Selection of housekeeping genes for gene expression studies in human reticulocytes using real-time PCR
© Silver et al; licensee BioMed Central Ltd. 2006
Received: 27 June 2006
Accepted: 06 October 2006
Published: 06 October 2006
Control genes, which are often referred to as housekeeping genes, are frequently used to normalise mRNA levels between different samples. However, the expression level of these genes may vary among tissues or cells and may change under certain circumstances. Thus, the selection of housekeeping genes is critical for gene expression studies. To address this issue, 7 candidate housekeeping genes including several commonly used ones were investigated in isolated human reticulocytes. For this, a simple ΔCt approach was employed by comparing relative expression of 'pairs of genes' within each sample. On this basis, stability of the candidate housekeeping genes was ranked according to repeatability of the gene expression differences among 31 samples.
Initial screening of the expression pattern demonstrated that 1 of the 7 genes was expressed at very low levels in reticulocytes and was excluded from further analysis. The range of expression stability of the other 6 genes was (from most stable to least stable): GAPDH (glyceraldehyde 3-phosphate dehydrogenase), SDHA (succinate dehydrogenase), HPRT1 (hypoxanthine phosphoribosyl transferase 1), HBS1L (HBS1-like protein) and AHSP (alpha haemoglobin stabilising protein), followed by B2M (beta-2-microglobulin).
Using this simple approach, GAPDH was found to be the most suitable housekeeping gene for expression studies in reticulocytes while the commonly used B2M should be avoided.
Reticulocytes are juvenile enucleated red cells produced during erythropoiesis and spend approximately 24 hours in the bone marrow before entering the peripheral circulation. Reticulocytes persist for a few days in the circulation before forming the slightly smaller, mature red blood cell at which time any residual RNA the reticulocytes still possessed, is lost . Peripheral blood reticulocytes are not only readily accessible, but remnants of RNA they still harbour are likely to represent gene expression profiles of patients. This makes them potentially very useful in looking for subtle changes in gene expression, particularly for quantitative traits involved in the phenotypic outcome of haemoglobinopathies. In order to reach the sensitivity needed to detect such subtle changes in gene expression, quantitative real time PCR (Q-RT-PCR) was employed.
When comparing gene expression in different samples it is crucial to consider experimental variations such as amount of starting material, RNA extraction and reverse transcription efficiencies. To account for these, accuracy of Q-RT-PCR relies on normalisation to an internal control, often referred to as a housekeeping gene [2, 3]
The prerequisite of a suitable housekeeping gene is that it should, of course, be adequately expressed in the tissue of interest, but most importantly, that it shows minimal variability in expression between samples and under the experimental conditions used [2, 4]. A study carried out in 1999 by Suzuki et al  reported that over 90% of the RNA transcription analysis published in high impact journals used a single housekeeping gene, and this is still common practise today. It is abundantly clear, however, that many of these control genes can show unacceptable variability in expression [6–10]. Appropriate validation of housekeeping genes in any new experimental system is therefore crucial.
Various attempts to find the most appropriate method for the selection of housekeeping genes have been made. One software-based approach, developed by Vandesompele et al  involves the use of normalising to more than one housekeeping gene, where, very briefly, an algorithm based computer program, geNorm is used to determine the most stable control genes from a panel of candidate housekeeping genes via a stepwise exclusion or ranking process, and this is then followed by geometric averaging of a selection of the most stable control genes. Other software based approaches include BestKeeper , an Excel-based tool and Normfinder  an add-in for Microsoft Excel which adds the NormFinder functionality directly to the Excel software package. Although accurate and reproducible, these approaches require the use of specialist programs, and small quantities of starting material in some experiments can be prohibitive and as a result may not be appropriate for reticulocyte RNA. A further, more common strategy for the investigation of gene expression involves the standardisation of starting mRNA whereby a constant amount of RNA is added to each reverse transcription reaction . In some instances, it is impossible to quantify this parameter, for example, for reticulocyte RNA, only minimal amounts of RNA are obtained from patient samples and errors in measurement can occur due to protein contamination in the RNA sample. We have adopted a method, similar to that described by Vandesompele et al , whereby 'pairs of genes' are compared using a simple ΔCt approach.
Panel of 7 candidate housekeeping genes selected and HBG2.
mRNA accession number
Cytoskeletal structural protein
Alpha haemoglobin stabilising factor
Chaperone for human alpha globin
Cytoskeletal protein involved in cell locomotion
Hypoxanthine guanine phosphoribosyl transferase I
Involved in the metabolic salvage of purines in mammals
Succinate dehydrogenase complex, subunit A, flavoprotein
Involved in the oxidation of succinate
FP, SDH2, SDHF
Gamma globin (γ-globin)
Oxygen binding and transport
Expression profiling of housekeeping genes
Sequence information for primers and probes used.
Sequence – 5' to 3'
Assay by design (ABI)
2 to 3
2 to 4
2 to 3
Assay by design (ABI)
4 to 5
Assay by design (ABI)
Assay by design (ABI)
Q-RT-PCR efficiency and intra- and inter- assay variability
Quantitative real-time PCR was used to measure the RNA transcription level of various candidate housekeeping genes. To compare different RNA transcription levels, the Ct values were compared directly. The Ct is defined as the number of cycles needed for the fluorescence to reach a specific threshold level of detection and is inversely correlated with the amount of template nucleic acid present in the reaction . To ensure comparability between the 7 PCR assays, efficiency of each individual assay was determined by measuring serial dilutions of 500 ng cDNA from K562 cells in triplicate. Inter-assay variation was investigated in three independent runs performed on three consecutive days. Only Ct values < 40 were used for calculation of the PCR efficiency from the given slope generated in the SDS 2.2 software according to the equation: PCR efficiency = (10[-1/slope]-1) × 100. All PCR assays displayed efficiency between 92% and 100%.
Intra-assay variation was < 0.67% and inter-assay variation < 0.4% for all assays.
Approaches used for housekeeping gene selection ΔCt Approach
It is not always satisfactory to rely on the precision and reproducibility of total RNA standardisation alone for Q-RT-PCR applications, as a very stringent estimate of concentration is required. To bypass this potential source of error we used a similar technique to Vandesompele et al. However, without compromising accuracy, we kept the level of mathematical methodology to a minimum in order to allow non-specialist personnel to easily follow a similar route in housekeeping discovery. Here we employed a ΔCt method by comparing relative expression of 'pairs of genes' within each sample to confidently identify useful housekeeping genes. If the ΔCt value between the two genes remains constant when analysed in different samples of reticulocyte RNA, it means either both genes are stably expressed among those samples, or co-regulated (here we assume the stability of both genes). However, if the ΔCt fluctuates, then 1 or both genes are variably expressed. Introduction of a third, fourth, fifth gene into the comparisons will provide more information on which pairs show less variability and hence which gene(s) has stable expression among samples tested. These can then be ranked or discarded. Using this technique of 'process of elimination' means relatively large panels of genes can be compared against one another and either selected or eliminated on the basis of ΔCt. Ultimately, a housekeeping gene(s) that is appropriate to use in the experimental system of interest can be selected.
Relative Stability of housekeeping genes
Candidate housekeeping gene comparisons.
GAPDH vs SDHA
GAPDH vs HPRT1
GAPDH vs HBS1L
GAPDH vs AHSP
GAPDH vs B2M
SDHA vs GAPDH
SDHA vs HBS1L
SDHA vs HPRT1
SDHA vs AHSP
SDHA vs B2M
HPRT1 vs GAPDH
HPRT1 vs SDHA
HPRT1 vs B2M
HPRT1 vs HBS1L
HPRT1 vs AHSP
HBS1L vs GAPDH
HBS1L vs SDHA
HBS1L vs AHSP
HBS1L vs HPRT1
HBS1L vs B2M
AHSP vs GAPDH
AHSP vs SDHA
AHSP vs HBS1L
AHSP vs HPRT1
AHSP vs B2M
B2M vs HPRT1
B2M vs GAPDH
B2M vs SDHA
B2M vs HBS1L
B2M vs AHSP
Based on the expression stability, the data suggests that GAPDH would be the most suitable housekeeping gene to use in reticulocyte studies. A popular housekeeper, B2M, should be avoided. In addition, comparison of GAPDH against the other 5 genes in the gene panel not only indicates its level of stability but also shows this level of stability cannot be due to co-regulation as each gene has a different function.
Evaluation and validation of selected candidate housekeeping genes using γ-globin
Following identification of the most stable housekeeping genes from the full gene panel of 7 genes, a method was needed for their validation. This validation process was based on protein (Hb F, α2γ2) levels as measured by high performance liquid chromatography (HPLC, BioRad Variant).
As seen in figure 3A, when γ-globin is normalised to B2M no relationship can be seen (R = 0.1221, P value (two-tailed) 0.5129), indicating B2M inappropriate as a housekeeping gene. However, there is a considerable improvement in the correlation when SDHA is used for normalisation; SDHA R = 0.7050, P value (two-tailed) P < 0.0001. The third gene exampled (figure 3C) in this validation process was GAPDH, which improves the correlation even further between γ-globin and Hb F; R = 0.8535, P value (two-tailed) P < 0.0001.
Reticulocytes are present in small numbers in peripheral blood, can be readily isolated, and still harbour remnants of genetic material that likely represent gene expression profiles of patients. This makes them a good source of biological material in which to study differential gene expression. Quantitative trait loci involved in the phenotypic outcome of haemoglobinopathies, which are mapped and localised by linkage and association studies, usually exert their effect through subtle changes in gene expression. To detect these subtle changes, a sensitive method of detection is needed. Real time PCR is one of the most sensitive and reproducible quantification methods for gene expression analysis. It provides simultaneous measurement of gene expression in many different samples for a number of genes. However, many different factors in real-time PCR may effect the results, including the selection of the housekeeping genes. The 'ideal' housekeeping gene should be constantly transcribed in all cell types and tissues and remain stable between samples taken from different time points and under different experimental conditions. However, it is impossible to find a 'universal' housekeeping gene having stable expression under all these conditions [4, 16]. For example, ACTB, GAPDH and 18S are the most commonly used housekeeping genes, but a number of studies have provided evidence that their transcription levels vary considerably between different individuals, different cell types, different developmental stages and under different experimental conditions [4, 5, 9]. Therefore, for accurate analysis of gene expression, thorough validation of candidate housekeeping genes is crucial.
Another crucial factor in gene expression analysis concerns using the most appropriate method for housekeeping gene selection. A significant consideration when it comes to this is the small quantities of mRNA obtained from reticulocytes. Attempts to find the most appropriate method has involved an approach developed by Vandesompele et al . This approach relies on the principle that two perfect housekeeping genes would be identical in all samples in all experimental conditions or cell types. Variation in expression ratios between different samples reflects the fact that one or both of the genes are not stably expressed. Therefore, increasing variation in this ratio corresponds to decreasing expression stability. A visual basic application for Microsoft Excel – termed geNorm – has been written that uses an algorithm to calculate M, a gene expression stability measure, which is the mean pairwise variation for a gene compared with all other tested control genes. Genes with higher M values have greater variation in expression, and via a stepwise exclusion process genes can be ranked. For an accurate measure of expression levels, normalisation by multiple housekeeping genes is suggested. Consequently, a normalisation factor based on the expression levels of the best performing housekeeping genes must be calculated via averaging of the control genes using the geometric mean. It is suggested that 3 stable control genes should suffice for accurate normalisation of samples with relatively low expression variation, whereas other tissue panels require a fourth, or even a fifth control gene to capture the variation. Further software based approaches include BestKeeper  and Normfinder . BestKeeper, an Excel-based tool also uses 'pair-wise' correlations, and can determine the best suited standards, out of 10 candidates, and combine them into an index. Where the earlier presented geNorm software is restricted to housekeeping gene analysis only, the index generated by BestKeeper can then be compared with a further 10 genes to decide whether they are differentially expressed, for example, under an applied treatment. Here, all data processing is based on crossing points . Normfinder, whose strategy is rooted in a mathematical model of gene expression, enables estimation not only of the overall variation of the candidate normalisation genes but also of the variation between sample subgroups of the sample set. Notably, the strategy provides a direct measure for the expression variation, enabling the user to evaluate the systematic error introduced when using the gene. It has also been reported to show relatively low sensitivity towards coregulation of the candidate normalisation genes . Although very accurate, these methods rely on the design of specialist software programs, and where the use of multiple housekeeping genes are suggested, it is difficult to obtain enough reticulocyte sample to realistically use 3 or more housekeeping genes for normalisation. A further common strategy for the investigation of gene expression involves the standardisation of starting mRNA whereby a constant amount of total RNA is added to each reverse transcription reaction . In some instances however, it is impossible to quantify this parameter, for example, because only minimal amounts of RNA are obtained from patient samples, errors in measurement can occur due to protein contamination in the RNA sample. We have therefore adopted a ΔCt approach; a method similar to that described by Vandesompele et al, whereby 'pairs of genes' are compared. This simple ΔCt approach bypasses the need to accurately quantify input RNA and instead uses ΔCt comparisons between genes.
Results show that the level of RNA transcription for some of the candidate housekeeping genes varied considerably and in some cases RNA transcription levels were too low to quantify accurately or reproducibly. This was particularly the case with ACTB, and this would therefore be completely inappropriate to use as a housekeeping gene. Out of all the genes tested, those that scored highest in their requirements as housekeeping genes were GAPDH, SDHA and HPRT1. However, when carrying out ΔCt comparisons, in order to provide the highest level of accuracy, the Ct values compared should not be vastly different. HPRT1 expression was found to be relatively low and it would therefore be inappropriate to compare this against a relatively highly expressed gene, for example, against γ-globin. GAPDH fulfilled most criteria as a suitable housekeeping gene in that it was strongly expressed, displayed minimal fluctuation and is likely to be independent of genes (i.e. not co-ordinately expressed) commonly involved in reticulocyte studies. In addition, B2M, which has been found to be a stable housekeeping gene in other biological systems [14, 17] showed one of the highest levels of variation in this tissue. To confirm the validity of the approach used in the housekeeping gene selection process, γ-globin levels normalised against the gene panel were compared with their corresponding protein level. The strongest correlation was observed between GAPDH-normalised γ-globin expression and HPLC-measured Hb F levels, supporting GAPDH as an appropriate housekeeping gene for reticulocytes. The second highest ranking housekeeping gene from the gene panel was SDHA; SDHA-normalised γ-globin expression was then correlated with Hb F levels.
A large number of studies have been carried out concerning the validation of housekeeping genes in many different tissues and cell types. However, it has been difficult to find information on appropriate housekeeping genes for use in reticulocytes. To our knowledge, the only reported gene for normalisation in reticulocytes is RPS19 . For accurate quantitation of a target gene, it is recommended that the housekeeper should have similar expression levels. This increases the sensitivity to detect subtle differences in gene expression. However, RPS19 is very highly expressed, as are many of the ribosomal genes, and would therefore be more appropriate for normalising very highly expressed genes such as α and β globins, but preferably not for genes with lower expression. Reticulocytes are often used in haematological studies and as a result, this information on alternative housekeeping genes may be very useful, if not essential for any future studies carried out involving this cell type.
Peripheral reticulocytes are a readily accessible source of biological material for gene expression studies. For this reason, a survey of a small panel of genes including some of the more popular housekeeping genes was carried out. This was done using a simple ΔCt method based on comparing pairs of genes. We found at least one gene, GAPDH, which shows a good level of expression and stability. ACTB showed background levels of expression, and B2M was the least stable.
Clinical samples were collected from healthy unrelated adult volunteers (36 to 73 years old, median age 56; 5 male and 26 female) recruited through the St Thomas' UK Twin Registry . This entailed collection of venous blood in both heparinized and EDTA-containing BD vacutainers™. Informed consent was obtained from all participants prior to collection. All donors had normal haemoglobin levels and normal RBC indices. Fetal haemoglobin (Hb F, α2γ2) percentages were measured by HPLC (Bio-Rad Laboratories, Inc., Hercules, CA). The study was approved by the Research Ethics Committee of St Thomas' Hospital, London, UK (E04/015).
Isolation of peripheral blood reticulocyte RNA and reverse transcription
Reticulocytes were isolated from 10 ml of peripheral blood in heparin following a process of leukodepletion . In brief, the reticulocyte-enriched, leukocyte-depleted red cell eluate was centrifuged to recover the reticulocytes which were then resuspended in 2 ml of cold PBS and 5 ml of TRI reagent (T-9424, Sigma-Aldrich) and stored at -80°C until RNA extraction.
RNA extraction was carried out on the thawed samples according to the manufacturer's protocol. The precipitated RNA was resuspended in 50 μl DEPC water and 1 microgram of RNA was reverse transcribed with 200U SuperScript™ III Reverse Transcriptase (Invitrogen) for 60 min at 50°C using Oligo-dT in a 20 μl volume.
Purity of reticulocytes, assessed by a Sysmex automated blood cell analyser, was 1 leukocyte per 35.5 × 103 RBCs and no detectable platelets.
Selection of primers and probes
Primer and probe sequences are shown in table 2. These were synthesised by Applied Biosystems and included both customised primer and probe sets, designed via the Applied Biosystems Primer Express™ 2.0 software as well as pre-designed, gene-specific TaqMan® probe and primer sets (TaqMan® Gene Expression Assays, Applied Biosystems). Primers were used at 300 nM and probes at 200 nM in a 25 μl reaction. A selection of housekeeping genes were chosen, paying close attention to selecting genes that belong to different functional classes. This was aided by information provided by the human gene expression index database (HuGE) . This was used to assess expression, variability and reproducibility of potential housekeeping genes.
Real-Time quantitative RT-PCR
PCR was performed using the ABI Prism® 7900HT Sequence Detection System (Applied Biosystems) in 96 well microtitre plates using a final volume of 25μl. Amplifications were performed starting with a 2 min activation step for AmparaseUNG at 50°C, 10 min template denaturation step at 95°C, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min.
- ACTB :
- AHSP :
alpha haemoglobin stabilising protein
- B2M :
- ΔCt (dCt):
delta cycle threshold
- HBG2 :
- GAPDH :
glyceraldehyde 3-phosphate dehydrogenase
- Hb F:
haemoglobin F (fetal haemoglobin)
- HBS1L :
- HPRT1 :
no template control
polymerase chain reaction
quantitative real-time PCR
- SDHA :
This work was supported by a grant from the Medical Research Council (G0000111, ID51640) to S L Thein.
- Koepke JF, Koepke JA: Reticulocytes. Clin Lab Haematol. 1986, 8: 169-179.View ArticlePubMed
- Dheda K, Huggett JF, Bustin SA, Johnson MA, Rook G, Zumla A: Validation of housekeeping genes for normalizing RNA expression in real-time PCR. Biotechniques. 2004, 37: 112-119.PubMed
- Karge WHIII, Schaefer EJ, Ordovas JM: Quantification of mRNA by polymerase chain reaction (PCR) using an internal standard and a nonradioactive detection method. Methods Mol Biol. 1998, 110: 43-61.PubMed
- Thellin O, Zorzi W, Lakaye B, De BB, Coumans B, Hennen G, Grisar T, Igout A, Heinen E: Housekeeping genes as internal standards: use and limits. J Biotechnol. 1999, 75: 291-295. 10.1016/S0168-1656(99)00163-7View ArticlePubMed
- Suzuki T, Higgins PJ, Crawford DR: Control selection for RNA quantitation. Biotechniques. 2000, 29: 332-337.PubMed
- Deindl E, Boengler K, van RN, Schaper W: Differential expression of GAPDH and beta3-actin in growing collateral arteries. Mol Cell Biochem. 2002, 236: 139-146. 10.1023/A:1016166127465View ArticlePubMed
- Hamalainen HK, Tubman JC, Vikman S, Kyrola T, Ylikoski E, Warrington JA, Lahesmaa R: Identification and validation of endogenous reference genes for expression profiling of T helper cell differentiation by quantitative real-time RT-PCR. Anal Biochem. 2001, 299: 63-70. 10.1006/abio.2001.5369View ArticlePubMed
- Zhong H, Simons JW: Direct comparison of GAPDH, beta-actin, cyclophilin, and 28S rRNA as internal standards for quantifying RNA levels under hypoxia. Biochem Biophys Res Commun. 1999, 259: 523-526. 10.1006/bbrc.1999.0815View ArticlePubMed
- Glare EM, Divjak M, Bailey MJ, Walters EH: beta-Actin and GAPDH housekeeping gene expression in asthmatic airways is variable and not suitable for normalising mRNA levels. Thorax. 2002, 57: 765-770. 10.1136/thorax.57.9.765PubMed CentralView ArticlePubMed
- Radonic A, Thulke S, Mackay IM, Landt O, Siegert W, Nitsche A: Guideline to reference gene selection for quantitative real-time PCR. Biochem Biophys Res Commun. 2004, 313: 856-862. 10.1016/j.bbrc.2003.11.177View ArticlePubMed
- Vandesompele J, De PK, Pattyn F, Poppe B, Van RN, De PA, Speleman F: Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol. 2002, 3: RESEARCH0034- 10.1186/gb-2002-3-7-research0034PubMed CentralView ArticlePubMed
- Pfaffl MW, Tichopad A, Prgomet C, Neuvians TP: Determination of stable housekeeping genes, differentially regulated target genes and sample integrity: BestKeeper--Excel-based tool using pair-wise correlations. Biotechnol Lett. 2004, 26: 509-515. 10.1023/B:BILE.0000019559.84305.47View ArticlePubMed
- Andersen CL, Jensen JL, Orntoft TF: 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. 2004, 64: 5245-5250. 10.1158/0008-5472.CAN-04-0496View ArticlePubMed
- Dydensborg AB, Herring E, Auclair J, Tremblay E, Beaulieu JF: Normalizing genes for quantitative RT-PCR in differentiating human intestinal epithelial cells and adenocarcinomas of the colon. Am J Physiol Gastrointest Liver Physiol. 2006, 290: G1067-G1074. 10.1152/ajpgi.00234.2005View ArticlePubMed
- Walker NJ: Tech.Sight. A technique whose time has come. Science. 2002, 296: 557-559. 10.1126/science.296.5567.557View ArticlePubMed
- Haberhausen G, Pinsl J, Kuhn CC, Markert-Hahn C: Comparative study of different standardization concepts in quantitative competitive reverse transcription-PCR assays. J Clin Microbiol. 1998, 36: 628-633.PubMed CentralPubMed
- Zhang X, Ding L, Sandford AJ: Selection of reference genes for gene expression studies in human neutrophils by real-time PCR. BMC Mol Biol. 2005, 6: 4- 10.1186/1471-2199-6-4PubMed CentralView ArticlePubMed
- Bonafoux B, Lejeune M, Piquemal D, Quere R, Baudet A, Assaf L, Marti J, guilar-Martinez P, Commes T: Analysis of remnant reticulocyte mRNA reveals new genes and antisense transcripts expressed in the human erythroid lineage. Haematologica. 2004, 89: 1434-1438.PubMed
- Spector TD, Macgregor AJ: The St. Thomas' UK Adult Twin Registry. Twin Res. 2002, 5: 440-443. 10.1375/136905202320906246View ArticlePubMed
- Pal S, Nemeth MJ, Bodine D, Miller JL, Svaren J, Thein SL, Lowry PJ, Bresnick EH: Neurokinin-B transcription in erythroid cells: direct activation by the hematopoietic transcription factor GATA-1. J Biol Chem. 2004, 279: 31348-31356. 10.1074/jbc.M403475200View ArticlePubMed
- Haverty PM, Weng Z, Best NL, Auerbach KR, Hsiao LL, Jensen RV, Gullans SR: HugeIndex: a database with visualization tools for high-density oligonucleotide array data from normal human tissues. Nucleic Acids Res. 2002, 30: 214-217. 10.1093/nar/30.1.214PubMed CentralView ArticlePubMed
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.