Limits...
The dilution effect and the importance of selecting the right internal control genes for RT-qPCR: a paradigmatic approach in fetal sheep.

Xu H, Bionaz M, Sloboda DM, Ehrlich L, Li S, Newnham JP, Dudenhausen JW, Henrich W, Plagemann A, Challis JR, Braun T - BMC Res Notes (2015)

Bottom Line: Our interest in prenatal glucocorticoid (GC) effects on fetal growth has resulted in our investigation of suitable ICGs relevant in this model.In order to account for confounding effects on the geNorm analysis due to co-regulation among ICGs tested, network analysis was performed using Ingenuity Pathway Analysis software.Raw RT-qPCR data of all the tested ICGs was significantly reduced across gestation.

View Article: PubMed Central - PubMed

Affiliation: Departments of Obstetrics and Division of Experimental Obstetrics, Charité - University Berlin, Augustenburger Platz 1, Berlin, Germany. xuhuaisheng@gmail.com.

ABSTRACT

Background: The key to understanding changes in gene expression levels using reverse transcription real-time quantitative polymerase chain reaction (RT-qPCR) relies on the ability to rationalize the technique using internal control genes (ICGs). However, the use of ICGs has become increasingly problematic given that any genes, including housekeeping genes, thought to be stable across different tissue types, ages and treatment protocols, can be regulated at transcriptomic level. Our interest in prenatal glucocorticoid (GC) effects on fetal growth has resulted in our investigation of suitable ICGs relevant in this model. The usefulness of RNA18S, ACTB, HPRT1, RPLP0, PPIA and TUBB as ICGs was analyzed according to effects of early dexamethasone (DEX) treatment, gender, and gestational age by two approaches: (1) the classical approach where raw (i.e., not normalized) RT-qPCR data of tested ICGs were statistically analyzed and the best ICG selected based on absence of any significant effect; (2) used of published algorithms. For the latter the geNorm Visual Basic application was mainly used, but data were also analyzed by Normfinder and Bestkeeper. In order to account for confounding effects on the geNorm analysis due to co-regulation among ICGs tested, network analysis was performed using Ingenuity Pathway Analysis software. The expression of RNA18S, the most abundant transcript, and correlation of ICGs with RNA18S, total RNA, and liver-specific genes were also performed to assess potential dilution effect of raw RT-qPCR data. The effect of the two approaches used to select the best ICG(s) was compared by normalization of NR3C1 (glucocorticoid receptor) mRNA expression, as an example for a target gene.

Results: Raw RT-qPCR data of all the tested ICGs was significantly reduced across gestation. TUBB was the only ICG that was affected by DEX treatment. Using approach (1) all tested ICGs would have been rejected because they would initially appear as not reliable for normalization. However, geNorm analysis (approach 2) of the ICGs indicated that the geometrical mean of PPIA, HPRT1, RNA18S and RPLPO can be considered a reliable approach for normalization of target genes in both control and DEX treated groups. Different subset of ICGs were tested for normalization of NR3C1 expression and, despite the overall pattern of the mean was not extremely different, the statistical analysis uncovered a significant influence of the use of different normalization approaches on the expression of the target gene. We observed a decrease of total RNA through gestation, a lower decrease in raw RT-qPCR data of the two rRNA measured compared to ICGs, and a positive correlation between raw RT-qPCR data of ICGs and total RNA. Based on the same amount of total RNA to performed RT-qPCR analysis, those data indicated that other mRNA might have had a large increase in expression and, as consequence, had artificially diluted the stably expressed genes, such as ICGs. This point was demonstrated by a significant negative correlation of raw RT-qPCR data between ICGs and liver-specific genes.

Conclusion: The study confirmed the necessity of assessing multiple ICGs using algorithms in order to obtain a reliable normalization of RT-qPCR data. Our data indicated that the use of the geometrical mean of PPIA, HPRT1, RNA18S and RPLPO can provide a reliable normalization for the proposed study. Furthermore, the dilution effect observed support the unreliability of the classical approach to test ICGs. Finally, the observed change in the composition of RNA species through time reveals the limitation of the use of ICGs to normalize RT-qPCR data, especially if absolute quantification is required.

Show MeSH

Related in: MedlinePlus

A-D: Comparison of a target gene mRNA expression normalized with different subsets of internal control genes (ICGs).NR3C1 raw RT-qPCR data were normalized to different subsets of ICGs: (A) raw RT-qPCR data geometrical mean of HPRT1, PPIA, RNA18S, and RPLPO, the most reliable normalization factor as uncovered by the use of geNorm; (B) “flat gene” = RNA18S and (C) “flat gene” = HPRT1; and (D)ACTB as the least reliable ICGs as uncovered by using geNorm (besides being among the most popular used ICG). Data were analyzed by MANOVA with treatment, gender and day of gestation as main effect, followed by a pairwise comparison (Holm’s Sidak) when main effects were p < 0.05. Different letters indicate significant differences in day of gestation; stars indicate significant differences in DEX treatment. The final data were obtained by rescaled normalized expression: Qnormalized/rescaled = (Qsample/NFsample)/Min (Qsample/NFsample) (geNorm v3.5 manual [51]).
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4352295&req=5

Fig4: A-D: Comparison of a target gene mRNA expression normalized with different subsets of internal control genes (ICGs).NR3C1 raw RT-qPCR data were normalized to different subsets of ICGs: (A) raw RT-qPCR data geometrical mean of HPRT1, PPIA, RNA18S, and RPLPO, the most reliable normalization factor as uncovered by the use of geNorm; (B) “flat gene” = RNA18S and (C) “flat gene” = HPRT1; and (D)ACTB as the least reliable ICGs as uncovered by using geNorm (besides being among the most popular used ICG). Data were analyzed by MANOVA with treatment, gender and day of gestation as main effect, followed by a pairwise comparison (Holm’s Sidak) when main effects were p < 0.05. Different letters indicate significant differences in day of gestation; stars indicate significant differences in DEX treatment. The final data were obtained by rescaled normalized expression: Qnormalized/rescaled = (Qsample/NFsample)/Min (Qsample/NFsample) (geNorm v3.5 manual [51]).

Mentions: We have compared the results of normalizing RT-qPCR data of glucocorticoid receptor (NR3C1), used as target gene, using the geometrical mean of the 4 best ICGs as indicated by approach 2 (Figure 4A), the two best ICGs (or the more “flat” ICGs HPRT and RNA18S) indicated by the approach 1 (Figure 4B and 4C) and ACTB, the ICG with the lowest average expression stability among the one tested but also one of the most used ICGs in literature (Figure 4D). The statistical differences observed between comparisons on the quantity of NR3C1 mRNA expression levels obviously differed between subsets of ICGs used for normalization. For example, NR3C1 mRNA expression significantly increased between 50 and 100dG and further increased between 100 and 140dG when normalized by the geometrical mean of HPRT1, PPIA, RNA18S and RPLPO (Figure 4A). However, when normalized only to RNA18S, the increase between 100 and 140dG was not significant (Figure 4B). Normalizing to ACTB resulted in a significant increase of NR3C1 mRNA expression between 50 and 125dG (Figure 4D). Normalizing to HPRT1, the ICG with the least time effect on the raw RT-qPCR data resulted in a significant reduction of the quantity of NR3C1 mRNA expression levels at 140dG in DEX compared to controls (Figure 4C), which was not significant when normalized to the other subsets of ICGs (Figure 4A, B, D).Figure 4


The dilution effect and the importance of selecting the right internal control genes for RT-qPCR: a paradigmatic approach in fetal sheep.

Xu H, Bionaz M, Sloboda DM, Ehrlich L, Li S, Newnham JP, Dudenhausen JW, Henrich W, Plagemann A, Challis JR, Braun T - BMC Res Notes (2015)

A-D: Comparison of a target gene mRNA expression normalized with different subsets of internal control genes (ICGs).NR3C1 raw RT-qPCR data were normalized to different subsets of ICGs: (A) raw RT-qPCR data geometrical mean of HPRT1, PPIA, RNA18S, and RPLPO, the most reliable normalization factor as uncovered by the use of geNorm; (B) “flat gene” = RNA18S and (C) “flat gene” = HPRT1; and (D)ACTB as the least reliable ICGs as uncovered by using geNorm (besides being among the most popular used ICG). Data were analyzed by MANOVA with treatment, gender and day of gestation as main effect, followed by a pairwise comparison (Holm’s Sidak) when main effects were p < 0.05. Different letters indicate significant differences in day of gestation; stars indicate significant differences in DEX treatment. The final data were obtained by rescaled normalized expression: Qnormalized/rescaled = (Qsample/NFsample)/Min (Qsample/NFsample) (geNorm v3.5 manual [51]).
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4352295&req=5

Fig4: A-D: Comparison of a target gene mRNA expression normalized with different subsets of internal control genes (ICGs).NR3C1 raw RT-qPCR data were normalized to different subsets of ICGs: (A) raw RT-qPCR data geometrical mean of HPRT1, PPIA, RNA18S, and RPLPO, the most reliable normalization factor as uncovered by the use of geNorm; (B) “flat gene” = RNA18S and (C) “flat gene” = HPRT1; and (D)ACTB as the least reliable ICGs as uncovered by using geNorm (besides being among the most popular used ICG). Data were analyzed by MANOVA with treatment, gender and day of gestation as main effect, followed by a pairwise comparison (Holm’s Sidak) when main effects were p < 0.05. Different letters indicate significant differences in day of gestation; stars indicate significant differences in DEX treatment. The final data were obtained by rescaled normalized expression: Qnormalized/rescaled = (Qsample/NFsample)/Min (Qsample/NFsample) (geNorm v3.5 manual [51]).
Mentions: We have compared the results of normalizing RT-qPCR data of glucocorticoid receptor (NR3C1), used as target gene, using the geometrical mean of the 4 best ICGs as indicated by approach 2 (Figure 4A), the two best ICGs (or the more “flat” ICGs HPRT and RNA18S) indicated by the approach 1 (Figure 4B and 4C) and ACTB, the ICG with the lowest average expression stability among the one tested but also one of the most used ICGs in literature (Figure 4D). The statistical differences observed between comparisons on the quantity of NR3C1 mRNA expression levels obviously differed between subsets of ICGs used for normalization. For example, NR3C1 mRNA expression significantly increased between 50 and 100dG and further increased between 100 and 140dG when normalized by the geometrical mean of HPRT1, PPIA, RNA18S and RPLPO (Figure 4A). However, when normalized only to RNA18S, the increase between 100 and 140dG was not significant (Figure 4B). Normalizing to ACTB resulted in a significant increase of NR3C1 mRNA expression between 50 and 125dG (Figure 4D). Normalizing to HPRT1, the ICG with the least time effect on the raw RT-qPCR data resulted in a significant reduction of the quantity of NR3C1 mRNA expression levels at 140dG in DEX compared to controls (Figure 4C), which was not significant when normalized to the other subsets of ICGs (Figure 4A, B, D).Figure 4

Bottom Line: Our interest in prenatal glucocorticoid (GC) effects on fetal growth has resulted in our investigation of suitable ICGs relevant in this model.In order to account for confounding effects on the geNorm analysis due to co-regulation among ICGs tested, network analysis was performed using Ingenuity Pathway Analysis software.Raw RT-qPCR data of all the tested ICGs was significantly reduced across gestation.

View Article: PubMed Central - PubMed

Affiliation: Departments of Obstetrics and Division of Experimental Obstetrics, Charité - University Berlin, Augustenburger Platz 1, Berlin, Germany. xuhuaisheng@gmail.com.

ABSTRACT

Background: The key to understanding changes in gene expression levels using reverse transcription real-time quantitative polymerase chain reaction (RT-qPCR) relies on the ability to rationalize the technique using internal control genes (ICGs). However, the use of ICGs has become increasingly problematic given that any genes, including housekeeping genes, thought to be stable across different tissue types, ages and treatment protocols, can be regulated at transcriptomic level. Our interest in prenatal glucocorticoid (GC) effects on fetal growth has resulted in our investigation of suitable ICGs relevant in this model. The usefulness of RNA18S, ACTB, HPRT1, RPLP0, PPIA and TUBB as ICGs was analyzed according to effects of early dexamethasone (DEX) treatment, gender, and gestational age by two approaches: (1) the classical approach where raw (i.e., not normalized) RT-qPCR data of tested ICGs were statistically analyzed and the best ICG selected based on absence of any significant effect; (2) used of published algorithms. For the latter the geNorm Visual Basic application was mainly used, but data were also analyzed by Normfinder and Bestkeeper. In order to account for confounding effects on the geNorm analysis due to co-regulation among ICGs tested, network analysis was performed using Ingenuity Pathway Analysis software. The expression of RNA18S, the most abundant transcript, and correlation of ICGs with RNA18S, total RNA, and liver-specific genes were also performed to assess potential dilution effect of raw RT-qPCR data. The effect of the two approaches used to select the best ICG(s) was compared by normalization of NR3C1 (glucocorticoid receptor) mRNA expression, as an example for a target gene.

Results: Raw RT-qPCR data of all the tested ICGs was significantly reduced across gestation. TUBB was the only ICG that was affected by DEX treatment. Using approach (1) all tested ICGs would have been rejected because they would initially appear as not reliable for normalization. However, geNorm analysis (approach 2) of the ICGs indicated that the geometrical mean of PPIA, HPRT1, RNA18S and RPLPO can be considered a reliable approach for normalization of target genes in both control and DEX treated groups. Different subset of ICGs were tested for normalization of NR3C1 expression and, despite the overall pattern of the mean was not extremely different, the statistical analysis uncovered a significant influence of the use of different normalization approaches on the expression of the target gene. We observed a decrease of total RNA through gestation, a lower decrease in raw RT-qPCR data of the two rRNA measured compared to ICGs, and a positive correlation between raw RT-qPCR data of ICGs and total RNA. Based on the same amount of total RNA to performed RT-qPCR analysis, those data indicated that other mRNA might have had a large increase in expression and, as consequence, had artificially diluted the stably expressed genes, such as ICGs. This point was demonstrated by a significant negative correlation of raw RT-qPCR data between ICGs and liver-specific genes.

Conclusion: The study confirmed the necessity of assessing multiple ICGs using algorithms in order to obtain a reliable normalization of RT-qPCR data. Our data indicated that the use of the geometrical mean of PPIA, HPRT1, RNA18S and RPLPO can provide a reliable normalization for the proposed study. Furthermore, the dilution effect observed support the unreliability of the classical approach to test ICGs. Finally, the observed change in the composition of RNA species through time reveals the limitation of the use of ICGs to normalize RT-qPCR data, especially if absolute quantification is required.

Show MeSH
Related in: MedlinePlus