Novel reference genes in colorectal cancer identify a distinct subset of high stage tumors and their associated histologically normal colonic tissues
- 156 Downloads
Reference genes are often interchangeably called housekeeping genes due to 1) the essential cellular functions their proteins provide and 2) their constitutive expression across a range of normal and pathophysiological conditions. However, given the proliferative drive of malignant cells, many reference genes such as beta-actin (ACTB) and glyceraldehyde-3-phosphate-dehydrogenase (GAPDH) which play critical roles in cell membrane organization and glycolysis, may be dysregulated in tumors versus their corresponding normal controls
Because Next Generation Sequencing (NGS) technology has several advantages over hybridization-based technologies, such as independent detection and quantitation of transcription levels, greater sensitivity, and increased dynamic range, we evaluated colorectal cancers (CRC) and their histologically normal tissue counterparts by NGS to evaluate the expression of 21 “classical” reference genes used as normalization standards for PCR based methods. Seventy-nine paired tissue samples of CRC and their patient matched healthy colonic tissues were subjected to NGS analysis of their mRNAs.
We affirmed that 17 out of 21 classical reference genes had upregulated expression in tumors compared to normal colonic epithelial tissue and dramatically so in some cases. Indeed, tumors were distinguished from normal controls in both unsupervised hierarchical clustering analyses (HCA) and principal component analyses (PCA). We then identified 42 novel potential reference genes with minimal coefficients of variation (CV) across 79 CRC tumor pairs. Though largely consistently expressed across tumors and normal control tissues, a subset of high stage tumors (HSTs) as well as some normal tissue samples (HSNs) located adjacent to these HSTs demonstrated dysregulated expression, thus identifying a subset of tumors with a potentially distinct and aggressive biological profile.
While classical CRC reference genes were found to be differentially expressed between tumors and normal controls, novel reference genes, identified via NGS, were more consistently expressed across malignant and normal colonic tissues. Nonetheless, a subset of HST had profound dysregulation of such genes as did many of the histologically normal tissues adjacent to such HSTs, indicating that the HSTs so distinguished may have unique biological properties and that their histologically normal tissues likely harbor a small population of microscopically undetected but metabolically active tumors.
KeywordsColorectal reference genes High stage tumors And molecular abnormalities in tumor adjacent tissues
Coefficients of variation
Fragments per kilobase of transcript per million mapped reads
Hierarchical clustering analyses
Normal tissue adjacent to high stage tumor
High stage tumor
Normal tissue adjacent to low stage tumor
Low stage tumor
Next generation sequencing
Principal component analyses
Reference gene coexpressed gene
Basic cellular functions are supported by guaranteed expression of genes encoding proteins mediating important proteins for cellular integrity. Such genes have been referred to as “housekeeping” genes or, for purposes of comparison of gene expression levels across different cell populations, as “reference” genes. While all cells require the functions of proteins encoded by such genes, the uniformity of expression levels in distinct cells and tissues is not confirmed, as diverse physiological conditions and disease states impose different metabolic and structural requirements [1, 2, 3, 4]. Often utilized or “classical” reference genes have been identified with roles in essential biological processes including molecular transport, RNA metabolism, oxidative phosphorylation, proteolysis, protein translation, regulation of protein metabolism and cell cycle control . Although various tools like Genorm, NormFinder, or BestKeeper have each defined a suitable set of classical reference genes for specific qPCR studies, recent cancer studies found that normalization of gene expression levels using classical ACTB and GAPDH introduced artifacts in qPCR results because of non-uniformity of reference gene expression in mouse fibroblasts  and in human cancer lines . Since NGS is quantitated directly as fragments per kilobase per of transcript per million mapped reads (FPKM), without the need for normalization by reference genes, we used NGS to examine both relative and absolute gene expression levels of 21 classical reference genes in CRC and their respective normal tissues. We also inquired into the presence of no novel reference genes, better suited for quantitative purposes in PCR based assays, based on limited CV across 79 CRC tumor pairs.
Original CRC cohort
Seventy-nine paired-tissues (79 tumor and 79 normal controls, Additional file 1: Table S1) of pretreatment CRCs were collected from 38 male and 41 female patients by Indivumed GmbH (Germany) for mRNA sequencing. The purchase of these samples was approved by U. S Food & Drug Administration Institutional Review Boards and Research Involving Human Subjects Committee. To evaluate tumor content, hematoxylin and eosin stained microscopic slices were examined by pathologists to determine the tumor cell and normal cell areas, respectively. Histologically, tumor samples had 50–70% content of cancer cells while normal samples had 0% content of cancer cells. Normal tissues were collected from a site at a minimum of 5 cm from the tumor margin. Ischemia time was 6–11 min. This short cold ischemia reduces postsurgical tissue processing artifacts . According to the medical pathology report, tumors were classified as well, moderately, and poorly differentiated tumors following international guideline UICC TNM-classification . For the convenience of analysis, 26 stage I and II tumors were considered as low stage tumors (LSTs), while 53 stage III and IV tumors were considered as HSTs. In this study, a normal control adjacent to a low stage tumor is referred as LSN. The ratio of high stage tumors vs. low stage tumors is 2 to 1. Among 26 low stage tumors, there were 2 either lymph node (LN) or lymphatic vessel (LV) positive tumors while among 53 high stage tumors, there were 28 either LN/LV positive tumors. For tumor grades, there were 17 well (low grade) differentiated, 36 moderately (medium grade) differentiated, and 26 poorly (high grade) differentiated tumors. Clinical and histopathological characteristics of the patients as well as tumor location are summarized in Additional file 1: Table S1. Among these 80 tumor pairs, 79 pairs were sequenced except the T7/N7 pair [10, 11, 12].
TCGA CRC validation cohort
Because, we studied reference genes across both tumor and normal samples, we only selected the patient matched 50 CRC pairs (100 samples) available from TCGA38 from OncoLand (TCGA38 contains 50 paired CRCs and 589 unpaired CRCs). In respect of reference genes as potential biomarkers for HST/HSN, we specifically compiled tumor stage information for 50 CRCs. Due to the fact that single data banks, such as CBioPortal, do not contain all of the relevant CRC information, we had to extract tumor staging information for 50 CRCs from three different data banks (the Human Protein Atlas, the Stanford Cancer Genome Atlas Analysis of colorectal cancer and cBbioPortal) (https://www.proteinatlas.org/news/tag/tcga, http://genomeportal.stanford.edu/tcga-crc/get_feature_samples?filename=Y_COADREAD_2013-01-16_CancerGenes_Integrative_ClinicalStage.txt and https://www.cbioportal.org). As result, this 50 CRC cohort contains 32 low stage (I/II) CRC pairs (64 LST/LSNs) and 18 high stage (III/IV) CRC pairs (36 HST/HSNs) (Additional file 1: Table S2). To validate results obtained from our 79 paired samples, gene expression (FPKM) information related to 6 CRC hallmark genes, 21 classical reference genes, 42 novel reference genes and 8 reference gene coexperssed genes of 50 CRC pairs were downloaded.
RNA quality was assessed using the Agilent 2100 Bioanalyzer, with cellular RNA analyzed using the RNA 6000 Nano Kit (Agilent). Samples with an RNA Integrity Number (RIN) of 7 or higher were processed to generate libraries for mRNA sequencing following the Illumina® TruSeq Stranded mRNA Sample Preparation Guide. In this method, poly-A mRNAs were purified from 0.5 μg total RNA, fragmented and reverse-transcribed into cDNAs. Double strand cDNAs were adenylated at the 3′ ends and ligated to indexed sequencing adaptors, followed with briefly amplification for 15 cycles. One femtomole of the sequencing libraries (median size ~ 260 nt) were denatured and loaded onto a flow cell for cluster generation using the Illumina cBot. Every six samples were loaded onto each lane of a rapid run flow cell. Paired-end sequencing was carried out on HiSeq 2500 sequencer (Illumina, San Diego, CA, USA) for 100 × 2 cycles. For each sample, we obtained ~ 50 million 100-bp reads that passed preset filtering parameters [10, 11, 12].
Sequencing data analysis
For mRNA sequencing, Tophat V.2.0.11 was used to align reads in fastq files to the UCSC human hg19 reference genome. Cufflinks V.2.2.1 was used to assemble the transcriptome based on the hg19 reference annotation, and Cuffquan/Cuffnorm (part of Cufflinks) were used in calculating relative abundance of each transcript reported as FPKM. ANOVA test was conducted (on Partek genomics suite) to identify mRNAs with differential expression between tumors and matched normal adjacent tissues using the threshold False Discovery Rate (FDR) ≤ 0.05. The unsupervised hierarchical clustering analysis (HCA) and principal component analysis (PCA) were used to explore the gene expression profiles on ArrayTrack (the National Center for Toxicological Research, U.S. Food and Drug Administration). The FPKMs from samples were log2 transformed and then z-score transformed for HCA and PCA plot. We determined tumor and normal sample outliers in PCA results as in our previous study [11, 12]. In brief, we manually picked a center point and used L2 distance to determine whether one node is inside or outside a boundary marked by a dashed circle. Then, CHITEST (excel 2016) was used to determine the differential location between HST/HSNs and LST/LSNs in PCA. The Student’s t-test (excel 2016) was used to detect differential CV between low and high stage tumors while Pearson correlation analysis was to detect the correlation between NGS and qPCR. The reference gene co-expression analyses were carried by Partek NGS & microarray data analysis software. Correlations were transformed to Fisher’s z-score using online tool (http://onlinestatbook.com/calculators/fisher_z.html) before averaging and retransforming with an inverse Fisher-Z. Gene ontology (GO) Integrated Discovery (DAVID) v6.7 (https://david.ncifcrf.gov/), NIAID/NIH. False Discovery Rate (FDR) ≤ 0.05 was used as the criteria for GO category enrichment.
NGS gene expression landscape of CRC
A total of 25,761 genes were detected. Since genes with higher FPKM values may generally confer more biological impacts, we focused on genes with FPKM > 1 . There were 10,255 genes (40% of total genes) with average FPKM > 1 and differential expression between tumors and normal controls (False Discovery Rate (FDR) < 0.05 in ANOVA). A total of 3893 genes (15% of total genes) with average FPKM > 1 show no differential expression between tumor and normal controls with FDR (ANOVA) > 0.05 [10, 11, 12].
TaqMan quantitative PCR (qPCR) quantification
cDNAs from T16 to T35 pairs (20 tumor pairs) were synthesized from total RNA (0.5 μg) using random primers and High Capacity cDNA Reverse Transcription Kit (ABI Part#4368813). qPCR was performed using an Applied Biosystems 7300 Sequence Detection system. The 10 μl PCR reaction included 0.67 μl cDNA, 1 μl 1× TaqMan Universal PCR master mix, 1 μl primers, and probe mix of the TaqMan Assay protocol (PE Applied Biosystems). The reactions were incubated in a 96-well optical plate at 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60° for 10 min. The threshold cycle (Ct) is defined as the fractional cycle number at which the fluorescence passes the fixed threshold. The Ct data were determined using default threshold settings. The average Ct values were 32 for DKC1, 25 for RRP1B, 32 for BOP1, 29 for C1orf43, 27 for RAB7A, 32 for HEBP4 and 26 for ACTB. Individual data points represent mean ± SD of “three biological replicates” in at least separate two experiments. The expression levels of BOP1 (block of proliferation 1), DKC1 (dyskerin pseudouridine synthase 1), and RRP1B (ribosomal RNA processing 1B) were examined with genes ACTB (beta-actin), RAB7A (ras-related protein Rab-7a), HEBP2 (heme binding protein 2), and C1orf43 (chromosome 1 open reading frame 43) as reference genes. ABI PCR primers for qPCR are listed below: BOP1: Hs00374884_m1; DKC1: Hs00154737_m1; RRP1B: Hs00380154_m1; ACTB: Hs01060665_g1; RAB7A: Hs01115139_m1; HEBP2: Hs00204872_m1; and C1orf43: Hs00367486_m1. All of these 7 primers were used in qPCR assays from other studies ( [13, 14, 15, 16, 17, 18]). The relative expression levels of target genes in tumor samples over normal controls were estimated using 2-ΔΔCt calculation ( [19, 20]).
Confirmation of NGS data accuracy with tumor landmark genes
Confirmation of dysregulation among classical reference genes in CRCs
Identification of novel colorectal reference genes
The differential expression of 18 classical reference genes in CRC vs. healthy intestinal tissue led us to explore whether there were reference genes, in addition to the three already identified, that may be more consistently expressed among diverse colonic tissues and could potentially be used for normalization purposes. The candidates are genes with high expression and low variance among tumor and normal samples. By the criteria of FDR > 0.05, CV < 30%, and average FPKM > 100, we found 42 potential colorectal reference genes (Additional file 1: Table S5) that are more consistently expressed among tumor and normal samples (ie, CV across combined 79 tumors and 79 normal tissues were between 15 and 29%). In addition, these 42 newly identified reference genes have smaller variance (average STDEV = 35, CV = 23%) than the 21 classical reference genes (average STDEV = 494, CV = 36%) (Additional file 1: Table S4) in the 79 CRC cohort. The functions of these 42 reference genes include cellular cargo transportation (20 genes in Additional file 1: Table S6), cellular structure (13 genes in Additional file 1: Table S7) and activity in various metabolic pathways (9 genes in Additional file 1: Table S8).
Subtyping of CRC by novel colorectal reference genes and their coexpressed genes
Molecular indicators of tumor in ‘histological normal’ tissues detected by the novel reference gene panel
Since the PCA signatures of the 8 novel reference genes (Fig. 3b, c) clustered 1 LSN (N1) and 6 HSNs (N8, N45, N46, N47, N56, N58) along with a subset of HST, it was important to verify whether these so-called normal samples may contain undetected tumor, as revealed by similar gene expression profiles as their respective tumors. We thus evaluated and compared the expression levels of the 8 reference genes as well as the RGCOEX genes in these histologically normal samples and their respective tumor samples. Interestingly, 5 HSNs (N45, N46, N47, N58, N8) manifest a similar downregulation of gene expression in genes marking normal enterocytes, but to an even greater extent, than their respective tumors (T45, T46, T47, T58, T8) while N1, and N56 manifest distinctly different patterns from their associated tumors (T1, T56) (Additional file 1: Figure S5a-f). The reason for the greater extent of gene downregulation in the 5 normal samples compared to their paired tumors is intriguing and needs further evaluation. To further our evaluation of the potential presence of tumor within these 7 HSNs, we additionally examined the expression of genes characteristic of tumor microenvironments including the following: desmoplasia (dense fibrosis around a neoplasm) genes, including those pertaining to the collagen (COL6A1, COL6A2, COL1A2, COL1A1); neutrophil/myeloid-derived suppressor cell infiltration (CXCR1, CXCR2); cell proliferation (MYC, CDK4) and tumor invasion (MMP2, MMP9, MM14), (Additional file 1: Figure S6a-d). Strikingly among 7 HSNs, N8 (adjacent to a stage 4 poorly differentiated tumor with LN metastasis/R0) had the highest expression of MMP2, MMP14, COL6A1, COL6A2, COL1A2, COL1A1, CXCR1 and CXCR2 across all 79 normal samples while N58 (adjacent to a stage 3 well differentiated tumor with LN metastasis/R2) had the highest expression MYC and CDK4 across all 79 normal samples. These data indicate that histologically normal tissues, may contain undetected tumor, profoundly altering patient prognosis and strongly indicating that rapid evaluation of tumor margins by more advanced technologies may improve surgical resection of malignant tissue and confer improved patient survival.
Computing normalization with 6 typical reference genes in NGS data
As most classical reference genes were found to be differentially expressed in CRC versus adjacent normal tissues (Additional file 1: Table S3, Fig. 2a, b), we investigated whether the CV values of reference genes impacted their ability to serve as reference genes for normalization of gene expression. Considering that the genes related to ribosome biogenesis are uniformly and significantly elevated in the CRC cohort (Additional file 1: Figure S1a, 1b), we simulated normalization of ribosome biogenesis related genes, with CRC reference genes of differing CVs. In doing so, we normalized the FPKM values of 15 ribosome biogenesis related genes using the FPKM values of 6 reference genes (C1orf43, RAB7A, HEBP2 (Heme Binding Protein 2), ACTB, TFRC (Transferrin Receptor) and HSP90AB1 (Heat Shock 90kD Protein 1, Beta)) whose CVs range from 16 to 75% (Additional file 1: Table S3, S4) and compared the separation of tumors and normal samples in PCAs (Additional file 1: Figure S7). The analysis revealed that normalization with a “hypothetical” reference gene with 0% CV across all 79 tumor and 79 normal samples (FPKM = 100, CV = 0%) maintained the separation of tumors from normal samples as did normalization with RAB7A (CV = 17%) and C1orf43 (CV = 16%). However, normalization using HSP90AB1 (CV = 75%) completely abrogated the separation of tumors and normal samples including high stage tumors, while normalization using genes of intermediate CV, TFRC (CV = 52%), ACTB (CV = 30) and HEBP2 (CV = 21%), separated tumors from normal samples to a variable extent. Collectively, these observations indicate that reference genes with lowest CVs, such as RAB7A and C1orf43, could serve as better reference genes for gene expression normalization in PCR based assays.
Experimental normalization with 4 typical reference genes in qPCR
To further test the fitness of the above reference genes for use as normalization values in qPCR based assays, we examined the expression of three ribosome biogenesis related genes (BOP, DKC1, and RRP1B) by a TaqMan qPCR assay using selected reference genes ACTB (CV = 34%), HEBP2 (CV = 21%), RAB7A (CV = 17%), and C1orf43 (CV = 16%) for gene expression normalization in 20 CRC pairs (T16 to T35). Compared to the NGS data in which all 20 tumors displayed upregulated BOP, DKC1 and RRP1B, the qPCR assays (Additional file 1: Figure S8a1–3) revealed upregulated BOP1, DKC1, and RRP1B in only 13 of 20 tumor samples, regardless of which individual reference gene or combinations of 4 reference genes were used for normalization. The expression correlation between the NGS experiments and qPCR assays (Additional file 1: Figure S8b1–3) was weak (Pearson’s correlation coefficients (cc) < 0.4), with the average Pearson’s correlation coefficient values for expression normalized by ACTB, HEBP4, RAB7A, and C1orf43 of 0.064, 0.167, 0.357 and 0.327, respectively. The data demonstrate that experimental normalization of actual qPCR data using reference genes with smaller CVs (C1orf43 and RAB7A) is comparable to normalization using those with larger CVs (ACTB and HEBP2) and discordant with NGS findings. Thus, we found inconsistent normalization results derived from the same reference genes in different assays (NGS and qPCR).
Validation of both classical and CRC reference gene sets in The Cancer Genome Atlas (TCGA)
The emergence of NGS enables absolute quantitative analyses of the transcriptome across different biological samples in a highly sensitive and precise manner, with consequent direct analysis and comparison of gene expression [12, 31]. Here we found that by NGS analysis, 21 classical reference genes (including GAPDH, ACTB, RPLP0, PPIA and B2M) pertaining to biological functions and processes including cell cycle, ribosome biogenesis, glycolysis, angiogenesis, apoptosis and inflammation and commonly used for the normalization of gene expression in qPCR studies, had differential expression in CRC tumors vs. normal tissues. We then identified 42 CRC reference gene candidates, distinct from the 21 reference gene panel, that had lower CVs and minimal differential expression in tumors vs histologically normal tissues. These 42 CRC reference genes have been frequently cited in published CRC studies or recommended by NormFinder for colon tissue studies [32, 33, 34, 35, 36]. The differential expression of 21 classical reference genes and minimal differential expression of the novel 42 CRC reference genes were further evaluated and validated in a TCGA cohort. Despite the more homogenous expression of the 42 CRC reference genes between tumor and normal tissue, PCA of 8 of these CRC reference genes identified a distinct subset of HST/HSN which may have distinct biological properties. This 8 reference gene subset was validated for detection of the distinct HST/HSN subset of tumors by PCA of highly coexpressed genes. Because this unique subset of colonic tissue reference genes mainly pertains to intracellular transport, the downregulation of these genes in HST/HSN likely indicates loss of brush border nutrient transport, a major physiological function of normal enterocytes. Furthermore, some additional reference genes may act as tumor suppressors since CLTC (vesicle traffic protein) (Additional file 1: Table S6), ACTR2 (cytosolic transport related protein) (Additional file 1: Table S6) and RAB1B (ras-like shuttle protein) (Additional file 1: Table S6) were also positively (cc > 0.6) coexpressed (same trend) with 11 tumor suppressor genes mainly related to rho guanine nucleotide exchange factor, succinate dehydrogenase complex and TP53 and negatively (cc < − 0.6) correlated with expression of 4 oncogenes as well as 11 ribosome biogenesis genes (Additional file 1: Table S15). Thus, the downregulation of such reference genes could be translated as a shift from normal cellular functions and well-behaved growth inhibition to highly proliferative cells equipped for tumor metastases in this subset of HSTs. Building on our previous study which captured increased cell proliferation, glycolysis, inflammation, collagen catabolism, and decreased lipid metabolism, colonic cellular transportation and detoxification as indispensable hallmarks for CRC , the novel colorectal reference genes mainly related to intracellular/cytosolic transport, here identified as highly dysregulated in a subpopulation of HST/HSN, may have identified tumors with unique biological characteristics with clinical implications. Further study is clearly needed. Moreover, this study has important implications for defining “clear” tumor margins, as despite having histologically normal tumor margins, two HSNs highly likely contained significant tumor content as assessed by the downregulated reference genes and the upregulation of genes relating to cell proliferation, invasion, fibrosis and neutrophil infiltration highly characteristic of a tumor microenvironment. Interestingly, there was concomitant downregulation of 3 reference genes (RAB1B, ACTR2 and CLTC) and 3 tumor suppressor genes (neurofibromatosis type 1 (NF1), DEAD-Box helicase 5 (DDX5) and CAMP responsive element binding protein 1 (CREB1)) in 3 out of 6 HSNs detected by the 8 reference gene PCA (Additional file 1: Figure S10 and Additional file 1: Table S15). These 3 histologically normal tissues were adjacent to either poorly differentiated HST or HST with lymph node metastasis. Moreover, the FPKM patterns of the reference genes distinctly revealed clonal similarities between five “normal” tissues and their poorly differentiated or local lymph node infiltrated tumors. This phenomenon could be caused by comparable genetic or epigenetic changes pertaining to undetected tumor infiltration or modification of the tumor microenvironment. As the tumor margin impacts overall survival [36, 37], a reference gene evaluation of tumor adjacent tissues, rather than sole reliance on a histological determination, may better determine truly negative margins. Since molecular CRC subtyping could have potential in cancer management [38, 39, 40, 41, 42], the clinical outcomes of patients in which tumors expressed profound reference gene dysregulation require further study.
Regarding data normalization, although simulated normalization by reference genes with smaller CVs suggested that such genes may be better reference genes, in actual qPCR assays, the normalized profiles showed very weak correlation with the NGS profiles regardless of the magnitude of CV values of the reference genes. The main reason for this lack of correlation could be intrinsic differences between NGS and qPCR in aspects of sensitivity, specificity and variability [43, 44]. Another likely reason could pertain to differences in the species of mRNA evaluated by the respective assays. NGS gene expression detection is dependent on the poly-A tail “intactness” of mRNA since only pure poly-A mRNA was used as templates for cDNA synthesis, while qPCR gene expression detection is independent of poly-A tail since mRNAs with or without poly-A tail were used for cDNA synthesis by the random sequence primers and sizes of standard amplicons are very short (75–150 bp).
In summary, we demonstrated the differential expression of 21 classical reference genes in CRC samples vs their histologically normal respective tissues and identified 42 novel reference genes with minimal variability between tumor and normal tissues. From these 42 reference genes, we further determined an 8 gene panel which distinguished a subset of HST/HSN with potentially unique biological properties. In comparing NGS with qPCR, we further demonstrated the clinical potential advantage of using NGS to capture the classical hallmarks of CRC, such as upregulated cell proliferation and downregulated cell differentiation, together with hallmarks of a subset of “high risk” CRC, such as downregulated vesicular transport to potentially improve patient outcome.
The authors would like to thank Drs. Ashutosh Rao (FDA) and Joseph Ziegelbauer (NIH) for his critical review and comments on this manuscript.
RW, WW, JP and RFS carried out experiments. RW, HL, LHW, VS, WLA and LX performed data analysis. AR, LX, RW, LP, HL and HJ designed experiments and interpreted results. AR, RW. HL and LX wrote the manuscript, and all authors edited it. AR is the principal investigator of this study. All authors reviewed and approved the final manuscript.
This work was supported by FDA intramural program funds awarded to Amy Rosenberg (FDA/OPQ/OBP/PDUFA/2016, 2017). The funding body play no direct role in the design of the study, and collection, analysis, and interpretation of data, and in writing the manuscript.
Ethics approval and consent to participate
The purchase of these tumor and normal RNA samples was waived by FDA Internal Standard Operating Procedures for the Research Involving Human Subjects Committee. Written informed consent was obtained from all patients involved in the study via the procuring laboratory at Indivumed (Hamburg, Germany).
Consent for publication
The authors declared that they have no competing interests.
- 6.Rebouças EDL, Nascimento Costa JJD, Passos MJ; José Renato de Sousa Passos JRD; Hurk RVD; Viana Silva JR. Real time PCR and importance of housekeepings genes for normalization and quantification of mRNA expression in different tissues. Brazilian Arch Biol Technol. 2013;56(1). On-line version ISSN 1678–4324.Google Scholar
- 9.Puppa G, Sonzogni A, Colombari R, Pelosi G. TNM staging system of colorectal carcinoma: a critical appraisal of challenging issues. Arch Pathol Lab Med. 2010;6:837–52.Google Scholar
- 12.Xu L, Wang R, Ziegelbauer J, Wu WW, Shen RF, Juhl H, Zhang Y, Pelosof L, Rosenberg A. Transcriptome analysis of human colorectal cancer biopsies reveals extensive expression correlations among genes related to cell proliferation, lipid metabolism, immune response and collagen catabolism. Oncotarget. 2017;8(43):74703–19.CrossRefGoogle Scholar
- 17.Frittoli E, Palamidessi A, Marighetti P, Confalonieri S, Bianchi F, Malinverno C, Mazzarol G, Viale G, Martin-Padura I, Garré M, Parazzoli D, Mattei V, Cortellino S, Bertalot G, Di Fiore PP, Scita G. A RAB5/RAB4 recycling circuitry induces a proteolytic invasive program and promotes tumor dissemination. J Cell Biol. 2014;206(2):307–28.CrossRefGoogle Scholar
- 20.Xu L, Elkahloun AG, Candotti F, Grajkowski A, Beaucage SL, Petricoin EF, Calvert V, Juhl H, Mills F, Mason K, Shastri N, Chik J, Xu C, Rosenberg AS. A novel function of RNAs arising from the long terminal repeat of human endogenous retrovirus 9 in cell cycle arrest. Virol. 2013;87(1):25–36.CrossRefGoogle Scholar
- 23.Ikeda K, Monden T, Tsujie M, Izawa H, Yamamoto H, Ohnishi T, Ohue M, Sekimoto M, Tomita N, Monden M. Cyclin D, CDK4 and p16 expression in colorectal cancer. Nihon Rinsho. 1996;54(4):1054–9.Google Scholar
- 25.Dalerba P, Kalisky T, Sahoo D, Rajendran PS, Rothenberg ME, Leyrat AA, Sim S, Okamoto J, Johnston DM, Qian D, Zabala M, Bueno J, Neff NF, Wang J, Shelton AA, Visser B, Hisamori S, Shimono Y, van de Wetering M, Clevers H, Clarke MF, Quake SR. Single-cell dissection of transcriptional heterogeneity in human colon tumors. Nat Biotechnol. 2011;29(12):1120–7.CrossRefGoogle Scholar
- 38.Wu Y, Wang X, Wu F, Huang R, Xue F, Liang G, Tao M, Cai P, Huang Y. Transcriptome profiling of the cancer, adjacent non-tumor and distant normal tissues from a colorectal cancer patient by deep sequencing. PLoS One. 2012;7(8):e4100.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.