HSP27 associates with epithelial-mesenchymal transition, stemness and radioresistance of salivary adenoid cystic carcinoma.

Epithelial-mesenchymal transition (EMT) has been shown to associate with cancer stem cells and radioresistance. However, it is obscure whether EMT itself or specific EMT regulators play causal roles in these properties of salivary adenoid cystic carcinoma (SACC). Here, we exhibited that overexpression of HSP27 drove the migration and invasion, induced EMT, as well as mediated TGF-β1-induced EMT in SACC cells, accompanying the up-regulation of Snail1 and Prrx1. Conversely, HSP27 silencing reduced the migration and invasion and contributed to MET of SACC cells. HSP27 indirectly down-regulates the expression of E-cadherin through activating Snail1 and Prrx1 expressions. Overexpression of Snail1 or Prrx1 restored the migration and invasion in HSP27 knockdown cells. Enforced expression of HSP27 enhanced colony formation, CD133+ /CD44+ population and radioresistance of SACC cell lines. In addition, HSP27 expression was positively associated with radioresistance and poor prognosis of SACC patients as well as with the expression of Prrx1 or Snail1 in SACC tissues. The data confirm an important function for HSP27 in SACC progression through regulating EMT and stemness, and they imply the possible association between EMT and radioresistance of SACC.

Introduction

Epithelial–mesenchymal transition (EMT) is malignant progression in cancer, during which epithelial cells lose epithelial characteristics and gain mesenchymal feature, thereby becoming more invasive and mobile 1. EMT has been shown to be a potent mechanism that enhances the detachment of cells from primary tumours during malignant tumour progression, but it is also implicated in cancer therapy resistance 2. It has been hypothesized that carcinoma cells that have undergone EMT acquire cancer stem cell (CSC) properties and CSC theory suggests that a fraction of cancer cells with self‐renewal and multipotential differentiation abilities are responsible for radioresistance 3, 4. However, it is unclear whether EMT itself or specific EMT regulators play causal roles in these properties.

Salivary adenoid cystic carcinoma (SACC) is the second most common malignancy in salivary gland, accounting for 10–25% of patients 5, 6, 7. SACC composed of luminal epithelial cells, abluminal myoepithelial cells and stromal cells has three histological subtypes including cribriform, tubular and solid. The insidious invasion into adjacent tissue and haematogenous spread to distant organs (lung, bone and liver) is the characteristic of SACC. The 15‐ and 20‐year survival rates are 35–40% and 10%, respectively 8, 9. Thus, elucidating the mechanism underlying disease malignant progression is urgently needed.

The small heat‐shock protein of 27 kDa (HSP27) is an ATP‐independent molecular chaperone induced under heat shock or other chemical stimulus 10. Besides its chaperone function, HSP27 has been reported to be overexpressed in various kinds of human cancers 11, 12, 13 and inhibited cancer cells apoptosis 14. The interaction between HSP27 and STAT3 regulated the HSP27‐mediated apoptosis inhibition during the androgen ablation of prostate cancer cells 15. Overexpression of HSP27 also protected MDA‐MB‐231 breast cancer cells from doxorubicin‐induced apoptosis 16. Moreover, HSP27 increased the migratory and invasive capabilities of breast cancer cells 17, regulated EMT in lung cancer cells 18, have been regarded as an important inducer of EMT in kidney fibrosis 19 and may serve as novel therapeutic targets. However, until now, the role of HSP27 in EMT of SACC cells is still unclear.

In this study, we exhibited that HSP27 induced EMT, mediated TGF‐β1‐induced EMT, as well as enhanced the capabilities of migration and invasion of SACC cells. Overexpression of HSP27 increased colony formation, cancer stem cell markers and radioresistance of SACC cell lines. In addition, HSP27 expression was positively associated with radioresistance and poor prognosis of SACC patients as well as with Prrx1 or Snail1 in SACC tissues. These verified HSP27 as an EMT regulator, conferring to stemness, and Prrx1 and Snail1 had involved in this process. And these might provide a cue to explore the association between EMT and radioresistance of SACC.

Materials and methods

Ethics statement

This study was approved by the Institutional Ethics Committee of Sichuan University with the following reference number: SKLODLL2013A103.

Patients and specimens

The cohort was assembled from patients who were histologically diagnosed with SACC at West China Hospital of Stomatology, Sichuan University, between 1998 and 2002. Exclusion criteria included recurrence at presentation, preoperative radiotherapy, chemotherapy, or hormone therapy, and incomplete medical records. Follow‐up time was calculated from the date of surgery to the 180th month, loss of follow‐up or the date of death, whichever came first (median follow‐up period of 146 months). Finally, 67 of the 70 cases (33 male and 34 female; median age, 52 years; range 20–87) were recruited in this study. The principal clinical and pathologic characteristics of the patient cohort are summarized in Table 1.

Table 1

Clinicopathologic features of the SACC patients and their primary tumours and their association with HSP27 expression (n = 67)

Clinicopathologic features	No. of cases	HSP27 expression		P value
		Negative	Positive
Age at diagnosis, year
≥60	39	15	24	0.621
<60	28	10	18
Gender
Female	33	13	20	0.462
Male	34	12	22
Complaints, month
≥12	41	15	26	0.540
<12	26	10	16
Site
Minor salivary gland	41	14	27	0.338
Major salivary gland	26	11	15
Histological subtype
Tubular/Cribriform	45	21	24	0.021
Solid	22	4	18
Resection margins
Free	51	20	31	0.396
Affected	15	5	11
TNM stage
I–II	32	22	10	0.001
III–IV	35	13	32
Perineural invasion
Yes	40	20	20	0.008
No	27	5	22
Local regional recurrence
Yes	24	5	19	0.033
No	43	20	23
Distant metastasis
Yes	27	3	24	0.000
No	40	22	18

Immunohistochemical staining

Tissue sections (4 μm) were deparaffinized in xylene and rehydrated, and endogenous peroxidase was blocked with 3% H2O2. Antigen retrieval was accomplished by 0.01 mol/l citrate buffer solution (pH 6.0) in a microwave oven at 700 W for 15 min. The sections were washed in PBS for 5 min. twice and then incubated with 5% normal goat serum for 20 min. The slides were incubated with the primary antibody in a humid chamber for 2 hrs at 37°C and then exposed overnight at 4°C. The subsequent steps were performed according to the manufacturer's instructions (streptavidin–peroxidase kit; Zymed Laboratories, San Francisco, CA, USA). The ratio of positive cells was expressed as the percentage of 100 tumour cells counted within 4–6 microscopic fields at ×200 magnification and semiquantitatively graded as follows: negative (0–9%), positive (>10%). Quantification of positive cells was conducted by two independent observers (XH and ZM).

Cell culture and virus infection

SACC cells lines, SACC‐LM and SACC‐83 were obtained from the State Key Laboratory of Oral Disease, Sichuan University. Short pairs of sense and antisense DNA oligonucleotide encoding a sense‐loop‐antisense sequence to HSP27 genes (accession number NM 001540) were synthesized for the validated corresponding siRNAs. The sequence of human sh1‐HSP27 (start 701): upper: GATCCGatccgatgagactgccgccaaTTCAAGAGAttggcggcagtctcatc ggatTTTTTTACGCGTG, lower: AATTCACGCGTAAAAAAatccgatgagactgccgccaaTCTCTTGAAttggcggcagtctcatcggatCG, and the hair sequence is TTCAAGAGA.

For overexpression of HSP27, the full‐length cDNA for human HSP27 was subcloned into the lentiviral vector pLVX‐AC‐GFP‐N1 at the BamHI and XhoI sites. Two vectors were created for study: pLVX‐AC‐GFP‐N1HSP27 and pLVX‐AC‐GFP‐N1.

OGX‐427 treatment

OGX‐427, provided by Biochempartner, Shanghai, China, was dissolved in DMSO to a stock concentration of 50 μm. Cells were incubated for 12 hr with 50 nm OGX‐427.

Ionizing radiation treatment of cells

5 × 105 cells (SACC‐LM and SACC‐83) were exposed to 6‐Gy irradiation for 48 hr with a Cs‐137 irradiator (HWM D‐2000; Siemens AG, Munich, Germany). At 2 days after irradiation, cells were washed with PBS three times and then further incubated in serum‐free media without antibiotics for 36 hr. CM were collected and centrifuged to remove any residual cells, after filtered through a 0.2‐μm syringe filter.

Real‐time RT–PCR

Total RNA was isolated with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and treated with RNase‐free DNase I (Takara, Dalian, China) to avoid genomic DNA contamination. PCR amplification of the cDNA template was carried out using Thunderbird SYBR qPCR mix (TOYOBO, Osaka, Japan) on ABI PRISM 7300 sequence detection system (Applied Biosystems, Foster City, CA, USA). Reactions were run in triplicate, and results were averaged. Each value was normalized to GAPDH as the housekeeping gene to control for variations in the amount of input cDNA. The relative expression level of the genes was calculated using the 2−ΔΔCT method.

Colony formation assay

After colonies were visible, the colonies were fixed by methanol and stained by crystal violet (Sigma‐Aldrich, St. Louis, MO, USA). The percentage of cells that formed into a clone was calculated.

Sphere formation assays

Single cells were plated at 10,000 cells/ml on 6‐well ultra‐low attachment plates (Corning Inc., Corning, NY, USA) in serum‐free DMEM/F12 supplemented with 20 ng/ml bFGF, 20 ng/ml EGF, 4 mg/ml insulin, 4 mg/ml heparin, 1 mg/ml hydrocortisone, 0.4% BSA and B27.

Flow cytometry

CD133‐APC and CD44‐PE (BD Biosciences, San Diego, CA, USA) were added to cell suspension and incubated on ice for 30 min. Cells were washed and resuspended in 500 ml FACS buffer and analysed using a FACS.

Immunofluorescence

Cells were incubated overnight with rabbit anti‐E‐cadherin (1:100) and then incubated with FITC‐conjugated goat anti‐rabbit IgG (1:200; Zhongshan Goldenbridge Biotechnology) at 37°C for 1 hr. Cells were counterstained with 4′, 6‐diamidino‐2‐phenylindole (DAPI; 1 μg/μl) and examined using a fluorescence microscope (Olympus BX51, Tokyo, Japan).

Luciferase assay

Cells were transfected with pE‐cadherin‐luc vector (Clontech, Mountain View, CA, USA) and pRL‐TK Vector (Promega, Madison, WI, USA) using the Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) and then were lysed and assayed using the dual‐luciferase reporter assay system (Promega, Madison, WI, USA).

Invasion assay

In vitro invasion assay was performed using 24‐well Transwell unit with polycarbonate filters (Corning Costar, Cambridge, MA, USA). Triplicate filters were used per condition, and the experiments were repeated three times. The values obtained were calculated by averaging the total number of cells from three filters.

Wound healing assay

Scratch wounds were made in confluent cell monolayers using a pipette tip. Cell migration was recorded in five different microscopic fields, and the number of migrating cells was calculated.

Xenografts in nude mice

The nude mice (female, 6 weeks of age) were obtained from the Laboratory Animal Center of Sichuan University (Chengdu, Sichuan, China). Sixty mice were randomized and divided into 10 groups (control, shRNA, shRNA‐neg, EV and overexpression), six mice each. Lentivirus‐transfected cells with green fluorescent protein were then injected s.c. (5 × 106 cells/200 μl PBS/mouse) on the abdomen of mouse. Tumour size was monitored by measuring diameters using vernier calliper weekly and was calculated as described previously 20. Tumours were harvested and fixed by 4% paraformaldehyde and then embedded by paraffin for immunohistochemistry analyses.

Statistical analysis

All the statistical analyses were performed using SPSS 13.0 (SPSS Inc., Chicago, IL, USA). Statistical analysis was considered to be significant when the probability value is <0.05.

Results

Overexpression of HSP27 induced EMT of SACC cell lines

To evaluate the role and significance of HSP27 in human SACC cells, HSP27 was up‐regulated in HSP27‐overexpressed SACC‐LM (Fig. 1A), which was confirmed by immunoblotting and real‐time PCR. Observation of culture morphology under phase‐contrast and immunofluorescence microscopy revealed that overexpression of HSP27 in SACC cell lines reduced tight primary cell nests and ring‐like structure with intercellular adhesion contact and caused a switch from a cobblestone‐like morphology in mock‐treated cells to a spindled fibroblastic morphology in HSP27‐expressed cells (Fig. 1B). This morphology conversion of EMT was accompanied by a loss of E‐cadherin (Fig. 1C), which prompted us to examine the protein and mRNA expression of EMT relative transcription factors. The data signified that the overexpression of HSP27 significantly increased the expression of mesenchymal markers like Vimentin and N‐cadherin and reduced the expression of E‐cadherin at both protein and mRNA levels. The protein and mRNA levels of Snail1, Slug, Prrx1 and c‐kit were significantly up‐regulated in HSP27‐overexpressing cells, compared with control cells (Fig. 1D and E). As shown in Fig. 1F and G, HSP27‐expressed SACC‐LM cells dramatically enhanced their migratory and invasive behaviours by approximate 2.0‐fold and 3.0‐fold, respectively. And HSP27‐expressed SACC‐LM cells with OGX‐427, HSP27 antisense drug, inhibited the migration and invasion abilities of HSP27‐expressed SACC‐LM cells and restored to the level of control cells. Similar data were obtained in SACC‐83 cells (Fig. 1F and G). These results indicated that HSP27 may be an EMT inducer and promotes the migration and invasion of SACC cells.

Figure 1

Ectopic expression of HSP27 induced an EMT programme in SACC cells. (A) Immunoblotting assessment of the ectopic HSP27 protein expression after transfection in SACC‐LM cells. ß‐Actin loading control was shown. The transcription level of HSP27 overexpression in SACC‐LM cells, relative to GAPDH, was determined by quantitative RT–PCR. Each experiment was repeated three times. Error bars represent the mean ± SD of triplicate experiments (*P < 0.05). (B) Morphologic change in SACC‐LM and SACC‐83 cells expressing HSP27, HSP27+ ‐β1 or empty vector. HSP27+ ‐β1 group was as a positive control. Scale bar, 100 μm. (C) Immunofluorescence staining for the epithelial markers E‐cadherin in HSP27‐overexpressed SACC‐LM and SACC‐83 cells. Scale bar, 100 μm. (D) Immunoblotting analysis of expression of the epithelial marker E‐cadherin, the mesenchymal markers Vimentin and N‐cadherin, and EMT transcription factors (Snail1, Slug, Twist, Prrx1 and c‐kit) in SACC‐LM cells. (E) The mRNA expressions of E‐cadherin, Vimentin, N‐cadherin, Snail1, Slug, Twist, Prrx1 and c‐kit were assessed by real‐time PCR in SACC‐LM cells. The value was first relative to GAPDH and then relative to the control. Each experiment was repeated three times. Error bars represent the mean ± SD of triplicate experiments (*P < 0.05). (F and G) Invasion (F) and migration (G) assays in SACC‐LM and SACC‐83 cells with stably overexpressing HSP27, 50 nm OGX‐427 and 50 nm OGX‐427+ HSP27. Representative images of invaded cells were shown. The means were derived from cell counts of five fields, and each experiment was repeated three times (*P < 0.05).

We further applied a luciferase reporter assay to explore whether HSP27 could transcriptionally adjust the expression of E‐cadherin, Snail1, Slug and Prrx1. The data showed that the relative luciferase activity of Snail1 and Prrx1 was promoted proportionally to the increasing amount of HSP27 in SACC‐LM cells, but E‐cadherin and Slug were not changed (Figure S1). This indicated that E‐cadherin reduction during HSP27‐induced EMT may contribute to the induction of Snail1 and Prrx1.

Silencing HSP27 resulted in mesenchymal‐to‐epithelial transition (MET) in SACC cells

To further investigate the role of HSP27 in EMT formation of SACC cells, we applied HSP27 targeting shRNA (sh1‐HSP27) to dramatically reduce the level of HSP27 of SACC‐LM cell line (Fig. 2A). Knockdown of HSP27 in SACC‐LM cell line reduced the expression of Vimentin and N‐cadherin at the mRNA level and increased the expression epithelial markers E‐cadherin. Moreover, suppression of HSP27 caused EMT‐associated transcription factors Snail1 and Prrx1 expression reduction at the mRNA level (Fig. 2B). These changes were accompanied with a reduction in the migration (~0.8‐fold) and invasion (~0.8‐fold) ability of SACC‐LM cells relative to that of non‐transfected HSP27 and the control (Fig. 2C and D). Further, we stably overexpressed Snail1 or Prrx1 and found that the overexpression of Snail1 or Prrx1 successfully rescued the capacities of migration and invasion in HSP27‐silenced cells (Fig. 2C and D). However, down‐regulation of HSP27 did not cause significant cellular morphologic changes in SACC‐LM cells (Fig. 2E). All in all, the above data indicated that silencing HSP27 might reverse EMT in SACC cells, though no obvious morphologic features.

Figure 2

Silencing HSP27 lead to MET in SACC‐LM cell lines. (A) SACC‐LM cells were transfected with HSP27‐shRNA1, HSP27‐shRNA2 or control. Western blot analysis of HSP27 silencing by HSP27‐shRNA1 and HSP27‐shRNA2 in SACC‐LM cells. ß‐Actin loading control was shown. Transcription levels of HSP27 silencing by HSP27‐shRNA1 and HSP27‐shRNA2 in SACC‐LM cells, relative to GAPDH, were determined by quantitative RT–PCR. Each experiment was repeated three times. Error bars represent the mean ± SD of triplicate experiments (*P < 0.05). (B) The mRNA expression of E‐cadherin, Vimentin, N‐cadherin, Twist, Snail1, Slug, c‐kit and Prrx1 was assessed by real‐time PCR in HSP27 silencing SACC‐LM cells. The value was first relative to GAPDH and then relative to the control. Error bars represent the mean±SD of triplicate experiments *P < 0.05). (C and D) Invasion (C) and migration (D) assays in SACC‐LM cells. Representative images of migrated and invaded cells were shown under inverted microscopy. Each experiment was repeated three times. Error bars represent the mean ± SD of triplicate experiments. (E) Representative morphologic feature of SACC‐LM cells with sh‐HSP27 or empty vector. Scale bar, 100 μm.

Overexpression of HSP27 enhanced colony formation, sphere formation and cancer stem cell markers of SACC cell lines

EMT has seemed to be a condition to induce cancer stem cells (CSCs) and conferred a cancer stem cell phenotype 21. Hence, the sphere‐forming assay and fluorescence activating cell sorter (FACS) were applied to examine the self‐renewal of CSCs to underline the relationship between EMT and stemness in HSP27‐expressing SACC cell lines. The data exhibited that HSP27‐expressing SACC‐LM cells heightened the colony formation (Fig. 3A) compared with the untreated group and negative group. SACC cells treated with HSP27‐expressing vector formed more and bigger spheres than the counterpart (Fig. 3B), indicating overexpression of HSP27 lifting self‐renewal of SACC cells in vitro. Importantly, we observed that the HSP27‐expressing SACC‐LM cells enhanced the CD133+/CD44+ stem cell population compared with the control (Fig. 3C). The sphere‐forming assay and FACS were quite repeatable in SACC‐83 cell line (Figure S2). To further verify whether overexpression of HSP27 up‐regulated the protein expressions of CSC markers, we performed a Western blot analysis and RT–PCR in SACC‐LM and SACC‐83 cell lines with or without HSP27‐expressing treatment. We found that HSP27 up‐regulation consistently enhanced the protein and mRNA levels of stem cell markers (CD133, CD44, Nanog, Sox2 and OCT4) (Fig. 3D and E) compared with the negative counterpart. These results supported the hypothesis that HSP27 overexpression can effectively raise the self‐renewal capacity of SACC cells in vitro.

Figure 3

Overexpression of HSP27 enhanced colony formation and cancer stem cell markers in SACC‐LM cell line. (A) Overexpression of HSP27 in SACC‐LM cell line increased anchor‐dependent colony formation; (B) Overexpression of HSP27 in SACC‐LM cell line promoted sphere formation; scale bar = 100 μm; (C) FACS analysis of cell‐surface markers CD133 and CD44 in SACC‐LM cells expressing HSP27 or empty vector. Percentages of mean CD133+/CD44+ subpopulation ±SD based on triplicate experiments were indicated. (D) Western blot analysis revealed that the protein levels of CD133, CD44, Nanog, Sox2 and OCT4 in SACC‐LM cells were enhanced in different degrees after HSP27 up‐regulation. β‐Actin was regarded as a loading control. Representative of three independent experiments was shown. (E) RT–PCR analysis revealed that the mRNA levels of CD133, CD44, Nanog, Sox2 and OCT4 in SACC‐LM cells were enhanced in different degrees after HSP27 up‐regulation, relative to GAPDH. Error bars represent the mean ± SD of triplicate experiments (*P < 0.05).

HSP27‐mediated TGF‐β1 induces EMT in SACC cell lines

TGF‐β1, one of the major EMT inducers, has been shown to an inducer of EMT during embryonic development, the pathogenesis of fibrotic disorders and cancer progression in various kinds of cancer including breast carcinoma, lung cancer and head and neck squamous cell carcinoma 22. Here, we first sought to determine whether TGF‐β1 induces EMT in SACC‐LM cell line. Real‐time PCR demonstrated that the SACC‐LM cells exposed to TGF‐β1 exhibited a significant down‐regulation of E‐cadherin; meanwhile, the mesenchymal markers Vimentin and Fibronectin were dramatically up‐regulated (Fig. 4A). Western blot also revealed that the protein expression E‐cadherin reduced, and Vimentin and Fibronectin increased in TGF‐β1‐treated SACC SACC‐LM cells (Fig. 4B). And TGF‐β1 treatment of SACC‐LM cells also added the expression of HSP27, Prrx1 and Snail1 (Fig. 4B). The data indicated that TGF‐β1 could induce EMT process of SACC‐LM cells, and HSP27, Prrx1 and Snail1 pathway were activated in TGF‐β1‐induced EMT in SACC‐LM cells.

Figure 4

HSP27‐mediated TGF‐β1 induced an EMT programme in SACC‐LM cell line. (A) The mRNA level of E‐cadherin, Fibronectin and Vimentin was assessed by real‐time PCR in the SACC‐LM cells exposed to TGF‐β1. And the mRNA expression of known EMT inducers (HSP27, Prrx1 and Snail1), relative to GAPDH and the control, was also examined. Error bars represent the mean ± SD of triplicate experiments (*P < 0.05). (B) Immunoblotting analysis of protein expression of the epithelial markers E‐cadherin, and the mesenchymal markers Fibronectin and Vimentin, as well as HSP27, Prrx1 and Snail1 in the SACC‐LM cells exposed to TGF‐β1. (C), Morphologic change in HSP27 knockdown SACC‐LM cells with the treatment of TGF‐β1. Scale bar, 100 μm. (D) Immunofluorescence staining for the epithelial markers E‐cadherin in HSP27 knockdown SACC‐LM cells with the treatment of TGF‐β1. Scale bar, 100 μm. (E) The mRNA of E‐cadherin, Fibronectin, Vimentin, HSP27, Prrx1 and Snail1, relative to GAPDH, was assessed by real‐time PCR in HSP27 silencing SACC‐LM cells treated with and without TGF‐β1. Error bars represent the mean ± SD of triplicate experiments (*P < 0.05). And the expression of their proteins was carried out by immunoblotting analysis. ß‐Actin loading control was shown.

To investigate the potential role of HSP27 in TGF‐β1‐induced EMT, we analysed epithelial and mesenchymal markers as well as Prrx1 and Snail1 expressions in TGF‐β1‐stimulated SACC‐LM cells with HSP27 stable knockdown. TGF‐β1 treatment of SACC‐LM cells induced a morphological change to a fibroblastic phenotype in control, but not in HSP27 knockdown cells (Figs 2E and 4C). This effect on cell morphology correlated with changes in EMT markers. In control cells, TGF‐β1 treatment added Vimentin and Fibronectin expressions, Prrx1 and Snail1 levels, and reduced E‐cadherin level (Fig. 4A). In HSP27 knockdown SACC‐LM cells, E‐cadherin level increased with TGF‐β1 treatment (Fig. 4D), and the protein and mRNA levels of Vimentin and Fibronectin and Prrx1 and Snail1 levels were markedly reduced almost to undetectable status, compared to control cells in both the presence and absence of TGF‐β1 (Fig. 4E). These data strongly suggest that HSP27 is required for both TGF‐β‐dependent and TGF‐β‐independent EMT in SACC cells.

HSP27 contributed to tumour progression and malignancy in mouse model of SACC

Further, we wanted to determine whether the expression of HSP27 affected tumour progression and malignancy in xenograft using SACC‐LM cells transfected, respectively, with shRNA‐neg, shRNA‐HSP27, EV, overexpression‐HSP27 and control. As shown in Fig. 5A and B, no difference in tumour volume was observed among the shRNA‐HSP27, the overexpression‐HSP27 and control groups in the first three weeks, but the growth of tumour significantly slowed down in the shRNA‐HSP27 compared with the shRNA‐neg group and the growth of tumour obviously speed up in the HSP27 compared with the EV group since the 4th week. Moreover, the volume of the overexpression‐HSP27‐treated tumour was significantly bigger than that of the shRNA‐HSP27‐treated tumour since the 4th week, and with the growth of tumour and time extension, this difference between shRNA‐HSP27 and the overexpression‐HSP27 was gradually obvious. Although it was still significantly slower in the shRNA‐HSP27‐treated tumour compared with the shRNA‐neg‐treated tumour, the volume of the tumour was a tendency to speed up after the 5th week. Additionally, the shRNA‐HSP27‐treated tumour kept a much slower growth and the overexpression‐HSP27‐treated tumour kept a much faster growth, compared with other four groups during the whole period (Fig. 5A and B).

Figure 5

HSP27 contributed to tumour progression and malignancy in mouse model of SACC (A, B), Stable HSP27‐shRNA and overexpression‐HSP27 SACC‐LM cells were subcutaneously injected into nude mice. Individual tumour volume was measured at the 8th week after injection, and growth curve of xenograft tumours was shown. (C) CTS in the circulation of mice using GFP–PCR analysis. The data showed that the SACC‐GFP cells in the overexpression‐HSP27 group presented in the circulation of the mice at the 14th day, and other groups had no CTS presentation. (D, E), The protein (D) and mRNA (E) levels of Snail1 and Prrx1 examined in tumour tissues by Western blot and RT–PCR. The results showed that the protein and mRNA levels of Snail1 and Prrx1 were quite higher in overexpressing‐HSP27 group than the control group (*P < 0.05). ß‐Actin loading control was shown in Western blotting, and GAPDH was in RT–PCR. And there was no difference in the protein and mRNA levels of Snail1 and Prrx1 between shRNA1‐HSP27 group and EV and control group (P > 0.05). Each experiment was repeated three times. Error bars represent the mean ± SD of triplicate experiments.

Then, we applied GFP–PCR analysis to examine the time course of metastasized SACC cells intravasated into the circulation of mice. We found SACC‐GFP circulation tumour cells (CTS) in nucleated cells from peripheral blood of over 80% mice at 14th day. The SACC‐GFP cells in the overexpression‐HSP27 group presented in the circulation of the mice at the 14th day and other groups had no CTS presentation (Fig. 5C). These results indicated that HSP27 could enhance the dissemination of SACC in vivo.

To further verify whether HSP27 can regulate the metastasis of SACC on mice model, we determined the protein and mRNA expression levels of Snail1 and Prrx1 by Western blot and RT–PCR in tumour tissues of each group. The results showed that the protein and mRNA levels of Snail1 and Prrx1 were quite higher in overexpressing‐HSP27 group than the control group (P < 0.001, Fig. 5D). And there was no difference in the protein and mRNA levels of Snail1 and Prrx1 between shRNA‐HSP27 and EV group (P > 0.05, Fig. 5E).

Overexpression of HSP27 enhanced radioresistance of SACC cell lines

Radiation can deliver biologic information by intratumoral communication through direct cell‐to‐cell contacts and paracrine signalling. We sought to investigate whether conditioned media(CM)derived from irradiated cells can affect SACC cells. We first measured the proliferation capacity of SACC cell lines (SACC‐83 and SACC‐LM cells) after irradiation by a colony‐forming assay. The proliferation capacity of SACC cells decreased along with the increasement dose of irradiation (0, 2, 4, 6, 8, 10 Gy), respectively (Fig. 6A). And there was no difference in cell proliferation capacity among these cell lines in response to irradiation (P > 0.05). Then, CMs from irradiated and non‐irradiated SACC‐LM cells were prepared. Invasion assays showed that irradiated‐derived CM significantly increased the invasion of SACC cells compared with non‐irradiated‐derived CM (Fig. 6B). In accordance with the invasion results, cell migration was remarkably promoted by irradiated‐derived CM compared with non‐irradiated‐derived CM (Fig. 6C). These results suggest that the radiation microenvironment can modulate the migration and invasion capabilities of SACC cells.

Figure 6

HSP27 promoted the migration, invasion and irradiation of SACC cells. (A) Irradiation dose response of cell survival in SACC cell lines (SACC‐83 and SACC‐LM cells). The proliferation capacity of SACC cells decreased along with the increasement dose of irradiation. No difference in cell proliferation capacity among these cell lines was shown in response to irradiation (P > 0.05). (B) Invasion assay in SACC cells with CM from irradiated and non‐irradiated SAC cells. Left. representative images of invaded cells. Right, quantitative analysis of the cell number. Data represent three experiments performed in triplicate.*P < 0.05. (C) Migration assay in SACC cells with CM from irradiated and non‐irradiated SACC cells. Left, representative images of migrated cells. Right, quantitative analysis of the cell number. Data represent three experiments performed in triplicate.*P < 0.05. (D) Relative expression levels of HSP27 in SACC cell lines (SACC‐83 and SACC‐LM cells) 0, 1, 4, 8, 12 or 24 hr after 6 Gy irradiation. Data were presented as means ± SD of three independent experiments. *P < 0.001 compared to non‐irradiated controls (0 hr group). (E) Secreted extracellular HSP27 in CM obtained from these cells (SACC‐83 and SACC‐LM cells) was analysed by Western blotting. ß‐Actin loading control was shown. (F) Quantitative analysis of the cell number using invasion assay in SACC‐LM cells with CM from treatment of sh‐HSP27, HSP27 or OGX427 prior to irradiation. Data represent three experiments performed in triplicate. *P < 0.05. (G) Wound healing assay in SACC‐LM and SACC‐83 cells with CM from treatment of sh‐HSP27, HSP27 or OGX427 prior to irradiation. Migration quantitative analysis of the cell number. Data represent three experiments performed in triplicate. *P < 0.05.

To identify the involvement of HSP27 of SACC cells in radioresistance, migration and invasion in response to radiation, we measured the expression of HSP27 in response to radiation in SACC cell lines. Expression of HSP27 increased in irradiated SACC‐LM and SACC‐83 cells (Fig. 6D), and secreted extracellular HSP27 was detected in the media obtained from these cells (Fig. 6E). To study the function of HSP27 derived from irradiated cells in contributing to the migration and invasion of SACC cells, SACC cells were cultured with CM obtained from SACC‐LM cells treated with HSP27‐specific shRNA or HSP27 vector prior to irradiation and control media under radiation exposure. As shown in Fig. 6F, knockdown of HSP27 significantly decreased the invasion of SACC‐LM cells compared with control media. The migration of cells treated with HSP27 shRNA was also remarkably inhibited (Fig. 6G). These results were recovered by treatment of HSP27‐overexpressed vector. In addition, treatment of SACC‐LM cells with OGX‐427, a HSP27 inhibitor prior to irradiation, resulted in the reduced migration and invasion of SACC‐LM cells, levels similar to that of HSP27 shRNA under radiation exposure. These results indicated that HSP27 promoted the migration, invasion and radioresistance of SACC cell lines.

Expression of HSP27 associated with radioresistance and the poor prognosis of SACC patients

To further evaluate the clinical relevance of HSP27 in SACC patients, we have collected the major pathological features in 67 SACC patients, including site of the primary tumour, histological grade, local invasiveness and metastasis (Table 1) and carried out immunohistochemical staining of HSP27. Representative immunohistochemical images are shown in Fig. 6A. We found that the cytoplasmic or nuclear staining of HSP27 was detected in 42 of 67(62.69%) SACC specimens. The positive staining of HSP27 was significantly associated with histological grade (P = 0.021), TNM stage (P = 0.001), neural invasion (P = 0.008), local recurrence (P = 0.033) and metastasis (P = 0.000) of SACC patients. The Kaplan–Meier survival analysis revealed a significantly worse overall survival for patients who had positive staining of HSP27 compared those negative HSP27 staining (P = 0.049, Fig. 6B). Simultaneously, we detected several EMT‐related proteins in SACC specimens, including Snail1, Slug, c‐kit, Prrx1, which are potentially relevant to EMT and cancer metastasis (Fig. 7A). The positive staining of Snail1, Slug, c‐kit, Prrx1 was obviously associated with the poor prognosis of SACC patients (P = 0.0142, P = 0.333, P = 0.366, P = 0.0334, respectively, Fig. 7B). The correlation analysis signified that there was a close positive correlation between the expression of Prrx1 and HSP27, Snail1 and HSP27 in SACC patients, respectively (P < 0.0001, Fig. 7C). Taken together, these data strongly suggest that overexpression of HSP27 may predict a worse survival of patients, and played a critical role in the EMT process of human SACC.

Figure 7

Expression of HSP27 associated with the radioresistance of SACC patients. (A) Representative images of immunohistochemical staining for HSP27, Snail1, Slug, c‐kit, Prrx1 in serial sections of SACC patients. Scale bar, 50 μm. (B) Kaplan–Meier survival analysis in patients with SACC. Overexpression of HSP27, Snail1, Slug, c‐kit, Prrx1 in SACC was associated with a shorter overall survival in the respective group. (C) Relative level of HSP27 according to the expression levels of Snail1 and Prrx1 expressions of SACC by immunohistochemical staining and counting. Error bars represent the mean ± SD of triplicate experiments. The data showed that expression of HSP27 associated with the expression of Snail1 and Prrx1 in SACC patients. (D) Relative level of HSP27 assessed by the positive cell counting result of immunohistochemical staining, RT–PCR and Western blot in SACC with radiotherapy and without radiotherapy. Error bars represent the mean ± SD of triplicate experiments. The data showed that expression of HSP27 in SACC patients after radiotherapy was higher than those without radiotherapy.

Then, to further investigate the protein expression of HSP27 in post‐radiotherapy recurrence SACC, we detected the protein and mRNA levels of HSP27 in 10 recurrence cases after radiotherapy and five SACC cases without radiotherapy. These patients came from Department of Oral Pathology, West China Hospital of Stomatology, Sichuan University, between 2014 and 2015. The age of patients ranged from 55 to 65, and the primary tumour site included five parotid glands, four sublingual glands, three submandibular glands, one palatine gland, one buccal and one mouth floor. There were 10 tubular or cribriform cases and five solid patients. The positive cell counting result of HSP27 immunohistochemical staining, RT–PCR assay and Western blot showed a significant increase in HSP27 expression in recurrent SACC after radiotherapy (Fig. 7D). These indicate that there may be relation between HSP27 and radioresistance in SACC patients.

Discussion

Identifying the determinants of radioresistance to obtain effective radiosensitizers of cancer is a key gating step to overcome the intrinsic and therapy‐induced radioresistance of cancer cells 23, 24. Increasing evidence implicated EMT and cancer stem cells in the acquisition of radioresistance and drug resistance 25, 26. Here, our data confirm an important function for HSP27 in SACC progression through regulating EMT and stemness, and there might be relationship between EMT and radioresistance of SACC.

First, we demonstrated that enhanced expression of HSP27 up‐regulated the protein and mRNA expressions of some EMT‐inducing transcription factors, and silencing HSP27 down‐regulated several known regulators of the EMT programme, notably the Snail1 and Prrx1. Loss of E‐cadherin is considered a hallmark of EMT 27, 28, 29. Our results confirmed that HSP27 may indirectly down‐regulate the transcription of E‐cadherin, which was supported by the reports that EMT inducers such as Snail1 and Slug could transcriptionally block E‐cadherin expression either directly or indirectly 30. Moreover, this study defined that HSP27 indirectly down‐regulated E‐cadherin through activating Snail1 and Prrx1, and the overexpression of Snail1 or Prrx1 successfully rescued the capacities of migration and invasion in HSP27‐silenced cells, suggesting Snail1 and Prrx1 may be a molecular target in HSP27 inducing EMT. These data indicated that HSP27 contributed to EMT of SACC through Snail1 and Prrx1, which was in line with the following previous reports. In prostate cancer, HSP27 (HSPB1) drove EMT and was an important regulator of IL‐6‐dependent and IL‐6‐independent EMT 31. In idiopathic pulmonary fibrosis, HSP27 stabilized snail and induced EMT, and HSP27 knockdown lead to Snail proteasomal degradation 32. In breast cancer, HSP27 regulated the EMT process and NF‐κB activity to contribute to the maintenance of CSC 33. Then, we showed that the autocrine TGF‐β1 signalling was activated in HSP27‐induced EMT, TGF‐β1 stimulation‐induced HSP27 expression and silencing HSP27 blocked TGF‐β1‐induced EMT in SACC cells. The data signified that HSP27 has also involved in and required for TGF‐β1‐induced EMT. These results suggest that HSP27 might function as a trigger for EMT to drive SACC progression.

CD133 or CD44 has been known to be valued as an important CSC or CSC‐like marker in human SACC 34. This study has found that ectopic expression of HSP27 in SACC cells added the CD133+/CD44+ subpopulation and increased the mammosphere forming ability, a property of the stem cells. The data were in line with the notion that EMT process can promote non‐cancer stem cells to acquire CSC‐like properties 35. This sustained that HSP27 was an EMT inducer and contributed to the formation of cancer stem‐like cells. Further, we showed the increase in HSP27 protein in SACC in correlation with the aggressive phenotype of SACC, as indicated by large tumour burden, solid histological type, perineural invasion, local recurrence and distant metastasis. Importantly, remarkable associations were found between HSP27 and Prrx1 and Snail1 in human SACC tissue. The high expression of HSP27 in SACC tissues with post‐radiotherapy was in agreement with the phenotype of CSC, thereby indicating that HSP27 may be associated with CSC of SACC. This correlation was further validated by direct in vitro overexpression of HSP27, which indicated that enforced HSP27 promoted the function and the self‐renewal markers of CSC in SACC. This finding is in agreement with several reports, which stated that in other cancers (breast, prostate and ovarian cancer), HSP27 was a marker that was applied to determine negative prognosis 36, 37, 38.

Recently, studies have demonstrated that HSP27 involved in radioresistance in nasopharyngeal cancer (NPC) cells 34. Abrogation of HSP27 function may be a candidate target for overcoming radioresistance and chemoresistance 39. In contrast, in human breast epithelial cells MCF10A cells, neither one‐time nor repeated single or combined plus radio frequency radiation exposure significantly altered HSP27 and ERK1/2 phosphorylations 40. In this article, we showed that radiation enhanced HSP27 levels in SACC cells. Depletion of HSP27 rescued invasion and metastasis of SACC cells. Hence, we suggest that HSP27 may involve in radioresistance of SACC cells. Due to the variations among different types of tumours, there may be tissue‐specific mechanisms of HSP27 regulation in response to radiation.

The differences are statistically significant; however, CM from HSP27‐specific shRNA before irradiation had a moderate effect on the inhibition of cell migration and invasion, and the effects of HSP27 vector on tumour EMT and stemness are also modest. The reason is that there might be other factors, other than HSP27, which are also playing major roles in these processes. Whether the other heat‐shock proteins, such as HSP70 and HSP90, induced EMT and contributed to cell radioresistance, the invasion and migration needs to be investigated in the further studies.

In conclusion, our findings confirm that HSP27 can confer EMT and stemness of SACC cells and play a critical role in the malignant progression of SACC. Additionally, overexpression of HSP27 has been observed in human SACC with the radiation therapy, which indicated that the expression of HSP27 may associate with the radioresistance of SACC. Much should be done to explore the mechanism underlying association between EMT and radioresistance of SACC in their future.

Conflict of interest

The authors declare no conflicts of interest.

Figure S1 HSP27 regulated E‐cadherin transcription through Snail1 and Prrx1.

Click here for additional data file.

Figure S2 Overexpression of HSP27 enhanced colony formation and cancer stem cell markers in SACC‐83 cell line.

Click here for additional data file.

Click here for additional data file.