DNA covalently bound to an uncharged nylon membrane was used for consecutive amplifications of several different genes by PCR. Successful PCR amplifications were obtained for membrane-bound genomic and plasmid DNA. Membrane-bound genomic DNA templates were re-used at least 15 times for PCR with specific amplification of the desired gene each time. PCR amplifications of specific sequences of p53, p16, CYP1A1, CYP2D6, GSTM1 and GSTM3 were performed independently on the same strips of uncharged nylon membrane containing genomic DNA. PCR products were subjected to restriction fragment length polymorphism analysis, single-strand conformational polymorphism analysis and/or dideoxy sequencing to confirm PCR-amplified gene sequences. We found that PCR fragments obtained by amplification from bound genomic DNA as template were identical in sequence to those of PCR products obtained from free genomic DNA in solution. PCR was performed using as little as 5 ng genomic or 4 fg plasmid DNA bound to membrane. These results suggest that DNA covalently bound to membrane can be re-used for sample-specific PCR amplifications, providing a potentially unlimited source of DNA for PCR.
Since its first report by Mullis and colleagues in 1987 (1 ), PCR has become one of the most powerful and widely used techniques in molecular biology. Thousands of scientific papers are published each year using this technique as a tool for molecular research as well as for clinical applications, such as disease diagnostics (2 ,3 ), occult tumor cell detection (4 -6 ) and prenatal diagnosis of a variety of human genetic disorders (7 ,8 ). Therefore, refinements in PCR protocols can improve DNA analysis and detection for a broad spectrum of medical and research purposes.
The amount of genomic DNA required for PCR ranges, in general, from 50 to 500 ng, with each amplification requiring fresh DNA as template. As the amplification of different genes often requires optimization and standardization, suitable quantities of DNA are required for analysis. However, circumstances may limit the availability of DNA, particularly for DNA isolated from individual tissue samples, including biopsies and archival tissue, where specimen quantities are usually finite and low in quantity. Therefore, re-use of a DNA sample for repetitive PCR amplifications and for amplification of DNA sequences from different genes would decrease the difficulties associated with limited DNA availability for all aspects of genetic testing by PCR.
In this report we describe a simple, novel PCR methodology where DNA, covalently bound to non-charged nylon membranes, serves as template for PCR amplification. We demonstrate that membrane-bound DNA can serve as template for repetitive PCR amplifications of multiple genes without cross-contamination between samples and can be utilized for: (i) genotyping polymorphic alleles, such as those present in specific xenobiotic metabolizing enzyme genes (XMEs); (ii) mutational analysis of oncogenes or tumor suppresser genes, such as p53 and p16.
Squamous cell carcinomas of the oral cavity were obtained by surgical resection from patients at Memorial Sloan-Kettering Cancer Center (New York, NY) as part of a large-scale study examining molecular markers of oral cancer susceptibility as previously described (9 ,10 ). These samples were selected based upon their unique genetic characteristics with respect to gene mutations and/or XME genotypes (see Table 1 ). All tumors were immediately frozen at -70oC after surgical resection. Slices (100-500 mg) of partially thawed tumor tissue were snap-frozen in liquid nitrogen and pulverized prior to homogenization in proteinase K (0.1 mg/ml)-containing TE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA). DNA was extracted as previously described prior to quantitation by spectrophotometery (11 ).
Plasmid pHp53E1CAT, which contains human p53 exon 1 sequences immediately upstream of the chloramphenical acetyltransferase gene (CAT), was kindly provided by Stephen Strudwick (Temple University). Plasmid DNA was purified using a DNA purification kit (Qiagen) as per the manufacturer's instructions and quantitated by spectrophotometery.
Duralon-UV membranes (Stratagene) were initially soaked in deionized, double-distilled water, followed by rinsing in 6* 0.9 M NaCl, 90 mM sodium citrate, pH 7.0, (SSC) buffer prior to slot blot assembly. Wells were washed with 500 [mu]l 6* SSC containing 0.05% bromophenol blue dye under low vacuum. DNA samples (either genomic or plasmid DNA, as indicated in the text) were diluted to 0.1 [mu]g/[mu]l in TE and denatured in 0.3 M NaOH at 60oC for 1 h. Denatured DNA was cooled on ice, diluted 1:1 (v/v) with 12* SSC and applied to slot blot wells under low vacuum in a total volume of 200 [mu]l. An equal volume (200 [mu]l) of 6* SSC buffer (-DNA) was applied to negative control wells. Membrane-bound DNA was immobilized by UV crosslinking in a UV Stratalinker 2400 (Stratagene).
Slot blot membranes containing plasmid or genomic DNA were hybridized with 32P-labeled antisense CAT (+50 to +70 nt relative to the translation initiation start site; GTACATTGAGCAACTGACTG) or human [beta]-actin (+3 to +34 nt relative to the translation initiation start site; CGTTGTCGACGACGAGCGCGGCGATATCATCA) probes respectively, using standard procedures (11 -13 ). Membranes were washed with 6* SSC for 2 min at 22oC, followed by 10 min at 45oC and then exposed for autoradiography.
All PCR amplifications were performed in a GenAmp 9600 thermocycler (Perkin Elmer) using Taq DNA polymerase (Boehringer Mannheim). The standard PCR was composed of a 100 [mu]l reaction volume containing either 100 ng free DNA in solution or DNA blotted and covalently bound to nylon membranes as described above. For covalently bound DNA, membrane fragments corresponding to individual slots were cut using an autoclaved razor. The razor was flamed before each use, with the negative control slot without DNA cut last to assure that cross-contamination between samples was not a factor.
p53 exons 5 and 8 were individually PCR amplified as previously described (9 ,14 ). PCR amplification of p53 exon 5- exon 7 sequences was performed using the Expandtm Long Template PCR System (Boehringer Mannheim) at an annealing temperature of 64oC with the following primers: sense, -43 to -24 nt relative to the 5'-end of exon 5, TTGTGCCCTGACTTTCAACT; antisense, +8 to +27 nt relative to the 3'-end of exon 7, TGTGCAGGGTGGCAAGTGGC. PCR amplification of p53 exon 1 was performed using pHp53E1CAT as template with reaction conditions similar to those described for p53 exon 8 (9 ,14 ). Primers used for p53 exon 1 were: sense, -220 to -200 nt relative to the translation start site, TGTCATGGCGACTGTCCAGCT; antisense, -44 to -20 nt relative to the translation start site, CCAATCCAGGGAAGCGTGTCACCGT. p16 exon 2 was PCR amplified using the following primers: sense, -26 to -7 nt relative to the 5'-end of p16 exon 2, TCTGACCATTCTGTTCTCTC; antisense, +32 to +53 nt relative to the 3'-end of p16 exon 2, CTCTGAGCTTTGGAAGCTCTCA. p16 PCR amplifications were performed as described above for p53 exon 8 (9 ,14 ) in the presence of 5% formamide with an annealing temperature of 58oC for 20 s. CYP1A1 exon 7, GSTM1 exons 4-5 and GSTM3 exons 6-7 sequences were PCR amplified for 36 cycles as previously described (10 ,15 ). CYP2D6 exon 5 sequences were PCR amplified as performed for p53 exon 5 using variable cycle numbers (as indicated in the text) with the following primers: sense, -25 to -7 nt relative to the 5'-end of exon 5, TGAGACCCCGTTCTGTCTG; antisense, +2 to +20 nt relative to the 3'-end of exon 5, ACCGTGGCAGCCACTCTCA. CYP2D6 intron 3-exon 5 sequences were PCR amplified using the same antisense primer in combination with the sense primer GCCTTCGCCAACCACTCCG (-18 to +1 nt relative to the 3'-end of exon 3), utilizing the Expandtm Long Template PCR System as described above. Between 10 and 25% of all amplifications were electrophoresed on 8% polyacrylamide gels to verify the integrity of the PCR bands. Gels were visualized over UV light after staining with ethidium bromide. Where indicated, PCR band intensities (PBI) were determined by computer scanning of gel images using the Photoshop/NIH Image 1.61 analysis system for McIntosh.
To prevent cross-contamination during PCR, all amplifications were performed using fresh, sterile, autoclaved tips, tubes and double-distilled water. Careful attention was given throughout to prevent cross-contamination between samples during DNA purification and isolation. All equipment utilized for tissue blending and homogenization were washed in a bath of concentrated chromic:sulfuric acid, rinsed three times in autoclaved double-distilled water and once in 70% ethanol, air dried and autoclaved. Forceps for membrane handling were autoclaved between experiments and flamed between the handling of each membrane sample. The membrane strips containing immobilized DNA were stripped twice with 150 [mu]l double-distilled water at 65oC for 10 min after each PCR amplification. DNA-bound membrane strips were stored in water at 4oC. Extreme care was taken to avoid cross-contamination between membrane samples.
Mutations in p53 exons 5 and 8 and p16 exon 2, as well as genotyping analysis of the polymorphic CYP1A1Val allele (codon 462 of exon 7) were screened by single-strand conformational polymorphism (SSCP) analysis as previously described (9 ,10 ,14 ). p53 and p16 mutations were verified by dideoxy sequencing as previously described (15 ). The GSTM3 YY1 polymorphism was assessed by restriction fragment length polymorphism analysis of GSTM3 PCR-amplified fragments digested with MnlI (1 U/25 [mu]l reaction) for 16 h at 37oC (16 ). The GSTM1 null (0/0) polymorphism was detected by co-amplification with the homologous GSTM4 gene [serving as a positive control for PCR amplification for GSTM1 (0/0) samples] in a three primer-based assay as previously described (10 ,17 ).
Serial dilutions (10-0.16 [mu]g) of pHp53E1/CAT were blotted onto an uncharged nylon membrane, UV crosslinked and subjected to hybridization with 32P-labeled CAT primer as described in the Materials and Methods. A correlation was observed between hybridization signal intensity and quantity of DNA, as shown in Figure 1 A. Membrane pieces corresponding to each slot were carefully cut and used for PCR amplifications of p53 exon 1 sequences. PCRs using these plasmid DNA-bound membrane strips as template resulted in amplification of the desired product (219 bp; Fig. 1 B). Upon further titration, a decreased intensity of PCR banding after gel electrophoresis and ethidium bromide staining was observed, beginning at 40 pg pHp53E1/CAT DNA (Fig. 1 C), although a detectable 219 bp band was observed for membranes containing as low as 4 fg plasmid DNA. p53 exon 1 sequences were verified by dideoxy sequencing (results not shown). These data suggested that the same membrane-bound DNA samples can be utilized for both hybridization studies as well as PCR.
DNA samples prepared from three oral cavity tumors (MSK73, MSK75 and MSK80) were chosen for this study and were analyzed for mutations in p53 and p16 gene sequences and known polymorphisms present in the GSTM1, CYP1A1 and GSTM3 genes. These samples were chosen based upon the fact that each DNA sample exhibited unique genetic characteristics with respect to gene mutations and/or XME genotypes. The mutation/genotype status of each DNA sample, elucidated in PCRs using free, unbound DNA as template, is shown in Table 1 . Ten micrograms of each genomic DNA sample were blotted and crosslinked to nylon membrane as described in the Materials and Methods. Membrane strips corresponding to each slot were carefully cut and used for successive PCR amplification of several different gene sequences. The membrane pieces with bound DNA were stripped between each PCR amplification and used for 15 consecutive PCRs. The results of alternate amplifications are shown in Figure 2 A. An initial PCR was performed for simultaneous amplification of GSTM1 (275 bp) and GSTM4 (202 bp) (lanes 3-5). Subsequent amplifications using the same DNA-bound membrane strips were also performed for p53 exon 8 (223, lanes 6-8), p16 exon 2 (384 bp, lanes 9-11), p53 exon 5 (261 bp, lanes 12-14), GSTM3 (lanes 15-17) and CYP1A1 exon 7 (see SSCP analysis, Fig. 3 A; the electrophoresis of CYP1A1 exon 7 PCR product on 8% non-denaturing polyacrylamide is not shown). For GSTM3, electrophoresis was performed after restriction enzyme digestion with MnlI. To determine the re-usability of membrane-bound DNA template, some gene sequences were amplified more than once (data not shown). PCR fragments of the appropriate size were amplified in all rounds of PCR for all membrane-bound DNA samples. Negative control membrane pieces were also re-used in PCRs to control for cross-contamination between samples. No amplification was observed for negative control membrane strips (i.e. without DNA) for any round of PCR (round 1 amplification shown in Fig. 2 A, lane 2; results not shown for negative control amplifications for other PCR rounds).
Each of the tumors used for this analysis were from individuals exhibiting specific genotypes for polymorphic XME genes (see Table 1 ). PCR fragments generated by amplification of DNA-bound membrane strips were screened by RFLP or SSCP followed by dideoxy sequencing to confirm gene sequences. As shown in Figure 2 A (lanes 3-5), electrophoresis of PCR-amplified products for the GSTM1 polymorphic gene shows that sample MSK73 is homozygous null for GSTM1 (0/0), while both samples MSK75 and MSK80 exhibit GSTM1-positive genotypes, as indicated by the presence of the GSTM1-specific 275 bp band. The non-polymorphic GSTM4 gene was amplified in all samples as an internal control for PCR amplification (17 ). Some background amplification was observed in GST PCR amplifications (see Fig. 2 A and B, lanes 3-5) due, in all probability, to the use of three primers in this multiplex PCR assay. However, this background was observed for PCRs using both membrane-bound and free DNA as template. In addition, RFLP analysis of GSTM3 gene sequences was performed after MnlI digestion of PCR- amplified product (Fig. 2 A, lanes 15-17). The A/A genotype (digested fragments at 11, 51, 86 and 125 bp) was exhibited for MSK73 and the A/B genotype was observed for samples MSK75 and MSK80 (digested fragments at 11, 51, 86, 125 and 134 bp). The upper 200 bp fragment observed after RFLP of MnlI-digested PCR amplifications of both membrane-bound and free DNA templates exhibiting the AB genotype is similar to that reported previously and has been suggested to be due to amplification of GSTM3-like sequences (16 ). Also, SSCP analysis of CYP1A1 exon 7 PCR-amplified sequences showed that DNA samples MSK73 and MSK75 exhibit a homozygous (Ile/Ile) genotype, whereas MSK80 is heterozygous (Ile/Val), as shown in Figure 3 A. Together, these results are in agreement with PCR results obtained for all polymorphic XME genes using free, unbound DNA for all samples tested (see Table 1 ).
We also studied these membrane-bound tumor DNA samples for mutations in the p53 and p16 tumor suppresser genes, since each tumor was shown to possess a specific mutational spectrum, as shown in Table 1 . SSCP analysis of p53 and p16 showed identical shifts with membrane-bound versus free DNA in PCR amplifications of both p16 (Fig. 3 B) and p53 (data not shown). The SSCP shifted bands were sequenced for both p53 and p16. Identical sequences were observed for PCR amplifications performed with either membrane-bound or free DNA as template, with MSK75 exhibiting a C -> T transition in exon 8 of the p53 gene and MSK80 exhibiting a G -> A transition in exon 2 of the p16 gene (Fig. 3 C). MSK73 exhibited both wild-type p53 and p16 (results not shown). Similar to the situation observed when using free DNA samples as template for PCR, wild-type sequences were also observed for PCR amplifications of p53 exon 5 for all three membrane-bound tumor DNA samples (results not shown). No significant artifactual band shifts or sequencing results were observed for any of the PCR samples amplified using membrane-bound PCR. Together, these results demonstrate that the sequences obtained for PCR-amplified bands using membrane-bound DNA as template were identical to those obtained using free, unbound DNA as template.
PCR is a method of choice for molecular biologists and is widely used for several types of research and clinical applications. In this report we describe a novel methodology, where membrane-bound DNA template can be re-used for PCR amplification of several different gene sequences. To our knowledge this is the first study reporting the use of a re-usable source of DNA as a template for repetitive PCR amplification. We demonstrate that DNA crosslinked by UV light to a solid support such as a nylon membrane could be used for multiple PCR amplifications without any effect from previous reactions. By SSCP screening and DNA sequencing we showed that PCR of membrane-bound DNA results in identical sequences to free DNA PCR amplifications without artifactual polymorphic amplifications caused by UV crosslinking of the DNA to the membrane. In addition, successful amplification was obtained for PCRs with as low as 5 ng membrane-bound DNA and band intensities from PCRs using lower amounts of membrane-bound DNA approached that of PCRs using higher amounts of membrane-bound DNA simply by increasing the cycle number. Furthermore, the efficiency of PCR at lower DNA amounts (i.e. 5-20 ng) approached that of free DNA. The relatively inefficient amplification observed when using higher amounts of membrane-bound DNA for PCR may be due to the fact that although more DNA is loaded into the well, the well surface area remains unchanged. Therefore, less surface area would be available for DNA binding, potentially resulting in increased secondary and tertiary formations within the membrane-bound DNA, which could significantly decrease the relative amounts of DNA available for PCR amplification. Inefficient PCR from membrane-bound DNA was reported previously (18 ), but this was for membrane-bound PCR using a relatively high amount (2 [mu]g) of a cDNA preparation (and not genomic DNA) as template. Together, the results obtained in this study suggest that membrane-bound PCR can be a highly sensitive and efficient technique.
A drawback to this methodology may be that, due to the nature of DNA binding to a solid support such as a nylon membrane, PCR amplification of longer fragments (i.e. >850 bp) appears to be much less efficient than for shorter sequences (<300 bp). We have not fully tested the parameters of membrane-bound PCR for amplification of longer fragments, but PCR amplification of longer fragments is likely to be dependent upon a number of factors, including length of sequence, sequence specificity, DNA quality, PCR conditions, etc. Our data suggest that although membrane-bound PCR can be used for amplification of longer DNA fragments, it appears to be more problematical than when amplifying shorter fragments. More extensive analysis is required to more fully assess the utility of this technique for amplification of larger PCR fragments.
We demonstrate that PCR using membrane-bound DNA as template is highly specific and can be utilized for both genotyping studies as well as mutational analysis. In addition, we demonstrate that one can combine this PCR procedure with membrane hybridization techniques utilized in dot blot or Southern analysis. Therefore, this method may be useful for a variety of clinical and research purposes, particularly for studies involving PCR amplification of DNA isolated from archival tissue specimens or tissue biopsies, where the quantities of isolated DNA are often limited.
We thank Stimson P.Schantz for providing tumor tissue samples, Stephen Strudwick for kindly supplying plasmid pHp53E1/CAT and helpful discussions and Taji Kaur and Jong Park for supplying primers. We also thank Roy White for use of his McIntosh computer and Photoshop/NIH Image 1.61 gel analysis system. This work was supported by NIH grants CA59917 (P.L.) and DE/CA12206 (P.L.).
*To whom correspondence should be addressed. Tel: +1 215 707 5114; Fax: +1 215 707 2781; Email: plazarus@vm.temple.edu
REFERENCES