Which Reference Gene?
Quantitative gene-expression studies performed with RT-qPCR require normalization of target genes using reference genes that are stable and independent of variables tested in experiments. However, there is no one reference gene that is universally applicable to every study, says Virginia Rebecca Falkenberg, Ph.D., microbiologist, Centers for Disease Control.
“Although a number of reference genes have been described for a wide variety of tissues and experimental conditions, they are not universally suitable or necessarily even present in all samples. For example, common reference genes are lacking for RNA derived from whole blood as well as isolated peripheral blood mononuclear cells (PBMC). Further, there are no suitable reference genes for analysis of many chronic and neuropsychiatric diseases for which blood may be the only sample to study gene expression as a systemic sensor of multisystem pathology.”
Dr. Falkenberg and associates initially were working on large-scale clinical studies for biomarker identification for various diseases. “We had received both PBMC and whole blood from patients that were obtained at the same time points. It is very unusual to have access to both. We set out to identify a suitable common reference gene. We first performed a literature review and analyzed our unpublished microarray datasets. We generated a set of six possible reference genes. We then used two algorithms (geNorm and NormFinder) to find the most stable genes for both PBMC and whole blood RNA.”
Next, Dr. Falkenberg changed an additional analysis program in an important and instructive way. “I modified a method called single control normalization error (E). Data is collected for several potential reference genes. To select the best (most stable) genes usually one takes each proposed reference gene and performs statistical analysis comparing one gene at a time to that potential reference gene. I decided to perform analyses by comparing the average of 2–3 genes simultaneously with the proposed reference gene.
“To allow the selection of a reference gene that works in both whole blood and PBMC, I took a single gene or average and the comparison was between the two different tissues instead of two different genes. This greatly improved the process allowing us to identify Phosphoglycerate kinase 1 (PGK1) as one of the optimal normalization genes for both whole blood and PBMC RNA.”
The idea of being able to utilize a single reference gene provides a much more cost-effective means for characterizing gene-expression profiles. “Using a single reference gene also improves the ability to compare gene-expression results using blood RNA collected and processed by different methods with the intention of biomarker discovery.”
Computation is a crucial step to measure gene expression for RT-qPCR. “Data analysis is not straightforward since it requires many iterative computations for data normalization and optimization,” suggests Daijun Ling, Ph.D., scientist, USC Davis School of Gerontology, University of Southern California.
“Oftentimes, PCR machines provide instrumental software that also functions in data analysis. Many just use what the companies provide rather than more suitable software.”
Dr. Ling developed an all-in-one computer program based on statistical software called SAS (SAS Institute). “I called this SASqPCR. It provides multiple macros to assess PCR efficiencies, validating reference genes, optimizing data normalizers, normalizing confounding variations across samples, and statistical comparison of target gene expression in parallel samples.”
SASqPCR does not require the user be an expert programmer. Rather, the program provides a dynamic interface for user-controllable customization based on specific research aims, data quality, and experimental design. Users can easily test a variety of combinations of different analytical scenarios or even customized analytical processes.
“The advantage of the SASqPCR program for analysis of RT-qPCR data is its versatility. First, SASqPCR has no limitation to the size of the dataset. It also can be used for profiling transcriptomes generated by RT-qPCR, as well as population genetic studies with hundreds of subjects.
“Second, SASqPCR can be easily combined with other SAS procedures for advanced statistical analysis, for example, cluster analysis for transcriptome profiling, or association studies for connecting gene-expression variation with particular biomedical conditions. Users are the final decision-makers in analysis of their data.”
Multiple combined datasets or PCR array data can be analyzed. “Users can manage and analyze data in a more traditional way instead of having to rely on proprietary instruments or even plate-based data formats. The final results are basically numeric values that provide a much more straightforward and accurate means to determine how to visualize their results.”
Malaria (Plasmodium) continues to remain one of the most devastating infectious diseases on the planet. While new initiatives aim to eradicate the disease, one important aspect of these projects is the necessity for improving diagnosis and surveillance methods. Despite inroads in molecular approaches, the gold standard for malaria diagnosis, epidemiology, and clinical trial efficacy evaluation continues to be microscopy, in particular the Giemsa-stained thick blood smears.
“Microscopy suffers from limitations such as sensitivity, with the average microscopist able to detect about 100 parasites per microliter of blood, but the threshold for fever and clinical disease is less than 10 parasites per microliter for nonimmune patients,” says MAJ Edwin Kamau, Ph.D., chief of the department of molecular diagnostics, Malaria Vaccine Branch at Walter Reed Army Institute of Research.
“As we move into an era working to control and eliminate malaria, it is critical that we develop a more sensitive and higher throughput means for detecting smaller amounts of the parasite useful for even subclinical infections. The malaria vaccine and drug development program at the Walter Reed Army Institute of Research is seeking better means of detection, particularly for those subclinically infected. To do this, we sought to develop a highly sensitive genus-specific RT-qPCR assay to detect Plasmodium.
“A number of such assays have been developed that can measure infection more than 1,000 times more sensitively than microscopy or even antigen detection tests. However, one challenge is that most qPCRs target the DNA of the multicopy 18S rRNA genes, which have enough genetic variation to be problematic. It is therefore important that qPCR target amplification of conserved regions of 18S rRNA genes.”
Dr. Kamau and colleagues developed a genus-specific reverse transcriptase RT-qPCR assay to detect Plasmodium. “The assay detects both rRNA and DNA and takes advantage of the high copy numbers of rRNA in the genome of the parasite. It can detect 0.002 parasites per microliter of blood, one of the lowest levels detected so far.”
According to Dr. Kamau, introduction of the reverse transcriptase step was key. “We were able to improve the assay by approximately 10-fold by adding this step. We verified our hypothesis that RT-qPCR was more reliable than microscopy at lower parasite densities, but not at higher densities. Thus another important finding was that dilution ranges of the sample are important.”
Before molecular diagnostic approaches replace microscopy, there are a number of remaining challenges to overcome including the need for stringent standardization of the methodology.