The clinical significance of interleukin 24 and its potential molecular mechanism in laryngeal squamous cell carcinoma

Abstract

Interleukin 24 (IL24) has been documented to be highly expressed in several cancers, but its role in laryngeal squamous cell carcinoma (LSCC) remains unclarified. In this study, to reveal the function and its clinical significance of IL24 in LSCC, multiple detecting methods were used comprehensively. IL24 protein expression was remarkably higher in LSCC ( $n=$ 49) than non-cancerous laryngeal controls ( $n=$ 26) as detected by in-house immunohistochemistry. Meanwhile, the IL24 mRNA expression was also evaluated based on high throughput data from Gene Expression Omnibus, The Cancer Genome Atlas, ArrayExpress and Oncomine databases. Consistently with the protein level, IL24 mRNA expression level was also predominantly upregulated in LSCC ( $n=$ 172) compared to non-cancerous laryngeal tissues ( $n=$ 81) with the standard mean difference (SMD) being 1.25 and the area under the curve (AUC) of the summary receiver operating characteristic (sROC) being 0.89 (95% CI $=$ 0.86–0.92). Furthermore, the related genes of IL24 and the differentially expressed genes (DEGs) of LSCC were intersected and sent for Gene ontology (GO) enrichment, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway, and the protein-protein interaction (PPI) analyses. In the GO annotation, the top terms of biological process (BP), cellular component (CC) and molecular function (MF) were extracellular matrix organization, extracellular matrix, cytokine activity, respectively. The top pathway of KEGG was ECM-receptor interaction. The PPI networks indicated the top hub genes of IL24-related genes in LSCC were SERPINE1, TGFB1, MMP1, MMP3, CSF2, and ITGA5. In conclusion, upregulating expression of IL24 may enhance the occurrence of LSCC, which owns prospect diagnostic ability and therapeutic significance in LSCC.

1. Introduction

Laryngeal carcinoma is a common kind of head and neck carcinoma and ranks second in the incidence of head and neck carcinomas [1]. In 100,000 people, an average of nearly 6 have laryngeal carcinoma. The predominant type is laryngeal squamous cell carcinoma (LSCC) [2, 3, 4, 5, 6, 7], seen in 90 of 100 laryngeal carcinoma cases (90%). HPV and smoking are common causes of LSCC [5, 8, 9].

Table 1

Details of the IHC staining scoring criteria

Staining intensity	Total score (staining intensity * percentage of positive cells)
	0 ( $<$ 5%)	1 (5%–25%)	2 (26%–50%)	3 (51–75%)	4 (76–100%)
0 (no staining)	0	0	0	0	0
1 (light staining)	0	1	2	3	4
2 (medium staining)	0	2	4	6	8
3 (strong staining)	0	3	6	9	12

In clinical practice, early LSCC often manifests as chronic laryngitis or leukoplakia. And dysphonia is a major symptom in early LSCC, in particular those glottic LSCCs. With no obvious symptoms, the best time for treatment may be missed and LSCC may already be advanced in many cases. At present, endoscopies and biopsies are performed for diagnosis [10]. Surgery and radiotherapy are used for treatment in clinical work [11, 12, 13]. The potential risks related to LSCC surgery include loss of normal speech and/or swallowing, throat, larynx stenosis, and breathing difficulties. The disadvantages of radiotherapy are that treatment cycles are long and symptoms such as the loss of taste and smell as well as dry mouth may appear [14, 15].

The mechanism of LSCC is not clear. Therefore, screening out relative genes of LSCC and studying their molecular mechanism have great significance for early diagnosis and prognosis of LSCC. Interleukin 24 (IL24) is a newly-identified member of the IL-10 family (IL-10, 19, 20, 22, 24, 26) [16], which is located on human chromosome 1q32-33. Its signaling pathways have many biological functions related to cell differentiation, proliferation, development, apoptosis, and inflammation [17, 18]. Research has shown that IL24 is capable of inhibiting the growth of tumor cells while having no influence on normal cells. Studies have shown that IL24, produced by immune cells, can induce apoptosis of cancer cells in some respects such as inhibiting angiogenesis, inducing specific mitochondrial dysfunction in cancer cells, inducing autophagy of cancer cells, and so on [19, 20, 21, 22]. The authors of some studies have examined IL24 as a potential treatment for cancer such as oral squamous cell carcinoma (OSCC) [23], prostate cancer [24], and hepatocellular carcinoma [25, 26], among others, with consideration for its biological functions [16, 27].

However, few research reports related to the expression of IL24 in LSCC tissues are available. Hence, in this study, we explored the relationship between the occurrence and development of LSCC and IL24 expression from the molecular level and relating to clinical significance, which potentially has diagnostic ability and therapeutic significance in LSCC. The results of our study suggest that IL24 may be used as a prospective diagnostic indicator because its increased expression may indicate the occurrence of cancer and cause immune resistance.

2. Materials and methods

2.1 Immunohistochemistry

In-house immunohistochemistry (IHC) was performed on tissue arrays, which were purchased from Pantomics, Inc (Richmond, CA; HNT961, HNT962, and HNT1021) including 49 LSCC samples and 26 normal samples The corresponding clinical parameters were also collected. The difference of IL24 expression among the various groups was analyzed by independent samples $t$ -test. The study was approved by the Ethics Committee of the First Affiliated Hospital of Guangxi Medical University in Nanning, China, and all patients signed the informed consent to use the samples for the study. The scores of IHC staining results were based on the product of staining intensity and the percentage of positive cells. See Table 1 for details of the IHC staining scoring criteria.

2.2 Collection of the RNA sequencing data in the TCGA database

We mined the TCGA (The Cancer Genome Atlas) database, searched the samples of LSCC and the corresponding clinical parameters, and processed the obtained data by matrix. Subsequently, we extracted the expression value and clinical information of LSCC from these processed data. A total of 111 LSCC tissues and 44 non-cancerous head and neck squamous epithelial tissues were obtained, as TCGA does not specify the exact location for each non-cancerous sample, we took all 44 cases as the control group. These data were used for evaluating the potential diagnostic functions of IL24 in LSCC by the receiver operating characteristic (ROC) curve; its prognostic significance was revealed by Kaplan-Meier curves. In addition, the clinical parameters of IL24 in LSCC were also analyzed.

2.3 Collection of the microarrays data in GEO, ArrayExpress, and Oncomine databases

We searched and downloaded IL24-related microarrays data in the GEO (Gene Expression Omnibus), ArrayExpress, and Oncomine databases. The search keywords were as follows: (laryngeal OR larynx) AND (cancer OR carcinoma OR tumor OR squamous cell carcinoma OR LSCC) AND (mRNA OR gene). The search results were filtered to meet the following criteria: (1) The collected samples were diagnosed as LSCC and contained both cancer tissues and control tissues; (2) the microarrays provided the IL24 mRNA expression; (3) the tissues were from Homo sapiens; and (4) the collected data had not been processed by knockout, splicing, etc. Finally, a total of four microarrays that met the requirements of the study were obtained.

2.4 Statistical processing

In this study, SPSS 24.0 (Chicago, IL, USA) and STATA software Version 16.0 (StataCorp, College Station, TX, USA) were used for statistical analysis of IL24 expression in LSCC. Then, R 3.5.2 (St. Louis, Missouri, USA) and PRISM Version 7 (San Diego, CA, USA) were used for visual analysis and for drawing a graph of the calculated results. The extracted data were normalized and the log2 was transformed. The mRNA expression of IL24 and clinical parameters in LSCC compared with normal tissue were analyzed using an Independent samples $t$ -test. When the standard mean difference (SMD) $>$ 0 and the 95% confidence interval (CI) $>$ 0, this suggested that IL24 was overexpressed in LSCC. Using the chi-squared or I ${}^2$ test to choose the applicable model, if the $P<$ 0.05 (I ${}^2>$ 50%), it was determined that heterogeneity existed and the random effect model was applied. Otherwise, the fixed-effects model was preferred. Related scatter plots and receiver operating characteristic (ROC) curves were created.

In addition, Begg’s and Egger’s funnel plots were used to detect publication bias. In order to illustrate the ability of IL24 expression to distinguish LSCC from normal tissue, we drew additional summary ROC (sROC) curves along with their areas under the curve (AUC). We then used the positive and negative likelihood ratio as a composite indicator of specificity and sensitivity to illustrate the diagnostic significance of IL24 expression in LSCC and normal tissues.

2.5 IL24-related genes and different expression genes in laryngeal squamous cell carcinoma

We combined 4 microarrays obtained from the GEO database and analyzed the RNA-Seq obtained from the TCGA. First, we used R 3.5.2 to calculate the IL24 related genes in LSCC and normal tissues (Spearman’s correlation coefficient $⩾$ 0.5). Then, we used R 3.5.2 to search for up-regulated and down-regulated genes ( $P<$ 0.05) in LSCC and to draw volcano plots of different expression genes (DEGs). Finally, the results of DEGs and IL24-related genes were intersected and the obtained 49 genes were used for further pathway and enrichment analysis.

Table 2

The clinicopathologic parameters of 49 LSCC samples and 26 cases of normal samples by IHC

Parameters	$n$	Mean $\pm$ SD	$t$	$P$ -value
Tissue
LSCC	49	9.020 $\pm$ 1.8313	20.949	$<$ 0.0001
non-LSCC	26	1.654 $\pm$ 1.1981
Gender
Male	48	9.042 $\pm$ 1.8446	0.559	0.579
Female	1	8.000
Age
$⩾$ 60	21	9.143 $\pm$ 1.8516	0.402	0.691
$<$ 60	28	8.929 $\pm$ 1.8445
Grade
G1–G2	33	8.667 $\pm$ 1.6329	$-$ 1.849	0.077
G3–G4	16	9.751 $\pm$ 2.0494
Pathological stage
I–II	15	8.133 $\pm$ 1.18723	$-$ 2.826	0.007
III–IV	34	9.418 $\pm$ 1.94029
Pathologic T
T1–T2	29	8.621 $\pm$ 1.6128	$-$ 1.813	0.078
T3–T4	20	9.601 $\pm$ 2.0105
Pathologic N
N0	22	8.455 $\pm$ 1.5032	$-$ 2.071	0.044
N1	27	9.482 $\pm$ 1.9684
Pathologic M
M0	49	9.020
M1	0

Note: T: tumor; N: node; M: metastasis.

Table 3

The clinicopathologic features of the 111 cases of LSCC patients and 44 cases of normal patients

Parameters	$n$	Mean $\pm$ SD		$t$	$P$ -value
Tissue
LSCC	111	4.004	$\pm$ 0.1904	7.304	$<$ 0.0001
Non-LSCC	44	3.808	$\pm$ 0.1306
Gender
Male	91	4.002	$\pm$ 0.1958	$-$ 0.165	0.869
Female	20	4.010	$\pm$ 0.1684
Age
$⩾$ 60	73	3.988	$\pm$ 0.1837	$-$ 1.191	0.236
$<$ 60	38	4.033	$\pm$ 0.2018
Grade
G1–G2	11	3.987	$\pm$ 0.2651	$-$ 0.171	0.864
G3–G4	85	3.997	$\pm$ 0.1797
Pathological stage
I–II	14	3.984	$\pm$ 0.2478	$-$ 0.378	0.706
III–IV	93	4.005	$\pm$ 0.1819
Pathologic T
T1–T2	20	3.998	$\pm$ 0.2249	$-$ 0.100	0.921
T3–T4	87	4.003	$\pm$ 0.1832
Pathologic N
N0	55	3.995	$\pm$ 0.1913	$-$ 0.268	0.789
N1	50	4.005	$\pm$ 0.1921
Pathologic M
M0	104	4.0001	$\pm$ 0.19036	$-$ 0.769	0.444
M1	2	4.1053	$\pm$ 0.3059

Note: T: tumor; N: node; M: metastasis.

Table 4

The basic features of the microarrays and RNA-seq of Interleukin 24 (IL24) expression profiling included in this study

				Tumor	Normal
ID	Platform	Tumor	Normal	Mean $\pm$ SD	Mean $\pm$ SD	$P$ -value
GSE51985	GPL10558	10	10	5.3814 $\pm$ 1.78051	3.8853 $\pm$ 1.61941	0.065
GSE59102	GPL6480	29	13	7.1657 $\pm$ 2.08407	3.5030 $\pm$ 1.46564	$<$ 0.0001
GSE29330	GPL570	13	5	6.3920 $\pm$ 2.36597	3.7372 $\pm$ 0.77795	0.028
GSE84957	GPL17843	9	9	5.2758 $\pm$ 1.88798	3.1558 $\pm$ 0.70734	0.010
TCGA (LSCC)		111	44	4.0036 $\pm$ 0.19041	3.8084 $\pm$ 0.13063	$<$ 0.0001

2.6 Gene set enrichment analysis and protein-protein interaction (PPI) network construction

The intersection of DEGs and IL24-related genes were used to perform bioinformatics analyses with the ClusterProfiler R package, including the Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis and the Gene Ontology (GO) annotation. These were also used to construct a protein-protein interaction (PPI) network by use of the STRING database (confidence score $>$ 0.9). The hub genes were then screened by Cytoscape v3.6.1 software (San Diego, CA, USA).

3. Results

3.1 Clinical significance of IL24 in LSCC by immunohistochemistry

The relationship between IL24 expression at the protein level in LSCC and corresponding clinicopathological parameters of 49 LSCC samples and 26 normal samples were showed in Table 2, including gender, age, stage, grade, and so on. Compared to non-cancerous squamous epithelium samples, the IL24 protein expression was significantly highly expressed in LSCC samples ( $P<$ 0.0001). IL24 showed markedly higher expression in LSCC patients who were stages III–IV ( $P=$ 0.007) or pathologic stage N1 ( $P=$ 0.044). Figure 2A and B shows the IHC staining pictures of IL24 in LCSS and non-cancerous tissues.

3.2 Clinical significance of IL24 in LSCC by RNA-Seq

By mining the TCGA database, we were able to download and extract the expression data and corresponding clinical information that met the conditions of this study. In comparing the IL24 mRNA expression in 111 LSCC patients with that of 44 normal patients, we found that IL24 was significantly highly expressed in LSCC patients ( $P<$ 0.0001). In addition, we also analyzed the clinical information of LSCC patients such as gender, age, stage, and grade, but no statistically significant difference was found; details are shown in Table 3. The Kaplan-Meier curves suggest that LSCC patients with low IL24 expression might have a poor prognosis; the results, however, were not statistically significant [P (log-rank) $=$ 0.5743] (Fig. 6F).

3.3 Analysis of the IL24 mRNA expression between LSCC and normal tissues by microarrays and RNA-Seq data

In this study, a total of 4 microarrays (GSE51985, GSE59102, GSE29330, and GSE84957) from the GEO database and RNA-Seq from the TCGA database were included. The filtering process for this data is shown in Fig. 1A and the relevant information and calculation results are shown in Table 4. Each profile suggested that IL24 was highly expressed in LSCC; only GSE51985 was not statistically significant ( $P=$ 0.065). The results of each profile were presented in detail using scatter plots, ROCs, and their AUCs (Fig. 3A–E). In order to comprehensively analyze the IL24 mRNA expression in LSCC, a forest plot was used to show that IL24 was also highly expressed in LSCC compared to normal tissues. This was done by combining the expression data mentioned above (SMD $=$ 1.25, 95% CI $=$ 0.94–1.55, Fig. 4A). No significant heterogeneity was found (I ${}^2=$ 2.7%, $P=$ 0.391), so the fixed-effect model was chosen (Fig. 4B). By using Begg’s test ( $P=$ 0.806) and Egger’s test ( $P=$ 0.524), we also found no significant publication bias (Fig. 4C and D).

Figure 1.

(A) Flow chart of the study design. (B) Flow chart of the study search and selection.

Figure 2.

The protein expression levels of IL24 in laryngeal squamous cell carcinoma (LSCC) and corresponding non-cancerous tissues. (A) LSCC tissues for IL24, immunohistochemistry ( $\times$ 100, $\times$ 200, $\times$ 400). (B) Normal laryngeal tissues for IL24, immunohistochemistry ( $\times$ 100, $\times$ 200, $\times$ 400).

Figure 3.

(A)–(D): The expression data of IL24 and corresponding the receiver operating characteristic (ROC) curves in laryngeal squamous cell cancer (LSCC) tissues and normal tissues from Gene Expression Omnibus (GEO). (E) The expression data of IL24 and corresponding the receiver operating characteristic (ROC) curves in laryngeal squamous cell cancer (LSCC) tissues and normal tissues from The Cancer Genome Atlas (TCGA) datasets.

Figure 4.

(A) The forest plot of 5 studies evaluating standard mean difference (SMD) of Interleukin 24 (IL24) expression between laryngeal squamous cell cancer (LSCC) tissues and normal tissues based on Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases. (B) Sensitivity analysis of analysis of the expression level of IL24 in the LSCC and non-tumor groups. (C) Begg’s funnel plot for publication bias test. (D) Egger’s funnel plot for publication bias test.

Figure 5.

(A) The summary receiver operating characteristic curves analysis of IL24 for discriminating LSCC from normal tissues based on GEO databases. (B) Prior probability and post-probability positive and negative of the included studies. The prior probability, post-probability positive, and post-probability negative reached 20%, 56%, and 6%, respectively (LR: likelihood ratio). (C) and (D) Sensitivity and specificity values of the included studies. The sensitivity and specificity values of the included studies were 0.79 (0.67–0.88) and 0.84 (0.71–0.92), respectively.

To further illustrate the significance of IL24 expression in distinguishing between LSCC and non-cancerous tissue, sROC curves were used to calculate the AUC and its 95% CI. As shown in Fig. 5A, C and D, the AUC was 0.89 (95% CI $=$ 0.86–0.92); the sensitivity was 0.79 (95% CI $=$ 0.67–0.88); and the specificity was 0.84 (95% CI $=$ 0.71–0.92), which demonstrated a certain diagnostic capability.

We also calculated likelihood ratios to account for the authenticity of the specificity and sensitivity. As shown in Fig. 5B, the prior probability was 20%. The likelihood ratio positive (LRP $+$ ) was 56% which indicated that if IL24 was highly expressed, the probability of the patient being diagnosed with LSCC was 56%. The likelihood ratio negative (LRN $-$ ) was 6% which indicated that if the IL24 expression was low, the probability of the patient being misdiagnosed as having LSCC was 6%.

Table 5

A total of 49 intersectional genes

Gene name
IL24	TMC7	SERPINE1	OSM	LOXL2	ITGA5	HOMER3	FEZ1	CYP27B1	C6ORF141
TREM1	TGFBI	PTHLH	MYO1B	LAMC2	ITGA3	GPRIN1	FADS3	CSF2	C1QTNF6
TPBG	TGFB1	PLAU	MMP3	LAMA3	INHBA	GJC1	EXT1	COL13A1	BMP1
TNFRSF12A	SPRY4	PI15	MMP1	KLF7	IL1A	FST	DFNA5	CD276	AIM2
TMEM184B	SH2D5	PDPN	LPCAT1	ITGA6	IL11	FSCN1	DCBLD1	CAV1	AGRN

Figure 6.

Volcano plots of the differential expression genes (DEGs) in the GEO and TCGA databases. The red represents the up-regulated genes while the blue dot represents the down-regulated genes. (A) The volcano plots of the DEGs in GSE51985. (B) The volcano plots of the DEGs in GSE59102. (C) The volcano plots of the DEGs in GSE29330. (D) The volcano plots of the DEGs in GSE84957. (E) The volcano plots of the DEGs in TCGA. (F) The Kaplan-Meier survival curve of IL24 in laryngeal squamous cell cancer (LSCC) by RNA-Seq.

Table 6

The GO annotation and KEGG pathway enrichment analysis of the 49 intersectional genes

ID	Description	$P$ -value	Count	Gene ID
Biological process
GO: 0030198	Extracellular matrix organization	6.92E-17	16	TGFBI/TGFB1/SERPINE1/PDPN/MMP3/MMP1/LOXL2/LAMC2/ LAMA3, etc.
GO: 0043062	Extracellular structure organization	6.72E-16	16	TGFBI/TGFB1/SERPINE1/PDPN/MMP3/MMP1/LOXL2/LAMC2/ LAMA3, etc.
GO: 0022617	Extracellular matrix disassembly	2.40E-08	6	TGFB1/PDPN/MMP3/MMP1/FSCN1/BMP1
GO: 0097191	Extrinsic apoptotic signaling pathway	5.99E-08	8	TNFRSF12A/TGFB1/SERPINE1/ITGA6/INHBA/IL1A/CSF2/CAV1
GO: 2001236	Regulation of extrinsic apoptotic signaling pathway	9.21E-08	7	TNFRSF12A/SERPINE1/ITGA6/INHBA/IL1A/CSF2/CAV1
Cellular component
GO: 0031012	Extracellular matrix	2.03E-08	11	TGFBI/TGFB1/SERPINE1/MMP3/MMP1/LOXL2/LAMC2/ LAMA3, etc.
GO: 0005604	Basement membrane	9.33E-08	6	TGFBI/LOXL2/LAMC2/LAMA3/ITGA6/AGRN
GO: 0062023	Collagen-containing extracellular matrix	9.42E-08	9	TGFBI/TGFB1/SERPINE1/LOXL2/LAMC2/LAMA3/ITGA6/ COL13A1/AGRN
GO: 0030175	Filopodium	4.43E-06	5	PDPN/MYO1B/ITGA6/ITGA3/FSCN1
GO: 0071437	Invadopodium	7.48E-06	3	PDPN/ITGA3/FSCN1
Molecular function
GO: 0005125	Cytokine activity	6.01E-08	8	IL24/TGFB1/OSM/INHBA/IL1A/IL11/CSF2/BMP1
GO: 0048018	Receptor ligand activity	2.24E-06	9	IL24/TGFB1/PTHLH/OSM/INHBA/IL1A/IL11/CSF2/BMP1
GO: 0008083	Growth factor activity	3.34E-06	6	TGFB1/OSM/INHBA/IL11/CSF2/BMP1
GO: 0005126	Cytokine receptor binding	5.61E-06	7	TGFB1/OSM/ITGA5/INHBA/IL1A/IL11/CSF2
GO: 0050840	Extracellular matrix binding	1.13E-05	4	TGFBI/ITGA6/ITGA3/AGRN
Kyoto Encyclopedia of Genes and Genomes
hsa04512	ECM-receptor interaction	7.53E-07	6	LAMA2/LAMA3/ITGA6/ITGA5/ITGA3/AGRN
hsa05323	Rheumatoid arthritis	1.05E-06	6	TGFB1/MMP3/MMP1/IL1A/IL11/CSF2
hsa04640	Hematopoietic cell lineage	1.51E-06	6	ITGA6/ITGA5/ITGA3/IL1A/IL11/CSF2
hsa04060	Cytokine-cytokine receptor interaction	9.30E-06	8	IL24/TNFRSF12A/TGFB1/OSM/INHBA/IL1A/IL11/CSF2
hsa04510	Focal adhesion	8.30E-05	6	LAMA2/LAMA3/ITGA6/ITGA5/ITGA3/CAV1

Figure 7.

The columnar graphs of biological process (A), cellular component (B), and molecular function (C) of Gene Ontology (GO) annotation. The columnar graphs of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment (D). The color represents the adjusted $P$ -value. The smaller the adjusted $P$ -value, the higher the degree of gene enrichment and the more the color is biased toward red. On the contrary, it is biased towards blue ( $P$ . adjust: adjusted $P$ value).

3.4 Gene set enrichment analysis and PPI network construction

Through the analysis of microarrays and RNA-Seq based on the GEO and TCGA databases, we used R 3.5.2 to obtain the IL24-related genes and the DEGs of LSCC and finally obtained 49 genes through the intersection of the two, which are shown in Table 5. The volcano plots of the DEGs are shown in detail in Fig. 6A–E. Then we used the 49 genes for KEGG pathway enrichment analysis and GO annotation by use of the ClusterProfiler R package. In the biological process (BP) of GO analysis, it was suggested that the genes of difference expression in LSCC and those associated with IL24 were mainly involved in extracellular matrix organization, extracellular structure organization, and extracellular matrix disassembly, among other items. In the cellular component (CC) of GO analysis, the genes took part in the extracellular matrix, the basement membrane, and the collagen-containing extracellular matrix, among others. In the molecular function (MF) of GO analysis, it showed that the genes were associated with cytokine activity, receptor ligand activity, growth factor activity, and other types of activities. KEGG analysis revealed that the genes were connected with ECM-receptor interactions, rheumatoid arthritis, and hematopoietic cell lineage as well as other conditions. The top five results of KEGG pathway enrichment analysis and GO annotation are shown in Table 6; results of the top 10 are illustrated in bubble charts, shown in Fig. 7A–D.

The 49 genes mentioned above were also used to construct the PPI network by use of the STRING database (Fig. 8). After that, using Cytoscape v3.6.1 software, it indicated that hub genes were SERPINE1, TGFB1, MMP1, MMP3, CSF2, and ITGA5, along with others. The top 10 hub genes are shown in Fig. 9 by screening according to degree value.

Figure 8.

Protein-Protein Interaction network of the 49 correlated genes of IL24 constructed by STRING Tool for the Retrieval of Interacting Genes online database. Nodes represent proteins and edges represent protein-protein associations.

Figure 9.

The top 10 hub genes screened according to degree value.

4. Discussion

In this study, we found for the first time that the mRNA and protein levels of IL24 are both upregulated in laryngeal squamous cell carcinoma tissues as compared to non-cancerous tissues. Multiple detecting methods were used to verify the finding, including immunohistochemistry with clinical samples, gene chips, RNA-sequencing, and an integrated meta-analysis. More importantly, we also observed that the co-expressed genes of IL24 were mostly enriched in the pathways related to the extracellular matrix, which may assist with explaining its clinical significance.

LSCC is a common subgroup of HNSCC. At present, the main well-documented carcinogenic factors of the disease are HPV, smoking, and consuming alcohol [28, 29, 30, 31]. Due to the lack of clinical features of early LSCC, about 60% of patients are diagnosed as having advanced cancer and the five-year survival rate of LSCC is declining [32, 33]. So early diagnosis and timely treatment of LSCC are particularly important.

IL24 is a newly-identified member of the IL-10 family [16]. Research has shown that IL24 produced by immune cells is capable of inhibiting the growth of tumor cells by different ways, such as inhibiting angiogenesis, inducing specific mitochondrial dysfunction in cancer cells, inducing autophagy of cancer cells, and so on [19, 20, 21]. For example, recombinant IL24 could synergize with cisplatin to suppress the angiogenesis by down-regulating the expressions of VEGF (vascular endothelial growth factor), VEGF-C and PDGF-B (Platelet-derived growth factor) [34]. Recombinant sIL24 (smallest isoform of IL24) peptide could mediate the apoptosis of cancer cells by regulating Bax-2 and Bcl-2 [22]. A hybrid gene delivery vector called AdLTR2EF1 $\alpha$ -based vector was used to transfer IL-24 (AdLTR2EF1 $\alpha$ -IL-24) to treat a human oral squamous cell carcinoma cell line KB and immortal human keratinocyte cell line HaCaT [23]. This AdLTR2EF1 $\alpha$ -IL-24 treatment could lead to autophagy, which was interrupted by autophagy inhibitor 3-MA [23]. Similar effect of IL24 was reported [35], that a viral vector expressing mda-7/IL-24 (Ad. IL-24) infection could reduce miR-221, which subsequently increased toxic autophagy in MDA-MB-231 breast cancer cells [35]. Besides the direct effect on the cellular function of tumor, IL24 has also been reported to own therapeutically potential to treat various tumor cells [20, 36], including melanoma, carcinomas of breast, liver, prostate, ovarian, and nasopharyngeal [37]. For instance, the combination of oncolytic adenovirus armed with IL24 (ZD55-IL24) and docetaxel (DTX) showed synergistic inhibitory effects on prostate cancer cell viability and invasion [38]. Lentivirus-mediated IL24 expression was found to inhibit hepatocellular carcinoma (HCC) cells proliferation and colony formation [39]. Adenovirus-mediated human interleukin 24 (Ad-hIL24) transfection reduced cisplatin resistance in lung carcinoma cell line A549/DDP [40]. Thus, IL24 itself, or combined with other therapeutic methods, could become a target for the treatment of malignant tumors. There were also some similar studies that explored the treatment of human laryngeal cancer with adenovirus-mediated IL24 (Ad-IL24) in vivo and in vitro. IL24 (Ad-IL24) could reduce the expression of a proangiogenic factor: vascular endothelial growth factor which was involved in the inhibition of tumor angiogenesis and thus led to apoptosis of laryngeal cancer cells via intrinsic apoptotic pathways. Ad-IL24 had a strong selective killing activity in laryngeal cancer cells and was a potential candidate for laryngeal cancer gene therapy [41, 42]. But there have been no studies of the relationships between IL24 expression and LSCC in the protein or mRNA expression level so far, the purpose of this study was to explore the clinical significance and molecular mechanism of IL24 expression in LSCC, also to provide potential indicator for the prediction of prognosis and treatment effect of LSCC.

First, we mainly evaluated the clinical significance of IL24 in LSCC from the protein expression level and the mRNA expression level, respectively. In the protein expression level, a total of 49 LSCC samples and 26 normal samples were collected for immunohistochemical staining. The results showed that the IL24 protein expression was significantly higher in LSCC tissues than that in the controls ( $P<$ 0.0001). In the mRNA expression level, by mining the data from TCGA, GEO, SRA, ArrayExpress, and Oncomine databases and retrieving a large number of references, we obtained four microarrays and an RNA-Seq dataset that met the inclusion criteria of this study. A total of 172 LSCC and 81 non-LSCC cases were achieved to compare the IL24 mRNA expression. IL24 was consistently highly expressed ( $P<$ 0.0001) in LSCC. Furthermore, up-regulation of IL24 level had good diagnostic significance for distinguishing cancer from non-cancer patients by use of the sROC curve [AUC $=$ 0.89 (95% CI $=$ 0.86–0.92)]. IL24 is produced by a variety of immune cells, such as myeloid cells, lymphocytes, and monocytes, in response to treatment with lipopolysaccha-rides inhibitors or specific cytokines. In Th2 lymphocytes and T cells, in particular CD4 $+$ naive cells and memory cells activated by anti-CD3 monoclonal antibody, IL24 is caused by stimulation by phorbol myristate acetate and ionomycin. In monocytes, IL24 is induced by antigenic stimulation of lipopolysaccharide, concanavalin A and cytokines. In B-lymphocytes, IL24 expression can also be triggered by stimulating B-cell receptor signaling. In addition, nonlymphoid cells can also express IL24 by responding to cytokines secreted by immune cells [16, 21]. When cancer occurs, the body may protect itself by expressing IL24 and producing related cytokines. Therefore, IL24 was determined to be highly expressed at both the mRNA and protein levels in some LSCC patients; this could provide a potential reference for the screening of LSCC. We also analyzed the clinical role of IL24 expression in LSCC by using independent samples t-tests for clinicopathological factors such as age, sex, grades, stages, etc. With statistically significant difference, we found in these clinical parameters, the IL24 protein expression tended to be higher in LSCC patients who were in the stage III–IV ( $P=$ 0.007) or pathologic stage N1 ( $P=$ 0.044). It is interesting to observe that IL-24 expression in LSCC tissue was enhanced in advanced stages, which contracted its inhibiting cancer capacity in other reports. Consistent reports from literature could partially explain the enhancing effect of IL-24 in LSCC development. Treatment of human endothelial HECV cell with recombinant human interleukin IL-24 (rhIL-24) could boost cell migration [43]. It is possible that IL-24 can also promote the LSCC cell migration and cause tumor progression into more advanced clinical stage. Another example was in cutaneous squamous cell carcinoma (CSCC). Remarkably increased expression level of IL-24 was reported in invasive CSCC as compared to non-invasive ones, and the authors proposed a probable involvement of IL-24 to CSCC invasion via promoting focal expression of MMP7 [44]. Our current data also provide evidence that IL-24 may play a pivotal role in the metastasis of LSCC cells to lymph node, so as to accelerate LSCC to advanced stage. Survival analysis showed that LSCC patients with low IL24 expression might have a poor prognosis according to Kaplan-Meier estimator, but these were not statistically significant [P (log-rank) $=$ 0.5743]. The prognostic value of IL24 in LSCC remains to be first investigated with larger sample size. To summarize the above findings, IL24 may show higher level in the LSCC tissues, and it increases when the tumor is progressing.

In order to comprehensively and systematically explore the molecular mechanism of IL24, we not only calculated the IL24-related genes but also screened the differentially-expressed genes in LSCC by the microarrays and RNA-Seq datasets mentioned above. Then we took the intersection of the two because these intersection genes could more accurately illustrate the molecular mechanism of the link between IL24 expression and LSCC. The intersection genes were then used for KEGG pathway enrichment analysis and GO annotation. In the GO annotation, the top terms of BP, CC, and MP were extracellular matrix (ECM) organization (GO: 0030198), extracellular matrix (GO: 0031012), and cytokine activity (GO: 0005125), respectively. The top three items of KEGG pathway enrichment were ECM-receptor interaction (hsa04512), rheumatoid arthritis (hsa05323), and hematopoietic cell lineage (hsa04640). The GO and KEGG analyses indicated that the pathways related to extracellular matrix was the most powerful one for the biological function of IL24. As a crucial component of the cancer cell niche, the ECM provides the mechanical support for the tissue and mediates the cell-microenvironment interactions. Significantly, collagens which have been implicated in many aspects of neoplastic transformation, are one of the major proteins found within the ECM [45, 46], so ECM pathways play a key role in tumor shedding, adhesion, degradation, metastasis and proliferation [47]. Thus, the up-regulation of IL24 could play its accelerating role in the development of LSCC related to ECM.

Some well-known oncogenic or tumor suppressive genes may work together with IL24 by their interactions. The PPI network indicated that the hub genes of the co-expressed genes were SERPINE1 (serpin family E member 1), TGFB1 (transforming growth factor beta 1), MMP1 (matrix metallopeptidase 1), MMP3 (matrix metallopeptidase 3), CSF2 (colony stimulating factor 2), and ITGA5 (integrin subunit alpha 5). Among these hub genes, SERPINA1 has been reported with high abundance in saliva of patients with oral squamous cell carcinoma (OSCC), suggesting that SERPINA1 may be related to the occurrence of OSCC [48]. As OSCC and LSCC both originate from squamous cells, SERPINA1 may play consistent function in LSCC. The expression of TGFB1 was related to the pathological grade and clinical stage of LSCC [49]. Furthermore, inhibition of MMP1 could reduce the invasion ability of HNSCC cells [50]. Hence, the interactions between IL24 and SERPINA1, TGFB1 and MMP1 may exist in the progress of LSCC. However, due to the lack of direct evidence of the functions of the other three hub genes MMP3, CSF2, and ITGA5, their relationships with IL24 in LSCC remains to be explored.

It can be argued that IL24 is an immunomodulatory cytokine and an anti-tumor molecule with a wide range of activities including tumor-specific induction of apoptosis and inhibition of tumor angiogenesis. IL24 can be used as a tumor treatment drug to induce apoptosis, enhance the activation and function of T cells, and play a certain adjuvant role in the chemotherapy of many tumors. This also provides a new research idea for future clinical trials of IL24 as an adjuvant therapy for LSCC. In conclusion, our study displays that high expression IL24 has a certain relationship with the occurrence and progression of LSCC, which has potential diagnostic ability and therapeutic significance in LSCC.