5@ CelPress
Heliyon
journal homepage: www.cell.com/heliyon
Heliyon
H
ColPress
Research article
Multidimensional pan-cancer analysis of HSPA5 and its validation in the prognostic value of bladder cancer
YaXuan Wanga, Jinfeng Wanga, Yang Liu ª, XiaoLin Wangb,”, MingHua Ren
ª Department of Urology, The First Affiliated Hospital of Harbin Medical University, Harbin, 150001, China
b Department of Urology, Nantong Tumor Hospital, Nantong, 226361, China
| ARTICLE INFO | ABSTRACT |
|---|---|
| Keywords: HSPA5 ER stress Prognosis BLCA | Endoplasmic reticulum (ER) stress-related genes are closely related to the occurrence, develop- ment, and immunotherapy response of tumors. This study provides a comprehensive assessment of HSPA5 from a pan-cancer perspective using multi-omics data. We analyzed the function of HSPA5 in multiple tumor types using multiple databases. Finally, immunohistochemistry was used to examine the relationship between HSPA5 expression in tissue microarrays from 100 patients with bladder cancer and the prognosis of patients with bladder cancer. Using the TCGA database, we were able to determine that HSPA5 is significantly elevated in a number of common malignancies and is linked with a bad prognosis. Cox regression analysis showed that the high expression of HSPA5 was correlated with OS, progression free survival (PFS), disease free survival (DFS), and disease special survival (DSS) of adrenocortical carcinoma (ACC). In addition, we discovered significant disparities in HSPA5 methylation and phosphorylation levels between various malignancies and normal tissues. HSPA5 expression was significantly correlated with the levels of infiltrating cells and immune checkpoint genes. HSPA5 is highly expressed in bladder cancer and patients with high HSPA5 expression have a poor prognosis. Our study provides a basis for further understanding of the role of ER stress-related gene HSPA5 in different tumor genesis and development. HSPA5 has also been shown to be a prognostic biomarker for bladder cancer patients. |
1. Introduction
Cancer is one of the main causes of mortality and disability worldwide, making it a major public health concern [1]. Cancer may be treated in a number of ways today, some of which include cutting into the patient. Despite these medicines’ successes in combating cancer, a sizable percentage of patients get relatively little from them. This serious problem emphasises the need for a comprehensive knowledge of tumour genesis pathways [2]. New targets and biomarkers for the diagnosis and treatment of cancer are urgently needed. Through their ongoing development, the TCGA and GEO databases have simplified the analysis of individual gene expression, gene connection, clinical prognosis, and signal pathway regulation.
Secretory protein synthesis, folding, and modification all take place in the endoplasmic reticulum (ER), a key organelle [3]. Even though the endoplasmic reticulum (ER) tightly regulates protein processing and synthesis, ER stress, characterized by the
* Corresponding author.
accumulation of misfolded or unfolded proteins, may be caused by a variety of external factors and intracellular processes [4]. Cancer cells’ function, destiny, and survival are all adversely affected by ER stress, which is caused by the tumor microenvironment and abnormal transcription and metabolism [5]. Targeting HSPA5 was shown to induce endoplasmic reticulum stress in both melanoma and hepatocellular carcinoma [6,7]. HSPA5 is a crucial chaperone protein in the normal unfolded protein response (UPR). However, this effect can easily lead to resistance of cancer cells to treatment [8]. Through many molecular processes, HSPA5 promotes the proliferation and survival of cancer cells. Its overexpression leads to primary cancer cell growth and metastasis, increased medication and treatment resistance, and worse clinical outcomes overall [9]. Inhibiting HSPA5 also boosts the potency of cancer drugs like chemotherapy. For instance, mantle cell lymphoma cells were made more sensitive to the anticancer proteasome inhibitor bortezomib by the small molecule Hsp 90 inhibitor retaspimycin hydrochloride [10]. As consequently, HSPA5 is pivotal in tumour development. Recent studies have shown that HSPA5 regulates proliferation, metastasis and iron death of bladder cancer cells through the P53/SLC7A11/GPX4 pathway [11]. And our study further confirmed the important value of HSPA5 in bladder cancer through clinical samples. It has also been shown that HSPA5 should play an important role in the entry of SARS-Cov-2 into cancer patients through the lungs. In addition, HSPA5 showed a high correlation with COVID-19 [12,13]. Chimeric antigen receptor T (CAR-T) cell therapy has been successfully applied to treat hematologic malignancies but faces many challenges in solid tumors. Cell surface GRP78-targeted CAR-T cells have also been shown to be effective in treating human pancreatic cancer [14]. In addition, HSPA5 has been shown to affect the stemness profile of cancer cells in esophageal, glioblastoma, gastric and breast cancers [15-18].
This is the first research to investigate HSPA5’s possible molecular pathways in the aetiology of various malignancies by analysing its gene expression, survival, and mutation using the TCGA datasets.
2. Materials and methods
2.1. Samples and datasets
Using information from the GTEx database and the Cancer Genome Atlas (TCGA), we compared HSPA5 expression in distinct cancer types and normal tissues. In addition, between June 2012 and March 2018, the Nantong Tumour Hospital collected data from 100 patients diagnosed with bladder cancer and 41 cases of normal bladder tissue undergoing partial and radical cystectomy. After surgery, the follow-up period for each patient ranged from one to six years and lasted until August 2019. All subjects provided their informed assent in writing. The Ethics Committee of Nantong Cancer Hospital approved the study.
2.2. Prognostic analysis of HSPA5 in different tumors
Forest plots and Kaplan-Meier curves were used to examine the association between HSPA5 expression and the various survival rates of patients with multiple tumors. A univariate survival analysis was performed in order to calculate the hazards ratios (HR) and the confidence intervals for 95%.
2.3. Analysis of methylation and phosphorylation levels of HSPA5 in different tumors
The study compared HSPA5 methylation and phosphorylation levels in cancer and surrounding tissues using the UALCAN database.
2.4. Gene mutation analysis of HSPA5 in different tumors
The somatic mutation, structural variation, amplification, lack of depth, and numerous alterations in HSPA5 in malignancies were evaluated using the Cbioportal for Cancer Genomics website. On top of that, TCGA patients with and without HSPA5 gene mutations were compared for OS, DFS, PFS, and RFS.
2.5. Functional enrichment analysis of HSPA5 gene
RNAseq data (level 3) and corresponding clinical information for the eight tumors required for analysis were obtained from the TCGA dataset. Two categories, high and low HSPA5 expression, were used to classify the samples. Limma, a R tool for analysing differential gene expression, was used to analyze the HSPA5 mRNA levels (version: 3.40.2). Threshold mRNA differential expression screening was specified as “log2 (fold change) >2 or log2 (fold change) ← 2.” ClusterProfiler package (version: 3.18.0) in R was employed to analyze the gene ontology (GO) function of potential targets and enrich the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway [19].
2.6. Correlation analysis of HSPA5 expression with tumor immune infiltration-related cells and immunoassay site genes
Accurate evaluation of immune scores for these tumors was performed using Immuneeconv, a R programme that incorporates the most recent TIMER algorithms. Using TIMER methods, we analyzed heat maps of the Spearman connection between HSPA5 and genes associated with immune examination locations and immunological scores. TISCH was used to obtain immune cell-related single-cell data.
A
HSPA5 Expression Level (log2 TPM)
10.0
7.5
5.0
ACC.Tumor (n=79)
BLCA.Tumor(m408)
BLCA.Normal (n=19)
BRCA.Tumor (n=1093)
BRCA.Normal (n=112)
BRCA-Basal.Tumor (n=190)
BRCA-Her2.Tumor (n=82)
BRCA-LumA.Tumor (n=564)
BRCA-LumB. Tumor (n=217)
CESC.Tumor (n=304)
CESC.Nonmal ( == 3)
CHOL. Tumor (n=36)
CHOL Normal (n 9)
COAD.Tumor (n=457)
COAD.Normal (n-41)
DLBC.Tumor (n-48)
ESCA.Tumor (n-184)
ESCA.Normal (n=11)
GBM. Tumor (n=153)
GBM.Normal (n=5)
HNSC.Tumor (n=520)
HNSC.Normal (n=44)
HINSC-HIPV+. Tumor (n=97)
HNSC-HPV -. Tumor (n=421)
KICH.Tumor (n=66)
KICH.Normal (I=25)
KTRC.Tumor (n=533)
KIRC.Normal (n=72)
KIRP.Tumor (n=290)
KIRP.Normal (n=32)
LAML. Tumor (n=1 73)
I.GG.Tumor (n=516)
LIHC.Tumor (n=371)
LIHC.Normal (n=50)
LUAD.Tumor (n=515)
LUIAD.Normal (n=59)
LUSC.Tumor (m-501)
LUSC.Normal (r=51)
MESO.Tumor (n=87)
OV. Tumor (n=303)
PAAD.Tumor (n=178)
PAAD.Normal (IF4)
PCPG.Turnor (n=179)
PCPG.Normal (n 3)
PRAD.Tumor (n=497)
PRAD.Normal (n=52)
READ.Tumor (n=166)
READ.Normal (n=10)
SARC.Tumor (n=259)
SKCM.Tumor (n=103)
SKCM.Metastasis (n=368)
STAD.Tumor ((=415)
STAD,Normal (n=35)
TGCT.Tumor (n=150)
THCA.Tumor (n=501)
THCA.Normal (n=59)
THYM.Tumor (n=120)
UCEC.Tumor (n=545)
UCEC.Normal (n=35)
UCS.Tumor (n=57)
UVM. Tumor (n=80)
B
C
HSPA5 log2(TPM+1)
Protein expression of HISPAS in Ovarian cancer p=2.52e-14
Protein expression of HISPAS in Colon cancer p=2.16c-11
3
3
Tumor Normal
0
2
2-
Z-value
Z-value
1”
0-
5
-1
-1
ACC(T=79;N=258)
BLCA(T=406;N=40)
BRCA(T=1101:N=572)
CESC(T=306;N=22)
CHOL(T=35;N=9)
COAD(T=455;N=820)
DLBC(T=48:N=929)
ESCA(T=163;N=1456)
-2.
-2
3
Normal (n=25)
Primary tumor (n=100)
.3
Normal (n=100)
Primary tumor (n 97)
TCGA+GTEx
Protein expression of HISPAS in Clear cell ROC p=3.90e-19
Protein expression of HSPAS in GBM p=4.28c-11
3-
3
*
HSPA5 log2(TPM+1)
12.5
1.
2
10.0
1
Z-value
1
Z-value
7.5
0-
1
-I
.1
5.0
-2.
-2
2.5
GBM(T=153;N=2647)
HNSC(T=504;N=44)
KICH(T=65;N=114)
KIRC(T=532;N=161)
KIRP(T=290;N=121)
LAML(T=150;N=0)
LGG(T=513;N=2642)
LIHC(T=371;N=276)
3
Nomal (n=84)
Primary tumor (n=110)
3
Normal (I=10)
Primary tumor (n=99)
Protein expression of HSPA5 in UCEC p=1.66c-26
Protein expression of HISPAS in LUAD
3
3-
p-2.68e-32
TCGA+GTEx
-
2-
2-
HSPA5 log2(TPM+1)
12
Z-value
1-
Z-value
1
10
0-
0-
8
-I-
-1-
6
-2.
-2.
4
LUAD(T=516;N=637)
LUSC(T=501;N=627)
MESO(T=87;N=0)
OV(T=376;N=180)
PAAD(T=179;N=332)
PCPG(T=181;N=3)
PRAD(T=498;N=297)
READ(T=165;N=789)
-3
Normal (n=31)
Primary tumor (n-100)
-3
Normal (n-111)
Primary tumor (n=111)
Protein expression of HSPAS in HNSC P=2.28e-14
Protein expression of HSPA5 in PAAD D=1.91e-26
3
12.5
TOGA+GTEX
-
2
10
HSPA5 log2(TPM+1)
12
1
7.5
Z-value
Z-value
9
0-
5.
.1.
2.5
6
-2.
2.
SARC(T=260;N=2)
SKCM(T=471;N=1810)
STAD(T=375;N=391)
TGCT(T=134;N=361)
THCA(T=512;N=712)
THYM(T == 120;N=2)
UCEC(T=545:N=177)
UCS(T=57;N=142)
UVM(T=80;N=0)
-3
Normal (n=71)
Primary tumor (n=108)
-2.5
Normal (n=74)
Primary turner (n=137)
TCGA+GTEx
2.7. Correlation analysis between HSPA5 gene expression and TMB or MSI, and the efficacy of immune checkpoint blockade
The publication of the article “The Immune Landscape of Cancer” by Vesteinn Thorsteinn et al., in 2018 served as the inspiration for the development of the TMB protocol. A 2017 publication by Russell Bonneville et al., titled “Landscape of Microsatellite Instability Across 39 Cancer Types,” is the source for MSI. Two-group data have been typically analyzed using the Wilcoxon test unless otherwise
| A Cancer | Pvalue | Hazard Ratio(95% CI) | B | 1.00 |
|---|---|---|---|---|
| ACC | 0.0111 | 2.81(1.27,6.24) | ||
| BLCA | 0.0071 | 1.51(1.12,2.03) | N probability | 0.75 |
| BRCA | 0.1377 | 1.27(0.93,1.75) | ||
| CESC | 0.0317 | 1.68(1.05,2.69) | survival H | 0.50- |
| CHOL | 0.8406 | 0.91(0.36,2.31) | Overall | |
| COAD | 0.6463 | 1.10(0.74,1.62) | 0.25 | |
| DLBC | 0.9082 | 1.09(0.27,4.35) | ||
| ESCA | 0.1526 | 1.43(0.88,2.35) | ||
| GBM | 0.0096 | 1.63(1.13,2.35) | H | 1.0 |
| HNSC | 0.0062 | 1.46(1.11,1.91) | ||
| KICH | 0.1010 | 3.73(0.77,18.03) | 0.8 | |
| KIRC | 0.0811 | 0.77(0.57,1.03) | probability | |
| KIRP | 0.0327 | 1.98(1.06,3.69) | 0.6 | |
| LAML | 0.1629 | 0.74(0.48,1.13) | survival | 0.4 |
| LGG | 0.0747 | 1.39(0.97,2.10) | Overall | |
| ALIHC | 0.0136 | 1.55(1.09,2.19) | 0.2 | |
| LUAD | 0.1551 | 1.23(0.92,1.65) | ||
| LUSC | 0.5034 | 1.10(0.84,1.44) | 1.00 | |
| MESO | 0.5230 | 1.16(0.73,1.85) | ||
| OV | 0.8630 | 1.02(0.79,1.32) | 0.75 | |
| PAAD | 0.3147 | 1.24(0.82,1.88) | probability | |
| PCPG | 0.6485 | 1.42(0.32,6.36) | survival | 0.50 |
| PRAD | 0.5032 | 1.55(0.43,5.57) | ||
| READ | 0.3247 | 0.67(0.30,1.49) | Overall | 0.25 |
| SARC | 0.2002 | 1.30(0.87,1.92) | 0.00 | |
| SKCM | 0.5885 | 1.08(0.82,1.41) | ||
| STAD | 0.8321 | 1.04(0.75,1.44) | 1.00 | |
| TGCT | 0.4994 | 2.20(0.22,21.81) | ||
| THCA | 0.1451 | 2.13(0.77,5.89) | 1.75 probability | |
| THYM | 0.6010 | 1.45(0.36,5.93) | ||
| UCEC | 0.3657 | 0.823(0.54,1.25) | survival | 1.50 |
| UCS | 0.2540 | 1.48(0.76,2.89) | ||
| UVM | 0.0082 | 3.52(1.38,8.97) | Overall 0.25- |
L
HISPAS
1.0
HSPA5
Low
High
Overall survival probability
0.8-
0.6-
0,4
ACC
P=0.0111
BLCA
0.2
P=0.0071
0
1000
2000
3000
4000
0
1000
2000
3000
Time (days)
4000
5000
Time (days)
HSPA5
1.00
HSPA5
Low
High
Overall survival probability
High
0.75
0.50
0.25
CESC
GBM
P=0.0317
0.00
0,0096
0
2000
4000
6000
0
500
1000
1500
2000
2500
Time (days)
Time (days)
HSPA5
1.0.
HSPAS
Low
High
Low
Overall survival probability
High
0.8-
0.6-
IINSC
KIRP
P=0.0062
0.4
P 0.0327
0
2000
4000
6000
0
1000
00 2000 3000 4000 5000 6 Time (days)
6000
Time (days)
HSPA5
1.0
HISPAS
Low
Low
High
0.8
High
Overall survival probability
0,6
0.4
0.22
LIHC
5
10
15
2022
P=0.0136
0.2
UVM
P=0.0082
Hazard Ratio
0
1000
2000
3000
0
500
1000
1500
2000
2500
C
Time (days)
HNSC
Time (days)
ACC
1
F
u
High em
₮
High emp
Alive
III
Alive
II
M
.60
TW
Low onp
M
Dead
Dend
540
TV
Age
Omder
PINM_stage
HISPAS
Statını
Ago
Gender
pTNM_stage
HISPAS
Status
BLCA
KIRP
0-60
F
II
F
haiph ong
Alne
0-10
High eng
I
Alive
M
M
Dead
>60
LƯƠN EXP
IV
Dead
Age
Kender
PTNM_stage
HSPAS
Mtafın
Ago
Gender
p’TNM_ntage
HESPAS
Status
CESC
LIHC
ASIAN
BLACK
M
F
1
Jagh ong
0-60
I
Bagh woop
Alive
0-60
Alive
WHITE
M
1
260
360
Dead
IV
Age
Race
PINM_stage
HSPAS
Stanın
Age
Gender
PTNM_stage
HISPAS
Stata
GBM
UVM
Alive
₸
0-60
1-60
I
II
Karla wow
Alive
Dead
M
360
>60
M
Dead
TV
Age
Gender
Bew Tumor
HISPAS
Statını
Age
Gender
PTNM_stage
HISPA5
Status
specified. The TIDE method was used to provide a prediction of the potential ICB reaction [20].
2.8. Immunohistochemistry
The tissue microarrays underwent several steps for processing. First, they were baked in an oven at 85 ℃ for 10 min. Then, they were soaked in xylene for 15 min and hydrated in a series of ethanol concentrations: 100%, 95%, 80%, and 70%. After that, the chips were treated with citric acid solution in an autoclave for antigen repair. Once cooled, the tissue chips were washed with PBS and incubated with hydrogen peroxide for 20 min. Following this, HSPA5 antibody (1:2000, ab21685) was added and incubated for 2 h at room temperature. After the completion of the above procedure, the tissue microarrays were washed 3 times with PBS and incubated with an immunohistochemical secondary antibody for 20 min at room temperature. The microarrays were washed 3 times with PBS again, and then DAB was added for staining, followed by staining with hematoxylin. Subsequently, the microarrays were dehydrated in a series of ethanol concentrations: 70%, 80%, 90%, and 100%. Finally, they were immersed in xylene for 8 min, and the microarrays
A
Kruskal-Wallis test p=3.6e-09
Kruskal-Wallis test p=1.9e-10
Kruskal-Wallis test p=3.9e-09
ACC
16
BLCA
+
16
CESC
.. ++
.
20
…
ns …
ns
ns
14
IS
HSPA5 expression
ns
HSPA5 expression
*
ns
TIS
12
HSPA5 expression
…
15
*
DIS
ns
12
ns
18
ns
10
8
10
8
5
4
Stage I Stage II Stage III Stage IV Normal
Stage I+II Stage III Stage IV Normal
Stage I Stage II Stage III Stage IV Normal
Kruskal-Wallis test p=2.9e-14
Kruskal-Wallis test p=3.2e-16
KIRP
Kruskal-Wallis test p=2.6e-49
16
HNSC
…
17.5
…
++++
..
ns ++++
ns ++++
20
LIHC
…
ns
14
15.0-
BIS
ns
ns
HSPA5 expression
ns
…
HSPA5 expression
ns
++++
HSPA5 expression
ns
…
ns
12.5
ns
15
2
1IS
ns
ns
LIS
ns
+
ns
10.0-
10
10
7.5-
8
5.0
5
Stage I Stage II Stage III Stage IV Normal
Stage I Stage II Stage III Stage IV Normal
Stage I Stage II Stage III Stage IV Normal
B
Normal
Cancer
Normal
Cancer
BLCA HPA038845
HPA038845
KIRP
CESC
HPA038845
HPA038845
LIHC
HNSC
HPA038845
HPA038845
GBM
were blocked after all of the aforementioned processes. Immunostaining intensity scores ranged from 0 to 3 (0, no reaction; 1, weak reaction; 2, moderate reaction; 3, strong reaction). Scales were scored as 1 (0%-25%), 2 (26%-50%), 3 (51%-75%) and 4 (76%- 100%). The final scores were obtained by multiplying the strength scores and the proportional scores. The results are as follows: 0-5: low expression; 6-12: high expression.
2.9. Statistical analysis
Expression levels of HSPA5 were compared in cancerous and healthy tissues using T-tests. With the use of a univariate Cox regression, we were able to calculate the HR and the p-value for the survival analysis. The p < 0.05 threshold was used in all statistical analyses. * p < 0.05, ** p < 0.01, *** p < 0.001.
A
Promoter methylation level of HSPA5 in BLCA
B
Promoter methylation level of HSPA5 in CESC
C
Promoter methylation level of HSPA5 in GBM
0,14
0.14
0.12
0.12
0.12
0.1
0.1
0.1
Beta value
Beta value
Beta value
0.08
0.08
0.08
0.06
0.06
0.06
0.04
0.04
0.02
P-2.07e-07
0.02
P=8.40e-01
P=9.72e-01
0.04
Normal (n=21)
Primary tumor (n=418)
Normal (n=3)
Primary tumor (n=307)
Normal (n=2)
Primary tumor (n=140)
D
TCGA samples
E
TCGA samples
F
TCGA samples
Promoter methylation level of HSPA5 in HNSC
Promoter methylation level of HSPA5 in KIRP
Promoter methylation level of HSPA5 in LIHC
0.14
0.12
0.12
0.12
0.1
0.1
0.1
Beta value
Beta value
Beta value
0,08
0.08
0.08
0.06
0.06
0.06
0.04
P=4.58e-09
P=3.39e-11
P-1.71e-02
0.02
0.04
Normal (n-50)
Primary tumor (n=528)
0.04
Normal (n=45)
Primary tumor (n=275)
Normal (n=50)
Primary tumor (n=377)
TCGA samples
TCGA samples
TCGA samples
G
Protein expression of HSPA5 (NP_005338.1:T648) in GBM
H
Protein expression of HSPA5 (NP_005338.1:S607) in HNSC
I Protein expression of HSPA5 (NP_005338.1:T643) in HNSC
3
4
2
2
3
1.
1
2
z-value
z-value
z-value
0
1
0
-1
0
-1-
-2-
-1
P=6.64c-01
P-3.16e-03
P-9.18e-01
-3
Normal
Primary tumor (n=99)
-2
-2
Normal (n=70)
Primary tumor (n=108)
Normal (n=70)
Primary tumor (n=108)
(1-10)
CPTAC samples
CPTAC samples
CPTAC samples
J
Protein expression of HSPA5 (NP_005338.1:T648) in HNSC
K Protein expression of HSPA5 (NP_005338.1:T648) in LIHC
2
3
2
1
1
z-value
z-value
0
0
-1
-1-
-2
P=1.29e-01
P=7.40e-05
-2
-3
Normal ( =- 70)
Primary tumor (n-108)
Normal (n=165)
Primary tumor (n=165)
CPTAC samples
CPTAC samples
3. Result
3.1. Expression of HSPA5 mRNA in tumor
First, we investigated the frequency of HSPA5 mRNA expression in 34 different types of common human malignant tumors using data from the TIMER database [21]. Our findings revealed that BLCA, BRCA, CESC, CHOL, COAD, ESCA, GBM, HNSC, LIHC, LUAD, LUSC, STAD, and UCEC exhibited significantly higher HSPA5 mRNA expression levels compared to the corresponding normal samples (Fig. 1A). Furthermore, we also analyzed HSPA5 expression using TCGA + GTEx data and observed that the expression of HSPA5 mRNA in most tumors remained significantly higher than that in normal tissues, except for LAML, MESO, PCPG, SARC, THYM, and UVM (Fig. 1B). To further investigate the expression of HSPA5 protein in tumors, we examined the CPTAC database. Our analysis revealed that the expression of HSPA5 was significantly higher in OV, COAD, KIRC, GBM, UCEC, LUAD, and HNSC compared to normal tissues. However, in the case of PAAD, the expression of HSPA5 in CPTAC was significantly lower than that in the corresponding normal tissues, possibly due to the smaller sample size (Fig. 1C).
3.2. Prognostic analysis of HSPA5 function in tumor
Next, we looked at how HSPA5 expression was linked to survival rates across different malignancies. Survival indicators in our analyses included OS, PFS, DFS, and DSS. Cox regression analysis showed that HSPA5 expression was significantly associated with OS in 8 of the 33 cancers, including ACC, BLCA, CESC, GBM, HNSC, KIRP, LIHC, and uveal melanoma (UVM). In all eight tumors, the
A
B
5%
HSPA5
ReSeq:NM_005347
CCDS:CCDS6863
Alteration Frequency
4%
Ensembl:ENST00000324460
Uniprot:BIP HUMAN
3%
2%
1%-
Structural variant data +
ACC + + +
BRCA + + +
*
+
*
KICH + + +
COAD + + +
GBM + + +
HNSC + + +
ESCA + + +
LAML + + +
THCA +++
+
+
LIHC + + +
*
LUAD + + +
TGCT + + +
KIRC + + +
KIRP + + +
UCS + + +
THYM + + +
UVM + + +
MESO + + +
+ +
Mutation dala
3
*
+
+
*
*
CNA data
*
UCEC
SKCM +
BLCA
OV
PAAD
PRAD + STAD
LUSC + +
LGG
SARC
CESC
PCPG
CHOL
DLBC +
C
D
E
100%
100%
100%
Altered group
Altered group
Altered group
90%
90%
Probability of Overall Survival
Unaltered group
Unaltered group
90%
Unaltered group
80%
Probability of Overall Survival
80%
Probability of Overall Survival
80%
70%
70%
70%-
80%
60%
60%
50%
50%
50%
10%
40%
40%
30%
30%
30%-
20%-
20%
20%
10%
ACC
BLCA
10%
10%
HNSC
Logrank Test P-Value: 8.270e-3
Logrank Test P-Value: 0.193
Logrank Test P-Value: 0.411
0%
a
10
20
0 30 40 50 00 70 80 90 100 110 120 130 140 150 Overall Survival (Months)
0%
D
10
20
30
40
50
60
70 80 90 100 110 120 130 140 150 160
0%
0
10
0
30
40
50
60
Overall Survival (Months)
70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 Overall Survival (Months)
F
G
100%
100%
Altered group
Altered group
90%
Unaltered group
00%
Probability of Overall Survival
Unaltered group
50%
Probability of Overall Survival
80%
70%
70%
50%
60%
50%
50%
40%
40%
30%
30%
20%
20%
KIRP
LIHC
10%
Logrank Test P-Value: 0.772
10%
Logrank Test P-Value: 0.950
0%
0
10
20
30
40
50 60 70 80 90 100 110 120 130 140 150 150 170 180 190 Overall Survival (Months)
0%
D
10
20
30 40 50 60 70 80 90 100 110 120 Overall Survival (Months)
prognosis of patients in the HSPA5 high-expression group was worse than that in the HSPA5 low-expression group. Kaplan-Meier survival curves were used to demonstrate the results (Fig. 2A and B). Based on the above conclusions, we further analyzed the cor- relation between HSPA5 expression and patient age, sex, race, and TNM stage in eight tumors, which revealed a significant correlation to poor prognosis in overall survival in Fig. 2A. Using Sankey diagrams, we found that in these eight tumors, there were always more deaths in samples with high HSPA5 expression than in samples with low HSPA5 expression (Fig. 2C). Finally, the correlation between HSPA5 and PFS, DFS, and DSS in these 33 tumors was analyzed by Cox regression. HSPA5 expression was significantly correlated with PFS in ACC, BLCA, CESC, HNSC, LGG, LUSC, THCA, and UVM (Supplementary Fig. 1A), and was significantly correlated with DFS in HNSC, LUSC, mesothelioma (MESO), and THCA (Supplementary Fig. 1B). Moreover, HSPA5 expression was significantly correlated with DSS in ACC, BLCA, ESCA, GBM, HNSC, KIRP, LGG, LIHC, LUSC, and UVM (Supplementary Fig. 1C).
3.3. Clinical significance and protein expression of HSPA5 in different tumors
The analysis of HSPA5 mRNA expression in various tumors and its prognostic significance reveals its essential role in ACC, BLCA, CESC, GBM, HNSC, KIRP, LIHC, and UVM biology. Further analysis was conducted to examine the expression of HSPA5 in different stages of these tumors. The results demonstrated significant differences in HSPA5 expression across stages of ACC, BLCA, CESC, HNSC, KIRP, and LIHC. In ACC, the expression of HSPA5 was significantly higher in Stage IV compared to Stage I and Stage III. In BLCA, the expression of HSPA5 in Stage III and Stage IV was significantly higher than that in Stage I + II. In CESC, HNSC, and KIRP, the expression of HSPA5 was higher in all stages (I, II, III, and IV) compared to normal samples. Similarly, in LIHC, HSPA5 expression was significantly
A
ACC
B
BLCA
KEGG (Lập)
tiO(Ii
KELKi ILE)
GOTUP
miNA surveillance pa.brway
·
vinil transcription
Vimal protein irleration with
L’biqukin mediasod proczolysis
viral gere expression ribesame leagazin
cylokrie and solo sine receptor Viral ryocarii lis
skin developesent
response lo interferon s.r.a
Spl coorTIL
Type 1 diabetes mellitus
Ast itive regulation of cytokine production
Siricnella infection
ribouclevprotein ova,les. biogenesis regulatica of mRNA metabolic process
Systent’e lupus erythematisus
pusilive regulation of cell cohesion
Ribosome
lng 10tadji s.)
RNA denduion
Coux:
rRNA processing
Caan
Staphylococcus aureus infection
noutrophil degranu ation
Rheumatoid a thritis
neutrophil chemolaxis
-4
I’ndeki processing i endup agric reticulum
9
ERNA melabolk process
Protein digestion and absorpilot
neutrophil activation involved in immune response meg live nquellior if iramireg syren process
Panercadie cuncer
=
procein torgauny
Nackencytoplasmic Ininsport
prolracine medmed ubiquitin depeudes, protein
Tiagotore
2
230
Leisbor, animals
” C
mycloid leukocyte migration
Neleh signa.ing pi.fmvey
projeasonal prueit calabolit process
Iaman T cell leucemia virus | infection
meyelidl la.kocyte differentiation
1
Mitoptagy - animal
-kg131)aģjusq
nuclear-orunseribed maRNA catabolic process
neRNA processing
lletnalupvielic vell lineage
-kg KXpaljas;
leukocyte chamotaxis Paulocyte migration
Hippo signa ing pathway
I’mlucyiuris
noKNA metabolic pnicevs
keltín-alivs!
Gratt-versus-host disease
Ligt
DNA replication
maRNA cutubelic process
Lued aubesiun
2
granulocyte chamotardis
5
Celorcetal cancer
hi stone modification
FCM-receptor Interaction
:
extraccilular struchar myas iration
extracellular matrix organization
Cuoricryeluid leu seinia
om en chnimatin modification
Cumplerer, amal emgulation carteles
2
$
2
Cellules senescence
RNA splicing, via fransesezrification reactions
Call adhesion moleculas
epidomis development
Cell eyelo
RNA splicing
Aaleiemune Thymic digrave
collagen libril organization
?
Autoplay animal
RNA calaboli prices
Amigen processing and presentation Annehiasis
cellu’a’ response to ‘interferon garima
Amyotrophic lateral sclerosis
DNA replication
Cell-substrale ceteston
era Rax dese Farc’man E.min
Allogruft rejection
antimicrobial hursoral response
-
nacamindi Kati>
C
GBM
KICK; (Lp)
CESC
D
Viral protein interaction with
GO (LP)
KFGG partyway
GOtem:
cytidine and cytokine reeepoor
respuese lo vitanin D
PSJ stur Ling pullway
INF signaling puhway
saponse to unfolded protein
“Transcriptional misregulation in cancer
response to atcyyen levels
Rosarise In hypoxia-
Small cell lung cicer
response lu riolecuir of baader al origin
INFagra ing paliway
response lo glecuce licoid-
Rheumnulvid ar.brit>
response to lipopolysacharido
Small cell big cunver
response to decreased oxygen levels-
Pertussis
vespoest lat chiisme’sine
Rheumatoid ardyritis
noutrophil migration
Retain spraling pathway
Response In arrioslavil
Count
NOD Like zeurptor signaling puibovary
neutrophil chemolacis
Proteoglycans in cancer
positive regulation of magiogenesis-
NP-kappo B signaling pathway
kelspaclasse
bgl 2tp-aljuc)
assification-
MickNAS in cmterr
humoral immune response
Protein digestie and absorption
Lipid end cfacrosclerosis
.. 6
reyoulive regulation of peptidase net vily negative regulativa of ly drelese okuvily
Teginnellosis
granulocyte migration
PIJK Aki sigra ing pathway
Kapes i sorcomo-asne ineć lempeseiras
grannyte chenelais
N’-kappa D signaling puiliway
9
multi-multicellular organismn procona-
somcellules structure organization
Kapusi sarcina associated herpesvirus Inindien
Matt
#
publim
zalrocellu ar matrix papanization
Reaktion of primary gerun Mayer.
kellízadivx]
II-17 signaling partway
Iluerin piquillomini’ts nectior
endopluatnic roliceluen enfolded protein responso
-a
IL-17 signaling puiliway
Le, Tikpalin-1)
fomale preg nancy1
HI M
Focal adhesion.
M
Coax:
chemokine-mediated signaling pathway
Human papillomavirus incostioe
4
calmaallelar x radias reprisal in
Helicobacter pylori Intector.
+ 2
cellules response to unfohled protein
*
Fucal athesier
6
extracellular matrix organization-
M
LCM-Taspior inkerator.
:
calinlar resporec to chemokine
ECM-receptor Interaction.
endodermal cell differentiation
Cyinking zyinkine receptre interaction
4
Bladder ca :22r
Cheiroline signaling pu.bwvny Amochasis
topologically inconeci protein
etalulema fonction *
microbial Furioral
antimicrobial humoral immune response
Amochiasi
African trypancsomiasis
endederin development -*
MGR RACHE; sheclicks per trovare
mediated by orti microbial peptide
IREI-mediated infolifed protein response
AGE RAGE styrking palliway in diabete complications
collagen rctahn ic process
in diebelic om aplications
collagen libril regarisation-
626 836
Q15 45: RIS Larichomore Este
340 633 338
E
HNSC
F
KIRP
KIXKi (LE)
00(h)
KIKKi prihva)
p53 signaling pullway
Tinoplasmosis
00 xenn
Thy roid bocrime synbasis
Inanamar brine receptor protein, serijpethegarine
Small cell lung cancer
response to transforming growth factor hera
Relaxia signaling porl way
regulation of’cel. substrale adhesion
uslucologrowth factor bera
Regubakri of artin cyhusky louri
xcraidosmosome assembly
Thyroid hormone signaling pathway
gastrulasica
Sonoll cell lung cancer
ME SUIS
Proteoglyceras in cancer
formation of primary gerin leyer
Regulation of setia cytoskeleton
PUIK Akt sigmalire pall ve.ne
esuravelkilar structure orunianica
Pro eoglycan in ciner
retine vascularure development in camera-type eye
Protein digestion and absorption
Protir processing in casoplatili
megetve lo lepologically xicorrect protein
bug. Ogg.adljuset
micidum
0;
Couit
MicroRNAs in caover
extrzecTular nieari’s organizacica
Trabalemal cell differentiation
Phassume
regulation of transme yleinen metal ion
IL-17 siganling pathway
kElDía -Juni
Count
Pancreatic cancer
tous
Ily permph & cardiomyopathy
endoderm. kermatica
endoderm development
.
PISK Akl sunalny pathway
Human papillomavirus infection
pas trans ational protein modifica in
Ilumen papillomavirus inlechun
2
imaintenance of blood-brain barrier
Hematopoiede cell linzege
4
conredline lissac desopment
2
collagen meubule procese
.
Gap junction
Focal adhesion
interin mediated ahauling pabrody
kop
L’M recepler intradiva
-legiiBp.ocjanbi
gland ir.orphogenesis
Com
Focal adhesion
colagen .br l organizasica
EÇA mamite i prachin
extracellular sinecure ungsuiza.jun
”
Dilmed curdisnyoparty
growth Leber beta sti pules
Bladder Cancer
.*
extracellular matrix orgerization
Bladder cancer
Count
cel xabarale unchien sinsmission
Arbyihmoyenne Derer
D
cellu ar respurse layryaldel protein
5
Bacterial invasion of epil telial cells.
cell-substrat junction assembly
Atriplarogaine ceapa in
*
col’-substrate adesica
Apclin signaling pathway
pellier response to
Dell matrix adresien
Amuebiasis
11
Dopoles ly juvenil pelein Tell”saharaleachesun
2
at
Arnica51
:
AGE-RAGE sigling podway
cell jaaction assembly
AGE-RAGE signaling pathway in diabetic complications
ezil-nutrise adhesion
in diabetic coraplications
cartilage development
cell.allusion medalt by irbegrin
bone remodeling
Forkant Biển
Leteimuni Batia
Intamin Kais
G
LITIC
KIKKI (Lp)
GOLFI
H
KEGG (I’m)
UVM
00(lp:
Viral peorzin interaction with
respeise te unfolded protein
Viral wwyncarditis ..
type I interferen signaling parlay
Stoking and sypoland Toseeor Toll-like receptor signaling pathway
regulation of mitotic sister chrecnatid separation
Type 1 diabetes mellitus.
response lo unfolded proleir.
Toxoplasmosis.
response to type 1 interferor
TNF signaling: pathway
1b17 cell differentiation ..
Rheumnu.oid anhrais
regulation of peutas Stephise Incenticyt
Thl and Th2 coll diffaeriaior .
response to logologically nemed proleir
response to Loicfiren-pa-na
CinaT
Psalcin export
regulation of’ chromosom - et paratikk. SPENTAS THICK OFFUHLOU hun vald mien respmiss
Protein processing in cadoplasmic.
prokaryne mediale thajafin deprecis Enviem
Ictiechas
catabolic process
Queytt Inditras
sincerely wer signaling pathway
Phagesconc.
pokasomal pedein catabolic pinces.
Pertussis .
inkricren-gunine-modialed signaling pabay.
0
Nitrogen ractabolism’
NF kappa B signaling parliway
Count
Imkclic: natid separation : sister chiny har na pe tonight mimic coll cycle chuckpn nt
legitipause)
melaphase mar ane Jos tion is cell cycle Healdas membre cell cycle
NOD-like receptor signaling pathway IJpid and athenschiusis.
kglbip.sciasty
establishment of uriel e localostior
.
.
Lipid erd albensc lerosis
1.8
axlamembrane dyston orgievizatior.
+
12
LegionzTosis
II. 17 sholing pushway
2
1.4
Lcisameniusis.
D
4
¥
nellalar response to type l’interferor
cellu a response to Icenteron-
integrated stress respirise signaling
Horari T cell leukem’a virus | infedior.
llepa.itks 13
LogiCup.adiasly
Hematopoietic cell lineage.
1
-laglit vadjus)
=
REPORTAGE
creloplasmic reticulum un allal protcin response
Chromosome separatiMi
Craft veros host disease
3D
Lpe.cin-Barr virus Infection. .
artigen processing and presents Fly Seffe
cellular response to alohled prolei
Cal
ahtigen, v & MHC class I
₹
Cytokine-cytokine iscedfor internetica
1*
Cell adhesion molecules.
Chemakins: signaling patlowvay
topologically jocorreci protein
Autoinenune thy roid dievitee .
E
artimicrobial homoral response
3
Cellular senescence
.
Celleyale
PLILK-med:mued un.folded protein response
4
Auth na ..
artigen processing
perlide cultiver
Binkler wir
IREI-mediated unfolded procein response
Anliver processing and presentation .
IR maslows signaling pathway
.
Allegraft rejection +
Amoebimis
urligen processing and
I presentafor.
CHI MEN Kanichtle. Kilis
Tela
Finition Italia
Tullian Pris
higher in Stage I than in Stage II (Fig. 3A). We further clarified the expression level of HSPA5 protein in these tumors from The Human Protein Altas, showing the HSPA5 protein in BLCA, CESC, HNSC, KIRP, LIHC, and GBM tissues to be significantly higher than in the normal ones (Fig. 3B).
3.4. Analysis of HSPA5 methylation and phosphorylation levels in different tumors
Using the UALCAN and TCGA datasets, we looked into the DNA methylation and HSPA5 phosphorylation levels in these eight tumors. BLCA, HNSC, UCEC, KIRP, and LIHC revealed substantially lower HSPA5 methylation levels compared to normal tissues in the UALCAN database, however CESC and GBM showed no significant change from normal tissues (Fig. 4A-F). Subsequently, Primary tumour tissues and normal tissues were compared for their HSPA5 phosphorylation levels. In HNSC, HSPA5 phosphorylation at S607 was considerably decreased compared to levels seen in normal tissues. In LIHC, the phosphorylation level of HSPA5 at T648 was significantly lower than that in normal tissues. Moreover, the phosphorylation level of HSPA5 in other tumors was not significantly different from that in normal tissues (Fig. 4G-K).
A
Correlation
B
0.50
* p<0.05
0.25
UVM
**
**
**
**
**
** p < 0.01
0.00
-0.25
*** p<0.001
LIHC
HỆ HỘI
ĐỘNG ĐỘNG
**
**
*
**
0.50
KIRP
**
#
**
**
**
云南
**
T cell CD8+
*
*
HNSC
*
*
**
*
T cell CD4+
**
TIMER
Neutrophil
2):
**
GBM
**
Myeloid dendritic cell
**
CESC
*
*
Macrophage
BLCA
**
**
**
**
**
**
**
**
B cell
*
**
ACC
ACC
BLCA
CESC
GBM
HNSC
KIRP
LIHC
UVM
CD274
CTLA4
HAVCR2
LAG3
PDCDI
PDCDILGZ
SIGLECIS
TIGIT
C
F
BLCA_GSE130001
HSPA5
HNSC_GSE103322
HSPA5
3
Mast
-3
4
Fbrob aste
Ennathaliet
Caltype trajor-inaagel
CDAT.
-4
Fibotans’s!
Celype timajur-ineaua)
Myocyle
MonoManiu
5
ombrotdas
CEPT
-
Myulibroslasiz
COSTOOnly
Erdochsiial
-
· Mpriserhink
.
CD TER
bly ant
PARDe
.
2
grant
Nonstegs
TP-3CH15
-2
1
Plasma
. #5.8
”
niżolhalial
0
0
D
CESC_GSE168652
HSPA5
G
LIHC_GSE98638
HSPA5
dothellal
&
CDAT
$
-15
Monn Macris
50
ablast
3.0
Celtype (majar-Ineage)
CUM
-2.5
S
Caltype (mejor-losepo)
-2
ficaTate storral alla
. COSTIN
2.5
Malignant
aprocess
-1.0
COAT
COFTer
-2.0
Việt Hunni
Cher
witry
15
TarGli
ENG
Trop
1.5
-10
-1.0
Jamesdal stromal cells
03
-C.S
E
Glioma_GSE103224
HSPA5
H
UVM_GSE138433
HSPA5
bro Macro
=
-3.0
Neuron
1
*
-4
08-lkn Mpignant
Cellype (mejor-Ineage)
2.5
AC-The Mallgrant
part
Endomalial
Malignant
Galtype (mejoriroage)
JE
AG ike Malign
· Mike Ma’am
MAL.HtF
*
A
CUS
· KlheMia
COBT
Encencia
Mainart
-1.5
DE4105 pm
-
· Mare/Kano
1.0
€+
.
EnDanial
00
3.5. Mutation analysis of HSPA5 in different tumors
We investigated HSPA5 mutations in different tumors using the cBioPortal database. In Fig. 5A, Mutation is shown in green, amplification in red, and a large deletion in blue. Among ACC, BLCA, CESC, GBM, HNSC, KIRP, LIHC, and UVM, HSPA5 had the highest mutation frequency in BLCA. Mutations were also predominant in CESC, and LIHC. HSPA5 was mainly amplified in ACC, GBM, HNSC, and KIRP. Subsequently, we performed a three-dimensional structural analysis of HSPA5, depicted by a schematic (Fig. 5B). Finally, an analysis of the impact of HSPA5 mutations on the overall survival of patients with these tumors showed them to be significantly related to the poor prognosis of ACC patients, despite no significant difference in other tumors (Fig. 5C-G).
3.6. GSEA analysis of the potential role of HSPA5 in tumors
To further analyze the pathogenic mechanism of HSPA5 in different tumors, we used KEGG analysis and GO analysis. In the ACC, BLCA, CESC, GBM, HNSC, KIRP, LIHC, and UVM, we divided the samples into two groups depending on HSPA5 high and low expression, for differential analysis, and the obtained differentially expressed genes were analyzed for gene enrichment analysis. In ACC, several cell cycles signaling pathways related to HSPA5 were significantly enriched with upregulated gene sets, suggesting that this gene may affect ACC progression by affecting ACC cell cycle. (Fig. 6A). GSEA results of down-regulated gene sets were also enriched in various signaling pathways, such as PPAR signaling pathway (Supplementary Fig. 2A). In BLCA, GSEA results of up- regulated gene sets indicate that HSPA5 regulates Phagosome, Focal adhesion, and immune-related signaling pathways (Fig. 6B), the results of the down-regulated gene set enrichment analysis were also enriched for the PPAR signaling pathway (Supplementary Fig. 2B). In CESC, the enrichment analysis of up-regulated gene sets revealed several well-known signaling pathways, such as TNF, NF- Kappa B signaling pathway (Fig. 6C). The results of down-regulated gene set enrichment analysis indicated that the Estrogen signaling pathway may be regulated by HSPA5 (Supplementary Fig. 2C). In GBM, due to the low number of down-regulated genes, we only performed functional analysis on up-regulated genes. The HSPA5-mediated tumor mechanisms were similar to those in CESC (Fig. 6D).
A
Correlation(TMB)
B
Correlation(MSI)
ACC r=0.39 p=4.38e-04
-log10(p-value)
-r ==- 0.04 p=0.726
-log1((p-value)
BLCA -r=0.16 p=0.002
9
T=0.06 p=0.233
5
4
6
3
HNSC -r=0.03 p=0.479
r =- 0.03-p=0.541
2
3
1
LIHC -r=0.02 p=0.786
r =— 0.02 p=0.714
GBM -r=0.32 p=7.79e-17
Correlation
T =— 0.38 p=2.25e-24
Correlation
0.1
0.1
KIRP -r =- 0.12 p=0.04
0.2
r-0.01 p-0.816
0.2
0.3
0.3
CESC -r =- 0.15-p=0.077
r=0.02 p=0.721
UVM r =- 0.2 p=0.076
r=0.14 p=0.226
0.2
0.0
0.2
0.4
-0.2
0.0
0.2
C
D
E
F
Responder
True
7
13
Responder
True
50
18
Responder
True
37
58
Responder
True
97
211
False
33
26
False
53
98
False
116
95
False
230
122
wilcox.tests p=0.69
wilcox.tests p=1.6c-08
wilcox.tests p-0.03
3
wilcox tests p=6.lc-22
ns
.
2
.I
3
-
2
2
2
TIDE Score
1
TIDE score
TIDE. score
2
TIDE score
*
1
1
Q
0
0
0
-1
-1
-1
..
ACC
-2
-2
BLCA
-2
-2
GBM
.
.
CESC
G
High HSPA5 exp
Low HSPAS exp
Low HSPA5 exp
High HSPA5 exp
Low HSPA5 exp
H
High HSPA5 exp
Low HSPA5 exp
High HSPA5 exp
I
J
Responder
True
53
99
Responder
True
40
63
Responder
True
47
64
Responder
True
11
16
False
199
5
False
105
82
False
13
121
False
29
24
wilcox.tests p=1.2c-07
wilcox.tests p=0.077
wilcox.tests p=0.01
wilcox.tests p=0.26
3
ns
#
4
ns
2
2
2
TIDE score
..
TIDE score
TIDE score
1
1
TIDE score
2
:
0
0
.
U
0
:
-1
-1
..
-2
HNSC
-2
KIRP
LIHC
UVM·
-2
-2
High HSPA5 exp
Low HSPAS exp
High HSPA5 cxp
Low HSPA5 exp
High HSPA5 exp
Low HSPA5 exp
High HSPA5 cxp
Low HSPA5 cxp
In HNSC, enrichment analysis showed the probability of HSPA5 affecting the PI3K-Akt and P53 signaling pathway, and the results of the enrichment analysis of down-regulated gene sets all indicated that the IL17 signaling pathway plays an important role (Fig. 6E, Supplementary Fig. 2D). In KIRP, the number of down-regulated genes was similarly low. GSEA showed significant enrichment of various well-known signaling pathways, such as PI3K-Akt and AGE-RAGE signaling pathway (Fig. 6F). Similarly, we hypothesized that HSPA5 may affect LIHC progression by affecting the cell cycle and immune microenvironment (Fig. 6G, Supplementary Fig. 2E).
A
Normal tissue
NMIBC
MIBC
HSPA5 expression
₹10
×10
×10
×20
×20
×20
B
C
D
P<0.05
20
12-
-
50
3
HSPA5 expression
15
HSPA5 expression
10-
40
20
8
10
6-
30
group Dead Alive
5
4
20
34
43
0
2
10
0-
Tumor (n=51)
Normal (n=49)
Normal (n=41)
Tumor (n=41)
0
High HSPA5 exp
Low HSPA5 exp
E
F
3000
≤65
≤G2
Low
male
≤5
≤T2
Survival time
2000
1000
>65
>G2
High
0
female
>5
>T2
Age
Tumor size
Tumor stage
Tumor grade HSPA5 expression
High HSPA5 exp Low HSPA5 exp
G
H
Points
0
20
40
60
80
100
1.0
Overall survival probability
Low HSPA5 expression
Tumor size
5
0.9
High HSPA5 expression
4
HSPA5 expression
High
0.8
Low
Total Points
0.7
S
40
80
120
160
200
Linear Predictor
0.6
-1.5
5
-0.5
0.5
1.5
2.5
1-year Survival Probability
0.5
Log-rank P < 0.001
0.9
0.8 0.7 0.6
3-year Survival Probability
0
500
1000
1500
2000
0.8
0.6
0,4
Time (Days)
5-year Survival Probability
Low
46
43
30
18
4
0.8
0.6
0,4
High
54
36
19
7
6
Finally, in UVM, enrichment analysis showed that HSPA5 could affect disease progression through NOD-like receptor, JAK-STAT signaling pathways (Fig. 6H, Supplementary Fig. 2F). In addition, we show a volcano map of HSPA5-related differential genes in these eight tumors (Supplementary Fig. 2G).
3.7. Correlation analysis of HSPA5 expression with immune cell infiltration and immunoassay sites
The association between HSPA5 expression and immune infiltration was investigated using the TIMER algorithm. HSPA5 expression showed significant correlation with immune infiltration in all tumor cells except for CESC. In ACC, HSPA5 was positively correlated with T cell CD8+ infiltration. In BLCA, HSPA5 was positively correlated with T cell CD8+, Neutrophil, Myeloid dendritic cell, and Macrophage infiltration. In GBM, HSPA5 was negatively correlated with B cell infiltration and positively correlated with Myeloid dendritic cell infiltration. In HNSC, HSPA5 was negatively correlated with T cell CD8+ and B cell infiltration, and positively correlated with Neutrophil infiltration. In KIRP, HSPA5 was positively correlated with Neutrophil, Myeloid dendritic cell, Macrophage, and B cell infiltration. In LIHC, HSPA5 was positively correlated with T cell CD4+, Neutrophil, Myeloid dendritic cell, and B cell infiltration. In UVM, HSPA5 was positively correlated with T cell CD8+ infiltration and negatively correlated with T cell CD4+ and Myeloid dendritic cell infiltration (Fig. 7A). Furthermore, we conducted an analysis on the correlation between HSPA5 and immune checkpoints. Our findings revealed a significant correlation between HSPA5 expression in BLCA and the expression of all immune checkpoint-associated genes. However, in the case of ACC, the opposite was observed (Fig. 7B). To explore the correlation of immune cell distribution and ratio with HSPA5 expression levels at the single-cell level, we obtained relevant data for BLCA, CESC, GBM, HNSC, LIHC, and UVM from the Tumor Immune Single-cell Hub 2 (TISCH2). This database does not contain ACC and KIRP-related data, so we only analyzed the other six tumors (Fig. 7C-H). In Summary, these results indicated that HSPA5 may be involved in immune regu- lation, which may influence immunotherapy response.
3.8. Analysis of HSPA5 correlation with TMB, MSI while predicting potential immunotherapeutic response using TIDE algorithm
TMB and MSI are two new biomarkers for assessing the effectiveness of immunotherapy. First, we investigated the relationship between HSPA5 expression and TMB, and we found that HSPA5 expression and TMB were positively correlated in ACC, BLCA, and GBM, while HSPA5 expression and TMB were negatively correlated in KIRP (Fig. 8A). We then analyzed the correlation between HSPA5 expression and MSI, and we found that only in GBM HSPA5 expression was negatively correlated with MSI. (Fig. 8B). High TIDE score, ineffective ICB therapy, and a short survival time are all indicators of tumor immune escape via the two mechanisms of
| Characteristics | Total(N) | Univariate analysis | Multivariate analysis | ||
|---|---|---|---|---|---|
| Hazard ratio (95% CI) | P value | Hazard ratio (95% CI) | P value | ||
| Age | 100 | 0.278 | |||
| ≤65 | 39 | Reference | |||
| >65 | 61 | 0.633 (0.279-1.436) | 0.274 | ||
| Gender | 100 | 0.880 | |||
| male | 81 | Reference | |||
| female | 19 | 0.921 (0.313-2.707) | 0.881 | ||
| Tumor size | 100 | 0.023 | |||
| ≤5 | 76 | Reference | Reference | ||
| > 5 | 24 | 2.713 (1.188-6.196) | 0.018 | 4.010 (1.666-9.654) | 0.002 |
| Tumor stage | 100 | 0.004 | |||
| ≤T2 | 81 | Reference | Reference | ||
| >T2 | 19 | 3.820 (1.634-8.933) | 0.002 | 1.446 (0.483-4.324) | 0.510 |
| Tumor grade | 100 | 0.112 | |||
| ≤G2 | 40 | Reference | |||
| > G2 | 60 | 2.052 (0.808-5.207) | 0.130 | ||
| Vascular invasion | 100 | 0.032 | |||
| No | 87 | Reference | Reference | ||
| Yes | 13 | 3.101 (1.215-7.917) | 0.018 | 1.187 (0.391-3.600) | 0.762 |
| Lymph node metastasis | 100 | 0.820 | |||
| No | 66 | Reference | |||
| Yes | 34 | 1.105 (0.468-2.610) | 0.819 | ||
| Recurrence | 100 | 0.032 | |||
| No | 62 | Reference | Reference | ||
| Yes | 38 | 2.465 (1.078-5.638) | 0.033 | 1.834 (0.682-4.935) | 0.230 |
| Distant metastasis | 100 | 0.153 | |||
| No | 90 | Reference | |||
| Yes | 10 | 2.383 (0.807-7.037) | 0.116 | ||
| HSPA5 expression | 100 | < 0.001 | |||
| Low | 46 | Reference | Reference | ||
| High | 54 | 7.240 (2.147-24.413) | 0.001 | 7.804 (2.148-28.360) | 0.002 |
dysfunction of tumor-infiltrating cytotoxic T lymphocytes (CTLS) and rejection of CTLS by immunosuppressors used in the tumor Immune Dysfunction and Exclusion (TIDE) algorithm [22]. On the basis of HSPA5 mRNA expression profile data, the TIDE algorithm was used to provide predictions about the degree to which individual samples will respond to immune checkpoint inhibitors. In BLCA, CESC, GBM, HNSC, and LIHC, the TIDE score of the HSPA5 high expression group was significantly higher than that of the HSPA5 low expression group. This suggests that patients with high HSPA5 expression have a worse outcome when treated with immune check- point inhibitors compared to patients in the low HSPA5 expression group. However, in ACC, KIRP and UVM, there was no significant difference in direct TIDE scores between the HSPA5 high and HSPA5 low expression groups (Fig. 8C-J). In conclusion, our results suggested that in most tumors, when HSPA5 is highly expressed, immune checkpoint blockade therapy has poor efficacy and survival is short after receiving ICB therapy.
3.9. Prognostic value of HSPA5 in bladder cancer patients by tissue microarray analysis
Based on these studies, we found that among the many tumors with significant differences in HSPA5 expression. It has been shown that HSPA5 has a significant impact on the prognosis of bladder cancer as well as the immune microenvironment. Furthermore, in addition to its prognostic value for bladder cancer, HSPA5 correlated with all of the immune checkpoint genes included in the study, and after grouping based on HSPA5 expression, we found that the difference in responsiveness scores to receiving immune checkpoint inhibitor therapy was most significant between the HSPA5 high-expression group and the HSPA5 low-expression group. Therefore, we chose to further validate its role in bladder cancer. We collected samples from 100 bladder cancer patients to further validate the prognosis and expression differences of HSPA5 in bladder cancer samples. Expression of HSPA5 was shown to be considerably greater in bladder cancer compared to normal bladder tissue using immunohistochemical staining, and expression of HSPA5 was further found to be significantly higher in muscle-infiltrating bladder cancer compared to non-muscle-infiltrating bladder cancer using the same method. (Fig. 9A-B). Furthermore, we compared the levels of HSPA5 expression in 41 pairs of bladder cancer and normal bladder tissues and found that, once again, the cancerous tissues expressed much more of this gene (Fig. 9C). In addition, we found that the ratio of dead to surviving population was significantly higher in the HSPA5 high expression group than in the HSPA5 low expression group (Fig. 9D). Patient survival is shown in a Sankey diagram alongside the frequency of high and low HSPA5 protein expression in bladder cancer samples of varying stage grades, ages, and other clinical features (Fig. 9E). Patient survival was also considerably lower in the high HSPA5 expression group compared to the low HSPA5 expression group (Fig. 9F). Univariate analysis showed that HSPA5 expression (p<0.001) was linked with overall survival, as were tumor size (p = 0.023), tumor stage (p = 0.004), vascular invasion (p = 0.032), and recurrence (p = 0.032). Multivariate study of bladder cancer patients revealed that tumor size (p = 0.002) and HSPA5 expression (p = 0.002) were independently associated with prognosis (Table 1). Based on the results of the multi-factor analysis we have drawn the line graphs (Fig. 9G). Finally, we subdivided individuals with bladder cancer based on HSPA5 expression, finding that those with high HSPA5 expression were more likely to have a worse prognosis than those with low HSPA5 expression (Fig. 9H). We have demonstrated that HSPA5 is highly expressed in bladder cancer and that its high expression is substantially associated with a poor prognosis.
4. Discussion
One of the most perilous illnesses endangering human health is cancer, the disease burden of which is increasing worldwide [23]. Understanding the molecular process of tumour genesis and progression and looking into possible new biomarkers for cancer diagnosis and prediction of cancer treatment outcomes is essential for cancer prevention and control methods. Malignant cells and invading immune cells are subjected to persistent ER stress caused by a number of metabolic and carcinogenic abnormalities in the tumour microenvironment (TME), which in turn constitutes a sexually active ER stress response that allows malignant cells to adapt to these stresses. At the same time, Cancer development is promoted by a coordinated set of immune regulatory systems [24]. HSPA5, as an ER stress-related gene, is highly expressed in various malignant tumor tissues such as colorectal [25], liver [26], and breast cancers [27], which may lead to malignant biological behaviors such as proliferation, invasion, induction of chemotherapy resistance and immune escape of tumor cells. However, there is a lack of thorough investigations of the HSPA5 functional role to highlight the similarities and variations between various malignancies that would provide key insights to bolster cancer prevention and develop individualized treatment plans. Recent research has placed a premium on genome-wide carcinomatosis analysis for early cancer identification and treatment in an effort to pinpoint gene mutations, RNA changes, and cancer drivers involved in cancer genesis and progression.
The expression and function of HSPA5 in all malignancies were analyzed for the first time in this work. Based on our findings, HSPA5 expression was much greater in tumors than in normal tissues (Fig. 1). In some cancers, the upregulation of HSPA5 was significantly associated with poor OS, PFS, DFS, and DSS (Fig. 2). Through the above analyses, we identified a vital oncogenic role of HSPA5 in ACC, BLCA, CESC, GBM, HNSC, KIRP, LIHC, and UVM, which also confirmed these eight tumors are our choices for carrying out the follow-up work. The methylation and phosphorylation levels of HSPA5 were significantly down-regulated in LIHC, suggesting an active HSPA5 status that may affect the progression of LIHC (Fig. 4). In addition, Gene enrichment analysis showed that HSPA5 might affect a variety of signaling pathways, leading to the malignant progression of tumor pairs, with the P53, TNF, NF-kappa B, PI3k- Akt signaling-related and immune-related gene sets being significantly enriched in multiple tumors. This observation indicates that HSPA5 can not only regulate the development of malignant tumors but also affect the tumor immune microenvironment to create a more conducive niche for tumor growth (Fig. 6). P53 is a major regulator of multiple cellular biological behaviors, including apoptosis, senescence, and autophagy [28]. The study by Kamil M et al. demonstrated that HSPA5 can regulate autophagy in lung adenocar- cinoma cells by affecting p53 localization [29]. One of the most investigated routes in relation to inflammation, metastasis, cell
proliferation, and cellular senescence is the nuclear factor kappa B (NF-KB) signalling system [30,31]. Interleukin-17 is also a factor closely related to inflammation. This indicates that HSPA5 may promote the progression of various malignant tumors by regulating inflammatory factors. In addition, GRP78 has been shown to improve the therapeutic efficacy of MSCs against hemorrhagic shock-induced liver injury via the NF-KB pathway [32]. Through PI3K/Akt signaling, HSPA5 has also been shown to improve the sensitivity of ovarian cancer cells to paclitaxel [33]. Moreover, several immune-related signaling pathways were enriched, demon- strating HSPA5’s crucial function in the immunological milieu of cancerous tumors. The enrichment analysis of down-regulated genes demonstrated, in addition, that HSPA5 has the ability to govern the appearance of malignant tumors as well as their progression via a variety of signaling pathways. For example, JAK-STAT, PPAR signaling pathway. The role of immune-related cells such as cancer-related fibroblasts (CAFs) in cancer development, treatment resistance, and disease transmission has been well-studied [34]. Recent research has indicated that activation of the UPR is necessary for the formation of a favorable tumor microenvironment, which includes the differentiation of cancer-associated fibroblasts (CAFs) [35]. Meanwhile, CAFs have been demonstrated in prior research to raise the expression of HSPA5, which is a factor that leads to the invasion of non-small cell lung cancer cells [36]. It is imperative that more research be conducted into the role and function of the UPR as a regulator of the tumor microenvironment and the actions of immune cells in order to ameliorate the immunological dysfunctions associated with cancer. Therefore, we investigated how HSPA5 regulated immune infiltration and immune checkpoint markers, which yielded beneficial results (Fig. 7). Our study showed that HSPA5 affected TMB and MSI in a variety of tumors, which led us to further use TIDE analysis to determine the effect of HSPA5 on the responders of ICB. The results collectively showed that HSPA5 can also affect ICB responders in various tumors. In conclusion, as a result of our research, HSPA5 has been identified as a candidate for a new prognostic biomarker in several types of cancer. It does this by modulating the immune milieu inside tumors. Our findings highlight the necessity of the development of therapeutic regimens targeting HSPA5.
Together, our findings suggest a critical function for HSPA5 in the immunological microenvironment and prognosis of bladder cancer. As a result, we choose to verify HSPA5’s expression and prognostic importance in samples from patients with bladder cancer. Based on our findings, HSPA5 is a promising predictive biomarker for individuals with bladder cancer. However, there are definitely caveats to our work, and further tests are needed to confirm HSPA5’s efficacy in large tumors.
5. Conclusions
HSPA5 expression upregulation correlates with increased immune cell infiltration and poor prognosis in a variety of malignancies. We also observed reduced levels of phosphorylation and methylation of HSPA5 in many types of cancer. HSPA5 expression was shown to be substantially linked with immunological checkpoint-related gene expression. Tissue microarray research verified that HSPA5 is substantially expressed in bladder cancer and has potential as a predictive biomarker for individuals with this disease. Future experimental investigations of HSPA5 expression and function-mediated immune cell infiltration in various cancer populations may give more significant insights to decipher the underlying tumor development mechanisms and develop novel treatments targeting HSPA5 to boost the therapeutic efficacy of immunotherapies.
Availability of data and materials
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.
Ethics approval and consent to participate
The studies involving human participants were reviewed and approved by the Ethics Committee of Nantong Tumour Hospital. Ethics approval number: 2022-053. The patients provided their written informed consent to participate in this study.
CRediT authorship contribution statement
YaXuan Wang: Writing - original draft, Formal analysis, Data curation. Jinfeng Wang: Writing - original draft, Conceptualiza- tion. Yang Liu: Writing - original draft, Data curation. XiaoLin Wang: Writing - review & editing, Funding acquisition. MingHua Ren: Writing - review & editing, Validation, Funding acquisition.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This study was supported by the Scientific research project of Jiangsu Provincial Health Commission (M2021005) and the Natural Science Foundation of Heilongjiang Province (No. LH2019H030).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.heliyon.2024.e27184.
References
[1] R.L. Siegel, K.D. Miller, H.E. Fuchs, et al., Cancer statistics, 2022, CA A Cancer J. Clin. 72 (1) (2022) 7-33.
[2] R.P. Graf, V. Fisher, J. Mateo, et al., Predictive genomic biomarkers of hormonal therapy versus chemotherapy benefit in metastatic castration-resistant prostate cancer, Eur. Urol. 81 (1) (2022) 37-47.
[3] S. Parashar, S. Ferro-Novick, Architecture of the endoplasmic reticulum plays a role in proteostasis, Autophagy 18 (4) (2022) 937-938.
[4] W. Jiang, L. Chen, X. Guo, et al., Combating multidrug resistance and metastasis of breast cancer by endoplasmic reticulum stress and cell-nucleus penetration enhanced immunochemotherapy, Theranostics 12 (6) (2022) 2987-3006.
[5] P. Wadgaonkar, F. Chen, Connections between endoplasmic reticulum stress-associated unfolded protein response, mitochondria, and autophagy in arsenic- induced carcinogenesis, Semin. Cancer Biol. 76 (2021) 258-266.
[6] M. Cerezo, S. Rocchi, New anti-cancer molecules targeting HSPA5/BIP to induce endoplasmic reticulum stress, autophagy and apoptosis, Autophagy 13 (1) (2017) 216-217.
[7] M.N. Cai, D.M. Chen, L.X. Xiao, et al., COLEC10 induces endoplasmic reticulum stress by occupying GRP78 and inhibits hepatocellular carcinoma, Lab. Invest. 103 (7) (2023) 100130.
[8] J. Wang, J. Lee, D. Liem, et al., HSPA5 Gene encoding Hsp 70 chaperone BiP in the endoplasmic reticulum, Gene 30 (618) (2017) 14-23.
[9] W .- T. Chen, G. Zhu, K. Pfaffenbach, et al., GRP78 as a regulator of liver steatosis and cancer progression mediated by loss of the tumor suppressor PTEN, Oncogene 33 (42) (2014) 4997-5005.
[10] F.M. Uckun, S. Qazi, Z. Ozer, et al., Inducing apoptosis in chemotherapy resistant B lineage acute lymphoblastic leukaemia cells by targeting HSPA5, a master regulator of the anti apoptotic unfolded protein response signalling network, Br. J. Haematol. 153 (6) (2011) 741-752.
[11] Q. Wang, S. Ke, Z. Liu, et al., HSPA5 promotes the proliferation, metastasis and regulates ferroptosis of bladder cancer, Int. J. Mol. Sci. 24 (6) (2023) 5144.
[12] T. Li, J. Fu, J. Cheng, et al., New progresses on cell surface protein HSPA5/BiP/GRP78 in cancers and COVID-19, Front. Immunol. 14 (2023) 1166680.
[13] J. Fu, C. Wei, J. He, et al., Evaluation and characterization of HSPA5 (GRP78) expression profiles in normal individuals and cancer patients with COVID-19, Int. J. Biol. Sci. 17 (3) (2021) 897-910.
[14] Y. Yuan, J. Fan, D. Liang, et al., Cell surface GRP78-directed CAR-T cells are effective at treating human pancreatic cancer in preclinical models, Transl Oncol 39 (2024) 101803.
[15] R. Lin, M. Ma, B. Han, et al., Esophageal cancer stem cells reduce hypoxia-induced apoptosis by inhibiting the GRP78-perk-eIF2x-ATF4-CHOP pathway in vitro, J. Gastrointest. Oncol. 14 (4) (2023) 1669-1693.
[16] S. Wang, W. Wei, Y. Yuan, et al., Chimeric antigen receptor T cells targeting cell surface GRP78 efficiently kill glioblastoma and cancer stem cells, J. Transl. Med. 21 (1) (2023) 493.
[17] J.L. Chen, Y.S. Tai, H.Y. Tsai, et al., Betulinic acid inhibits the stemness of gastric cancer cells by regulating the GRP78-TGF-$1 signaling pathway and macrophage polarization, Molecules 28 (4) (2023) 1725.
[18] Y. Zheng, N. Wang, S. Wang, et al., Cefoselis enhances breast cancer chemosensitivity by directly targeting GRP78/LRP5 signalling of cancer stem cells, Clin. Transl. Med. 13 (2) (2023) e1119.
[19] G. Yu, L.G. Wang, Y. Han, et al., clusterProfiler: an R package for comparing biological themes among gene clusters, OMICS 16 (5) (2012 May) 284-287.
[20] P. Jiang, S. Gu, D. Pan, et al., Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response, Nat Med 24 (10) (2018) 1550-1558.
[21] F. Chen, Y. Fan, P. Cao, et al., Pan-cancer analysis of the prognostic and immunological role of HSF1: a potential target for survival and immunotherapy, Oxid. Med. Cell. Longev. 2021 (2021) 5551036.
[2][2] Q. Wang, M. Li, M. Yang, et al., Analysis of immune-related signatures of lung adenocarcinoma identified two distinct subtypes: implications for immune checkpoint blockade therapy, Aging (Albany NY) 12 (4) (2020) 3312-3339.
[23] H. Sung, J. Ferlay, R.L. Siegel, et al., Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries, CA A Cancer J. Clin. 71 (3) (2021) 209-249.
[24] X. Chen, J.R. Cubillos-Ruiz, Endoplasmic reticulum stress signals in the tumour and its microenvironment, Nat. Rev. Cancer 21 (2) (2021) 71-88.
[25] N. Rahmani-Kukia, M. Zamani, P. Mokaram, ERMP1 facilitates the malignant characteristics of colorectal cancer cells through modulating PI3K/AKT/ß-Catenin pathway and localization of GRP78, Cell J 25 (7) (2023) 470-482.
[26] Y. Zou, H. Shi, H. Lin, et al., The abrogation of GRP78 sensitizes liver cancer cells to lysionotin by enhancing ER stress-mediated pro-apoptotic pathway, Cell Stress Chaperones 28 (4) (2023) 409-422.
[27] Y. Zheng, N. Wang, S. Wang, et al., Cefoselis enhances breast cancer chemosensitivity by directly targeting GRP78/LRP5 signalling of cancer stem cells, Clin. Transl. Med. 13 (2) (2023) e1119.
[28] Y. Stein, V. Rotter, R. Aloni-Grinstein, Gain-of-Function mutant p53: all the roads lead to tumorigenesis, Int. J. Mol. Sci. 20 (24) (2019) 6197.
[29] G. Yu, L.G. Wang, Y. Han, et al., clusterProfiler: an R package for comparing biological themes among gene clusters, OMICS 16 (5) (2012 May) 284-287.
[30] M. Haga, M. Okada, Systems approaches to investigate the role of NF-KB signaling in aging, Biochem. J. 479 (2) (2022) 161-183.
[31] S. Mirzaei, S. Saghari, F. Bassiri, et al., NF-KB as a regulator of cancer metastasis and therapy response: a focus on epithelial-mesenchymal transition, J. Cell. Physiol. 237 (7) (2022) 2770-2795.
[32] J. Han, D. Jia, H. Yao, et al., GRP78 improves the therapeutic effect of mesenchymal stem cells on hemorrhagic shock-induced liver injury: involvement of the NF-KB and HO-1/Nrf-2 pathways, FASEB J 38 (1) (2024) e23334.
[33] L.Y. Zhang, J.Y. Yu, Y.L. Leng, et al., MiR-181c sensitizes ovarian cancer cells to paclitaxel by targeting GRP78 through the PI3K/Akt pathway, Cancer Gene Ther. 29 (6) (2022) 770-783.
[34] Y. Zhu, X. Li, L. Wang, et al., Metabolic reprogramming and crosstalk of cancer-related fibroblasts and immune cells in the tumor microenvironment, Front. Endocrinol. 13 (2022) 988295.
[35] Y. Peng, Z. Li, Z. Li, GRP78 secreted by tumor cells stimulates differentiation of bone marrow mesenchymal stem cells to cancer-associated fibroblasts, Biochem. Biophys. Res. Commun. 440 (4) (2013) 558-563.
[36] T. Yu, Z. Guo, H. Fan, et al., Cancer-associated fibroblasts promote non-small cell lung cancer cell invasion by upregulation of glucose-regulated protein 78 (GRP78) expression in an integrated bionic microfluidic device, Oncotarget 7 (18) (2016) 25593-25603.