Screening and Quantification of Multiple Chromosome Translocations in Human Leukemia

Abstract

Background: Characterization of fusion gene transcripts in leukemia that result from chromosome translocations provides valuable information regarding appropriate treatment and prognosis. However, screening for multiple fusion gene transcripts is difficult with conventional PCR and state-of-the-art real-time PCR and high-density microarrays.

Methods: We developed a multiplex reverse transcription-PCR (RT-PCR) assay for screening and quantification of fusion gene transcripts in human leukemia cells. Chimeric primers were used that contained gene-specific and universal sequences. PCR amplification of fusion and control gene transcripts was achieved with use of an excess of universal primers to allow the ratio of abundance of fusion gene to endogenous or exogenous controls to be maintained throughout PCR. Multiplex RT-PCR products analyzed by an ABI 310 Genetic Analyzer were consistent with those of duplex RT-PCR (single analytical sample plus control). In addition, multiplex RT-PCR results were analyzed by an assay using an oligonucleotide microarray that contained probes for the splice-junction sequences of various fusion transcripts.

Results: The multiplex RT-PCR assay enabled screening of >10 different fusion gene transcripts in a single reaction. RT-PCR followed by analysis with the ABI Prism 310 Genetic Analyzer consistently detected 1 fusion-transcript-carrying leukemia cell in 100–10 000 cells. The assay covered a 1000-fold range. Preliminary results indicate that multiplex RT-PCR products can also be analyzed by hybridization-based microarray assay.

Conclusions: The multiplex RT-PCR analyzed by either ABI Prism 310 Genetic Analyzer or microarray provides a sensitive and specific assay for screening of multiple fusion transcripts in leukemia, with the latter an assay that is adaptable to a high-throughput system for clinical screening.

Disease-specific RNA is often expressed in leukemia cells as a result of chromosome translocations that join two gene coding regions, leading to expression of an in-frame chimeric mRNA product that participates in leukemogenesis (1). Identification of these disease-specific RNA markers offers a detailed molecular fingerprint that can be used to identify and classify specific leukemia or lymphoma subtypes (2)(3) for better treatment and for predicting outcome.

Current leukemia diagnosis and stratification involve karyotyping and surface immunophenotype analysis. In addition, detection of specific fusion gene transcripts is required for optimal differential diagnosis and treatment strategies.

To screen for specific fusion genes to achieve leukemia subtyping and risk stratification, an ideal assay system should be highly sensitive to accommodate the limited amount of patient sample, especially when RNA is the starting material. In addition, because there are >40 common, distinct chromosome translocations characteristic of various leukemia subtypes, it would be highly advantageous that the test be carried out without previous knowledge of specific translocation types. One of the most practical ways to accomplish this is to detect multiple or all targets of interest in a single reaction. Multiplexing is therefore a highly desirable goal.

Reverse transcription-PCR (RT-PCR) has been shown to be a sensitive tool in detecting transcripts of chromosome translocations (4)(5)(6). Although the improvement of various RT-PCR approaches for translocation transcript detection is ongoing, the use of RT-PCR for fusion gene transcript screening is still labor-intensive, costly, and time-consuming if separate reactions are performed for screening and detecting individual fusion transcripts. Furthermore, because of the nature of PCR, i.e., its logarithmic amplification phase followed by a plateau, PCR with gene-specific primers alone and high cycle numbers would not allow quantification of multiple targets.

We describe here a multiplex, quantitative RT-PCR method for detecting fusion gene transcripts in total RNA samples from human leukemia cell lines. In this assay, chimeric primers were used that contained both gene-specific and universal sequences. PCR was carried out with only the universal primers in large excess over chimeric primers; amplification was mainly achieved by universal primers. The priming and amplification strategy allowed the expression ratio of fusion genes compared with control genes to be maintained throughout amplification so that fusion genes were quantified relative to added exogenous control transcripts as well as to endogenous housekeeping gene controls. The method provides sensitivity and specificity with multiplexing capability in the optimum range of detecting 10–40 fusion-transcript targets.

The ABI Prism 310 Genetic Analyzer is capable of differentially detecting by size the RT-PCR products and their relative quantification. However, to use the assay for clinical sample screening and diagnosis, additional confirmation of product sequences would be necessary. We designed a prototype oligonucleotide microarray for analysis and confirmation of the multiplex RT-PCR products. A fraction of the RT-PCR mixture is used in an in vitro transcription (IVT) assay to generate biotin-labeled single-strand RNA (cRNA). Hybridization of the labeled cRNA with a microarray containing capture probes with sequences complementary to that of the fusion transcript splice junction as well as to several controls would provide sequence-specific signals and quantification in the range of the assay. The combination of RT-PCR and a hybridization-based microarray assay can thus serve as a clinical diagnostic tool with the desired specificity and sensitivity for screening and quantification of fusion genes in leukemia as well as molecular targets of other diseases.

Materials and Methods

leukemia cell lines and rna preparation

Cell lines with unique translocations were used in the study (Table 1 ). All cell lines were cultured at 37 °C in RPMI-1640 supplemented with 100 mL/L fetal calf serum and antibiotics. Additional supplements were used according to recommendations of the original depositors. Cells were continuously passaged for no longer than 8 weeks before isolation of cellular RNA. An overview of the characteristics of these cell lines has been published previously (7). Leukemia cell extract was prepared by the Trizol (Life Technologies) homogenization method according to the manufacturer’s recommendations. Total RNA was prepared by subsequent extraction and precipitation with chloroform–iso-amyl alcohol (24:1 by volume; EM Science) and isopropanol, respectively. Total RNA was resuspended in RNase-free water after washing with 750 mL/L ethanol, and its concentration and quality were determined by ultraviolet spectroscopy and electrophoretic analysis on an Agilent Bioanalyzer and RNA 6000 reagent set (Agilent) or by 2% denaturing agarose gel electrophoresis. Only total RNA with an ultraviolet absorbance ratio (A₂₆₀/A₂₈₀) >1.6 and ribosomal RNA (28S/18S) ratio of 1.8–2.2 was used. The total RNA stock solutions containing >1.0 g/L RNA were stored at −80 °C at all times.

Table 1.

Human leukemia cell lines with specific translocations and fusion genes¹ used in the study.

Cell line/Translocation	Source2	Genes3	Fusion splice junction
BV-173	ATCC	BCR (22)/ABL (9)	b2a2
t(9;22)(q34;q11)			b3a2
K-562 t(9:22)(q34;q11)	ATCC	BCR (22)/ABL (9)	b3a2
			b2a2
Kasumi-1	DSMZ	AML1 (21)/ETO (8)	AML1 exon5/EOT exon2
t(8;21)(q22;q22)
ML-2 t(6;11)(q27;q23)	DSMZ	MLL (11)/MLLT44 (6)	MLL exon9/MLLT4 exon2
Mono-Mac-6	DSMZ	MLL (11)/MLLT35 (9)	MLL exon11/MLLT3 central exon
t(9;11)(p22;q23)			MLL exon10/MLLT3 central exon
THP-1 t(9;11)(p22;q23)	ATCC	MLL (11)/MLLT3 (9)	MLL exon9/MLLT3 central exon
NB4 t(15;17)(q22;q21)	DSMZ	PML (17)/RARA (15)	PML exon6/RARA exon3
REH t(12;21)(p13;q22)	ATCC	TEL (12)/AML1 (21)	TEL exon5/AML1 exon20
RS4;11 t(4;11)(q21;q23)	ATCC	MLL (11)/MLLT26 (4)	MLL exon10/MLLT2 exon4
MV4-11	ATCC	MLL (11)/MLLT2 (4)	MLL exon9/MLLT2 exon5
t(4;11)(q21;q23)

1

Gene names adhere to the official nomenclature used by CGAP Mitelman Database of Chromosome Aberrations in Cancer.

2

ATCC, American Type Culture Collection; DSMZ, German Collection of Microorganisms and Cell Cultures.

3

Chromosomes on which genes are located are in parentheses.

4

MLLT4 = AF6.

5

MLLT3 = AF9.

6

MLLT2 = AF4.

fusion gene sequence confirmation

Gene-specific primers for fusion genes in leukemia cell lines were designed with the primer analysis software OLIGO 6.0 (National Biosciences Inc.), based on sequence data of fusion partner genes deposited in GenBank or from publications. HPLC-purified oligonucleotide primers were purchased from Life Technologies. Fusion gene sequences for each cell line were determined from amplicons of at least three separate RT-PCRs using a unique pair of primers.

primer design for multiplex rt-pcr

The portions of the gene-specific sequences of the chimeric primers were selected with use of Primer3 (Whitehead Institute for Biomedical Research). Bacteriophage promoter T7 or SP6 sequence was added as a universal priming site; each upper (forward) and lower (reverse) primer therefore contained a T7 and SP6 sequence, respectively, at the 5′ end (Table 1 in the Data Supplement that accompanies the online version of this article athttp://www.clinchem.org/content/vol49/issue7/). The specificity of chimeric primers was confirmed by RT-PCR. In addition, the universal primer, T7, was 5′-labeled with 6-carboxyfluorescein so that amplicons were fluorescently labeled and could be analyzed with the ABI Prism 310 Genetic Analyzer (Applied Biosystems). The reverse primer for the housekeeping gene β-actin was a mixture of primer with and without a phosphate group at the 3′ end, at a concentration ratio of 39:1. This modification was necessary because of the high abundance of the β-actin gene in the cells; such an attenuated primer for β-actin can limit β-actin amplification to within the dynamic range of the assay. The mixture of all primer pairs was designed to generate RT-PCR products with size differences of at least 3 bp in each reaction; the products thus were readily distinguishable by size by ABI Prism 310 Genetic Analyzer analysis.

multiplex rt-pcr

RT-PCR was carried out in two steps. For the reverse transcription step, the following were added to a total volume of 5 μL: 20 nM reverse primers, 1 mM deoxynucleotide triphosphates, 2.5 units of RNasin ribonuclease inhibitor (Promega), 10 U of Moloney murine leukemia virus reverse transcriptase (Promega), 10 mM Tris (pH 8.3), and 2.25 mM MgCl₂. The reverse transcription mixture contained 20 ng of total RNA. Heating and incubation were performed as follows: 48 °C for 1 min, 37 °C for 5 min, 42 °C for 30 min, and 95 °C for 5 min. For the PCR step, the following were added to the reverse transcription mixture: 10 nM forward primers, 1 mM each of the upper (forward) and lower (reverse) universal primers (T7 and SP6), 0.375 mM deoxynucleotide triphosphates, 1 U of AmpliTaq Gold (Perkin-Elmer), 10 mM Tris (pH 8.3; contained in the 10× Taq buffer), 50 mM KCl, and 8 mM MgCl₂. The final reaction volume was 20 μL. PCR cycles included 1 cycle at 95 °C for 5 min and 35 cycles of 30 s at 94 °C, 30s at 60 °C, and 1 min at 72 °C. Reactions were left at 68 °C for an additional 60 min before cooling to 4 °C.

6-Carboxyfluorescein-labeled RT-PCR products were detected and quantified with use of the ABI Prism 310 Genetic Analyzer and GeneScan collection software (Applied Biosystems). Briefly, 1.5 μL of cDNA from the multiplex reaction was added to 12 μL of pure formamide and 0.5 μL of ROX 350 size markers (Applied Biosystems). Samples were denatured at 95 °C for 5 min and cooled on ice for 5 min. The following conditions were used for capillary electrophoresis: 5-s injection at 15.0 kV; electrophoresis voltage, 15 kV (resulting current, 7–9 A); column temperature, 60 °C; electrophoresis time, 25 min. RT-PCR product size calling and quantification were performed by GenoTyper software (Applied Biosystems).

microarray design and manufacturing, and analysis of multiplex rt-pcr products

For each fusion gene transcript present in the cell lines, 10 probes (30mer) were selected that tiled across each gene fusion splice-junction sequence at an interval of 2 nt (Table 2 in the online Data Supplement). Probes for several external controls and housekeeping genes were designed with use of the Motorola Life Sciences probe-selection algorithm according to properties such as melting temperature, GC content, position of RT-PCR products, and free energy of probe-target hybrid. Leukemia microarrays containing these probes were manufactured according to Motorola Life Science standard operating procedures, using Surmodics porous standard glass slides (Surmodics Inc.). In brief, 30mer oligonucleotide probes were presynthesized that contained a hexylamine linker at the 5′ end. Probes were HPLC-purified, and quality control was carried out before probes were dispensed on glass slides. A prepolymer of acrylamide was photocoupled to the prepared Surmodics slide, yielding a low-density cross-linking polymer film. Activated ester had been added to the prepolymer to provide attachment sites for C-6 amino oligonucleotides. The 5′-amine-capped oligonucleotides were dispensed and deposited on the polymer-coated slides described above by a piezoelectric dispensing robot. Oligonucleotides were codispensed with a fluorescein-derivative dye to account for each spot and its size by scanning as part of the postdispensing quality-control process. Unbound sites on the slides were blocked, and slides were washed, rinsed, and dried before attachment of a proprietary polypropylene hybridization chamber.

Because all amplicons carried the SP6 promoter sequence at the 5′ end of the antisense strand of the double-stranded cDNA, an IVT reaction was performed with the MegaScript SP6 IVT reagent set (Ambion) and 1–5 μL of reaction mixture (of a total of 20 μL) from the RT-PCR reaction described above. Briefly, the following components were added to up to 5 μL of the RT-PCR mixture: 4 μL of 10× IVT reaction buffer, 4 μL each of 75 mmol/L ATP and GTP, 3 μL each of CTP and UTP, and 7.5 μL of 10 mmol/L Biotin-11-CTP and Biotin-11-UTP (HPLC-purified; NEN Life Science Products). RNase-free water was added to adjust the volume to 36 μL. Finally, 4 μL of enzyme was added, and the reaction was kept at 37 °C for 6 h. Biotin-labeled single-stranded RNA (cRNA) was purified with use of the RNeasy reagent set (Qiagen), eluted in 30 μL of RNase-free H₂O, and quantified by ultraviolet spectroscopy. Hybridization of target cRNA to the microarray was carried out in Hybridization Buffer (Motorola Life Sciences) at 37 °C for 14 h. Posthybridization washing and staining were performed according to the Motorola CodeLink Manual Target Preparation protocol. Streptavidin-Alexa 647 (Molecular Probes) was diluted 1:500 in 0.10 mol/L Tris-HCl (pH 7.5)–0.15 mol/L NaCl–5 mL/L Blocking Reagent (NEN Life Science) for 30 min. Slides were washed with 0.10 mol/L Tris-HCl (pH 7.5)–0.15 mol/L NaCl–0.5 mL/L Tween 20, rinsed with distilled water, and dried before being scanned on an Axon GenePix Scanner (Axon Instruments). Images of the slides were analyzed by CodeLink Expression Analysis Software (Motorola Life Sciences).

Results and Discussion

confirmation of fusion sequence

Different primer pairs were designed for each fusion transcript. RT-PCR was carried out using total or poly(A) RNA from cell lines containing known fusion transcripts. Fusion gene sequences flanking the fusion junction for each cell line were obtained from sequencing of at least three RT-PCR products with use of distinct primer pairs. The sequence data were used in primer design for the multiplex RT-PCR (Table 1 in the online Data Supplement) as well as in the tiling probe selection for the oligonucleotide microarray (Table 2 in the online Data Supplement). Our fusion splice-junction sequence data for 7 of the 10 cell lines used matched those published or deposited at the National Center for Biotechnology Information (8)(9)(10)(11)(12)(13)(14). For cell lines THP-1, BV-173, and Mono-Mac-6, which have no sequence data deposited or published in detail, our sequence data matched those of patients carrying the same types of fusion transcripts. NB4 and Mono-Mac-6 cells also carry multiple splice variants, but for simplicity, only one fusion sequence of each cell line was used for the assay. The multiple RT-PCR products generated with use of a single primer pair in NB4 and Mono-Mac-6 cells were most likely the result of short sequence duplications within the primer-targeted splice-junction regions.

detection and quantification of fusion transcripts by multiplex rt-pcr

The assay was tested using cell lines containing known types of gene fusions as the result of specific chromosome translocations. To amplify translocation-specific transcripts present in leukemia cells, primers contained sequences specific to those flanking the junction of fusion genes as well as the T7 and SP6 universal sequence portion. These primers were selected to have similar melting temperatures (60 °C) and other properties, such as length, GC content, and free energy, by Primer3 to reduce differential priming and nonspecific amplification of targets. Primers were redesigned until specific amplicon(s) were successfully detected in single or duplex (with β-actin) reactions before being combined in the 15-plex RT-PCR (including the primer pair T7 and SP6) for testing. In addition to design, selection, and quality control of the primers, the multiplex (15-plex) assay was also optimized by evaluating various RT-PCR temperatures and reaction times and finding the optimum concentrations for reagents such as salt, deoxynucleotide triphosphates, and primers. Although multiplex RT-PCR with 10–20 primer pairs was ideal, our preliminary data indicated that multiplex RT-PCR with primer pairs in excess of 20 was achievable with substantial assay optimization effort (data not shown). However, the probability that formation of nonspecific PCR products and primer–dimers would increase with increased numbers of primer pairs limited the maximum number of primers pairs.

The multiplex assay described here is atypical in the sense that it does not give rise to the number of specific amplicons equivalent to the number of primer pairs present when leukemia cell RNA is used. Because only one or two fusion transcripts may be present in cells of a particular type of leukemia, the assay is therefore best suited for screening of patient samples for the most likely fusion transcripts and their variants. In practice, more that one set of multiplex RT-PCRs can be devised to encompass more than 40 fusion transcripts. The multiplex RT-PCR for the 10–20 most prevalent fusion gene transcripts can be used as a first screening with additional tests to be carried out for the less common fusion gene transcripts and their variants.

The total RNA of fusion-positive cells was first mixed with the cloned bacteria gene transcripts entF, fixB, and araB (Ambion). Universal primers (containing only T7 or SP6 sequence) were used at a 100-fold excess over chimeric primers to carry out amplification of the multiple targets of interest in the reactions. We optimized the multiplex RT-PCR assay to give sensitive and specific amplifications by varying the annealing and elongation temperatures and the length of the PCR steps. The detection limits and dynamic range were evaluated with 10-fold limiting dilutions of the total RNA of gene-fusion-positive cell lines in human liver total RNA from healthy donors (Clontech). The assay covered a 1000-fold range (Fig. 1 ). For the endogenous control (β-actin), attenuated reverse primer was used, and the amplification of β-actin was well within the assay dynamic range. The same approach could be applied to other targets of interest if their expression is high.

Figure 1.

Dynamic range of the quantitative multiplex RT-PCR assay.

Bacterial gene transcripts fixB, entF, and araB were added to 20 ng of a wild-type human total RNA sample to the concentration of 0.1 pg/μL. The total RNA added was serially diluted to 1 in 10, 100, 1000, and 10 000 in the same total RNA without bacterial genes. Multiplex RT-PCR was performed in a two-step reaction (for details see Materials and Methods), and results were analyzed by the ABI Prism 310 Genetic Analyzer. The integrated absorbance (IOD) represents the mean peak area of the products from at least three repeat injections on the ABI Prism 310 Genetic Analyzer.

Detection and quantification of the specific translocation transcript of each cell line were performed with a multiplex RT-PCR approach and analysis by the ABI Prism 310 Genetic Analyzer (see Materials and Methods for details). Quantitative multiplex RT-PCR results were compared with those of the RT-PCR assay under the same conditions but with only the primer pairs for the translocation specific to each cell line plus the primer pair for β-actin. The limit of detection for each cell line was determined in 10-fold serial dilutions of total RNA from the cell line in human liver total RNA from healthy donors. Several series of multiplex RT-PCR experiments consistently showed positive amplicon detection by the ABI Prism 310 in the 1:100 to 1:10 000 dilutions, indicating that the assay can detect at least 1 fusion-gene-carrying cell in 100–10 000 normal cells (Table 2 ). The wide sensitivity range is a crude reflection of the variable abundance of specific fusion gene transcripts in cell lines and is also indicated by the ratio of the abundance of fusion gene to that of the housekeeping gene, in this case, β-actin. Because cellular concentrations of most so-called housekeeping genes (including β-actin) fluctuate considerably among different tissue samples, exogenous bacterial DNA was added, and entF was used in addition to β-actin for comparison (Table 2 ). There was good agreement between the ratios of fusion genes in each leukemia cell line to both entF and β-actin, owing largely to the fact that the β-actin concentration was consistent in these cells, which were from the same tissue type, namely, peripheral blood.

Table 2.

Comparison of RT-PCR results for fusion transcripts.

Human leukemia cell line	Size, bp	Mean (SD) ratio to β-actin, %			Multiplex RT-PCR
				Multiplex1	Duplex2	Mean (SD) ratio to entF, %	Detection limit, dilution
BV-173	142	66 (7.7)	48 (7.0)	410 (44)	1:1000
K-562	115	76 (26)	55 (13)	380 (15)	1:1000
Kasumi-1	161	85 (3.0)	67 (5.0)	470 (55)	1:10 000
ML-2	183	12 (0.5)	11 (0.3)	120 (11)	1:100
Mono-Mac-6	264	8.4 (0.2)	6.1 (1.1)	60 (8)	1:100
THP-1	230	11 (0.3)	4.3 (0.7)	70 (10)	1:1000
NB4	210	3.0 (0.0)	2.0 (0.3)	40 (5)	1:100
REH	197	13 (0.3)	8.1 (1.1)	95 (8)	1:100
RS 4;11	189	23 (3.3)	18 (1.5)	110 (10)	1:1000
MV 4-11	150	13 (0.5)	3.0 (0.4)	130 (8)	1:1000

1

Multiplex RT-PCR with a total of 14 chimeric primers pairs, including bv173, k562, kasumi1, ml2, mm6_m, thp1, nb4, reh, rs411, mv411, bactin, entF, araB, and fixB (see Table 1 in the online Data Supplement for primer sequences), plus T7 and SP6.

2

Duplex RT-PCR with chimeric primer pair for one of the fusion sequences and chimeric primer pair for β-actin.

confirmation of the results of the multiplex rt-pcr by microarray analysis

All multiplex RT-PCR amplicons carry a fully functional SP6 phage promoter sequence on the antisense strand and T7 phage promoter sequence on the sense strand. An IVT reaction was carried out to generate antisense single-stranded cRNA in the presence of biotin-UTP/-CTP. The biotin-labeled cRNA can be readily hybridized to a microarray containing sense strand-capturing probes. The posthybridization process was carried out to wash away nonspecific hybridization and initiate fluorescent color development in the presence of streptavidin-Alexa 647 (Molecular Probes). Subsequent washing, slide scanning, and data analysis were carried out. Usually 0.1–0.4 μg of cRNA was needed for single leukemia array hybridization. A fraction of the RT-PCR reaction mixture (1–5 μL of a total of 20 μL) was used to generate an adequate amount of cRNA (in the range of 0.1–1.0 μg) for hybridization, indicating the high sensitivity of this method. Because the length of cRNA is determined by the size of the RT-PCR cDNA reaction products, which are relatively short, we omitted the sample fragmentation step that typically is included in target preparation for microarray gene expression profiling. In addition, the usually lengthy IVT step can be significantly shortened because incorporation of a RT-PCR step generates a sufficient amount of template for IVT.

Analysis of RT-PCR products from leukemia cell lines by microarray provided specific and sensitive results, confirming the results of the ABI Prism 310 Genetic Analyzer analysis. Moreover, microarray analysis was able to detect both known splice variants in K-562 and BV-173 cells (Fig. 2 ) (6)(15). Preliminary results also were consistent with previous findings that K-562 carried predominantly the major type of BCR-ABL fusion (b3a2), whereas BV-173 carried mostly the minor type of BCR-ABL fusion (b2a2). Signals in K-562 samples appeared to be saturating for spots on the microarray containing capture probes for the major forms of fusion transcripts, whereas in BV-173 cells, which carried predominantly the minor form, saturation had been reached for spots carrying capture probes for the minor forms. More experiments are needed to determine the optimum amount of RT-PCR products to be used in target preparation for microarray analysis.

Figure 2.

Detection and quantification of fusion gene transcripts.

Multiplex RT-PCR was performed in two steps (see Materials and Methods for details) with total RNA. A 5-μL PCR reaction mixture was used in the IVT reaction with the MegaScript IVT reagent set (Ambion) to generate single-stranded RNA (cRNA) labeled with biotin (through biotin-UTP/-CTP). Hybridization of RNA to the microarray was carried out according to the manufacturer’s recommended conditions (see Materials and Methods for further details). The y axis represents signal intensity (IOD) given as the mean of 10 tiling probes for each specific fusion carried by the cell line. The x axis represents samples for hybridization from each cell line. The z axis represents probes for each specific fusion splice junction.

The array results gave satisfactory signal-to-noise ratios with low background signals, indicated by the results of wild-type RNA samples with no known fusion transcripts. Because this is the prototype microarray for proof-of-principle studies, it has been simplified compared with the reality of testing of a patient sample. Some complexity of patient samples is to be expected because they may carry chromosome translocation types that produce different types of fusion transcripts and their alternative splice variants. However, microarrays designed with probes targeting most or all types of fusion genes and their splice variants can accommodate such clinical complexity in a single microarray, which would therefore add to the attractiveness of such a screening assay.

There were sizeable background signals from probe Kasumi1_5 for sample REH and probe Reh_6 for sample ML-2 (Fig. 2 ). Despite the extensive assay optimization, nonspecific products of wild-type partner genes could be present. As a result, the prepared target might contain labeled wild-type partner gene products, which could give rise to nonspecific signals on the microarray. Further optimization is necessary to reduce nonspecific amplification in the multiplex RT-PCR. In addition, more stringent microarray hybridization conditions and/or the use of longer oligonucleotide probes (50mers–60mers) in the microarray may help to better differentiate fusion gene amplicons from those of wild-type partner genes.

Chromosome translocations and their fusion gene transcripts are increasingly important molecular markers for leukemia/lymphoma subtyping and disease stratification. Usually, only ∼1–2 μg of total RNA can be isolated from 1 mL of fresh peripheral blood, and much less can be isolated from frozen blood or bone marrow biopsies from patients. Thus, the starting material for any RNA-based diagnostic assay is very limiting and is susceptible to further RNA degradation if not handled quickly and properly. Patient clinical presentations and other diagnostic tests, such as cytogenetic results, in general may be compatible with several possible subtypes of leukemia/lymphoma. Without previous knowledge of the specific translocation type, it is difficult or impractical to carry out a series of single analytical detection assays when total RNA is limiting. Very few existing assays can perform multi-analyte tests in a single assay with precision and accuracy with limited amounts of starting material [total RNA or poly(A) RNA]. In addition, as a diagnostic tool, definitive analysis of molecular markers through sequence confirmation is essential for unambiguous results. With a conventional approach, molecular target identification by RT-PCR amplicon size is insufficient. Individual amplification and sequencing of targets would be too time-consuming and costly and thus would be deemed impractical for general clinical laboratories.

Fluorescence in situ hybridization is widely used clinically for detecting chromosome translocations on a DNA level, but it requires previous knowledge of the translocation type to be able to select the appropriate DNA probes for detection (16). Real-time RT-PCR using TaqMan^® technology and the ABI 7700 (Applied Biosystems) or LightCycler (Roche Diagnostics) system is considered very sensitive and can quantify fusion gene transcripts in leukemia patient samples, but this methodology lacks multiplexing capabilities beyond two to four samples because of the signal crossover and complexity that occur when multiple fluorescent labels are used and the increased possibility for error when multiple fluorescent signals are being analyzed (17)(18). It would therefore be ideal only for monitoring minimal residual disease after screening of fusion gene transcripts is carried out and the specific translocation is known. The method described here overcomes both of these obstacles because of its increased sensitivity: as little as 20 ng of total RNA or less is needed (whereas other assays use 1–5 μg), and 10–40 analyses can be multiplexed in a single assay with the capability for quantification.

Because some of the translocation fusion splice-junction sites may be a few kilobases distant from the 3′ poly(A) tail on the mRNA, use of a microarray assay alone is not possible at this stage because the reverse transcription step is unable to generate cRNA long enough to reach the fusion splice-junction site. The target preparation step for microarray hybridization follows a protocol that generates cDNA from the 3′ poly(A) site with use of poly(dT) primers. cDNA covers primarily sequences at the 3′ end; therefore, cRNA would contain mainly mRNA sequences at the 3′ end. Incorporation of a microarray step serves as a sequence confirmation/detection step after the multiplex RT-PCR assay that provides increased sensitivity and specificity. The microarray tiling probe approach is unique for detection of fusion gene transcripts and provides unambiguous and definitive results for products of different chromosome translocations as well as detection of splice variants from the same type of chromosome translocation. The signal intensity was averaged for all 10 tiling probes in this study. Because all 10 probes center on the fusion splice junction, in theory, they should give specific signals when fusion transcripts with the correct fusion junction sequence are present with no substantial hybridization with wild-type partner genes. This indeed was demonstrated in the assay (Fig. 2 ). Obtaining and maintaining quantitative results of the multiplex RT-PCR assay on microarrays was also possible, which was shown in our preliminary results and remains to be tested further with various starting materials for target preparation.

Several applications of RT-PCR in combination with microarrays have been reported (19)(20)(21). An attempt has also been made to analyze leukemia fusion transcripts by multiplex RT-PCR followed by acrylamide gel-pad-based oligonucleotide microarray detection (22). However, these multiplex RT-PCR assays used only gene-specific primer pair mixtures, amplifying multiplex targets or fusion transcripts in a single assay. This strategy basically excludes the quantification or semiquantification aspect of the assay because of the logarithmic amplification that occurs in PCR and its plateau thereafter. The subsequent microarray assay merely serves as a sequence confirmation step. In contrast, our approach uses universal primers for amplification. The multiplex RT-PCR itself is quantitative for target relative to controls. The subsequent microarray assay not only serves as a sequence confirmation step through hybridization but also a quantification step.

On the basis of the initial results of our assay integrating RT-PCR with microarray analysis, we have developed a method that moves in the direction of a microarray with lower probe density for specific diagnostic applications. As more molecular markers associated with cancer and other diseases are discovered, the approach described here can be readily applied to these markers and serves as new diagnostic tools in those areas.

This work was supported by Motorola Corporate Laboratories (Ft. Lauderdale, FL) and NIH Grant CA84405 Kiwanis (to J.D.R.).