Selection and validation of reference genes for RT-qPCR normalization in dormant cancer cells

doi:10.21203/rs.3.rs-4857207/v1

Download PDF

Article

Selection and validation of reference genes for RT-qPCR normalization in dormant cancer cells

https://doi.org/10.21203/rs.3.rs-4857207/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Recent findings have indicated that pharmacological inhibition of the mTOR kinase can become a widely used experimental approach to generate dormant cancer cells in vitro. However, the suppression of mTOR, which is responsible for global translation, can significantly rewire basic cellular functions influencing the expression of housekeeping genes. To prevent incorrect selection of a reference gene in dormant tumor cells, we analyzed the expression stability of the widely used housekeeping genes GAPDH, ACTB, TUBA1A, RPS23, RPS18, RPL13A, PGK1, EIF2B1, TBP, CYC1, B2M, and YWHAZ in T98G, A549, and PA-1 cancer cell lines treated with the dual mTOR inhibitor AZD8055. It has been revealed that the expression of the ACTB gene, encoding the cytoskeleton, and the RPS23, RPS18, and RPL13A genes, encoding ribosomal proteins, undergoes dramatic changes, and these genes are categorically inappropriate for RT-qPCR normalization in cancer cells treated with dual mTOR inhibitors. B2M and YWHAZ were determined to be the bestl reference genes in A549 cells, and the TUBA1A and GAPDH genes were the best reference genes in T98G cells. The optimal reference genes among the 12 candidate reference genes were not revealed in the PA-1 cell line. Validation of the stability of the 12 investigated genes demonstrated that the incorrect selection of a reference gene led to a significant distortion of the gene expression profile in dormant cancer cells.

Biological sciences/Cancer

Biological sciences/Cell biology

Biological sciences/Molecular biology

dormant cancer cells

mTOR

reference genes

ribosomal proteins

A549 cancer cells

T98G cancer cells

PA-1 cancer cells

The mTOR kinase is a clinically recognized key target for the elimination of cancer cells with increased activity of PI3K/mTOR signaling contributing to tumor growth and proliferation^1,2. According to preclinical and clinical studies, effective suppression of mTOR by dual inhibitors leads to a reduction in the size of solid tumors in vivo^1–3 and patient stabilization^4,5. However, these promising results have a significant shadow: pharmacological mTOR suppression may lead to the generation of numerous dormant cancer cells^6–9. The dormant state of cancer cells is a poorly characterized cellular phenotype. A key property of known dormant tumor cells is reversible cell cycle arrest in the G1/G0 phase¹⁰, but knowledge of specific signaling pathways and markers is still limited. Recent studies have revealed that suppression of the mTOR kinase can be a molecular determinant of dormant cancer cells, with pharmacological inhibition of mTOR being the mechanistic basis for the production of dormant tumor cells in vitro^6–8,11. In a widely known study, Rehman et al. treated colorectal cancer cells with the dual mTOR inhibitor INK128 and obtained a culture of dormant cancer cells that resumed proliferation after withdrawal without loss of viability⁶. The authors have demonstrated a simple strategy for the generation of dormant cancer cells in vitro, enabling the search for key characteristics of this pathological condition. Worth to mention that mTOR-inhibited cancer cells can undergo extensive proteome changes caused by the shutdown of global mTOR-dependent mRNA translation and the activation of alternative translation pathways¹². Dramatic mTOR-dependent alterations in proteostasis can induce responsive changes in basic cellular functions, modulating the stable expression of housekeeping genes under a dormant phenotype. Analysis of the stable expression of housekeeping genes has not been reported in the literature, despite the numerous data published on cancer cells treated with dual mTOR inhibitors. To prevent potential errors in the interpretation of gene expression results in dormant cancer cells, ensuring the availability of relevant reference genes for RT-qPCR data normalization in tumor cells after mTOR suppression is necessary.

Generation of dormant cancer cells in vitro by pharmacological suppression of mTOR kinase

Chemotherapy is the mainstay treatment for most types of cancer. While killing the bulk of tumor cells, chemotherapeutic agents blocking mTOR kinase can provoke the emergence of a drug-resistant population of cancer cells^6,8. To demonstrate this phenomenon in vitro, AZD8055, a highly ATP-competitive inhibitor of mTOR kinase activity³, has been used to convert proliferating cancer cells into the dormant state without inducing cell death. Three tumor cell lines of different etiologies and pathogenesises were cultured for a long period of time at high doses of the agent. A549 (lung adenocarcinoma), T98G (glioblastoma), and PA-1 (ovarian teratocarcinoma) cells were grown for 1 week under AZD8055 treatment at concentrations of 0.5, 1, 2.5, 5, and 10 µM. After that, the antiproliferative effect of AZD8055 was subsequently reversed by changing the culture medium to one containing no inhibitor. Tumor cell repopulation was recorded after one and two weeks (Fig. 1a). According to the data obtained, A549, T98G, and PA-1 cancer cell populations were dose-dependently restored after one and two weeks of AZD8055 treatment. Propidium iodide staining of A549, T98G, and PA-1 living cells did not reveal dead cells permeable to this dye (Fig. 1b). The dose-dependent cancer cell repopulation may reflect the differing abilities of tumor cells to recover from the dormant phenotype, or the so-called depth of cellular quiescence, which was recently reported in the literature^13,14. Assuming that tumor cells treated with the highest concentration of AZD8055 undergo the greatest intracellular changes and a greater depth of quiescence, an AZD8055 concentration of 10 µM was chosen as the most appropriate to obtain dormant cancer cells in vitro and to select suitable reference genes in these hibernating cells.

Candidate reference gene selection and primer design

To identify the most relevant reference genes for normalization of RT-qPCR data in dormant cancer cells, the 12 candidate genes were selected from among widely used references, according to the literature: GAPDH, ACTB, TUBA1A, RPS23, RPS18, RPL13A, PGK1, EIF2B1, TBP, CYC1, B2M, and YWHAZ. To correctly determine the level and expression stability of candidate reference genes, appropriate primers were designed, and their specificity was assessed (Table 1). The specificity of the primers was evaluated with the coefficient of determination (R²) and the coefficient of efficiency (E), appended by melt curve analyses (see Supplementary Fig. S1 and S2). These coefficients were derived from the Ct values obtained from the qPCR of the analyzed primers in serial dilutions of cDNA 1:1, 1:10, 1:100, 1:1000, and 1:10000. R² measures a linear data distribution, predicting Y(Ct) as a linear function of X (log (DNAcopies)). E measures the rate at which the polymerase converts the reagents to amplicon. The dilution series produces a linear standard curve with the goal of 90–110% reaction efficiency. Therefore, according to generally accepted recommendations, R² > 0.99 indicates a good linearity of the dilution, within which E reaches values of 90–110%¹⁵. R² > 0.99 was obtained for all investigated primers. E was within the range of 90–110% for all investigated primers, with the exceptions of PGK1 (E = 113.5%), RPL13A (E = 111.9%), and RPS23 (E = 114.2%). High reaction efficiency (> 110%) is generally the result of amplicons or primer–dimers¹⁵, but the ones were not observed in the melting curves; therefore, the used primers were determined to be highly specific.

Table 1

Selected candidate reference genes, primer sequences and PCR amplification characteristics.
Gene	Gene name	NCBI sequence	Sequence of forward and reverse primers	Product size (bp)	Efficiency (%)	R²
ACTB	actin beta	NM_001101.5	CTTCGCGGGCGACGAT	104	91,538	1,000
ACTB	actin beta	NM_001101.5	CCACATAGGAATCCTTCTGACC	104	91,538	1,000
B2M	beta-2-microglobulin	NM_004048.4	CATGGAGGTTTGAAGATGCCGC	239	93,945	0,998
B2M	beta-2-microglobulin	NM_004048.4	CCCCACCTCTAAGTTGCCAGCC	239	93,945	0,998
CYC1	cytochrome c1	NM_001916.5	GAGGTGGAGGTTCAAGACGG	160	91,038	0,995
CYC1	cytochrome c1	NM_001916.5	TAGCTCGCACGATGTAGCTG	160	91,038	0,995
EIF2B1	eukaryotic translation initiation factor 2B subunit alpha	NM_001414.4	CTACTCCAGAGTGGTCCTGAGA	130	105,081	1,000
EIF2B1	eukaryotic translation initiation factor 2B subunit alpha	NM_001414.4	GTTGAGGTGGCAGAGGGCTTTG	130	105,081	1,000
GAPDH	glyceraldehyde-3-phosphate dehydrogenase	NM_002046.7	GAGGTCAATGAAGGGGTCAT	100	95,649	0,998
GAPDH	glyceraldehyde-3-phosphate dehydrogenase	NM_002046.7	AGTCAACGGATTTGGTCGTA	100	95,649	0,998
PGK1	phosphoglycerate kinase 1	NM_000291.4	AACCAGAGGATTAAGGCTGC	195	113,529	0,995
PGK1	phosphoglycerate kinase 1	NM_000291.4	GCCTACACAGTCCTTCAAGA	195	113,529	0,995
RPL13A	ribosomal protein L13a	NM_012423.4	CCTGGAGGAGAAGAGGAAAGAGA	126	111,903	0,995
RPL13A	ribosomal protein L13a	NM_012423.4	TTGAGGACCTCTGTGTATTTGTCA	126	111,903	0,995
RPS18	ribosomal protein S18	NM_022551.3	GGATGAGGTGGAACGTGTGA	151	103,411	0,998
RPS18	ribosomal protein S18	NM_022551.3	CAGGTCTTCACGGAGCTTGT	151	103,411	0,998
RPS23	ribosomal protein S23	NM_001025.5	TGGCACAGTTTCCCTCATCC	177	114,227	0,999
RPS23	ribosomal protein S23	NM_001025.5	AACTGAAGTGTTGCACCCGA	177	114,227	0,999
TBP	TATA-box binding protein	NM_003194.5	TGTATCCACAGTGAATCTTGGTTG	124	99,445	0,998
TBP	TATA-box binding protein	NM_003194.5	GGTTCGTGGCTCTCTTATCCTC	124	99,445	0,998
TUBA1A	tubulin alpha 1a	NM_006009.4	GAAGCAGCAACCATGCGTGA	98	97,046	0,995
TUBA1A	tubulin alpha 1a	NM_006009.4	CCGTGTTCCAGGCAGTAGAG	98	97,046	0,995
YWHAZ	tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein zeta	NM_003406.4	ACCGTTACTTGGCTGAGGTTGC	130	100,403	0,999
YWHAZ		NM_003406.4	CCCAGTCTGATAGGATGTGTTGG	130	100,403	0,999

Expression levels and variabilities of candidate reference genes

Using the characterized primers and qPCR method, Ct values of the indicated candidate reference genes were obtained for A549, T98G, and PA-1 tumor cell lines untreated and treated with 10 µM of AZD8055 for 1 week. The obtained data are presented as the mean Ct values derived from the sum Ct values from untreated and treated samples and their standard deviations (SD) for each selected gene within each cell line (Fig. 2a). The raw Ct data are included in the appendix (see Supplementary Fig. S3). The main requirements for reference genes are minimal variability and a high expression level. The observed mean Ct values indicated that the 12 investigated genes were highly expressed, satisfying one of these requirements for a reference gene. To compare three different datasets with different scales, the coefficient of variation (CV) was calculated. The CV shows the relative dispersion of data points around the mean Ct, allowing to compare the percentage variation from the mean in percentage. The estimated expression of the 12 genes varied from 0.71–5.22% in A549 cells, from 1.63–8.54% in PA-1 cells, and from 2.03–5.69% in T98G cells (Fig. 2b). The expression of the RPS18, RPS23, RPL13A, and ACTB genes exhibited the greatest variation in each cell line (Fig. 2b). The variations in the B2M expression (0.97% in A549; 8.54% in PA-1; 5.18% in T98G) and the PGK1 expression (2.05% in A549; 5.38% in PA-1; 2.03% in T98G) differed depending on the cell type. The expression variability of the remaining genes, such as YWHAZ, GAPDH, TUBA1A1, TBP, EIF2B1, and CYC1, mostly overlapped among the three cell lines. YWHAZ (0.71%), B2M (0.97%), and GAPDH (1.39%) were identified as the three top genes with the smallest CVs in A549 cells; TBP (1.63%), CYC1 (2.49%), and YWHAZ (2.50%) were identified in PA-1 cells; and PGK1 (2.03%), TUBA1A (2.04%), and YWHAZ (2.06%) were identified in T98G cells. The CV analysis is used to obtain a rough rank of the most stable reference genes but does not examine the relationship between two or more variables. The measurement of intragroup variation can reveal co-regulated genes and systematic errors in the PCR method. Therefore, at this stage, genes are anticipatory classified as stable or unstable and then subjected to the NormFinder and BestKeeper algorithms for verification. Comparative ∆Ct and geNorm approaches, which have been criticized in the literature,^16,17 were used as additional methods for assessing the data.

Expression stability of candidate reference genes

NormFinder and BestKeeper analyses

The NormFinder approach calculates the stability of candidate reference genes based on inter- and intragroup variation. NormFinder calculates a “stability value” (SV) as a final measure of these two parameters for each gene, with the smallest value signifying the most stable reference gene. NormFinder identified B2M, YWHAZ, and TBP as the three most stable genes in A549 cells, partially coinciding with the top of CV calculations for these genes (Fig. 3 and Fig. 2b). The YWHAZ, TBP, and TUBA1A genes were defined as the most stable genes by NormFinder in the PA-1 cell line, partially overlapping with the CV results (Fig. 3 and Fig. 2b). TUBA1A, YWHAZ, and GAPDH were ranked as the three top genes according to NormFinder in T98G cells, that was also partially overlapped with the CV calculations (Fig. 3 and Fig. 2b). A striking result obtained from NormFinder is the demonstration of genes related to the translation pathway (RPS18, RPS23, and RPL13A) as a highly inappropriate internal controls in each cancer cell line. The ACTB gene was also categorically excluded as a good reference gene by NormFinder in A549, PA-1, and T98G cells, reflecting the fact that the cytoskeleton undergoes major changes in dormant cancer cells^18,19. To confirm the results obtained from NormFinder in the A549, PA-1, T98G cell lines, BestKeeper analysis was performed.

BestKeeper is an Excel-based tool that detects the most stable reference gene by applying descriptive statistics methods (calculating the standard deviation (SD) and the coefficient of variation (CV) for each gene) and pairwise correlation analysis²⁰. The stability rating of candidate genes is ordered on the basis of SDs. The authors defined a threshold of SD = 1, and genes exhibiting SD > 1 were considered highly variable and removed from further analysis. BestKeeper confirmed the NormFinder results, identifying B2M (SD = 0) and YWHAZ (SD = 0) as the most stable genes in A549 cells (Fig. 4). In PA-1 cells, BestKeeper also confirmed the NormFinder results by calculating the TBP (SD = 0.375), YWHAZ (SD = 0.500), and TUBA1A (SD = 0.563) genes as the three top genes but with rearranging their ranks (Fig. 4). TUBA1A (SD = 0.219) and GAPDH (SD = 0.250) were the genes with the lowest SD values in T98G cells, which is consistent with the NormFinder data (Fig. 4). The SD values from BestKeeper were greater than 1 for CYC1, ACTB, RPS18, and RPS23 in A549; for ACTB, RPL13A, RPS18, PGK1, RPS23, and B2M in PA-1; and for ACTB, RPS18, and B2M in T98G. Therefore, the BestKeeper results indicated that ACTB and genes related to translational pathways were not suitable for normalizing RT-qPCR data in cancer cells treated with dual mTOR inhibitors.

Dormant tumor cells appear to undergo extensive intracellular changes affecting the majority of basic cellular functions to adapt to extreme conditions. To assess the scale of co-regulation between the 12 studied genes, the Pearson correlation coefficient (R) was applied. The Pearson correlation coefficient is an index ranging from − 1.0 to 1.0, which reflects the degree of linear relationship between two datasets. It has been widely accepted that |R| > 0.7 indicates a strong linear relationship between two experimental variables. A high level of correlation (|R| > 0.7; p < 0.05) was revealed for almost all the candidate reference genes in A549, T98G, and PA-1 cells, indicating overall co-dependent expression between most pairs of housekeeping genes (Fig. 5). The greatest correlation of expression between the 12 investigated genes was identified in PA-1 cells, showing that the PA-1 cell line underwent the greatest intracellular changes compared with A549 and T98G cells. TBP and TUBA1A were determined to be the least correlated genes within the selected pool of genes but correlated with each other in the PA-1 cell line. NormFinder and BestKeeper calculated TBP, YWHAZ, and TUBA1A as the best reference genes in PA-1 cells. Based on the Pearson correlation coefficient, YWHAZ was excluded from this trio of suitable reference genes in PA-1 cells. B2M and YWHAZ, recognized by NormFinder and BestKeeper as the best reference genes, were the only genes whose expression did not correlate with that of the other genes in A549 cells. GAPDH, TUBA1A, and YWHAZ were identified as the three top reference genes in T98G cells. Considering that TUBA1A and GAPDH demonstrated no correlation with any genes unlike YWHAZ, the last was excluded from this trio as optimal reference gene for normalization qPCR data in the T98G cell line.

Comparative ∆Ct method and geNorm analyses

Comparative ∆Ct method and geNorm analysis use pairwise approaches but with different mathematical algorithms and outcomes. The ∆Ct method is based on the comparison of relative expression of candidate gene pairs within each sample. First, the Ct values of each gene are compared with those of all other genes in pairs, and the ΔCt values, means, and standard deviations are calculated. Then, based on the comparisons of all possible pairs, the mean standard deviation (SD) is obtained for each gene. If the ΔCt value between two genes remains constant in all samples, it means that these genes are either stably expressed under specific conditions or co-regulated. An SD below 1 indicates very stable gene expression.

∆Ct method ranked 12 genes into two discrete groups with SDs up to 1 (B2M, YWHAZ, TBP, GAPDH, TUBA1A, PGK1, RPL13A) and above (EIF2B1, CYC1, ACTB, RPS23, RPS18) in the A549 cell line (Fig. 6a). The B2M, YWHAZ, and TBP genes were ranked the same as in previous statistical analyses and were determined to be the top reference genes in A549 cells. ACTB, RPS23, and RPS18 were categorically excluded as reference genes from this cell line. In the PA-1 cell line, the SD values for 12 genes were greater than 1, indicating that all investigated candidate reference genes were unstable (Fig. 6a). Additionally, considering the revealed general correlation of expressions, the 12 investigated candidate genes had poor statistical indicators to be good reference genes. The ∆Ct method also confirmed that the TUBA1A, YWHAZ, and GAPDH genes were the optimal reference genes in T98G cells, and defined these genes as the only with SD ˂1 (Fig. 6a). The remaining genes were defined as unstable with SD˃1 in the T98G cell line. The ∆Ct method also revealed that ACTB, RPS18, RPS23, and RPL13A are not suitable for reference genes in dormant cancer cells.

GeNorm ranks genes based on pairwise correlation. GeNorm algorithm calculates M values for all candidate genes, stepwise eliminating the genes with the highest M value and recalculating the stability of M value for the remained genes to identify two candidates with the lowest M value. The M value threshold for gene expression stability was established as 1.5. According to the results obtained via geNorm, all 12 candidate genes had the M˂1.5 and were considered as stable in the A549, T98G, and PA-1 cell lines (Fig. 6b). These results can be explained by a limitation that ranking by geNorm may depend on the degree of correlation between genes^17,21. Nevertheless, geNorm algorithm substantively confirmed previous results that B2M and YWHAZ were the most suitable reference genes among the 12 candidate genes in A549 cells; TUBA1A was the optimal internal control for T98G cells; and RPS23, RPS18 and ACTB were unreliable references for dormant cancer cells.

Validation of expression stability of candidate reference genes

To validate the stability of the 12 candidate reference genes, we measured and analyzed the mRNA expression levels of the CDK1 and CCND1 gene, which transcribe the CDK1 and CyclinD1 proteins, accordingly, among three cancer cell lines that were untreated and treated with the mTOR inhibitor AZD8055 (Fig. 7). It is known that proliferative and quiescent signals converge on cell cycle regulators, the analysis of which predicts dormant or cycling cellular fates. CyclinD1 and CDK1, which control G1/S phase and G2/M cell cycle transitions, respectively, can be used as relevant markers for validation of candidate reference genes in dormant cancer cells.

Using the 12 investigated genes as references showed different fold reductions in CDK1 expression in A549 cells (Fig. 7a). Through comprehensive statistical analysis, B2M and YWHAZ were determined to be the best reference genes among the 12 investigated genes in A549 cells. Normalization by B2M and YWHAZ led to similar reductions in CDK1 expression in 47.5 and 44.7 fold, respectively. Using TBP gene, expression of which correlated with YWHAZ, as internal control resulted in a similar level of CDK1 expression, with 44.5-fold change. TUBA1A and GAPDH genes, which were ranked behind the top genes, normalized CDK1 expression to a similar level, with 73.2- and 61.7-fold changes, but this result was already error-prone. Applying RPS23 as internal control resulted in an enormous decrease in CDK1 expression, with 266.2-fold change. The use of RPS18 and RPL13A also led to false positive results, with decreases in CDK1 transcription of 175.5 and 93.7-fold, respectively. The 11.1- to 24.4-fold decrease in relative CDK1 expression in A549 cells based on the EIF2B1, PGK1, CYC1, and ACTB reference genes should also not be considered as true results. Indeed, using CYC1, ACTB, and PGK1 for normalization of CCND1 relative expression resulted in CCND1 upregulation, a molecular event that is unlikely to occur in dormant cells. The application of the EIF2B1 gene as an internal control revealed no changes in CCND1 relative expression. According to the statistical data, the accurate result was a decrease in the CCND1 expression level, with 1.66- to 1.77-fold change, normalized to the B2M, YWHAZ, and TBP reference genes. TUBA1A and GAPDH, as internal controls, resulted in an incorrect decrease in CCND1 of 2.3-2.73-fold, whereas the RPS23, RPS18, and RPL13A reference genes led to false positive results of 3-9-fold. Therefore, a comprehensive statistical analysis was consistent with the experimental results confirming B2M and YWHAZ as the best reference genes among the 12 investigated candidate genes in the A549 cell line.

The 12 investigated genes used as references showed different fold reductions in CDK1 expression levels in T98G cells treated with AZD8055 (Fig. 7b). The TUBA1A and GAPDH were determined to be the two top reference genes among the 12 investigated genes in T98G cells. Normalization of relative CDK1 expression by TUBA1A and GAPDH led to similar decreases of 34.2- and 31.2-fold, respectively. Normalization by TUBA1A and GAPDH also resulted in similar levels of CCND1 downregulation (6.21 and 5.67-fold, respectively). The application of the RPS18 and B2M reference genes led to enormous decreases in CDK1 and CCND1 expression in 168.3-159.7-fold and 30.58-29.03-fold, respectively, in T98G cells. Using RPL13A and RPS23 as internal controls also led to falsification of the results, indicating significant downregulation of CDK1 in 94.8- and 41.3-fold and CCND1 in 17.23- and 7.5-fold. The obtained statistical and experimental data demonstrated that the ACTB gene was not a suitable reference gene in dormant A549 cells. Similarly, this gene was not relevant as a reference gene in T98G cells, demonstrating slight downregulation of CDK1 and unchanged relative expression of CCND1. The observed 18.8-11.7-fold and 3.42-2.12-fold decrease in CDK1 and CCND1 expression, respectively, normalized for the reference genes YWHAZ, CYC1, EIF2B1, PGK1 and TBP in T98G cells should also not be considered true results according to the statistical data obtained. Therefore, the experimental results confirmed that TUBA1A and GAPDH are the best references genes among the 12 investigated candidate genes in the T98G cell line.

In turn, validation of the 12 candidate reference genes in the PA-1 cell line revealed that none of the investigated genes were relevant for obtaining correct results for CDK1 and CCND1 expression (Fig. 7c). The use of TBP and TUBA1A as the best reference genes led to the questionable outcome of normalization of the CDK1 expression level. TBP, as an internal control, caused slight CDK1 downregulation, whereas TUBA1A caused no change in CDK1 expression. Normalization by the RPS23, RPS18, and RPL13A reference genes, which are categorically unsuitable, led to substantial CDK1 downregulation which was considered as a false positive result. The use of the remaining genes as references revealed an increase in the CDK1 mRNA transcription level of 1.8-3.1-fold, excluding B2M. The application of B2M as an internal control led to substantial upregulation of CDK1 expression, with 12.5-fold change. Similarly, normalization of CCND1 gene expression by the 12 reference genes was observed questionable in PA-1 cells. The application of TBP and TUBA1A as internal controls resulted in the upregulation of CCND1 relative expression, with 5.5-8.3-fold change, respectively, in PA-1 cells treated with AZD8055. Normalization by B2M demonstrated an extreme increase in CCND1 expression, with 100-fold change. The use of the remaining reference genes led to erroneous CCND1 upregulation of 12.5-20-fold. Therefore, despite a comprehensive statistical analysis, the practical validation of the 12 investigated candidate reference genes did not reveal the correct reference gene.

Suppression of mTOR kinase induces diapause in embryonic stem cells (ESCs)^22,23. According to RNA-sequencing data, dormant cancer cells treated with the dual mTOR inhibitor INK128 demonstrated a similar transcriptome profile with diapaused ESCs^6,7. Obtained fact can identify that non-proliferating cancer cells can undergo through a fundamental intracellular changes to acquire a new phenotype as a part of the pro-survival program. A state of dormancy appears to be maintained at both translational and transcriptional levels. The state of mitotic arrest in cancer cell lines can alter the basal levels of physiological functions. Gained data confirmed this hypothesis, clearly showing that the majority of housekeeping genes did not maintain their expression after the conversation of tumor cells to a dormant state. The altered expression of the RPS23, RPS18 and RPL13A genes demonstrated changes in the translational network of the mTOR-suppressed cancer cells. Therefore, genes participating in translation cannot be relevant as reference genes in dormant cancer cells. The ACTB gene has also been categorized as an ineligible internal control in A549, T98G, and PA-1 cancer cell lines treated with AZD8055. This result was also confirmed by data in the literature, which demonstrated that the transition from dormancy to proliferation of tumor cells is dependent on reorganization of F-actin^18,19. Additionally, dormant tumor cells leverage mechanisms allowing them to avoid immunosurveillance by downregulating major histocompatibility complex I (MHC I), which is essential for T cell recognition²⁴. B2M, as an important subunit of MHC class I, can also be downregulated in dormant cancer cells. Therefore, despite the definition of B2M as a stable reference gene in A549 cells, the reliable expression of this gene in other dormant cancer cell lines is questionable. This proposition is supported by the results revealing the non stable expression of B2M in the T98G and PA-1 cancer cell lines after AZD8055 treatment. Therefore, the use of RPS23, RPS18, RPL13A, ACTB, and B2M (in some cancer types) as reference genes can lead to significant errors in the interpretation of RT-qPCR data, generating false positive or false negative results. For example, the decrease in CDK1 expression, with 266.2-fold change in A549 cells and 168.3-fold change in T98G cells after AZD8055 treatment, looks like a very reasonable measurement of the gene expression in dormant cancer cells. However, a more realistic decrease in CDK1 expression is 47.5-44.7-fold change in A549 cells (based on B2M and YWHAZ) and 34.2-31.2-fold change in T98G cells (based on TUBA1A and GAPDH). Data obtained from three tumor cell lines showed how easy it is to make the mistake of choosing the wrong reference. Taking TUBA1A, GAPDH, YWHAZ, and TBP as internal controls may indicate a more appropriate ranges of gene transcription for most dormant cancer cells, but with the possibility of errors. At least two reference genes should be applied to avoid fatal errors in RT-qPCR normalization in tumor cells treated with dual mTOR inhibitors, until the most relevant reference gene has not yet been revealed.

One of the results of this work is the indication of cardinal intracellular changes occurring in cancer cells entering dormancy. For example, the reference gene was not selected from the 12 studied candidate reference genes in the PA-1 cell line. The absence of a suitable reference gene resulted in a lack of understanding of the transcriptional activity of the CDK1 and CCND1 genes in PA-1 cells: being increased, decreased, or remained unchanged. Using an incorrect reference gene in this case led to a significant distortion of the interpretation of gene expression of cell cycle regulators in dormant cancer cells. Obviously, the transcription of CDK1 and CCND1 must be greatly reduced after 1 week of high dose AZD8055 treatment in PA-1 cells, similarly to A549 and T98G cells. However, this conclusion is correct if downregulation of these cell cycle regulators is an obligatory molecular event in all types of dormant cancer cells. For this reason, searching for suitable reference genes is an important step in understanding the biology of cancer cell dormancy.

Cell culture and treatment

T98G (glioblastoma), A549 (lung adenocarcinoma), and PA-1 (ovarian teratocarcinoma) tumor cell lines were cultured at 37°C and 5% CO₂ in DMEM medium (HiMedia, India) supplemented with 10% fetal bovine serum (HyClone, USA), 1 g/l glucose, and 5 mM glutamine (Paneco, Russia). Cells were obtained from the shared research facility “Vertebrate cell culture collection” (Institute of Cytology). The ATP-competitive inhibitor of mTOR kinase activity AZD8055 and Actinomycin D (Selleckchem, USA) were used.

Analysis of cancer cell repopulation

Cells from each tumor cell line were seeded in parallel into 3 6-well plates. A total of 3000 cells per well were seeded and treated with 0.5 µM, 1 µM, 2.5 µM, 5 µM, and 10 µM of AZD8055 for 1 week. After 7 days of AZD8055 incubation, one plate for each cell line was analyzed; the culture medium containing AZD8055 was replaced with clear medium in the two remaining plates, which were analyzed after 7 and 14 days after AZD8055 removal. Cells were washed with PBS, fixed with 4% formalin in PBS for 10 minutes at room temperature, and stained with crystal violet for 2 hours. The plates were left to dry after which images were taken via the SNAPSCAN600 device (Agfa, Belgium).

Cell death analysis

Adherent A549, T98G, and PA-1 tumor cells were harvested by trypsinization, stained with 10 µg/mL propidium iodide, and analyzed using a flow cytometer the CytoFLEX (Beckman Coulter, Brea, CA, USA) with a PE filter. The obtained data were analyzed using CytExpert software version 2.0 (Beckman Coulter, Brea, CA, USA). A minimum of 10000 cells were analyzed in each sample.

RNA extraction

Total RNA was isolated using ExtractRNA reagents (Evrogen, Russia) according to the manufacturer's protocol. Briefly, the cells were lysed in 1 ml of ExtractRNA reagent, and 250 µl of 1-bromo-3-chloropropane was added. To precipitate RNA in the aqueous phase, 600 µl of isopropanol was added to 1 ml of the reagent. The precipitate was washed with 75% ethanol three times, dried, and dissolved in diethylpyrocarbonate (DEPC)-treated water. The concentration of the isolated RNA was measured using NanoDrop 2000 (Thermo Scientific, USA). The absorbance ratios of all samples at OD260/OD280 were between 1.88 and 1.92 and at OD260/230 were between 2.1 and 2.2.

cDNA synthesis

Reverse transcription was performed using the MMLV RT kit (Evrogen, Russia) in 20 µl of total volume using 3 µg of total RNA, according to the manufacturer’s protocol. 3 µg of total RNA, 1.5 µg random hexaprimes, and deionized water were mixed and incubated at 70°C for 2 minutes. A reaction mixture containing 4 µl of 5x buffer for cDNA synthesis, 2 µl of DTT, 2 µl of dNTP mixture, deionized water, and 1 µl of MMLV reverse transcriptase was added to the solution. A reverse transcription reaction was performed at 42°C for 90 minutes. Synthesis was stopped by heating the samples to 70°C for 10 minutes.

qPCR

Quantitative reverse transcription polymerase chain reaction (qRT-PCR) was conducted using the qPCRmix-HS SYBR PCR kit (Evrogen, Russia). Each reaction consisted of 2 µl of cDNA, 5 µl of 5x qPCRmix-HS SYBR buffer, 1 µl of forward and reverse primers, 16 of µl deionized water. The PCR protocol involved amplification initiation at 95°C for 5 minutes, followed by 40 cycles of denaturation (95°C, 30 seconds), annealing (60°C, 30 seconds), elongation (72°C, 30 seconds). The Ct threshold cycle data of gene candidates was measured on the CFX96 Real-time PCR Detection System (Bio-Rad Laboratories, CA, USA). The primers for candidate reference genes were designed in NCBI Primer-BLAST (Table 1). The following primers for validation of reference genes were used: CDK1-F, CAGGTCAAGTGGTAGCCATGAA; CDK1-R, TTGCAGTACTAGGAACCCCTTC; CCND1-F, GCTGCGAAGTGGAAACCATC; CCNG1-R, CCTCCTTCTGCACACATTTGAA.

Analysis of gene expression stability and statistical analysis

The stability of the 12 selected candidate reference genes was estimated by the coefficient of variation, the Pearson correlation coefficient, the comparative NormFinder²¹, BestKeeper²⁰, ΔCt method²⁵, and geNorm²⁶. For convenience, the results of all the statistical algorithms were combined into a single file (see Supplementary Fig. S4). The input data were the mean Ct values obtained from eight independent qPCR experiments including four untreated and four treated samples from three tumor cell lines. The average Ct value for one independent qPCR experiment was obtained from four technical replicates. The normality of the PCR data distribution was verified using the Shapiro-Wilk test. The coefficient of variation (CV) was calculated as the standard deviation of the Ct values divided by the mean of these values. The amplification efficiency (E) for each reference gene was calculated according to the following equation: E (%) = (10^{− 1/slope}–1)×100, where the slope is the angular coefficient k of the straight line equation y = k*x + b. The relative expression was conducted with the comparative 2^-∆∆Ct method. To assess the significant differences between the two groups, Student’s t-test was performed in GraphPad Prism version 8 software package. The Pearson correlation coefficients were also generated in GraphPad Prism version 8 software package, and differences with p < 0.05 were regarded as statistically significant.

Competing interests

The authors declare no competing interests.

Funding

This work was supported by the Russian Science Foundation (grant number 23-25-10017) and the St. Petersburg Science Foundation (grant number 23-25-10017).

Author Contribution

I.S. designed and planned the study; E.I. and K.A. carried out the experiments; E.I., K.A., and I.S. analyzed and interpreted data; I.S. wrote the manuscript; E.I. and K.A. prepared the pictures and edited the text; All authors reviewed the manuscript.

Data Availability

All data generating during this study are included in this published article and in the Supplementary files.

Fang, D. D. et al. Antitumor Efficacy of the Dual PI3K/mTOR Inhibitor PF-04691502 in a Human Xenograft Tumor Model Derived from Colorectal Cancer Stem Cells Harboring a PIK3CA Mutation. PLoS One 8, e67258 (2013).
Sl, F. et al. MTORC1/2 Inhibition as a Therapeutic Strategy for PIK3CA Mutant Cancers. Molecular cancer therapeutics 18, (2019).
Chresta, C. M. et al. AZD8055 is a potent, selective, and orally bioavailable ATP-competitive mammalian target of rapamycin kinase inhibitor with in vitro and in vivo antitumor activity. Cancer Res 70, 288–298 (2010).
Powles, T. et al. A Randomised Phase 2 Study of AZD2014 Versus Everolimus in Patients with VEGF-Refractory Metastatic Clear Cell Renal Cancer. Eur Urol 69, 450–456 (2016).
Morscher, R. J. et al. First-in-child phase I/II study of the dual mTORC1/2 inhibitor vistusertib (AZD2014) as monotherapy and in combination with topotecan-temozolomide in children with advanced malignancies: arms E and F of the AcSé-ESMART trial. Eur J Cancer 157, 268–277 (2021).
Rehman, S. K. et al. Colorectal Cancer Cells Enter a Diapause-like DTP State to Survive Chemotherapy. Cell 184, 226–242.e21 (2021).
Dhimolea, E. et al. An Embryonic Diapause-like Adaptation with Suppressed Myc Activity Enables Tumor Treatment Persistence. Cancer Cell 39, 240–256.e11 (2021).
Liu, Y., Azizian, N. G., Sullivan, D. K. & Li, Y. mTOR inhibition attenuates chemosensitivity through the induction of chemotherapy resistant persisters. Nat Commun 13, 7047 (2022).
Aleksandrova, K. V., Vorobev, M. L. & Suvorova, I. I. mTOR pathway occupies a central role in the emergence of latent cancer cells. Cell Death Dis 15, 1–11 (2024).
Weston, W. A. & Barr, A. R. A cell cycle centric view of tumour dormancy. Br J Cancer 129, 1535–1545 (2023).
Khalil, B. D. et al. An NR2F1-specific agonist suppresses metastasis by inducing cancer cell dormancy. J Exp Med 219, e20210836 (2022).
Shin, S. et al. mTOR inhibition reprograms cellular proteostasis by regulating eIF3D-mediated selective mRNA translation and promotes cell phenotype switching. Cell Rep 42, 112868 (2023).
Kwon, J. S. et al. Controlling Depth of Cellular Quiescence by an Rb-E2F Network Switch. Cell Rep 20, 3223–3235 (2017).
Fujimaki, K. et al. Graded regulation of cellular quiescence depth between proliferation and senescence by a lysosomal dimmer switch. Proc Natl Acad Sci U S A 116, 22624–22634 (2019).
Taylor, S., Wakem, M., Dijkman, G., Alsarraj, M. & Nguyen, M. A practical approach to RT-qPCR—Publishing data that conform to the MIQE guidelines. Methods 50, S1–S5 (2010).
Robledo, D. et al. Analysis of qPCR reference gene stability determination methods and a practical approach for efficiency calculation on a turbot (Scophthalmus maximus) gonad dataset. BMC Genomics 15, 648 (2014).
Sundaram, V. K., Sampathkumar, N. K., Massaad, C. & Grenier, J. Optimal use of statistical methods to validate reference gene stability in longitudinal studies. PLoS One 14, e0219440 (2019).
Barkan, D. et al. Inhibition of Metastatic Outgrowth From Single Dormant Tumor Cells by Targeting the Cytoskeleton. Cancer Res 68, 6241–6250 (2008).
Barkan, D., Green, J. E. & Chambers, A. F. Extracellular matrix: a gatekeeper in the transition from dormancy to metastatic growth. Eur J Cancer 46, 1181–1188 (2010).
Pfaffl, M. W., Tichopad, A., Prgomet, C. & Neuvians, T. P. Determination of stable housekeeping genes, differentially regulated target genes and sample integrity: BestKeeper – Excel-based tool using pair-wise correlations. Biotechnology Letters 26, 509–515 (2004).
Andersen, C. L., Jensen, J. L. & Ørntoft, T. F. Normalization of real-time quantitative reverse transcription-PCR data: a model-based variance estimation approach to identify genes suited for normalization, applied to bladder and colon cancer data sets. Cancer Res 64, 5245–5250 (2004).
Bulut-Karslioglu, A. et al. Inhibition of mTOR induces a paused pluripotent state. Nature 540, 119–123 (2016).
Hussein, A. M. et al. Metabolic Control over mTOR-Dependent Diapause-like State. Dev Cell 52, 236–250.e7 (2020).
Pommier, A. et al. Unresolved endoplasmic reticulum stress engenders immune-resistant, latent pancreatic cancer metastases. Science 360, eaao4908 (2018).
Silver, N., Best, S., Jiang, J. & Thein, S. L. Selection of housekeeping genes for gene expression studies in human reticulocytes using real-time PCR. BMC Molecular Biology 7, 33 (2006).
Vandesompele, J. et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biology 3, research0034.1 (2002).

No competing interests reported.

Download PDF

Reviews received at journal
15 Sep, 2024
Reviewers agreed at journal
27 Aug, 2024
Reviewers agreed at journal
26 Aug, 2024
Reviewers agreed at journal
23 Aug, 2024
Reviewers invited by journal
21 Aug, 2024
Editor assigned by journal
21 Aug, 2024
Editor invited by journal
21 Aug, 2024
Submission checks completed at journal
17 Aug, 2024
First submitted to journal
04 Aug, 2024

You are reading this latest preprint version

Selection and validation of reference genes for RT-qPCR normalization in dormant cancer cells

Status:

Version 1

Abstract

Figures

Introduction

Results

Discussion

Materials and methods

Declarations

Competing interests

Funding

Author Contribution

Data Availability

References

Additional Declarations

Supplementary Files

Status:

Version 1