Minimizing DNA recombination during long RT-PCR

Journal of Virological Methods 76 (1998) 139 – 148

Minimizing DNA recombination during long RT-PCR Guowei Fang a, Guan Zhu a, Harold Burger a,b, Janet S. Keithly a,c, Barbara Weiser a,b,* a

Wadsworth Center, New York State Department of Health, PO Box 22002, 120 New Scotland A6enue, Albany, NY 12201, USA b Albany Medical College, Albany, NY 12208, USA c School of Public Health, SUNY, Albany, NY 12201, USA Received 5 May 1998; received in revised form 18 September 1998; accepted 18 September 1998

Abstract Recent developments have made it possible to reverse transcribe RNA and amplify cDNA molecules of \10 kb in length, including the HIV-1 genome. To use long reverse transcription combined with polymerase chain reaction (RT-PCR) to best advantage, it is necessary to determine the frequency of recombination during the combined procedure and then take steps to reduce it. We investigated the requirements for minimizing DNA recombination during long RT-PCR of HIV-1 by experimenting with three different aspects of the procedure: conditions for RT, conditions for PCR, and the molar ratios of different templates. We used two distinct HIV-1 strains as templates and strain-specific probes to detect recombination. The data showed that strategies aimed at completing DNA strand synthesis and the addition of proofreading function to the PCR were most effective in reducing recombination during the combined procedure. This study demonstrated that by adjusting reaction conditions, the recombination frequency during RT-PCR can be controlled and greatly reduced. © 1998 Elsevier Science B.V. All rights reserved. Keywords: HIV-1 RT-PCR; HIV-1 recombination; Long RT-PCR

1. Introduction Recent improvements in the polymerase chain reaction (PCR) have made it possible to amplify DNA molecules of up to 35 kb in length (Barnes, * Corresponding author. Tel.: +1-518-473-3546; fax: + 1518-473-4110; e-mail: [email protected].

1994; Cheng et al., 1994). This increase in the capability of PCR stems primarily from improvements in two major areas, cloned thermostable DNA polymerases and thermocycling conditions, which together permit the amplification of long fragments of cloned as well as complex mixtures of DNA (Barnes, 1994; Cheng et al., 1994; Salminen et al., 1995). The combination of reverse

0166-0934/98/$ - see front matter © 1998 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 6 - 0 9 3 4 ( 9 8 ) 0 0 1 3 3 - 5

140

G. Fang et al. / Journal of Virological Methods 76 (1998) 139–148

transcription and PCR (RT-PCR) has expanded the power of the PCR procedure so that it now encompasses the analysis of rare transcripts, tissue-specific gene expression, and viral RNA genomes including HIV-1 (Nathan et al., 1995; Fang et al., 1996; Gow et al., 1996; Holterman et al., 1996). Long RT-PCR is a particularly useful tool for studying the HIV-1 genome. During HIV-1 infection in vivo, the plasma viral RNA is more representative of the replicating virus pool at any point in time than the cell-associated proviral DNA; it also represents the pathogenetically significant species (Coffin, 1995; Chun et al., 1997). For these reasons, we developed a method of using RTPCR to clone the full-length HIV-1 genome as a single molecule directly from plasma viral RNA (Fang et al., 1996). HIV-1, like many RNA viruses, exhibits a high degree of genetic diversity stemming from the high misincorporation rate of the reverse transcriptase (RTase) (Coffin, 1995). As a consequence, populations of HIV-1 RNA genomes in vivo, known as quasispecies, are often very heterogeneous (Eigen, 1993). For accurate characterization of viral RNA genomes derived from HIV-1-infected individuals, therefore, it is necessary to limit artifactual recombination between quasispecies which may occur during laboratory analysis. Experimental evidence for recombination between quasispecies in vitro exists; in a previous report, PCR co-amplification of a 294 bp DNA fragment from two distinct HIV-1 tat gene sequences led to the formation of recombinant DNA molecules constituting up to 5.4% of the total product (Meyerhans et al., 1990). If recombination is a frequent event during long RT-PCR amplification (\4 kb), it may pose a problem for studying heterogeneous genetic material such as HIV-1 RNA in vivo. PCR-generated recombinants are molecules composed of parts of two or more different sequences brought together during PCR (Saiki et al., 1988). Recombinants presumably arise due to the presence of incompletely extended primers annealing to a heterologous target sequence (Meyerhans et al., 1990). During the RT-PCR procedure, the RNA is copied into cDNA by RT, and the cDNA is then used as a template for PCR

amplification. Numerous steps in this process may produce incomplete DNA strands which in turn may result in PCR-mediated recombination. First, because RNA is more labile than DNA, RNA may be readily damaged or degraded during extraction and storage; the damaged RNA molecule may then result in incomplete cDNA synthesis during the RTase reaction that follows. Second, a portion of the cDNAs synthesized during RT may be incomplete because of the low efficiency of retroviral RTases. Finally, the PCR amplification of the cDNA may terminate prematurely, also resulting in incomplete DNA strands. Every incomplete cDNA or DNA strand that is produced during each step of the RT-PCR procedure may lead to PCR-mediated recombination and therefore has the potential to compromise the fidelity of this method. We therefore performed a systematic investigation of the requirements for minimizing recombination during long RT-PCR.

2. Materials and methods

2.1. Viral strains and 6iral RNA isolation HIV-1 virions were obtained from molecular clones of HIV-1 strains HIV-1JR-CSF and HIV1NL4-3 (Myers et al., 1996) by transfecting plasmid DNA into normal human dermal fibroblasts (NHDF 710, Clonetics, San Diego, CA) and cocultivating with peripheral blood mononuclear cells (PBMCs) (Fang et al., 1995). Virus cultures were centrifuged at 18 500× g for 1 h at 10°C to pellet virions. Viral RNA was then extracted from the virus (Fang et al., 1995, 1996). The RNA was treated with RQ2 RNase-free DNase (Stratagene, La Jolla, CA, USA) to remove any contaminating DNA.

2.2. Re6erse transcription quantitati6e competiti6e PCR To determine the number of HIV-1 RNA copies, we used a modified reverse transcription quantitative competitive PCR (RT QC-PCR) method which has been described in detail previously (Fang et al., 1995).

G. Fang et al. / Journal of Virological Methods 76 (1998) 139–148

141

Fig. 1. Strategies for determining recombination frequency in long RT-PCR. Viral RNAs were mixed, and reverse transcription was initiated from the 3% end of the RNA genome using an oligo dT primer. The 3% half of the cDNA product was PCR-amplified with primers FGF46 and FGR91. The 4.5 kb PCR product was cloned into a TA vector and probed with probes NL-1, NL-6, CSF-1 and CSF-6. Colonies that hybridized with NL-1 and CSF-6 or CSF-1 and NL-6 were designated recombinants. Colonies that hybridized with NL-1 and NL-6 or CSF-1 and CSF-6 were considered as wild type. Colonies that hybridized with only one of the probes were considered as incomplete and were not counted.

2.3. Long RT-PCR and molecular cloning The strategy for long RT-PCR is illustrated in Fig. 1. The long RT-PCR reaction was carried out as described previously (Fang et al., 1996), with the following changes: (1) three PCR temperatures were used instead of two (94°C for denaturation, 55°C for annealing, and 72°C for elongation); (2) the concentration of PCR primers used was 0.02 mM; (3) approximately 100 000 copies of HIV-1 RNA genome were used as templates. The procedure used Moloney murine leukemia virus (M-MuLV) reverse transcriptase (BioLabs, Beverly, MA, USA) and rTth XL DNA polymerase (Perkin Elmer, Foster City, CA, USA). For long PCR of plasmid DNA, the GeneAmp

XL PCR kit with rTth XL DNA polymerase and the GeneAmp PCR Reagent kit with AmpliTaq DNA polymerase (Perkin-Elmer) were used following the manufacturer’s instructions. The reaction was carried out in a MicroAmp Reaction tube (Perkin Elmer) in a volume of 100 ml containing 1X XL buffer II or 1× PCR buffer II, 200 mM dNTP mixture, 1.2 mM Mg(OAc)2 or MgCl2 solution, 4 U of rTth XL DNA polymerase or AmpliTaq DNA polymerase, 4 pg of DNA template, and 0.02 mM of each primer. The mixture, in a final volume of 100 ml, was subjected to 35 cycles of amplification in a GeneAmp PCR System 9600 thermal cycler (Perkin Elmer). After 35 cycles, the samples were extended for another 10 min at 72°C. The PCR products were assayed on 0.8% SeaKem GTG agarose gel (FMC, Rock-

142

G. Fang et al. / Journal of Virological Methods 76 (1998) 139–148

land, ME, USA). A hot-start protocol (Chou et al., 1992) using AmpliWax® PCR Gem beads (Perkin Elmer) was used in each PCR to minimize undesired primer interaction. All template concentrations used are stated in the table legends. The study was focussed on the 3% half of the genome because it encompasses regions of sequence variation between the two HIV-1 strains under investigation, NL4-3 and JR-CSF (Myers et al., 1996), and therefore is useful to assess recombination. The primer pair chosen to amplify the 3% half (4.5 kb fragment) of the HIV-1 genome was: FGF46 (5%-GCA TTC CCT ACA ATC CCC AAA G-3%) and FGR91 (5’-TAG TTC TGC CAA TCA GGG AAG TAG C-3%). Both of the test HIV-1 strains contain these sequences, and when assayed, were both amplified equally well using identical PCR conditions. After verification of fragment size, the PCR products were run on a 0.8% SeaKem GTG agarose gel; desired DNA bands were then cut out of the gel and DNA was purified using a QIAquick Gel Extraction (Qiagen, Chatsworth, CA, USA). The purified DNA fragments were directly ligated into a phagemid vector (TA plasmid) by using the Original TA Cloning Kit (Invitrogen, Carlsbad, CA, USA). After transformation of E. coli strain DH5a, white colonies from X-Gal plates were screened for recombinants.

In the same way, CSF-1 (5%-TAA GGG ATG GAT TTA TAA ACA TC-3%) and CSF-6 (5%-AAC ATA GTG TGC CTG GAT GGT-3%) were specific to the HIV-1 JR-CSF strain and located at nucleotide 5157–5179 and 8803–8823, respectively (Fig. 1). Oligonucleotides were labelled with [32P]-ATP using T4 polynucleotide kinase (Promega, Madison, WI, USA), and cleaned with G-25 Quick SpinTM columns (Boehringer Mannheim, Indianapolis, IN, USA). To screen for recombinants, each nitrocellulose replicate sheet was first pre-hybridized for 4–6 h and then hybridized to one of the four probes overnight at 50°C in the hybridization buffer containing 6 × SSC, 0.5% SDS, 5 × Denhardt’s solution and 0.1 mg/ml salmon sperm DNA. Membranes were washed with 6× SSC/0.1% SDS twice at room temperature for 10 min and once at 60°C for 15 min, and then exposed to X-ray film overnight. Colonies that hybridized with either NL-1 and CSF-6 or CSF-1 and NL-6 were considered as recombinants; colonies that hybridized only with NL-1 and NL-6 or only CSF-1 and CSF-6 were considered as wild type. Colonies that hybridized with only one of the probes were considered as incomplete products and were not counted.

3. Results

2.4. Screening for recombinants About 100–200 white bacterial colonies from each reaction were picked and transferred to a nitrocellulose filter (Schleicher & Schuell, Keene, NH, USA). The membranes were laid onto LB agar/ampicillin (50 mg/ml) plates and incubated at 370C overnight. At least 4 nitrocellulose replicates of the bacterial colonies were made for each reaction. Two distinct sets of oligonucleotide probes were used to detect sequences specific for each of the two HIV-1 strains under investigation (Myers et al., 1996). NL-1 (5%-TAA GGA CTG GTT TTA TAG AGA TC-3%) and NL-6 (5%-AAA GTA GTG TGA TTG GAT GGC-3%) were specific to the HIV-1 NL4-3 strain and located at nucleotide positions 5145–5167 and 8806 – 8827, respectively.

The strategy for long RT-PCR is illustrated in Fig. 1 and is carried out as described in Section 2. To examine the requirements to minimize recombination during long RT-PCR, we experimented with multiple conditions affecting RT and PCR as well as with different ratios of two distinct templates, HIV-1JR-CSF and HIV-1NL4-3. Recombinants were detected using strain-specific probes as described in Section 2.

3.1. RNA RT-PCR 6ersus DNA PCR PCR-mediated recombination has been previously studied in DNA PCR amplification. What has not yet been determined, however, is the recombination frequency for the combined procedure of RT followed by PCR. In particular, it is

G. Fang et al. / Journal of Virological Methods 76 (1998) 139–148

not known how the recombination frequency for PCR amplification of a given DNA sequence compares with the recombination rate for RTPCR using the same RNA sequence as template. To differentiate the frequencies of recombination resulting from RT-PCR versus amplification of cDNA by PCR alone, 4.5 kb fragments were amplified from equimolar mixtures of HIV-1JR-CSF RNA and HIV-1NL-43 RNA using RT-PCR and from plasmid DNA templates of identical sequence using DNA PCR. Table 1 shows the resultant frequencies of recombination. Under identical PCR conditions, RT-PCR using an RNA template resulted in much higher recombination frequencies than PCR using plasmid DNA templates (6.49 vs. 2.65%). The higher recombination frequency may be related to the presence of partial RNA templates that result from the labile and fragile nature of RNA as well as from incomplete DNA strands. The results suggest that in RT-PCR, the factors affecting the RT step, such as the RNA quality and RT conditions, influence the degree of recombination more than the factors affecting the cDNA amplification step.

3.2. RT reaction time Having established that the RT step itself plays a large role in recombination during RT-PCR, we Table 1 Recombination frequencies of RT-PCR and plasmid DNA PCRa Procedure

Number of recombinants Total colonies counted Recombination rate (%)

RNA RT-PCR

Plasmid DNA PCR

12

4

185

151

6.49

2.65

143

Table 2 Relationship between RT reaction time and recombination frequencya RT reaction time (min)

Number of recombinants Total colonies counted Recombination rate (%)

15

60

120

21 168 12.50

12 185 6.49

10 170 5.88

a

Recombination frequencies of RT-PCR using 100 000 copies each of HIV-1JR-CSF and HIV-1NL4-3 RNA templates and different RT reaction times. The synthesized cDNA was subjected to 35 cycles of PCR with 6 min PCR extension.

set out to identify which elements of the process may be adjusted to reduce the recombination rate. To determine the relationship between the RT reaction time and recombination frequency following cDNA amplification by PCR, equal quantities of HIV-1 RNA (100 000 copies) extracted from two HIV-1 strains, JR-CSF and NL4-3, were mixed and incubated for 15, 30, 60, and 120 min under the RT conditions described previously (Fang et al., 1995). An oligo dT primer was used to start cDNA synthesis from the 3% end of the RNA genome. The synthesized cDNAs were subjected to 35 cycles of PCR using a 6 min extension time. (Six min has been determined to be the optimal PCR extension time under these conditions, with data presented below). Table 2 demonstrates the frequencies of RTPCR mediated recombination in relation to RT reaction time. Recombinants were identified after all reaction times, but fewer recombinants were detected after using longer RT reaction times. The frequency of recombination was 12.50% when the RT reaction time was 15 min and only 6.49% when the reaction lasted 60 min. Increasing the reaction time even further to 120 min did not result in a significant decrease in recombination frequency.

a

Equimolar mixtures of HIV-1JR-CSF and HIV-1NL4-3 RNA (100 000 copies each) or plasmid DNA (4 pg each) were used as templates. The primer pair chosen for amplification of the 4.5 kb fragment from the 3% half of the HIV-1 genomes was FGF46 and FGR91. The cDNA and plasmid DNA templates were subjected to 35 cycles of PCR (94°C, 15 s for denaturation; 55°C, 30 s for annealing; and 72°C, 6 min for extension).

3.3. Molar proportions of mixed templates In infected individuals, different HIV-1 quasispecies are rarely present in equimolar proportions (Meyerhans et al., 1989; Eigen, 1993). We there-

144

G. Fang et al. / Journal of Virological Methods 76 (1998) 139–148

Table 3 Relationship of molar ratios of mixed templates and RT reaction time to recombination frequencya JR-CSF:NL4-3 ratio

RT reaction time (min) Number of recombinants Total colonies screened Recombination rate (%)

1:1

1:10

10:1

1:1

1:10

10:1

15 21 168 12.50

15 4 132 3.03

15 3 147 2.04

60 12 185 6.49

60 2 129 1.55

60 0 144 0

a

HIV-1JR-CSF and HIV-1NL4-3 template RNAs were mixed in quantities of 100 000:100 000 copies as a 1:1 ratio, 100 000:10 000 as a 10:1 ratio, and vice versa. The synthesized cDNA was subjected to 35 cycles of PCR with 6 min PCR extension.

fore determined the recombination frequency for RT-PCR using varying proportions of RNA template from two distinct virus strains; we also varied the RT reaction times in this series of experiments. HIV-1JR-CSF RNA and HIV- 1NL4-3 RNA were mixed in proportions of 10:1, 1:10, and 1:1 (100 000:10 000 copies, 10 000:100 000 copies and 100 000:100 000 copies). The template mixtures were then RT-PCR amplified using RT reaction times of 15 or 60 min and cloned as described above. Mixtures containing unequal proportions of the two RNA templates led to recombination frequencies that were greatly reduced as compared to frequencies using equimolar proportions (Table 3). Furthermore, experiments using both unequal proportions of template RNA and the longer RT reaction time resulted in the greatest reduction in the recombination frequency (Table 3).

HIV-1 NL4-3 (4 pg each) was subjected to 35 cycles of PCR under identical conditions using 2, 4, 6, or 8 min extension times. Amplified 4.5 kb fragments were cloned into TA vectors and screened for recombinants. Short extension times of 2 and 4 min resulted in relatively high recombination frequencies of 10.46 and 7.07% (Table 4), but by increasing the extension time to 6 min the recombination frequency was reduced to only 2.65%. Additional prolongation of PCR extension time beyond 6 min did not, however, reduce the recombination rate even further. An extension time of 8 min resulted in a recombination rate of 3.45%, and unspecified PCR products accumulated as well, probably because the enzyme activity decreased over the long extension time. Because a 6 min extension time gave the lowest recombination frequency after

3.4. PCR extension time

Table 4 Relationship between PCR extension time and recombination frequencya

We next examined factors that may affect recombination during the PCR process. To avoid PCR-mediated recombination, it is necessary to achieve complete strand synthesis. It was reported that when PCR extension time is increased, the frequency of recombination is reduced (Meyerhans et al., 1990). To identify the extension time necessary to achieve minimum recombination without compromising yield and specificity, we determined the recombination frequencies for long DNA PCR using a range of extension times. The rTth XL polymerase was used. A series of equimolar mixtures of plasmid HIV-1JR-CSF and

Extension time (min) 2 Number of recombinants Total colonies counted Recombination rate (%)

4

6

8

9

7

4

3

86

99

151

87

10.46

7.07

2.65

3.45

a Equimolar mixtures of HIV-1JR-CSF and HIV-1NL4-3 plasmid DNA (4 pg each) were used as templates. The DNA template was subjected to 35 cycles of PCR with 6 min for extension.

G. Fang et al. / Journal of Virological Methods 76 (1998) 139–148 Table 5 Recombination frequencies of PCR using different polymerase combinationsa Polymerases Number of recombinants Total colonies counted Recombination rate (%)

rTth+Vent

Taq

Taq + Pfu

4

9

7

151

69

112

2.65

13.04

6.25

a Equimolar mixtures (4 pg each) of HIV-1JR-CSF and HIV1NL4-3 DNA templates were amplified using different polymerase combinations. The plasmid DNA templates were subjected to 35 cycles of PCR with 6 min for extension.

amplification of the 4.5 kb HIV-1 fragment, it was used in the other experiments.

3.5. rTth XL 6ersus Taq polymerase (DNA PCR) We next compared recombination frequencies of long DNA PCR using two PCR kits from Perkin Elmer: the GeneAmp XL PCR kit, which uses rTth and Vent DNA polymerases, and the GeneAmp PCR Core reagent kit, which uses AmpliTaq DNA polymerase. In addition, we tested a combination of AmpliTaq and Pfu DNA polymerases (Stratagene). Using these different polymerases and a 6 min extension time, we amplified 4.5 kb HIV-1 sequences from equimolar mixtures of two different plasmid DNAs. As seen in Table 5, the GeneAmp XL PCR kit has a much lower recombination rate (2.65%) than the GeneAmp PCR Reagent kit (13.04%). The GeneAmp XL PCR kit is designed specifically to amplify long DNA fragments and is supplied with a small amount of Vent polymerase in addition to rTth polymerase. These two polymerases work in concert to permit completion of strand synthesis. Vent polymerase possesses a thermostable 3%- to 5%-exonuclease activity and removes mismatched 3%-terminal bases, which may cause synthesis to terminate prematurely. The dominant polymerase, rTth, can then complete strand synthesis (Barnes, 1994; Cheng et al., 1994). In addition, the rTth polymerase itself performs better in long PCR than the Taq polymerase (Myers and

145

Gelfand, 1991; Foord and Rose, 1994). In this way, the GeneAmp XL PCR kit has the capacity to make longer PCR products. The standard GeneAmp PCR Reagent kit with Taq polymerase, by contrast, lacks proofreading activity. We observed in the experiment described here that the yield of complete 4.5 kb PCR products when using the GeneAmp PCR Reagent kit was B 1% of the yield when using the GeneAmp XL PCR kit; this result suggests that a significant portion of the PCR products amplified by Taq polymerase may represent prematurely terminated strands. These strands may be caused by a mismatched 3’ terminal base resulting from the lack of 3%- to 5%-proofreading activity. These incomplete strands may lead in turn to increased recombination. To ask if the low recombination frequency seen with the GeneAmp XL PCR kit is at least partially due to its proofreading activity, a small amount (0.02 U) of Pfu polymerase, which has a proofreading function, was added to a PCR reaction using Taq polymerase (2 U) and the Gene Amp PCR Reagent kit. As expected, the addition of Pfu effectively reduced recombination (6.26 vs. 13.04%, Table 5) and increased the yield of PCR product. Nonetheless, the GeneAmp XL PCR kit (rTth polymerase with addition of Vent polymerase) performed better, with a lower recombination frequency. Our results suggest that to increase efficiency and decrease recombination, Taq DNA polymerase itself is not a good choice for long PCR amplification (\ 4 kb).

4. Discussion To evaluate the conditions that influence recombination during RT-PCR, experiments were carried out using conditions that favor formation of recombinants, namely: two distinct HIV-1 strains, a high initial input of RNA, an equimolar mixture of the two types of template molecules, and a limited primer concentration. These conditions were chosen in order to identify the factors predisposing to recombination. As such, our data represent the upper estimates for recombinational events in an otherwise typical RT-PCR reaction.

146

G. Fang et al. / Journal of Virological Methods 76 (1998) 139–148

The experiments presented here suggest that with the reagents available at this time, some degree of DNA recombination during long RTPCR is inevitable, but the frequency of recombination may be controlled and greatly reduced. By assessing DNA recombination frequencies using two different HIV-1 strains, we first showed that recombination frequencies resulting from the combined procedure of long RT and PCR are more than twice those of the single procedure of DNA PCR alone even when using the same sequences as template (Table 1). We then set out to examine conditions necessary to minimize recombination, with the experiments guided by previous work on recombination during in vitro RT and PCR. The formation of recombinant molecules may be due to one of three potential problems. First, the RNA template may be damaged or degraded during extraction and storage; second, the cDNA strands produced by RT may be incomplete; and third, the PCR may terminate prematurely, resulting in incomplete DNA strands. Because an intact RNA template is essential for accurate and complete transcription, obtaining high quality, intact RNA is the first and the most critical step in undertaking long RT-PCR. The extraction of RNA is potentially more problematic than extraction of DNA because RNA is more labile, especially at elevated temperature, high pH, or both. RNA is also prone to endonucleolytic and exonucleolytic attack from a variety of ribonucleases. Despite these difficulties, it is possible to isolate RNA consistently and reliably, even from tissues with very high levels of endogenous ribonuclease activity. Various methods have been described for isolation of high-quality RNA for use as RT-PCR template (Chomczynski and Sacchi, 1987; Boom et al., 1990; Muir et al., 1993) and we have used methods to isolate high-quality HIV-1 RNA from plasma virions (Fang et al., 1995, 1996). To reduce the problem of incomplete strand synthesis and consequent suboptimal template for amplification, we first examined RT reaction time. The results indicated that RT reaction time is a factor affecting recombination frequency and can be adjusted to reduce this unwanted effect (Table 2). Short RT reaction times (e.g. 15 min) resulted

in higher recombination rates, probably because incomplete cDNA strands produced during the short reaction time may anneal to longer strands during cDNA amplification by PCR. It was not possible, however, to eliminate completely recombination by prolonging reaction times to 120 min; multiple experiments in our lab suggest that a 60 min RT reaction time is suitable for most cDNA synthesis. The failure to reduce the recombination frequency by increasing RT reaction times suggests that some of the template RNA molecules are incomplete. These incomplete templates may have been damaged or degraded during extraction and storage, probably by physical or chemical factors such as elevated temperature or pH or by ribonucleases. In additional studies aimed at producing complete strand synthesis of DNA, we examined PCR extension time. Our data confirm that PCR extension time is directly related to the frequency of recombination. Although a short extension time results in more incomplete DNA strands, excess extension time causes formation of non-specific PCR products, and both may lead to recombination. The optimal PCR extension time should be determined for each template, polymerase, and set of primer pairs. In our experiments, a 6 min PCR extension gave the lowest recombination rate when amplifying a 4.5 kb cDNA using rTth polymerase. Based on several considerations, we suggest that an extension time of 1.5–2 min/kb of template DNA will be effective in minimizing recombination in most long PCRs. This suggestion is based on the results presented here as well as the Taq and rTth DNA polymerase extension rates previously determined at up to 60 nucleotides per second in standard PCR (Gelfand, 1989; Myers et al., 1996). In addition, the auto-extension feature, which prolongs later cycles, was also used to reduce the formation of incomplete and non-specific PCR products (Cheng et al., 1994). Another means of ensuring complete strand synthesis is the introduction of proofreading function for DNA polymerase reactions. In our experiments, the combination of rTth plus Vent DNA polymerase or Taq plus Pfu DNA polymerase resulted in a higher PCR efficiency and a lower

G. Fang et al. / Journal of Virological Methods 76 (1998) 139–148

recombination frequency than experiments using Taq polymerase alone. These results most likely occurred because of proofreading activity provided by Vent and Pfu polymerases, which leads to fewer incomplete PCR products and reduced recombination. Our data indicate that complete strand synthesis, a requirement for reducing recombination, is facilitated by addition of a proofreading activity. This activity removes mismatched nucleotides and permits the predominant polymerase activity to complete strand synthesis (Barnes, 1994; Cheng et al., 1994). Taq polymerase has been used in long PCR when supplemented with a low level of an additional thermostable proofreading DNA polymerase. Experiments presented both here and by others (Cheng et al., 1994), however, provide evidence that the rTth and Vent polymerase combination produces the most reliable polymerase activity (Cheng et al., 1994) and the lowest frequencies of recombination in long PCR. On a practical level, a previous investigation of DNA PCR as a single procedure found that viral recombination only affected the characteristics of HIV-1 proviral populations to a small degree (Meyerhans et al., 1989). That study determined that in a PCR reaction, priming of oligonucleotides, which are usually added in high concentration, competes extremely effectively with priming of overlaps of damaged DNA strands, which occur at low concentrations and lead to DNA recombination (Colgan, 1993). In our RT-PCR experiments, using unequal amounts of two input RNAs resulted in much less recombination than using equimolar mixtures of templates. Under optimum reaction conditions, long RT-PCR is unlikely to produce significantly higher recombination rates than PCR of shorter molecules. There may be occasions when PCR-mediated recombination presents an opportunity rather than a problem. Meyerhans et al. (1990) suggested that under optimized conditions for the recovery of recombinants, PCR-mediated recombination can be exploited to produce chimeric molecules. Several papers have described directed joining of DNA fragments by PCR-mediated recombination (Horton et al., 1989; Yolov and Shabarova, 1990; Gu et al., 1991; Klug et al., 1991; Sandhu et al.,

147

1992). We recently developed methods in the laboratory to create full-length HIV-1 chimeric molecules based on PCR-mediated recombination (manuscript in preparation). Chimeric molecules may be useful in HIV-1 research and other fields as well.

Acknowledgements This work was supported in part by grants from the National Institute of Allergy and Infectious Diseases and the National Institute on Drug Abuse (RO1A133334 and U01AI35004). We thank Ellen Shippey and Anne Klugo for help in preparing the manuscript and the Wadsworth Center Molecular Genetics Core Laboratory for oligonucleotide synthesis for PCR.

References Barnes, W.M., 1994. PCR amplification of up to 35-kb DNA with high fidelity and high yield from lambda bacteriophage templates. Proc. Natl. Acad. Sci. USA 91, 2216 – 2220. Boom, R., Sol, C.J.A., Salimans, M.M., Jansen, C.L., Wertheim-van Dillen, P.M.E., Van der Noordaa, J., 1990. Rapid and simple method for purification of nucleic acids. J. Clin. Microbiol. 28, 495 – 503. Cheng, S., Fockler, C., Barnes, W.M., Higuchi, R., 1994. Effective amplification of long targets from cloned inserts and human genomic DNA. Proc. Natl. Acad. Sci. USA 91, 5695 – 5699. Chomczynski, P., Sacchi, N., 1987. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenolchloroform extraction. Anal. Biochem. 162, 156 – 159. Chou, Q., Russell, M., Birch, D.E., Raymond, J., Bloch, W., 1992. Prevention of pre-PCR mis-priming and primer dimerization improves low-copy-number amplifications. Nucl. Acids. Res. 20, 1717 – 1723. Chun, T.W., Carruth, L., Finzi, D., et al., 1997. Quantification of latent tissue reservoirs and total body viral load in HIV-1 infection. Nature 387, 183 – 187. Coffin, J.M., 1995. HIV population dynamics in vivo: implications for genetic variation, pathogenesis, and therapy. Science 267, 483 – 489. Colgan, D.J., 1993. DNA regeneration in the polymerase chain reaction. J. Theor. Biol. 162, 289 – 307. Eigen, M., 1993. Viral quasispecies. Sci. Am. 269, 42 – 49. Fang, G., Weiser, B., Visosky, A., Townsend, L., Burger, H., 1996. Molecular cloning of full-length HIV-1 genomes directly from plasma viral RNA. J. AIDS 12, 352 – 357.

148

G. Fang et al. / Journal of Virological Methods 76 (1998) 139–148

Fang, G., Burger, H., Grimson, R., et al., 1995. Maternal plasma human immunodeficiency virus type 1 RNA level: a determinant and projected threshold for mother-to-child transmission. Proc. Natl. Acad. Sci. USA 92, 12100–12104. Foord, O.S., Rose, E.A., 1994. Long-distance PCR. PCR Methods Appl. 3, 149–161. Gelfand, D.H., 1989. Taq DNA polymerase. In: Erlich, H.E. (Ed.), PCR Technology—Principles and Applications for DNA Amplification. Stockton Press, New York, pp. 17–22. Gow, J.W., Margaret, M.M., Behan, W.M.H., Behan, P.O., 1996. Long RT-PCR amplification of full–length enterovirus genome. BioTechniques 20, 582–584. Gu, H., Planas, J., Gomez, R., Wilson, D.J., 1991. Full length mouse glycophorin gene constructed using recombinant polymerase chain reaction. Biochem. Biophy. Res. Commun. 177, 202 – 208. Holterman, L., Mullins, J.I., Haaijman, J.J., Heeney, J.L., 1996. Direct amplification and cloning of up to 5-kb lentivirus genomes from serum. BioTechniques 21, 312–319. Horton, R.M., Hunt, H.D., Ho, S.N., Pullen, J., Pease, L.R., 1989. Engineering hybrid gene without the use of restriction enzymes: gene splicing by overlap extentsion. Gene 77, 61– 68. Klug, J., Wolf, M., Beato, M., 1991. Creating chimeric molecules by PCR directed homologous DNA recombination. Nucl. Acids. Res. 19, 2793. Meyerhans, A., Vartanian, J.P., Wain-Hobson, S., 1990. DNA recombination during PCR. Nucl. Acids. Res. 18, 1687– 1691.

Meyerhans, A., Cheynier, R., Albert, J., et al., 1989. Temporal fluctuations in HIV quasispecies in vivo are not reflected by sequential isolations. Cell 58, 901 – 910. Muir, P., Nicolson, F., Jhetam, M., Neogi, S., Banatvala, J.E., 1993. Rapid diagnosis of enterovirus infection by magnetic bead extraction and polymerase chain reaction detection of enterovirus RNA in clinical specimens. J. Clin. Microbiol. 31, 31 – 38. Myers, G., Korber, B., Foley, B., Jeang, K.T., Mellors, J., Wain-Hobson, S. (Eds.), 1996. Human Retroviruses and AIDS. Los Alamos National Laboratory, Los Alamos, NM Myers, T.W., Gelfand, D.H., 1991. Reverse transcription and DNA amplification by a Thermus thermophilus DNA polymerase. Biochemistry 30, 7661 – 7666. Nathan, M., Mertz, L.M., Fox, D.K., 1995. Optimizing long RT-PCR. Focus 17, 78 – 80. Saiki, R.K., Gelfand, D.H., Stoffel, S., et al., 1988. Primer-directed enzymatic amplication of DNA with a thermostable DNA polymerase. Science 239, 487 – 491. Salminen, M.O., Koch, C., Sanders-Buell, E., et al., 1995. Recovery of virtually full – length HIV-1 provirus of diverse subtypes from primary virus cultures using the polymerase chain reaction. Virology 213, 80 – 86. Sandhu, G.S., Aleff, R.A., Kline, B.C., 1992. Dual asymmetric PCR: one-step construction of synthetic genes. BioTechniques 12, 14 – 16. Yolov, A.A., Shabarova, Z.A., 1990. Constructing DNA by polymerase recombination. Nucl. Acids. Res. 18, 3483 – 3486.

. .