Potential biomarkers for the early diagnosis of colorectal adenocarcinoma – transcriptomic analysis of four clinical stages

Abstract

BACKGROUNDS:

Colorectal cancer is the third most common cancer in economically developed countries. Molecular studies and, in particular, gene expression have contributed to advances in the diagnosis and treatment of many cancers. Genes can be molecular and therapeutic markers, but because of the large molecular diversity in colorectal cancer the knowledge is not yet fully established. Probably one of the most crucial processes during early cancer development is inflammation. The inflammatory response in the tumor is an important indicator of molecular etiology and later of cancer progression.

OBJECTIVE:

The aim of this work is to identify potential biomarkers for early stage of colorectal adenocarcinoma in patients’ bowel tissues using transcriptomic analysis.

METHODS:

Expression of the inflammatory response genes of colorectal cancer at all clinical stages (I–IV) and control of the bowel were evaluated by oligonucleotide microarrays.

RESULTS:

Based on statistical analysis many differentially expressed genes were selected. LCK (LCK Proto-Oncogene, Src Family Tyrosine Kinase), GNLY (granulysin), SLC6A6 (Solute-Carrier Family 6 Member 6) and LAMP2 (Lysosomal Associated Membrane Protein 2) were specific for the early stage of the disease. These genes had the properties of the good biomarkers.

CONCLUSIONS:

The expression of LCK, GNLY, SLC6A6 and LAMP2 genes could be valuable potential diagnostic biomarkers of the early stage of colorectal adenocarcinoma.

Keywords

Colorectal adenocarcinoma clinical stage microarray LCK GNLY SLC6A6 LAMP2 biomarkers

1. Introduction

Colorectal cancer (CRC) ranks third among all the cancers in the United States. A downward trend continues for both sexes, however still new cases are diagnosed every year among younger people [33]. There are similar epidemiological trends in Poland, since 1980 there has been an increasing number of new diagnoses, a 4-fold increase for males and for females about a 3-fold increase. Malignant tumours in the colon are responsible for more than 12% of cancer mortality [40]. The most commonly diagnosed CRC is adenocarcinoma. In spite of better and more accessible screening methods, it is most commonly recognized in advanced clinical stages. Current tests for the early detection of colorectal cancer do not give the expected results [15]. A lower risk for bowel cancer is visible for people taking non-steroidal anti-inflammatory drugs or who drink green tea, thus confirming the role of inflammatory mediators in this disease [32, 34].

Inflammatory mediators may create a microenvironment that promotes growth, cell migration and neoangiogenesis [29]. Oxidative stress in such cells and the presence of the tumour necrosis factor (TNF- $\alpha$ ) can cause DNA damage [41]. The nuclear factor kappa-light-chain-enhancer of activated B cells (NF- $\kappa$ B) can regulate the expression of proinflammatory cytokines, thus intensifying the inflammatory response [30, 35]. Cascade changes of many genes controlled via these transcription factors can activate processes leading to the inhibition of apoptosis, cell survival, and cell differentiation [10]. It is well known that after long-lasting inflammation there is a high probability of carcinogenesis being initialized in the tissues. Previous studies had shown that the regulatory effects of inflammatory factors on tumor growth and survival are partly due to cytokine synthesis, such as interleukin (IL)-6, IL-17A, IL-21. These interleukins contribute to the development of a tumor-promoting microenvironment. In turn, interferon (IFN)- $\gamma$ is expected to exert a dampening effect on the tumor [21]. The response to inflammation is closely related not only to colorectal cancer aetiology, but also with progression [35]. Detailed knowledge about the molecular mechanisms of colorectal cancer and various intestinal inflammatory conditions is based on mouse model studies [20]. However, cases of colorectal cancer are not sufficient and the commonly used markers are not consistent [28]. Based on these results and other cancer studies, whole genome analysis were performed showing various molecules, which are important for the development of colorectal cancer, such as: TGF $\beta$ , NF $\kappa B$ , TNF, WNT, IGF1 [3, 16]. Inflammation-related molecular markers in colorectal cancer are evident in many new therapies [18]. These markers have improved the predicted five-year survival in colorectal cancer patients, however the etiological and molecular differences contribute to many difficulties. In spite of this, no differences were developed between colon and rectum tumours or correlated with the sex of the patients [28]. CRC is a genetically heterogeneous disease and there is a necessity for a more detailed look at a particular pathway (e.g. inflammatory response) in patients at all stages.

This study is a part of the comprehensive subject of the CRC inflammatory microenvironment. It seems that this topic has been extensively developed, as currently evidenced by many review papers on the subject. Many sets of genes which may have significant meaning as diagnostic, prognostic or therapeutic factors have been stated in colorectal cancer. But, it is not enough, due to the molecular differences between subtypes of CRC. Many more studies are at the clinical trial stage and are not yet evaluated. Looking at the current knowledge on the topic and the continuous searching of new molecules playing a key role in the inflammatory response pathway (not limited to cancer), we have attempted to identify the biomarkers of early stage CRC which are differentially expressed compared to the control tissues. The results from oligonucleotide microarrays for the query – “inflammatory response pathway” based on the three sets of genes available in three bioinformatics databases were compared. The aim of our work was to identify genes that differentiate colorectal cancer from the microscopically healthy gut tissues (both unaccompanied by other diseases or therapies) in the inflammatory response pathway that characterizes its microenvironment. It is very important to find early markers of this disease that could provide unequivocal diagnostic information, but that could also be therapeutic targets. Thus, the selected-genes will be examined for their suitability as diagnostic markers and/or therapeutic targets in colorectal cancer patients.

2. Materials and methods

The material consisted of CRC and healthy colon tissue samples from 84 patients aged 51–86 years who underwent a standard diagnostic method (colonoscopy), followed by surgical resections due to CRC at Clinical Department of General, Colorectal, and Trauma Surgery of the Medical University of Silesia. The study was approved by the Bioethics Committee of the Medical University of Silesia (consent no. KNW/0022/KB1/ 21/I/10) and conforms to the standards set by the Declaration of Helsinki (printed in the British Medical Journal, 1964). All participants provided understanding and written consent for the use of their material in the present study.

Figure 1.

Self-Organizing Map of the investigated CRC samples and fragment of examined genes. (Explanations: CC – control colon; CSI – clinical stage I of CRC, CSII – clinical stage II of CRC, CSIII – clinical stage III of CRC, CSIV – clinical stage IV of CRC; colours in the heat map show the intensities of gene expression).

Table 1

Number of significant mRNAs in ANOVA test and Tukey post hoc test with Benjamini-Hochberg correction of inflammatory response pathway in different stage of CRC tissues

All mRNA	ANOVA	Clinical stage (CS) vs control (CC)
	$P<$ 0.05	CSI	CSII	CSIII	CSIV
(A) 5376	874	120	163	84	394
SuperPath 59	11	4	1	1	4
(B) 2598	413	63	76	42	178
(C) 740	111	15	18	10	49

Explanation: Three panels of mRNAs: (A) based on GeneCard database (additionally presented SuperPath of 30 genes related to 59 mRNAs) and (B) – based on NCBI Gene database, and (C) – based on Affymetrix database; CRC (colorectal cancer) in fourth clinical stages (CS) and control colon (CC) tissues; ANOVA – one-way analysis of variance; $P$ – probability value.

The criteria for exclusion from the study group included patients: diagnosed with a form of cancer other than CRC, with unclear histological confirmation of CRC, re-operated on due to the underlying disease, with genetic, metabolic or systemic conditions (as obesity, hyperlipidaemia, hyperglycaemia), after previous radio- and/or chemotherapy and undergoing or after hormone replacement therapy completed within the 5 years prior to the operation. The criteria for inclusion were CRC in histopathological assessments of tumour material obtained in the colonoscopies and post-surgery of patients in all stages of disease undergoing elective classical surgical procedures. The criteria for inclusion and exclusion found 37 study cases (18 CRCs and 19 controls), which come only from the Silesia area of Poland. Taking into consideration of the 7 ${}^{\rm th}$ edition of the AJCC/UICC staging system of CRC, the cancer tissues divided into 4 groups: Clinical Stage I (CSI, $n=$ 4), Clinical Stage II (CSII, $n=$ 5), Clinical Stage III (CSIII, $n=$ 6) and Clinical Stage IV (CSIV, $n=$ 4). Samples were from both men and women. All samples were stored at the temperature of $-$ 80 ${{}^{\circ}}$ C prior to the molecular analysis.

2.1 RNA isolation

The frozen postoperative control colon (CC) and CRC samples at different stage (CSI–CSIV) were homogenized. Total RNA was extracted using TRIzol ${}^{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}$ reagent (Invitrogen) and an RNeasy Mini kit (Qiagen), according to the manufacturer’s instructions. Then the obtained RNA was purified using columns an RNaseFree DNase Set (Qiagen) according to the manufacturer’s instructions. The quality and quantity of the RNA was tested spectrophotometrically with the use of GeneQuant II (Pharmacia BioTech) and electrophoretically using 0.8–1.0% agarose gel stained with ethidium bromide. Additionally, the degree of the total RNA integrity was estimated based on the RIN (RNA Integrity Number) parameter – rRNA ratio (28 s/18 s).

2.2 Microarray performance

Gene expression profiles were performed using an oligonucleotide microarray technique with GeneChip® Human Genome U133A (Affymetrix, actually Thermo Fisher Scientific), according to the manufacturer’s instructions (cDNA synthesis, biotinylated cRNA fragmentation and hybridized with the HGU133A microarray, stained with streptavidin-phycoerythrin). Quality control of genetic material at the subsequent steps of the experiment was performed by using 1% agarose gel electrophoresis.

The plate obtained from this process was scanned using GeneChip Scanner 3000 7G (Affymetrix). The data was processed for fluorescence signal values using the Affymetrix GeneChip Command Console Software (Affymetrix). Analysis of the 22,283 mRNA was performed and the degradation index of RNA was assessed using a 3’/5’ ratio (signal intensity ratio of the 3’ probe set over the 5’ probe set) of the GAPDH and ACTB (housekeeping genes probes: AFFXBioB_at, AFFXBioC_at, AFFXBioDn_at, AFFXCreX_at,AFFXr2EcBioB_at, AFFXr2EcBioC_at, AFFXr2EcBioD_at, AFFXr2P1cre_at). Then, the followingmRNA panels of probes (August 2017) in the “inflammatory response pathway” query: of 5376 sequences (3381 genes) and 59 mRNA (representing 30 genes) of SuperPath, signed in the GeneCards ${}^{{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}}$ : The Human Gene Database (http://www.genecards.org), and of 2598 (1624 genes) signed in the NCBI database (National Center for Biotechnology Information, http://www.ncbi.nlm.nih.gov/gene), and 740 signed in the Affymetrix NetAffx™ (http://www.affymetrix.com/analysis/index.affx) Analysis Centre database were selected.

Table 2
Comparison the Venn diagrams results of differentially expressed genes (DEGs; Tukey post hoc, $P<$ 0.05 and $-$ 2.0 $>$ FC $>$ 2.0) of fourth clinical stages (CS) of the CRC tissues compared to colon control (CC)

Panel	CSI vs CC	CSII vs CC	CSIII vs CC	CSIV vsCC
A	68 mRNA, 32 DEGs: $\uparrow$ CXCL2, CXCL1,CLDN18, SCL12A2,SRSF6, CD55, CXCL11,MUC5AC, MIK67, TAP1,PSMB9, BIRC5, KPNA2,KYNU, WISP3, FEN1,STAT1, GZMB, ETS2,LMNB2, WARS, LAMP3,GNLY, CD26; $\downarrow$ EPHX1, CUX1, FOXO1,TJP1, CPQ, PTN; $\uparrow\uparrow\uparrow\downarrow$ LCK, GNLY.	61 mRNA, 17 DEGs: $\uparrow$ MMP1, SCD, MYC,MMP3, F12, IFITM1,MET; $\downarrow$ AQP8, KIT, PLCE1,FGFR2, MMP28; $\uparrow\uparrow\uparrow\downarrow$ PROC2,ANAXA2(NHP2L1); $\downarrow\uparrow\uparrow\uparrow$ NQO1, SLC6A6,LAMP2.	40 mRNA, 16 DEGs: $\uparrow$ CDH11, THY1, VCAN,FAP, CXCR4, LTF, CD93,MMP19, IFITM3, WISP1; $\downarrow\uparrow\uparrow\uparrow$ COMP, GJA1, TIE1,CLEC11A; $\downarrow\downarrow\uparrow\uparrow$ IL1R1, SERPINA3	275 mRNA, 61 DEGs: $\uparrow$ THBS, TACSTD2,CYP2B6, TNFAIP6,TREM1, JUP(KRT17),COL10A1, NOX4,ALOX5, PKM, VCAN,CHI3L1, NACA, RBCK1,CA9, C2, VEGFA,COL7A1, CD44, INHBB,ITGAX, RGS16, FOSL1,APOC2, TLR2; $\downarrow$ TXNIP, FHL1, HBB,XK, RAP1A, TCF7L2,SCARB2, SLC25A4,ABCD3, PTP4A1,HADHA, CAST, SKP1,RNF11, PTP4A1,SMAD2, TCF7L2,GNA11, PDCD4,ACADM, COX7B,PGRMC1, PPP1CB,BAG1, SOCS2, LRRFIP2,GNAQ, C21orf33, PTEN; $\uparrow\downarrow\downarrow\downarrow$ KLF4, HSP90B1,PTP4A1, RARRES3,BCL2; $\downarrow\uparrow\uparrow\uparrow$ BMP4
B	34 mRNA, 19 DEGs: $\uparrow$ CXCL2, CXCL1, CD55,CXCL11, MUC5AC,NMU, DUSP4, PSMB9,BIRC5, KYNU, FEN1,STAT1, GZMB, EST2; $\downarrow$ EPHX1, CUX1, DST,FOXO1, PTN.	28 mRNA, 16 DEGs: $\uparrow$ MMP1, MET, MMP3,MYC, KCNN4, CHI3L1,SLC5A1, KLK6, GSTP1,BCL2L1, WNT5A, LYN; $\downarrow$ KIT, DST; $\downarrow\uparrow\uparrow\uparrow$ NQO1; $\uparrow\uparrow\uparrow\downarrow$ PROCR.	20 mRNA, 7 DEGs: $\uparrow$ FAP, CXCR4, LTF; $\uparrow\uparrow\uparrow\downarrow$ SELP; $\downarrow\downarrow\uparrow\uparrow$ IL1R1; $\downarrow\uparrow\uparrow\uparrow$ GJA1, MSR1	126 mRNA, 27 DEGs: $\uparrow$ TREM1, NOX4,ALOX5, PKM, CHI3L1,RBCK1, VEGFA,INHBB, ITGAX, RGS16,FOSL1, TLR2; $\downarrow$ TPM1, MTUS1,SMAD2, PDCD4,SMAD4, CAT, PSEM,SOCS2, LRRFIP2, PTEN; $\downarrow\uparrow\uparrow\uparrow$ BMP4; $\uparrow\downarrow\downarrow\downarrow$ KLF4, HSP90B1,RARRES3, BCL2
C	8 mRNA, 4 DEGs: $\uparrow$ CXCL2, CXCL1, CD55,CXCL11.	7 mRNA, 4 DEGs: $\uparrow$ F12, WNT5A, GSTP1; $\downarrow$ KIT.	6 mRNA, 2 DEGs: $\uparrow$ CXCR4, IL1R1.	35 mRNA, 7 DEGs: $\uparrow$ IL8, ALOX5, CD44, C2,TLR2; $\downarrow$ KLF4, TMED7

Explanation: The results from panels of 5376 (A) and 2598 (B), and 740 mRNAs (C) associated to term “inflammatory response pathway” in GeneCards and NCBI and Affymetrix database, respectively.

Table 3

Differentially expressed genes associated with the inflammatory response pathway (5376 mRNAs) with significant changes in the early stage of CRC

Probe set ID	Gene symbol	Gene name [39]	FDR $P$ -value	Fold change (CS vs CC)
				CSI	CSII	CSIII	CSIV
204891_s_at	LCK	LCK Proto-Oncogene, Src Family Tyrosine Kinase	0.013	3.06	1.00	1.91	$-$ 1.23
205495_s_at	GNLY	Granulysin	0.009	2.27	1.00	1.10	$-$ 1.04
205920_at	SLC6A6	Solute Carrier Family 6 Member 6	0.003	$-$ 1.06	1.86	1.30	1.58
200821_at	LAMP2	Lysosomal Associated Membrane Protein 2	0.009	$-$ 1.07	1.72	1.35	1.40

Explanations: ID – number identified probe on HG-U133A chip; FDR – false discovery rate; $P$ – probability; CS – clinical stage of examined colorectal cancer tissues; CC – control colon.

Figure 2.

List of significant biological processes in inflammatory response pathway (panel A; 874 significant mRNAs in the CRC, $P<$ 0.05). (Explanations: gene ontology over-representation test; highlighted bars of $P<$ 0.000000).

Figure 3.

Venn diagrams of differentially expressed genes in the CRC. (Explanations: DEGs involved in the inflammatory response pathway at four clinical stages (CS) of the CRC tissues compared to control colon (CC) in the results of 5376 (A) and 2598 (B), and 740 mRNAs (C) belonging to GeneCards and NCBI and NetAffx database, respectively).

Figure 4.

Normalized intensities of distinguished gene expression: LCK (A), GLNY (B), SLC6A6 (C), LAMP2 (D). (Explanations: Expression of genes are presented as normalized values, mean and standard error; whiskers show the 95% confidence interval).

2.3 Statistical analysis

Statistical analysis of the obtained data was performed using Statistica 12 software (StatSoft), Excel 2006 (Microsoft) and GeneSpring 13.0 software (Agilent Technologies). Values of 22,283 fluorescence signals of mRNA for the each HG-U133A chip were normalized using the RMA (Robust Multichip Average) method. Microarray quality control tests were carried out. The Shapiro-Wilk test was used to check the normality of the data distribution in selected panels of genes. Similarities and differences between samples were checked by the Self Organizing Map (SOM). Differentially expressed genes (DEGs) were determined using a one-way ANOVA test with asymptotic computation of $P$ -values and the post hoc Tukey HSD (High Significance Difference) with the Benjamini and Hochberg correction with a false discovery rate (FDR $P<$ 0.05) controlling procedure. Additionally, the criteria used for DEGs required the value of the fold change (FC) parameter greater or less than 2.0 ( $|$ FC $|$ $>$ 2.0). The cut-off for $P$ values was required to be less than 0.05 ( $P<$ 0.05). The visualization of high significance differences obtained in the post hoc Tukey HSD of the DEGs was presented as four-set Venn diagram. Analysis was completed by carrying out an over-representation test using PANTHER 12.0 software (Protein ANalysis THrough Evolutionary Relationships, http://www.pantherdb.org). All differences were statistically significant if the $P$ value was less than 0.05 ( $P<$ 0.05).

3. Results

To obtain normally distributed data the fluorescence signals were log-transformed. The similarities and differences between all 5 samples groups (CC, CSI, CSII, CSIII and CSIV) and genes were shown by implementing a SOM (Self Organising Map), the results can be seen in Fig. 1. A grouping algorithm was applied to both the tissues and the genes, revealing coherent patterns that suggest a high degree of gene expression in these tissues. The dendrogram shows similarities between CRC samples in different clinical stages and the heat map differences in expression of genes. The results indicate that the examined transcripts are divided into 5 groups differentiated by the stage of the cancer.

The next step was to perform comparative analysis of three mRNA panels consisting of panels A, B and C. The ANOVA and Tukey post hoc tests determined the number of statistically significant mRNAs in the examined panels of the “inflammatory response pathway”, the larger the sequences panel, the more statistically significant mRNAs are (Table 1). A similar trend was observed in a mutual comparison of transcriptomes between the groups at different stages of CRC and the control colon tissue.

Based on significant mRNAs (Table 1) an overrepresentation test was performed (Fig. 2). The obtained results show the following processes: response to stimulus, cellular processes, signal transduction, cell communication, response to stress, cell proliferation, cytokine – mediated signaling pathways and developmental, which have great significance in CRC carcinogenesis.

Additionally, based on the statistical analysis (the self-organizing map and one-way ANOVA combined with the post hoc Tukey test with the Benjamini-Hochberg correction) the obtained results were specified in a Venn diagram (the numbers of differentially expresses mRNAs in particular clinical stages of the CRC tissues compared to control colons are described within an ellipse), Fig. 3. The individual panels confirming a difference in the number of genes and in the DEGs are shown.

The overexpressed and downregulated genes in particular transcriptomes were mentioned in the Table 2, only genes with $-$ 2.0 $>$ FC $>$ 2.0 (marked with an arrow) are shown. The underlined genes in the Table 2 stand out in only one of the analyzed panels and show differences between individual mRNA one sets. Panel A comprises 70 underlined genes, panel B 16 and panel C 2.

All the distinguished genes are very important for the inflammatory response pathway in the CRC. But genes used for further analysis of potential biomarkers in the adenocarcinoma of CRC are shown in Table 3.

Selected DEGs from the early stage of CRC, namely CSI and CSII, are represented by LCK and GNLY, as well as, SLC6A6 and LAMP2 (Table 3). They have the features of potentially good markers. The values of FC for LCK and GNLY are much higher in the first stage of CRC development than in the controls, and this is visible in the subsequent decreases. However, in the case of SLC6A6 and LAMP2 which have a reduced expression in CSI, their value is strongly increasing in CSII.

Expression of each distinguished gene in fourth stages of CRC and healthy colon (CC) was presented on Fig. 4. Expression of each distinguished gene in fourth stages of CRC and healthy colon (CC) was presented on Fig. 4. Statistical significance was stated between CC and CSI for LCK (Fig. 4A; $P=$ 0.007) and GNLY (Fig. 4B; $P=$ 0.001) as well as between CC and CSII for SLC6A6 (Fig. 4C; $P=$ 0.004) and LAMP2 (Fig. 4D; $P=$ 0.005) in Tukey HSD post hoc test. All mRNAs were significantly different in comparison CSI to CSII ( $P<$ 0.04). Additionally, LCK presented significance between CSI and CSIV (Fig. 4A; $P=$ 0.01) and GNLY between CSI and CSIII, and CSIV (Fig. 4B; $P<$ 0.02). All significant differences presented in genes expression additionally confirm the results obtained earlier.

4. Discussion

The presented study shows that the number of genes differs in various databases in spite of entering the same query (Table 1), thus making it an important factor for the identification of DEGs. Hence, the most extensive database result was used for the over-representation test, which allowed ongoing biological processes in the analysed samples to be recognized. There were 36 important biological processes identified. However only 8 were highlighted due to high statistical significance related to the ongoing inflammatory response in the analyzed CRC thus all have an effect on inflammation in the tumor microenvironment. Although these molecules may be attributed to different signaling pathways, they are not unreasonably incorporated into the inflammatory response pathway. Specific genes that differentiate subsequent stages of the CRC were selected from all the database panels shown (FC $>$ 2.0, $P<$ 0.05 are shown in Fig. 3). The majority of DEGs presented in Fig. 3 and in Table 2 were described in literature as key molecules in CRC development. Amongst them were cytokines, chemokines, metalloproteinase and molecules associated with angiogenesis [18]. In this study, cytokines and chemokines were overexpressed mainly in CSI (CXCL1, CXCL2, CXCL11), but also in CSIII (CXCR4). These early stage inflammatory mediators are described as promoters of angiogenesis and stimulators of metastasis in CRC [14]. Metalloproteinase expression was higher than in the control group (MMP1, MMP3 upregulated and MMP28 downregulated in CSII and MMP19 upregulated in CSIII; Fig. 3), as well as angiogenesis associated proteins (e.g. VEGFA in CSIV). Their role in cancer progression is crucial and can result in a poor prognosis [10, 43]. Agesen et al. 2012 has included MMP3 in the 13-genes, whose expression is classified as a predictive model for relapse in stage II CRC patients(ColoGuideEx test). The ColoGuideEx test confirms patients have a good prognosis when gene expression is high [2]. We can also see in this study important tumour suppressor genes known as SMAD2 and SMAD4 are downregulated in the CSIV, the loss of which leads to the ability to evade apoptosis and deregulation of the cell cycle. These genes are proposed as predictive markers for patient outcome [22]. All DEGs mentioned in the CSI play a key role in carcinogenesis. MIK67 is a marker of proliferation and a prognostic factor in premenopausal women with node-negative early breast cancer [6]. Similarly to breast cancer, MIK67 was included in the OncoDX (the Oncotype DX Colon Cancer Assay) test as one of the seven recurrence genes [7]. The FDA approved gene-expression based multi-gene prognostic tests for stage II-III colorectal cancer, but because of its heterogeneity it is unknown how many molecular subtypes describe these tumours. All mentioned tests are in clinical trials, time will show their usefulness.

The analysis of the Venn diagram results (Table 2) allowed four genes which can be good diagnostic and/or therapeutic markers of early stage CRC adenocarcinoma to be chosen. Amongst these genes are LCK, GNLY, SLC6A6 and LAMP2 (Table 3).

The LCK gene encoding lymphocyte-specific protein tyrosine kinase belongs to the SRC family which consists of such genes as SRC, FYN, YES, LYN, HCK, FGR and BLK. The gene plays a crucial role during signal transduction after T-cell receptor or FGFR2 activation. Furthermore, the FGF receptor 2 was down-regulated in the analysed tissue. Moreover, the LYN (up-regulated in CSII; Table 2, panel B) can activate FGFR2 and mediate cell division, growth and differentiation [31]. These results comply with other studies where FGFR2 expression is changed, however, strong expression was at the invasive front of CRC [23].

After activation, the LCK interacts with a large number of molecules among which are CD55, LYN and KIT. Whereas induced expression of CD55 is correlated with poor prognosis of CRC patient [9]. In turn, LYN expression is correlated with patient survival [11]. However, KIT expression in most cancers is reduced due to mutations in this gene [37]. All of these molecules are visible in our studies (Table 2) as DEGs.

Studies performed on LCK-knockout mice show a dramatic reduction in the number of peripheral T-cells. Furthermore, defective expression of LCK causes severe combined immunodeficiency in humans, but the oncogenic potential of LCK is visible through dysregulated expression in various cancers [1]. In non-small-cell lung cancer, LCK expression was determined as a good prognostic factor [5]. The observed overexpression of LCK (FC $=$ 3.06; FDR $P=$ 0.013) and CD55 (FC $=$ 3.72; FDR $P=$ 0.029) in stage I (CSI) can be seen as a contrary effect due to LCK giving a good prognosis and CD55 indicating poor patient survival. Hence, the prognosis probably does not depend on a single molecule. However, elevated LCK expression demonstrate the good immune status of the patient, as is apparent at CSI. However, its decline is observed in subsequent stages of the CRC. This demonstrates that LCK is an indicator of the patient’s immune status and it is important that we should care about its level in bowel of the patient during oncological therapy. It is a gene of our special interest, so we nominated it as a potentially good biomarker in CRC.

In the advanced stages of CRC (CS IV), the GNLY gene (granulysin, cytolytic protein), which is present in cytotoxic T cells and natural killer cells, was down-regulated [13], similar to our results. In gastric carcinoma its protein is a prognostic factor and was observed only in clinical stage IV [27]. Camus et al. 2009 indicated that the co-expression of GNLY and Th1 adaptive immune responses (IRF1) accurately predicted patient survival independently to the metastatic status and high intratumoral VEGF mRNA expression. Furthermore, it was associated with significantly reduced survival rates in CRC patients expressing high mRNA levels of granulysin [4]. Our results showed the high expression of GLNY in the CSI, moderate in II and III and low in IV (Table 3), therefore it can be a “good” indicator of prognosis. This gene encodes a protein with antibacterial properties. Due to the different etiology of CRC (changed intestinal microbiome) and, among other things, many subtypes of this tumour, the gene may be a key biomarker. Nevertheless, in this case, supplementary studies at protein level should be carried out.

In contrast to the previously presented genes, the SLC6A6 expression is reduced in the CSI and is rapidly rising in CSII and afterwards this stops and it maintains its level in advanced clinical stages. The gene plays an important role in cell survival and maintenance [42]. In large B cell lymphoma its expression is a robust prognostic factor [17]. Furthermore, it has a neurotransmitter transporter function through which it mediates the uptake of 5-aminolevulinic acid by cancerous cells and causes the accumulation of protoporphyrin IX and cell photodamage [36]. The lack of reports about SLC6A6 in CRC tissues indicate that our findings are the first to show its relation to control tissues and its potential as a biomarker in colorectal adenocarcinoma.

LAMP2 is a not well-known gene encoding membrane glycoprotein which provides selectins with a carbohydrate ligand. Its high expression causes poor prognosis in patients with esophageal squamous cell carcinoma and correlates with tumor histology [19]. In prostate cancer LAMP2 was indicated as a potential biomarker and target for the neuroendocrine subtype [24]. Furthermore, the expression of LAMP2 was significantly higher in salivary gland adenoid cystic carcinoma [12]. Through elevated expression of LAMP2 in breast tumour tissue, the survival mechanism via chaperone-mediated autophagy is observed [26]. Our findings are the first to indicate LAMP2 as a potential biomarker in CRC.

Additional confirmation of the gene expression results (Fig. 4) indicate the statistical significance of the differences between the control and CRC in the early stages.

The limitation of this work is a small group of patients meeting the inclusion criteria for the study. In addition, this group comes from one surgical and geographical center. In turn, this can be an advantage, because the group is highly homogeneous. An additional limitation is the lack of independent validation and validation of selected genes for individual stages of CRC. However, each gene has already been mentioned or analysed in another studies regarding CRC [8, 25, 38, 42]. These genes are known to be expressed in colorectal cancer, and their importance is described but not in the context of potential biomarkers. Another limitation but perhaps an advantage is that the analysis of the microarrays in the search for biomarkers only concerns the inflammatory response pathway. The limitation of this work is also that the highlighted genes were not estimated in the peripheral blood of patients.

Both LCK and GNLY are associated with a lymphocytic inflammatory response and have a similar expression in colorectal cancer tissue samples, but differ in their role. The first one is involved in T lymphocyte activation and maturation and the second in the placing of cytotoxic granules of T cells and encoding protein which has antibacterial properties. Their expression is elevated in stage I but in stage IV it is strongly reduced. In turn, SLC6A6 and LAMP2 are down-regulated in stage I, and in stage II they are overexpressed. All of the distinguished genes indicate significant differences between control colon and clinical stage I or II of the CRC.

In conclusion, the mentioned genes are good candidates to become potential diagnostic biomarkers of the early stage of colorectal adenocarcinoma.

Footnotes

Acknowledgments

This study was supported by the Medical University of Silesia statutory work (KNW-1-013/K/7/0 and KNW-1-019/K/6/0). This research was supported in part by PL-Grid Infrastructure.

References

Abraham

K.M.

Levin

S.D.

Marth

J.D.

Forbush

K.A.

and Perlmutter

R.M.

, Thymic tumorigenesis induced by overexpression of p56lck, Proceedings of the National Academy of Sciences of the United States of America 88 (1991), 3977–3981.

Ågesen

T.H.

Sveen

Merok

M.A.

Lind

G.E.

Nesbakken

Skotheim

R.I.

and Lothe

R.A.

, ColoGuideEx: A robust gene classifier specific for stage II colorectal cancer prognosis, Gut 61 (2012), 1560–1567.

Benson

A.B.

and Hamilton

S.R.

, Path toward prognostication and prediction: An evolving matrix, Journal of Clinical Oncology 29 (2011), 4599–4601.

Camus

Tosolini

Mlecnik

Pages

Kirilovsky

Berger

Costes

Bindea

Charoentong

Bruneval

Trajanoski

Fridman

W.H.

and Galon

, Coordination of intratumoral immune reaction and human colorectal cancer recurrence, Cancer Research 69 (2009), 2685–2693.

Chen

H.Y.

S.L.

Chen

C.H.

Chang

G.C.

Chen

C.Y.

Yuan

Cheng

C.L.

Wang

C.H.

Terng

H.J.

Kao

S.F.

Chan

W.K.

H.N.

Liu

C.C.

Singh

Chen

W.J.

Chen

J.J.W.

and Yang

P.C.

, A five-gene signature and clinical outcome in non-small-cell lung cancer, New England Journal of Medicine 356 (2007), 11–20.

Clahsen

P.C.

van de Velde

C.J.

Duval

Pallud

Mandard

A.M.

Delobelle-Deroide

van den Broek

Sahmoud

T.M.

and van de Vijver

M.J.

, P53 protein accumulation and response to adjuvant chemotherapy in premenopausal women with node-negative early breast cancer, Journal of Clinical Oncology 16 (1998), 470–479.

Clark-Langone

K.M.

Sangli

Krishnakumar

and Watson

, Translating tumor biology into personalized treatment planning: Analytical performance characteristics of the Oncotype DX Colon Cancer Assay, BMC Cancer 10 (2010), 691.

Huang

Peng

Yao

Zhao

Yang

Wang

Cao

Zhu

W.G.

and Gu

, 5-Fluorouracil targets histone acetyltransferases p300/CBP in the treatment of colorectal cancer, Cancer Letter 400 (2017), 183–193.

Durrant

L.G.

Chapman

M.A.

Buckley

D.J.

Spendlove

Robins

R.A.

and Armitage

N.C.

, Enhanced expression of the complement regulatory protein CD55 predicts a poor prognosis in colorectal cancer patients, Cancer Immunology and Immunotherapy 52 (2003), 638–642.

10.

Goos

J.A.C.M.

V de Cuba

E.M.

Coupe

V.M.H.

Diosdado

Delis-Van Diemen

P.M.

Karga

Belien

J.A.M.

Menke-Van der Houven van Oordt

C.W.

Geldof

A.A.

Meijer

G.A.

Hoekstra

O.S.

and Fijneman

R.J.A.

, Glucose transporter 1 (SLC2A1) and vascular endothelial growth factor A (VEGFA) predict survival after resection of colorectal cancer liver metastasis, Annals of Surgery 263 (2016), 138–145.

11.

Hao

J.M.

Chen

J.Z.

Sui

H.M.

Si-Ma

X.Q.

G.Q.

Liu

J.L.

Ding

Y.Q.

and Li

J.M.

, A five-gene signature as a potential predictor of metastasis and survival in colorectal cancer, Journal of Pathology 220 (2010), 475–489.

12.

Huang

C.F.

Deng

W.W.

Zhang

W.F.

and Sun

Z.J.

, Expression of LC3, LAMP2, KEAP1 and NRF2 in salivary adenoid cystic carcinoma, Pathology and Oncology Research 22 (2016), 109–114.

13.

Huang

L.P.

Lyu

S.C.

Clayberger

and Krensky

A.M.

, Granulysin-mediated tumor rejection in transgenic mice, Journal of Immunology 178 (2007), 77–84.

14.

Itatani

Kawada

Inamoto

Yamamoto

Ogawa

Taketo

M.M.

and Sakai

, The role of chemokines in promoting colorectal cancer invasion/metastasis, International Journal of Molecular Science 17 (2016), E643.

15.

Keane

M.G.

and Johnson

G.J.

, Early diagnosis improves survival in colorectal cancer, Practitioner 256 (2012), 15–18.

16.

Kennedy

R.D.

Bylesjo

Kerr

Davison

Black

J.M.

Kay

E.W.

Holt

R.J.

Proutski

Ahdesmaki

Farztdinov

Goffard

Hey

McDyer

Mulligan

Mussen

O’Brien

Oliver

Walker

S.M.

Mulligan

J.M.

Wilson

Winter

O’Donoghue

Mulcahy

O’Sullivan

Sheahan

Hyland

Dhir

Bathe

O.F.

Winqvist

Manne

Shanmugam

Ramaswamy

Leon

E.J.

Smith

W.I.J.

McDermott

Wilson

R.H.

Longley

Marshall

Cummins

Sargent

D.J.

Johnston

P.G.

and Harkin

D.P.

, Development and independent validation of a prognostic assay for stage II colon cancer using formalin-fixed paraffin-embedded tissue, Journal of Clinical Oncology 29 (2011), 4620–4626.

17.

Kim

S.J.

Sohn

I.-G.

Jung

S.H.

Y.H.

Yoo

H.Y.

Paik

and Kim

W.S.

, Gene expression profiles for the prediction of progression-free survival in diffuse large B cell lymphoma: Results of a DASL assay, Annals of Hematology 93 (2014), 437–447.

18.

Lasry

Zinger

and Ben-Neriah

, Inflammatory networks underlying colorectal cancer, Nature Immunology 17 (2016), 230–240.

19.

Wang

Zhang

Liu

and Huang

, High expression of LAMP2 predicts poor prognosis in patients with esophageal squamous cell carcinoma, Cancer Biomarkers 19 (2017), 305–311.

20.

Long

A.G.

Lundsmith

E.T.

and Hamilton

K.E.

, Inflammation and colorectal cancer, Current Colorectal Cancer Reports 13 (2017), 1–11.

21.

Mantovani

Allavena

Sica

and Balkwill

, Cancer-related inflammation, Nature 454 (2008), 436–444.

22.

Marmol

Sanchez-de-Diego

Pradilla Dieste

Cerrada

and Rodriguez Yoldi

M.J.

, Colorectal carcinoma: A general overview and future perspectives in colorectal cancer, International Journal of Molecular Science 18 (2017), E197.

23.

Matsuda

Ishiwata

Yamahatsu

Kawahara

Hagio

Peng

W.X.

Yamamoto

Nakazawa

Seya

Ohaki

and Naito

, Overexpressed fibroblast growth factor receptor 2 in the invasive front of colorectal cancer: A potential therapeutic target in colorectal cancer, Cancer Letter 309 (2011), 209–219.

24.

Morell

Bort

Vara-Ciruelos

Ramos-Torres

Altamirano-Dimas

Diaz-Laviada

and Rodriguez-Henche

, Up-regulated expression of LAMP2 and autophagy activity during neuroendocrine differentiation of prostate cancer LNCaP cells, PLoS One 11 (2016), e0162977.

25.

Pagès

Berger

Camus

Sanchez-Cabo

Costes

Molidor

Mlecnik

Kirilovsky

Nilsson

Damotte

Meatchi

Bruneval

Cugnenc

P.H.

Trajanoski

Fridman

W.H.

and Galon

, Effector memory T cells, early metastasis, and survival in colorectal cancer, New England Journal of Medicine 353 (2005), 2654–2666.

26.

Saha

, LAMP2A overexpression in breast tumors promotes cancer cell survival via chaperone-mediated autophagy, Autophagy 8 (2012), 1643–1656.

27.

Saigusa

Ichikura

Tsujimoto

Sugasawa

Majima

Kawarabayashi

Chochi

Ono

Kinoshita

Seki

Ogawa

and Mochizuki

, Serum granulysin level as a novel prognostic marker in patients with gastric carcinoma, Journal of Gastroenterology and Hepatology 22 (2007), 1322–1327.

28.

Sanz-Pamplona

Berenguer

Cordero

Riccadonna

Solé

Crous-Bou

Guinó

Sanjuan

Biondo

Soriano

Jurman

Capella

Furlanello

and Moreno

, Clinical value of prognosis gene expression signatures in colorectal cancer: A systematic review, PLoS One 7 (2012), e48877.

29.

Scarpa

Castagliuolo

Castoro

Pozza

Scarpa

Kotsafti

and Angriman

, Inflammatory colonic carcinogenesis: A review on pathogenesis and immunosurveillance mechanisms in ulcerative colitis, World Journal of Gastroenterology 20 (2014), 6774–6785.

30.

Schottelius

A.J.

and Dinter

, Cytokines, NF-kappaB, microenvironment, intestinal inflammation and cancer, Cancer Treatment Research 130 (2006), 67–87.

31.

Sefton

B.M.

, The lck tyrosine protein kinase, Oncogene 6 (1991), 683–686.

32.

Shirakami

Ohnishi

Sakai

Tanaka

and Shimizu

, Prevention of colorectal cancer by targeting obesity-related disorders and inflammation, International Journal of Molecular Science 18 (2017), E908.

33.

Siegel

R.L.

Miller

K.D.

and Jemal

, Cancer statistics, 2017, CA Cancer Journal for Clinicians 67 (2017), 7–30.

34.

Smalley

W.E.

and DuBois

R.N.

, Colorectal cancer and nonsteroidal anti-inflammatory drugs, Advances in Pharmacology 39 (1997), 1–20.

35.

Terzic

Grivennikov

Karin

and Karin

, Inflammation and colon cancer, Gastroenterology 138 (2010), 2101–2114.e5.

36.

Tran

T.T.

Adachi

and Taketani

, Neurotransmitter transporter family including SLC6A6 and SLC6A13 contributes to the 5-aminolevulinic acid (ALA)-induced accumulation of protoporphyrin IX and photodamage, through uptake of ALA by cancerous cells, Photochemistry and Photobiology 90 (2014), 1136–1143.

37.

Wain

H.M.

Bruford

E.A.

Lovering

R.C.

Lush

M.J.

Wright

M.W.

and Povey

, Guidelines for human gene nomenclature, Genomics 79 (2002), 464–470.

38.

Wang

Sun

Yang

and Zhou

, SCF/ C-Kit/JNK/AP-1 Signaling pathway promotes claudin-3 expression in colonic epithelium and colorectal carcinoma, International Journal of Molecular Science 18 (2017).

39.

Watanabe

Kobunai

Yamamoto

Kanazawa

Konishi

Tanaka

Matsuda

Ishihara

Nozawa

Eshima

Muto

and Nagawa

, Prediction of liver metastasis after colorectal cancer using reverse transcription-polymerase chain reaction analysis of 10 genes, European Journal of Cancer 46 (2010), 2119–2126.

40.

Wojciechowska

and Didkowska

, Cancers in Poland, Krajowy Rejestr Nowotworów, Centrum Onkologii – Instytut im. Marii Skłodowskiej-Curie (2017). http://onkologia.org.pl/raporty/4/8/2017.

41.

Yan

Wang

Rabbani

Z.N.

Zhao

Yuan

Dewhirst

M.W.

and Li

C.Y.

, Tumor necrosis factor-alpha is a potent endogenous mutagen that promotes cellular transformation, Cancer Research 66 (2006), 11565–11570.

42.

Yasunaga

and Matsumura

, Role of SLC6A6 in promoting the survival and multidrug resistance of colorectal cancer, Science Reports 4 (2014), 4852.

43.

Herazo-Maya

J.D.

Nukui

Romkes

Parwani

Juan-Guardela

B.M.

Robertson

Gauldie

Siegfried

J.M.

Kaminski

and Kass

D.J.

, Matrix metalloproteinase-19 promotes metastatic behavior in vitro and is associated with increased mortality in non-small cell lung cancer, American Journal of Respiratory and Critical Care Medicine 190 (2014), 780–790.

Potential biomarkers for the early diagnosis of colorectal adenocarcinoma – transcriptomic analysis of four clinical stages

Abstract

BACKGROUNDS:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Introduction

2. Materials and methods

2.2 Microarray performance

Table 2 Comparison the Venn diagrams results of differentially expressed genes (DEGs; Tukey post hoc, P < 0.05 and - 2.0 > FC > 2.0) of fourth clinical stages (CS) of the CRC tissues compared to colon control (CC)

3. Results

4. Discussion

Footnotes

Acknowledgments

References

Table 2
Comparison the Venn diagrams results of differentially expressed genes (DEGs; Tukey post hoc, $P<$ 0.05 and $-$ 2.0 $>$ FC $>$ 2.0) of fourth clinical stages (CS) of the CRC tissues compared to colon control (CC)