A Reliable Strategy for Validation of Real-Time Quantitative RT-PCR Amplicons from Linearly Amplified Human Hematopoietic Stem Cell mRNA.

Real-time quantitative RT-PCR (qRT-PCR) is a powerful tool for measuring or validating gene expression. Enrichment for the most primitive hematopoietic progenitors yields small cell numbers and thus small amounts of mRNA. Linear amplification (required for gene expression analysis) amplifies the 3′ end of mRNA and imposes limitations on qRT-PCR primer design that may require the use of sub-optimal primers that can form primer dimers (PDs). Indeed, using linear amplification and qRT-PCR to assay expression of 31 genes of the bone morphogenetic protein (BMP) signaling cascade in CD34⁺/CD38⁻/lin⁻ human umbilical cord blood (UCB) progenitors, we found that 17/31 (55%) of products were false positives. To distinguish true target amplicons from PDs, we therefore used a five-step sequential strategy to initially generate a Primer Profile for each primer set:

1. water blank and positive control

2. dissociation curve analysis

3. serial dilution

4. gel electrophoresis and

5. product sequencing, which we describe for one gene (BMP-2) that we found is variably expressed in UCB progenitors under different culture conditions.

Total RNA was obtained from 5–20 x 10³ UCB progenitors cultured in conditions that induced detectable BMP-2 expression (BMP-2⁺) and those that did not (BMP-2⁻), mRNA was linearly amplified (RiboAmp) and cDNA was synthesized using Invitrogen Superscript III. qRT-PCR was performed using Invitrogen SYBR Green qPCR SuperMix on an ABI 7900 HT Sequence Detection System. (1) Though primers were designed for the 3′ end of BMP-2 using the Primer 3 website (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www_slow.cgi) to identify optimal primers, an amplified product was seen in the water blank and in BMP-2⁻ cDNA. Nevertheless, the primers also amplified the correct product in the positive control (SaOS-2 human osteosarcoma cells) and in BMP-2⁺ cDNA. (2) Although the genuine and the false positive products appeared similar on amplification (ΔR_n vs cycle) plots, analysis of the derivative dissociation curves showed that their dissociation temperatures were distinct, being lower for the false positive PD (77°C vs 81°C). (3) Positive control and BMP-2⁺ cDNA yielded products with the expected one-cycle increase in cycle threshold (CT) for each 2-fold serial dilution, whereas PD product CTs did not change with serial dilution of BMP-2⁻ cDNA. Importantly, PDs also formed at higher dilutions (≥ 1:16) of BMP-2⁺ cDNA, possibly because of low target mRNA abundance. (4) Agarose gel electrophoresis of the amplified products from the positive control and BMP-2⁺ cDNA showed the expected amplicon size of 107 bp, whereas the PD product was smaller (50 bp). (5) Finally, sequence analysis of the products (at the Univ. of Minnesota DNA Sequencing and Analysis Center) confirmed that the 107 bp product was identical with the 3′ end of human BMP-2, whereas the PD product did not yield any sequence. Similar differences were seen between genuine amplicons and PDs for several other genes examined. Thus, for each primer set, the Primer Profile provided the melting temperatures of the genuine amplicon and the PDs. For subsequent experiments, we were then able to reliably predict which product was a genuine amplicon by inspection of its melting temperature. These findings demonstrate the critical importance of using such a strategy to detect false positive qRT-PCR results due to PDs when using linear amplification of mRNA from primitive hematopoietic progenitors and other rare cell subpopulations.