INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Recombinations in retroviral genes play important roles in retroviral carcinogenesis and in the AIDS epidemic. Two plausible models for retrovirus recombination were first proposed 16 years ago. A third model has recently been proposed. However, none of these models have been critically tested. The first model (Fig. 1A to C) proposed that recombination occurred during minus-strand synthesis using RNA as the template (7). This model proposed that retroviral genomes contain considerable numbers of breaks within the RNA molecule and that when reverse transcriptase encounters such breaks, it switches to the homologous sequence on the other RNA molecule and continues synthesis. This model is called the forced-copy-choice model. The second model (Fig. 1D to F) proposed that recombination occurs during plus-strand DNA synthesis (15, 16). This model proposed that two minus-strand DNAs are made in one virion using both RNA templates (retrovirus packages two RNA molecules in each virion). Since plus-strand DNA synthesis is initially discontinuous, a fragment of product DNA might be displaced by continuous DNA synthesis. The displaced DNA fragment may then hybridize to the minus-strand DNA synthesized from the other molecule of viral RNA used as the template. This model is called the plus-strand replacement model. The third model differs from the original copy choice models because it does not require preexisting breaks in the RNA (6) (Fig. 1G to I). Since reverse transcriptase has a relatively low processivity and because the RNA template is degraded approximately 18 nucleotides from the point of DNA synthesis, it is proposed that an extended single-stranded DNA molecule is left tailing behind. This DNA molecule is free to form a hybrid duplex with another RNA molecule. If the reverse transcriptase molecule detaches from the template, it is likely that the short (18-bp) hybrid with the original template will be displaced by branch migration of the second RNA and that this RNA molecule will then become the template for continued synthesis. This model is called the minus-strand replacement model.

View larger version (10K): [in this window] [in a new window]   FIG. 1.   Models of retroviral recominations. The light lines represent RNA molecules, and the heavy lines represent DNA molecules. (A to C) Model of forced copy choice for retroviral recombination. This model proposes that retroviral genomes contain breaks within the RNA molecule (A) and that when reverse transcriptase encounters such breaks, it switches to the homologous sequence on the other genome (B) and continues minus-strand DNA synthesis (C). (D to F) Model of strand displacement-assimilation for retroviral recombination. This model proposes that two minus-strand DNAs are made within one virion (D). Since plus-strand DNA synthesis is initially discontinuous, a fragment of product DNA might be displaced by continuous DNA synthesis (E). The displaced DNA fragment may then hybridize to the other minus-strand DNA (F). (G to I) Model of minus-strand exchange for retroviral recombination. An extended single-stranded DNA molecule is left tailing behind during synthesis of minus-strand DNA (G). This DNA molecule forms a hybrid duplex with another RNA molecule (H). The reverse transcriptase molecule detaches from the template and is displaced by the second RNA (I).

Data have demonstrated that poly(A) sequences exist in some recombinants of one naturally occurring acute oncogenic virus (14) and in three other experimental systems, including the one we have described previously (22, 25, 28). The poly(A) tracts exist only on the RNA molecules and not on the minus-strand DNA molecules during reverse transcription, since the minus-strand strong stop DNA jumps to the R region [upstream of the poly(A) tract] to continue synthesis of minus-strand DNA. Therefore, some, if not all, recombination events should use an RNA molecule as the template. Since minus-strand DNA synthesis uses RNA as a template, these results suggest that some retroviral recombinations occur during minus-strand DNA synthesis.

The retroviral RNA molecules are positive sense in polarity, equivalent to mRNA. During reverse transcription, the first strand of DNA synthesized is minus in polarity since it is synthesized from the positive-sense RNA molecule, which is used as the template. The minus-strand DNA is complementary to the plus or sense viral genomic RNA. After the minus-strand DNA is polymerized, the plus-strand DNA is synthesized using the minus-strand DNA as the template. By this rationale, if a change results from recombination that occurs during minus-strand DNA synthesis, the plus-strand DNA would also contain this change, because during synthesis of the plus-strand DNA, the minus-strand DNA is used as the template. However, if changes occurs during plus-strand DNA synthesis, the double-stranded DNA would contain a mismatch. A mismatch within retroviral double-stranded DNA can also be introduced by a preexisting mutation within the primer binding site (PBS) of the positive-sense viral RNA genome (3, 8). The PBS is the binding site for a host tRNA within the viral RNA molecule. The PBS is 18 bases in length in Moloney leukemia virus (MLV). The 18-base sequence is reverse transcribed into a minus-strand DNA using the viral RNA genome as the template, while the plus-strand DNA is synthesized using the host tRNA 3' end as the template. If there is a preexisting mutation within this PBS on the viral RNA molecule, the minus-strand DNA contains the mutation while the plus strand remains the wild type (wt). Therefore, the double-stranded DNA should contain a mismatch within the PBS.

Retroviral double-stranded DNA is then integrated into the cellular chromosome DNA to form a provirus. Most cells contain a system to repair different types of mismatches. Some cell lines, such as the HCT 116 cell line, contain defects in their DNA repair systems and are incapable of repairing mismatches (10, 20). To date it is not clear if the mismatch in the provirus can be repaired by the host cellular DNA repair system. If the target cells does not repair the mismatch before an infected cell divides, one daughter cell will contain this mutation while the other daughter cell will not contain this mutation. Therefore, the offspring of the infected cells should carry two different proviruses. However, in the host cell, these two different proviruses would be integrated into the same site of the host chromosome if the integration resulted from a single event. This report demonstrates that D17 cells (target cells) are unable to repair the majority of mismatches within the retroviral double-stranded DNA. This property of D17 cells, therefore, is useful for determining at which step(s) retroviral recombinations occur.

Previous studies that were designed to determine the recombination rate of a retrovirus have used selectable genes and cell culture systems (12, 28). In this report, drug selection markers have been exchanged for an unselected color reporter gene, gfp (green fluorescent protein) (5). Cells containing the wt gfp reporter gene appear green under fluorescence microscopy, while cells with a mutant gfp appear colorless or clear (24, 27). A helper cell containing two proviruses that encoded two mutated gfp genes was constructed. We reasoned that recombination between the two mutations would produce a functional gfp gene. Because the plus-strand DNA is synthesized from the minus-strand DNA templatethe recombination occurs during minus-strand DNA synthesisthe plus-strand DNA will also carry the functional gfp sequence. After the cell divides, all the offspring cells will be green under a fluorescence microscope. However, if the recombination occurs during plus-strand DNA synthesis, only the plus-strand DNA will contain the functional gfp sequence and the minus-strand DNA will contain the mutated gfp. The double-stranded DNA, which contains a mismatch, integrates into the host chromosomal DNA to form a provirus. Because D17 cells are unable to repair this mismatch, after the cell divides into two daughter cells, one of the daughter cells will encode wt gfp and the other daughter cell will encode a mutated gfp. Growth of the offspring of these two daughter cells will produce a colony. Under the fluorescence microscope, this colony should be a mixture of green and clear cells.

However, a colony of a mixture of green and clear cells is not necessarily a result from a recombination (one event) occurring during plus-strand DNA synthesis. Another possibility for the formation of a mixed colony is an infection with two viruses of two daughter cells that have just undergone division from one cell (or two cells that were plated together on a dish) (two separate events). One virus is a parent-type virus (clear), and the other virus is a recombinant formed during minus-strand DNA synthesis (green). If a colony of cells results from two infection events, the resulting colony should be a mixture of two different cells, each of which contains a different provirus resulting from an independent integration event. However, in this scenario these two proviruses will have integrated into two different sites. If a colony of mixed cells results from a recombination occurring during plus-strand DNA synthesis, however, the provirus in the green cells should be integrated into the same site as the clear cells in their mixed colony. Using this rationale, we demonstrate that more than 98%, if not all, of retroviral recombinations occur during minus-strand DNA synthesis. In other words, the model of copy choice or minus-strand displacement is the mechanism of most, if not all, retroviral recombination events.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Nomenclature. Plasmids are designated, for example pJZ425 and pJZ468; viruses made from these plasmids are designated, for example, JZ425 and JZ468.

Vector constructions. All recombinant techniques were carried out according to conventional procedures (23). All vector sequences are available upon request.

(i) Construction of pLT2. To construct a retroviral vector that contains one base mutation in its PBS region, three fragments were ligated. The first fragment was derived from pLN (19) and contained two MLV long terminal repeats (LTRs) and a neomycin resistance gene (neo). pLN was digested with PshAI and BstEII. PshAI digested 17 bp upstream of the PBS, while BstEII digested within the package signal, 537 bp downstream of the PBS. The second fragment was a double-stranded DNA created by annealing two artificially synthesized oligonucleotides. The first oligonucleotide was GGGTCTTTCATTTGGGGGCTCGc, while the second oligonucleotide was CCGGgCGAGCCCCCAAATGAAAGACCC (the lowercase letters represent the nucleotides that are different from those of the wt PBS sequence). The 5' end of this DNA was blunt and compatible with the PshAI-digested DNA fragment. The 3' double-stranded DNA contained a sticky end, which was identical to the sticky end formed by digestion with XmaI. This double-stranded DNA was designated the PshAI-XmaI fragment. The third fragment was also derived from pLN. Sequences of pLN were amplified by PCR using two primers. The first primer (GGGGGCTCGcCCGGGATTT) was located at the PBS; however, the thymidine (T) within the PBS at position 744 was changed to cytosine (C) so that the mutated PBS would be digested with XmaI (cCCGGG) (underlined). The amplified fragment was digested with XmaI and BstEII. The BstEII-PshAI fragment isolated from pLN, the double-stranded DNA artificially synthesized with PshAI and XmaI and the XmaI- and BstEII-digested PCR-amplified fragment were ligated. The resulting plasmid was designated pLT2, which was identical to pLN except that the T at position 744 was changed to C so that this plasmid would contain an additional SmaI site at the PBS. (SmaI recognizes the same CCCGGG as XmaI does, except that the SmaI-cut fragment is blunt ended while the XmaI-cut fragment forms a sticky end.)

(ii) Construction of a retroviral vector with deletion of the downstream PBS. The EcoRI fragment which contained the MLV 5' LTR was cloned into pSP73 (Promega, Madison, Wis.). To delete the downstream PBS, the MLV vector was amplified by PCR using two primers. The first primer (TAGAACCAGATCTGATATCATC) was located on pSP73, and the second primer (GACGAGgtCgacAATGAAAGACCCCCGTCGTGGGT) was located at the U5 region of the 3' LTR. The first six nucleotides (CCCCCA) of the PBS region were changed to (gtCgac), so that the amplified fragment could be digested with SalI. The amplified fragment was digested with BglII and SalI, and this BglII-SalI fragment did not contain the downstream PBS.

(iii) Construction of a retroviral vector which contains an additional SmaI site at its PBS and deletes the downstream PBS. The pLN plasmid contains two MLV LTRs and a neo gene. Plasmid pLT5 is similar to pLN except that pLT5 contains a TC transition at the PBS and the PBS downstream of the 3' LTR is deleted (Fig. 2A). Plasmid pLT5 was constructed by ligations of five fragments from three vectors. The five fragments are as follows: the SalI-EcoRI fragment (positions 5324 to 2232), which contains the pUC18 (26) backbone; the EcoRI-AscI fragment of pLN (positions 2233 to 2816), which contains half of the 5' LTR; the AscI-BstEII fragment (positions 2817 to 3519) of pLT2, which contains the other half of the 5' LTR and a mutation within the PBS region; the BstEII-XbaI fragment (positions 3520 to 5027) from pLN, which contains the neo gene and half of the 3' LTR; and the XbaI-SalI fragment of pLN (positions 5028 to 5323), which contains the second part of the 3' LTR and the deleted downstream PBS region.

View larger version (1038K): [in this window] [in a new window]   FIG. 2.   (A) Structure of a retrovirus vector which contains a transition mutation on its PBS. The retroviral vector LT5 contains the neo gene between two MLV LTRs. Retroviral replication uses host tRNA as its primer. This 18-base PBS (GGGCTCGTCCGGGAT) is complementary to the tRNAPro 3' end on the acceptor arm. The T at the 11th nucleotide (underlined) was changed to C so that the 11th to 16th nucleotides within the PBS region could be digested with SmaI (cCCGGG) (the lowercase letter represents the nucleotide that is different from the nucleotide in the wt PBS sequence.) The PBS downstream from the 3' LTR was deleted. The 18-base sequence is reverse transcribed into a minus-strand DNA using the viral RNA genome as the template; the plus-strand DNA is synthesized using the host tRNA 3' end as the template. Because of this preexisting mutation within this PBS region on the viral RNA molecule, the minus-strand DNA contains the mutation (i.e., C) while the plus-strand sequence remains wt (i.e., T). Therefore, the double-stranded DNA should contain a mismatch (C or T) within the PBS region. (B) Electrophoresis of the PCR-amplified fragment of the PBS region of DNA isolated from D17 cells infected with LT5. The cellular DNA of each Neor colony infected with LT5 was amplified by PCR, followed by digestion with SmaI. After digestion, the amplified parental fragment was cut into two fragments, 686 and 207 bp in length. If the amplified mutant PBS fragments contained an additional SmaI site within the PBS region and this amplified fragment was digested with SmaI, there would be three fragments, which would be 558, 207, and 128 bp in length. Lane 1, SmaI-digested amplified DNA fragments from plasmid pLT5; lane 2, SmaI-digested amplified DNA fragments from plasmid pLN; lanes 3 to 26, SmaI-digested amplified DNA fragments from DNA isolated from individual Neor colonies infected with pLT5.

(iv) Introduction of restriction enzyme sites within the gfp gene. The gfp gene was isolated from the jellyfish Aequorea victoria (wt GFP; Clontech, Palo Alto, Calif.). The wt gfp open reading frame contains an NcoI site at the 5' end and a BstBI site at the 3' end (Fig. 3A). Three amino acids between the two sites have been changed so that the gfp gene emits brighter light. In addition, wt gfp has also been engineered to contain silent mutations in all of its amino acid codons to fit the human codon usage preference in order to increase its translational efficiency (Clontech). In this study, the gfp gene containing these silent mutations is designated h-gfp (Fig. 3B).

View larger version (17K): [in this window] [in a new window]   FIG. 3.   Structures of retroviral vectors used for determining in which step(s) the recombinations in retroviral replication occur. (A) wt gfp gene. The gfp gene was isolated from the jellyfish A. victoria, which contains an NcoI site at the 5' end and a BstBI site at the 3' end. (B) h-gfp gene. wt gfp has also been engineered to contain silent mutations in all of its amino acid codons to fit the human codon usage preference in order to increase the translational efficiency and is designated h-gfp. (C) Chimeric gfp-BstBI gene. The gfp-BstBI gene contains a BstBI site, which was constructed by PCR with the h-gfp sequence to allow the construction of an h-gfp-wt-gfp chimeric gene. The 5' end upstream of the BstBI site remains the h-gfp sequence. (D) Chimeric gfp-NcoI gene. The gfp-NcoI gene contains an NcoI site in wt gfp, and the construction in the h-gfp sequence upstream of the NcoI site was replaced with the wt gfp gene to produce this chimeric gene. (E) Structure of the retroviral vector pJZ425. pJZ425 contains a selectable marker (hyg) and an unselectable color reporter gene, gfp-BstBI. The gfp gene is transcriptionally expressed from the MLV 5' LTR (containing the promoter and enhancer), and the selectable gene (hyg) is expressed from an IRES. A frameshift mutation is at the 3' end of the gfp gene. (F) Structure of the retroviral vector pJZ468. pJZ468 is similar to pJZ425; however, the neo gene is replaced with the hyg gene and a frameshift mutation is at the 5' end of the gfp-NcoI gene.

(v) Introduction of a frameshift mutation into the gfp gene. To introduce a frameshift mutation into the gfp gene, the gfp-BstBI gene sequence was digested with BstBI, followed by repair with Klenow fragment. BstBI digestion creates two DNA ends that contain a 2-base overhang. When the overhang was repaired by the Klenow fragment, it created two blunt ends. Ligation of these two blunt ends with T4 ligase created a 2-bp insertion, which shifted the gfp open reading frame by two (+2). The gfp-NcoI was digested with NcoI, followed by repair with the Klenow fragment and ligation. NcoI digestion created two DNA ends that contained a 4-base overhang. As a result, this shifted the gfp open reading frame by one (+1). The first gfp gene contained a frameshift mutation at BstBI within the 3' end of the gfp gene, and the second gfp gene contained another frameshift mutation at the NcoI site within the 5' end of the gfp gene. The two mutated gfp genes contained identical sequences between their individual frameshift mutation sites, which were separated by about 0.55 kb. The gfp gene is approximately 0.7 kb in length.

(vi) Construction of pJZ425 and pJZ468. JZ425 (Fig. 3E) contained the frameshift mutation gfp-BstBI gene, the hygromycin resistance gene (hyg), and an internal ribosome entry segment (IRES) sequence located between the two genes. The gfp gene was expressed by the conventional ribosome scanning model from the viral 5' R region, while the hyg gene in JZ425 was expressed by the IRES (1, 4). In pJZ425, the 4.6-kb ClaI-BamHI fragment (from positions 4055 to 1630) was isolated from pLNCX (19) and contained the two MLV LTRs. The 0.7-kb BamHI-SalI fragment (from positions 1631 to 2389) was derived from gfp-BstBI (Fig. 3C) and contained a frameshift mutation at the BstBI site. The 0.6-kb SalI-NdeI fragment (from positions 2390 to 2991), which contained the IRES sequence, was isolated from pCITE-1 (Novagen, Madison, Wis.). The 1.1-kb NdeI-ClaI fragment (from positions 2992 to 4054) was isolated from pJD214 Hyg (9) and contained the hyg sequence.

Preparation of virus LT5. DNA of pLT5 was used to transfect the xenotropic helper cell line PG13 (18) using Lipofectamine Plus (Life Technologies, Grand Island, N.Y.) according to the manufacturer's protocol. Three hours after transfection, cells were washed four times with Dulbecco modified Eagle medium to remove the residual DNA from the plates. Viruses were detected as early as 6 h by infection of D17 cells and selection for neomycin resistance (Neo^r). The viruses were collected 10 to 12 h after transfection.

Isolation of Neo^r cellular DNA and amplification of the PBS. D17 cells infected with LT5 were selected for Neo^r with visible colonies formed about 12 days after infection. Well-separated colonies were cloned into 24-well plates. The cells were harvested once they covered the bottom of the well. Cellular chromosomal DNAs were isolated with a Wizard genomic DNA purification kit (Promega) according to the manufacturer's protocol.

Cells, transfection, and infection. The processing of D17 cells (ATCC CRL-8468), PA317 helper cells (ATCC CRL-9078), and PG13 helper cells (ATCC CRL-10686); DNA transfections; virus harvesting; and virus infections were performed as previously described (28).

Fluorescence microscopy. An inverted fluorescence microscope (Zeiss Axiovert 25) with a mercury arc lamp (100 W) and a fluorescence filter set (CZ909) consisting of a 470- by 40-nm exciter, a 515-nm emitter, and a 500-nm beam splitter was used to detect green fluorescent protein in living cells.

Relations between hyg and neo RNA frequencies in the step 3 virion. The relations between hyg and neo RNA frequencies in the step 3 virion were described previously (28). The relation between hyg and neo RNA frequencies in step 3 virions can be described in algebraic terms by means of the Hardy-Weinberg equation as follows.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of a retroviral vector that contained a mutation within the PBS. Retroviral replication uses host tRNA as its primer. A host tRNA^Pro binds to an 18-base sequence of the MLV genome. This 18-base sequence is called the PBS (8). The 18-base sequence, TGGGGGCTCGTCCGGGAT, is complementary to the tRNA^Pro 3' end on the acceptor arm. The T at the 11th nucleotide (underlined) was changed to C using a PCR technique so that the 11th to 16th nucleotides within the PBS could be digested with SmaI (CCCGGG). This MLV vector was derived from pLN (19). The pLN plasmid contains a neomycin resistance gene between two MLV LTRs. The 3' LTR of pLN was cloned from a 5' LTR that contained a PBS positioned downstream of the 3' LTR. Fifteen percent of the progeny virus utilized the downstream 3' LTR PBS to initiate minus-strand DNA synthesis, and a complete minus-strand DNA can be transcribed without a strong stop DNA jump (J. Zhang, unpublished data). To force all progeny viruses to use their natural PBS, the PBS downstream from the 3' LTR was deleted using standard PCR techniques. The vector designated LT5 (Fig. 2A), which contained a deletion of this 3' PBS, was constructed so that it contained a SmaI site within its normally positioned PBS.

Majority of mismatches within MLV double-stranded DNA was not corrected by D17 cells. DNA containing the vector LT5 was transfected into a murine retroviral helper cell line PG13 (18). The virus-containing cell supernatants were collected 10 h after transfection. The viruses were used to infect D17 cells. Since the target cells did not contain the retrovirus structural genes (gag-pol and env), virus was not released from infected target cells after infection with the retrovirus vector (28). Therefore, this infection assay represented only a single round of replication. Since pLT5 (DNA) contained a mutation in its PBS region (Fig. 2A), the virus RNA transcribed from the viral DNA also contained this mutation. After this viral vector entered D17 cells, the single-stranded RNA was reverse transcribed into double-stranded DNA. The PBS within the minus-strand DNA was copied with the viral RNA molecule as its template, while the plus-strand PBS DNA was synthesized with the host tRNA^Pro as its template (8). As a consequence, the viral double-stranded DNA contained a mismatch within this PBS. The viral double-stranded DNA was then integrated into the host chromosome as a provirus. Infected D17 cells were selected for Neo^r, and resistant colonies appeared in about 12 days. Well-separated individual colonies were cloned, and the DNAs from cells of these single colonies were isolated. The cellular DNA of each colony was amplified using two primers. The upstream primer (U32770) was located within the U3 region of the LTR, while the downstream primer was located within the packaging signal (MLV BstEII). The amplified fragment was 893 bp in length.

Construction of vectors for the determination of the mechanisms of retroviral recombination. Two additional MLV vectors were constructed. The first vector contained a drug resistance gene (hyg) and an unselected color reporter gene (gfp-BstBI) between the two MLV LTRs and is designated JZ425 (Fig. 3E). The gfp-BstBI gene was expressed downstream of the viral 5' R region, while the hyg gene in JZ425 was expressed downstream of an IRES (1, 4). This IRES allows internal ribosome binding to the transcript expressed from the 5' LTR retroviral promoter. gfp-BstBI contained a frameshift mutation near its 3' end, inside the gene's open reading frame. As a result of this frameshift, the gfp gene was no longer functional. The second vector, JZ468 (Fig. 3F), contained another selectable marker, neo, and an unselected color reporter gene, gfp-NcoI, between the two MLV LTRs. gfp-NcoI contained a frameshift mutation at the NcoI site. Cells containing either JZ425 or JZ468 did not emit green light (data not shown). These two frameshift mutations are separated by 0.55 kb, while the gfp gene is about 0.7 kb in length.

The rates of backward frameshift mutation of JZ425 and JZ468 were very low. pJZ425 was used for transfection into the MLV amphotropic helper cell line PA317 (17). PA317 cells containing pJZ425 were designated step 1A cells (Fig. 4). Viruses from step 1A cells were used to infect the xenotropic helper cell line PG13 (18). Infected cells were selected for Hyg^r. Four individual Hyg^r cell clones were isolated and designated step 2 cells. The pJZ468 vector was used for transfection into fresh PA317 cells. PA317 cells containing JZ468 were designated step 1B cells (Fig. 4). Viruses from step 1B cells were used to infect fresh PG13 cells. Infected cells were selected for Neo^r. Four individual Neo^r cell clones were isolated.

View larger version (30K): [in this window] [in a new window]   FIG. 4.   Outline of an experimental approach for the determination of the rate of recombination during a single cycle of retroviral replication. No plasmid backbone sequences are shown. Transfections are indicated by test tube shapes. Infections are indicated by virion shapes. The different backgrounds represent the indicated cell lines. The lines in the LTR separate the U3, R, and U5 regions. The provirus in step 4 cells is the construct of Neor or Hygr recombinants, which result from recombination between the two frameshift mutations located on pJZ425 and pJZ468 (step 3).

Determination of the rate of retroviral recombination. Figure 4 shows the outline of an experimental approach for the determination of the rate of retroviral recombination. The virus from step 1B cells was used to infect step 2 Hyg^r cells. The two vector proviruses were introduced into a single helper cell line by selection for Neo^r. Individual well-separated Neo^r PG13 cells were cloned. PG13 cells containing JZ425 and JZ428 were designated step 3 cells (Fig. 4). Viruses from this step 3 helper cell line can contain two nonidentical RNA molecules, one transcribed from provirus JZ425 and the other transcribed from provirus JZ468. Viruses from this helper cell line were used to infect target D17 cells. Infected cells were selected for Hyg^r and Neo^r, respectively. Unlike the helper cell lines, these target cells do not contain the retroviral structural genes (gag-pol and env), so that after infection with the retrovirus vector, no virus was released from the infected target cells (28). Thus, this infection represented only a single round of replication. Since both proviruses in PG13 cells contain a frameshift mutation within their gfp genes, if not mutation or recombination events occur, no green Neo^r (or Hyg^r) colonies should be observed. Green Neo^r (and Hyg^r) colonies can occur by two plausible mechanisms. The first possibility is that a reverse mutation occurs to overcome the frameshift mutation, but this is less likely since the reverse frameshift mutation rate is very low (<10⁵) (see above). The second possibility that may cause a green colony to form is that a recombination event occurs between the two frameshift mutation sites of these two vectors. The two frameshift mutations are located at a distance of 0.55 kb from one another. The ratio of the number of green colonies to the number of virions which contain both hyg (JZ425) RNA and neo (JZ468) RNAs is the rate of recombination during a single round of retroviral replication. To limit double infections, the multiplicity of infection was lower than 1:1,000. To determine if green colonies resulted from a single infection, 11 green colonies were randomly isolated. The DNA of each colony was digested with BamHI and hybridized with a gfp probe. BamHI digested JZ425 and JZ468 proviral DNAs 5' of the gfp gene and in cellular flanking sequences 3' of the gfp gene. All of 11 colonies analyzed contained only one copy of the gfp gene, indicating that most green cells did not result from double infections (data not shown).

Most retroviral recombinations occur during minus-strand DNA synthesis. If recombination occurred during minus-strand DNA synthesis, the minus-strand DNA would contain a functional gfp sequence, and since the plus-strand DNA uses the minus-strand DNA as a template, the plus-strand DNA would contain the functional gfp sequence. After the cell divides, all the cells in a Neo^r (or Hyg^r) colony would be green. However, if the recombination occurs during plus-strand DNA synthesis, then only the plus-strand DNA would contain a functional gfp sequence and the minus-strand DNA would still carry a frameshift mutation. The double-stranded DNA containing this mismatch subsequently integrates into the host chromosomal DNA of the D17 cells, which are unable to repair the majority of mismatches within retroviral DNA. After the cell divides, one daughter cell would carry a functional gfp sequence while the other daughter cell would contain a frameshift mutation in the gfp sequence. Under fluorescence microscopy, one-half of the cells in this colony should be green and one-half of the cells in this colony should be clear.

View larger version (68K): [in this window] [in a new window]   FIG. 5.   Microscopic analyses of D17 cells infected with viruses from step 3 cells. (A) Visible light microscopy of a Hygr colony, which contains the parental type of Hygr provirus. (B) Fluorescence microscopy of a Hygr colony, which contains the parental type of Hygr provirus. The same colony as that shown in panel A was examined under a fluorescence microscope. (C) Visible light microscopy of a Hygr colony which resulted from a recombination between JZ425 and JZ468. (D) Fluorescence microscopy of a solid green Hygr colony, which resulted from a recombination between JZ425 and JZ468. The same colony as that shown in panel C was examined under fluorescence microscopy. (E) Visible light microscopy of a Hygr mixed cell colony, which contains both green and clear cells. (F) Fluorescence microscopy of a Hygr mixed cell colony, which contains both green and clear cells. The same colony as that shown in panel C was examined under fluorescence microscopy.

View larger version (71K): [in this window] [in a new window]   FIG. 6.   Southern blot analysis of recombinants between JZ425 and JZ468. Chromosomal DNAs of Hygr (H) or Neor (N) step 4 clones were used. Cellular DNAs isolated from green (g) cells and clear (c) cells were digested with BamHI and hybridized to a gfp gene probe.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Eukaryotic cells possess several distinct mismatch repair pathways. A mismatch can be introduced in retroviral double-stranded DNA by a preexisting mutation within the PBS of the viral RNA genome. MLV-based vectors with a mutation in their PBSs were used to infect mismatch-repair-competent cell lines as well as mismatch-repair-deficient cell lines. Almost all mismatch-repair-deficient cells colonies analyzed (cell lines HCT 116 and PMS2^/) (2, 10, 20, 21) contained both the wild-type and mutated PBS. Therefore, mismatches within retroviral double-strand DNA could not be repaired by the mismatch-repair-deficient cells. Only 2 of 45 colonies analyzed showed a pattern corresponding to the presence of a repaired PBS. The two exceptions probably represent an infected cell that divided into two daughter cells and one daughter cell died so that the resulting colony contained only either the wild-type or the mutant PBS (11). Two of 24 (10%) D17 colonies analyzed showed a pattern corresponding to the presence of a repaired PBS, suggesting that D17 is a mismatch-repair-deficient cell line.

In contrast, approximately 25% of the mismatch-repair-competent cell clones analyzed (cell lines HeLa and PMS2^+/+) were repaired while 75% were not. Therefore, the normal cellular mismatch repair system is not able to repair the majority of mismatches within the viral double-stranded DNA. One possible explanation is that the cells do not have enough time to repair the mismatch after integration of the double-strand DNA before the cell divides. After a retrovirus enters host cells, the single-strand RNA genome is reverse transcribed into double-strand DNA within the retroviral capsid in the host cell cytoplasm. Before retroviral integration, the retroviral double-stranded DNA exists as a structure called the preintegration complex (PIC) (8). This PIC contains retroviral capsid protein and other proteins. The cellular mismatch repair system is probably unable to recognize a DNA mismatch within this PIC. MLV, like other simple retroviruses, can infect only dividing cells, and it has been hypothesized that the PIC enters the nucleus only during mitosis (8, 11). Therefore, the window for the cellular mismatch repair system to repair a mismatch within the retroviral double-stranded DNA would be constrained to mitosis. Only a very low percentage of mismatches within retroviral double-stranded DNA were repaired in these experiments.

Previous reports suggested that a recombinant may be a result of one or more crossover events (13). In our system, one crossover is sufficient to form a Neo^r recombinant. The reverse transcription growing points synthesize the neo gene and the 3' end of the gfp gene using JZ468 as a template and then shift to the JZ425 RNA to complete the functional gfp gene (Fig. 3E and F). However, the formation of a Hyg^r recombinant requires at least two crossover events. The reverse transcription growing points at first synthesize the hyg gene using JZ425 as the template and then shift the template to copy the 3' end of the gfp gene from JZ468. The template is shifted once again to copy the 5' end of the gfp gene of JZ425. Results in this report indicate that the Hyg^r recombinants formed with approximately equal frequencies as Neo^r recombinants, which indicates that the two crossover events occurred at the same rate as one crossover event during the minus-strand DNA synthesis. Furthermore, these results suggest that after one recombination occurs, the frequency of a second recombination rate must be much higher than the average recombination rate.

In this report, we have estimated that most, if not all, recombinations occur during minus-strand DNA synthesis. Therefore, the mechanism of retroviral recombination is either the copy choice model or the minus-strand replacement model. Further work needs to be done to distinguish between these two minus-strand DNA synthesis models.