Relative real-time reverse transcription PCR (RT-PCR) has become an important tool for quantifying changes in messenger RNA (mRNA) populations following differential development or stimulation of tissues or cells. However, the best methods for conducting such experiments and analyzing the resultant data remain an issue of discussion. In this report we describe an appropriate experimental methodology and the computer programs necessary to generate a meaningful statistical analysis of the combined biological and experimental variability in such experiments. Specifically, logarithmic transformations of raw fluorescence data from the log-linear portion of real-time PCR growth curves for both target and reference genes are analyzed using a SAS/STAT Mixed Procedure program specifically designed to give a point estimate of the relative expression ratio of the target gene with associated 95% confidence interval. The program code is open-source and is printed in the text.

Introduction

Relative real-time reverse transcription PCR (RT-PCR) has become an important method for determining messenger RNA (mRNA) expression levels, both in its own right (1,2) and as a tool for the validation of microarray experiments (3). However, the best approach for the analysis of such data remains a topic of discussion (4,5). Recently, we published a procedure for determining the relative expression ratio for a target mRNA under two different treatment conditions directly from the raw real-time RT-PCR data without the use of a standard curve (6). Our method exploited a modification of the procedure originally proposed by Gentle et al. (7) for determining PCR efficiency and produces results that are both accurate and statistically verified. In the process of testing our method we demonstrated small but statistically significant day effects, which may call into question the use of a standard curve to determine efficiency when that efficiency will be applied to PCRs run on a different day. Because our procedure determines relative expression using slope and intercept data from each experiment individually rather than using a cycle threshold (C_t) and a general efficiency, this problem is effectively circumvented (6).

However, the statistical method used in our previously published procedure assumes that there are no confounding correlations between any pair of target and reference gene values. While this simple assumption is valid for the analysis of sample replicates, it is not valid when analyzing data from multiple independent samples (e.g., when each RT-PCR derives from a distinct RNA, perhaps isolated from individual mice or tissue samples that have been exposed to distinct treatment conditions), a situation frequently encountered in gene expression studies and important for establishing the true biological variability in the system. In this case, each target and reference gene pair is entangled in a “sample effect” that should be taken into account when calculating variance.

Here we present a new method for calculating the point estimate of a relative gene expression ratio between two treatment populations and for assigning appropriate 95% confidence intervals to the estimate. Our procedure exploits Mixed Procedure algorithms built-in to the commercially available statistical program SAS, and thus should be of general applicability to anyone performing such experiments.

Materials and Methods

Animals

Parental mice were purchased from Jackson Laboratories (Bar Harbor, ME, USA) and bred at the Animal Core of the Nathan Shock Center at the University of Texas Health Science at San Antonio. All mice were fed ad libitum (AL) Harlan Teklad LM-485 mouse/rat sterilizable diet 7912 (Madison, WI, USA) until 6 weeks of age. At 6 weeks, half of the mice were allowed to continue on this diet until sacrificed. The remaining mice were calorie-restricted (CR) by limiting them to 60% of the mean food intake of group AL until sacrificed. All procedures involving the use of mice were approved by the Institutional Animal Care and Use Committee of the University of Texas Health Science Center and the Subcommittee for Animal Studies at the Audie L. Murphy Memorial Veterans Hospital.

Tissue Collection and RNA Preparation

Livers from 4- to 6-month-old male mice were collected. The tissues were quickly frozen in liquid nitrogen and stored at -80°C until RNA extraction. Total RNA was extracted from each liver as previously described (8).

Real-Time RT-PCR

The reverse transcription reaction was performed using 1 µg of DNase I (Invitrogen, Carlsbad, CA, USA) digested total RNA, random primers, and SuperScript® II RT (Invitrogen) in a total volume of 20 µL according to the protocol of the manufacturer. The cDNA was diluted to 6-fold for the real-time RT-PCR. Primers were designed using the OligoPerfect™ Designer (Invitrogen) and purchased from Invitrogen. The 18S ribosomal RNA (rRNA) was used as the reference gene for the target gene (NF κ light chain; NFK) normalization. PCR was carried out using a Smart Cycler® thermal cycler (Cepheid, Sunnyvale, CA, USA). Each PCR included 3.0 µL diluted cDNA, 2.5 µL 10× PCR buffer without MgCl₂ (Sigma, St. Louis, MO, USA), 1.0 µL 25 mmol/L MgCl₂, 0.5 µL 10 mmol/L dNTPs (Invitrogen), 1.0 µL 0.5 µmol/ L each primer, 0.25 µL Taq DNA polymerase (Sigma), 0.2 µL 300 g/L bovine serum albumin (BSA; Sigma), 2.5 µL SYBR® Green I (Molecular Probes, Eugene, OR, USA) at a concentration of 1:4000 of the commercial stock, and 13.05 µL AccuGENE® Molecular Biology Grade water (Cambrex Bio Science, Rockland, ME, USA). The thermal cycling parameters included a 94°C heating step for 1 min at the beginning of every run. The tubes were then cycled 40 times at 94°C for 30 s, annealed at the optimal annealing temperature of each primer set (18S = 62°C; NFK = 59°C) for 60 s, and extended at 72°C for 60 s. Optical data were collected during the annealing step. The specificity of the reaction was monitored by melting curve analysis to avoid nonspecific signals.