Screening of key genes and prediction of therapeutic agents in Arsenic-induced lung carcinoma

Abstract

BACKGROUND:

Evidence indicates that inorganic arsenic (iAs) can directly damage cells and result in malignant transformation with unclear complicated mechanisms. In the present study, we aimed to explore the possible molecules, pathways and therapeutic agents by using bioinformatics methods.

METHODS:

Microarray-based data were retrieved and analyzed to screen the differentially expressed genes (DEGs) between iAs-treated lung cells and controls. Then, the functions of DEGs were annotated and the hub genes were filtrated. The key genes were selected from the hub genes through validation in The Cancer Genome Atlas (TCGA) cohorts. Possible drugs were predicted by using CMAP tool.

RESULTS:

Two datasets (GSE33520 and GSE36684) were retrieved, and 61 up-regulated and 228 down-regulated DEGs were screened out, which were enriched in various pathways, particularly metabolism-related pathways. Among the DEGs, four hub genes including MTIF2, ACOX1, CAV1, and MRPL17, which might affect lung cancer prognosis, were selected as the key genes. Interestingly, Quinostatin was predicted to be a potential agent reversing iAs-induced lung cell malignant transformation.

CONCLUSION:

The present study sheds novel insights into the mechanisms of iAs-induced lung cell malignant transformation and identified several potential small agents for iAs toxicity prevention and therapy.

Keywords

Arsenic differentially expressed genes TCGA lung carcinoma bioinformatics

1. Introduction

Lung cancer is the most commonly diagnosed cancer and the leading cause of cancer death in the world [1]. Epidemiological evidence indicates that cigarette smoking, alcohol consumption, ionizing radiation, and chronic inflammation are risk factors. Besides, genetic variations can also contribute to lung carcinoma risk [2]. Moreover, environmental carcinogens in the environment might have a long-term effect on human health. With the development of industry, heavy metal exposure has become a problem that can not be ignored.

Evidence showed that cadmium exposure may result in a number of cancers, including lung cancer [3]. A study conducted in Slovenia et al. showed that mercury exposure could increase lung cancer risk [4]. In addition, lead exposure might also have an association with an elevated risk of lung carcinoma [5]. Besides these heavy metals, Arsenic, a toxic metalloid presents in the earth’s crust, has attracted much attention. Inorganic arsenic (iAs) has been classified as a human carcinogen since 1980 [6]. People can be exposed to iAs through air, food and drinking water [7]. Evidence showed that long-term iAs exposure may have a correlation with a variety of diseases, such as cardiovascular disease, diabetes, and cancers [8]. Thus, iAs has become a serious public concern.

A meta-analysis indicated that people who exposed to iAs may have a 2.44-fold increased risk of lung cancer relative to that of non-exposed people [6]. Even a low-dose of iAs exposure could significantly increase lung cancer risk [9]. The molecular mechanisms of iAs carcinogenesis remain poorly understood. Several mechanisms have been presented, such as genotoxicity, altered cell proliferation, oxidative stress, changes to the epigenome, and disturbances of signal transduction pathways [10]. However, only a few studies have been devoted to the molecular mechanisms, with inconclusive results generated.

In recent years, high-throughput tools such as microarray and sequencing have been widely used to explore the gene variations in a variety of diseases, particularly carcinomas [11]. Exploring and mining massive data may help us screen important genes or pathways associated with the diseases. Hence, in the present study, we aimed to retrieve and analyze relevant microarray-based data and screen the differentially expressed genes (DEGs) that might play a role in iAs-induced bronchial epithelial cells damage. Next, the functions of the candidate genes were annotated. Then, hub genes among the DEGs were screened out and their roles in lung cancer were further assessed. Afterward, the agents that might rescue the cells from iAs-induced damage were predicted and evaluated.

2. Methods

2.1 Data source

A search in the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/) was conducted in order to get datasets regarding iAs-treated lung cells. Datasets were restricted to human sapiens, and the study type was confined to Expression profiling by array.

2.2 Screening of DEGs

To obtain the DEGs between iAs-treated lung cell group and the control group, the data of the selected gene expression profiles were analyzed by GEO2R [12]. The results were downloaded in text format, and the $|$ log fold-change $|$ values were converted to fold-change (FC) ones. Genes that met the cut-off criteria of $P<$ 0.05 and a $|$ FC $|$ of $>$ 1.2 were selected as DEGs.

If there were more than or equal to two datasets that met the inclusion criteria, the intersections of the up-regulated and down-regulated DEGs from each dataset, respectively, were obtained by using Venn analysis [13].

2.3 Functional enrichment analysis of DEGs

To evaluate the possible functions of the DEGs, the functional enrichment analyses of the DEGs, including gene ontology (GO) function analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were carried out by using Kobas database [14].

In the GO analysis, genes can be classified into hierarchical categories and the gene network can be presented according to biological processes [15]. An adjusted $P$ -value of less than 0.05 was regarded as statistically significant.

The KEGG pathways database is a collection of graphical diagrams (pathway maps) for the biochemical pathways, whose efforts have been made to develop and improve the cross-species annotation procedure for linking genomes to the molecular networks through the KEGG Orthology system [16]. Fisher’s exact test and chi-square test was used to select the significant pathways. A $P$ -value of less than 0.05 was regarded as statistically significant.

Table 1
The most significant up-regulated and down-regulated DEGs in GSE33520 and GSE36684 (Top ten, iAs-treated cells versus controls)

GSE33520				GSE36684
Probe set	Gene symbol	LogFC	$p$ value	ID	Gene symbol	LogFC	$p$ value
Down-regulated
A_23_P31407	AGR2	$-$ 6.84735	0.000161	8022295	PIEZO2	$-$ 2.83978	6.01E-09
A_23_P72096	IL1A	$-$ 6.40363	2.45E-05	8022692	DSC3	$-$ 2.70029	5.93E-03
A_24_P913146	HOPX	$-$ 6.39072	4.1E-05	8092541	LIPH	$-$ 2.62833	1.24E-07
A_23_P254507	HOPX	$-$ 6.37473	7.29E-05	8056222	DPP4	$-$ 2.61876	1.67E-08
A_24_P152968	AKR1C1	$-$ 5.93177	0.013263	8046895	FAM171B	$-$ 2.54776	3.04E-07
A_23_P83098	ALDH1A1	$-$ 5.83302	0.014787	8140579	CACNA2D1	$-$ 2.43092	1.07E-07
A_23_P4536	EPB41L3	$-$ 5.34117	0.010543	8022283	PIEZO2	$-$ 2.33265	1.22E-06
A_24_P220947	AKR1C1	$-$ 5.26816	0.042389	8046906	GULP1	$-$ 2.30225	1.04E-03
A_23_P204847	LCP1	$-$ 5.17445	0.001355	8096176	PTPN13	$-$ 2.17979	6.57E-05
A_23_P110204	CXCL5	$-$ 5.16023	2.47E-05	7951271	MMP1	$-$ 2.14495	3.46E-04
Up-regulated
A_23_P154605	SULF2	5.110408	0.001318	8051583	CYP1B1	2.672109	3.34E-04
A_23_P10127	SFRP1	4.460879	0.017158	8130867	THBS2	2.569728	2.74E-05
A_23_P366366	SCRN1	4.414129	0.022151	7936494	GFRA1	2.245493	3.08E-05
A_23_P7642	SPARC	4.3732	0.02113	8168691	DIAPH2	2.179322	5.14E-06
A_24_P743802	ZNF618	4.345001	2.62E-05	8015387	KRT17	2.165334	4.57E-05
A_23_P144796	PDLIM4	4.1085	0.013957	7933312	ANXA8L1	2.130435	2.25E-06
A_23_P15174	MT1F	3.891799	2.32E-05	8172425	SLC38A5	1.943904	3.57E-04
A_23_P50946	RAMP1	3.885401	0.011786	8005449	KRT17	1.896057	1.05E-06
A_23_P72737	IFITM1	3.72218	0.044222	7976812	SNORD113-4	1.822497	1.42E-03
A_23_P55251	ITGA3	3.671474	0.013959	7947338	PAX6	1.752537	3.62E-05

2.4 Protein-protein interaction (PPI) network construction and assessment

To find the possible hub genes/proteins that might play a key role in the biological process, the DEGs were analyzed by STRING and Cytoscape tools in order to predict the interaction relationship among them. A combined score of not less than 0.4 (median confidence score) was used as a threshold. The hub protein was selected based on its association with other proteins, which were sorted by the degree values in the network.

To explore the possible roles and prognostic values of candidate hub genes in lung cancer, two cohorts in the TCGA database were used for evaluation [17]. Thus, the data concerned two subgroups, namely, lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC), respectively. The expression levels and prognostic values were assessed for each hub gene. The genes whose expression levels affected the prognosis of lung cancer may be regarded as the key genes.

2.5 Identification of small molecules that reverse iAs-induced biological alterations

To predict the small agent for prevention of iAs-induced cell damage, the connectivity mapping (CMAP) database that contains whole genomic expression profiles for small active molecular inferences was used [18]. The DEGs were submitted to CMAP for analysis. Small molecular compounds with negative connectivity enrichment scores were evaluated and selected as potential therapeutic molecules.

3. Results

3.1 Identification of DEGs from the selected gene expression profiles

Two gene expression profiles, GSE33520 and GSE36684, were retrieved from the GEO database. GSE33520 contained seven specimens (three specimens of lung epithelial cells exposed to 2.5 $\mu$ M arsenic trioxide for six months, and four controls); GSE36684 contained eighteen specimens (seven specimens of lung epithelial cells exposed to 2.0 $\mu$ M NaAsO ${}_{2}$ for two months, and eleven controls). In each dataset, a human immortalized lung epithelial cell line, BEAS-2B, was used.

Figure 1.

Intersection of the up-regulated (a) and down-regulated (b) DEGs from GSE33520 and GSE36684, respectively. The intersection included 61 up-regulated and 228 down-regulated genes.

The DEGs were screened out from the comparison of the two groups, respectively. As a result, 632 up-regulated and 1594 down-regulated genes have been screened out from GSE33520, while 1293 up-regulated and 2230 down-regulated genes have been generated from GSE36684 (Table 1).

Figure 2.

The top 10 terms of Go analysis (a) and KEGG enrichment analysis (b), respectively.

To get the shared DEGs in both datasets, Venn analysis was used to obtain the intersection of the DEGs profiles. As shown in Fig. 1, there were 61 up-regulated and 228 down-regulated genes in the intersections, respectively.

3.2 Functional annotation and pathway enrichment of DEGs

To explore the biological functions of the DEGs, they were clustered through GO analysis. GO analysis showed that the DEGs were enriched in 183 GO terms. The top items were cell part, intracellular part, cytoplasmic part, cellular process and binding (Fig. 2a).

KEGG pathway analysis showed that the DEGs were enriched in 23 pathways. The top items were Mineral absorption, Biosynthesis of unsaturated fatty acids, Metabolic pathways, Pentose and glucuronate interconversions, and Fatty acid metabolism (Fig. 2b).

3.3 PPI network construction

The DEGs in the intersection were submitted to STRING website for analysis. The nodes with a confidence of 0.4 were selected for constructing PPI networks.

In this network, 15 node genes/proteins, whose degrees were more than five, showed a strong association with other node proteins, respectively (Table 2, Fig. 3). Therefore, the screened hub genes (proteins) might play critical roles in iAs-induced lung epithelial cell malignant transformation.

Table 2
The hub genes in the protein-protein interaction network and validation in TCGA cohorts

Node_name	Degree	Betweenness		LUAD vs Normal	Log-rank test	LUSC vs Normal	Log-rank test
				( $P$ value)	(high vs median $+$ low)	( $P$ value)	(high vs median $+$ low)
					$P$ value		$P$ value
ALDH3A2	12	1588	.524	$P>$ 0.05	0.96	$P>$ 0.05	0.39
JUN	9	4983	.727	Down ( $P<$ 0.05)	0.68	Down ( $P<$ 0.05)	0.053
RAB1A	7	1571	.9	Up ( $P<$ 0.05)	0.098	Up ( $P<$ 0.05)	0.49
MTIF2	7	1158	.167	Up ( $P<$ 0.05)	0.027	Up ( $P<$ 0.05)	0.44
MRPS9	7	1275	.081	Up ( $P<$ 0.05)	0.32	Up ( $P<$ 0.05)	0.77
NQO1	6	629	.5606	Up ( $P<$ 0.05)	0.18	Up ( $P<$ 0.05)	0.9
PECR	6	318	.0818	Down ( $P<$ 0.05)	0.93	Down ( $P<$ 0.05)	0.6
HIBCH	6	1928	.248	Up ( $P<$ 0.05)	0.098	Up ( $P<$ 0.05)	0.77
NDUFAB1	6	1017	.826	Up ( $P<$ 0.05)	0.4	Up ( $P<$ 0.05)	0.068
ME1	6	3719	.899	Up ( $P<$ 0.05)	0.22	Up ( $P<$ 0.05)	0.76
ACOX1	6	3203	.8	Down ( $P<$ 0.05)	0.041	Down ( $P<$ 0.05)	0.44
MRPL10	6	928	.0714	Up ( $P<$ 0.05)	0.055	Up ( $P<$ 0.05)	0.6
CAV1	6	2265	.669	Down ( $P<$ 0.05)	0.027	Down ( $P<$ 0.05)	0.017
MRPL44	6	460		Up ( $P<$ 0.05)	0.083	Up ( $P<$ 0.05)	0.96
MRPL17	6	610	.8024	Up ( $P<$ 0.05)	0.0011	Up ( $P<$ 0.05)	0.63

LUAD: Lung Adenocarcinoma; LUSC: Lung Squamous Cell Carcinoma; Up (Down): The gene was up-regulated (down-regulated) in cancer tissues relative to normal tissues.

Figure 3.

Protein-Protein Interaction network of the hub genes (proteins). The genes in the green modules were the top 15 hub genes sorted by the degree value.

3.4 Assessments of the hub genes in TCGA cohort

To explore the expressions and prognostic values of the hub genes in lung carcinoma, we validated these issues with two cohorts in TCGA.

In the TCGA database, the data regarding lung cancer were divided into two cohorts (LUAD and LUSC), respectively. The cohort of LUAD contained 515 cases and 59 normal controls; while that of LUSC comprised of 503 cases and 52 controls.

As shown in Table 2, most of the hub genes, except for ALDH3A2, were dysregulated in lung cancer tissues compared to normal tissues. Nevertheless, among the 15 hub genes, only four of them (MTIF2, ACOX1, CAV1, and MRPL17) might act as prognostic factors for LUAD. Notably, only one hub gene, CAV1, might have an association with LUSC prognosis (Fig. 4).

Figure 4.

The expression levels (A) and prognostic values (B) of the four key genes in lung carcinoma based on TCGA cohorts. (a) MTIF2 in LUAD; (b) ACOX1 in LUAD; (c) MRPL17 in LUAD; (d) CAV1 in LUAD; (e) CAV1 in LUSC. $P<$ 0.05 for all comparisons. LUAD: Lung Adenocarcinoma; LUSC: Lung Squamous Cell Carcinoma.

To further assess the prognostic value of these genes in lung cancer, we conducted Cox regression analyses by using the TIMER tool [19] that was also based on TCGA cohorts. As shown in Table 3, the results failed to show the four genes (MTIF2, ACOX1, CAV1, and MRPL17) as independent prognostic factors for lung carcinoma.

Table 3

Multivariate analyses of several hub gene expressions and other clinical prognostic markers related to overall survival in lung cancer

	Coeffecient	HR	95%CI-lower	95%CI-upper	$P$ -value
LUAD
Age	0.009	1.010	0.991	1.028	0.314
Gendermale	$-$ 0.044	0.957	0.680	1.347	0.802
RaceBlack	16.209	10949342.80	0	–	0.994
RaceWhite	16.436	13740364.61	0	–	0.994
Stage2	0.884	2.422	1.605	3.653	0.000
Stage3	0.993	2.698	1.752	4.155	0.000
Stage4	1.064	2.897	1.496	5.610	0.002
Tumor purity	0.703	2.02	0.977	4.177	0.058
CAV1	0.069	1.072	0.955	1.203	0.239
MTIF2	0.338	1.402	0.971	2.025	0.072
ACOX1	$-$ 0.086	0.917	0.654	1.287	0.617
MRPL17	0.229	1.258	0.945	1.673	0.115
LUSC
Age	0.015	1.015	0.997	1.034	0.103
Gendermale	0.402	1.495	1.023	2.184	0.038
RaceBlack	$-$ 0.001	0.999	0.302	3.308	0.999
RaceWhite	$-$ 0.459	0.632	0.208	1.917	0.417
Stage2	0.231	1.260	0.873	1.819	0.217
Stage3	0.632	1.882	1.233	2.872	0.003
Stage4	0.705	2.024	0.422	9.710	0.378
Tumor purity	$-$ 0.311	0.733	0.357	1.505	0.397
CAV1	0.070	1.073	0.949	1.213	0.261

HR: hazard ratio; CI: confidence interval; LUAD: Lung Adenocarcinoma; LUSC: Lung Squamous Cell Carcinoma.

3.5 CMAP analysis

To predict small molecules capable of reversing the iAs-induced lung cell malignant transformation, the DEGs were submitted to CMAP for analysis. The perturbagens from the CMAP were analyzed according to their permutated results, $P$ values, and enrichment scores. Consequently, the top 5 negatively related small molecules with highly significant correlations, in turn, were listed in Table 4.

Table 4
Predicted small molecules in CMAP (top 5)

Cmap name	Mean	$n$	Enrichment	$p$
Quinostatin	$-$ 0.67	2	$-$ 0.968	0.00229
Alcuronium chloride	$-$ 0.656	2	$-$ 0.919	0.01332
5230742	$-$ 0.599	2	$-$ 0.91	0.01612
Proscillaridin	$-$ 0.649	3	$-$ 0.899	0.00192
Edrophonium chloride	$-$ 0.576	5	$-$ 0.882	0.00006

As shown in this table, Quinostatin, Alcuronium Chloride, 5230742, Proscillaridin and Edrophonium Chloride might have a potential reversing the iAs-induced biological process, indicating that these agents need to be further validated as preventive or therapeutic drugs for iAs-associated lung cancer.

4. Discussion

Heavy metal exposure has become a public concern for years. iAs exposure mainly through drinking water may result in cell damage or malignant transformation, thus increasing cancer risk in the exposed population. The mechanisms of iAs-induced cell malignant transformation have remained not clear. Previous reports have focused on the carcinogenic effects of the iAs metabolic products, such as MMA ${}^{+3}$ , MMA ${}^{+5}$ , DMA ${}^{+3}$ , and DMA ${}^{+5}$ , on cells [20]. However, the evidence regarding gene variations in this biological process is limited. In the present study, microarray-based data were analyzed, and the DEGs that might play a role in the iAs-induced cell transformation were screened out. Through PPI network construction, we found that a total of 15 genes may be the hub genes. After a validation analysis using TCGA cohorts, we found that four of them (MTIF2, ACOX1, CAV1, and MRPL17) may be the key genes that might be critical in both the onset and the development of iAs-induced lung carcinoma. Lastly, using CMAP tool, several small molecules that might be potential drugs for reversing iAs-induced lung cell malignant transformation had been predicted.

GO analysis showed that the DEGs were enriched in 183 GO terms, such as cell part, intracellular part, cytoplasmic part, cellular process and binding, while pathway analysis showed that the DEGs were enriched in 23 pathways such as Mineral absorption, Biosynthesis of unsaturated fatty acids, Metabolic pathways, Pentose and glucuronate interconversions, and Fatty acid metabolism. The data indicated that the screened DEGs in the present study might be involved in multiple aspects of the biological process, and they might also be enriched in multiple signaling pathways. Of these pathways, cell metabolism may be worth noting because several top pathways were related to ‘metabolism’. This is consistent with previous evidence that metabolite of iAs may exert a toxic effect on lung cells [8].

Most of the hub genes were dysregulated in lung cancer tissues relative to normal tissues, indicating that these genes may be crucial in the iAs-induced tumorigenesis of lung cells. However, not all genes can affect the prognosis of lung cancer. After a validation analysis in TCGA cohorts, four genes, including MTIF2, ACOX1, CAV1, and MRPL17, were selected as prognostic factors for lung cancer, suggesting that the expressions of these genes may have an influence on the survival time of lung carcinoma patients. MTIF2 (Mitochondrial translation initiation factor 2) may initiate the translation of proteins encoded by the mitochondrial genome [21]. Little evidence has revealed its roles in cancer. Nevertheless, a recent report indicated that MTIF2 may be cell proliferation-related genes in glioma [22]. ACOX1 (acyl-CoA oxidase 1) is the first and rate-limiting enzyme in fatty acid $\beta$ -oxidation and is linked to peroxisomal disorders and carcinogenesis [23]. Misregulated ACOX1 expression in tumor alters lipid metabolomes, thus eliciting an intracellular signaling change to enhance cell motility [24]. CAV1 (Caveolin 1) is over-expressed in bladder tissues, and its down-regulation leads to inhibition of cell proliferation and invasion [25]. CAV1 may be involved in RANKL-induced cell migration in lung, renal and breast cancer cells [26]. MRPL17 is a mitospecific component of the mitoribosomal central protuberance [27]. Few investigations have been devoted to its roles in cancers. However, a recent study indicated that MRPL17 was up-regulated in oral pre-cancer and cancer pathologies, implicating this gene as a crucial player in oral carcinogenesis [28]. Taken together, these four key genes may play critical roles in the development of lung cancer. The data also showed that iAs-induced lung tumorigenesis included various aspects of biological variations. Regulation of these key gene expressions might have a potential reversing or attenuating the iAs-induced malignant transformation. Since these key genes were hub genes, they might act as potential therapeutic targets for iAs-induced lung cancer therapy. Notably, besides LUAD, only one of these four key genes (CAV1) may be related to the prognosis of LUSC. This discrepancy might be due to the reasons that the carcinogenicity of iAs on lungs is cell-type specific, where the association between adenocarcinoma and arsenic exposures through inhalation appeared to be stronger than that of squamous cell carcinoma [29]. However, since none of these key genes have been indicated to be independent prognostic factors for lung cancer, future investigations using cohorts with large sample sizes are needed to address this controversy.

It is worth noting that both arsenic trioxide and NaAsO2 can often be used as Inorganic arsenic representatives in toxicological studies. The exposure of lung cells to Inorganic arsenic may usually result in two major events: cell apoptosis or cell malignant transformation. However, the mechanisms of these two Inorganic arsenic representatives on carcinogenesis are very complicated. Few investigations have been devoted to the comparison of the carcinogenesis mechanisms between them. Moreover, the goal of our paper was to find the possible key genes in iAs-related lung cancer by using bioinformatics methods. Thus,‘similarities and differences’ of arsenic trioxide and NaAsO2 can only be addressed in future studies.

To predict the small agent for reversing the iAs-induced cell malignant transformation, CMAP tool was utilized. As a result, a list of agents was predicted. The top molecules included Quinostatin, Alcuronium Chloride, 5230742, Proscillaridin, and Edrophonium Chloride. Among these molecules, Quinostatin had the highest negative enrichment score. Evidence indicated that Quinostatin can directly target the lipid-kinase activity of the catalytic subunits of class Ia PI3Ks, and inhibit mTOR network [30]. Quinostatin was also regarded as an inhibitor of cellular S6 phosphorylation [31]. In a recent study, Quinostatin has been predicted as a potential supplementary drug for treating pediatric acute lymphoblastic leukemia [32]. Therefore, since S6 phosphorylation, PI3K, mTOR signaling pathways might be involved in the iAs-induced cell transformation, Quinostatin might act as a potent agent for the prevention or treatment of iAs-associated lung carcinoma.

In conclusion, the present study provides preliminary research on the molecular mechanisms underlying iAs-induced bronchial epithelial cell transformation. High throughput data were analyzed to screen the DEGs by computational bioinformatics methods. The DEGs were mostly enriched in a variety of pathways, particularly cell metabolism-associated pathways. Then, several key hub genes that may play key roles in iAs-induced cell malignant transformation have been screened out. Using the CMAP tool, an mTOR inhibitor, Quinostatin, was predicted to have a potential reversing the iAs-induced carcinogenesis and development. However, future validation experiments are warranted to confirm the results.

Footnotes

Acknowledgments

This study was partially supported by the National Natural Science Foundation of China (No. 81672290).

Conflict of interest

The authors declare that they have no conflict of interest.

References

Lim

Ridge

C.A.

Nicholson

A.G.

and Mirsadraee

, The 8(th) lung cancer TNM classification and clinical staging system: review of the changes and clinical implications, Quant Imaging Med Surg 8 (2018), 709–718.

Malhotra

Malvezzi

Negri

La Vecchia

and Boffetta

, Risk factors for lung cancer worldwide, Eur Respir J 48 (2016), 889–902.

Mezynska

and Brzoska

M.M.

, Environmental exposure to cadmium-a risk for health of the general population in industrialized countries and preventive strategies, Environ Sci Pollut Res Int 25 (2018), 3211–3232.

Zadnik

and Pompe-Kirn

, Effects of 500-year mercury mining and milling on cancer incidence in the region of Idrija, Slovenia, Coll Antropol 31 (2007), 897–903.

Liao

L.M.

Friesen

M.C.

Xiang

Y.B.

Cai

Koh

D.H.

B.T.

Yang

H.L.

Locke

S.J.

Rothman

Zheng

Gao

Y.T.

Shu

X.O.

and Purdue

M.P.

, Occupational lead exposure and associations with selected cancers: the shanghai men’s and women’s health study cohorts, Environ Health Perspect 124 (2016), 97–103.

Freeman

E.W.

and Sammel

M.D.

, Anxiety as a risk factor for menopausal hot flashes: evidence from the Penn Ovarian Aging cohort, Menopause 23 (2016), 942–949.

Lamm

S.H.

Ferdosi

Dissen

E.K.

and Ahn

, A systematic review and meta-regression analysis of lung cancer risk and inorganic arsenic in drinking water, Int J Environ Res Public Health 12 (2015), 15498–15515.

Kuo

C.C.

Moon

K.A.

Wang

S.L.

Silbergeld

and Navas-Acien

, The association of arsenic metabolism with cancer, cardiovascular disease, and diabetes: a systematic review of the epidemiological evidence, Environ Health Perspect 125 (2017), 087001.

Steinmaus

Ferreccio

Yuan

Acevedo

Gonzalez

Perez

Cortes

Balmes

J.R.

Liaw

and Smith

A.H.

, Elevated lung cancer in younger adults and low concentrations of arsenic in water, Am J Epidemiol 180 (2014), 1082–1087.

10.

Zhou

and Xi

, A review on arsenic carcinogenesis: epidemiology, metabolism, genotoxicity and epigenetic changes, Regul Toxicol Pharmacol 99 (2018), 78–88.

11.

Surrey

L.F.

Luo

Chang

and Li

M.M.

, The genomic era of clinical oncology: integrated genomic analysis for precision cancer care, Cytogenet Genome Res 150 (2016), 162–175.

12.

Fan

Huang

Zheng

Zhou

Zeng

J.F.

Zhu

and Yi

, Analysis of microarray-identified genes and microRNAs associated with idiopathic pulmonary fibrosis, Mediators Inflamm 2017 (2017), 1804240.

13.

J.W.

Mai

Cui

Y.L.

and Kong

L.Y.

, Genes and pathways identified in thyroid carcinoma based on bioinformatics analysis, Neoplasma 63 (2016), 559–568.

14.

Xie

Mao

Huang

Ding

Dong

Kong

Gao

C.Y.

and Wei

, KOBAS 2.0: a web server for annotation and identification of enriched pathways and diseases, Nucleic Acids Res 39 (2011), W316–322.

15.

Tian

Wang

Guo

Liu

and Teng

, An improved method for functional similarity analysis of genes based on Gene Ontology, BMC Syst Biol 10 (2016), 119.

16.

Kanehisa

Goto

Sato

Furumichi

and Tanabe

, KEGG for integration and interpretation of large-scale molecular data sets, Nucleic Acids Res 40 (2012), D109–114.

17.

Chandrashekar

D.S.

Bashel

Balasubramanya

S.A.H.

Creighton

C.J.

Ponce-Rodriguez

Chakravarthi

and Varambally

, UALCAN: a portal for facilitating tumor subgroup gene expression and survival analyses, Neoplasia 19 (2017), 649–658.

18.

Gillbro

J.M.

Merinville

Olsson

Al-Bader

Klack

Visdal-Johnsen

and Mavon

, The use of gene arrays and corresponding connectivity mapping (Cmap) to identify novel anti-ageing ingredients, Int J Cosmet Sci 37(Suppl 1) (2015), 9–14.

19.

Fan

Wang

Traugh

Chen

Liu

J.S.

and Liu

X.S.

, TIMER: a web server for comprehensive analysis of tumor-infiltrating immune cells, Cancer Res 77 (2017), e108–e110.

20.

Gamboa-Loira

Cebrian

M.E.

Franco-Marina

and Lopez-Carrillo

, Arsenic metabolism and cancer risk: a meta-analysis, Environ Res 156 (2017), 551–558.

21.

Overman

R.G.

, Jr. Enderle

P.J.

Farrow

J.M.

, 3rd Wiley

J.E.

and Farwell

M.A.

, The human mitochondrial translation initiation factor 2 gene (MTIF2): transcriptional analysis and identification of a pseudogene, Biochim Biophys Acta 1628 (2003), 195–205.

22.

Minchenko

O.H.

Tsymbal

D.O.

Minchenko

D.O.

Riabovol

O.O.

Ratushna

O.O.

and Karbovskyi

L.L.

, Hypoxic regulation of the expression of cell proliferation related genes in U87 glioma cells upon inhibition of ire1 signaling enzyme, Ukr Biochem J 88 (2016), 11–21.

23.

Chen

X.F.

Tian

M.X.

Sun

R.Q.

Zhang

M.L.

Zhou

L.S.

Jin

Chen

L.L.

Zhou

W.J.

Duan

K.L.

Chen

Y.J.

Gao

Cheng

Z.L.

Wang

Zhang

J.Y.

Sun

Y.P.

H.X.

Zhao

Y.Z.

Yang

Liu

W.R.

Shi

Y.H.

Xiong

Guan

K.L.

and Ye

, SIRT5 inhibits peroxisomal ACOX1 to prevent oxidative damage and is downregulated in liver cancer, EMBO Rep 19 (2018).

24.

Lai

Y.H.

Liu

Chiang

W.F.

Chen

T.W.

Chu

L.J.

J.S.

Chen

S.J.

Chen

H.C.

and Tan

B.C.

, MiR-31-5p-ACOX1 Axis Enhances Tumorigenic Fitness in Oral Squamous Cell Carcinoma Via the Promigratory Prostaglandin E2, Theranostics 8 (2018), 486–504.

25.

Zhou

Dai

Zhang

Wang

and Liu

, MicroRNA-124 inhibits cell proliferation, invasion and migration by targeting CAV1 in bladder cancer, Exp Ther Med 16 (2018), 2811–2820.

26.

Wang

Song

Che

Zhang

Wang

Zhang

Fan

Hou

Liu

and Qu

, Caveolin1 enhances RANKLinduced gastric cancer cell migration, Oncol Rep 40 (2018), 1287–1296.

27.

Box

J.M.

Kaur

and Stuart

R.A.

, MrpL35, a mitospecific component of mitoribosomes, plays a key role in cytochrome c oxidase assembly, Mol Biol Cell 28 (2017), 3489–3499.

28.

Khowal

Naqvi

S.H.

Monga

Jain

S.K.

and Wajid

, Assessment of cellular and serum proteome from tongue squamous cell carcinoma patient lacking addictive proclivities for tobacco, betel nut, and alcohol: case study, J Cell Biochem 119 (2018), 5186–5221.

29.

Kuo

Y.C.

Y.S.

and Guo

H.R.

, Lung cancer associated with arsenic ingestion: cell-type specificity and dose response, Epidemiology 28(Suppl 1) (2017), S106–S112.

30.

Yang

Shamji

Matchacheep

and Schreiber

S.L.

, Identification of a small-molecule inhibitor of class Ia PI3Ks with cell-based screening, Chem Biol 14 (2007), 371–377.

31.

Sato

Murata

Shirakawa

and Uesugi

, Biochemical target isolation for novices: affinity-based strategies, Chem Biol 17 (2010), 616–623.

32.

Kong

Zhang

and Zhou

, Potential therapeutic targets and small molecular drugs for pediatric B-precursor acute lymphoblastic leukemia treatment based on microarray data, Oncol Lett 14 (2017), 1543–1549.

Screening of key genes and prediction of therapeutic agents in Arsenic-induced lung carcinoma

Abstract

BACKGROUND:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1. Introduction

2. Methods

2.1 Data source

2.2 Screening of DEGs

2.3 Functional enrichment analysis of DEGs

Table 1 The most significant up-regulated and down-regulated DEGs in GSE33520 and GSE36684 (Top ten, iAs-treated cells versus controls)

2.5 Identification of small molecules that reverse iAs-induced biological alterations

3. Results

3.1 Identification of DEGs from the selected gene expression profiles

3.3 PPI network construction

Table 2 The hub genes in the protein-protein interaction network and validation in TCGA cohorts

Table 4 Predicted small molecules in CMAP (top 5)

Footnotes

Acknowledgments

Conflict of interest

References

Table 1
The most significant up-regulated and down-regulated DEGs in GSE33520 and GSE36684 (Top ten, iAs-treated cells versus controls)

Table 2
The hub genes in the protein-protein interaction network and validation in TCGA cohorts

Table 4
Predicted small molecules in CMAP (top 5)