| Nucleic Acids Research | Pages |
Identification of fragments of human transcripts froma defined chromosomal region: representational difference analysis of somatic cell hybrids
Introduction
Materials And Methods
Oligonucleotides
Cell lines, mRNA and cDNA isolation
cDNA RDA
Analysis of difference products
`Normalization by PCR' of difference products
Suppression with difference products
Results
Discussion
Acknowledgements
References
Identification of fragments of human transcripts froma defined chromosomal region: representational difference analysis of somatic cell hybrids
ABSTRACT
INTRODUCTION
Representational difference analysis of cDNAs (cDNA RDA) (1) is an effective method to identify fragments of differentially expressed genes (2-7). This technique has been shown to be highly sensitive and was used to identify genes which are expressed in only a very small fraction of the cells from which the Tester was derived (1). This implies that genes with very low levels of expression should also be amenable to isolation, provided they are not expressed in the cells used to prepare the Driver. Therefore, using cDNA RDA, it should be possible to isolate genes from defined chromosomal fragments present in somatic cell hybrids, by comparing these hybrids with the parental cell lines from which they are derived.
To test this possibility, we have applied cDNA RDA to the somatic cell hybrid Q1Z (8), by comparing it with the Chinese hamster ovary cell line Y21 (9), from which it was derived. Cell hybrid Q1Z carries the region of the X chromosome distal to the Fra(X) site at Xq27.3 as the only human chromosomal fragment. The Q1Z cell line was previously used for the isolation of Xq28-derived genes by screening cDNA libraries with Q1Z-derived cosmid clones. This has resulted in the isolation of QM, which is highly expressed in Q1Z (10), and RENBP (11). The Xq27-28 region is one of the most gene dense regions of the human genome (12) and harbours a large number of genes implicated in human disease (13).
In this report we demonstrate that cDNA RDA can be used effectively to isolate gene fragments from defined chromosomal regions. Suppression with major products from highly expressed genes appeared to be advantageous for the isolation of larger number of fragments from less abundant transcripts.
MATERIALS AND METHODS
Oligonucleotides
Sequences of oligonucleotides used in cDNA RDA were as follows: R-Bgl-24, 5[prime]-AGCACTCTCCAGCCTCTCACCGCA-3[prime]; R-Bgl-12, 5[prime]-GATCTGCGGTGA-3[prime]; J-Bgl-24, 5[prime]-ACCGACGTCGACTATCCATGAACA-3[prime]; J-Bgl-12, 5[prime]-GATCTGTTCATG-3[prime]; N-Bgl-24, 5[prime]-AGGCAACTGTGCTATCCGAGGGAA-3[prime]; N-Bgl-12, 5[prime]-GATCTTCCCTCG-3[prime] (1).
Cell lines, mRNA and cDNA isolation
The hamster-human somatic cell hybrid Q1Z (8) and the parental hamster cell line Y21 (9) were kindly provided by Dr S. Warren. Cells were grown in Dulbecco's modified Eagle's medium (DMEM), 10% fetal calf serum, 50 µg/ml gentamycin in a humidified atmosphere of 5% CO2 at 37°C. Poly(A)+ RNA was isolated (Fast-Track 2.0 kit; Invitrogen) and converted into double-stranded cDNA (Copy kit; Invitrogen). In order to avoid differences between samples caused by variations in growth conditions or downstream processing, Q1Z and Y21 were grown side by side and mRNA isolation and cDNA construction were done in parallel. The quality of the cDNAs obtained was tested by hybridization with a probe for the QM (10) and [beta]-actin genes, respectively.
cDNA RDA
Except for a few modifications, the protocol of Hubank and Schatz (1), which was based on the original protocol of Lisitsyn et al. (14), was followed. Detailed protocols were kindly provided by N. Lisitsyn and D. Schatz, respectively. Briefly, cDNA was digested with the four-cutter DpnII (GATC) to obtain maximum representation, ligated to R-BglII adaptors and amplified by PCR to generate amplicons. Adaptors were removed by digestion with DpnII followed by S300-HR spin column chromatography (Pharmacia). To generate the Tester, J-BglII adaptors were ligated to the Q1Z amplicons and excess adaptors were removed using S300 HR spin columns. After subtractive hybridization, the hybridization product was re-amplified as described, to generate DP1, the first difference product. After changing J-BglII adaptors for N-BglII adaptors, a new round of subtractive hybridization was performed to generate DP2, etc. In the first round of subtractive hybridization, a Driver:Tester ratio of 100:1 was used, in the second round 800:1 and in the third round 40 000:1.
PCR was always performed for 20 cycles on a PCT-100 Thermal Cycler (MJ Research). Taq DNA polymerase was obtained from Gibco BRL. Prior to the preparative PCR reactions, analytical PCR reactions were performed to determine the proper amount of input DNA (Fig.
Figure 1. Analytical PCR reactions to determine the optimal amount of input DNA required for preparative PCR of amplicon DNA. Serial 1:5 dilutions of linker-ligated DpnII-digested cDNAs were used as input in 30 µl PCR reactions. Starting concentrations varied from ~1000 pg/µl (lanes 1 and 8) to 0.064 pg/µl (lanes 7 and 14). Twenty µl of each reaction was analysed on a 2% agarose gel which was subsequently blotted onto a nylon membrane. (A) Ethidium bromide staining; (B) hybridization with QM cDNA as probe; (C) hybridization with [beta]-actin cDNA as probe.
Analysis of difference products
Difference products were cloned in BamHI-digested pZero-1 (Invitrogen) and analysed by PCR with M13 sequencing primers (Pharmacia). In one experiment, the difference product was `normalized' prior to subcloning (see below). Plasmid DNA was isolated using Wizard minipreps (Promega) and sequenced using a T7 sequencing kit, according to the instructions of the manufacturer (Pharmacia). Southern blots with Tester and Driver amplicons or with BamHI-digested genomic DNA of Q1Z and Y21 were hybridized with individual difference products as probes. Probes were random prime labelled (Multiprime labelling kit; Amersham). Hybridizations were done overnight in Church hybridization solution (15) at 65°C.
`Normalization by PCR' of difference products
Normalization of difference products was achieved as follows. The difference product was first separated on a standard 2% agarose gel and small amounts (1-5 µl) of agarose, each containing a different size fraction, were obtained by simply stabbing the gel with standard 200 µl pipette tips. The small agarose plugs retained in the end of the tips were pipetted out in 100 µl distilled water. After boiling to dissolve the agarose, 5 µl of each size fraction was used to start a new 30 µl PCR reaction with J-Bgl-24 as primer and 5 µl of each reaction was analysed on gel. Based on the yields obtained, amounts of each PCR were pooled in such a way that a `normalized' difference product was obtained, which was then subcloned.
Suppression with difference products
In order to suppress the appearance of difference products isolated in previous subtraction experiments, these fragments were amplified from the pZero clones with M13 primers, digested with DpnII and 500 ng each were added to the Driver.
Figure 2. Ethidium bromide staining of difference products after electrophoresis on a 2% agarose gel. Lane 1, Q1Z amplicons (Tester); lane 2, Y21 amplicons (Driver); lanes 3 and 4, experiment 1, DP1 and DP2; lanes 5-7, experiment 2, DP1, DP2 and DP3. Experiment 2 differed from experiment 1 in that the two major products obtained in experiment 1 (the two strong bands in lane 4) were added to the Driver prior to subtractive hybridization, to prevent their reappearance.
RESULTS
In order to isolate human Xq28-derived gene fragments, cDNA RDA was performed according to the procedure of Hubank and Schatz (1). To determine the efficacy of the procedure, hybridization with human QM cDNA was used as a specific assay in the first cDNA RDA experiment. The QM gene is specifically expressed in the somatic cell hybrid Q1Z (10) used to generate the Tester and harbours a 510 bp DpnII fragment which should become enriched throughout the cDNA RDA procedure.
The first important step of the RDA is the generation of amplicons by PCR of linker-ligated DpnII-digested cDNA fragments. Titration to determine the proper amount of input DNA for the preparative PCR reactions was achieved by performing a series of so-called analytical PCR reactions in which the amount of input DNA was varied (16). Seven PCR reactions with 1:5 serial dilutions of linker-ligated cDNA fragments as input enabled us to scan a concentration range spanning four orders of magnitude (from ~1000 to 0.064 pg/µl; Fig.
Figure 3. Hybridization of the MP2 fragment (see Fig. 2, lane 4) to Tester and Driver amplicons (A) and to BamHI-digested Q1Z and Y21 genomic DNA (B). The result of the first subtraction experiment is shown in lanes 1-4 of Figure Contrary to expectations, DP2 appeared to be complex. One hundred and seven randomly selected clones were analysed by T-tracking and ~50 clones were sequenced. From the 123 DpnII fragments present in these clones (16 clones contained multiple DpnII fragments as a result of co-ligation during subcloning), 17 fragments were identical to MP1 and 27 fragments were more or less identical to MP2 (see below). The remaining 79 fragments were unique, except for three fragments which were found twice and one fragment which was found three times. The MP2 fragment hybridized strongly to Q1Z amplicons, but no detectable hybridization to Y21 amplicons could be observed (Fig. Four additional Xq28-derived cDNA fragments were identified on the basis of sequence homology with published sequences. Two fragments of 200 and 440 bp were derived from the actin binding protein gene Filamin (FLN1), one 350 bp fragment from the GDP dissociation inhibitor gene (XAP-4) and one 250 bp fragment from the iduronate 2-sulfatase gene (IDS). That these fragments were indeed derived from human Xq28 was experimentally confirmed for both FLN1 fragments by hybridization to BamHI-digested genomic DNA of Y21 and Q1Z. In both cases hybridization was observed to a 10 kb band present in Q1Z only (not shown). Figure 4. cDNAs from the Xq28-derived IDH, IDS, L1-CAM and FLN1 genes. Vertical lines indicate DpnII sites, based on the GenBank or EMBL files with accession nos Z68907 (IDH), M58342 (IDS), X58776 (L1-CAM) and X53416 (FLN1). The cDNA fragments isolated in experiments 1 (DP2) and 2 (DP3) are shown as closed and open rectangles, respectively. The sizes of the isolated cDNA fragments in bp are given at the bottom of each fragment. For all other cDNA fragments, BLAST searches either revealed some degree of homology with various known genes or did not detect any homology with known sequences. In order to reveal whether fragments were of human or hamster origin, 43 different fragments were hybridized against Driver and Tester amplicon blots. Nine fragments failed to hybridize. Twenty-three fragments hybridized to both Tester and Driver and were thus derived from the hamster gene `background'. Three of these cDNA fragments hybridized much more strongly to Tester than to Driver and most likely represent hamster genes which were up-regulated in Q1Z. For 11 fragments, significant hybridization could be detected to Tester amplicons only. However, hybridization of five of these fragments to genomic Southern blots showed that they all hybridized to the same bands in both Y21 and Q1Z and thus also represented up-regulated hamster-derived genes. In order to increase the yield of Xq28-derived sequences, a second subtraction experiment was performed in which the two main products of the first subtraction experiment were added to the driver in order to prevent their reappearance. Also, three rounds of subtraction were performed. In Figure Table 1. Analysis of difference products
DISCUSSION
Using cDNA RDA (1) we have been able to isolate fragments of genes from a defined chromosomal region. Our results indicate that cDNA RDA can be very useful for this purpose, especially because it does not require genomic DNA clones covering the chromosomal region of interest, as is necessary for the direct selection of cDNAs (18,19). This was also shown by a concurrent study of Tajima et al. (20), who were able to identify chromosome 17-specific cDNAs using human-mouse somatic hybrid cell lines.
Once cDNA RDA is operational, performing it is straightforward and a difference product can be obtained in <2 weeks. In our experience, the most time-consuming part of cDNA RDA is the analysis of the difference products obtained. But this is equally true and even more so for other techniques used for the identification of differentially expressed sequences, such as differential display PCR (21) or AFLP-based mRNA fingerprinting (22). In comparison, cDNA RDA allows for much easier cloning of difference products and does not require testing of many different primer combinations and many gel runs. Moreover, these techniques suffer from a bias against rare mRNAs (23,24). In principle, cDNA RDA is also sensitive to this bias, but iteration greatly increases its sensitivity (see below).
Currently, comprehensive techniques for the analysis of differential gene expression have been developed, such as serial analysis of gene expression (25) and hybridization to microarrays (26) or to high density oligonucleotide arrays (27). The latter two techniques require prior sequence information and all three techniques require special equipment and are (still) very expensive. In comparison, cDNA RDA is relatively inexpensive and can also be carried out by small laboratories.
Few small, but in our view important, changes to the protocols developed by Lisitsyn et al. (14) and Hubank and Schatz (1) were made. We incorporated so-called analytical PCR reactions prior to each preparative PCR reaction, which greatly increased the speed and predictability of the procedure. For removal of adaptors we used spin column chromotography instead of purification of amplicons from an agarose gel, which is a potential source of contamination. This procedure was fast and worked well in our hands.
Our first experiment yielded only two major products, the Xq28-derived QM fragment and the MP2 fragment derived from a hamster gene or gene family which was strongly up-regulated in Q1Z.
Repeating the experiment after addition of these two products to the Driver completely suppressed the reappearance of MP1 and MP2 and yielded 14 additional subtraction products, 12 of which came from Xq28 (Table 1), clearly demonstrating the effectiveness as well as the necessity of iteration. To appreciate this, the effect of kinetic enrichment must be considered. Kinetic enrichment is likely the most important factor in achieving the very high enrichment factors (up to 1 000 000-fold after three cycles of subtractive hybridization) observed in RDA (14). In each round of subtractive hybridization, the kinetic enrichment of a given fragment is proportional to the square of its concentration. Initially small differences in concentration will thus become very large after a few rounds of RDA. This has profound implications in the case of cDNA RDA, where amplicons are generated from mRNA populations consisting of mRNA species present at greatly differing concentrations. Because of these differences, cDNA RDA is expected to result in the recovery of only those differentially expressed fragments which were initially present at the highest concentration. Other fragments are lost, despite the fact that the latter fragments may represent genuine subtraction products. The isolation of only two major products, derived from highly expressed genes, in our first experiment can be explained in this way. Without iteration, this would be considered a very disappointing result. However, the isolation of 14 additional subtraction products against a low background in the second experiment clearly demonstrates the effectiveness of this approach. This is in agreement with previously published results (1,5,7).
In both experiments one or more cDNA fragments derived from the same three genes, FLN1, IDS and XAP-4, were isolated, demonstrating the reproducibility of the procedure. Two overlapping fragments of 279 bp (DP3-13) and 62 bp (DP3-37) were isolated from FLN1 (Fig.
From four Xq28 genes, IDH, IDS, L1-CAM and FLN1, more different DpnII fragments were isolated (Fig.
In our view, the most effective future improvement of cDNA RDA would be the simultaneous construction of a cDNA library from the cDNA population used to generate the Tester. After each cycle of cDNA RDA, this library should then be screened with the normalized DP3 or perhaps even DP4 (DP2 will likely still contain too much background), to yield full-length cDNAs rather than cDNA fragments. This would make identification of the genes involved much easier and would also make suppression with previously identified products much more effective. For instance, the addition of the two FLN1 fragments from our first experiment to the Driver would not have suppressed the appearance of any of the four FLN1 fragments in the second experiment, whereas addition of the complete FLN1 cDNA would have suppressed them all.
ACKNOWLEDGEMENTS
We gratefully acknowledge Dr N. Lisitsyn and Dr D. Schatz for providing detailed protocols and Dr M.Schutte for advice. This work was supported by the Dutch Foundation for Scientific Research (NWO) (project-grant no. 901-04-136).
REFERENCES
This article has been cited by other articles:
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 19 Sep 1998
Copyright©Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
B. van de Sluis, J. Rothuizen, P. L. Pearson, B. A. van Oost, and C. Wijmenga
Identification of a new copper metabolism gene by positional cloning in a purebred dog population
Hum. Mol. Genet.,
January 1, 2002;
11(2):
165 - 173.
[Abstract]
[Full Text]
[PDF]
This Article 
Abstract
Print PDF (174K)
Alert me when this article is cited
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Search for citing articles in:
ISI Web of Science (9)
Request Permissions
Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Articles by Groot, P. C.
Articles by van Oost, B. A.
Search for Related Content
PubMed
PubMed Citation
Articles by Groot, P. C.
Articles by van Oost, B. A.
Social Bookmarking
What's this?