| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Illumina, Inc., San Diego, CA
2 Veterans Affairs Hospital, University of California-San Diego, San Diego, CA 92161
aaddress correspondence to this author at: Illumina, Inc., 9885 Towne Centre Dr., San Diego, CA 92121-1975; fax 858-202-4680, e-mail jfan{at}illumina.com
Gene expression profiling using microarrays has revolutionized the analysis of biological samples. In clinical applications, microarray data have been used to successfully distinguish among patients exhibiting similar symptoms (1)(2). Early demonstrations of this power were in the diagnosis of subtypes of acute leukemia (3) and diffuse large B-cell lymphomas (4), and such analyses are gradually gaining acceptance for diagnostic and prognostic applications (5)(6)(7). Investigators are currently accumulating microarray data for a broad assortment of such studies but are limited by the requirement of fresh/frozen tissues for sample preparation and labeling (8). This limitation requires the accumulation of fresh samples throughout the course of the disease, which may involve years of monitoring. However, formalin-fixed, paraffin-embedded (FFPE) tissues are widely available and have the advantage of a known patient outcome and drug response history. RNAs derived from these samples are commonly badly degraded and have not been useful for conventional microarray studies (9)(10). We applied a novel expression assay to simultaneously monitor 502 cancer-related genes in RNAs derived from FFPE samples, using microarrays assembled on fiber optic bundles. Our results suggest that this approach can be used for extending microarray analyses to RNAs derived from archival tissue samples.
We have recently developed a gene expression method called the DASLTM assay (cDNA-mediated annealing, selection, extension, and ligation) (11). This assay targets gene-specific sequences, using pools of chimeric query oligonucleotides. The oligonucleotides all share common primer landing sites so that once the upstream oligonucleotide is extended and ligated to the downstream oligonucleotide, an amplifiable product is generated. One PCR primer pair is used to amplify all of the amplifiable templates and generate amplicons of similar size (
100 bp). This uniformity minimizes potential bias during amplification of many different targets. Currently, the DASL assay can be multiplexed to monitor hundreds of genes (11).
To allow the use of universal microarrays, the downstream query oligonucleotides also contain a unique address sequence that is associated with each gene. This address sequence allows the amplified product, which is labeled during PCR with a fluorescent primer, to hybridize to a microarray bearing the complementary address sequences. This feature provides ready flexibility: to change the genes being monitored, the address sequences can be reassigned and the query oligonucleotide pool resynthesized, using the same arrays.
Finally, the cDNA synthesis step is performed with both oligo(dT) and random priming, which frees the assay from dependence on an intact polyA tail, unlike the usual T7 promoter-oligo(dT) priming method for microarray sample preparation (12). The use of random hexamers or nonamers in the cDNA synthesis allows representation of targeted cDNA sequences despite RNA degradation. Less than 50% of target probes in FFPE samples were detected when only oligo(dT) primer was used for cDNA synthesis, compared with the use of both oligo(dT) and random primers.
To monitor gene expression in FFPE samples, we prepared RNA from 5-µm human tissue sections mounted on microscope slides, obtained from BioChain Institute according to an Institutional Review Board-approved protocol. Briefly, the slides were deparaffinized in xylenes, and tissue samples were then scraped off with razor blades and held in xylenes until subsequent processing. RNA was extracted by use of the High Pure RNA Paraffin Kit (Roche Applied Science) and quantified by use of RiboGreen (Molecular Probes). When measured on Bioanalyzer (Agilent) RNA Pico Chips, the size of the RNA fragments ranged from 100 to >500 nucleotides, with a peak maximum at 130 nucleotides. On average, 1–2 µg of total RNA was isolated from five 5-µm tissue sections. For each sample, 200 ng of total RNA was converted to cDNA and processed in the DASL assay as described previously (11)(13). Oligonucleotides targeting 502 cancer-related genes were used in these experiments, at a density of three nonoverlapping probes per gene, giving a 1506-plex measurement for each sample. Our previous study showed that three probes per gene lend the assay sufficient sensitivity and reproducibility for quantitative detection of differential expression in FFPE tissues (13). Mean signal values were computed for each gene by determining the mean signal for the three representative probes (11). Because DASL uses random priming in the cDNA synthesis, the probes can be designed to target any unique regions of the gene without the need to limit the selection of optimal probes to the 3' end of transcripts.
We profiled 16 FFPE samples of four tissue types—prostate, colon, breast, and lung—with each tissue represented by one nondiseased and several cancer samples. As shown in Fig. 1
, the DASL assay gave highly reproducible intensity measurements for FFPE samples preserved for 1.5–3 years. The correlations (r2) between the expression profiles of replicate experiments were 0.997 and 0.998 for the healthy prostate tissue and prostate adenocarcinoma samples and 0.995 and 0.994 for healthy colon and colon adenocarcinoma samples, respectively. Shown in the far right panels of Fig. 1
is a comparison of the intensity data across these samples, indicating significant differences in gene expression between the healthy and disease states. In an independent study (13), we have shown that highly reproducible tissue- and cancer-specific gene expression profiles can be obtained with as little as 50 ng of total RNA (r2 = 0.97) isolated from formalin-fixed tissues that had been stored from 1 to longer than 10 years. In addition, gene expression profiles generated with 50 ng of input RNA were quite comparable with those generated with 500 ng RNA (r2 = 0.95) (13).
|
Differentially expressed genes were determined by a t-test with a P value (0.01) cutoff. Of the 502 genes monitored, 86 were higher in adenocarcinoma relative to healthy prostate, whereas 111 genes were down-regulated in adenocarcinoma relative to healthy tissue sample. The up-regulated genes included IL6, IL8, LIF, and OSM, as well as other cytokines previously associated with prostate cancer progression (14), and MMP7, a known marker for a variety of tumors (15). Differential expression analysis of colon adenocarcinoma and healthy colon tissue showed higher expression of 71 genes in cancer compared with healthy tissue, whereas 64 genes had lower expression in adenocarcinoma. Highly up-regulated genes in colon adenocarcinoma included SERPINE1, previously known as a gene associated with tumor invasion (16)(17), and several matrix metalloproteinases, including MMP7 (15).
In a separate study, we used the DASL assay to monitor expression of 510 prostate cancer-related genes in >140 cancer and benign prostate hyperplasia FFPE samples. We also performed laser capture microdissection (LCM) on two of the FFPE tissues (both with <20% tumor content) and compared gene expression in these two heterogeneous samples with expression in their corresponding cancer and stroma LCM samples. Although the data obtained from the LCM samples was noisier (presumably because of the lower input of RNA in the assay, 10 ng), with the correlation (r2) between technical replicates ranging from 0.92 to 0.95, we observed increased expression of tumor- and stroma-specific markers, originally identified from analysis of undissected tissues, in the respective LCM samples. This demonstrates the feasibility of using heterogeneous FFPE tissues for biomarker discovery.
In interpreting these data, it is important to recognize that the output of the DASL assay is a reflection of the extended and ligated query oligonucleotide pool. Because this is an indirect measurement of expression that is dependent on competition among the resulting products for labeling in the PCR amplification, changes in hybridization signal may not accurately reflect changes in transcript abundance. Thus, measurement of absolute changes in cancer-specific gene expression during cancer progression requires independent experimental determination. Nevertheless, we found that intensity in the DASL assay generally tracked with RNA expression, with a correlation coefficient (r2) of 0.8–0.9 in a comparison between the DASL assay and qPCR for fold difference detection among samples (11).
In summary, we have demonstrated that gene expression profiling of RNAs from FFPE samples is possible, despite their extensive degradation, by use of the DASL assay and universal microarrays. The DASL assay combines the advantages of array-based gene expression analysis with those of multiplexed qPCR, thereby offering much higher multiplexing capacity and substantial throughput and cost-saving advantages. It uses 200 ng of total RNA to analyze 500 genes, 5- to 100-fold less than that required by qPCR, which usually takes 2–50 ng per reaction (per gene) (9)(18)(19)(20)(21). In addition, the small size of the targeted gene sequence (50 nucleotides) and the use of random primers in cDNA synthesis allow detection of RNAs that are otherwise too degraded for conventional microarray analysis. Our results suggest that archival tissue samples may be used to provide sufficient expression information to monitor multiple stages in disease progression as well as response to treatment. The capacity of the DASL assay to monitor hundreds of genes simultaneously in RNAs from FFPE samples may also accelerate the identification of potential biomarkers as prognostic indicators, given that the patient’s history is already known for many of these samples. Furthermore, the parallel processing of 96 samples at once, using the Sentrix® Array Matrix fiber optic array platform (22), should permit rapid analyses of expression patterns in hundreds of clinical samples.
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  S. Badve Reviving the Dead: DASL Expression Analysis from Paraffin Am. Assoc. Cancer Res. Educ. Book, April 12, 2008; 2008(1): 531 - 533. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. Loudig, E. Milova, M. Brandwein-Gensler, A. Massimi, T. J. Belbin, G. Childs, R. H. Singer, T. Rohan, and M. B. Prystowsky Molecular restoration of archived transcriptional profiles by complementary-template reverse-transcription (CT-RT) Nucleic Acids Res., August 1, 2007; 35(15): e94 - e94. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. S. Scicchitano, D. A. Dalmas, M. A. Bertiaux, S. M. Anderson, L. R. Turner, R. A. Thomas, R. Mirable, and R. W. Boyce Preliminary Comparison of Quantity, Quality, and Microarray Performance of RNA Extracted From Formalin-fixed, Paraffin-embedded, and Unfixed Frozen Tissue Samples J. Histochem. Cytochem., November 1, 2006; 54(11): 1229 - 1237. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
