ABSTRACT
A novel method combining elements of suppression subtractive hybridization with high throughput differential screening permits the efficient and rapid cloning of rarely transcribed differentially expressed genes. The experimental strategy virtually excludes the possibility of isolating false positive clones. The potential of the method is demonstrated by the isolation of 625 differentially expressed cDNAs from the metastatic adenocarcinoma cell line Bsp73-ASML when subtracted from its non-metastatic counterpart Bsp73-1AS. Northern analysis of 72 randomly selected clones demonstrated that 68 were differentially expressed with respect to Bsp73-ASML, indicating a true positive rate of 94%. Additionally, a large proportion of these clones represented rare transcripts as determined by the exposure time required to detect a signal. Sequence data indicated that of the 625 clones obtained, 92 clones scored perfect or near perfect matches with already known genes. Two hundred and eighty one clones scored between 60 and 95% homology to known human and mouse genes, whereas 252 clones scored no match with any sequences in the public databases. The method we describe is ideally suited whenever subtle changes in gene expression profiles need to be determined.
In almost any area of biology or medicine questions arise that make it desirable to know as many possible differences in gene expression between two types of cells or between two conditions.
There exists a large repertoire of techniques that aim at producing an inventory of differential transcripts between two populations of mRNAs. Identification and isolation of differentially expressed transcripts is generally achieved by one of the following methods: differential display and related techniques (1 ,2 ); representational difference analysis (RDA) (3 ); enzymatic degradation subtraction (4 ); linker capture subtraction (5 ); techniques involving physical removal of common sequences (6 ,7 ). Despite the fact that all these methods have proven successful in isolation of differentially expressed genes, they all possess some specific intrinsic drawbacks. For instance, differential display restricts the analysis to differences at the 3'-end of cDNAs, so that differences in the 5'-portion of cDNAs (e.g. variants of alternatively spliced genes) are not detected. Additionally, variable reproducibility of the differential band patterns and the significant incidence of false positives make it difficult to isolate rare transcripts that are differentially expressed. Another common feature of the methods mentioned above also represents an obstacle to isolation of rare transcripts: the disproportion of concentrations of differentially expressed genes is maintained in the subtraction. RDA requires multiple rounds of subtraction, as the method fails to take into account the large differences in relative abundance of individual mRNA transcripts.
The ideal system for subtractive cloning would generate an equalized representation of differentially expressed genes irrespective of their relative abundance, would permit the monitoring of subtraction efficiency prior to the time consuming screening work and would minimize, if not completely exclude, the isolation of false positive clones.
Recently, a novel technique called subtraction suppression hybridization (SSH; 8 ) has been described that combines a high subtraction efficiency with an equalized representation of differentially expressed sequences: The method is based on a specific form of PCR that permits exponential amplification of cDNAs which differ in abundance, whereas amplification of sequences of identical abundance in two populations are suppressed. We have developed a system that combines the advantages of SSH with high throughput differential screening. The system allows rapid identification of differentially expressed genes and virtually excludes isolation of false positive and false negatives clones.
The rat pancreatic adenocarcinoma cell lines Bsp73-1AS, Bsp73-10AS and Bsp73-ASML were cultured as described (9 ).
Polyadenylated RNA was isolated as previously described (10 ). Double-stranded cDNA was synthesized using Superscript reverse transcriptase (Gibco BRL) as follows. Aliquots of 2 [mu]g poly(A)+ RNA with 500 ng oligo(dT30) primer in a volume of 11 [mu]l was heated to 70oC for 10 min in a thermal cycler (Perkin Elmer 2400) and rapidly chilled on ice. The reaction mixture was made up to 20 [mu]l by adding 4 [mu]l 5* first strand reaction buffer (provided with the reverse transcriptase), 2 [mu]l 0.1 M DTT and 1 [mu]l dNTP mix (10 mM each dATP, dGTP, dCTP and dTTP). Reverse transcription was started by adding 2 [mu]l Superscript reverse transcriptase. From this reaction mixture, 1 [mu]l was immediately taken out and added to a labelling mixture to monitor incorporation (tracer reaction; see below). Both reaction mixtures were incubated at 42oC for 1 h. The tracer reaction was then stopped by adding EDTA to a final concentration of 20 mM.Tracer reaction. The efficiency of the first strand reaction was monitored by adding 1 [mu]l of the first strand synthesis to 1 [mu]l of a mixture containing 0.3 [mu]Ci [32P]dCTP. Specific incorporation of dCTP into high molecular weight nucleic acid was determined by the TCA precipitation procedure (11 ). Typically ~28% of the label was incorporated in the first strand reaction.
Following first strand synthesis, second strand synthesis was performed by adding 91.8 [mu]l sterile, bidistilled water, 32 [mu]l 5* second strand buffer [94 mM Tris-HCl, pH 6.9, 453 mM KCl, 23 mM MgCl2, 750 [mu]M [beta]-NAD and 50 mM (NH4)2SO4], 3 [mu]l 10 mM mixed dNTPs stock, 6 [mu]l 0.1 M DTT, 2 [mu]l Escherichia coli DNA ligase (7.5 U/[mu]l), 4 [mu]l E.coli DNA polymerase I (10 U/[mu]l) and 0.7 [mu]l E.coli RNase H (2 U/[mu]l). The contents were mixed and 5 [mu]l were removed for the radioactive incorporation assay (as above). Both reactions were incubated at 16oC for 2.5 h. The double-stranded cDNA in the unlabelled reaction was blunted by the addition of T4 polymerase, followed by further incubation at 16oC for 20 min before both reactions were stopped by addition of EDTA. As before, labelled second strand cDNA synthesis was monitored by TCA precipitation and by resolving the product on an alkaline agarose gel (11 ).
SSH was performed between Bsp73-1AS (`driver') and Bsp73-ASML (`tester') using the PCR-Selecttm cDNA Subtraction Kit (Clontech) according to the manufacturer's recommendations, except for modifications of the PCR and hybridization conditions. All PCR and hybridization steps were performed on a Perkin-Elmer 2400 thermal cycler.
For the first hybridization the mixture of driver and tester cDNAs was denatured at 100oC for 20 s and then cooled over 1 min to 68oC and maintained at that temperature for ~8 h. For the second hybridization, driver cDNA was denatured at 100oC for 20 s then added directly to the pooled mix of the two previous hybridizations and allowed to incubate at 68oC for 20 h. It was necessary to alter the PCR conditions (see below) such that amplification of unwanted sequences was kept to a minimum. All other procedures for generation of the subtracted library were done according to the manual provided with the cDNA subtraction kit (Clontech).
Double-stranded cDNA from the tester (Bsp73-ASML) and the driver (Bsp73-1AS) were separately digested with RsaI (a four base cutter, as used for the construction of the initial subtracted library). HindIII linkers were added to the tester cDNA and EcoRI linkers to the driver cDNA (HindIII linker, 5'-ATCGTCAAGCTTCAAGTTAGCATCG-3', 5'-GCTAACTTGAAGCTTGACGAT-3'; EcoRI linker, 5'-TAGTCCGAATTCAAGCAAGAGCACA-3', 5'-CTCTTGCTTGAATTCGGACTA-3'). Free linkers were removed by preparative agarose gel electrophoresis and the cDNA mixture was amplified as follows. An aliquot of 1 [mu]l adaptor-ligated cDNA was diluted into 1 ml H2O and amplified in a 50 [mu]l reaction using the appropriate primers (HindIII and EcoRI primers), a standard PCR buffer (Pharmacia), 200 [mu]M dNTPs and 2 U Taq polymerase (Pharmacia). Cycling parameters were as follows: 30 cycles of 94oC for 20 s, 52oC for 20 s and 72oC for 2 min. For the subtracted cDNA, PCR conditions were as follows: 27 cycles each of 94oC for 30 s, 68oC for 30 s and 72oC for 2 min. Only the subtracted cDNA was subjected to a second round of PCR (nested), using the same PCR conditions with the exception that 12 cycles were performed. Equal amounts of amplified cDNA from the driver, the tester and of the subtracted library were resolved on a 1.5% agarose gel, blotted and transferred onto nylon membrane. Hybridizations were carried out under stringent conditions in 0.5 M Na2PO4, pH 7.2, 7% SDS at 65oC (12 ). Filters were washed twice in 2* SSC, 0.5% SDS at 68oC, then once in 0.1* SSC, 0.1% SDS at 68oC.
After evaluation of the subtraction efficiency the subtracted library cDNA was cloned directly into pCRII.1 (TA Cloning Kit, Invitrogen). Prior to ligation the subtracted PCR cDNA mix was incubated for a further 1 h at 72oC with additional dATP and Taq DNA polymerase (Pharmacia) to ensure that most of the cDNA fragments contained `A overhangs'. Approximately 100 ng PCR-amplified cDNA were ligated without further purification into 50 ng vector and the ligation mixture was introduced into Electromax bacterial strain DH10B (Gibco BRL) by electroporation (1.8 kV) using an E.coli pulser (BioRad). The library was plated onto 22 * 22 cm ampicillin-containing agar plates and bacteria were grown until colonies were visible. Bacteria were then washed off in LB medium, aliquoted and frozen in 10% DMSO. For library screening the titre was determined and bacteria were plated onto 22 * 22 cm agar plates containing 100 [mu]g/ml ampicillin, 100 [mu]M IPTG and 50 [mu]g/ml X-Gal. Plates were incubated at 37oC until small colonies were visible then incubated further at 4oC until blue/white staining could be clearly distinguished.
A total of 5000 individual recombinant clones were picked and used to inoculate 52 sterile 96-well microtitre plates containing LB medium and ampicillin at 100 [mu]g/ml. After incubation of bacteria on a gyratory shaker for 4 h at room temperature, 5 [mu]l bacterial culture were transferred into 15 [mu]l sterile water in PCR tubes. (This part of the protocol was done in a 96 tube format using a Perkin-Elmer 9600 thermal cycler. Pipetting of PCR mixes was done using a multichannel pipette.) The bacteria were lysed by heating to 100oC for 5 min. Samples of 5 [mu]l bacterial lysate were used to PCR amplify cloned inserts in 50 [mu]l reactions using standard PCR buffer (Pharmacia), 200 [mu]M dNTPs, 2 U Taq polymerase (Pharmacia) and 10 [mu]mol M13 rev and M13(-20) primers (which flank the multiple cloning site of pCRII.1) under the following conditions: 30 cycles each of 94oC for 20 s, 48oC for 20 s and 72oC for 45 s.
After amplification, 12 [mu]l were loaded onto high density gels (Centipedetm gel electrophoresis chambers; Owl Scientific, Woburn, MA). PCR products were denatured and alkaline blotted onto nylon membranes. The filters were hybridized under stringent conditions (7% SDS in 0.5 M NaPO4, pH 7.2; 12 ) with equivalent amounts of 32P-labelled double-stranded cDNA of approximately equal specific activity derived from driver and tester mRNA respectively. Filters were washed under stringent conditions (see above) and exposed to phosphorimager plates overnight (equivalent to 8 days exposure to conventional film). In addition, the filters were exposed to conventional film for up to 12 days at -80oC and the signals of like clones compared.
Aliquots of 4 [mu]g poly(A)+ RNA were fractionated on a 1.4% formaldehyde-agarose gel and blotted in 10* SSC overnight onto Hybond N+ membrane (Amersham). Filters were hybridized at 65oC with the probes using QuickHybtm (Stratagene) and washed according to the manufacturer's guidelines. Probes were generated by PCR amplification of the insert of interest, gel purifying and 32P-labelling the cDNA (ReadyPrime; Amersham). Unincorporated label was removed prior to hybridization using an Elutip (Schleicher & Schüll) according to the manufacturer's specifications. Filters were then exposed to autoradiography film at -80oC for from 4 h to >4 days.
Phenotypic differences between cells of the same organism are determined by differential expression of identical genes. Identification of the genes responsible for these phenotypic differences requires methods which rapidly and efficiently compare the transcripts expressed in the two cell types and which permit isolation of those transcripts found in only one cell type. The recent description of a novel equalizing cDNA subtraction method, termed suppression subtractive hybridization (SSH; 8 ), provides the technical basis for such a comparison. In an attempt to identify changes in gene expression patterns that accompany the progression from a non-metastatic to a metastatic tumour, we subtracted the highly metastatic adenocarcinoma cell line cell line Bsp73-ASML from its non-metastatic counterpart Bsp73-1AS. Both cell lines have previously been used for isolation of a metastasis-associated gene, CD44v (13 ).
The flow diagram of the resulting screening system is depicted in Figure 1 . It starts with the two populations of mRNAs that are to be subtracted. The `driver' is defined as the population of mRNAs that will be eliminated during subtraction, whereas the `tester' population of mRNAs contains in addition the differentially expressed genes of interest. Both mRNA populations are reverse transcribed and subjected to the SSH procedure as described by Diatchenko et al. (8 ). One important feature of the method is that the efficiency of the subtraction process can be easily monitored before any further processing of the subtracted cDNAs is undertaken. The degree of subtraction efficiency can be determined by either monitoring depletion of transcripts common to both populations or enrichment of sequences specific to one population after subtraction.
This work was supported by a grant of the Dr Mildred Scheel Stiftung für Krebsforschung (W53/94/Ho1). The authors would like to thank Harald König, Jonathan Sleeman and Klaus Störmann for valuable discussions and hints and Conny Leicht for her excellent technical assistance.
*To whom correspondence should be addressed. Tel: +49 7247 822781; Fax: +49 7247 823354; Email: hofmann@igen.fzk.d400.de
REFERENCES