Overexpression of HSP10 correlates with HSP60 and Mcl-1 levels and predicts poor prognosis in non-small cell lung cancer patients.

BACKGROUND: HSP60 and its partner HSP10 are members of heat shock proteins (HSPs) family, which help mitochondrial protein to fold correctly. Mcl-1, a member of the Bcl-2 family, plays a crucial role in regulation of cell apoptosis. Aberrant expression of HSP10, HSP60 and Mcl-1 is involved in the development of many tumors.
OBJECTIVE: To examine the association between expression of HSP10, HSP60 and Mcl-1 and clinicopathological features of non-small cell lung cancer (NSCLC).
METHODS: Tissue microarrays including 53 non-cancerous lung tissues (Non-CLT) and 354 surgically resected NSCLC were stained with anti-HSP10, anti-HSP60 and anti-Mcl-1 antibodies respectively by immunohistochemistry.
RESULTS: Higher expression of HSP10, HSP60 and Mcl-1 was found in NSCLC compared with Non-CLT. Both individual and combined HSP10 and HSP60 expression in patients with clinical stage III was higher than that in stage I ∼ II. Expression of HSP10 showed a positive correlation with HSP60 and Mcl-1. Overall survival time of NSCLC patients was remarkably shorter with elevated expression of HSP10, HSP60 and Mcl-1 alone and in combination. Moreover overexpression of HSP10 and Mcl-1 was poor independent prognostic factor for lung adenocarcinoma patients.
CONCLUSIONS: High expression of HSP10, HSP60 and Mcl-1 might act as novel biomarker of poor prognosis for NSCLC patients.

Introduction

Lung cancer, particularly non-small cell lung cancer (NSCLC) is the most frequently diagnosed cancer and the leading cause of cancer-related death worldwide [1]. Approximately 80%–85% of diagnosed lung cancer patients are NSCLC, among which adenocarcinoma (ADC) and squamous cell carcinoma (SCC) are the most two common histological subtypes [2]. Early detection of NSCLC is of critical importance for a favorable outcome after surgical resection, since the five-year survival rate of patients with stage IB could reach 68% [3]. Unfortunately, most of the patients are in advanced stages at the time of diagnosis [4]. Although recent years have seen great progress in treatment for patients with late stage lung cancer, the survival rate is still very low [3, 4]. Therefore, it is of great necessity and urgency to discover and identify new biomarker(s) for the prediction, diagnosis, prognosis judgment and targeted therapy of NSCLC, in turn to lead more effective treatment and reduction of mortality.

Heat shock proteins (HSPs), also known as stress proteins, are highly conserved and maintain cellular homeostasis under various stress conditions, such as hyperthermia and hypoxia [5]. The 10 kDa heat shock protein (HSP10), predominantly located in mitochondria, is a co-chaperone protein of the 60 kDa heat shock protein (HSP60). They interact to fold and transport proteins to the mitochondria, in order to maintain the stability of the mitochondrial protein [6]. Aberrant expression of HSP10 and HSP60 is involved in the progression of several major kinds of tumors, including colorectal adenocarcinoma and glioblastoma [7, 8, 9]. In addition, HSP10 and HSP60 are also involved in cell apoptosis. It is well known that the activation of Bak/Bax is considered to be the most critical step of mitochondrial pathway which leads to cell apoptosis [10]. Low expression of HSP10 can promote testosterone-induced apoptosis in mouse ovarian granulosa cells by upregulating pro-apoptotic factor Bax [11]. HSP60 can also form complexes with Bax and Bak to play an anti-apoptotic function [12, 13].

Myeloid Cell Leukemia 1 (Mcl-1) is known as a member of B-cell Lymphoma 2 (Bcl-2) family that strictly controls the intrinsic apoptotic pathways [14]. Accumulating evidences suggest that Mcl-1 plays crucial roles in suppressing mitochondria-dependent apoptosis [15] through regulation of the balance between the long and short variants of the Mcl-1 gene to regulate mitochondrial fusion and fission machinery [16]. Moreover, elevated Mcl-1 has also been found in different types of cancer, such as breast cancer, acute myeloid leukemia, hepatocellular carcinoma and lung cancer [17, 18, 19, 20].

It still remains unclear whether the expression of HSP10, HSP60 and Mcl-1 is related to clinicopathological features of NSCLC patients. In this study, we carried out the present retrospective study by using NSCLC tissue microarrays (TMAs) and immunohistochemistry (IHC) to evaluate the potential correlation between HSP10, HSP60 and Mcl-1, and to investigate whether there is a relationship between their expression, clinicopathological parameters and the prognostic impact in NSCLC.

Materials and methods

Ethical statement

All protocols related to the research were approved by the Institutional Human Experiment and Ethics Committee of the Second Xiangya Hospital of Central South University (approval No. S039/2011). Informed consent was obtained from all subjects before surgery.

Clinical data

A total of 53 non-cancerous lung tissues (Non-CLT) and 354 NSCLC tissues resected from 2003 to 2013 in the Thoracic Surgery Department of the Second Xiangya Hospital were obtained from the Department of pathology, the Second Xiangya Hospital of Central South University. No one received radiation therapy or chemotherapy. All patients obtained the exact histological diagnosis which was confirmed by well-experienced pathologists. The clinical stage of each case was confirmed according to the Eighth Edition Lung Cancer Stage Classification [21]. Clinicopathological data (Supplemental Table 1) were extracted from the clinical record. In this study, we constructed NSCLC TMAs by using the TMA technology as previously described [22, 23].

Table 1

Analysis of the association between expression of HSP10, HSP60 and Mcl-1 proteins and clinicopathological features of NSCLC ( 354)

Clinicopathological features (n)	HSP10			HSP60			Mcl-1			HSP10/HSP60^#			HSP10/HSP60/Mcl-1^#
	High (%)	Low (%)	P-value	High (%)	Low (%)	P-value	High (%)	Low (%)	P-value	P^+ (%)	N^- (%)	P-value	P^+ (%)	N^- (%)	P-value
Age (years)
⩽ 50 (n= 97)	60 (61.9)	37 (38.1)		68 (70.1)	29 (29.9)		63 (64.9)	34 (35.1)		50 (51.5)	47 (48.5)		37 (38.1)	60 (61.9)
> 50 (n= 257)	149 (57.6)	109 (42.4)	0.545	191 (74.3)	66 (25.7)	0.423	166 (64.6)	91 (35.4)	1.000	133 (51.8)	124 (48.2)	1.000	95 (37.0)	162 (63.0)	0.902
Gender
Male (n= 271)	158 (58.3)	113 (41.7)		198 (73.1)	73 (26.9)		176 (64.9)	95 (35.1)		139 (51.3)	132 (48.7)		101 (37.3)	170 (62.7)
Female (n= 83)	50 (60.2)	33 (39.8)	0.800	61 (73.5)	22 (26.5)	1.000	53 (63.9)	30 (36.1)	0.896	44 (53.0)	39 (47.0)	0.803	31 (37.3)	52 (62.7)	1.000
Clinical stages
Stage I–II (n= 151)	78 (51.7)	73 (48.3)		101 (66.9)	50 (33.1)		90 (59.6)	61 (40.4)		67 (44.4)	84 (55.6)		47 (31.1)	104 (68.9)
Stage III (n= 203)	130 (64.0)	73 (36.0)	0.022^*	158 (77.8)	45 (22.2)	0.029^*	139 (68.5)	64 (31.5)	0.092	116 (57.1)	87 (42.9)	0.018^*	85 (41.9)	118 (58.1)	0.045^*
LN status
LNM (n= 213)	133 (62.4)	80 (37.6)		166 (77.9)	47 (22.1)		140 (65.7)	73 (34.3)		118 (55.4)	95 (44.6)		83 (39.0)	130 (61.0)
No LNM (n= 141)	75 (53.2)	66 (46.8)	0.098	93 (66.0)	48 (34.0)	0.014^*	89 (63.1)	52 (36.9)	0.650	65 (46.1)	76 (53.9)	0.103	49 (34.8)	92 (65.2)	0.434
Histological type
SCC (n= 161)	94 (58.4)	67 (41.6)		105 (65.2)	56 (34.8)		90 (55.9)	71 (44.1)		76 (47.2)	85 (52.8)		51 (31.7)	110 (68.3)
ADC (n= 193)	114 (59.1)	79 (40.9)	0.914	154 (79.8)	39 (20.2)	0.003^*	139 (72.0)	54 (28.0)	0.002^*	107 (55.4)	86 (44.6)	0.135	81 (42.0)	112 (58.0)	0.048^*
Pathological grade
Well (n= 6)	2 (33.3)	4 (66.7)		2 (33.3)	4 (66.7)		6 (100.0)	0 (0.0)		1 (16.7)	5 (83.3)		1 (16.7)	5 (83.3)
Moderate (n= 146)	84 (57.5)	62 (42.5)		105 (71.9)	41 (28.1)		94 (64.4)	52 (35.6)		71 (48.6)	75 (51.4)		50 (34.2)	96 (65.8)
Poor (n= 202)	122 (60.4)	80 (39.6)	0.384	152 (75.2)	50 (24.8)	0.067	129 (63.9)	73 (36.1)	0.188	111 (55.0)	91 (45.0)	0.113	81 (40.1)	121 (59.9)	0.309
Survival status
Alive (n= 194)	99 (51.0)	95 (49.0)		131 (67.5)	63 (32.5)		110 (56.7)	84 (43.3)		84 (43.3)	110 (56.7)		53 (27.3)	141 (72.7)
Dead (n= 160)	109 (68.1)	51 (31.9)	0.001^*	128 (80.0)	32 (20.0)	0.011^*	119 (74.4)	41 (25.6)	0.001^*	99 (61.9)	61 (47.9)	0.001^*	79 (49.4)	81 (50.6)	0.000^*

Chi-square test, statistically significant difference ( 0.05). Abbreviations: LNM, lymph node metastasis; SCC, squamous cell carcinoma; HSP10/HSP60 P, common high expression of HSP10 and HSP60, P, other combination of these two proteins; HSP10/HSP60/Mcl-1 P, common high expression of these three proteins, P, other combination of expression of these three proteins.

Antibody selections

Staining was done using primary antibodies as follows: HSP10 (Catalog: SC-376313; Santa Cruz Biotechnology, CA; 1:1000 dilution); HSP60 (Catalog: #12165, Cell Signaling Technology; 1:1500 dilution); Mcl-1 (Catalog: #94296, Cell Signaling Technology; 1:300 dilution).

Immunohistochemistry and scores

The staining for HSP10, HSP60 and Mcl-1 in TMAs of NSCLC and non-CLT was carried out as described previously and the staining conditions of each antibody was adjusted in the light of previous laboratory experience [24, 25].

Immunostaining was scored independently by QW and SF. The score was based on intensity and distribution of staining described as follows: the total score is percentage score plus intensity score. Specifically, the intensity of staining signal was scored as 0 (negative), 1 (mild expression), 2 (moderate expression) and 3 (strong expression). According to the percentage of positive cells, the percentage score included 0 (0%), 1 (1–25%), 2 (26–50%), 3 (51–75%) and 4 (76–100%). The total score ranges from 0 to 7. According to NSCLC patients’ overall survival (OS) using the log-rank test, the best cutoff levels for the three proteins are 5, 4 and 4, respectively. The total scores more than 5, 4 and 4 were considered to be high expression of these three proteins, others were considered low expression, respectively. Agreement between the two pathologists is 95%, and borderline and equivocal cases are solved through remained and discussion.

Statistical analysis

Statistical analysis was performed by using SPSS 24.0 software. The difference expression of proteins between NSCLC and Non-CLT was analyzed by Chi square test, which was also used for analyzing the relationship between HSP10, HSP60 and Mcl-1 and the clinicopathological features of NSCLC. Spearman’s rank correlation coefficient was performed on the correlation between HSP10, HSP60 and Mcl-1. Kaplan-Meier analysis was performed for survival rate, and comparisons of survival rate curve were analyzed by log-rank test. Cox proportional hazards regression model was carried out to assess independent prognostic factors. Statistical significance was set at 0.05 (Two-sided).

Results

Expression of HSP10, HSP60 and Mcl-1 increased significantly in NSCLC

The expression and subcellular localization of HSP10, HSP60 and Mcl-1 was detected in NSCLC and Non-CLT tissues by IHC. These three proteins were stained in cytoplasm of lung SCC and ADC cells (Fig. 1A, B, D, E, G and H), but were absent from Non-CLT (Fig. 1C, F and I). The high expression percentage of HSP10 and HSP60 was 58.4% (94/161) and 65.2% (105/161) in lung SCC, 59.1% (114/193) and 79.8% (154/193) in lung ADC, and 11.3% (6/53) and 26.4% (14/53) in Non-CLT, respectively. For Mcl-1, the

Figure 1.

Expression of HSP10, HSP60 and Mcl-1 in lung ADC, lung SCC and Non-CLT were detected by IHC. Strong positive staining of HSP10 (A), HSP60 (D) and Mcl-1 (G) was found in cell cytoplasm of lung SCC cells. Strong positive staining of HSP10 (B), HSP60 (E) and Mcl-1 (H) was also showed in cell cytoplasm of lung ADC cells. Negative staining of HSP10 (C), HSP60 (F) and Mcl-1 proteins was found in Non-CLT (200 , IHC, DAB staining).

percentage of high expression was 55.9% (90/161), 72.0% (139/193) and 28.3% (15/53) in lung SCC, ADC and Non-CLT. As shown in Fig. 2, the level of HSP10, HSP60 and Mcl-1 expression in lung SCC and ADC was remarkably increased (all 0.001).

Figure 2.

The comparison of expression of HSP10, HSP60 and Mcl-1 in lung SCC and lung ADC compared to the Non-CLT. The expression of HSP10, HSP60 and Mcl-1 in lung SCC and lung ADC was significantly higher than those in Non-CLT (all 0.001).

Association between expression of proteins and clinicopathological features of NSCLC

We further analyzed the association between elevated expression of HSP10, HSP60 and Mcl-1 and clinicopathological features. The data in Table 1 shows that the percentage of overexpression of HSP10 and HSP60 in patients with clinical stage III was higher than that in clinical stage I II ( 0.022, 0.029, respectively). Of note, NSCLC patients with lymph node metastasis (LNM) had higher expression of HSP60 than that without LNM ( 0.014), and lung ADC patients had higher HSP60 expression than lung SCC patients ( 0.003). As for Mcl-1, lung ADC patients had obviously higher expression of Mcl-1 than lung SCC patients ( 0.002). Compared to NSCLC patients in clinical stage I II, a conjoint analysis indicated that patients in clinical stage III had higher HSP10 and HSP60 expression, or higher expression of HSP10, HSP60 and Mcl-1 in combination ( 0.018, 0.045, respectively). Moreover, patients with individual or combined high expression of HSP10, HSP60 and Mcl-1 suffered a lower OS rate than that with other phenotypes of these proteins (all 0.05).

Correlation between HSP10, HSP60 and Mcl-1 expression in NSCLC

We analyzed the correlation between HSP10, HSP60 and Mcl-1 expression in NSCLC. As data shown in Table 2, the overexpression of HSP10 was positively correlated with expression of HSP60 and Mcl-1 in NSCLC ( 0.399, 0.001; 0.137, 0.010, respectively). Similar correlation can be found between HSP60 and Mcl-1 expression in NSCLC ( 0.219, 0.001).

Table 2

The pairwise association between expression of HSP10, HSP60 and Mcl-1 in 354 cases of NSCLC

	HSP10	HSP60	Mcl-1
HSP10
Spearman’s correlation coefficient	1	0.399	0.137
Sig. (2-tailed)		0.000^*	0.010^*
HSP60
Spearman’s correlation coefficient	0.399	1	0.219
Sig. (2-tailed)	0.000^*		0.000^*
Mcl-1
Spearman’s correlation coefficient	0.137	0.219	1
Sig. (2-tailed)	0.010^*	0.000^*

Spearman’s rank correlation test, statistically significant difference ( 0.05).

Status of expression of HSP10, HSP60 and Mcl-1 on patients’ prognosis

Moreover, we used Kaplan-Meier analysis to estimate the association between expression of the three proteins and patients’ survival rate with lung SCC and ADC alone or in different combination. As shown in Fig. 3, the survival rates of lung SCC patients with low HSP10 expression ( 0.017, Fig. 3A), HSP60 ( 0.012, Fig. 3B) or Mcl-1 ( 0.025, Fig. 3C) were higher than those with high expression. Lung ADC patients with low expression of HSP10 ( 0.007, Fig. 3D) and Mcl-1 ( 0.026, Fig. 3F) had longer survival time than these with high expression of this two proteins, there was no significant difference in survival regardless of HSP60 expression level in ADC patients ( 0.233, Fig. 3E). In combined condition, lung SCC and ADC patients with high HSP10 and HSP60 owned a lower survival rate than that of others ( 0.003, Fig. 4A; 0.012, Fig. 4C). Similarly, both lung SCC and ADC patients who had combined high expression of HSP10, HSP60 and Mcl-1 had obviously lower OS rates than patients with any other patterns ( 0.001, Fig. 4B; 0.003, Fig. 4D).

Figure 3.

Kaplan-Meier curves for overall survival of lung SCC and ADC patients with expression of HSP10, HSP60 and Mcl-1. (A) Lung SCC patients with high expression of HSP10 showed worse overall survival rates compared to patients with low HSP10 expression ( 0.017, two sided). (B) Lung SCC patients with high HSP60 expression showed worse overall survival rates compared to patients with low HSP60 expression ( 0.012, two sided). (C) Lung SCC patients with high Mcl-1 expression had worse overall survival rates than patients with low one ( 0.025, two sided). (D) Lung ADC patients with high expression of HSP10 had worse overall survival rates than that with low one ( 0.007, two sided). (E) High expression of HSP60 had not significantly correlation with overall survival rates of lung ADC patients ( 0.05, two sided). (F) Lung ADC patients with low Mcl-1 expression survived longer than those with high one ( 0.026, two sided).

Figure 4.

Kaplan-Meier curves for overall survival of lung SCC and ADC patients with combined expression of HSP10/HSP60 and HSP10/HSP60/Mcl-1. (A) Kaplan-Meier curves showed lung SCC patients with combined high expression of HSP60 and HSP10 had worse overall survival rates than these with others ( 0.003, two sided). (B) Lung SCC patients with high HSP10, HSP60 and Mcl-1 expression owned poor prognosis compared with that with other expression patterns of these three proteins ( 0.001, two sided). (C) Lung ADC patients with common high expression of HSP60 and HSP10 proteins showed worse overall survival rates compared with others ( 0.012, two sided). (D) Patients with combined high expression of HSP10, HSP60 and Mcl-1 all have higher survival rates than others ( 0.003, two sided).

Furthermore, we analyzed the correlation between survival rate and clinicopathological features. As the data shown in Tables 3 and 4, for lung SCC and ADC patients, the survival rate of patients in clinical stage III was lower than that of patients in clinical stage I II (all 0.001), and patients without LNM had a longer survival time than that of patients with LNM ( 0.005; 0.002). Similar trend also reflected in pathological grade, lung SCC patients with higher pathological grade owned a lower the survival rate ( 0.004).

Table 3

Summary of univariate/multivariate analysis for overall survival in patients with lung SCC ( 161)

Variables	Univariate analysis			Multivariate analysis
	Average survival time (SE)	95% CI	P-value	Exp (B)	95.0% CI	P-value
HSP10
High expression	48.766 (4.511)	39.925–57.607	0.017^*	1.463	0.840–2.550	0.179
Low expression	77.114 (8.269)	60.905–93.322
HSP60
High expression	51.516 (4.992)	41.733–61.300	0.012^*	1.609	0.873–2.976	0.128
Low expression	80.509 (8.967)	62.934–98.087
Mcl-1
High expression	56.467 (6.553)	43.623–69.310	0.025^*	1.333	0.784–2.256	0.290
Low expression	69.234 (6.155)	57.197–81.306
Clinical stages
Stage I–II	77.678 (5.901)	66.111–89.244	0.000^*	1.895	1.004–3.578	0.049^*
Stage III	53.732 (5.906)	42.157–65.307
LN status
LNM	46.512 (4.476)	37.739–55.258	0.005^*	1.709	0.958–3.051	0.070
No LNM	80.439 (8.389)	63.996–96.881
Pathological grade
Well/moderated	68.856 (5.510)	58.763–78.950	0.004^*	1.970	1.152–3.372	0.013^*
Poor	56.314 (6.201)	44.160–68.469
Age
⩽ 50	51.443 (5.738)	40.196–62.690	0.776	1.005	0.575–1.756	0.986
> 50	65.619 (6.273)	53.324–77.913
Gender
Female	64.358 (6.912)	50.836–77.159	0.063	0.318	0.075–1.342	0.119
Male	62.314 (5.533)	51.470–73.159

Abbreviations: CI, confidence interval; Exp(B), odds ratio; SE, standard error; LNM, lymph node metastasis; SCC, squamous cell carcinoma. 0.05.

Table 4

Summary of univariate/multivariate analysis for overall survival in patients with lung ADC ( 193)

Variables	Univariate analysis			Multivariate analysis
	Average survival time (SE)	95% CI	P-value	Exp (B)	95.0% CI	P-value
HSP10
High expression	44.716 (4.333)	36.224–53.208	0.007^*	1.763	1.086–2.862	0.022^*
Low expression	59.130 (4.821)	49.680–68.580
HSP60
High expression	49.987 (3.847)	42.447–57.526	0.233	0.795	0.433–1.457	0.457
Low expression	57.787 (6.878)	44.306–71.268
Mcl-1
High expression	45.140 (3.484)	38.311–51.970	0.026^*	1.845	1.109–2.256	0.018^*
Low expression	67.201 (6.419)	54.619–79.782
Clinical stages
Stage I–II	59.815 (4.579)	50.839–68.790	0.000^*	1.547	0.933–2.567	0.091
Stage III	43.371 (4.264)	35.013–51.728
LN status
LNM	45.793 (4.258)	37.447–54.139	0.002^*	1.741	1.016–2.984	0.044^*
No LNM	59.878 (4.831)	50.408–69.348
Pathological grade
Well/moderated	53.915 (4.572)	44.954–62.876	0.113	1.199	0.804–1.790	0.374
Poor	48.897 (4.585)	39.910–57.883
Age
⩽ 50	41.866 (4.407)	33.229–50.502	0.510	1.073	0.674–1.709	0.766
> 50	53.961 (4.113)	45.900–62.021
Gender
Female	52.183 (4.935)	42.510–61.855	0.457	0.864	0.555–1.344	0.516
Male	50.635 (4.274)	42.257–59.012

Abbreviations: CI, confidence interval; Exp(B), odds ratio; SE, standard error; LNM, lymph node metastasis; ADC, adenocarcinoma. 0.05.

We further investigated whether the high HSP10, HSP60 and/or Mcl-1 expression could act as independent prognostic marker for NSCLC patients via Cox proportional risk regression analysis. Table 3 showed that pathological grade ( 0.013) and clinical stage ( 0.049) might be independent prognostic factors for lung SCC patients. Table 4 showed that expression of HSP10 ( 0.022) and Mcl-1 ( 0.018) and LNM ( 0.044) were independent prognostic factors for patients with lung ADC.

Discussion

HSP10 is a co-chaperone protein of HSP60 in mitochondria where it is involved in folding and transporting proteins into mitochondria, and maintaining the stability of mitochondrial proteins [6]. In addition, HSP10 plays a crucial role in cancer progression and development [26], as indicated by its increased expression in many kinds of malignant tumors, including astrocytoma, colorectal cancer, and liver cancer [7, 27, 28]. It has suggested that HSP10 might be an independent prognostic factor for poor prognosis of nasopharyngeal carcinoma and invasive ductal breast carcinoma [29, 30]. In this study, we found higher HSP10, HSP60 and Mcl-1 expression in NSCLC compared with Non-CLT. Patients in clinical stage III had higher HSP10 and/or HSP60 than that in stage I II. On protein level, there existed a positive correlation between HSP10, HSP60 and Mcl-1. Moreover, OS time in NSCLC patients was remarkably shorter in cases with elevated expression of HSP10, HSP60 and Mcl-1 alone and in combination. High HSP10 and Mcl-1 could be poor independent prognostic factors for lung ADC patients.

Our results indicated that the expression of HSP10 was significantly increased in NSCLC, which is consistent with the previous discoveries. In addition, increased expression of HSP10 in patients with advanced clinical stage provides further evidence for its role in tumorigenesis and progression of NSCLC. The synergistic effects of HSP10 and HSP60 may occur not only in maintaining the stability of mitochondrial protein, but also in the process of carcinogenesis. Indeed, we found that both HSP10 and HSP60 were highly expressed in NSCLC compared with Non-CLT, and the percentage of patients with the combined high expression of HSP10 and HSP60 in clinical stage III was significantly higher than that in clinical stage I II. Patients with high co-expression of HSP10 and HSP60 also suffered a lower OS rate than those with other phenotypes of these two proteins. Since HSP10 and HSP60 are localized head to head and separated by a bidirectional promoter on chromosome 2 [31], the region between these two genes might act as a two-way promoter whose transcriptional activity could be triggered by the same stimulus. This could explain why there is a positive correlation between HSP10 and HSP60 expression. Moreover, co-expression of HSP10 and HSP60 might have a synergistic effect to against apoptosis in NSCLC cells. As reported, combined expression of HSP10 and HSP60 can protect cardiac myocytes from apoptosis coaxed by simulated ischemia and reoxygenation [32]. Mutation in the HSP10 mobile loop region impairs its ability to bind with HSP60 and therefore abolish the protective effect of HSP10 on cardiomyocytes [33]. In addition, HSP10, HSP60 and pro-caspase 3 can form a complex, while highly expressed HSP10 and HSP60 can inhibit the dissociation of pro-caspase 3 thus antagonize stress or other factors induced apoptosis [34].

Our data showed that both HSP10 and HSP60 expression were positively correlated with Mcl-1 expression. Patients with combined overexpression of HSP10, HSP60 and Mcl-1 owned lower OS rate than those with other phenotypes. All these results strongly indicate that HSP10 and HSP60 might be involved in the anti-apoptosis pathway of Mcl-1 protein. A plausible explanation is that the expression of HSP10, HSP60 and Mcl-1 could be regulated by same transcription regulator, such as signal transducer and activator of transcription 3 (STAT3), which expresses highly in NSCLC and plays an inhibitory role in the process of tumor cell apoptosis [35, 36]. Previous studies found that the inactivation of STAT3 can downregulate the transcriptional level of MCL-1 gene [36, 37]. Since the bidirectional promoter between HSP10 and HSP60 contains STAT3-responsive element, STAT3 can also associate with this binding site to induce the up-regulation of HSP10 and HSP60 levels [38]. Another likely explanation is that both HSP60 and Mcl-1 are involved in the same signal pathway. On one hand, as a highly unstable protein, Mcl-1 is very sensitive to activity of mechanistic target of rapamycin complex 1 (mTORC1)-dependent translation [39, 40]. On the other hand, HSP60 gene knockdown can activate AMPK which inhibit mTORC1 and its downstream targets, thereby reducing protein synthesis [9, 41]. As mentioned above, it indicates that HSP60 might affect Mcl-1 expression level through mTORC1 pathway. The regulatory network of Mcl-1 is very complicated and the underlying mechanisms of relationship between HSP10, HSP60 and Mcl-1 still need further study in the future.

In summary, the expression of HSP10, HSP60 and Mcl-1 in NSCLC patients increased significantly. HSP10 and HSP60 may play a synergistic role in the development of NSCLC and participate in the anti-apoptosis pathway of mitochondria in NSCLC. HSP10 and HSP60 might be regarded as novel biomarkers of poor prognosis for NSCLC patients.

Supplementary Table 1

Clinicopathological features of patients with NSCLC and non-cancerous control lung tissues

Patients characteristics	No. of patients (%)
NSCLC
Age (years)
⩽ 50	97 (27.4)
> 50	257 (72.6)
Gender
Male	271 (76.6)
Female	83 (23.4)
Clinical stages
Stage I	76 (21.5)
Stage II	75 (21.2)
Stage III	203 (57.3)
Lymph node status
N0	141 (39.8)
N1/N2/N3	213 (60.2)
Histological type
SCC	161 (45.5)
ADC	193 (54.5)
Pathological grade
Well	6 (1.7)
Moderate	146 (41.2)
Poor	202 (57.1)
Non-cancerous control lung tissues
Age (years)
⩽ 50	22 (41.5)
> 50	31 (58.5)
Gender
Male	27 (50.9)
Female	26 (49.1)