HSP27-Mediated Extracellular and Intracellular Signaling Pathways Synergistically Confer Chemoresistance in Squamous Cell Carcinoma of Tongue

Oral cavity carcinoma (OCC) is a major cause of morbidity and mortality in patients with head and neck cancer (1). Despite the progress of treatment options, the outcomes remain poor in patients with advanced oral cavity carcinoma (2). Squamous cell carcinoma of tongue (SCCT) is the most common type of the oral cavity carcinoma. In the United States, approximately 12,060 new cases and 2,030 deaths from SCCT were estimated in 2011 (3). The majority of patients with SCCT are treated with surgery, radiotherapy, and chemotherapy. In China, Pingyangmycin (PYM)- and/or Cisplatin (cDDP)-based chemotherapy demonstrates favorable outcomes, but often chemoresistance develops later on and results in failure of this therapy. Both basic researches and clinical studies demonstrate that chemoinsensitivity of SCCTs is correlated with more aggressive cancer behavior and a worse clinical outcome (4). Thus, there is an urgent need to fully understand the molecular mechanisms for chemoresistance of SCCT cells and to identify new therapeutic targets to improve the efficacy of chemotherapy.

Several mechanisms that mediate chemoresistance in cancer have been reported, including activation of antiapoptotic cellular defense, increased DNA repair, and induction of drug-detoxifying mechanisms (5). However, there are limited reports regarding the mechanisms for tongue cancer chemoresistance (6–8). The precise mechanisms of chemoresistance in SCCT remain elusive. In this study, we compared the protein expression profiles between established multidrug-resistant (MDR) SCCT cell line SCC-15/PYM and its parental cell line SCC-15, and identified that increased HSP27 level contributes to chemoresistance and predicts poor clinical outcome. Moreover, we demonstrated that both extracellular and intracellular HSP27 synergistically confers chemoresistance via enhancing TLR5/NF-κB signaling and interacting with BAX and BIM to inhibit the mitochondrial apoptotic pathway, respectively.

This study was reviewed and approved by the Ethnics Committees of the Central South University and Affiliated Xiangya Stomatological Hospital (Changsha, Hunan, China), and Guangzhou Medical University and Affiliated Cancer Hospital (Guangzhou, Guangdong, China). The study was conducted in accordance with the Declaration of Helsinki.

The human SCCT cell lines SCC-15, SCC-25, SCC-9, and CAL-27 were obtained commercially from the American Type Culture Collection at the beginning of this study. The cell lines were last authenticated by short tandem repeat DNA fingerprinting on May 28, 2014. Cells were cultured in RPMI-1640 medium (Gibco) supplemented with 10% FBS (Gibco) and incubated at 37°C in a humidified incubator containing 5% CO₂. The stable MDR-SCCT cell line SCC-15/PYM was established in our lab. To maintain the chemoresistance phenotype, SCC-15/PYM cells were cultured in the medium with 0.5 mg/L PYM. For establishing stable transfectants with knockdown or overexpression, cell lines were transfected with psi-LVRU6GP vectors with HSP27 shRNAs (target sequence for sh-1#: 5′-CCTCAAACGGGTCATTGCCATTAAT-3′, sh-2#: 5′-CATTGCCATTAATAGAGACCTCAAA-3′, sh-3#: 5′-GCGTGTCCCTGGATGTCAACCACTT-3′, sh-4#: 5′-TTCCGCGACTGGTACCCGCAT-3′) or pLEX-MCS vectors with HSP27-overexpressing constructs, and stable clones were selected using puromycin. In some experiments, related cells were cultured in medium with 10 μg/mL anti-HSP27 antibody (Enzo Life Sciences) or 1 μg/mL recombinant human protein (rhHSP27; R&D Systems). Primary tongue cancer tissue samples were obtained from 84 patients at the Affiliated Caner Hospital of Guangzhou Medical University, and 81 patients at the Xiangya Stomatological Hospital and School of Stomatology, Central South University. All samples were collected with informed consent from patients, and all related procedures were performed with the approval of the Internal Review and Ethics Boards of the two hospitals.

To compare different protein expression profiles between the drug-resistant SCC-15/PYM cells and SCC-15 cells, we first did two-dimensional gel electrophoresis analysis of total lysates prepared from these two cell lines. Three well-reproducible 2-DE gels were performed. Following coomassie blue staining, the gels were subjected to scan with MagicScan software on Imagescanner, and PDQuest analysis software was used for spot intensity calibration. Protein spots were subjected to MALDI-TOF-MS analysis and ESI-Q-MS analysis to identify differential proteins. Protein spots that had consistent differences (>2-fold) between the two cell lines in triplicate experiments were chosen as differential protein spots.

Caspase-3 activity was measured by a caspase-3 activity kit (Beyotime) according to the manufacturer's instruction. To evaluate the activity of caspase-3, cell lysates were prepared after related treatments. Assays were performed on 96-well plates by incubating 10 μL of cell lysate per sample in 80 μL of reaction buffer and 10 μL of caspase-3 substrate (Ac-DEVD-pNA, 2 mm). After incubation at 37°C for 4 hours, the absorbance at 405 nm was recorded for each well on the BioTek Synergy 2, and the relative caspase-3 activity was calculated.

Xenograft tumors were generated by subcutaneous injection of related cell lines at 1 × 10⁶ cells in 200 μL, respectively, into the armpit of 4- to 6-week-old female Balb/C athymic nude mice. All mice were housed and maintained under specific pathogen-free conditions at 27°C with 12:12-hour light:dark cycle and fed with sterilized food and water, and all animal experiments were approved by the Experimental Animal Ethics Committee of Guangzhou medical University and performed in accordance with institutional guidelines. Fifteen days later, the mice were grouped randomly and injected intraperitoneally with PYM (30 mg/kg) or Cisplatin (cis-diamminedichloridoplatinum, cDDP; 3 mg/kg). The treatment was administered every 3 days for 10 cycles. Tumor growth was examined every 3 days during the animal experiment. The mice were sacrificed with 120 μL 10% hydral (Sinopharm Chemical Reagent Co., Ltd) at the experimental endpoint, and the tumors were harvested and weighed.

MTS assay, Hoechst staining, Western blot, ELISA assay, immunohistochemistry, NF-κB activity assay, real-time PCR, coimmunoprecipitation (co-IP), and immunofluorescence were used in the study. The methods used are described in the Supplementary Experimental Procedures.

Statistical analyses were performed using SPSS version 16.0 (SPSS) and GraphPad Prism 6. Data are presented as mean ± SD. The difference between two groups for statistical significance was analyzed using the Student t test. To compare multiple groups, one-way ANOVA analysis was used. Pearson correlation analysis was performed to determine the correlation between two variables. The Pearson χ² test was used to analyze the clinical variables. Survival curves were plotted using the Kaplan–Meier method and compared using the log-rank test. A P value < 0.05 was considered statistically significant.

In order to investigate the mechanisms for chemoresistance in tongue cancer, we established an MDR human SCCT cell line SCC-15/PYM derived from SCC-15 by PYM induction. Here, we firstly confirmed the chemoresistant characteristics of SCC-15/PYM cells. After treatment with different concentrations of PYM or cDDP, the dose-response curves for cells were plotted and IC₅₀ values were calculated. SCC-15/PYM cells were significantly more resistant to chemotherapy-induced cytotoxicity as compared with the parental cells (Fig. 1A and Supplementary Fig. S1A). Because of the wide use of PYM and cDDP in chemotherapy for SCCT, we selected PYM and cDDP for our further experiments. Consistently, SCC-15/PYM cells possessed enhanced antiapoptotic ability in response to chemotherapy measured by Hoechest staining (Supplementary Fig. S1B). Moreover, chemotherapy significantly induced caspase-3 activation in SCC-15 cells, but caspase-3 activation was attenuated in SCC-15/PYM cells upon chemotherapy (Fig. 1B).

We further confirmed chemoresistance of SCC-15/PYM cells in vivo using a xenograft mouse model. We separately injected the SCC-15 and SCC-15/PYM cells subcutaneously into the armpit of athymic mice. After 15 days, mice were treated with PBS (Control), PYM, or cDDP every 3 days for 10 cycles. At the experimental end point, the mice were euthanized. The tumors were excised and weighted. We found that the tumors formed by SCC-15 and SCC-15/PYM cells grew at a similar rate. Chemotherapy persistently inhibited the growth of tumors formed by SCC-15 cells, but tumors formed by SCC-15/PYM cells showed only a short growth delay and then started to regrow (Fig. 1C and Supplementary Fig. S1C). Consistently, chemotherapy induced more caspase-3 activation (cleaved caspase 3) in tumors formed by SCC-15 cells compared with tumors formed by SCC-15/PYM cells in vivo (Fig. 1D). Together with the aforementioned in vitro study, these data verified the stable chemoresistant characteristics of SCC-15/PYM cells.

In order to identify key molecules involved into the chemoresistance, we compared the protein profiles between SCC-15/PYM and SCC-15 cell lines using the two-dimensional gel proteomic approach (Supplementary Fig. S1D and Supplementary Table S1). Among the differentially expressed proteins, HSP27, Alpha-enolase (Enolase-1), and Lamin-A/C (LAMN) were selected to be further validated (Supplementary Fig. S1E). We found that protein level of HSP27 is significantly higher in SCC-15/PYM cell line, compared with other SCCT cell lines (Fig. 1E). Because HSP27 is a secreted protein that is suitable to serve as a biomarker, we selected HSP27 to explore its roles and mechanisms for chemoresistance. We further determined the protein level of HSP27 in cell culture medium (CM) of SCCT cell lines. The ELISA result confirmed the significantly higher HSP27 protein level in CM of SCC-15/PYM cell line, as compared with the CM of the SCC-15 cell line (Fig. 1E). As a negative control, β-actin protein is nondetectable in the CM of SCCT cell lines, indicating that the HSP27 protein detected in the cell CM was extracellular protein that was secreted from the cells (Supplementary Fig. S1F). Taken together, these results verified the significant changes in HSP27 protein express in both its intracellular and extracellular levels in chemoresistant SCCT cells.

Given the increased HSP27 in chemoresistant SCCT cells, we then investigated whether the upregulation of both extracellular and intracellular HSP27 play a causal effect on the chemoresistance. We found that HSP27 knockdown with shRNA significantly enhanced the sensitivity of SCC-15/PYM cells to PYM and cDDP in vitro (Fig. 2A and Supplementary Fig. S2A). Inversely, the chemosensitivity of SCC-15 cells was dramatically decreased after HSP27 was overexpressed (Fig. 2B and Supplementary Fig. S2B). This effect of HSP27 on chemoresistance was also validated in another SCCT cell line SCC-25. Similarly, HSP27 overexpression (Supplementary Fig. S2C) also enhanced chemoresistance in SCC-25 cells (Fig. 2B).

We further validated the effect of HSP27 on chemosensitivity in the xenograft tumor models by subcutaneously injecting athymic mice with SCC-15/PYM cells with HSP27 knockdown, SCC-15 and SCC-25 cell lines with HSP27 overexpression or their related control cells. After 15 days, these mice were treated with PBS (Control), PYM, or cDDP every 3 days for 10 cycles. We found that HSP27 overexpression did not significantly influence tumor growth, but markedly affected the chemosensitivity in vivo. The tumors formed by control cells showed a short initial response to chemotherapy prior to regrowing, but chemotherapy led to sustained growth inhibition of tumors formed by SCC-15/PYM cells with HSP27 knockdown (Fig. 2D). The volume and weight of the tumors formed by SCC-15/PYM cells with HSP27 knockdown were decreased to significantly greater extent than that of control xenografts, in response to PYM and cDDP treatment (Fig. 2D and Supplementary Fig. S2C). On the other hand, the tumor volume and weight of xenografts formed by SCC-15 and SCC-25 cells with HSP27 overexpression showed an inverse effect upon chemotherapy (Fig. 2E and Supplementary Fig. S2D).

We further sought to assess the role of extracellular HSP27 in the cells with altered HSP27 expression. We observed that HSP27 knockdown also reduced HSP27 protein level in CM of SCC-15/PYM cells (Supplementary Fig. S2A), whereas HSP27 overexpression increased HSP27 protein level in CM of SCC-15 and SCC-25 cell lines (Supplementary Fig. S2B). To investigate whether extracellular HSP27 was also involved in chemoresistance, anti-HSP27 antibody was used to block extracellular HSP27 function, whereas recombinant human protein (rhHSP27) was used to stimulate SCCT cells. We found that anti-HSP27 antibody treatment also sensitized SCC-15/PYM cells to chemotherapy (Fig. 2A), but rhHSP27 treatment enhanced chemoresistance of SCC-15 (Fig. 2B) and SCC-25 (Fig. 2C) cell lines. In summary, these lines of evidence strongly suggest that both extracellular and intracellular HSP27 mediate chemoresistance in SCCT cells.

In order to elucidate the detailed molecular mechanism mediating the chemoresistance effect of HSP27, we examined NF-κB transactivation in a series of SCCT cells lines using NF-κB reporter assay. We found that NF-κB transactivation activity in SCC-15/PYM cells was significantly higher than that in the other SCCT cells lines (Fig. 3A). The expression of NF-κB target genes such as survivin and IL6 was also significantly upregulated in SCC-15/PYM cells (Fig. 3A). Moreover, compared with other SCCT cell lines, p65 nuclear translocation and p-IκBα level were also increased in SCC-15/PYM cells, but the total IκBα level was decreased (Fig. 3A). These data indicated that NF-κB signaling was markedly activated in SCC-15/PYM cells. As expected, HSP27 knockdown significantly repressed NF-κB transactivation, as reflected by decreased expression of NF-κB target genes survivin and IL6, decreased p65 nuclear translocation, decreased p-IκBα level, and increased total IκBα level (Fig. 3B). On the other hand, treatment with anti-HSP27 antibody in the CM showed the similar effect of HSP27 knockdown on NF-κB signaling in SCC-15/PYM cells (Fig. 3B). However, overexpression of HSP27 enhanced NF-κB signaling activation in SCC-15 and SCC-25 cell lines. Treating the cells with rhHSP27 exerted the similar effect of HSP27 overexpression (Fig. 3C).

In order to examine whether HSP27 regulates NF-κB signaling via modulating IκBα level, we used a dominant-negative model where a mutant gene encoding a nonphosphorylatable IκBα, which was cloned into the pBabe plasmid (pBabe-IκBα), resulted into the constitutive suppression of NF-κB signaling despite the presence of its activators (9). We found that pBabe-IκBα transfection or treatment with NF-κB inhibitor Bay11-7082 prevented IκBα degradation and impaired rhHSP27-induced NF-κB signaling activation (Fig. 3D). Together, these data indicated that the effect of HSP27 in chemoresistance is mediated by NF-κB signaling activation in SCCT cells, and both extracellular and intracellular HSP27 are involved in this process.

We sought to further explore the key mediator linking between HSP27 and the activation of NF-κB signaling. Given the high correlation between HSP27 function and the expression of Toll-like receptors (TLR), and that TLR5 has been reported to be highly expressed in tongue cancer (10), we examined TLR5 protein level in SCCT cell lines and found that TLR5 is highly expressed in SCCT cells (Supplementary Fig. S3A). To validate whether TLR5 was involved in extracellular HSP27-mediated chemoresistance, we knocked down TLR5 expression in SCCT cell lines with shRNA (Supplementary Fig. S3B). We then assessed whether extracellular HSP27 mediated chemoresistance via TLR5. TLR5 knockdown sensitized SCC-15/PYM cells to chemotherapy (Supplementary Fig. S3C). TLR5 knockdown impaired the chemoresistance induced by rhHSP27 in SCC-15 and SCC-25 cells (Supplementary Fig. S3D). Levels of NF-κB transactivation, p65 nuclear translocation, p-IκBα, and expression of survivin and IL6 were also significantly decreased after TLR5 knockdown in SCC-15/PYM cells, but the total IκBα level increased (Fig. 4A and Supplementary Fig. S3E).

Moreover, the efficiency of rhHSP27 treatment on NF-κB signaling activation was attenuated in SCC-15 cell after TLR5 knockdown (Fig. 4B and Supplementary Fig. S3F). We also confirmed the role of TLR5 in another SCCT cell line SCC-25 in response to rhHSP27 treatment (Fig. 4C and Supplementary Fig. S3G). Furthermore, Co-IP assay demonstrated that extracellular HSP27 could physically interact with TLR5 after rhHSP27 incubation in SCC-15 and SCC-25 cell lines (Fig. 4D). In addition, immunofluorescent staining also confirmed the colocalization of the two proteins on the cell membrane of SCC-15 cells after rhHSP27 incubation (Fig. 4E). These results indicated that extracellular HSP27 activates NF-κB signal via a TLR5-dependent manner to mediate chemoresistance in SCCT cells.

As shown in Fig. 2, the chemosensitizing effect of HSP27 knockdown was greater than anti-HSP27 antibody treatment in SCC-15/PYM cells (Fig. 2A). The chemoresistant effect of HSP27 overexpression was also greater than that of rhHSP27 treatment in SCC-15 and SCC-25 cell lines (Fig. 2B and C), suggesting that intracellular HSP27 may also play a role in chemoresistance. To further explore the mechanism underlying this effect, we further determined the activation of apoptotic executant caspase 3 in SCCT cells resulted from intracellular and extracellular interference of HSP27 function. We found that in SCC-15/PYM cells, anti-HSP27 antibody treatment enhanced PYM- or cDDP-induced caspase-3 activation and cleaved caspase-3 level, but HSP27 knockdown demonstrated stronger enhancement on caspase-3 activation upon chemotherapy (Supplementary Fig. S4A). Similarly, in SCC-15 and SCC-25 cell lines, rhHSP27 treatment attenuated PYM- or cDDP-induced caspase-3 activation and cleaved caspase-3 level, but the efficiency resulted from overexpressed HSP27 was greater than rhHSP27 treatment (Supplementary Fig. S4B). These findings implied that HSP27 mediates chemoresistance via both intracellular and extracellular mechanisms rather than extracellular-only mechanisms.

In order to investigate the role of HSP27 in apoptosis, we detected changes of proapoptotic proteins levels upon chemotherapy. We found that, regardless of HSP27 protein level, chemotherapy induced protein expression of BCL-2 family members BAX, BAD, and BIM in SCCT cell lines SCC-15 and SCC-25 (Fig. 5A), even in SCC-15/PYM cells (Fig. 5B). However, HSP27 overexpression repressed BAX and BIM mitochondria translocation, and cytochrome C (Cyto C) release from mitochondria, and subsequently repressed the increase of cytosolic Cyto C level upon chemotherapy in SCC-15 and SCC-25 cell lines (Fig. 5A). Inversely, in SCC-15/PYM cells, HSP27 knockdown enhanced the mitochondria translocation of BAX and BIM, and the release of Cyto C from mitochondria, and subsequently enhanced the increase of cytosolic Cyto C level (Fig. 5B). To assess whether intracellular HSP27 mediates the translocation of proapoptotic proteins, we detected HSP27 protein level in mitochondria fraction, but there was no detectable HSP27 protein in the mitochondrial fraction. Moreover, using Co-IP assay, we found HSP27 could physically interact with BAX and BIM, and the interacting level was increased when HSP27 was overexpressed in response to chemotherapy (Fig. 5C). In addition, immunofluorescent staining also confirmed the colocalization of HSP27, BAX, and BIM in SCC-15/PYM cells upon cDDP treatment (Supplementary Fig. S4C). These results suggest that intracellular HSP27 interacts with BAX and BIM to sequester them in the cytoplasm from mitochondria translocation upon chemotherapy.

In order to explore the role of HSP27 in chemotherapy, we examined the expression change of HSP27 protein in a series of SCCT cell lines in response to chemotherapy. We found that PYM or cDDP treatment induced increased levels of HSP27 protein in SCCT cells and cell CM (Fig. 6A). In the xenograft tumors formed by SCC-15 and SCC-25 cell lines and their HSP27-overexpressed cell lines (Fig. 2E and Supplementary Fig. S2D), PYM and CDDP treatment induced increased expression of HSP27 protein in xenograft tumors and mice serum (Fig. 6B). Notably, chemotherapy did not induce significant HSP27 protein expression in tumors formed by HSP27-overexpressed SCCT cells, but resulted in significant HSP27 protein increase in mice serum. In addition, we found that chemotherapy induced increased expression of HSP27 protein in the mice normal liver tissues collected from both mice group harboring tumors formed by HSP27-overexpressing and control cells, but not in the mice normal lung and kidney tissues (Supplementary Fig. S5). These data suggest that chemotherapy may also induce HSP27 secretion from other tissues rather than only from tumors.

In order to examine whether chemotherapy alters HSP27 expression, we determined HSP27 protein levels in the serum from 20 healthy controls, and 23 tongue cancer patients before and after chemotherapy, respectively. These patients underwent treatment with PYM- and/or cDDP-based chemotherapy. We found that HSP27 protein level was higher in the serum from tongue cancer patients as compared with that of healthy controls, and that chemotherapy significantly induced expression of HSP27 protein in the serum of 15 of the 23 (65.22%) tongue cancer patients, who showed poor response to chemotherapy (Fig. 6C). Moreover, we detected HSP27 protein level in the 23 paired SCCT tissues collected before and after PYM- and/or cDDP-based chemotherapy. HSP27 protein levels in the 15 tissues from patients who had HSP27 protein increased in serum after treatment and were dramatically increased after chemotherapy (Fig. 6C). These results demonstrated that chemotherapy could induce HSP27 expression in tongue cancer cells as well as in SCCT patients.

To further explore the clinical significance of HSP27 in tongue cancer treatment, we evaluated HSP27 status in two independent cohorts of human SCCT tissues collected from SCCT patients who received PYM- and/or cDDP-based chemotherapy. In the cohort 1 tissues, 84 SCCT tissues were obtained at Affiliated Cancer Hospital of Guangzhou Medical University (Guangzhou, Guangdong Province, China). In the cohort 2 tissues, 81 SCCT tissues were obtained at Xiangya Stomatological Hospital and School of Stomatology, Central South University (Changsha, Hunan Province, China). The SCCT patients from both cohorts 1 and 2 had no distant metastasis. Using immunohistochemistry staining, we found that HSP27 was highly expressed in 57.14% (48/84) and 53.09% (43/81) SCCT tissue samples of cohorts 1 and 2, respectively. High HSP27 levels were closely associated with tumor stage and recurrence of patients with SCCT (Supplementary Tables S2 and S3). TLR5 protein was usually highly expressed in SCCT tissues (77.38% for cohort 1 and 75.31% for cohort 2). Tissue samples with high HSP27 protein level possessed a high nuclear p65 protein level, but a low cleaved caspase-3 protein level (Supplementary Fig. S6). HSP27 protein level was positively correlated with p65 nuclear translocation, but negatively correlated with caspase-3 activation (Fig. 7A). Importantly, SCCT patients with high HSP27 protein level in their tumor samples had significantly shorter overall survival times (Fig. 7B). On the basis of our analysis, we conclude that high HSP27 protein levels predict worse treatment outcomes.

Our study reveals that HSP27 is a key modulator that is at least in part responsible for the chemoresistance of tongue cancer. Our findings provide several insights into the detailed mechanisms for the role of HSP27 in chemoresistance of SCCT: (1) Chemotherapy induced HSP27 protein level increase in SCCT cells and extracellular environment to (2) enhance chemoresistance. (3) Extracellular HSP27 interacts with TLR5 to activate NF-κB signaling, whereas (4) intracellular HSP27 sequesters BAX and BIM from mitochondria translocation to inhibit apoptotic signaling upon chemotherapy (Fig. 7D).

Our study for the first time revealed a novel function of HSP27 in cancers. Heat shock proteins (HSP) include several different protein families, which can be induced in cells by many physiologic and pathophysiologic stresses (11). HSPs are expressed in either a constitutive or inductive pattern in different cell types and subcellular compartments (12). Cancer cells usually possess high expression of chaperones to meet the high metabolic requirements and the abundant signal transduction, subsequently, to maintain cancer cells survival. HSPs mediate the oncogenic signal transduction by folding and stabilizing oncoproteins. Thus, treatment strategies based on HSP inhibition are recognized as an important approach to deplete oncoproteins and attack oncogenic signal pathways required for cancer development and progression (13). HSP27 is an important small HSP and is induced in response to varieties of cellular stresses, such as exposure to mitogens, growth factors, hormones, inflammatory cytokines, and anticancer treatment (14). It has been reported that HSP27 was elevated in some human cancer types, such as breast cancer and prostate cancer, and aberrant HSP27 expression correlates with tumor growth, metastasis, as well as therapy resistance (15, 16). In consistent with these previous observations, we found that the protein expression of HSP27 was significantly upregulated in MDR-SCCT cell line and its CM. Both HSP27 knockdown and HSP27-neutralizing antibody treatment sensitized MDR-SCCT cells to chemotherapy. Also, both ectopic HSP27 overexpression and recombinant human HSP27 protein treatment enhanced chemoresistance in SCCT cell lines. This observation further highlighted that HSP27 is a key player in both maintaining cancer cell function and, more importantly, facilitating cancer cells to escape the stress from chemotherapy.

Our study also delineated the detailed mechanism underlying the HSP27 signaling involved in both extracellular and intracellular function. The extracellular HSP27 level in CM from MDR-SCCT cell line was higher than that from other SCCT cell lines. Interestingly, chemotherapy induced the increase of HSP27 protein levels in CM from SCCT cell lines in vitro and in the serum of mouse model bearing tongue cancer. Notably, chemotherapy did not induce significant HSP27 protein increase in tumors formed by HSP27-overexpressed SCCT cells, but resulted in significant HSP27 protein increase in mice serum, suggesting that chemotherapy may induce HSP27 secretion from both SCCT cells and other tissues. As expected, the serum HSP27 levels are indeed elevated after chemotherapy in SCCT patients. The elevated serum HSP27 level also has been reported in breast cancer patients (17). These lines of evidence highlighted the potential role of HSP27 in the microenvironment of SCCT and possibly other cancers as well. This role in the microenvironment may be associated with both endocrine and paracrine mechanisms. It is worthwhile noting that this elevated HSP27 in both tumor microenvironment and entire circulatory system may globally affect cancer recurrence, metastasis, and drug resistance in general. This notion is supported by previously studies where extracellular HSP27 affects the biological behaviors of both cancer cells and stromal cell, and then contributes to cancer progression, for example, extracellular HSP27 in cancer microenvironment promotes VEGF secretion in endothelial cells depending on TLR3 and then promotes angiogenesis (18). Extracellular HSP27 also exhibited biological effects on monocytes, resulting in the secretion of IL10 and TNFα (19), blocking the differentiation of monocytes to normal dendritic cells (20), and driving the differentiation of monocytes to tumor-associated macrophages to promote cancer progression (21). Whether the role of HSP27 we have discovered in this study would be further extended to these cancer biological and physiologic processes in SCCT patients and further contributes to the chemoresistance and other cancer function remains to be further investigated.

Our study also for the first time identified the major signaling pathways downstream of HSP27 in the chemoresistance SCCT cells. Extracellular HSP27 executes roles intensely through TLR-activated NF-κB signaling (18, 22). We demonstrated that tongue cancer cells and tissues possess high TLR5 protein levels and that extracellular HSP27 triggered NF-κB signaling via binding to TLR5. Moreover, we found that chemotherapy induced expression of proapoptotic proteins BAX, BAD, and BIM which are major BH-3 domain containing BLC-2 family members. We demonstrated that HSP27 interacts with BAX and BIM to attenuate their translocation to mitochondria and subsequently blocks Cyto C release. Our observation is consistent with previous findings where HSP27 is able to prevent cells from apoptosis by interacting with Cyto C (23) or by interfering the apoptotic signaling upstream of the mitochondrial Cyto C release (24, 25). In addition, it was reported that HSP27 as a subunit of AUF1 protein complexes can also act as a RNA-binding protein to mediate mRNA decay (26). Enhanced HSP27 expression in response to oxidative stress binds to the 3′-untranslated region of BIM mRNA to repress its translation and subsequently prevent neuronal death in cerebellar granule neurons (27). Whether this could also be a potential mechanism in tongue cancer as well as other cancer types should be further investigated in future studies.

Last, our study has important translational implications. Our findings indicate that HSP27 and its associated key molecules can be critical targets for developing new therapeutics for tongue cancer. In particular, we showed that HSP27 is induced among SCCT patients by chemotherapy, which indicates that increased HSP27 may also be a useful biomarker for selecting patients toward developing precision medication for tongue cancer treatment. In addition, it is of importance to keep in mind that HSP27-targeting strategies should control both intracellular and extracellular functions of HSP27. Our study thus warranted continued investigation in these translational directions.

No potential conflicts of interest were disclosed.

Conception and design: G. Zheng, Y. Xiong, Z. He

Development of methodology: G. Zheng, Z. Zhang, L. Luo, X. Jia, H. Pan

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Zheng, Q. Zhang, N. Li, M. Lu, Y. Song

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Zheng, Z. Zhang, H. Liu, Y. Gu, W. Liu

Writing, review, and/or revision of the manuscript: G. Zheng, H. Liu, W. Liu, Z. He

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G. Zheng, L. Luo, C. Peng

Study supervision: J. Liu, Z. He

The authors thank Key Laboratory of Cancer Proteomics of Chinese Ministry of Health (Xiangya Hospital, Central South University, Changsha, China) for the technical support of proteomics.

This study was supported by grants from the National Natural Science Foundation of China [no. 81402196 (G. Zheng), no. 81672616 (G. Zheng), no. 81272450 (Z. He), and no. 81401989 (N. Li)], Guangdong Natural Science Funds for Distinguished Young Scholars (no. 2016A030306003; G. Zheng), Guangzhou key medical discipline construction project fund, Science and Technology Program of Guangzhou, China (201710010100; G. Zheng), and Guangzhou Municipal University Scientific Research project (1201610027; G. Zheng).

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