Introduction

The inherent or acquired radioresistance of colorectal tumors leads to failure of treatment and more toxic effects of ionizing radiation [1,2,3,4]. Using appropriate biomarkers of tumor radiosensitivity, we can predict the efficacy of radiotherapy and select a proper treatment strategy accordingly. There are several biomarkers, such as EGFR, p53, Bcl-2, Bax and p21, and several clinical factors that may be used as radiation response markers. However, the clinical application of these markers for the response of CRC to radiotherapy remains controversial and reliable identification of a patient’s response to radiotherapy is still impossible [5,6,7].

Recent studies on the molecular biology of CRC indicate that the origin of radioresistance of CRC is related to tumor microenvironment and dysregulation of specific genes which play a critical role in cell signaling pathways. These genes are entitled oncogenes or tumor suppressors [8]. Furthermore, miRNAs as epigenetic factors play a key role in the pathogenesis of CRC [9, 10]. MicroRNAs (miRNAs) are small (18–24 nucleotide), noncoding RNA molecules that down-regulate their specific target genes through specific binding to the 3′-untranslated region (3′-UTR) of target mRNAs, suppressing mRNA translation or mRNA degradation [11].

Transcription of miRNA genes yields a primary miRNA (pri-miRNA). After processing of pri-miRNA by Drosha, pre-miRNA (about 70 nucleotides) is generated. Further processing of pre-miRNAs in cytoplasm leads to mature miRNA generation [12, 13]. One strand of mature miRNA specifically enters an RNA-induced silencing complex (RISC), then the activated RISC complex targets the target mRNA [14]. MiRNAs can control different biological functions including metabolism, autophagy, differentiation, inflammation, apoptosis and DNA damage response [15]. Based on the results of recent studies, miRNAs play a significant role in pathogenesis, prognosis, and progression of cancers [3, 11]. MiRNAs regulate the response of tumors to radiation through interaction with critical factors in PI3K/AKT, MAPK/ERK, NF-κB, or TGF-β pathways, which are essential radiation-related signal transduction pathways [15].

In summary, due to different responses of CRC patients to preoperative radiotherapy, and lack of a specific biomarker for distinguishing radiosensitive from radioresistant CRC tumors, identification and development of a novel biomarker of radiosensitivity are vital [16, 17]. On the other hand, considering the function of miRNAs in regulating the essential genes that determine tumor cells’ radiosensitivity, we hypothesized that miRNA expression is different between radioresistant and radiosensitive CRC cell lines. Deregulation of miR-222 [18] (Gene ID: 407007) and miR-155 [19] (Gene ID: 406947) in several cancers and their dysregulation in response to radiation have been previously reported [20,21,22,23,24,25]. Therefore, the aim of this study was to investigate the expression changes of miR-222 and miR-155 and their candidate target genes in acquired radioresistant and parental radiosensitive colorectal cancer cell lines.

Materials and methods

Cell line and cell culture

The colorectal cancer cell line HCT 116 was purchased from the Pasteur Institute of Iran. The cell was cultured in high glucose DMEM medium (Gibco, USA), supplemented with 10% heat-inactivated fetal bovine serum (Gibco, USA) and 1% Pen-Strep (Gibco, USA) and incubated at 37 °C in a 5% CO2 atmosphere with high relative humidity.

Establishing radioresistant cell lines

Radioresistant cell lines were established based on the method recommended by Su et al. [26]. The cells were fractionally irradiated by 6-MeV X-ray radiation with a high-energy linear accelerator (Shimva, China) at a dose rate of 200 MU per min (1 MU equals 1 cGy of absorbed dose in water under specific calibration conditions for the medical linear accelerator). The cells were first grown to 50% confluence in 25-cm2 culture flasks. After irradiation with 1 Gy, the medium was changed immediately, and until 90% confluency, the cells were incubated at 37 °C. At 90% confluence, the adequate cells were sub-cultured into new flasks. When the new flask reached 50% confluence again, the cells were re-irradiated by 1 Gy (second fraction). These procedures were repeated 9 times (1 Gy 3 times, 2 Gy 3 times and 4 Gy 3 times) to a total dose of 21 Gy. When a radioresistant cell line with a total dose of 21 Gy was established, it was named a RR1 cell line. The RR1 sub-line was irradiated 4 more times with 6 Gy until a RR2 cell line with a total dose of 45 Gy was established. The parental cells underwent the same procedure under the same culture conditions, only without irradiation. Previous to all assays, radioresistant cells must be cultured without any intervention for at least 3 weeks after the last irradiation.

MicroRNA target genes prediction

The analysis of miR-222- and miR-155-predicted targets was performed using the following four algorithms: TargetScan (http://targetscan.org), miRanda (http://www.microrna.org/microrna/home.do), microT_CDS (http://diana.imis.athena-innovation.gr) and RNAhybrid (https://bibiserv2.cebitec.uni-bielefeld.de/rnahybrid).

In this study, the target genes were selected by considering their high prediction score and their confirmed role in cell cycle arrest or DNA damage repair and apoptosis signaling pathway.

Clonogenic assay (colony formation assay)

A clonogenic assay was applied to determine the radiosensitivity of each cell line. A predetermined number of viable cells, based on cell counting by trypan blue (900 cells for 0, 2 and 4 Gy, 1350 cells for 6 and 8 Gy), were seeded in 6-well culture plates and the plates were incubated at 37 °C for 24 h. The cells were irradiated with different doses of X-ray radiation (0, 2, 4, 6 and 8 Gy) and then incubated until the proper size of colonies was seen (10 days). The wells of the plate were then washed with PBS and stained with 0.5% crystal violet in 50% methanol. After washing and drying of plates, the colonies containing ≥ 50 cells were counted. Using the following formulas, the survival fractions (SF) were calculated, and the plating efficiency (PE) was calculated for cells which were not irradiated [27]. All tests were replicated 3 times.

$${\text{PE}} = \frac{{{\text{Number}}\;{\text{of}}\;{\text{colonies}}\;{\text{counted}}}}{{{\text{Number}}\;{\text{of}}\;{\text{cells}}\;{\text{seeded}}}} \times 100\;\;{\text{SF}} = \frac{{{\text{PE}}\;{\text{of}}\;{\text{irritated}}\;{\text{cells}}}}{{{\text{PE}}\;{\text{of}}\;{\text{control}}\;{\text{cells}}}} \times 100$$

After estimating the survival fraction at different radiation doses, the survival curve (log of survival fraction versus radiation dose) was plotted and the D 0 value for each cell line was calculated using the following equation:

$${\text{SF}} = 1 - \left( {1 - {\text{e}}^{{\frac{D}{{D_{0} }}}} } \right)^{n}$$

MTT assay

About 2 × 104 cells were seeded in 96-well plates at 100 μl/well. After 24 h of incubation at 37 °C, the cells were treated with a range of 6-MeV IR doses (0, 2, 4, 6 and 8 Gy). For a period of 144 to 192 h after irradiation, a linear relationship was observed between OD and number of live cells. An MTT assay was done following 168 h of irradiation. Briefly, 10 μl MTT solution [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide solution, 5 mg/ml MTT] was added to each well and the plates were incubated at 37 °C for 4 h. Following the incubation, the solution from each well was removed and 100 μl of DMSO was added to each well to dissolve the formazan crystals. Plates were shaken for 15 min on a plate shaker to ensure adequate formazan solubility. The absorbance of the plate was read at 570 nm wavelength using an automatic microplate reader. The experiments were repeated 3 times. Finally, survival fraction was calculated by absorbance readings at 168 h after radiation using the following formula [27]:

$${\text{SF}} = \frac{{{\text{Mean}}\;{\text{OD}}\;{\text{in}}\;{\text{test}}\;{\text{wells}} - {\text{Mean}}\;{\text{OD}}\;{\text{in}}\;{\text{cell}}\;{\text{free}}\;{\text{wells}}}}{{{\text{Mean}}\;{\text{OD}}\;{\text{in}}\;{\text{control}}\;{\text{wells}} - {\text{Mean}}\;{\text{OD}}\;{\text{in}}\;{\text{cell}}\;{\text{free}}\;{\text{wells}}}} \times 100$$

Sub-G1 fraction analysis

The fixation and staining of cells were done based on Pozarowski and Darzynkiewicz’s method (with some modifications) [28]. In brief, around 3 × 105 of each cell line were seeded in 6-well culture plates and treated with 4 and 6 Gy X-ray radiation following 24 h of incubation at 37 °C. These cells were harvested after 48 and 72 h of radiation. They were then washed with PBS and fixed with 70% ethanol and stored at +4 °C for at least 2 h. After washing with PBS, the cells were resuspended in a solution containing 50 µg/ml of propidium iodide and 100 µg/ml of RNase A, followed by a 20-min incubation at 37 °C. The fluorescence in FL2 was collected by a Partec Flow cytometer.

RNA extraction

Total RNA was prepared using RNX™-plus reagent (Cinnagen, Iran) according to the manufacturer’s instructions. The quality and integrity of the extracted RNA were verified by 1% gel electrophoresis. The concentration and purity of RNA were verified by optical density measurements (260/280 nm ratios).

Primer designing

Human gene-specific primers for PTEN, FOXO3a, and GAPDH (internal control) were designed using Allele ID software (version 6). The specificity of designed primer pairs was checked in NCBI primer blast. The resulting primers were as follows:

PTEN, forward: 5′AGTCCAGAGCCATT-T-CCATC3′;

PTEN, reverse: 5′GATAAATATAGGTCAAGTCTAAG-TCG3′;

FOXO3a, forward: 5′TGAGTGAGAGGCAATAGCATAC3′;

FOXO3a, reverse: 5′AGCACCTATACAGCACCAT-A-A-C3′;

GAPDH, forward: 5′AAGGCTGTGGGCAAGGTCATC3′;

GAPDH, reverse: 5′ GCGT-C-A-A-A-GGTGG-AGGAGTGG3′. Hsa-miR-222-3p, hsa-miR-155-5p and U6 snRNA (internal control) primer sets were purchased from Exiqon.

cDNA synthesis

Complementary DNA (cDNA) for mRNA expression analysis was synthesized using a First Strand cDNA Synthesis Kit (Fermentas, K1631) in a total 20-μl reaction mixture, according to the manufacturer’s recommendations. cDNA synthesis for miRNA expression analysis was performed using miRCURY LNA™ Universal RT cDNA Synthesis Kit (Exiqon, Cat. No 203301) according to the instructions provided with the kit.

Real-time PCR

All real-time PCR assays were carried out by the Roche LightCycler® 96 system. Firstly, real-time PCR efficiency for all reactions was calculated from the standard curves (log of the DNA concentration used vs the CT) using the following formula: \(E = 10^{{\left( {\frac{ - 1}{\text{slope}}} \right)}} - 1\). The efficiency of all reactions was close to 100%. The mRNA expression levels were evaluated using gene-specific primer pairs mixed with a SYBR® Premix ExTaq™ II Kit (Takara, Cat. No RR820B) in a final volume of 20 µl according to the manufacturer’s recommended protocol. The specificity of PCR products was confirmed by 1% gel electrophoresis, melting peak and dissociation curve analysis. To confirm the expected molecular weight (size of amplification product), sequencing was performed (Bioneer, Korea). The miRNA expression levels were measured using a microRNA LNA™ PCR primer set and ExiLENT SYBR® Green master mix, 2.5 ml (Exiqon, Cat. No 203403) in a final volume of 10 µl according to the manufacturer’s instructions. Specificity of miRNA real-time PCR product was confirmed by melting peak and dissociation curve analysis. Each experiment was performed in triplicate. The comparative Ct (cycle threshold) method was used to present real-time PCR data. Fold change of gene was calculated using the Eq. 2−ΔΔCt [29]. All statistical tests were performed based on ∆CT [Ct (gene, sample)–Ct (gene, control)]. Also, graphs show ∆CT on the ordinate, which has a reverse relationship with gene expression level.

Statistical analysis

All values are expressed in mean ± standard deviation. Comparisons between groups were analyzed with one-way ANOVA, Tukey post hoc, and independent samples t-test. A Pearson test was done to investigate the correlation between quantitative parameters. P < 0.05 was considered statistically significant.

Results

Morphological change of HCT 116 cell line after radiation

As seen in Fig. 1, the parental HCT 116 cell line has regular fusiform cell morphology, while the established RR1 cell line shows irregular cell morphology with sharp protrusion and increased intracellular particles. The RR2 cell line has oval/round morphology, granular appearance with increased cell volume, and appendixes are lost.

Fig. 1
figure 1

Morphological comparison of RR2 (c), RR1 (b) and parental cell line (a). Parental cell line has a spindle form, RR1 cell line has cytoplasmic redundancy and RR2 cell line has a globular form

Full size image

Evaluation of radiation sensitivity of colorectal cancer cell lines by colony formation assay

Radiosensitivities of the parental, RR1, and RR2 cell lines were investigated using a clonogenic assay. The D 0 (the dose that reduces the surviving fraction to 37%) of all cell lines are shown in Table 1. The D 0 was higher in the RR2 cell line than in the RR1 cell line. Furthermore, D 0 was higher in the RR1 cell line than in the parental cell line. Survival curves are shown in Fig. 2. SF2 (survival fraction at 2 Gy) of RR1, RR2, and parental cell lines were 61.75 ± 9.30, 58.16 ± 0.11, and 53.34 ± 1.38, respectively.

Table 1 D 0 dose value in RR2, RR1 and parental cell lines. RR2 cell line has higher D 0 value than RR1 and parental cell line
Full size table
Fig. 2
figure 2

Survival curve of RR2, RR1, and parental cell lines. Survival fraction of RR2 cell line is higher than RR1 and parental cell line at all doses

Full size image

Survival fraction by MTT assay

The survival fractions of all cell lines at 2, 4, 6 and 8 Gy doses, were also calculated by MTT assay (168 h after radiation treatment). MTT survival fractions of all cell lines are compared in Fig. 3. It should be noted that the numerical value of MTT survival fraction was higher than the clonogenic survival fraction in all cell lines and doses. However, the correlation between clonogenic and MTT assay was significant and linear (Pearson correlation = 0.991, P value = 0.00).

Fig. 3
figure 3

MTT survival fraction versus radiation dose for all cell lines (168 h after radiation treatment). One-way ANOVA test showed a significant difference in survival fraction of three cell lines in 2-, 4- and 6-Gy doses. Tukey test results showed that MTT survival fraction of RR2 cell line at 2, 4 and 6 Gy was significantly larger than parental cell line. Survival fraction of RR1 cell line was significantly higher than parental cell line at 4 and 6 Gy

Full size image

Sub-G1 fraction after radiation

In order to validate the radioresistance of the RR2 cell line, the sub-G1 fraction of parental and RR2 cell lines were evaluated and compared using the Partec Flow Max software. As seen in Fig. 4, the sub-G1 fraction of the parental cell line at 48 and 72 h following 4 Gy radiation and at 72 h after 6 Gy radiation was significantly higher than that of the RR2 cell line.

Fig. 4
figure 4

Sub-G1 fraction of the parental and RR2 sub-line, following 4 Gy (a) and 6 Gy (b) radiation at indicated time point. *P < 0.05 by independent sample t-test

Full size image

MiRNA target prediction results

Candidate genes were selected according to the following criteria: first, having a high prediction score at least in 3 of 4 databases; second, playing a role in cell radiation response (DNA damage response); and third, dysregulation in irradiated colorectal cancer cell lines. Four different prediction tools with different prediction algorithms were used for miRNA target evaluation. The miTG score. DIANA-microT-CDS is a weighted summation of the scores of all miRNA-recognition elements (MREs); the greater the MITG score is, the more probability of targeting is expected [30]. The TargetScan Total Context score predicts the relative repression of the mRNAs based on the targeting features such as site position, site number, site type, 3′-pairing contribution, and local AU content; a more negative score is interpreted as a better suppression [31]. The miranda-mirSVR score, which is a machine-learning method, evaluates the miRNA effect on the mRNA expression level; a more negative score means more suppression effect exists [32]. The RNAhybrid prediction tool was used for evaluating the minimum free energy (MFE) of miRNA and mRNA hybridization; a more negative MFE is a result of a more stable duplex [33].

Based on the above criteria, PTEN (Gene ID: 5728) and FOXO3a (Gene ID: 2309), which are important members of the phosphatidylinositol (PI)3-kinase pathway [18, 34], were selected as candidate target genes of miR-222 and miR-155 (respectively) for real-time PCR confirmation. Prediction scores for PTEN and FOXO3a as candidate target genes of miR-222 and miR-155 are shown in Table 2.

Table 2 Different prediction score for miRNA target genes in a different database
Full size table

Real-time PCR analysis

There was no difference between mean CT values of GAPDH and U6 snRNA in the three cell lines (GAPDH P value = 0.210, U6 P value = 0.082). Therefore, they were suitable as reference genes to normalize gene expression between these cell lines.

A single peak was observed on the melting curve analysis, confirming the specificity of primers. Also, a single band of real-time PCR products on agarose gel was matched with PTEN and FOXO3a primer products length in NCBI primer blast. The sequencing data of real-time PCR products of FOXO3a and PTEN primers showed sequence similarity up to 99% with their mRNAs in NCBI/blast N.

MiR-222 was upregulated in RR2 and RR1 cell lines in comparison with the parental cell line, with an average increase of 2.03- and 2.48-fold, respectively. The expression level of miR-222 in RR2 (4.35 ± 0.26) and RR1 (4.06 ± 0.08) cell lines showed a statistically significant increase compared to parental cells (5.37 ± 0.24). However, the expression level of miR-222 in RR1 and RR2 cell lines did not show a significant difference. The expression level of PTEN mRNA, a candidate target of miR-222, significantly decreased in RR2 (7.93 ± 0.08) and RR1 (7.17 ± 0.06) cell lines in comparison to the parental cell line (6.24 ± 0.34) with a fold change of −3.22 and −1.90, respectively. Also, PTEN expression in the RR2 cell line was significantly lower than the RR1 cell line (−1.69 fold) (Figs. 5, 6). ΔCt values of miR-222 and PTEN mRNA were compared with each other using the Pearson test. The results showed a significant negative correlation between miR-222 and PTEN expression (Pearson correlation = −0.866, P value = 0.026).

Fig. 5
figure 5

MiR-222 and miR-155 expression levels. MiR-222 and miR-155 have a significantly higher expression in RR2 cell line in comparison to parental cell line. MiR-222 expression did not significantly change in RR2 cell line compared with RR1, while miR-155 was significantly overexpressed in RR2 cell line compared with RR1 cell line. The graph shows ∆CT which has a reverse relationship with gene expression level. P < 0.05 was considered to be a significant level by one-way ANOVA and Tukey post hoc tests

Full size image
Fig. 6
figure 6

PTEN and FOXO3a expression levels. PTEN and FOXO3a have a significantly lower expression in RR2 cell line in comparison to parental cell line. PTEN expression was significantly decreased in RR2 cell line compared with RR1 cell line. The expression level of FOXO3a was decreased in RR2 compared with RR1 cell line but this change wasn’t statistically significant. The graph shows ∆CT which has a reverse relationship with gene expression level. P < 0.05 was considered to be a significant level by one-way ANOVA and Tukey post hoc tests

Full size image

The miR-155 expression level was significantly increased in the RR2 and RR1 cell lines compared with the parental cell line, with a change of 2.58- and 1.64-fold, respectively. Also, miR-155 expression difference between RR1 and RR2 was significant with RR2 1.57 times greater than RR1. The expression level of FOXO3a mRNA, a candidate target of miR-155, was upregulated in RR2 and RR1 cell lines compared with the parental cell line with an average change of −4.59- and −4.16-fold, respectively. The expression levels of FOXO3a in RR1 and RR2 cell lines were not significantly different (Figs. 5, 6). There was a significant negative correlation between expression of miR-155 and FOXO3a gene (Pearson correlation = −0.889, P value = 0.025).

Discussion

In this study, we established radioresistant HCT 116 cell lines with repeated X-ray radiation. After comparing the radiosensitivities of the established cell lines and the parental cell line, a real-time PCR analysis was performed to examine the miR-222/PTEN and miR-155/FOXO3a expression.

The survival fraction at 2 Gy (SF2) and D 0 of the two cell lines (RR1 and RR2) after repeated irradiation were significantly higher than the comparable values for the parental cell line. The results of the flow cytometry indicated that the sub-G1 fraction of the parental HCT 116 cell line after radiation was significantly higher than the RR2 cell line. Our results are in agreement with findings by Anastasov et al. [35] that showed apoptotic activity was significantly lower in radioresistant cell lines than in radiosensitive cell lines. Yang et al. [36] and Huang et al. [37] showed that D 0 and SF2 of HCT 116 as a radiosensitive cell line were about 0.8 Gy and 25%, respectively [36, 37]. On the other hand, D 0 and SF2 of SW-480 as an intrinsic radioresistant cell line were about 1.4 Gy and 60%, respectively. Results of another study [38] indicate that HCT 116 cells are significantly more sensitive to radiation (SF2 = 38.3%, D 0 = 2.23 Gy) when compared with HT-29 cells (SF2 = 61.4%, D 0 = 3.51 Gy). Results of our study indicate that radiosensitivity of RR1 and RR2 sub-lines such as intrinsic radioresistant cell lines (SW-480 and HT-29) were significantly lower than the parental HCT 116 cell line [37, 38]. However, the RR2 sub-line sustained a radioresistant phenotype for at least 3 months after termination of fractionated irradiation (data not shown).

MiR-222, as a well-recognized onco-miR, is frequently upregulated in several types of human tumors [39]. In our study, miR-222 was significantly upregulated in RR2 and RR1 cell lines compared with the parental cell line. In some studies [21, 22, 40], acute exposure to high dose ionizing radiation leads to up-regulation of miR-222. PTEN as a candidate target of miR-222 was significantly downregulated in RR2 and RR1 in comparison with the parental cell line and its expression showed a significant negative correlation with miR-222. PTEN functions as a tumor suppressor gene, specifically by negatively regulating the Akt/PKB signaling pathway. Reduced expression of PTEN results in Akt hyperactivation, thereby promoting cell proliferation, inhibition of apoptosis, and enhanced cell invasion and radioresistance [41,42,43,44]. So, we can postulate that miR-222 mediated radiation resistance of RR2 and RR1 cell lines by targeting PTEN via the PI3/Akt pathway. Although PTEN expression was significantly decreased in RR2 compared to the RR1 cell line, the expression level of miR-222 did not show a significant difference between the RR2 and RR1 cell lines. We assume other miRNAs (or other gene regulation systems) are involved in PTEN gene expression and PTEN has been suppressed by another miRNA in addition to miR-222. That hypothetical miRNA may be overexpressed in higher cumulative doses of X-radiation. According to some studies, one gene may be suppressed by several miRNAs [45, 46]. However, overall changes in PTEN expression level and miR-222 level between three cell lines were significant by one-way ANOVA test and a high and significant correlation was observed between miR-222 and PTEN.

MiR-155 as an onco-miR is over-expressed in various solid tumors [19, 47,48,49]. In this study, miR-155 expression was significantly increased in RR2 and RR1 cell lines compared with the parental cell line. Results of similar studies indicate that chronic and acute exposure irradiation lead to upregulation of miR-155 in cancerous cell lines [23, 24, 50, 51]. In our study, FOXO3a as a candidate target of miR-155, was significantly decreased in RR2 and RR1 cell lines in comparison with the parental cell line. Furthermore, a significant negative correlation between miR-155 and FOXO3a gene was observed. FOXO3a is a member of the forkhead family of transcription factors, which are downstream effectors of the PI3 K/PKB pathway and participate in a variety of cellular processes, such as cell cycle progression, programmed cell death, stress and DNA damage repair [52]. We can consider FOXO3a gene as a target of miR-155, and hence conclude that miR-155 leads to the radiation resistance of RR2 and RR1 cell lines by targeting FOXO3a via the PI3/Akt pathway. Although miR-155 expression level was significantly increased, the change in FOXO3a expression level between RR1 and RR2 was not significant. The explanation of this phenomenon is similar to miR-222/PTEN. It is also possible that miR-155 targets other genes in addition to FOXO3a. Furthermore, FOXO3a expression suppression mediated by miR-155 might have been compensated by downregulation of other miRNAs that target the FOXO3a gene. Based on the results of recent studies, one miRNA can suppress the expression of several target genes and each mRNA may be suppressed by several miRNAs [46, 53]. However, the one-way ANOVA revealed that the overall changes in miR-155 and FOXO3a expression level in the three cell lines were significant and there was a significant correlation between miR-155 and FOXO3a.

In previous studies, acquired radiation resistance of CRC has been explained by the following two mechanisms: first, adaptive response to fractional radiation, which leads to acquired radioresistance of tumor cells [37]; and second, radioresistance of tumor cells, which may originate from cancer stem cells (CSCs). Unlike CSCs, non-stem cancer cells are radiosensitive and will die under fractional radiation, thereby increasing the population of CSCs [54,55,56,57,58,59]. On the other hand, non-stem cancer cells can undergo dedifferentiation under fractional radiation, which induces generation of novel CSCs [60,61,62]. However, the mechanism of acquired radioresistance is outside the scope of the present study, and requires further investigation.

Conclusion

We have recognized that miR-222/PTEN and miR-155/FOXO3a mediate radiation resistance in colorectal cancer cell lines via the PI3/Akt pathway. After conducting the observational case–control study, in the future miR-222 and miR-155 can be used as novel biomarkers to predict the effectiveness of radiotherapy in clinical cases. MiR-222 and miR-155 can also be used as clinical targets for reducing radiation resistance, although more accurate investigations such as using miRNA inhibition are needed. Finally, we suggest that cross-resistance of radioresistant sub-lines to different chemotherapeutic agents requires further evaluation.