Detection of Prostate Cancer Biomarkers by Expression Microarray Analysis of Transplex® WTA2-Amplified FFPE Tissue RNA

By: Ken Heuermann and Brian Ward, Sigma-Aldrich Corporation

Abstract

The Transplex® Complete Whole Transcriptome Amplification (WTA2) Kit has been enhanced to effectively amplify damaged RNA. Total RNA samples isolated from matched FFPE (formalin-fixed paraffin-embedded) and frozen prostate tissues (malignant and normal) were amplified with the Transplex WTA2 Kit and analyzed on Agilent Whole Genome expression microarrays. A call rate of >65% was observed for combined FFPE and frozen expression differentials (Malignant/Normal), with >90% commonality between the FFPE and frozen data sets. Amplification reproducibility of experimental replicates was demonstrated by principal component analysis. Pair-wise comparison of FFPE malignant and normal expression (p<0.0001) resulted in detection of 66 known prostate cancer biomarkers. Similar microarray pair-wise analysis of the original matched FFPE RNAs amplified with the Nugen WT-Ovation™ FFPE Kit at a significantly lower level of confidence (p ≤ 0.05) revealed only a single array feature, disallowing a comprehensive comparison with Transplex WTA2. These results confirm the utility of the Transplex WTA2 Kit for amplification of damaged DNA.

Introduction

Over the last two decades, methods for amplification of impracticably small RNA samples have been developed to generate sufficient start material for microarray applications, including gene expression studies, genotyping, and amplification of sequences isolated by ChIP. Pursuant with this objective and for downstream qPCR studies, methods capable of rescuing and amplifying RNA isolated from formalin-fixed paraffin-embedded (FFPE) tissues have also been developed. The earliest popular platforms for RNA amplification were based on the linear model proposed by Van Gelder, et al.1 More recently, another linear method has been designed around the concept of a single amplification primer.2,3 Neither of these methodologies, however, is sufficiently sensitive for amplification of highly-degraded RNA or low copy-number transcripts.

Kamberov, et al4 adapted the single primer concept to exponential amplification. The result is a robust, sensitive PCR-based amplification method, marketed by Sigma-Aldrich Corporation as the Transplex WTA1 Whole Transcriptome Amplification Kit. Agilent Technologies recently posted an application note, Gene Expression Microarray Analysis of Archival FFPE Samples,5 including the Transplex WTA1 Kit in their workflow.

Transplex WTA methodology (Figure 1) offers clear advantages over competing platforms:

(1) the amplification product is double-stranded cDNA, stable for months at –20 C; (2) the procedure is fast, 4 hours versus 48 hours, providing ~1000X amplification; (3) the quasi-degenerate library synthesis primer design eliminates 3’ bias (data not presented here); (4) while a single non-self-complementary amplification primer provides optimal, uniform amplification.

Moreover, to provide for efficient amplification of degraded RNA, the Complete Transplex WTA2 Kit has been developed. Enhancements found in Transplex WTA2 Kit include (1) a re-engineered library synthesis primer design for efficient amplification of fragmented RNA, (2) amplification conditions optimized for polymerization through high AT- and GC-rich regions, and (3) redesign of the amplification primer and (4) reformulation of the amplification mix which allows for enhanced maintenance of the relative abundance of each transcript during the amplification process.

This study demonstrates the capability of the Transplex WTA2 Kit to amplify highly degraded RNA extracted from matched FFPE malignant and normal human prostate tissue samples. Microarray analysis of amplified RNA from FFPE malignant tissue identified known cancer biomarkers. Transplex WTA2 amplification results are also compared with those of the popular Nugen WT-Ovation™ FFPE RNA amplification kit.

Transplex WTA2 Methodology

Figure 1. Transplex WTA2 Methodology. Total RNA is combined with Library Synthesis Solution and heat-denatured. This is followed with the addition of Reaction Buffer and the strand-displacing Library Synthesis Enzyme, for “single-tube” reverse-transcription and second-strand cDNA synthesis (Omniplex™ Library). The 3’ end of the Library Synthesis Primer is “quasi-random” and substantially non-self-complementary, while the 5’ end is a single, constant non-self-complementary sequence that serves as the annealing site for the universal amplification primer.

Moreover, to provide for efficient amplification of degraded RNA, the Complete Transplex WTA2 Kit has been developed. Enhancements found in Transplex WTA2 Kit include (1) a re-engineered library synthesis primer design for efficient amplification of fragmented RNA, (2) amplification conditions optimized for polymerization through high AT- and GC- rich regions, and (3) redesign of the amplification primer and (4) reformulation of the amplification mix which allows for enhanced maintenance of the relative abundance of each transcript during the amplification process.

This study demonstrates the capability of the Transplex WTA2 Kit to amplify highly degraded RNA extracted from matched FFPE malignant and normal human prostate tissue samples. Microarray analysis of amplified RNA from FFPE malignant tissue identified known cancer biomarkers. Transplex WTA2 amplification results are also compared with those of the popular Nugen WT-Ovation™  FFPE RNA amplification kit.

Materials and Methods

Matched FFPE and frozen human prostate tissue samples were procured from the Human Tissue Resource Network, College of Medicine, Ohio State University. Highly degraded RNA was isolated from malignant and normal FFPE tissue following procedure provided in the Qiagen RNeasy™ RNA Extraction Kit for FFPE tissue samples (Qiagen No. 74404). Intact total RNA was isolated from matched frozen malignant and normal tissue using the GenElute™ Mammalian RNA Extraction Kit (Sigma-Aldrich Catalog No. RTN). In both instances, RNA was treated with RNase-free DNase (Sigma-Aldrich Catalog No. AMPD-1) to remove contaminating genomic DNA. Verification of DNA removal was confirmed by RT-PCR followed with qPCR utilizing cDNA-specific primer sets for β-actin and GAPDH.

Approximately 250 nanograms of RNA (Nanodrop) were evaluated by Agilent Bioanalyzer analysis. Five nanograms of intact RNA or 50 nanograms of FFPE RNA were amplified using the Transplex Complete WTA2 Kit. Each reaction was carried through two cycles beyond stationary amplification (“plateau”); all Transplex WTA2 reactions were stopped at cycle 17. Similarly, 50 nanograms of the same FFPE RNAs were amplified using the Nugen WT-Ovation™ FFPE System V2 Kit. Nugen RNA amplification and subsequent microarray analysis was performed by an independent commercial service provider who specializes in both Agilent and Nugen applications. Quantitative PCR was performed using SYBR® Green JumpStart™ Taq ReadyMix™ (Sigma-Aldrich Catalog No. S4438), with 250 nM primer concentration. Primer pairs, with the exception of the 18S rRNA set, are specific for cDNA detection.

Transplex-amplified RNA (double-stranded cDNA) was labeled with cy3 or cy5 using the Agilent Genomic DNA Labeling Kit Plus procedure (Agilent No. 5188-5309). Two amplified, differentially-labeled cDNA targets were combined, heat-denatured, applied to Agilent Whole Genome™ Arrays (Agilent No. G4112F), and incubated at 65 °C for 40 hours. Hybridization and subsequent wash procedures for the Agilent Oligonucleotide Array-Based CGH for Genomic DNA Analysis were followed, except for the omission of COT-1 DNA block during hybridization. Arrays were scanned (Agilent scanner model G2505B) and features extracted. Intensities representing non-uniform features or associated with local non-uniform background, in addition to “absent” and “saturated” array feature intensities, were removed using Microsoft Excel™. All intensities less than 100 were removed. Microarray features described as “statistically tested and analyzed” were then subjected to median normalization, Welch’s t-test, Benjamimi-Hochberg false discovery rate correction, and a ± 1.5-fold differential threshold using Genesifter™ (Geospiza) pair-wise comparison, except where otherwise indicated. Tested data was further analyzed using the Genesifter Intersector and Spotfire™ (TIBCO) software. Two-dimensional principal component analyses and cluster analysis of median-normalized intensities were also performed using Genesifter.

Agilent microarray probes corresponding to suspected prostate cancer biomarkers6 were determined. Biomarker probes detected for FFPE malignant sample transcripts and transcripts common to malignant and normal tissues associated were then identified and duplicates removed.

Results and Discussion

Preparation of Amplified RNA for Microarray Analysis. Total prostate RNA was extracted from matched frozen and FFPE tissue samples as described in Materials and Methods. Agilent Bioanalyzer traces of good quality frozen sample RNA are shown, with correspondingly high RIN values (Figure 2). Markedly degraded FFPE sample RNAs (RIN = 2.0) lacked any remnant of rRNA banding pattern. Following removal of contaminating genomic DNA, RNA samples were amplified using the Transplex Complete WTA2 Whole Transcriptome Amplification Kit (Figure 1). Reactions were allowed to proceed through two cycles of stationary amplification or “plateau”. GenElute mini-column-purified amplification product was evaluated by qPCR (Table 1) and agarose gel electrophoresis prior to microarray analysis (Figure 3). Amplified product from degraded template is of a perceptibly smaller size range, reflecting the degraded nature of the template.

Total prostate RNA

Figure 2. Total prostate RNA was extracted from matched frozen and FFPE tissue samples as described in Materials and Methods. Agilent Bioanalyzer RIN values for frozen sample RNA are shown. An RIN of 2.0 was determined for the highly degraded FFPE sample RNAs in a separate analysis.

 

C(t), 0.1
  18S β-actin GAPDH
Frozen Malignant 8.6 15.44 17.71
Frozen Normal 8.47 15.09 17.45
FFPE Malignant 11.32 22.6 23.41
FFPE Normal 10.65 20.56 21.82

Table 1. Quantitative PCR analysis of amplified RNA is performed for quality assessment. Five nanograms of frozen RNA samples were amplified using Transplex WTA2, while 50 nanograms of FFPE sample RNAs were amplified with the Transplex WTA2. Amplification reactions were stopped at cycle 17. Prior to labeling for microarray analysis, Transplex WTA2 amplification product was evaluated by qPCR using primers for 18S, GAPDH, and β-actin transcripts. C(t) differences between FFPE and frozen sample reflect reduced primer accessibility to intact template for the FFPE samples.

Electrophoresis analysis

Figure 3. Electrophoresis analysis shows the reduced size range for FFPE amplification product (1% agarose), reflecting the degraded condition of the RNA template prior to Transplex WTA2 amplification.

Principal component analysis

Figure 4. Principal component analysis (PCA) illustrates the reproducibility of Transplex WTA2 amplification. Replicate experiments (including separate amplification reactions) were performed on different days using the same RNA samples: Experiment 1 (n = 4), and Experiment 2 (n = 2). Amplification product was labeled and hybridized to Agilent Whole Human Genome Arrays, as described in Materials and Methods. PCA of all arrays were shown for Experiments 1 and 2 (A and B), and for both experiments combined (C). PCA is presented in (D) for experimental conditions for combined Experiments 1 and 2. (FP = FFPE; Fr = frozen, M = malignant; N = normal.)

Principal Component Analysis. Amplified RNA samples were labeled and hybridized to expression microarrays as described in Materials and Methods. Results of principal component analyses for the individual array targets (median-normalized intensities, >100) and experimental conditions are shown (Figure 4). Reproducibility for duplicate targets is most evident (Figure 4A, B). Good concordance is also observed for targets representing the same sample RNAs amplified on two separate occasions (Experiments 1 and 2, 4C, D). In addition vector component differences between Malignant and Normal targets or conditions were substantially maintained, when comparing results for FFPE and frozen samples. This global analysis demonstrates the reproducibility of the Transplex amplification process.

Differential Expression for FFPE and Frozen Matched Prostate Tissue Samples. Detected differential expression in FFPE and frozen prostate tissue samples (Experiment 1, log2 MalignantIntensities/NormalIntensities) was compared (Figure 5). Greater than 98 percent of FFPE expression differentials were detected in frozen tissue, with a total number of detected FFPE differentials approaching 62% of all unique biological probes (41,000) represented on the array. Stringent statistical testing (see Materials and Methods) was applied, and FFPE versus frozen differentials were plotted (Figure 6). For 41,000 microarray probes at a p ≤ 0.0001, 4.1 false reads are predicted; this was validated by a single reversed differential (Figure 5). Microarray features detected for malignant versus normal tissue samples for these FFPE and frozen data were also compared. A 57% concordance was found for FFPE and frozen malignant-vs.-normal intersections (Figure 7).

Differential expression for FFPE and frozen

Figure 5. Differential expression for FFPE and frozen prostate tissue samples (Experiment 1 log2 MalignantIntensities/NormalIntensities), comprising median-normalized intensities (>100) were compared using Genesifter Intersector (Experiment 1, n = 4). Greater than 98 percent of FFPE expression differentials were found to be common to frozen, whereas the differential call rate for FFPE and frozen were individually better than 60% of all unique probes (41000) represented on the array. (Note: Venn diagrams are not presented to scale.)

Differential expression for FFPE versus Frozen samples

Figure 6. Differential expression for FFPE versus Frozen samples (Experiment 1 log2 ratio of malignant and normal feature intensities) presented graphically, comprising median-normalized intensities >100, at an adjusted p ≤ 0.0001. Expression differentials were subjected to Welch’s t-test, Benjamini-Hochberg false discovery correction, and a threshold of ± 1.5. Four hundred 482 significant differentials were identified (R2 = 0.8944), with only a single data point indicating a reversal of expression polarity (arrow).

Venn diagrams are shown for detected FFPE and frozen malignant

Figure 7. Venn diagrams are shown for detected FFPE and frozen malignant versus normal microarray features (Experiment 1 median-normalized intensities >100), at an adjusted p value of 0.0001. Statistical analyses and corrections applied in Figure 5 were also applied here. At p ≤ 0.0001, a concordance of 57% exists for the two overlapping regions (malignant vs. normal) of the FFPE and frozen data sets comparisons.

Detection of Prostate Cancer Biomarkers. Microarray probes were identified for a list of proteins5 suspected to play a role in prostate malignancy (Table 2). FFPE array features associated with malignancy in Figure 6 were cross-referenced with these biomarker probes. Sixty-six markers were found to be detected in the tested data, from a combined 22 probes “unique for malignancy”, and 55 probes “common to malignant and normal” tissues. The discrepancy between 77 total detected probes and 66 identified markers lies in the fact that for several genes in the “common” set, as many as five probes encoding different regions of the transcript were detected. In addition, three markers were represented in both “unique” and “common” sets by at least one probe encoding different sequence than the probe or probes in the other set. Further investigation is warranted to determine whether this observation represents an alternative splicing event or genetic deletion, potentially related to prostate malignancy.

prostate-cancer-table

Table 2. Agilent microarray probes for proteins suspected to be markers for prostate cancer were identified. The panel of sixty-six biomarkers, above, represents detected microarray features derived from results in Figure 6: features common for FFPE malignant and normal samples, features unique to malignancy; and features detected in both data sets.

Comparison of the Transplex WTA2 and Nugen WT Ovation V.2 Kits. The capability of the Complete Transplex WTA2 Kit to amplify degraded RNA was compared with the leading linear single primer Isothermal amplification (SPIA) platform,2,3 the Nugen WT-Ovation V.2 kit. Amplification of FFPE sample RNA with the Nugen is observably less sensitive than with Transplex WTA2 (Figure 7). All array features detected by Nugen, except for a single probe, were also detected by Transplex WTA2. Transplex WTA2 also detected an additional 7150 features (27% more array features detected). Normalized intensities (>100) for malignant versus normal data were compared for Transplex WTA2- and Nugen-amplified FFPE RNA (Figure 9). A concordance of 63% was found for features represented in the intersecting regions.

Further evaluation of the Nugen kit proved to be impracticable, in that statistical testing was only able to detect a single probe that demonstrated significant variance between malignant and normal data sets at a p ≤ 0.05. This observation was in keeping with previous results indicating a lower sensitivity for the Nugen kit.

Cluster analysis (Figure 10) further emphasizes the overall dissimilarity in the data generated by the two amplification methods, and verifies the nominal difference between the malignant and normal data generated with the Nugen kit in comparison to Transplex WTA2 results.

Differential expression

Figure 8. Differential expression (Experiment 1 log2 Malignant/Normal, median-normalized intensities) for Transplex WTA2 frozen and FFPE were compared with Nugen-amplified FFPE RNA (same FFPE matched samples). All probes detected by the Nugen experiment, except one, were also detected in the Transplex FFPE data set, which included an additional 7150 features (27% additional data).

Overlap of malignant and normal detected array features

Figure 9. Overlap of malignant and normal detected array features, comprising median-normalized intensities >100, is shown for Transplex WTA2- and Nugen-generated microarray target. Though numerically similar, commonality between the Transplex WTA2 and Nugen overlapping regions is 63%. A statistical analysis of Nugen-amplified RNA, as performed for Transplex WTA2 in Figure 6 and Table 1, was not possible. Genesifter pair-wise analysis determined only a single detected array feature to be significantly variant between Nugen malignant and normal data at an adjusted p ≤ 0.05.

Cluster comparison of median-normalized intensities

Figure 10. Cluster comparison of median-normalized intensities (>100) for Transplex WTA2 (Experiment1) and the Nugen, looking at conditions, reveals variance between results achieved with the two amplification methods. Euclidean distance, Ward linkage, and gene centering were applied.

Conclusions

With respect to the additional array features detected by Transplex WTA2, it has been postulated that Transplex exponential amplification is capable of detecting basal expression. Stochastic in its nature,7 this residual expression can be readily eliminated by a combination of experimental replication, and stringent statistical testing and analysis.

In summary, the Transplex WTA2 Kit has been demonstrated to be exceedingly sensitive, reproducible, and capable of generating more statistically significant information than linear amplification. This has been demonstrated in this presentation with the detection of more than sixty known cancer markers in malignant FFPE prostate tissue. Finally, the Transplex WTA outperforms the SPIA amplification method.

Acknowledgements

We would like to acknowledge the members of Sigma-Aldrich Biotechnology R&D, in particular Patrick Sullivan, Ernie Mueller and current supervisor, Carol Kreader, for their assistance in the development of this product; and Sigma Marketing Communication Group for their assistance in the preparation of this poster.

Materials

     

References

  1. Van Gelder, R.N. et al. 1990. Amplified RNA synthesized from limited quantities of heterogeneous cDNA. Proc. Natl. Acad. Sci. USA 87: 1663-1667.
  2. Smith L, et al. 2003. Single primer amplification (SPA) of cDNA for microarray expression analysis Nucleic Acids Res. 31(3): e9.
  3. Dafforn A, et al. 2004. Linear mRNA amplification from as little as 5 ng total RNA for global gene expression analysis. Biotechniques. 37(5):854-7.
  4. Kamberov, E. et al. Amplification and analysis of whole genome and whole transcriptome libraries generated by a DNA polymerization process. US Patent Application 2007054311, 3/8/2007. (Previously, US Patent Application 20040209298, 12/14/2005.) EPO Application EP1604040, 12/14/2005. WIPO PCT WO2004081225, 9/23/2004).
  5. Anne Bergstom and Lucas Gary Lin. 2009. Gene Expression Microarray Analysis of Archival FFPE Samples. Application Note. Internet Address: http://www.chem.agilent.com/Library/applications/5990-3917en_hi.pdf . Agilent Technologies. Santa Clara, CA USA
  6. Supplemental: contact primary author for complete references for prostate cancer biomarkers: Ken Heuermann, kheuermann@sial.com
  7. Nygaard, V, and Hovig, E. 2006. Options available for profiling small samples: a review of sample amplification technology when combined with microarray profiling. Nucleic Acids Res. 34: 996-1014.

 

Related Links