Effects of creatine supplementation on housekeeping genes in human skeletal muscle using real-time RT-PCR
R. M. Murphy,
K. K. O. Watt,
D. Cameron-Smith,
C. J. Gibbons and
R. J. Snow
School of Health Sciences, Deakin University, Burwood, Victoria 3125, Australia
 |
ABSTRACT
|
---|
The present study examined the validity and reliability of measuring the expression of various genes in human skeletal muscle using quantitative real-time RT-PCR on a GeneAmp 5700 sequence detection system with SYBR Green 1 chemistry. In addition, the validity of using some of these genes as endogenous controls (i.e., housekeeping genes) when human skeletal muscle was exposed to elevated total creatine levels and exercise was also examined. For all except 28S, linear relationships between the logarithm of the starting RNA concentrations and the cycle threshold (CT) values were established for ß-actin, ß2-microglobulin (ß2M), cyclophilin (CYC), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). We found a linear response between CT values and the logarithm of a given amount of starting cDNA for all the genes tested. The overall intra-assay coefficient of variance for these genes was 1.3% and 21% for raw CT values and the linear value of 2-CT, respectively. Interassay variability was 2.3% for raw CT values and 34% for the linear value of 2-CT. We also examined the expression of various housekeeping genes in human skeletal muscle at days 0, 1, and 5 following oral supplementation with either creatine or a placebo employing a double-blind crossover study design. Treatments were separated by a 5-wk washout period. Immediately following each muscle sampling, subjects performed two 30-s all-out bouts on a cycle ergometer. Creatine supplementation increased (P < 0.05) muscle total creatine content above placebo levels; however, there were no changes (P > 0.05) in CT values across the supplementation periods for any of the genes. Nevertheless, 95% confidence intervals showed that GAPDH was variable, whereas ß-actin, ß2M, and CYC were the least varying genes. Normalization of the data to these housekeeping genes revealed variable behavior for ß2M with more stable expressions for both ß-actin and CYC. We conclude that, using real-time RT-PCR, ß-actin or CYC may be used as housekeeping genes to study gene expression in human muscle in experiments employing short-term creatine supplementation combined with high-intensity exercise.
gene expression; quantitative RT-PCR; endogenous controls; biological variability; PCR efficiency
 |
INTRODUCTION
|
---|
THE EMPLOYMENT of the reverse transcription polymerase chain reaction (RT-PCR) method has been widely used for the analyses of gene expression in a number of systems, including skeletal muscle. Over recent years the development of "real-time" RT-PCR equipment has introduced a technique with higher throughput and more quantitative capabilities. As with other methods such as Northern hybridization and ribonuclease (RNase) protection, real-time RT-PCR quantifies the relative abundance of the gene of interest against a constitutively (unregulated and constant), and often highly expressed gene, or a gene which has been shown to vary little across an intervention, referred to as an endogenous control or housekeeping gene. In RT-PCR, housekeeping genes are active references that are used to standardize between amounts of samples assayed. As such, their constitutive, or least changing, nature must be determined for the given experimental conditions.
Commonly used housekeeping genes for studies involving mRNA measurements in human skeletal muscle include 18S (10, 12) and 28S ribosomal RNA (rRNA) (4), ß-actin (29), or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (31). It has become apparent that there is no single housekeeping gene that is suitable for all experimental conditions, and a number of papers have been published looking at the viability of using certain genes as housekeeping genes in different systems using various methodologies (11, 13, 14, 27, 28, 33, 34).
Dietary creatine supplementation (CrS) is a widely used intervention in exercise science and, more recently, in clinical studies examining various muscular disease states (for review, see Ref. 32). Many studies have demonstrated that CrS results in an elevated total creatine content in skeletal muscle (9, 20, 26). Importantly, this increase may induce changes in muscle gene expression perhaps by altering the energy state of the cell or by cell volume-mediated processes. For example, changes in both gene and/or protein expression have been recently reported in models involving CrS and resistance training compared with a placebo group (15, 31). In studies examining gene expression, knowledge of the variations in the expression of housekeeping genes under given experimental conditions is required for valid data interpretation. Although real-time RT-PCR requires the use of a housekeeping gene for interpretation of the data, the relative constitutive nature of this gene must first be established, as must the use of raw data values obtained in real-time RT-PCR, before normalizing them to a housekeeping gene. Consequently, the initial aims of the present study were to determine whether a linear response existed for detection of 28S, ß-actin, ß2-microglobulin (ß2M), cyclophilin (CYC), and GAPDH gene expression in human skeletal muscle using raw data values and to characterize the real-time RT-PCR method in terms of intra- and interassay variability. We then aimed to identify genes, which in response to short-term CrS, were expressed with the least variation and to subsequently establish their use as a housekeeping gene in human skeletal muscle. The samples analyzed in this study were part of a larger study, and the protocol included a small number of high-intensity exercise bouts to investigate the potential ergogenic effects of Cr. Finally, to establish the power with which changes could be identified, we performed power analyses for the detection of the genes 28S, ß-actin, ß2M, CYC, and GAPDH using the real-time RT-PCR technique.
 |
MATERIALS AND METHODS
|
---|
Total RNA isolation, RT, and real-time PCR.
For all experiments total RNA was extracted from 510 mg of human skeletal muscle samples using a modification of the phenol/chloroform extraction, isopropanol precipitation protocol, using the FastRNA protocol (BIO 101, Vista, CA). Following a series of centrifugation and separation steps, a pellet of RNA was obtained and suspended in EDTA-treated water. Following quantification of the total RNA using the optical density at 260 nm (OD260), RNA was heated at 65°C for 10 min immediately prior to first strand cDNA being generated using AMV reverse transcriptase (kit A3500; Promega, Madison, WI) with oligo (dT)15 primers, in the presence of 1 mM of each dNTP and 20 U of recombinant RNasin ribonuclease inhibitor. A reverse transcription (RT) negative was obtained by not adding any RNA to an aliquot of the RT mix. The cDNA was stored at -80°C until subsequent analysis. All samples being compared underwent RT together. For the PCR step, reaction volumes of 20 µl contained SYBR Green 1 Buffer (Applied Biosystems, Foster City, CA), forward and reverse primers (see Table 1), and cDNA template (see below) per tube. Samples were run in triplicate on one or two separate occasions (see below). Real-time PCR was run for 1 cycle (50°C 2 min, 95°C 10 min) followed immediately by 40 cycles (95°C 15 s, 60°C 60 s), and fluorescence was measured after each of the repetitive cycles. The fluorescence resulted from the incorporation of SYBR Green 1 dye into the double-stranded DNA produced during the PCR reaction, and emission data were quantitated using the threshold cycle (CT) value. CT readings obtained represent measurements on the log scale. A melting point dissociation curve generated by the instrument was used to confirm that only a single product was present. To validate the specificity of a primer set, RNA (13 µg) and the RT negative were analyzed in triplicate to confirm that there was no fluorescence resulting from either genomic DNA contamination or from the RT step. Each PCR run also included triplicate wells of no template control (NTC) where RNase-free water was added to reaction wells.
Pilot study.
A pilot study was conducted to determine which housekeeping genes would be examined in the present study. RNA was extracted from several samples of skeletal muscle obtained from individuals exposed to various interventions used in our laboratory (e.g., exercise; dietary intervention, training etc). This RNA was pooled and converted to cDNA using random hexamers. Resultant cDNA was added in duplicate to the commercially available Applied Biosystems Endogenous Control Plate (Applied Biosystems) and underwent PCR according to the manufacturers instructions. For each sample, a CT value was obtained for each of the 12 genes present on the plate. The genes with the least variation in CT values across all the interventions were found to be ß-actin, ß2M, and CYC. In addition to these genes, GAPDH was examined, since it has been used in a previous study examining gene expression in human skeletal muscle following CrS (31). 28S was also examined, since it has been used as a representative ribosomal RNA previously (4, 8).
Primer designs.
To perform PCR, specific primers were designed for the genes 28S, ß-actin, ß2M, CYC, and GAPDH, using Primer Express software (Applied Biosystems) on sequences attained from GenBank (see Table 1 for details). Searching of the sequences using Blast (http://www.ncbi.nlm.nih.gov/BLAST/) confirmed their specificity. Primers were purchased from GeneWorks (Adelaide, SA, Australia).
Relationship between starting cDNA amounts and PCR results.
To determine the validity of using raw CT values as a measure of starting cDNA concentrations, real-time PCR was performed to measure the genes 28S, ß-actin, ß2M, CYC, and GAPDH. RNA extracted from a single human muscle sample was used to examine the dynamic range of response for a series of dilutions of cDNA generated from the RT step (see Table 2). Linear regression was used to analyze the response of CT vs. the logarithm of the cDNA concentration. Using the slopes of the lines, we calculated the efficiency (E) of target amplification with the equation E = (10-1/slope) - 1 (Ref. 18).
Linearity of RT step to PCR.
To ascertain whether, for the genes examined (28S, ß-actin, ß2M, CYC, and GAPDH), the RT step exhibited a linear response, RNA was extracted from another human muscle sample cut into six pieces. The RNA was then quantified, diluted to 0.25 µg/µl, and pooled. Following this, various volumes were added to the same amount of RT mix (optimal for RT of 1.0 µg RNA) to give 0.1 µg to 2.0 µg RNA, which subsequently underwent simultaneous RT followed by PCR. Linear regression was used to analyze the response of CT vs. the logarithm of the amount of input RNA to the RT step. Efficiency was calculated as previously described.
Intra-assay variability of RT-PCR.
To determine the intra-assay variability, the coefficient of variance (CV) was determined for samples run in triplicate for detection of ß-actin, ß2M, CYC, and GAPDH. CV values were calculated using the term 100 x (standard deviation/mean). A total of 94 triplicate readings were used over the 4 genes. The CV was determined for the raw CT values, and also for the term 2-CT, which converts the log scale CT value to the linear form.
Interassay variability of RT-PCR.
To determine interassay variability, the CV was calculated for triplicate readings of samples (n = 34), which were run over two separate occasions (Table 3). For the four genes examined, the two occasions were separated by up to 39 days (21 ± 3 days, mean ± SE, see Table 3). The CV was also calculated for the 2-CT values.
View this table:
[in this window]
[in a new window]
|
Table 3. Coefficient of variance for triplicate readings for intra-assay (column A) and interassay (column B) assay variability using raw CT values or 2-CT values
|
|
Biological and technical variability.
Technical variability incorporated inconsistencies due to dilutions, as well as instrumentation variation in OD260 readings and CT values obtained from the PCR. To determine the variability due to the dilutions, six aliquots of total RNA extracted from a single muscle sample were quantified on a spectrophotometer measuring the OD260. The variability in the OD260 readings was measured in the same samples (n = 3) over three occasions. To evaluate the technical variability due to the RT step, six aliquots of a single RNA sample underwent RT (0.5 µg total RNA per 10 µl reaction) followed by PCR in triplicate. To determine the error associated with biological variability, a single skeletal muscle sample was cut into seven pieces (7.1 ± 1.4 mg, mean ± SD), and total RNA was extracted simultaneously from each piece. Subsequent to quantification of the total RNA from all samples, 0.5 µg underwent RT in 10 µl. Following RT, the cDNA from all samples was diluted and underwent PCR in triplicate using ß-actin primers. To determine the combined technical and biological variability of the samples, the variation in the CT value obtained for the given samples was calculated. To establish the contribution due to biological variability, the average of the technical variability due to the RT and PCR steps involved in obtaining a CT value was calculated and subtracted from the overall variability.
Study design for the determination of an appropriate housekeeping gene when skeletal muscle Cr content was elevated by oral Cr supplementation.
Healthy, active male subjects, aged 25.4 ± 4.6 yr (mean ± SD) were used in the study, which was approved by Deakin University Human Ethics Committee and was part of a larger study being conducted by us. The study group consisted of habitual meat eaters (n = 4) and non-meat eaters (n = 4). Subjects (n = 8) underwent a double-blind, crossover study involving the ingestion of Cr (Cr, 0.4 g·kg body wt-1·day-1, plus glucose, 0.4 g·kg body wt-1·day-1; CR group) or a placebo (glucose, 0.8 g·kg body wt-1·day-1; placebo group). Treatment order was randomly assigned and was separated by 5 wk. The washout period of intramuscular Cr has previously been shown to be 4 wk (9, 20). Resting muscle biopsies were taken from the vastus lateralis as previously described (26) before the supplementation period (day 0) and following 1 day (day 1) and 5 days (day 5) of supplementation using the percutaneous needle biopsy method modified to include suction. The muscle samples were stored in liquid nitrogen until analyzed. Immediately after each biopsy, two 30-s all-out sprints, separated by 4 min of passive rest, were performed on a cycle ergometer (Excalibur ergometer; Lode, Groningen, The Netherlands). Consequently, the data correspond to the effects of the ingestion of Cr compared with a placebo when both groups perform this exercise.
Total Cr content.
To ensure that all subjects responded to CrS, intramuscular total Cr (TCr) measurements were made. Muscle samples were freeze-dried for 24 h, weighed and powdered, removing any visible connective tissue, and then extracted using 0.5 M perchloric acid and 1 mM EDTA, and neutralized with 2.1 M KHCO3. These extracts were subsequently analyzed for creatine phosphate (CrP) and Cr levels using an enzymatic fluorometric technique (19). TCr was taken as the sum of CrP and Cr.
Examination of housekeeping genes.
Based upon the results of the validity and reliability studies, the protocols for which are outlined above, four genes were chosen to establish their consistency of expression in human skeletal muscle biopsies at 0, 1, and 5 days postsupplementation with Cr or placebo. To reduce intersubject variation, samples were normalized to the day 0 sample of the placebo trial for each subject. For ß-actin, ß2M, CYC, and GAPDH, a
CT value was calculated for each sample by subtracting the CT value for a given sample from the CT value of the gene being treated as the housekeeping gene (except when this was the same gene). The relative expressions of these genes compared with the basal values were calculated using the expression 2-
CT. The efficiency of the RT-PCR for ß2M, ß-actin, and CYC was determined to be the same across a series of dilutions (data not shown), thus justifying their use as housekeeping genes.
Power analyses.
To reveal the differences measurable in the current experiments, power analyses were performed for each gene using the equation
where n = required sample size per group; S = estimate of standard deviation (average of triplicate readings from n = 24 for each gene);
= desired width of confidence interval (CI) in CT; (1 -
) = 95% CI; (1 - ß) = power; f(
,ß) was adopted for a two-tailed test.
Statistics.
BMDP software was used to perform two-way (treatment x time) ANOVAs. TCr results are expressed as means ± SE; 95% CI values, slopes, and y-intercepts were calculated using Prism GraphPad Software.
 |
RESULTS
|
---|
Amplification of all genes was as expected, with a region of exponential increase in fluorescence preceding a plateau phase. Following RT-PCR, the samples were subjected to a heat dissociation protocol over 6095°C, indicating the number of products present in each reaction well, and the temperature at which the products dissociated (Fig. 1). RNA, RT negative, and NTC produced no amplification (i.e., remained below threshold) during the 40 cycles of PCR, confirmed by neither a CT value nor a dissociation curve being obtained (data not shown). This substantiated the absence of genomic DNA or primer dimer formation, which could otherwise contribute to observed fluorescence. Threshold values were set to cross the exponential phase of each genes amplification, and from this the CT was determined. The amount of transcript present is inversely related to the observed CT, and for every twofold dilution in transcript a single increase in CT is expected to be observed.

View larger version (36K):
[in this window]
[in a new window]
|
Fig. 1. Real-time RT-PCR analysis of skeletal muscle mRNA content. Typical heat dissociation protocol for cyclophilin (CYC). Following PCR, the samples and no template control (NTC) were subjected to a heat dissociation protocol over 6095°C. Samples yielded a product, which dissociated at 83°C, whereas the NTC had no product present.
|
|
Linearity of various input cDNA amounts to PCR.
For each gene, the validity of using CT values was established, since a linear relationship existed between the CT values and the logarithm of the cDNA template concentration (Fig. 2). Dilutions of cDNA exhibited five orders of dynamic range for CT values for 28S, ß-actin, ß2M, CYC, and GAPDH (r = -0.9786, -0.9581, -0.9255, -0.9465, and -0.9066, respectively; Fig. 2). The mean efficiency (E) of the PCR reaction for 28S, ß-actin, ß2M, and GAPDH was 1.1 and for CYC was 0.9 (see table in Fig. 2). A 100% efficient PCR would produce a slope of -3.3 (E = 1), indicating that twice as many amplicons are being made for every PCR cycle, represented as a decrease of 1 CT. In addition, the 95% CI values for the slope values of the graphs (see table in Fig. 2) indicate a linear detection of the cDNA for all genes.
Linearity of RT step to PCR.
Gene expression of 28S exhibited a positive relationship (r = 0.5180) between input RNA amounts and CT values obtained (data not shown). As the amount of input RNA increased, the CT value increased which indicates a decrease in input template, where there was actually an increase in template. Consequently, 28S was excluded from subsequent experiments testing for a valid housekeeping gene. CT values were not correlated to the logarithm of the input RNA when either too little (
0.10 µg) or too much (
2.0 µg) RNA was added to the RT reaction (data not shown). When the upper and lower concentrations were omitted as outliers, ß-actin, ß2M, CYC, and GAPDH exhibited inverse correlations (r = -0.9230, -0.8787, -0.8130, and -0.7336, respectively) between CT values and the logarithm of the input RNA content (Fig. 3). The efficiencies of the PCR reactions were 1.4, 1.1, 1.3, and 1.2 for ß-actin, ß2M, CYC, and GAPDH, respectively (see table in Fig. 3). Examination of the 95% CI values for the slopes of each gene indicated that the efficiencies were not different.
Intra- and interassay variability of RT-PCR.
Calculation of the CV for CT values and the conversion of the CT values to their linear form using the expression 2-CT was determined for triplicate readings for ß-actin, ß2M, CYC, and GAPDH for intra-assay (Table 3, column A) and interassay (Table 3, column B) variabilities, respectively. Overall, the intra-assay variability was 1.3% and 21% and interassay variability was 2.3% and 34% for raw CT values and 2-CT, respectively.
Biological and technical variability.
The technical variation due to the OD260 readings obtained for a series of six aliquots of the same total RNA extract was 0.050 ± 0.003 (mean ± SD; CV, 6.2%; n = 6). This value incorporates a 5.0% variation due to the repeatability of OD260 readings. Once the validity of using raw CT values for ß-actin was determined, the biological variation seen in different extractions from the same piece of muscle was examined. Variability in the raw CT values for replicates of a single sample due to the RT step was 24.0 ± 0.6 (mean ± SD; CV, 2.6%; n = 7; Table 4). Across the samples, variability in the raw CT values obtained was 24.1 ± 1.5 (mean ± SD; CV, 6.2%; n = 7; Table 4), which accounts for both technical (incorporating inconsistencies in OD260 readings) and biological variability. To determine the biological variability, the variation of the two technical steps (RT 2.6% and PCR 1.3% = 1.9%) involved in obtaining a CT value were averaged, then subtracted from the overall variability. Using this procedure, we found that the biological variability accounted for 4.3% of the overall variability (Table 4).
Intramuscular total Cr content.
Compared with both CR and placebo day 0 concentrations, intramuscular TCr levels were increased in the CR trial (P < 0.05) at day 1 (+22 ± 4%) and day 5 (+54 ± 9%; Fig. 4). In the CR trial, day 5 intramuscular TCr was also higher (P < 0.05) than day 1 (+26 ± 6%). Although there were no differences between any of the time points in the placebo trial, the CR day 0 was lower than day 0 in the placebo trial, probably due to incomplete washout of intramuscular TCr stores in the four subjects who received Cr in the first trial. A 4-wk washout period has previously been shown to result in intramuscular TCr stores returning to pre-CrS values (9, 26). TCr values at days 1 and 5 in the CR trial were greater than all time points in the placebo trial (Fig. 4).
Examination of housekeeping genes using raw CT values.
Having determined the linear nature of CT values vs. the logarithm of cDNA and input RNA amounts, with no difference in the efficiencies between the genes, CT values were subsequently used as a measure of specific mRNA contents in samples. RNA was extracted from human skeletal muscle samples obtained at days 0, 1, and 5 from the CR and placebo trials and underwent RT-PCR in triplicate for ß-actin, ß2M, CYC, and GAPDH. mRNA expression was expressed as the fold change in CT values relative to the CT value at day 0 in the placebo trial (Fig. 5). There was a tendency for a main effect for time for GAPDH (P = 0.08). The 95% CI values have been plotted, which indicate that GAPDH was the most variable gene following CrS, while ß-actin, ß2M, and CYC showed similar variations of within
1 CT value.

View larger version (19K):
[in this window]
[in a new window]
|
Fig. 5. CT values at days 0, 1, and 5 from creatine (CR; solid symbols) and placebo (open symbols) trials (n = 8). All values have been normalized to day 0 of the placebo trial for the genes ß-actin (A), ß2M (B), CYC (C), and GAPDH (D). Values are means with 95% confidence indicated by dashed lines. P = 0.08, trend for time main effect.
|
|
Examination of housekeeping genes using normalized CT values.
To determine a normalized arbitrary value (2-
CT) for each gene, each data point was normalized to the least varying genes (i.e., ß-actin, ß2M, and CYC), as well as to the placebo day 0 CT value. CYC normalized to ß-actin (Fig. 6A) and ß-actin normalized to CYC (Fig. 6D) were not different between the trials. When normalized to ß-actin, ß2M gene expression was different (P < 0.05) between placebo day 0 and day 5 samples (Fig. 6B). There was significant interaction (P < 0.05) for GAPDH (Fig. 6C) when normalized to ß-actin. When normalized to CYC, there were tendencies for main effects for time for ß2M (P = 0.06; Fig. 6E) and GAPDH (P = 0.08; Fig. 6F). When normalized to ß2M, there were no differences in the expression of ß-actin (Fig. 6G), CYC (Fig. 6H), or GAPDH (Fig. 6I).

View larger version (19K):
[in this window]
[in a new window]
|
Fig. 6. Gene expression at days 0, 1, and 5 from creatine (CR; solid symbols) and placebo (open symbols) trials (n = 8). Data are expressed as 2- CT values obtained by normalizing CT values to day 0 of the placebo trial and to ß-actin (A, CYC; B, ß2M; and C, GAPDH), CYC (D, ß-actin; E, ß2M; and F, GAPDH), or ß2M (G, ß-actin; H, CYC; and I, GAPDH). Values are means with 95% confidence are indicated by dashed lines. *P < 0.05 different to day 0.
|
|
Power analysis.
The power with which a 95% CI could be obtained using the data set in the present study (i.e., n = 8) are listed in Table 5. The conventional power = 0.8 indicated that for a group of eight subjects a difference of 1.01.2 CT is measurable in all genes except 28S, for which a 3.1 CT difference could be detected.
 |
DISCUSSION
|
---|
The present study examined the validity and reliability of measuring the expression of various genes in human skeletal muscle using real-time RT-PCR. We examined the relationships between starting RNA concentrations and input cDNA amounts against the CT values obtained for 28S, ß-actin, ß2M, CYC, and GAPDH, as well as the intra- and interassay CV values. For data analysis, an endogenously expressed gene, or a gene that does not vary significantly across the given intervention, known as a housekeeping gene, is required for normalization of the genes of interest. As such, the technique assumes prior knowledge about the least changing behavior of the housekeeping gene with respect to the specified intervention. In the current study, the validity of using certain genes as housekeeping genes when human skeletal muscle was exposed to elevated total creatine levels was examined.
Linearity of various input cDNA amounts to PCR.
The first experiment investigated the dynamic range of quantitation of RT-PCR using CT values. Following the log conversion of a series of fold dilutions from the same cDNA, a linear response was demonstrated for all genes tested (28S, ß-actin, ß2M, CYC, and GAPDH; Fig. 2). Examination of the 95% CI showed that for all genes the slope of the logarithmic relationship included -3.3, which indicated that the PCR was able to detect twofold differences in input template, reflected by a single CT value. Hence, these data indicated that the amounts of cDNA added to a given PCR run were reflected by differences in CT values for 28S, ß-actin, ß2M, CYC, and GAPDH transcripts. This finding validates the use of the CT values for detection of absolute changes in the amount of input cDNA for these particular transcripts. These data are in agreement with other researchers who have reported the efficiency of real-time RT-PCR using SYBR Green 1 chemistry and various genes in different tissues (6, 21, 24).
Linearity of RT step to PCR.
Linear relationships between the logarithm of various input amounts of RNA into the RT reactions and the CT values of ß-actin, ß2M, CYC, and GAPDH were obtained (Fig. 3), once the lowest and highest input RNA amounts were omitted. The addition of either not enough (e.g.,
0.1 µg RNA) or too much (e.g.,
2.0 µg RNA) RNA into an RT reaction, optimal for the RT of 1.0 µg RNA, appeared to be inhibitory to the reaction (data not shown). Although the reactions were not 100% efficient for all the genes, the important feature is that since the 95% CI for the slopes of the graphs for each gene overlapped each other (see table in Fig. 3), the efficiency of the AMV reverse transcriptase enzyme was not different between ß-actin, ß2M, CYC, and GAPDH. These data suggest that for samples undergoing RT at the same time, RT efficiency appears to be the same for input RNA amounts between 0.5 to 1.5 µg and that the subsequent analyses for determination of an appropriate housekeeping gene(s) were valid. The response for 28S was different from the other genes examined, since it exhibited a positive relationship between input RNA amounts of 0.5 to 1.5 µg vs. CT values (data not shown). It is not known why 28S responded in this way, although we speculated that the use of the oligo dT primers, which would selectively reverse transcribe RNA species possessing a poly-A tail, would not be suitable for efficient translation of ribosomal RNA (i.e., 28S). To test that possibility, we undertook RT reactions using random hexamers (Applied Biosystems) using various input amounts (0.25 to 1.5 µg RNA) of pooled RNA. In our hands, this also failed to display an inverse linear relationship (Fig. 7), and as such we eliminated the further testing of 28S in the current experiments. We do not know why this was the case; however, it is possible that the relatively high abundance of the ribosomal RNA may somehow impede the detection of the transcripts in human skeletal muscle using SYBR Green 1 chemistry and the sensitive real-time RT-PCR methods. Consequently, the choice of 28S is not appropriate for real-time PCR under the conditions employed by us (i.e., SYBR Green 1 chemistry, oligo dT or random hexamers as cDNA priming oligo, and two-step RT-PCR). It is possible that 28S might be an appropriate housekeeping gene for human skeletal muscle under other specific conditions (e.g., use of gene-specific primers as cDNA priming oligo, use of TaqMan probe with one-step RT-PCR, or greater dilutions of RNA), although these possibilities were not examined in the current study. To our knowledge, the examination of the efficiency of the RT enzyme AMV on high- or low-abundance target genes has not been reported and might warrant further investigation. Interestingly, it has been demonstrated that the efficiency of the RT enzyme, Moloney murine leukemia virus (MMLV), fell from
20% to less than 6% when a low-abundance target template was used (7). Furthermore, it has recently been reported that the choice of cDNA synthesis conditions influences the sensitivity and accuracy of real-time PCR using SYBR Green 1 chemistry (18). Those authors also found that the efficiency is related to both choice of RT primer as well as transcript being investigated (18). As mentioned in the previous section, the use of fivefold dilutions of cDNA and measurement of CT values demonstrated the linearity of the system for determining the input amount of serial dilutions of 28S rRNA. The data obtained using different amounts of RNA input in the RT reaction along with the findings of Lekanne Deprez and colleagues (18) further highlight the importance, emphasized by numerous researchers previously (11, 14, 17, 22, 25), that investigators need to optimize parameters for individual experimental conditions. To our knowledge, this is the first study that has examined the linearity of the RT step to PCR using RNA obtained from skeletal muscle.

View larger version (12K):
[in this window]
[in a new window]
|
Fig. 7. Dose response curve following RT-PCR for 0.25 to 1.5 µg total RNA added to the RT reaction vs. CT for 28S. cDNA underwent RT using random hexamers primers. Values are means with 95% CI values indicated by dashed lines.
|
|
Intra- and interassay variability of RT-PCR.
CV values for triplicate CT readings were 1.3 to 2.3% for intra- and interassay variability, respectively. These values are in line with those reported by others (5, 23). Interestingly, Schmittgen and colleagues (23) pointed out that presentation of statistical data calculated using these raw CT values was incorrect due to the logarithmic nature of CT values. Using this approach, we recalculated the CV using the transformation of CT to the linear 2-CT value, where the variability was seen as 21 to 34% for intra- and interassays. These values are similar to that reported for intra assay CV using 2-CT (23).
Biological and technical variability.
The same piece of muscle cut into seven pieces resulted in a CV of 6.2%, which included both biological (within the same sample, 4.3%) and technical error (1.9%; see Table 4). It needs to be emphasized that this total variability is likely to be the minimum, since further intra-individual variation is likely when additional muscle samples from the same subject are analyzed, due variation in parameters such as fiber type (2). Nonetheless, these data highlight the areas of variability present within a single skeletal muscle sample when real-time RT-PCR is the tool of measuring mRNA expression. Although previous researchers have attributed large variability [in uncoupling protein (UCP) 2 and UCP3 gene expression in muscle samples obtained from the same subject] to biological variability, no quantitative data were provided for variation within a sample (3), and to our knowledge the current paper is the first to provide these data. Additionally, the current study did not measure intersubject variability, although CV values of 5070% have been reported for the expression of UCP2 and UCP3 gene expression in skeletal muscle amongst individuals (3).
Cr supplementation and gene expression.
Only a single study has examined gene expression following CrS (31). In that study it was reported that skeletal muscle myosin heavy chain (MHC) I, IIa, and IIx gene expression was higher following 12 wk of resistance training combined with CrS compared with resistance training and a placebo (31). This increase in MHC gene expression was attributed to a synergistic response between the resistance training and CrS. In terms of proteins, the ingestion of Cr resulted in differential protein expression of the myogenic transcription factors myogenin and MRF4 following a period of immobilization and 10 wk of resistance training (15). Data from these studies indicate that the examination of gene expression consequent to CrS warrants further investigation. It is important to note that outcomes of studies examining the effect of CrS on gene expression are dependent on the knowledge of an appropriate housekeeping gene.
Examination of housekeeping genes using raw CT values.
Following the characterization of real-time RT-PCR, we examined the response of ß-actin, ß2M, CYC, and GAPDH mRNA at days 0, 1, and 5 following ingestion of either CR or placebo interspersed with high-intensity exercise. The CR trial showed increased intramuscular TCr stores at days 1 and 5 compared with day 0, as well compared with the placebo trial, indicating a response to the supplementation period. Initial examination of gene expression by two-way (treatment x time) ANOVA showed no statistical differences for any of the genes (Fig. 5), although there was a trend (P = 0.08) for a main effect for time for GAPDH. Inspection of the 95% CI indicated that GAPDH was the most variable gene, with the remaining genes behaving in a similar manner. The variability in the gene expression of GAPDH would indicate that this is not a good choice of housekeeping gene and that its use may mask differences that are in fact present. This would suggest that ß2M, ß-actin, or CYC could be appropriate housekeeping genes for the present study.
Examination of housekeeping genes using normalized CT values.
Since initial examinations suggested that only GAPDH showed variable expression, ß2M, ß-actin, and CYC were used for the next step of normalizing the 2-CT data to 2-
CT values. Normalization of the ß-actin and CYC data to each other showed a similar stability in their expression. Interestingly, when the ß-actin and CYC were normalized to ß2M, their expression did not change (Figs. 6, G and H). Additionally, the expression of ß2M normalized to CYC was not different (Fig. 6E), although the expression of ß2M was different between days 0 and 5 in the placebo trial when normalized to ß-actin (Fig. 6B). The inconsistent expression of ß2M suggests that it would not be an appropriate choice of housekeeping gene compared with ß-actin and CYC. This discrepancy also highlights the importance of the correct choice of housekeeping gene, particularly when only small differences in the order of twofold changes may be present. It could in fact support the suggestion of employing two housekeeping genes (28), as well as using various baselines for normalizing data, as demonstrated by Hortobagyi and colleagues to validate their data (16). The change in ß2M, a gene involved in immune response (1), in the placebo trial at day 5, but not the CR trial, could suggest that this gene is responding to the exercise undertaken during the trial and that this effect is attenuated by the Cr. Examination of this possibility is outside the scope of this paper; however, further investigation is warranted. There were tendencies for main effects for time for GAPDH when normalized to either ß2M or CYC. A significant interaction with no changes detected by post hoc analysis was seen when GAPDH was normalized to ß-actin. Examination of the GAPDH data indicated it was the most variable gene measured in this study. The variable nature of this gene across various interventions and between individuals is highlighted in a recent review (5). As such, its use as a housekeeping gene has been questioned previously, and it would seem that CrS in combination with high-intensity exercise also introduces wide variation in the expression of GAPDH and it would not be recommended as a housekeeping gene for this intervention. Interestingly, the only other studies that have examined the effect of CrS on skeletal muscle gene expression have adopted GAPDH as their endogenous control (31). Justification of using GAPDH due to its constitutive nature was based on cultured cell data (31). Our results point to the possibility that the results may have been affected by large variations in the choice of housekeeping gene in those studies.
Power analyses.
RT-PCR is typically associated with large variability (5), although far less than other methods of mRNA detection. A previous report examining housekeeping genes suggested that for most genes, less than fourfold differences are probably not biologically significant (30); however, it should be added that we currently are unable to say what magnitude of fold difference in gene expression would be required to represent a biologically significant outcome. Our data indicate that, with real-time RT-PCR, approximately a twofold or greater difference in the gene expression of ß-actin, ß2M, CYC, or GAPDH is measurable with 0.8 power and 95% confidence, for n = 8. For 28S a six- to eightfold difference would be required for detection at this power level and sample size.
Conclusion.
We have demonstrated the quantitative and linear detection of amplified products from human skeletal muscle using SYBR Green 1 chemistry and real-time RT-PCR and the variability of triplicate readings using this method. When taking into account all experiments of the present study, it would appear that ß-actin and CYC would be the most valid choices of housekeeping genes for real-time RT-PCR analysis of human skeletal muscle samples exposed to short-term CrS and high-intensity exercise.
 |
ACKNOWLEDGMENTS
|
---|
We thank Dr. Brant Bassam for discussions of the project, Rebecca Tunstall for assistance in the pilot work in this study and reading of the manuscript, and Janelle Mollica for reading of the manuscript.
This work was supported by a grant from the School of Health and Behavioural Sciences, Deakin University, Australia.
Additional resources on the World Wide Web may be found at the Entrez Blast web site (http://www.ncbi.nlm.nih.gov/BLAST/) and at the Entrez Nucleotides database (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=Nucleotide).
 |
FOOTNOTES
|
---|
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: R. T. Snow, School of Health Sciences, Deakin Univ., Burwood, Victoria 3125, Australia (E-mail: rsnow{at}deakin.edu.au).
10.1152/physiolgenomics.00060.2002.
 |
References
|
---|
- Balint E. Role of beta-2-microglobulin in the immune response in renal osteodystrophy. Semin Dial 14: 113116, 2001.[ISI][Medline]
- Blomstrand E and Ekblom B. The needle biopsy technique for fibre type determination in human skeletal muscle: a methodological study. Acta Physiol Scand 116: 437442, 1982.[ISI][Medline]
- Boivin M, Camirand A, Carli F, Hoffer LJ, and Silva JE. Uncoupling protein-2 and -3 messenger ribonucleic acids in adipose tissue and skeletal muscle of healthy males: variability, factors affecting expression, and relation to measures of metabolic rate. J Clin Endocrinol Metab 85: 19751983, 2000.[Abstract/Free Full Text]
- Bonne G, Seibel P, Possekel S, Marsac C, and Kadenbach B. Expression of human cytochrome c oxidase subunits during fetal development. Eur J Biochem 217: 10991107, 1993.[Abstract]
- Bustin SA. Absolute quantification of mRNA using real-time reverse transcription polymerase chain reaction assays. J Mol Endocrinol 25: 169193, 2000.[Abstract/Free Full Text]
- Collantes-Fernandez E, Zaballos A, Alvarez-Garcia G, and Ortega-Mora LM. Quantitative detection of Neospora caninum in bovine aborted fetuses and experimentally infected mice by real-time PCR. J Clin Microbiol 40: 11941198, 2002.[Abstract/Free Full Text]
- Curry J, McHale C, and Smith MT. Low efficiency of the Moloney murine leukemia virus reverse transcriptase during reverse transcription of rare t(8;21) fusion gene transcripts. Biotechniques 32: 768, 770, 772, 754765, 2002.[ISI][Medline]
- de Leeuw WJ, Slagboom PE, and Vijg J. Quantitative comparison of mRNA levels in mammalian tissues: 28S ribosomal RNA level as an accurate internal control. Nucleic Acids Res 17: 1013710138, 1989.[ISI][Medline]
- Febbraio MA, Flanagan TR, Snow RJ, Zhao S, and Carey MF. Effect of creatine supplementation on intramuscular TCr, metabolism and performance during intermittent, supramaximal exercise in humans. Acta Physiol Scand 155: 387395, 1995.[ISI][Medline]
- Febbraio MA and Koukoulas I. HSP72 gene expression progressively increases in human skeletal muscle during prolonged, exhaustive exercise. J Appl Physiol 89: 10551060, 2000.[Abstract/Free Full Text]
- Foss DL, Baarsch MJ, and Murtaugh MP. Regulation of hypoxanthine phosphoribosyltransferase, glyceraldehyde-3-phosphate dehydrogenase and beta-actin mRNA expression in porcine immune cells and tissues. Anim Biotechnol 9: 6778, 1998.[ISI][Medline]
- Gavin TP, Spector DA, Wagner H, Breen EC, and Wagner PD. Nitric oxide synthase inhibition attenuates the skeletal muscle VEGF mRNA response to exercise. J Appl Physiol 88: 11921198., 2000.[Abstract/Free Full Text]
- Gong Y, Cui L, and Minuk GY. Comparison of glyceraldehyde-3-phosphate dehydrogenase and 28S-ribosomal RNA gene expression in human hepatocellular carcinoma. Hepatology 23: 734737, 1996.[ISI][Medline]
- Hamalainen HK, Tubman JC, Vikman S, Kyrola T, Ylikoski E, Warrington JA, and 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 299: 6370, 2001.[ISI][Medline]
- Hespel P, Eijnde BO, Van Leemputte M, Urso B, Greenhaff PL, Labarque V, Dymarkowski S, Van Hecke P, and Richter EA. Oral creatine supplementation facilitates the rehabilitation of disuse atrophy and alters the expression of muscle myogenic factors in humans. J Physiol 536: 625633, 2001.[Abstract/Free Full Text]
- Hortobagyi T, Dempsey L, Fraser D, Zheng D, Hamilton G, Lambert J, and Dohm L. Changes in muscle strength, muscle fibre size and myofibrillar gene expression after immobilization and retraining in humans. J Physiol 524: 293304, 2000.[Abstract/Free Full Text]
- Lee PD, Sladek R, Greenwood CM, and Hudson TJ. Control genes and variability: absence of ubiquitous reference transcripts in diverse mammalian expression studies. Genome Res 12: 292297, 2002.[Abstract/Free Full Text]
- Lekanne Deprez RH, Fijnvandraat AC, Ruijter JM, and Moorman AFM. Sensitivity and accuracy of quantitative real-time polymerase chain reaction using SYBR green I depends on cDNA synthesis conditions. Anal Biochem 307: 6369, 2002.[ISI][Medline]
- Lowry OH and Passoneau JV. A Flexible System of Enzymatic Analysis. New York: Academic, 1972.
- McKenna MJ, Morton J, Selig SE, and Snow RJ. Creatine supplementation increase muscle total creatine, but not maximal intermittent exercise performance. J Appl Physiol 87: 22442252, 1999.[Abstract/Free Full Text]
- Moody A, Sellers S, and Bumstead N. Measuring infectious bursal disease virus RNA in blood by multiplex real-time quantitative RT-PCR. J Virol Methods 85: 5564, 2000.[ISI][Medline]
- Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res 29: E45E45, 2001.[Medline]
- Schmittgen TD, Zakrajsek BA, Mills AG, Gorn V, Singer MJ, and Reed MW. Quantitative reverse transcription-polymerase chain reaction to study mRNA decay: comparison of endpoint and real-time methods. Anal Biochem 285: 194204, 2000.[ISI][Medline]
- Simpson DA, Feeney S, Boyle C, and Stitt AW. Retinal VEGF mRNA measured by SYBR green I fluorescence: a versatile approach to quantitative PCR. Mol Vis 6: 178183, 2000.[ISI][Medline]
- Slack JL, Bi W, Livak KJ, Beaubier N, Yu M, Clark M, Kim SH, Gallagher RE, and Willman CL. Pre-clinical validation of a novel, highly sensitive assay to detect PML-RARalpha mRNA using real-time reverse-transcription polymerase chain reaction. J Mol Diagn 3: 141149, 2001.[Abstract/Free Full Text]
- Snow RJ, McKenna MJ, Selig SE, Kemp J, Stathis CG, and Zhao S. Effect of creatine supplementation on sprint exercise performance and muscle metabolism. J Appl Physiol 84: 16671673, 1998.[Free Full Text]
- Suzuki T, Higgins PJ, and Crawford DR. Control selection for RNA quantitation. Biotechniques 29: 332337, 2000.[ISI][Medline]
- Thellin O, Zorzi W, Lakaye B, De Borman B, Coumans B, Hennen G, Grisar T, Igout A, and Heinen E. Housekeeping genes as internal standards: use and limits. J Biotechnol 75: 291295, 1999.[ISI][Medline]
- Wadley GD, Tunstall RJ, Sanigorski A, Collier GR, Hargreaves M, and Cameron-Smith D. Differential effects of exercise on insulin-signaling gene expression in human skeletal muscle. J Appl Physiol 90: 436440, 2001.[Abstract/Free Full Text]
- Warrington JA, Nair A, Mahadevappa M, and Tsyganskaya M. Comparison of human adult and fetal expression and identification of 535 housekeeping/maintenance genes. Physiol Genomics 2: 143147, 2000.[Abstract/Free Full Text]
- Willoughby DS and Rosene J. Effects of oral creatine and resistance training on myosin heavy chain expression. Med Sci Sports Exerc 33: 16741681, 2001.[ISI][Medline]
- Wyss M and Kaddurah-Daouk R. Creatine and creatinine metabolism. Physiol Rev 80: 11071213, 2000.[Abstract/Free Full Text]
- Yamada H, Chen D, Monstein HJ, and Hakanson R. Effects of fasting on the expression of gastrin, cholecystokinin, and somatostatin genes and of various housekeeping genes in the pancreas and upper digestive tract of rats. Biochem Biophys Res Commun 231: 835838, 1997.[ISI][Medline]
- Zhong H and 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 259: 523526, 1999.[ISI][Medline]