Introduction
Colorectal cancer (CRC) is one of the major causes of death worldwide. The carcinogenic processes of CRC involve complex mechanisms. Autophagy is a physiologic mechanism that utilizes self-digestion of cell organelles to promote cellular homeostasis in the setting of dysfunctional cellular components, cellular stress or energy-deprived states. An extensive network of proteins coordinates the process of autophagy.1
Autophagy can play an instrumental role in the pathogenesis of CRC.1 Autophagy reduces carcinogenesis and acts as a tumor suppressor by removing damaged organelles and reducing reactive oxygen species and DNA damage, and by promoting autophagic cell death. However, in established tumors, autophagy promotes cancer growth by providing tumor cells access to nutrients for growth and metabolism as well as increasing drug resistance and ensuring the maintenance of the cancer stem cell.2 Autophagy-related genes (ATGs) play an integral role in the complex cellular pathway of autophagy. Thirty-six such genes have been identified to date.3 Early clinical trials have shown the potential benefit of inhibiting parts of the autophagy pathway in many other cancers, including pancreatic cancer, multiple myeloma, and melanoma.3 Novel clinical biomarkers are being discovered via ongoing clinical studies to monitor autophagy in patients.4
However, little is known about the expression of autophagy genes in human colon cancer.5,6 In this study, we compared the expression levels of ATG between normal and colon cancer tissues using a predefined autophagy pathway array. We hypothesized that the genes involved in autophagy were differentially expressed in colon tumor tissue versus normal control tissues.
Methods
Study participants
The study population and research approaches were described previously.7 Our study was conducted at the affiliated hospitals of Baylor College of Medicine (BCM) and involved six patients who were diagnosed with colon cancer and subsequently underwent total curative surgical excision during the years of 2007–2013. None received pre-operative neoadjuvant chemotherapy. The resected tumor tissues and normal adjacent mucosa samples were snap-frozen in liquid nitrogen and stored in −80 °C. Additionally, we acquired colon mucosa samples from five unrelated patients who underwent a colonoscopy and were shown to have normal colons endoscopically at the Michael E. DeBakey Veterans Affairs Medical Center (MEDVAMC). These samples served as the control samples when the RNA quality of normal adjacent samples was not adequate for gene expression analysis. The tumor and its unrelated control tissue were matched in relation to colonic segment, race/ethnicity, age (±5 years), and sex. All patients provided informed consent to participate in a research and use their samples. The Institutional Review Boards of MEDVAMC and BCM approved the study protocol.
RNA extraction and gene expression analysis
In the genomic core lab at the Texas Medical Center Digestive Disease Center, using the NucleoSpin RNA isolation kit (MACHEREY-NAGEL Inc, Bethlehem, PA, USA), we obtained total RNA from 17 fresh frozen tissue biopsy samples (six biopsies from excised cancerous tissues, six biopsies from adjacent normal colon mucosa, and five from unrelated normal control mucosa). The RNA purity of samples was confirmed with a nucleic acid 260:280 ratio, which was above 2.0 for all. Nonetheless, for one tumor sample and five normal adjacent mucosa samples, the RNA integrity number (RIN) was <6.0. These samples were ultimately excluded from the experiment as they were deemed not appropriate for gene expression analysis. Thus, our analysis was restricted to the samples from five mucosal specimens with colon cancer and five biopsies from matched unrelated healthy controls.7
We utilized the Qiagen RT2 Profiler PCR Array (Valencia, CA, USA) to determine the expression of 84 key genes in the autophagy pathway (primer sequence is available upon request from Qiagen). The array includes genes encoding proteins for autophagic vacuole formation (AMBRA1 (NYW1), ATG12, ATG16L1, ATG4A, ATG4B, ATG4C, ATG4D, ATG5, ATG9A, ATG9B, BECN1, GABARAP, GABARAPL1, GABARAPL2, IRGM, MAP1LC3A, MAP1LC3B, RGS19, ULK1, WIPI1), vacuole targeting proteins (ATG4A, ATG4B, ATG4C, ATG4D, GABARAP), transport proteins (ATG10, ATG16L1, ATG16L2, ATG3, ATG4A, ATG4B, ATG4C, ATG4D, ATG7, ATG9A, GABARAP, GABARAPL2, RAB24), autophagosome-lysosome linkage proteins (DRAM1, GABARAP, LAMP1, NPC1), ubiquitination proteins (ATG3, ATG7, HDAC6), and proteases (ATG4A, ATG4B, ATG4C, ATG4D). At the core facility at BCM, the assays were conducted using the Bio-Rad Lightcycler RealTime PCR system (Bio-Rad, Hercules, CA, USA).
Analysis of mRNA microarray data from TCGA
We also downloaded mRNA expression data of 244 TCGA (The Cancer Genome Atlas) colon adenocarcinoma and rectum adenocarcinoma (COAD/READ) patients on June 14, 2018 (www.firebrowse.org ). This dataset included 22 normal samples and 222 CRC tissues. Using these data, we examined mRNA expression levels of ATG4A, ATG4B, ATG4C, ATG4D, GAA, GABARAP, CTSD, and CTSS genes in the CRC tissues versus the control tissues. The sequencing platform of the mRNA expression data was UNC_IlluminaHiseq RNASeqV2.
Statistical analysis
The RT2 Profiler PCR Array data analysis version 3.5 was utilized for data analysis. The gene expression levels were normalized to ACTB, GAPDH, and RPLP0. The Student’s t-test was used to compare the expression levels between diverse groups (overall cancer versus control; early- or late-stage cancer versus control) utilizing the normalized gene expression data. Fold-change (FC)/fold-regulation (FR) was calculated using the ΔΔCt method (wherein ΔCt was calculated between a gene of interest and an average of reference housekeeping genes, after that the ΔΔCt calculation was performed: ΔCt (Test Group)—ΔCt (Control Group)). Subsequently, we calculated FC using the 2∧ΔΔCt formula.
To visualize gene expression changes, we used a scatter plot to compare the normalized expression of every gene on the array between the tumor and normal tissues. On the plot, we used the middle line to denote unchanged gene expression. The selected FR threshold was denoted by the dotted lines above or below the middle line. For multiple testing for 84 genes, we adjusted it using the false discovery rate (FDR). An FDR q value <0.10 denoted statistical significance.7 For the downloaded TCGA data, we performed the Student’s t-test comparing CRC tissue with normal tissue.
Results
The five cancer patients with high-quality RNA included in this analysis were all non-Hispanic Caucasian males between 55 and 75 years old, with a mean age of 62 years. Three patients had early-stage tumors, and three patients had late-stage tumors. The characteristics of these study participants have been described.7
We observed significant down-regulation of 29 genes in the autophagy pathway (p < 0.05) in colon cancer tissues compared to the normal colonic mucosa (Fig. 1, Supplemental Table S1). The down-regulations were observed for both early- and late-stage tumors. Furthermore, after adjusting for multiple testing, the difference of 17 genes remained statistically significant. The FR was greater than 2 for all 17 genes, including ATG4A, ATG4C, ATG4D, CTSD, CTSS, ESR1, GAA, and GABARAP (FDR q < 0.10; Table 1). ATG9B was the only ATG gene that was up-regulated in cancer tissue compared with control tissue (FC=8.39, p=0.06). However, the up-regulation was statistically nonsignificant after multiple testing adjustment. The differential expression was consistently observed in early- or late-stage tumors (Table 1).
Table 1Differential expression of selected autophagy genes in human colon cancer tissues and control tissues
| Fold-regulation, All tumor | p | q | Fold-regulation, Early tumor | p | Fold-regulation, Late tumor | p |
---|
PIK3CG | −10.58 | 0.01 | 0.08 | −7.97 | 0.09 | −14.06 | 0.07 |
IGF1 | −5.84 | 0.0009 | 0.08 | −2.39 | 0.09 | −14.27 | 0.04 |
TNF | −4.59 | 0.01 | 0.08 | −4.1 | 0.07 | −5.14 | 0.05 |
ESR1 | −4.05 | 0.01 | 0.08 | −3.18 | 0.08 | −5.15 | 0.05 |
BCL2 | −3.89 | 0.01 | 0.08 | −3.21 | 0.12 | −4.7 | 0.03 |
RGS19 | −3.19 | 0.007 | 0.08 | −2.65 | 0.07 | −3.83 | 0.04 |
GAA | −3.11 | 0.01 | 0.08 | −4.39 | 0.02 | −2.21 | 0.11 |
TNFSF10 | −3.05 | 0.01 | 0.08 | −2.12 | 0.12 | −4.39 | 0.04 |
CTSD | −2.63 | 0.00085 | 0.07 | −2.08 | 0.02 | −3.31 | 0.01 |
PTEN | −2.43 | 0.01 | 0.08 | −2.67 | 0.06 | −2.2 | 0.07 |
ATG4A | −2.25 | 0.02 | 0.09 | −2.31 | 0.09 | −2.2 | 0.08 |
ATG4C | −2.31 | 0.01 | 0.10 | −2.40 | 0.07 | −2.22 | 0.06 |
ATG4D | −2.22 | 0.01 | 0.08 | −2.14 | 0.1 | −2.3 | 0.05 |
CTSS | −2.16 | 0.01 | 0.08 | −1.78 | 0.11 | −2.64 | 0.05 |
GABARAP | −2.12 | 0.0003 | 0.07 | −2.09 | 0.03 | −2.15 | 0.04 |
CASP3 | −2.05 | 0.02 | 0.08 | −1.58 | 0.17 | −2.65 | 0.04 |
TGFB1 | −2.01 | 0.01 | 0.08 | −1.92 | 0.08 | −2.1 | 0.03 |
ATG9B | 8.39 | 0.06 | 0.16 | 18 | 0.01 | 3.91 | 0.22 |
By comparing CRC tissue with normal tissue using the TCGA data, we also found significant down-regulation of ATG4A, ATG4C, ATG4D, GAA, GABARAP, CTSD, and CTSS (Fig. 2).
Discussion
This pilot study suggested differential expression of a number of genes in the autophagy pathway in both early-and late-stage colon cancer compared with normal colonic tissues. Down-regulation of ATG in colon cancer may comprise the host’s response in preventing the accumulation of genetic defects that accompany malignant transformations and therefore promote tumor progression. The analysis based on the TCGA data validated our major findings.
We found that multiple ATGs were down-regulated in colon cancer compared to normal mucosa after accounting for multiple testing, including ATG4A, ATG4C, and ATG4D. ATG4 genes play multiple roles in autophagy, including autophagic vacuole formation and targeting, and proteases. In yeast, the Atg4-family cysteine proteases play a crucial role in preparing Atg8 for conjugation to lipid membranes and for the deconjugation of Atg8 from the autophagosomes.8 There are eight human Atg8 homologues that belong to two subfamilies: LC3 and GABARAP. Atg4a is known to be a potent protease for the GABARAP family but not the LC3 family.9 In our study, the GABARAP gene was also significantly down-regulated. Contrary to our results, a recent study on gastric cancer reported the higher expression of ATG4A in cancer tissues and metastatic lymph nodes than normal tissues. That study also showed that ATG4A up-regulated the Notch signaling pathway-induced metastasis in gastric cancer cells.10
Of all the ATG4 proteases, ATG4B has been shown to be the most functionally dominant in autophagy regulation in mammals.11 Liu et al.12 showed that knock-down of the ATG4B in CRC cell lines induced autophagic flux and reduced cyclin D1 expression to inhibit G1/S phase transition of the cell cycle. However, in our research, the expression of ATG4B was statistically nonsignificantly down-regulated in tumors, with only a small FC. ATG4C gene knock-out in mice has been shown to aid in forming methylcholanthrene-induced fibrosarcomas, suggesting a regulatory impact on tumor cells.13 We found the ATG4C gene was down-regulated significantly, as was the ATG4D gene. In line with our finding, it has been shown that the expression of the ATG4D gene was significantly lower in breast cancer tissues than that in the adjacent normal tissues.14 The identification of diminished ATG4 gene expression in colon cancer suggested that the formation of autophagosomes may be compromised in colon cancer. In our study, ATG9B was the only gene to be highly up-regulated with an FC of 16. However, the difference between colon cancer and normal mucosa was not statistically significant.
Cathepsin is an aspartic endoprotease that is distributed in all lysosomes in the human body.15 Its primary function is to disassemble proteins and activate cascades in prelysosomal compartments.16 Cathepsin D (CTSD) and cathepsin S (CTSS) genes were significantly down-regulated in colon cancer compared to normal mucosa after accounting for multiple testing. Contrary results were reported by Sebzda et al.,17 which showed increased expression of CTSD in colon cancer compared to adjacent normal colon mucosa utilizing immunohistochemical (IHC) staining in 68 patients from Poland. Another study also utilizing IHC testing showed that cathepsins B, L, and D activities were higher in a series of malignant tissues compared with adjacent normal colorectal tissues.18 In our study, CTSS was also significantly down-regulated, even after accounting for multiple testing. Our data were incongruous to the prior studies possibly in part because we used mRNA levels to measure gene expression, whereas the prior studies used IHC or enzyme-linked immunosorbent assay. In addition, the function of autophagy genes in cancer development may be tissue-specific.
The research on autophagy and CRC was triggered by a genome-wide association study that identified an association between the single nucleotide polymorphism (SNP) rs2241880 of ATG16L1 and risk of inflammatory bowel disease.19 Hampe et al.19 showed a statistically significantly positive association between this SNP and Crohn’s disease in a German population. This SNP was also associated with an improved overall survival in CRC.20 However, we did not find the expression of the ATG16L1 to be associated with colon cancer in our sample.
BECN1 (Bectin 1), LC3B (MAP1LC3B), and p62 (SQSTM1) are three genes that have been widely reported in CRC research. They are expressed in normal colon mucosa as well.21 We found LC3B and BECN1 nonsignificantly down-regulated and p62 up-regulated in cancer tissues. The loss of expression of BECN1 has been reported for ovarian and esophageal cancers.22,23 Contrarily, prior studies showed mRNA and protein expression of BECN1 or LC3 were higher in CRC tissue compared with normal adjacent tissue.24–26 p62 is one of five autophagy cargo adaptor molecules that can interact with processed LC3 at the autophagosome. p62 accumulation has been observed in liver cancer progression.27 An IHC study also showed that higher p62 expression was an independent adverse prognostic marker in patients with CRC.28 p62 and LC3 also contribute to selective autophagy.29,30 For example, mitophagy is a selective mode of autophagy in which mitochondria are targeted for degradation.31 Mitophagy affects cell stemness, cell fate determination, inflammation, and DNA damage, and has also been implicated in cancer.32 It is likely that these genes can influence colon tumorigenesis through regulating mitophagy.33
BECN1 and PTEN are important in inducing autophagy in cancer. PTEN is a tumor suppressor gene in colon cancer and was down-regulated in our research.34 The down-regulation of autophagy inducers was in line with the overall down-regulation of autophagy pathway genes in colon cancer. A cross-talk exists between apoptosis and autophagy. BECN1 is also a direct substrate of caspase 3. Several genes related to apoptosis, including CAPS3 and BCL2, were also down-regulated in colon cancer in our study.
This was a preliminary study with multiple limitations. First, it was based on a small sample size and, thus, false negative and false positive findings could have occurred. The stratified analysis by tumor stage was exploratory given a limited sample size. Furthermore, transcription factor EB (TFEB) is probably one of the most important transcription regulators for autophagy and lysosome-related genes. TFEB could modulate the expression of a broad range of lysosomal genes. including CTSD and CSTS.35 TFEB over-expression has been seen in a variety of cancers.36–38 However, this gene was not included in the predefined panel in the present study. Lastly, one study shows that clinical implications of autophagy biomarkers, such as LC3 and p62, in CRC prognosis may depend on K-ras mutation status.39 As this was a preliminary study, we did not ascertain K-ras mutation status or evaluate protein expression.
In summary, the findings from our study support the tumor suppressor role of autophagy in colon cancer. In particular, the ATG4 proteases may play an important role in colon cancer development. The precise mechanisms inhibiting or leading to autophagy, and the downstream effectors of autophagy in carcinogenesis need to be fully elucidated in order to identify novel biomarkers for prognosis and target for cancer treatment.
Future research directions
We found that multiple autophagy-related genes were down-regulated in colon cancer tissues. We will validate our findings using larger clinical sample series and by including IHC staining. We will also identify modifiable factors and metabolic stress factors that can contribute to macroautophagy and mitochondria autophagy dysregulation in colon cancer.
Supporting information
Supplementary material for this article is available at https://doi.org/10.14218/ERHM.2018.00007 .
Supplemental Table S1
Differential expression of all autophagy genes in human colon cancer tissue and normal control tissues.
(DOCX)
Abbreviations
- ATG:
autophagy-related genes
- CD:
Crohn’s disease
- CRC:
colorectal cancer
- FC:
fold change
- FR:
fold regulation
- FDR:
false discovery rate
- IHC:
immunohistochemical
- SNP:
single nucleotide polymorphism
Declarations
Acknowledgement
This research was supported by Gillson Longenbaugh Foundation and Golfers Against Cancer Organization (PI: LJ), Cancer Prevention Research Institute of Texas (RP#140767, PI: LJ), and National Cancer Institute P30 Cancer Center (P30 CA125123) for support of the Human Tissue Acquisition and Pathology Shared Resources. This project was also supported in part by Public Health Service (grant P30DK056338), the expert assistance of Lisa D. White and the Specimen Collection Core E of Texas Medical Center Digestive Diseases Center, and Houston VA HSR&D Center for Innovations in Quality, Effectiveness and Safety (CIN13-413). Finally, we are thankful for Dr. Lang Rao’s comments on this manuscript.
Conflict of interest
The authors have no conflict of interests related to this publication.
Authors’ contributions
Study design (LJ), sample acquisition (MI, HBE), data analysis (LC, FC), data interpretation and manuscript writing (SG, TK, DYG, MI, NTE, HBE, LJ), critical intellectual input (SG, TK, LC, MI, NTE, HBE, LJ).