chiariello M, Yu JC, Teramoto H, Crespo P, Xu N, et al. The small 676
GTP-binding proteins Rac1 and Cdc42 regulate the activity of the JNK/SAPK 677
signaling pathway. Cell. 1995; 81: 1137-46. 678
23. Thomas EK, Cancelas JA, Chae HD, Cox AD, Keller PJ, Perrotti D, et al. Rac 679
guanosine triphosphatases represent integrating molecular therapeutic targets for 680
BCR-ABL-induced myeloproliferative disease. Cancer cell. 2007; 12: 467-78. 681
24. Zhou K, Rao J, Zhou ZH, Yao XH, Wu F, Yang J, et al. RAC1-GTP promotes 682
epithelial-mesenchymal transition and invasion of colorectal cancer by activation of 683
STAT3. Lab Invest. 2018; 98: 989-98. 684
32
25. Karpel-Massler G, Westhoff MA, Zhou S, Nonnenmacher L, Dwucet A, Kast RE, 685
et al. Combined inhibition of HER1/EGFR and RAC1 results in a synergistic 686
antiproliferative effect on established and primary cultured human glioblastoma cells. 687
Mol Cancer Ther. 2013; 12: 1783-95. 688
26. Boggs AE, Vitolo MI, Whipple RA, Charpentier MS, Goloubeva OG, Ioffe OB, 689
et al. alpha-Tubulin acetylation elevated in metastatic and basal-like breast cancer 690
cells promotes microtentacle formation, adhesion, and invasive migration. Cancer Res. 691
2015; 75: 203-15. 692
27. Bailey JM, Alsina J, Rasheed ZA, McAllister FM, Fu YY, Plentz R, et al. DCLK1 693
marks a morphologically distinct subpopulation of cells with stem cell properties in 694
preinvasive pancreatic cancer. Gastroenterology. 2014; 146: 245-56. 695
28. Saba NF, Magliocca KR, Kim S, Muller S, Chen Z, Owonikoko TK, et al. 696
Acetylated tubulin (AT) as a prognostic marker in squamous cell carcinoma of the 697
head and neck. Head Neck Pathol. 2014; 8: 66-72. 698
29. Song Y, Brady ST. Post-translational modifications of tubulin: pathways to 699
functional diversity of microtubules. Trends Cell Biol. 2015; 25: 125-36. 700
30. Batchu SN, Brijmohan AS, Advani A. The therapeutic hope for HDAC6 701
inhibitors in malignancy and chronic disease. Clin Sci (Lond). 2016; 130: 987-1003. 702
31. Kanno K, Kanno S, Nitta H, Uesugi N, Sugai T, Masuda T, et al. Overexpression 703
of histone deacetylase 6 contributes to accelerated migration and invasion activity of 704
hepatocellular carcinoma cells. Oncol Rep. 2012; 28: 867-73. 705
32. Yang HD, Kim HS, Kim SY, Na MJ, Yang G, Eun JW, et al. HDAC6 Suppresses 706
Let-7i-5p to Elicit TSP1/CD47-Mediated Anti-Tumorigenesis and Phagocytosis of 707
Hepatocellular Carcinoma. Hepatology. 2019; 70: 1262-79. 708
33. Qiu W, Wang B, Gao Y, Tian Y, Tian M, Chen Y, et al. Targeting HDAC6 709
Reprograms TH 17 Pathogenicity and Facilitates Immunotherapies for Hepatocellular 710
Carcinoma. Hepatology, in press. doi:10.1002/hep.30960. 711
34. Bahhaj Fe, Denis I, Pichavant L, Delatouche R, Collette F, Linot C, et al. Histone 712
Deacetylase Inhibitors Delivery using Nanoparticles with Intrinsic Passive Tumor 713
Targeting Properties for Tumor Therapy. Theranostics. 2016; 6: 795-807. 714
33
35. Yang CJ, Liu YP, Dai HY, Shiue YL, Tsai CJ, Huang MS, et al. Nuclear HDAC6 715
inhibits invasion by suppressing NF-kappaB/MMP2 and is inversely correlated with 716
metastasis of non-small cell lung cancer. Oncotarget. 2015; 6: 30263-76. 717
718
719
Figure legends 720
Figure 1. CAMSAP2 is significantly upregulated in HCC tissues and indicates a 721
poor prognosis. 722
(A) Relative CAMSAP2 mRNA levels in 20 normal liver samples and 90 paired HCC 723
and adjacent nontumorous samples (Left). Relative CAMSAP2 mRNA levels in HCC 724
patients with (n = 48) or without (n = 42) recurrence (Middle). Relative CAMSAP2 725
mRNA levels in 20 paired primary and metastatic HCC tissues (Right). **P < 0.01. 726
(B) Representative IHC staining images showing CAMSAP2 expression in HCC and 727
adjacent nontumorous tissues. Scale bars, 500 µm (upper), 100 µm (lower). (C) 728
Representative IF staining images of CAMSAP2 (green), α-tubulin (red), and DNA 729
(DAPI, blue) in HCC and matched adjacent nontumorous tissues. Scale bars, 50 µm 730
(left), 100 µm (right). (D) Representative IHC staining images of CAMSAP2 in HCC 731
and peritumoral area. Scale bars, 500 µm (upper), 100 µm (lower). (E) CAMSAP2 732
protein levels in 8 paired HCC and adjacent nontumorous tissues as detected by 733
western blotting. N, adjacent nontumorous tissue; T, tumor tissue. (F) IHC scores of 734
CAMSAP2 in two independent cohorts of human HCC (cohort I, n = 360; cohort II, n 735
= 178). (G) Kaplan–Meier analysis was used to determine the correlation between 736
CAMSAP2 expression and recurrence or OS in two independent cohorts. 737
34
738
Figure 2. CAMSAP2 promotes HCC invasion and metastasis in vitro and in vivo. 739
(A) Western blot analysis of CAMSAP2 in the indicated transfected HCC cells. (B) 740
Transwell assays of the indicated HCC cells. Migrating and invading cells were 741
quantified in (C). Data are the mean ± SEM from triplicate experiments. **P < 0.01. 742
Scale bar, 400 μm. (D) IF staining of ɑ-tubulin (green), F-actin (red), and DNA (DAPI, 743
blue) to monitor three-dimensional invasion morphology. Scale bar, 100 μm. (E) 744
Incidence of lung metastasis and bioluminescence imaging of each group at 10 weeks 745
after orthotopic xenografting with the indicated HCC cells. (F) OS of mice in the 746
different groups. (G) Representative H&E staining of lung tissues from each group. 747
Scale bars, 500 mm (upper), 500 µm (lower). (H) Number of metastatic lung nodules 748
observed in each group. **P < 0.01. 749
750
Figure 3. CAMSAP2-mediated noncentrosomal microtubule remodeling 751
contributes to directional cell migration in HCC cells. 752
(A) IF staining of CAMSAP2 (green) and α-tubulin (red) in MHCC97H and Huh7 753
cells. Scale bars, 100 µm, 20 µm (insert). (B) IF staining of CAMSAP2 (green), 754
α-tubulin (red), and DNA (DAPI, blue) in the wound-healing edge of MHCC97H 755
cells. Scale bars, 50 µm. (C) Endogenous co-immunoprecipitation assays revealed 756
that CAMSAP2 interacts with α-tubulin and EB1 in MHCC97H cells. IgG was used 757
as a negative control. (D) Double IF staining of α-tubulin (red) and 758
γ-tubulin/CAMSAP2 (green) in control and MHCC97H-shCAMSAP2 cells. Scale bar, 759
35
100 µm. (E) Cells were treated with 15 µM nocodazole at 4°C for 30 min, incubated 760
min at 37°C for another 10 min after drug washout, fixed, and immunostained for 761
α-tubulin (red) and γ-tubulin (green). Scale bars, 500 µm (left), 200 µm (right). (F) 762
Double immunostaining for GM130 (green) and α-tubulin (red) in control and 763
CAMSAP2-knockdown cells. Scale bars, 200 µm (left), 100 µm (right). (G) 764
Quantification of CAMSAP2 and GM130 distribution in CAMSAP2-depleted and 765
control cells. Distance from the center of GM130 is shown on the horizontal axis. 766
Twenty cells were analyzed per group. Intensities were normalized to the maximum 767
intensity of each cell. Data are the mean ± SEM from triplicate experiments. (H) 768
Reorientation of the Golgi apparatus towards the direction of cell migration after 12 h 769
of monolayer wound healing in control and MHCC97H-shCAMSAP2 cells. IF 770
staining of α-tubulin (red), the GM130 (green), and DNA (DAPI, blue). Scale bar, 100 771
µm. (I) Quantification of directional cell migration in control and 772
MHCC97H-shCAMSAP2 cells. A cell was considered reoriented (+) when GM130 773
was positioned in the 120° arc towards the wound edge. Approximately 120–150 cells 774
were analyzed for each condition. Data are the mean ± SEM from triplicate 775
experiments. *P < 0.05. 776
777
Figure 4. CAMSAP2 cooperates with EB1 to regulate microtubule dynamics and 778
invasive cell migration via Trio/Rac1 signaling. 779
(A) Microtubule fractionation was performed to detect depolymerized (soluble, S) and 780
polymerized (pelleted, P) fractions of α-tubulin and EB1. The ratio of polymerized 781
36
ɑ-tubulin/EB1 to the total amount is presented in the right panel. Data are the mean ± 782
SEM from triplicate experiments. *P < 0.05. (B) Endogenous co-immunoprecipitation 783
assays revealed that CAMSAP2 interacts with Trio, α-tubulin, and EB1 in HCC cells. 784
IgG was used as a negative control. (C) Western blot analysis of active-Rac1, total 785
Rac1, and α-tubulin in the indicated HCC cells. (D) Western blot analysis of 786
CAMSAP2, EB1, and Trio in the indicated transfected HCC cells. (E) Western blot 787
analysis of active-Rac1, total Rac1, and α-tubulin in the indicated HCC cells. 788
NSC23766, a specific inhibitor was used to inactivate Rac1. (F) Transwell assay of 789
the indicated HCC cells. Migrating and invading cells were quantified in the lower 790
panel. Data are the mean ± SEM from triplicate experiments. Scale bar, 400 μm. *P < 791
0.05. (G) Incidence of lung metastasis and bioluminescence imaging of each group at 792
10 weeks after orthotopic xenografting with the indicated HCC cells. (H) OS of mice 793
in the different groups. (I) Representative H&E staining of lung tissues from each 794
group. Scale bars, 500 mm (upper), 500 µm (lower). (J) Number of metastatic lung 795
nodules observed in each group. *P < 0.05, **P < 0.01. 796
797
Figure 5. Prognostic significance of the correlation between CAMSAP2 and EB1 798
or Trio expression in HCC tissues. 799
(A) Representative IHC images of CAMSAP2, EB1, and Trio expression in HCC and 800
adjacent nontumorous tissues. Scale bars, 500 µm (upper), 100 µm (lower). (B) 801
Relationship between CAMSAP2 expression and EB1 or Trio expression in two 802
independent cohorts. (C) Kaplan–Meier analysis was used to determine the 803
37
correlation between EB1 or Trio expression and recurrence or OS in the two cohorts. 804
(D) Kaplan–Meier analysis of the correlation between CAMSAP2/EB1 or 805
CAMSAP2/Trio coexpression and recurrence and OS in the two cohorts. 806
807
Figure 6. CAMSAP2 promotes microtubule acetylation to control invasive cell 808
migration via an inhibitory interaction with HDAC6. 809
(A) Western blot analysis of CAMSAP2 and acetylated α-tubulin in the indicated cells. 810
Ace-tubulin, acetylated α-tubulin. (B) IF staining of α-tubulin (red) and Ace-tubulin 811
(green) in control and MHCC97H-shCAMSAP2 cells. Scale bars, 200 µm. (C) IF 812
staining of α-tubulin (red), Ace-tubulin (green), and DNA (DAPI, blue) in indicated 813
cells at the edge of directional migration. Scale bars, 100 µm. (D) Western blot 814
analysis of Ace-tubulin in the indicated HCC cells. Tubacin and thiomyristoyl (TM) 815
were used to inactivate HDAC6 and sirtuin 2, respectively. (E) IF staining of 816
α-tubulin (red), Ace-tubulin (green), and DNA (DAPI, blue) in the indicated HCC 817
cells. Scale bars, 200 µm. Quantification of α-tubulin and Ace-tubulin distribution in 818
the indicated HCC cells. Distance from the microtubule-organizing center is displayed 819
on the horizontal axis. Intensities were normalized to the maximum intensity of each 820
cell. Twenty-five cells were analyzed for each condition. Data are the mean ± SEM 821
from triplicate experiments. (F) Transwell assays of the indicated HCC cells. 822
Migrating and invading cells were quantified in the lower panel. Data are the mean ± 823
SEM from triplicate experiments. *P < 0.05. Scale bar, 400 μm. (G) Protein levels of 824
CAMSAP2, HDAC6, Ace-tubulin and αTAT1 in the indicated cells as determined by 825
38
western blotting. (H) Western blot analysis of CAMSAP2, HDAC6, and Ace-tubulin 826
in the indicated HCC cells. (I) IF staining of Ace-tubulin (green) in the indicated cells. 827
Scale bars, 50 µm. (J) Transwell assays of the indicated HCC cells. Migrating and 828
invading cells were quantified in the right panel. Data are the mean ± SEM from 829
triplicate experiments. **P < 0.01. Scale bar, 400 μm (left). 830
831
Figure 7. Prognostic significance of the correlation between CAMSAP2 and 832
acetylated α-tubulin or HDAC6 expression in HCC tissues. 833
(A) Representative IHC images of CAMSAP2, Ace-tubulin, and HDAC6 expression 834
in HCC and adjacent nontumorous tissues. Scale bars, 500 µm (upper), 100 µm 835
(lower). (B) Relationship between CAMSAP2 expression and Ace-tubulin or HDAC6 836
expression in two independent cohorts. (C) Kaplan–Meier analysis was used to 837
determine the correlation between Ace-tubulin or HDAC6 expression and recurrence 838
or OS in the two cohorts. (D) Kaplan–Meier analysis of the correlation of 839
CAMSAP2/Ace-tubulin or CAMSAP2/HDAC6 coexpression with recurrence and OS 840
in the two cohorts. 841
842
Figure 8. CAMSAP2 activates c-Jun to induce transrepression of HDAC6 843
through Trio-dependent activation of Rac1/JNK pathway. 844
(A) Western blot analysis of CAMSAP2, HDAC6, Ace-tubulin, phosphorylated JNK, 845
and c-Jun in control and CAMSAP2-depleted HCC cells. (B) Western blot analysis of 846
active-Rac1, total Rac1, phosphorylated JNK, and c-Jun in the indicated HCC cells. 847
39
NSC23766, a specific inhibitor was used to inactivate Rac1. (C) Western blot analysis 848
of CAMSAP2, HDAC6, Ace-tubulin, phosphorylated JNK, and c-Jun in 849
pCMV-CAMSAP2 and control cells (denoted as ―pCMV-Taq‖) treated with 850
SP600125 or c-Jun knockdown lentivirus (denoted as ―shc-Jun‖). (D) Transwell 851
assays of the indicated HCC cells. Migrating and invading cells were quantified in the 852
right panel. Data are the mean ± SEM from triplicate experiments. Scale bar, 400 μm. 853
**P < 0.01, *P < 0.05 (E) HDAC6 mRNA levels in the indicated HCC cells as 854
measured by RT-qPCR. Data are the mean ± SEM from triplicate experiments. *P < 855
0.05. (F) HDAC6 promoter activity in the indicated HCC cells as measured by 856
luciferase reporter assay. Data are the mean ± SEM from triplicate experiments. *P < 857
0.05. (G) Relative luciferase activities were measured after serially truncated and 858
mutated HDAC6 promoter constructs were cotransfected with pCMV-CAMSAP2 859
plasmid. (H) Chromatin immunoprecipitation assay demonstrating the direct binding 860
of c-Jun to the HDAC6 promoter in HCC cells. (I) Summary of the orthotopic 861
metastatic model. (J) Incidence of lung metastasis and bioluminescence imaging of 862
each group at 10 weeks after orthotopic xenografting with the indicated HCC cells. (K) 863
OS of mice in the different groups. (L) Number of metastatic lung nodules observed 864
in each group. *P < 0.05, **P < 0.01. (M) Representative H&E staining of lung 865
tissues from each group. Scale bars, 500 mm (upper), 500 µm (lower). 866
867
868
869
40
Figure 1 870
871
872
873
874
875
41
Figure 2 876
877
878
879
880
42
Figure 3 881
882
883
884
43
Figure 4 885
886
44
Figure 5 887
888
45
Figure 6 889
890
891
892
46
Figure 7 893
894
47
Figure 8 895
896