Reference genes (RG) as sample internal controls for gene transcript level analyses by quantitative RT-PCR (RT-qPCR) must be stably expressed within the experimental range. A variety of in vitro cell culture settings with primary human hepatocytes, and Huh-7 and HepG2 cell lines, were used to determine candidate RG expression stability in RT-qPCR analyses. Employing GeNorm, BestKeeper and Normfinder algorithms, this study identifies PSMB6, MDH1 and some more RG as sufficiently unregulated, thus expressed at stable levels, in hepatocyte-like cells in vitro. Inclusion of multiple RG, quenching occasional regulations of single RG, greatly stabilises gene expression level calculations from RT-qPCR data. To further enhance validity and reproducibility of relative RT-qPCR quantifications, the ΔCT calculation can be extended (e-ΔCT) by replacing the CT of a single RG in ΔCT with an averaged CT-value from multiple RG. The use of two or three RG - here identified suited for human hepatocyte-like cells - for normalisation with the straightforward e-ΔCT calculation, should improve reproducibility and robustness of comparative RT-qPCR-based gene expression analyses.
Citation: Riedel G, Rüdrich U, Fekete-Drimusz N, Manns MP, Vondran FWR, Bock M (2014) An Extended ΔCT-Method Facilitating Normalisation with Multiple Reference Genes Suited for Quantitative RT-PCR Analyses of Human Hepatocyte-Like Cells. PLoS ONE 9(3): e93031. https://doi.org/10.1371/journal.pone.0093031
Editor: Birke Bartosch, Inserm, U1052, UMR 5286, France
Received: December 16, 2013; Accepted: February 27, 2014; Published: March 21, 2014
Copyright: © 2014 Riedel et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded by the REBIRTH-Cluster of Excellence (Deutsche Forschungsgemeinschaft), the German Centre for Infection Research (DZIF, TTU Hepatitis) and the Else Kröner-Fresenius-Stiftung (FWRV; 2010_A49). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The quantification of mRNA expression levels by reverse transcription quantitative PCR (RT-qPCR) is - if applied appropriately  - a powerful and straightforward tool. Most RT-qPCR analyses employ reference genes (RG), i.e. genes that are stably expressed within the experimental range, for normalisation purposes. Whereas for other tissues appropriate RG have already been verified –, to our knowledge no in-depth analysis has been performed for human hepatocyte-like cells in vitro so far.
A literature survey covering eight volumes of the journal HEPATOLOGY (years 2009–2012) for RT-qPCR-analyses of human hepatocyte-like cells – i.e. primary hepatocytes and hepatic cell lines - identified GAPDH as the most commonly used RG (in 79 of 192 papers, i.e. 41%), followed by ACTB (46/192, 24%). Also, of the 192 RT-qPCR-papers found, 91 analysed primary liver material from patients and 131 studies included hepatic cell lines, wherein HepG2 (59%) and Huh-7 (47%) cell lines were by far the most commonly used ones.
Hepatocytes and hepatoma-derived cell lines are used for a plethora of in vitro investigations mainly in the areas of basic, preclinical and pharmacological research - including metabolic assays, drug- and drug-toxicity testings, accompanied by optimisations of culture conditions towards enhanced cell maturity. Also, studies with infectious human hepatotropic agents are of major importance and more recently, in vitro gene therapeutic approaches using primary hepatocytes - requiring the issues of rapid dedifferentiation and almost complete absence of proliferation of primary hepatocytes in tissue culture to be overcome, possibly by the culture of proliferating hepatic stem cells and their subsequent maturation in the future.
Investigating the expression stability of 22 widely used RG in experimental in vitro settings with commonly used hepatocyte-like cell types, we identified in this study several reference genes suited for RT-qPCR-analyses of hepatocyte-like cells. Furthermore we were able to derive from the ΔCT-method a simple calculation allowing for the inclusion of multiple RG to significantly strenghten data normalisation in comparative, not absolute, gene expression analyses - without the need for laborious generation of qPCR-standard curves.
Materials and Methods
A more detailed description of materials and methods used can be found in File S1.
Briefly, the expression stability of 22 widely used RG (Table S2 in File S1) was investigated across a total of 32 experimental settings with hepatocyte-like cell types, including freshly isolated primary human hepatocytes  at defined time points in cell culture (subgroup “primary hepatocytes”, PH), and HepG2 and Huh-7.5 cells treated with Chloroquine, Actinomycin D (ActD) , Trichostatin A  and DMSO - commonly used drugs with significantly differing effects in tissue culture - for different durations without passaging (subgroup “drug and density”, DD), or cultured for 14 days under a variety of conditions altering cell maturity status (subgroup “culture conditions”, CC) ,  (all experimental settings listed in Table S1 in File S1). After RNA isolation and RT-qPCR, individual data sets of the samples, each containing Cycle Threshold (CT) values for all reference genes (primer details in Table S2 in File S1) and some exemplary genes of interest (target genes, TG; Table S3 in File S1) were further analysed in silico.
Results and Discussion
The individual RG rankings (Fig. 1A (AS) and Fig. 2 (PH, DD, CC)), the analyses generated by the software algorithms (Fig. 1B, Fig. S1 in File S1), and the cumulative rankings should only serve as estimates: Individual softwares qualitatively differ to various extents - geNorm and Normfinder are more similar to each other than to Bestkeeper - hindering more exact combined ranking. Also, in the “all sample” group (AS), the three subgroups are not equally represented, in particular PH contributing only 12 data sets to a total of 96. However, from a general and overall point of view, a strikingly clear picture arises, even when taking into account that the PH rankings differ significantly from the cell line based analyses (DD and CC): PSMB6 and MDH1 could be identified as stable RG for all hepatocyte-like cells tested and some more genes (especially ACTB, PPIA, HDDC2) consistently rank well in the subgroups, offering the option to use multiple RG for normalisation. Noteworthy, albeit not included in the systematic analyses presented here, PSMB1 and MDH1 are also almost entirely unregulated in the human liver progenitor cell line HepaRG  under a variety of experimental conditions investigating metabolic activity or differentiation/maturation potential (data not shown).
(A) Overall ranking of 22 RG using geNorm, Bestkeeper and Normfinder algorithms and calculation of the cumulative ranking (column 4). (B) geNorm derived M-values as measures for the average pairwise expression stabilities within AS (column 1), and in the subgroups DD (2), CC (3) and PH (4). (C) Box-whisker plots of all CT-values of reference genes and target genes examined. Median (central horizontal line), the 25th and 75th quartile (boxes) and whiskers for the total CT-range are shown.
(A) PH, primary hepatocytes, (B) DD, drug and density and (C) CC, culture conditions (also see Figure 1A, all samples).
All RG identified as suitable are measured in a CT-range above 20 (Fig. 1C), thus are reasonably close to the ranges of TG CT-values - a prerequisite for accurate normalisation, mainly to cover the possibility of exhaustive processes in the course of thermal cycling.
In many RT-qPCR applications relative changes of TG expression levels between experimental settings - and not absolute molecule numbers - are of interest. For such relative comparisons, the ΔCT method is most straightforward and often used , . So far, the ΔCT-method does not allow for the inclusion of multiple RG. Thus we extended the ΔCT calculation (e-ΔCT) to permit the use of multiple RG (Fig. 3A): Since fold changes of expression levels are calculated as ratios of two experimental settings (Fig. 3B), the absolute CT-values in the two ΔCT-calculations of the ratio are cancelled. Only CT-differences between numerator and denominator in the fraction render essential.
(A) Calculation of e-ΔCT, an extension of the ΔCT-method by first calculating the mean CT-value of multiple RG included for each sample. (B) Depiction of data processing after geNorm-/e-ΔCT-calculations of sample-specific G-values to calculate fold changes and their significance (p-values) between two experimental settings, each consisting of three fully independent samples. (C) Example (Huh-7, without vs. with ActD) comparing fold changes in gene expression levels calculated by geNorm (grey) and e-ΔCT (black) using: GAPDH, the most common RG, the AS best ranking PSMB6, MDH1 and ACTB, and inclusion of two (PSMB6 and MDH1; P_M) or three (PSMB6, MDH1 and ACTB; P_M_A) RG. (D) Examples (primary hepatocytes, 0 vs. 24 hrs) comparing use of most stable RG of the subgroup (RG group, i.e. PH) vs. overall RG (RG all, i.e. AS) via e-ΔCT. *: p-values <0.05.
Raw multiple RG CT-values obtained from one cDNA-sample can safely be regarded as being independent from each other with respect to choosing the method of calculating a mean CT-value. Thus, the arithmetic mean of the CTs of multiple RG is applied in e-ΔCT for variability-quenching and stabilisation purposes - with the advantages of using multiple RG discussed elsewhere  and also demonstrated by some of our examples: Even well-suited RG are regulated occasionally - thus not entirely stable - leading to misestimates of TG expression (Fig. 3D, Table S4 in File S1). For instance in PH, PSMB6 and CYC1 (Fig. 3D), both with similar M-values (Fig. 1B), lead to different results for the fold change - hence at least one has to be slightly regulated in primary hepatocytes within the first 24 hours of culturing. Further, regulation of ACTB, with a mediocre ranking in PH (Fig. 2) and obviously regulated, is sufficiently quenched by the other two RG (Fig. 3D, grey bars: ACTB vs. P_M_A). Thus, inclusion of multiple RG, which ideally serve independent cellular functions and therefore are least likely of being co-regulated in the same experimental systems, significantly buffers the effects of RG regulations.
In Fig. 3C (and Table S4 in File S1), e-ΔCT-derived expression level changes are compared with geNorm-derived values. geNorm, so far the only widely used method using multiple RG, estimates the absolute molecule numbers of a TG in a sample, but requires the availability of standard curve data that have to be established for each primer pair-template combination (with no fully defined method for generating standard curves being available , , but several methodical variants, each with its own accuracy-related shortcomings). ΔCT, not allowing determination of absolute mRNA molecule numbers, only requires the primer pairs chosen to be reasonably efficient - which can be verified quite easily . However, although generated by significantly different underlying formulas, the outcomes of e-ΔCT and geNorm - using one, two or three RG - never differ by more than 30%, with similar significance measures (p-values) (Fig. 3C, Table S4 in File S1).
Whereas PSMB1 and MDH1, ranking first in AS, appear suited also in all subgroups tested, sole use of ACTB would lead to misestimates, in particular in PH (Fig. 3D, Table S4 in File S1). However, we arbitrarily preferred ACTB to PPIA (same ranking in AS, Fig. 1A) as a third RG, for use together with PSMB6 and MDH1: Inclusion of ACTB - its gene product being widely used as standard in other, non PCR-based assays - would allow to cross-compare qPCR findings with other assays where β-Actin is used for calibration.
Having analysed a reasonably wide spectrum of hepatic cell types and experimental conditions, this study aims to provisionally recommend a set of RG to be used as a routine and default choice for RT-qPCR analyses with human hepatocyte-like cells in general. However, caution is needed and RG stability should be verified  specifically for each experimental setting: Certainly many experiments exist that require a separate and specific search for appropriate RGs.
In conclusion, RG suited for RT-qPCR-analyses of hepatocyte-like cells in vitro could be identified. Two or three RG used with the straightforward e-ΔCT calculation can greatly improve reproducibility and robustness of relative gene expression data generated by RT-qPCR.
The Huh-7.5 cell line was kindly provided by Thomas Pietschmann, Twincore, Hannover, Germany. The authors thank the Twincore Centre for Experimental and Clinical Infection Research, Hannover, for essential support of qPCR measurements.
Conceived and designed the experiments: GR UR NFD MPM FWRV MB. Performed the experiments: GR UR NFD FWRV. Analyzed the data: GR UR NFD MPM MB. Contributed reagents/materials/analysis tools: MPM FWRV. Wrote the paper: GR UR NFD MPM FWRV MB.
- 1. Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, et al. (2009) The MIQE guidelines: Minimum information for publication of quantitative real-time PCR experiments. Clin Chem 55: 611–622.
- 2. Wang Q, Ishikawa T, Michiue T, Zhu BL, Guan DW, et al. (2012) Stability of endogenous reference genes in postmortem human brains for normalization of quantitative real-time PCR data: Comprehensive evaluation using geNorm, NormFinder, and BestKeeper. Int J Legal Med 126: 943–952.
- 3. Stern-Straeter J, Bonaterra GA, Hormann K, Kinscherf R, Goessler UR (2009) Identification of valid reference genes during the differentiation of human myoblasts. BMC Mol Biol 10: 66–2199-10-66.
- 4. Andersen CL, Jensen JL, Orntoft TF (2004) 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.
- 5. Vondran FW, Katenz E, Schwartlander R, Morgul MH, Raschzok N, et al. (2008) Isolation of primary human hepatocytes after partial hepatectomy: Criteria for identification of the most promising liver specimen. Artif Organs 32: 205–213.
- 6. Tan SH, Shui G, Zhou J, Li JJ, Bay BH, et al. (2012) Induction of autophagy by palmitic acid via protein kinase C-mediated signaling pathway independent of mTOR (mammalian target of rapamycin). J Biol Chem 287: 14364–14376.
- 7. Yamashita Y, Shimada M, Harimoto N, Rikimaru T, Shirabe K, et al. (2003) Histone deacetylase inhibitor trichostatin A induces cell-cycle arrest/apoptosis and hepatocyte differentiation in human hepatoma cells. Int J Cancer 103: 572–576.
- 8. Koyama T, Ehashi T, Ohshima N, Miyoshi H (2009) Efficient proliferation and maturation of fetal liver cells in three-dimensional culture by stimulation of oncostatin M, epidermal growth factor, and dimethyl sulfoxide. Tissue Eng Part A 15: 1099–1107.
- 9. Tostoes RM, Leite SB, Serra M, Jensen J, Bjorquist P, et al. (2012) Human liver cell spheroids in extended perfusion bioreactor culture for repeated-dose drug testing. Hepatology 55: 1227–1236.
- 10. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, et al. (2002) Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol 3: RESEARCH0034.
- 11. Pfaffl MW, Tichopad A, Prgomet C, Neuvians TP (2004) Determination of stable housekeeping genes, differentially regulated target genes and sample integrity: BestKeeper—excel-based tool using pair-wise correlations. Biotechnol Lett 26: 509–515.
- 12. Cerec V, Glaise D, Garnier D, Morosan S, Turlin B, et al. (2007) Transdifferentiation of hepatocyte-like cells from the human hepatoma HepaRG cell line through bipotent progenitor. Hepatology 45: 957–967.
- 13. de Jonge HJ, Fehrmann RS, de Bont ES, Hofstra RM, Gerbens F, et al. (2007) Evidence based selection of housekeeping genes. PLoS One 2: e898.
- 14. Barber RD, Harmer DW, Coleman RA, Clark BJ (2005) GAPDH as a housekeeping gene: Analysis of GAPDH mRNA expression in a panel of 72 human tissues. Physiol Genomics 21: 389–395.
- 15. Rubie C, Kempf K, Hans J, Su T, Tilton B, et al. (2005) Housekeeping gene variability in normal and cancerous colorectal, pancreatic, esophageal, gastric and hepatic tissues. Mol Cell Probes 19: 101–109.
- 16. Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data using real-time quantitative PCR and the 2(-delta delta C(T)) method. Methods 25: 402–408.
- 17. Schmittgen TD, Livak KJ (2008) Analyzing real-time PCR data by the comparative C(T) method. Nat Protoc 3: 1101–1108.
- 18. Yang X, Hatfield JT, Hinze SJ, Mu X, Anderson PJ, et al. (2012) Bone to pick: The importance of evaluating reference genes for RT-qPCR quantification of gene expression in craniosynostosis and bone-related tissues and cells. BMC Res Notes 5: 222–0500-5-222.
- 19. Sivaganesan M, Seifring S, Varma M, Haugland RA, Shanks OC (2008) A bayesian method for calculating real-time quantitative PCR calibration curves using absolute plasmid DNA standards. BMC Bioinformatics 9: 120–2105-9-120.
- 20. Rutledge RG, Cote C (2003) Mathematics of quantitative kinetic PCR and the application of standard curves. Nucleic Acids Res 31: e93.