to BioTechniques free email alert service to receive content updates.
Comparison of mRNA gene expression by RT-PCR and DNA microarray
 
Wiguins Etienne1, Martha H. Meyer1, Johnny Peppers2, and Ralph A. Meyer1
Full Text (PDF)

Studies of the linearity of the PCR at a constant number of PCR cycles have suggested a linear relationship of concentration plotted against resulting amplimer signal levels in other laboratories (16,17,18) and in our own (19). On the other hand, studies of mRNA quantified by microarray have also found a linear relationship between the microarray signal and the mRNA concentration (15,20). The strategy in the design of our PCRs had been to ensure that the peak expression levels were on the linear range of amplification for each gene (9). This may have resulted in missing the detection of low expression levels by RT-PCR. It may be that the dynamic range of microarrays is better than that of PCR for some genes with moderate expression levels.

There was poor agreement between the two methods for some genes, and two hypotheses were explored to explain this. First, the effect of an increased number of absence calls by microarray was tested for the 16 genes with moderate expression levels and with PCR primers that overlapped the location of the microarray probe sets. For these genes, an increased number of absent calls (≥7) among the 10 data points for each gene significantly (P < 0.05) reduced the correlation coefficient (Table 2).





Second, a test was made of the effect of probe location. The PCR primers tended to be at the 5′ end of the coding region, while the microarray probe sets tended to be at the 3′ end. Only three genes had microarray probes outside the coding sequence in the 3′ untranslated region. For each gene, the PCR primers and the microarray probes were located, and the separation in number of nucleotides was measured. Eleven genes with moderate expression levels and with a low number of absent calls by microarray were studied. For these genes, a separation of more than 1000 nucleotides between PCR primer location and microarray probe set location resulted in a lowering of the correlation coefficient (Table 2) that approached significance (P = 0.06). In this analysis, there were no samples with a nucleotide separation between 400 and 1000 nucleotides. It would seem that measuring an intact mRNA that supports protein synthesis would yield approximately equal results regardless of the region being probed. However, if mRNA fragmentation or alternative splicing had occurred within the cells, some sequences may be more abundant than others. It is not clear whether this is the source for the variation in the results.

All 26 genes are shown in Figure 2 with the correlation coefficients plotted against mean microarray values. Points are identified that have a large number of absent calls, a large nucleotide separation between PCR and microarray probe sets, or both. All genes with low mean microarray fluorescence had poor agreement between the two methods. The best agreements came for genes with mean fluorescence levels between 100 and 5000. While increased absent calls or probe separation lowered the correlation coefficients for some genes, there were exceptions. This variation makes the interpretation gene-specific. Valid physiological interpretations can be drawn from the data for some genes, even though the signal level is low, while other individual genes may be difficult to interpret. While the microarray scanner and associated software assign a signal value to each gene, the finding of an absent call should give great caution that the data may not accurately reflect mRNA expression levels.





There was no apparent explanation for the poor agreement for osteocalcin (Figure 1C). This gene had no absent calls, and there was overlap between the PCR primers and the microarray probes. However, there was a high signal value. This was not at the upper limit of the fluorescence measurements, since genes regularly approach 50,000 units in fluorescence. Histograms of the data distribution did not reveal any indication of truncation at the high end of the data; thus, the detector was not saturated. It might be that the probe sets on the array are being saturated for this particular gene. It may be that saturation is not at a given microarray signal value, but rather is a function of biotin labeling of the cRNA. A gene with lower biotin labeling may saturate its probe sets at a lower fluorescence level than another gene with higher biotin labeling.

Relatively few studies have been done to compare results obtained by microarray to results obtained by other methods for the measurement of mRNA levels. Good agreement was found between commercial oligonucleotide microarrays and custom-made spotted microarrays, but, for genes that increased in abundance, both microarrays estimated smaller increases in mRNA levels than those measured by real-time PCR (4). When results from spotted microarrays were compared to results obtained by Northern analysis, good agreement was found between the two methods with similar sensitivity levels (6). Ueda et al. (5) studied the diurnal rhythm of gene expression in the suprachiasmatic nuclei of mice. Several of the genes, which had a diurnal rhythm on oligonucleotide arrays, also had similar diurnal rhythms by quantitative PCR (5). Serial analysis of gene expression has also been compared to oligonucleotide microarray, and genes were ranked in almost the same order of abundance by the two methods (21).

In conclusion, there was good agreement for the measurement of mRNA gene expression between RT-PCR and DNA microarray for genes with moderate levels of expression that have PCR primers located close to the microarray probes. Genes with very high or low levels of expression, or those with larger separation between the location of the PCR primers and microarray probes, often had reduced agreement between the two methods.

Acknowledgments

We thank the North Carolina Biotechnology Center and the Orthopaedic Trauma Association for their partial support for this study. We also thank Ms. Carolyn Ayers for her secretarial support.

References
1.) Jurecic, R., and J.W. Belmont. 2000. Longdistance DD-PCR and cDNA microarrays. Curr. Opin. Microbiol. 3:316-321.

2.) Meyer, R.A., M.H. Meyer, M. Tenholder, S. Wondracek, R. Wasserman, and P. Garges. 2003. Gene expression in older rats with delayed union of femoral fractures. J. Bone Joint Surg. Am. 85A:1243-1254.

3.) Freeman, W.M., D.J. Robertson, and K.E. Vrana. 2000. Fundamentals of DNA hybridization arrays for gene expression analysis. BioTechniques 29:1042-1055.

4.) Yuen, T., E. Wurmbach, R.L. Pfeffer, B.J. Ebersole, and S.C. Sealfon. 2002. Accuracy and calibration of commercial oligonucleotide and custom cDNA microarrays. Nucleic Acids Res. 30:1-9.

5.) Ueda, H.R., W. Chen, A. Adachi, H. Wakamatsu, S. Hayashi, T. Takasugi, M. Nagano, K.I. Nakahama. 2002. A transcription factor response element for gene expression during circadian night. Nature 418:534-539.

6.) Taniguchi, M., K. Miura, H. Iwao, and S. Yamanaka. 2001. Quantitative assessment of DNA microarrays—comparison with Northern blot analyses. Genomics 71:34-39.

7.) Meyer, R.A., P.J. Tsahakis, D.F. Martin, D.M. Banks, M.E. Harrow, and G.M. Kiebzak. 2001. Age and ovariectomy impair both the normalization of mechanical properties and the accretion of mineral by the fracture callus in rats. J. Orthop. Res. 19:428-435.

8.) Meyer, R.A., M.H. Meyer, L.S. Phieffer, and D.M. Banks. 2001. Delayed union of femoral fractures in older rats: decreased gene expression. BMC Musculoskelet. Disord. 2:2.

9.) Desai, B.J., M.H. Meyer, S. Porter, J.F. Kellam, and R.A. Meyer. 2003. The effect of age on gene expression in adult and juvenile rats following femoral fracture. J. Orthop. Trauma 17:689-698.

10.) Fodor, S.P.A., J.L. Read, M.C. Pirrung, L. Stryer, A.T. Lu, and D. Solas. 1991. Light-directed, spatially addressable parallel chemical synthesis. Science 251:767-773.

11.) Lipshutz, R.J., S.P.A. Fodor, T.R. Gingeras, and D.J. Lockhart. 1999. High density synthetic oligonucleotide arrays. Nat. Genet. (Suppl.) 21:20-24.

12.) Meyer, R.A., B.R. Desai, and M.H. Meyer. 2003. Age alters mRNA gene expression in the fracture callus of rats by microarray analysis. Trans. Orthop. Res. Soc. 28:489.

13.) Zar, J.H. 1984. Biostatistical Analysis, 2nd ed. Prentice-Hall, Englewood Cliffs.

14.) Bolstad, B.M., R.A. Irizarry, M. Astrand, and T.P. Speed. 2003. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 19:185-193.

15.) Irizarry, R.A., B. Hobbs, F. Collin, Y.D. Beazer-Barclay, K.J. Antonellis, U. Scherf, and T.P. Speed. 2003. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4:249-264.

16.) Hayward, A.L., P.J. Oefner, D.B. Kainer, C.A. Hinojos, and P.A. Doris. 1999.Kinetics of competitive reverse transcriptase-PCR. In M.A. Innis, D.H. Gelfand, and J.J. Sninsky (Eds.) PCR Applications, Protocols For Functional Genomics. Academic Press, San Diego:231-261.

17.) Piatak, M., K.C. Luk, B. Williams, and J.D. Lifson. 1997.Quantitative competitive polymerase chain reaction for accurate quantitation of HIV DNA and RNA species. In J.W. Larrick (Ed.) The PCR Technique: Quantitative PCR. Eaton Publishing, Westborough:43-66.

18.) Zhang, X.Y., and M. Ehrlich. 1997.Detection and quantitation of low numbers of chromosomes containing bcl-2 oncogene translocations using semi-nested PCR. In J.W. Larrick (Ed.) The PCR Technique: Quantitative PCR. Eaton Publishing, West-borough:297-310.

19.) Meyer, R.A., C.G. Young, M.H. Meyer, P.L. Garges, and D.K. Price. 2000. Effect of age on the expression of Pex (Phex) in the mouse. Calcif. Tissue Int. 66:282-287.

20.) Heller, R.A., J. Allard, F. Zuo, C. Lock, S. Wilson, P. Klonowski, H. Gmuender, H. Van Wart, and R. Booth. 1999.Gene chips and microarrays: applications in disease profiles, drug target discovery, drug action and toxicity. In M. Schena (Ed.) DNA Microarrays, A Practical Approach. Oxford University Press, New York:187-202.

21.) Gnatenko, D.V., J.J. Dunn, S.R. McCorkle, D. Weissmann, P.L. Perrotta, and W.F. Bahou. 2003. Transcript profiling of human platelets using microarray and serial analysis of gene expression. Blood 101:2285-2293.

  1    2    3