| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Abstracts of Oak Ridge Posters |
(Department of Pathology, Stanford University, Stanford, CA;
aaddress correspondence to this author at: Stanford University, Lane Building, L217, Stanford, California 94305; fax 650-725-6902, email hsussman{at}stanford.edu)
Microarray analysis, which enables the comprehensive examination of many thousands of genes in a single experiment, is a promising method for furthering understanding of disease states. Because of the large amounts of probe required, however, microarray analysis has not been possible for small excision biopsies, fine needle aspirates, and microdissected tissue samples. Linear amplification of target cDNA with T7 RNA polymerase (1) is currently the most common method for the amplification of RNA for microarray analysis and has been validated (2) and optimized(3)(4). Other linear RNA amplification strategies have been developed (5)(6), but these do not generate sufficient amounts of probe for microarray analysis. DNA fragments have been used for enriching populations (7), cloning differences (8), and subtractive screening(9). Representational fragment amplification (RFA) is a method that we have developed for global amplification of cDNA as universally primed fragments. The product of RFA is double-stranded DNA, which can be directly labeled for microarray analysis, screened for genetic variation with traditional probes, analyzed with PCR-based protocols, or stored for future analysis.
To perform RNA isolation, we processed paired nondiseased and diseased cervical biopsy samples from patients diagnosed with squamous cell carcinoma of the cervix as previously described (10). The specimens were anonymized by ILS Bio or Genomics Collaborative and collected with patient consent in compliance with the company Institutional Review Boards and with the Code of Federal Regulations (CFR) 45CFR46.101B. Briefly, tissue samples frozen with liquid nitrogen were ground to a fine powder, transferred to 6-mol/L guanidine-HCL at room temperature, and homogenized by multiple passes through a syringe equipped with an 18-gauge needle. We isolated RNA with a Qiagen RNeasy Midi Kit.
For RFA cDNA synthesis, we used 5 µg of RNA from nondiseased and diseased tissues as templates in the Roche cDNA Synthesis System, according to manufacturer’s instructions, substituting 2 mmol/L PolyT18_DpnII/NlaIII-V (5'-GAG AGT GAG TGA TCA TGT TTT TTT TTT TTT TTT TTV-3') as the primer.
For in vitro transcription synthesis, we used 10 µg of total RNA and followed the protocols for the Affymetrix in vitro transcription (IVT) Kit. For microscale cDNA synthesis, we used 10 ng total RNA from Human Universal Reference RNA (UHRR) (Stratagene, Inc.) and Human Breast Carcinoma T-47 Cell Line Total RNA (Ambion, Inc.). We established a template for cDNA synthesis through dilution in 10 mmol/L Tris pH 8.0 containing 30 mg/L polyinosinic acid carrier (11). One of the volumes of the Roche reagent set were used. For microscale cDNA synthesis, we added 20 µg of T4gp32 (12) immediately before the addition of reverse transcriptase. Microscale cDNA samples were heat killed and were not treated with RNase or proteinase K.
For RFA amplicon synthesis, cDNA fractions (1/6 to 1/12) were digested with DpnII or NlaIII for 90 min at 37 °C, heat killed at 65 °C for 90 min, and ligated to 5 µg of the appropriate preassembled linker (3 to 16 hrs). The DpnII linker was assembled with R-BGL-24, sequence 5'-AGC ACT CTC CAG CCT CTC ACC GCA-3', and R-BGL-12, sequence 5'-GAT CTG CGG TGA-3') (9) and the NlaIII linker is assembled with: R-BGL-28_NlaIII, sequence 5'-AGC ACT CTC CAG CCT CTC ACC GCA CAT G-3' and R-Bgl-08_NlaIII, sequence 5'-TGC GTGA-3'). Linker-ligated cDNA dilutions were the templates for amplifications. Amplification was performed on 4 to 6 identical 100-µL tubes containing diluted template, 100 pmol/L R-BGL-24 primer, and (final concentration) 66 mmol/L Tris-HCl pH 8.8 at 25 °C, 16 mmol/L (NH4)2SO4, 4 mmol/L MgCl2, and 0.2 mmol/L each dNTP. The amplification tubes were incubated at 72 °C for 3 min before the addition of 5 units of Taq polymerase. The 72 °C incubation continued for 10 min before 20–28 cycles at 95 °C for 15 s and 72 °C for 3 min. The DpnII and NlaIII amplicons were pooled, phenol/chloroform extracted, and isopropanol precipitated and resuspended in 100 µL TE–1 (1 mM Tris pH 8.0, 0.1 mmol/L EDTA). The RFA amplicons were diluted in water and quantitated by A260 and checked for purity by A260/A280 ratio.
For microscale RFA amplicon synthesis, linker ligations contained 1 µg of linker in a 25 µL volume. We calculated the target for 25 cycles of RFA to be 15.8 pg of mRNA. The yield of double-stranded probe was 25 µg: net 1.5 million– fold.
Our improved precipitation protocols were performed with equal-volume isopropanol precipitations with 0.3 mol/L sodium acetate, pH 5.3, incubated at –80 °C for at least 2 h. Ethanol washes were 85% ethanol.
The primer designs and protocols for real-time quantitative reverse transcriptase (RT)-PCR were taken from previously published experiments (10), and the amplified segments were free of DpnII and NlaIII restriction sites (CCNB1 primers; 3–2 PRIME-195F: TGG TCT GGG TCG GCC TC, 3–2 PRIME-263R: TCG ACA TCA ACC TCT CCA ATC TT, 3–2 213FT: ACC TTT GCA CTT CCT TCG GAG AGC ATC). We used specific fluorescein/tetramethylrhodamine probes, and cycling was 95 °C, 15 s and 65 °C, 3 min for 40 cycles. For each independent gene assay, we used actin diseased/nondiseased (D/N) ratios from cDNA to normalize the gene D/N ratios of the amplicon. The mean normalization factor was 30%.
For random primer biotinylation, we biotinylated RFA amplicons in 4–6 independent, replicate BioPrime® DNA-labeling reactions (0.5 µg target) (13), following the manufacture’s protocols (Invitrogen, Inc), and obtained a mean 10-fold yield.
For all microarray analysis experiments, we followed the manufacturer’s hybridization and processing protocols for the HU133 plus 2.0 chips. We used Affymetrix Microarray Analysis Suite (GCOS v1.0) and ArrayAssist (3.3) (Stratagene, Inc) to import Affymetrix CEL files and to generate intensity values based on the robust multiarray average method (14). The relationships between the different platforms were assessed by Pearson correlation coefficients (15)(16).
The RFA method uses 2 enzymes to fragment the cDNA in a known, reproducible manner (Fig. 1A
). To give every cDNA fragment a common priming site at both ends, the cDNA fragments are ligated to compatible adaptors containing the R-BGL-24 universal primer sequence. Amplification of these smaller cDNA fragments proceeds exponentially and can be used to generate expression signatures. We used the synthesis rate definition for 5 units of Taq DNA polymerase to establish the goal of 5 µg of DNA per 100 µL (40 µg/8 tubes).
|
Complete real-time quantitative RT-PCR analysis of 6 gene transcripts for 3 paired cervical biopsy samples (10) are presented in Table 1
. These quantitative results demonstrate that RFA is a robust methodology that can produce accurate DNA signatures from limited amounts of starting material. For each patient, we compared the gene-fragment concentrations from the DpnII and NlaIII amplicons with gene concentrations in the original cDNA to determine whether these amplicons maintained the relative D/N expression ratios seen in the cDNA. For all patients the RFA amplification was >100 000-fold, and the mean CV for the all analyses was 11.8%.
|
Yields of RFA amplicon were determined by harvesting identical tubes at different cycles of amplification. The net yield data from 5 tubes of pooled amplification are plotted in Fig. 1B
. For statistical comparison, we established replicate hybridization experiments, generating 6-tube and 24-tube pools from 24 cycles of amplification from identical target. The R2 values were 0.9971 for the duplicate hybridizations to Affymetrix U133A plus chips and 0.9934 for the 6-tube pool vs the 24-tube pool. Scatterplot representation of values >100 for the 6-tube vs 24-tube pools are shown in Fig. 1C
. Agilent 2100 Bioanalyzer analysis showed that most of the products of the RFA protocols were 100 to 700 bases in length.
To validate the independent repeatability of small-scale amplifications, we established microscale cDNA synthesis protocols (10 ng total RNA) that used independent IVT (T7) results to compare and contrast the independent 15-million–fold RFA results. We used UHRR as the control sample for the Breast Carcinoma Cell line (T-47) microarray experiments designed to compare the RFA and T7 platforms (17). Amplification was 200-fold for the duplicate pair of IVT samples, with a mean duplicate array correlation of 0.9822, and 15–million-fold for the duplicate microscale RFA samples, with a mean duplicate array correlation of 0.9798. Analysis of duplicate UHRR microarray results from both platforms revealed a mean correlation of 0.8296 for the 4 independent RFA vs T7 analyses (RFA-1/T7–1:0.8295; RFA-1/T7–2:0.8119; RFA-2/T7–1:0.8296; RFA-2/T7–2:0.8474).
There were 4305 T7 and 7560 RFA paired mean D/N ratios >2.0, with 2280 confirmed D/N values between the T7 and RFA platforms. We compared the D/N results from the 2 platforms using only the genes that were determined to be present in the T7 analysis by the Affymetrix software (19 460 D/N values). The correlation for mean log D/N values between the 2 platforms for the genes was 0.5583. The correlations remained high for T7 log D/N duplicate (0.8478) and RFA log D/N duplicate (0.7418). Analysis of mean D/N values within probe sets that were confirmed by both platforms revealed that 1454 D/N values were up-regulated more than 2.0-fold, 275 D/N values were down-regulated more than 2.0-fold, and 10 326 D/N values were within the bounds of 2.0-fold up-regulated to 2.0-fold down-regulated. Summarizing these results gives 62% agreement for differential expression calls between the 2 platforms for the 19 460 genes with confirmed present calls.
The 15–million-fold RFA D/N results are comparable to the 200-fold T7 D/N results and demonstrate good correlations for such high degrees of amplification. These results show that RFA robustly amplifies cDNA as small double-stranded fragments that are universally primed. Fragment cDNA retains accurate gene expression ratios for individual fragments from 2 populations, as demonstrated by real-time quantitative RT-PCR data, while enabling exponential amplification.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
