Evaluation and integration of normalization approaches and internal reference genes in real-time quantitative reverse transcription PCR in liver tissues of db/db mice*

LI Kai-ji, TIAN Zeng-you, TIAN Jing-rui, MEN Xiu-li, WU Jing

(Department of Pathophysiology, School of Basic Medical Sciences, North China University of Scienceand Technology, Hebei Key Laboratory for Chronic Diseases, Tangshan Key Laboratory for Preclinicaland Basic Research on Chronic Diseases, Tangshan 063000, China. E-mail: jingwu26@126.com)

?

·實(shí)驗(yàn)技術(shù)·

Evaluation and integration of normalization approaches and internal reference genes in real-time quantitative reverse transcription PCR in liver tissues of db/db mice*

LI Kai-ji, TIAN Zeng-you, TIAN Jing-rui, MEN Xiu-li, WU Jing△

(DepartmentofPathophysiology,SchoolofBasicMedicalSciences,NorthChinaUniversityofScienceandTechnology,HebeiKeyLaboratoryforChronicDiseases,TangshanKeyLaboratoryforPreclinicalandBasicResearchonChronicDiseases,Tangshan063000,China.E-mail:jingwu26@126.com)

AIM: Normalizing the results of real-time quantitative reverse transcription polymerase chain reaction (RT-qPCR) is essential for the accuracy of analysis. Commonly used approaches include input nucleic acid standardization (ΔCt method), normalization against a single internal reference gene (ΔΔCt method), and geometric averaging of multiple reference gene abundance using statistical software. We evaluated these approaches in the liver ofdb/dbmice, a typical model of fatty liver disease. METHODS: Seven reference genes, β-actin (ACTB), eukaryotic initiation factor (eIF) 5, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), hydroxymethylbilane synthase (HMBS), hypoxanthine-guanine phosphoribosyltransferase (HPRT) 1, polymerase (RNA) II (DNA directed) polypeptide A (Polr2A) and ribosomal protein P0 (RPLP0), were evaluated using software of geNorm and NormFinder. Hepatic lipogenesis genes, such as thyroid hormone-responsive protein (Thrsp), stearoyl-CoA desaturase (SCD) 1, sterol regulatory element-binding protein (SREBP) 1c and fatty acid synthase (FAS), were used as target genes of interest. RESULTS: The expression levels of all target genes and GAPDH were significantly elevated (P<0．05) indb/dbmouse livers by the ΔCt method. ACTB and HMBS were the most stable genes calculated by the software of geNorm. NormFinder analysis indicated that ACTB was the most stable gene, and the best combination of 2 genes was GAPDH and RPLP0. Normalization against a single internal reference gene of ACTB or RPLP0, the geometric mean of ACTB and HMBS, or GAPDH and RPLP0 showed similar results that the expression levels of Thrsp, SCD1 and FAS, but not SREBP1c increased (P<0．05) in the liver ofdb/dbmice. CONCLUSION: The ΔCt approach ensures a meaningful and biologically significant appraisal of gene expression. Use of the software like geNorm or NormFinder should be integrated with ΔCt method.

Real-time quantitative reverse transcription PCR; Normalization approaches; Internal reference genes;db/dbmice

The messenger RNA (mRNA) molecule as the link between DNA and protein is of central interest in bioscience and medicine. The profiling of mRNA transcription has become a popular research field in recent years[1]. Real-time quantitative reverse transcription polymerase chain reaction (RT-qPCR) is a sensitive and target-specific technique that has been widely used for the quantification of mRNA transcripts. One of the critical challenges of RT-qPCR is the availability of appropriate normalization approaches, which are commonly used as: (1) ΔCt, which standardizes against the input amount of nucleic acids, usually total RNA; (2) ΔΔCt, which normalizes the target gene transcript abundance against a single internal reference gene whose expression level keeps constant; and (3) geometric averaging of the abundance of multiple reference gene transcripts chosen using geNorm (or similar) software[2-3].

Normalizing target gene transcript abundance against the input of total RNA is simple, but is difficult to entirely remove the variation introduced by imbalance between mRNA and rRNA[4-5], or the difference in reverse transcriptase efficiency[3]. To date, internal reference genes are most frequently used to normalize the mRNA fraction, such as glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and β-actin (ACTB)[1]. However, the literatures show that the mRNA levels of certain internal reference genes are not constant[6]. Presently, an increasing number of investigators validate multiple reference genes with software, such as GeNorm, Normfinder and BestKeeper, and the geometric mean of multiple selected reference genes is used as an accurate normalization factor. Usually these kinds of software indicate different “stable” reference genes. Integration of these different results needs comprehensive discussion and reflection.

In addition to GAPDH and ACTB, the other 5 candidate reference genes were evaluated in the present investigation, including eukaryotic initiation factor 5 (eIF5), hydroxymethylbilane synthase (HMBS), hypoxanthine-guanine phosphoribosyltransferase 1 (HPRT1), polymerase (RNA) II (DNA directed) polypeptide A (Polr2A), and ribosomal protein P0 (RPLP0). These genes can be roughly divided into 4 different groups: (1) structure-related gene: ACTB; (2) metabolism-related genes: GAPDH, HPRT1, and HMBS; (3) transcription and translation-related genes: Polr2A and eIF5; and (4) the gene which can not be clearly categorized into one of these groupings: RPLP0[1].

Nonalcoholic fatty liver disease (NAFLD) is a common component of metabolic syndrome, which has become a worldwide epidemic as a result of improved living conditions, excessive food intake, and sedentary lifestyles. Thedb/dbmice are a common model of murine NAFLD[7-8]. Lots of RT-qPCR have been or are performed in the liver of this animal model. In the present study, our aim is to compare the results of different normalization approaches in the livers ofdb/dbmice, and do integration. Hepatic lipogenesis genes, thyroid hormone-responsive protein (Thrsp), stearoyl-CoA desaturase 1 (SCD1), sterol regulatory element-binding protein 1c (SREBP1c) and fatty acid synthase (FAS), were selected as target genes of interest.

MATERIALS AND METHODS

1 Animals and treatment

Maledb/dbmice (8～12 weeks of age) and age- and sex-matcheddb/mmice on a C57BKS background (Jackson Laboratory) were kindly provided by Dr. Youfei Guan of Peking University Health Science Center.

2 RNA extraction and RT-qPCR

Total RNA from the mouse liver was isolated using TRIzol reagent (Invitrogen). RNA quality and quantity were determined by measuring the absorbance at the wavelength of 260 nm (A260)/A280ratio (Eppendorf). The RNA sample (5 μg of each) was reversely transcribed using oligo (dT) primer and M-MuLV reverse transcriptase (Fermentas), and subsequently diluted with ddH2O to 10 mg/L cDNA. Real-time quantitative PCR amplification mixture (20 μL) containing 40 ng of template cDNA, 2× SYBR Green I SuperMix (10 μL) (Transgen) and 200 nmol/L of each forward and reverse primers. The reactions were run on Mastercycler EP Realplex System (Eppendorf). The amplification program was as follows: an initial denaturation step of 5 min at 95 ℃, followed by 40 cycles of denaturation at 95 ℃ for 15 s, annealing at 58 ℃ for 15 s, and elongation at 72 ℃ for 15 s. The primer sequences were listed in Table 1.

3 Statistical analysis

All measurements were performed in duplicates. Data were expressed as mean±SEM, and analyzed by Student’st-test.P<0．05 was considered statistically significant.

RESULTS

1 The stability of candidate reference genes in the liver ofdb/dbmice using ΔCt method

The expression levels were determined as the number of cycles needed for the amplification to reach a fixed threshold in the exponential phase of the PCR,

Table 1. The primer pairs used for RT-qPCR.

F: forward; R: reverse.

which was called cycle threshold (Ct) value. Ct values of the reference genes from RT-qPCR with cDNA of the liver samples from thedb/dband the controldb/mmice were represented with box-and-whisker plots (Figure 1A). Each box corresponding to an individual reference gene indicated the 25% and 75% percentiles, whiskers represented the maximum and minimum values, and the median was depicted by the line across the box.

The relative expression levels of the 7 candidate reference genes were obtained by normalizing against total RNA input using the ΔCt method. The expression levels of GAPDH and Polr2A were significantly increased in the livers ofdb/dbmice (P<0.01), as compared withdb/mmice. The values of relative expression levels and standard errors showed that RPLP0 may be the most stable reference gene, which had not only relative expression level much closer to 1, but also smaller standard error (Figure 1B).

2 The stability analysis of gene expression indb/dbmice livers by software

The stability of the gene expression for the candidate reference genes was evaluated by the software of geNorm and NormFinder. According to the algorithms of geNorm[2], the ascending order on the stability of the reference genes indb/manddb/dbmouse livers was HPRT< eIF5 < Polr2A < GAPDH < RPLP0 < ACTB and HMBS. In another word, HPRT was the least stable one with highest average expression stability (M) value, while ACTB and HMBS were the most stable ones with least M values (Figure 2A). The pairwire variation (V) of ACTB and HMBS was 0.106, less than the cut-off value of 0.15, indicating that there was no real need to include a 3rd gene in the normalization factor (Figure 2B).

The results of NormFinder analysis indicated that ACTB was the most stable gene over the whole set of samples. The best combination of 2 genes was GAPDH and RPLP0 (Figure 2A). GAPDH expression was significantly increased in the livers ofdb/dbmice norma-lized with ΔCt method (P<0．01) (Figure 1B).

Figure 1.The expression levels of candidate reference genes in the livers ofdb/dbmice. A: Ct values were represented with box-and-whisker plots; B: the relative expression levels of the 7 candidate reference genes using ΔCt method. Mean±SEM.n=6.**P<0．01vsdb/m.

3 Comparison of normalization methods: the expression alterations of the genes involved in hepatic lipogenesis

All comparisons of normalization methods were performed using Thrsp, SCD1, SREBP1c, and FAS as target genes of interest, which were all involved in hepatic lipogenesis (Figure 3). Normalization against total RNA input using the ΔCt method showed that the transcript levels of Thrsp, SCD1, SREBP1c and FAS were all significantly higher in the livers ofdb/dbmice, as compared withdb/mmice (P<0．05). By normalization against ACTB or RPLP0, the hepatic expression levels of Thrsp, SCD1 and FAS, but not SREBP1c increased indb/dbmice (P<0．05). And only Thrsp expression was significantly elevated indb/dbmouse liver (P<0．05), when using GAPDH as the internal control, since GAPDH itself had enhanced expression in the animal model (P<0.01) (ΔCt method) (Figure 1B). Normalization against geometric mean of ACTB and HMBS, or GAPDH and RPLP0 showed similar results as that of ACTB. Normalization against geometric mean of RPLP0 and eIF5 showed similar result as that of total RNA input using the ΔCt method.

Figure 2.The expression stability analysis of the reference genes in the livers ofdb/dbmice. A: average expression stability (M) of the 7 candidate reference genes calculated by geNorm (hollow square), and the stability value calculated by NormFinder (solid triangle). Low M values correspond to high expression stability. B: pairwire variation (V) between 2 sequential normalization factors containing an increasing number of genes. 0．15 is a cut-off value.

DISCUSSION

Thedb/dbmice are perfect animal models of type 2 diabetes, obesity and NAFLD that have been widely used. In the present study, we evaluated 7 candidate reference genes and compared 3 widely used methods of RT-qPCR normalization, including: (1) standardizing the input amount of total RNA; (2) normalizing target gene transcript abundance against a single internal reference gene; and (3) geometric averaging of the abundance of 2 reference gene transcripts chosen using geNorm or NormFinder software[2]. In our study, we showed that quantifying and standardizing the input level of total RNA and using the ΔCt method produced the most robust and biologically meaningful assessment of gene expression for the genes involved in hepatic lipogenesis. Normalization against a single internal reference gene is the method being most frequently used. GAPDH and ACTB are 2 of the typical internal reference genes widely used. The expression levels of target genes normalized against ACTB were significantly elevated, similar as that of ΔCt method, except SREBP1c.

Figure 3. The expression levels of target genes involved in hepatic lipogenesis normalized with different kinds of methods. Mean±SEM.n=6.*P<0．05,**P<0．01vsdb/m.

However, the expression of GAPDH was significantly enhanced in the liver ofdb/dbmice normalized with ΔCt method. The expression levels of target genes normalized against GAPDH were very different from that of ACTB and ΔCt method. This result admonished that the stability of internal reference genes should be verified before use, or it will generate misleading outcomes.

For this reason, some researchers recently suggest employing the geometric average of multiple reference genes and assessing gene stability with the support of validated mathematical models[9]such as geNorm and NormFinder[2]. geNorm is a user-friendly Microsoft Excel application that evaluates the expression stability of internal control genes based on the principle that the expression ratio of 2 stable reference genes should be the same in all samples. Using the same data, geNorm and NormFinder produced similar, but not exactly the same result. In this study, the geNorm expression stability analysis indicated that ACTB and HMBS had the highest stability value, while NormFinder showed that ACTB was the best gene, and GAPDH and RPLP0 were the 2 genes of best combination. Though the expression of GAPDH was significantly enhanced in the liver ofdb/dbmice determined by ΔCt method, geometric average of GAPDH and RPLP0, or ACTB and HMBS produced the similar results as that of ACTB.

A correct prediction of the geNorm model is based on assumptions that at least 2 stable genes are present and the transcriptional levels of 2 most stable genes are more similar to one another than to the expression of any other paired genes. So in some cases, things may happen as following: the expression of target gene A and several internal reference genes B～F were significantly upregulated in the treatment group, and only one internal reference gene G was expressed stably (normalized by ΔCt method). However, the program geNorm appraised B and F as the most stable reference genes, and the target gene A appeared unchanged when data were normalized against the geometric average of B and F[3]. In this study, the geNorm expression stability analysis indicated that ACTB and HMBS had the highest stability value. However, the values of relative expression levels and standard errors showed that RPLP0 may be the most stable reference gene, which had not only relative expression level much closer to 1, but also smaller standard error. Meanwhile, the expression levels of target genes normalized against RPLP0 were similar as that of ACTB. The gene eIF5 was the only gene that had the trend to be downregulated in the liver ofdb/dbmice. The geometric mean of RPLP0 and eIF5 was the only factor we used in this study that made the expression of SREBP1c significantly elevated, like ΔCt method. It has been proved by several separated research groups that SREBP1c has significantly increased expression in the liver ofdb/dbmice, versus that ofdb/mmice as control[10-12]. It seems that, geNorm should not be relied upon alone to select suitable reference genes. The candidate reference genes should also be evaluated by the ΔCt method including total RNA quantification within the RT-qPCR protocol. Pre-experiment is required to evaluate the expression patterns of up to 7 or more candidate reference genes to identify at least 1 and preferably 2 or 3 genes that are truly stable under the experimental conditions employed.

In conclusion, our results suggest that the ΔCt approach to RT-qPCR normalization ensures a meaningful and biologically significant appraisal of gene expression, and furthermore eliminates the risks of false outcomes associated with inappropriate use of the relatively limited number of reference gene(s). The use of software like geNorm or NormFinder should be integrated with ΔCt method.

[1] Radonic A, Thulke S, Mackay IM, et al. Guideline to reference gene selection for quantitative real-time PCR[J]. Biochem Biophys Res Commun, 2004, 313(4):856-862.

[2] Vandesompele J, De Preter K, Pattyn F, et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes[J]. Genome Biol, 2002, 3(7):RESEARCH0034.

[3] De Santis C, Smith-Keune C, Jerry DR. Normalizing RT-qPCR data: are we getting the right answers? An appraisal of normalization approaches and internal reference genes from a case study in the finfishLatescalcarifer[J]. Mar Biotechnol (NY), 2011, 13(2):170-180.

[4] Solanas M, Moral R, Escrich E. Unsuitability of using ribosomal RNA as loading control for Northern blot analyses related to the imbalance between messenger and ribosomal RNA content in rat mammary tumors[J]. Anal Biochem, 2001, 288(1):99-102.

[5] Schmittgen TD, Zakrajsek BA. Effect of experimental treatment on housekeeping gene expression: validation by real-time, quantitative RT-PCR[J]. J Biochem Biophys Methods, 2000, 46(1-2):69-81.

[6] Bustin SA. Quantification of mRNA using real-time reverse transcription PCR (RT-PCR): trends and problems[J]. J Mol Endocrinol, 2002, 29(1):23-39.

[7] Wu J, Wang C, Li S, et al. Thyroid hormone-responsive SPOT 14 homolog promotes hepatic lipogenesis, and its expression is regulated by liver X receptor alpha through a sterol regulatory element-binding protein 1c-dependent mechanism in mice[J]. Hepatology, 2013, 58(2):617-628.

[8] Li J, Chi Y, Wang C, et al. Pancreatic-derived factor promotes lipogenesis in the mouse liver: role of the Forkhead box 1 signaling pathway[J]. Hepatology, 2011, 53(6): 1906-1916.

[9] Bustin SA, Benes V, Garson JA, et al. The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments[J]. Clin Chem, 2009, 55(4):611-622.

[10]Matsumoto E, Ishihara A, Tamai S, et al. Time of day and nutrients in feeding govern daily expression rhythms of the gene for sterol regulatory element-binding protein (SREBP)-1 in the mouse liver[J]. J Biol Chem, 2010, 285(43):33028-33036.

[11]Fukui K, Wada T, Kagawa S, et al. Impact of the liver-specific expression of SHIP2 (SH2-containing inositol 5’-phosphatase 2) on insulin signaling and glucose metabolism in mice[J]. Diabetes, 2005, 54(7):1958-1967.

[12]Ueki K, Kadowaki T, Kahn CR. Role of suppressors of cytokine signaling SOCS-1 and SOCS-3 in hepatic steatosis and the metabolic syndrome[J]. Hepatol Res, 2005, 33(2): 185-192.

(責(zé)任編輯： 盧 萍， 羅 森)

db/db小鼠肝臟中實(shí)時(shí)定量逆轉(zhuǎn)錄PCR內(nèi)參照基因及標(biāo)準(zhǔn)化方法的評(píng)估與整合

李開(kāi)濟(jì)， 田增有， 田景瑞， 門(mén)秀麗， 吳 靜△

(華北理工大學(xué)基礎(chǔ)醫(yī)學(xué)院病理生理學(xué)系，河北省慢性疾病重點(diǎn)實(shí)驗(yàn)室，唐山市慢性病臨床基礎(chǔ)研究重點(diǎn)實(shí)驗(yàn)室，河北 唐山 063000)

目的：實(shí)時(shí)定量逆轉(zhuǎn)錄PCR(RT-qPCR)結(jié)果的標(biāo)準(zhǔn)化對(duì)于保證最終結(jié)果的準(zhǔn)確性尤其重要。常用的標(biāo)準(zhǔn)化方法包括用加入的核酸量校正(ΔCt法)、用單個(gè)內(nèi)參照基因校正(ΔΔCt法)和用統(tǒng)計(jì)學(xué)軟件計(jì)算多個(gè)內(nèi)參照基因的幾何平均值進(jìn)行校正。我們?cè)赿b/db小鼠肝臟中對(duì)各種校正方法進(jìn)行評(píng)估。方法：用geNorm 和 NormFinder兩種軟件評(píng)估ACTB、eIF5、GAPDH、HMBS、HPRT1、Polr2A和RPLP0共7個(gè)內(nèi)參照基因，以肝臟脂質(zhì)合成相關(guān)基因Thrsp、SCD、SREBP1c和FAS作為目的基因。結(jié)果：應(yīng)用ΔCt法，db/db小鼠肝臟中所有目的基因及GAPDH的表達(dá)顯著升高(P<0.05)。geNorm計(jì)算認(rèn)為ACTB和HMBS最穩(wěn)定。NormFinder計(jì)算認(rèn)為ACTB最穩(wěn)定，而GAPDH和RPLP0為最佳組合。以單個(gè)基因ACTB或RPLP0，以及ACTB與HMBS，或GAPDH與RPLP0的幾何平均值進(jìn)行校正，db/db小鼠肝臟中除SREBP1c以外，Thrsp、SCD1和FAS的表達(dá)均升高(P<0.05)。結(jié)論：用ΔCt法校正RT-qPCR的結(jié)果穩(wěn)定并具有生物學(xué)意義。統(tǒng)計(jì)學(xué)軟件geNorm或NormFinder應(yīng)與ΔCt法整合使用。

實(shí)時(shí)定量逆轉(zhuǎn)錄PCR； 標(biāo)準(zhǔn)化方法； 內(nèi)參照基因；db/db小鼠

R-331; R34

A

1000- 4718(2016)11- 2106- 07

R-331; R34 [Document code] A

10.3969/j.issn.1000- 4718.2016.11.033

[Received date] 2016- 05- 06 [Accepted date] 2016- 09- 21

*[Foundation item] Supported by National Natural Science Foundation of China (No. 81370477; No. 81370918); Science and Technology Research Outstanding Youth Fund of Hebei College (No. Y2011117); Undergraduate Innovation Project (No. 201410081061; No. X2014026).

△Corresponding author Tel: 0315-3725612; E-mail: jingwu26@126.com

雜志網(wǎng)址： http://www.cjpp.net