| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Regular Articles |
Faculty of Integrated Arts and Sciences, Hiroshima University, 739-8521, Japan
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Mechanisms of gene amplification in cancer cells have been extensively investigated and many different models have been proposed (reviewed in Ref. 12 ). Among these, BFB cycles initiated by chromosome breakage is one of the widely accepted models (reviewed in Ref. 13 ). This model purports that breakage followed by replication and end-to-end fusion of a sister chromatid generates a mitotically unstable dicentric chromosome, which leads to a second breakage in close proximity to the first one. Multiple cycles of BFB result in the multiplication of genes near the breakage. Recent work demonstrated that the break induced at the IgH locus by the recombination activating gene product, followed by recombination with the c-myc gene, initiates the BFB cycle leading to the formation of the IgH/c-myc coamplified structure (14 , 15) . However, the general mechanisms of BFB cycle initiation remain to be elucidated. The episomal model is another important model for gene amplification, which centers on extrachromosomal submicroscopic circular molecules (episomes; reviewed in Ref. 16 ). This model suggests that the episomes of tens to hundreds of kbp are excised from the chromosomes and that the mutual recombination of the episomes may lead to the formation of cytogenetically detectable DMs. The DMs, in turn, be inserted to a chromosome, leading to the formation of an HSR. Despite a plethora of studies, to date, a complete understanding of gene amplification at the molecular level remains elusive, possibly because there is no experimental system that artificially reproduces this process.
We previously reported that plasmids with a mammalian replication IR and a nuclear MAR frequently initiate an event similar to gene amplification in human cancer cells (17) . Namely, they are effectively amplified and integrated into preexisting DMs or alternatively induce the de novo generation of DMs or the HSR. Other researchers reported that a defined sequence of a few kb can be sufficient to direct initiation in the ectopic locus by using IRs from human c-myc (18) , ß-globin (19) , and DHFR (Ref. 20 ). Therefore, we hypothesized that the autonomous replication and extrachromosomal maintenance of the introduced plasmid are required for the generation of the amplified structure, at least in the short term. Consistent with this argument, EBV vector, which is autonomously replicated in mammalian cells, was integrated into DMs, although the plasmid never generated HSR (17 , 21) . On the contrary, a plasmid containing the Drosophila P-element transposon sequence was amplified to an extraordinarily high copy number in a mosquito cell line and generated both DMs and HSR (22) . Although it was unclear whether the plasmid replicated effectively in mosquito cells, it was recovered, in its original size, in Escherichia coli. Another article reported an amplification-promoting sequence in CHO cells, which, if ligated to the plasmid in cis-, promotes the amplification of the plasmid sequence under selective pressure (23) . Those authors found that the amplification-promoting sequence was homologous to the replication IR in the DHFR locus. Furthermore, transfection of a plasmid containing the DHFR structural gene in CHO cells, followed by stepwise selection with increasing concentrations of a DHFR inhibitor (methotrexate), resulted in amplification of the original plasmid (24) . Although this amplification was shown to be mediated by the BFB cycle, the initiation mechanism and influence of the cis-structure of the plasmid remain to be clarified.
Thus, the aim of the present study was to provide a mechanistic basis for our previous findings (17) , i.e., to elucidate how and why an amplified structure is generated by plasmids containing a mammalian IR. We show that the plasmid initially forms a multimer in which the plasmid sequences are organized into head-to-tail direct repeats. If this multimer is integrated with the chromosome arm, the BFB cycle is initiated, which leads to the generation of HSR. Importantly, the HSR is generated if the transcription from the drug resistance gene directly meets with replication from the hypothetical origin at the IR. Such conflict was controllable by targeted insertion of a poly(A) addition sequence, an orientation-specific RFB, or a MAR.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. pSFVdhfr has two drug resistance genes, Bsr and Hyg. We identified a sequence showing MAR activity inside the 4.6-kbp DHFR insert DNA (17)
. From this plasmid, we deleted the entire DHFR insert (pSFV-V) or inverted its orientation by NotI digestion (pINV.NotI). We made several plasmid constructs where a HSV poly(A) addition signal, a MAR sequence (AR1), or a RFB was inserted downstream of Bsr. To insert these sequences, p
BN and pB were constructed, whereas almost all of the Hyg transcription unit or the 202-bp portion of HSV poly(A) was deleted, respectively. pSV.SB2 was a generous gift from Dr. Friedrich Grummt (University of Wurzburg, Wurzburg, Germany). The plasmid contained a 110-bp sequence of RFB from the human rDNA repeat (27)
. Using the plasmid DNA as a template, the RFB sequence was amplified by PCR. The sequences of primers used were as follows: RFBnot1L (31 mer), 5'-GGGCGGCCGCGCTGGAGGTCGACCAGATGTC-3'; RFBnot1R (36 mer), 5'-GGGCGGCCGCAATTTAAAAAAAAAAAAAAAAAAAAA-3'. The PCR product was treated with NotI that cut the primer sequences and inserted into the NotI site of the pB plasmid. The orientation of the RFB sequence inside the plasmid was determined by PCR using either of the above primers and the primer with a sequence that flanks the NotI site in the pB plasmid. The source of pAR1 has been described previously (17)
. It has a 377-bp sequence from Ig
intron that shows strong MAR activity.
|
FISH.
Metaphase spreading, probe preparation and FISH were performed according to a previously published protocol (4)
. FISH analysis of PFA-fixed in situ growing cells has also been described previously (11)
. Chromatin fiber was prepared according to the published protocol (28)
with slight modifications. Briefly, cells were washed twice with PBS and suspended at a density of 1–2 x 106 cells/ml. Then, 5 µl of the suspension were spotted on the slide, and it was dried completely. The slide was immersed in SDS lysis buffer [0.5% SDS, 50 mM EDTA, and 200 mM Tris-HCl (pH 7.4)] prewarmed at 37°C in a coplin jar and incubated for 3 min. The slide was pulled up slowly at a constant speed, tilted at 45 degrees, and air-dried completely. It was fixed in methanol/acetic acid (3/1) for 5 min at room temperature and again air-dried. For the experiment reported in Fig. 1, F and G
, DIG-labeled DHFR IR DNA (4.6 kbp) excised from pSFVdhfr and biotin-labeled vector DNA (pSFV-V; 4.6 kbp) was mixed and simultaneously hybridized to the slide. For the experiment reported in Fig. 3
, the c-myc IR (2.4 kbp) and ß-globin IR (7.8 kbp) were excised from pNeo.Myc-2.4 and p6XN ß, respectively, in addition to DHFR IR DNA. To avoid contamination of the vector sequence, two consecutive rounds of electrophoresis followed by band excision were performed. This IR DNA and the entire 
phage DNA were differentially labeled by biotin or DIG. Before hybridization, a mixture of the labeled probe was prehybridized with a large excess of unlabeled vector DNA (i.e., restriction digest of pSFV-V) to suppress cross-hybridization caused by the contaminated vector sequence. Detection of the hybridized probe was performed using essentially the same protocol as reported previously, using Texas red-conjugated streptavidin (Vector Laboratories, Inc., Burlingame, CA) or FITC-conjugated anti-DIG antibody (Roche Diagnostics, Basel, Switzerland). The absence of cross-hybridization, which might be caused by the vector sequence, was confirmed; for example, hybridization of c-myc and ß-globin IR probe did not produce any signal in cells transformed with pSFVdhfr.
|
|
were as follows: L3 (19 mer), 5'-CAGTACATGCGGTCCATGC-3'; and R3 (20 mer), 5'-GGCCTCGAACTCACAGAGAT-3'. The L1, L2, R1, and R2 sequences are available from our laboratory upon request. PCR was performed according to a standard protocol using the GeneAmp PCR System 9700 (PE Applied Biosystems, Foster City, CA). The PCR product was subjected to base-sequencing analysis using either the L3 or R3 primer (Espec Oligo Service Corp., Tsukuba, Japan).
|
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. In contrast, in metaphase derived from cells transfected with the vector plasmid (p6XN), signals at DMs were barely detectable (3%; 1 of 35 metaphases examined), and no signals appeared at the HSR (0%; 0 of 40 metaphases examined). By simultaneous hybridization of the plasmid probe and the c-myc cosmid probe that detects an original amplicon in COLO 320DM, we previously showed that amplified plasmids were either integrated into preexisting DMs, generated independent DMs, or generated HSR independent of the original amplicon (17)
. These distinct structures may appear simultaneously in a single cell. In addition to these structures, we found extraordinarily large ring chromosomes (Fig. 1E)
. Although the frequency of this structure was low, its significance is discussed in a later section. The HSR or DMs containing the transfected plasmid sequences was stable once they were generated, and they persisted in the cells after >1 month without drug selection.
Molecular Structure of the Amplified Region.
We obtained 16 independent clones of COLO 320DM cells stably transformed with pSFVdhfr. Analysis by FISH led to the identification of clones in which the plasmid was localized to either DMs (10 clones), HSR (3 clones), or both (3 clones). Among those, we subsequently used clone numbers 12 (Fig. 1C)
and 14 as clones showing DM integration (DM clones) and clone 22 (Fig. 1D)
and 24 as showing HSR formation (HSR clones).
Genomic high molecular weight DNA isolated from such clones was treated with a restriction enzyme, and the products analyzed by Southern blot hybridization (Fig. 2, B and C)
. Initially, we observed that the copy number of plasmid sequences was surprisingly high. On the basis of signal intensity, we estimated the copy number as 2000–4000 copies/cell for DM clones or in excess of 10,000 copies/cell for HSR clones. Secondly, we observed that bands obtained from genomic digests were surprisingly sharp, especially for DM clones (clones 12 and 14). With one exception, each fragment corresponded well with a fragment generated from circular pSFVdhfr that had been subjected to a similar digestion procedure. Similar results were obtained using 6 independent clones of COLO 320DM cells transformed with pNeo.Myc-2.4 (data not shown). FISH analysis of PFA-fixed cells always indicated that the majority of the plasmid sequences were at the amplified region (data not shown). This implied that the plasmid sequences must be amplified in a highly ordered array, especially in DMs. Moreover, the single digestion experiment indicated this to be a head-to-tail tandem array. Although the molecular mechanisms remain to be elucidated, it has been demonstrated previously that transgenes are usually arranged as tandem arrays within a chromosomal integration site (29)
. The results obtained here suggest that amplification of plasmids containing IR may also be mediated by a similar mechanism, presumably in the initial step of transformation. However, it was remarkable that the head-to-tail direct repeat remained intact, even after the amplification of a few thousand plasmid sequences. This suggested that integration into DMs enabled a large increase in copy number without destabilizing the sequence.
Southern blot analysis of clone 12 revealed an unusual fragment that was thought to reflect a recombination product (Fig. 2B
, asterisk). Therefore, we used Southern hybridization, using several additional restriction enzymes, to identify the recombination point responsible for the generation of this fragment (data not shown). This analysis revealed a recombination between the region showing MAR activity in DHFR and the Hyg gene in a direct orientation. This suggested that breakage occurs at these two sites in a plasmid direct repeat and that this is followed by the rejoining of each end. Using a series of PCR primers that surrounded the recombination point (Fig. 2A)
and genomic DNA from clone 12, we performed PCR and subjected the shortest product (L3-R3) to sequencing. The sequence we obtained is shown in Fig. 2D
, along with the original pSFVdhfr sequence. There was no base insertion at the junction, as well as no obvious homology between recombined sequences, which could be interpreted as a product of nonhomologous end joining. A clear sequencing pattern was obtained from the PCR product, although the genomic DNA from cells bearing multiple DMs was used as a template. This suggested that the recombined sequences in DMs were stable during cell propagation.
To investigate the macroscopic structure of the amplified region, we performed FISH using stretched chromatin fiber (28)
. Representative images are shown in Fig. 1, F and G
. As is usually the case with this technique, the resultant signal resembled a string of beads. The green and red signals, which indicate the vector and DHFR insert, respectively, appeared almost alternately in every fiber where the specific hybridization signal appeared. This indicated that the amplified plasmid sequences span without interruption. We estimated the repeat length in clone 12 to be a few tens to 100 copies of the original plasmid/repeat based on the appearance of alternate green and red signals. Because clone 12 cells contained 
20–30 DMs, the integration of one to five copies of this repeat into DMs might explain the total copy number/cell (2000–4000), which was obtained from Southern blot analysis. As expected, the amplified plasmid sequences spanned a much longer length of the fiber prepared from HSR clone 22 (Fig. 1G)
.
Taken together, these results suggested that a large circular molecule composed of tens to hundreds copies of directly repeating plasmid was formed at the initial stage of transformation. We frequently observed the paired tiny signal among the metaphase spread hybridized with the plasmid probe, suggesting it might correspond to a submicroscopic circular plasmid multimer (data not shown). The multimer may recombine with preexisting DMs, recombine with each other to grow up to a size similar to DMs or integrate into the chromosomes.
Analysis of the Initial Multimerization Process by Cotransfection.
We cotransfected two plasmids having different IR and/or -phage DNA and examined how their amplified structures appeared. Previously, it was demonstrated that similar cotransfected sequences underwent a series of nonhomologous intermolecular recombination to form an intermixed structure, early after the transfection (30)
, however, their structures had not been examined by FISH. We prepared the metaphase spread from stable transformants at >4 weeks after the transfection and examined it by two-color FISH that independently detects both of the transfected sequences. Cotransfected sequences were always detected as a mixture in the amplified structure (Fig. 3, A and B)
. Furthermore, even -phage DNA was coamplified if it was cotransfected with a plasmid containing IR.
The coamplified structure was analyzed by chromatin fiber FISH. Representative images are shown in Fig. 3, C–H
. The sequence in the amplified region was always linked to the plasmid sequence, however, it usually formed a long uninterrupted stretch along the fibers, with a size corresponding to between one and >10 molecules of DNA (Fig. 3C)
. This most likely reflects the ligation between the COS site leading to the formation of a concatemeric molecule. Cotransfection of two types of plasmid with different IR also resulted in an intermixed structure, demonstrating that nonhomologous recombination occurred frequently, as has been suggested previously (30)
. However, we frequently detected structures in which either plasmid sequence continued along the fiber, the length of which corresponded to a few tens copies of plasmids (Fig. 3, C–E)
. This phenomenon might be explained by intermolecular homologous recombination. However, pSFVdhfr (11 kbp) and p6XNß (14 kbp) share a 6.3-kbp common sequence, and 2 kbp are shared between the above two plasmids and pNeo.myc-2.4 (9 kbp). Therefore, long uninterrupted continuation of either plasmid sequence is unlikely to reflect homologous recombination. Furthermore, this activity is known to be low in mammalian cells, unlike lower eukaryotes. On the contrary, analysis of some fibers indicated that the plasmid and phage sequences were arranged as a direct tandem repeat (Fig. 3, F and G)
. This suggested that DNA was ligated to mammalian IR by recombination with an IR-bearing plasmid at the initial stage of transfection. Intermolecular recombination could not satisfactorily account for such tandem repeats given that a large number of molecules with different structures were transfected. This might suggest that autonomous replication of the recombined sequence occurs along with multimerization by the intramolecular process associated with its replication. It is also possible that the continuation of either plasmid sequence along the fiber, described above, might result from multimerization by a replication-mediated process. Previously, dimerization of circular genomic DNA was observed in bacterial cells containing a replication termination mutation (31
, 32)
.
HSR Is Generated by the BFB Cycle Initiated at the Plasmid Repeat.
In examining cells bearing plasmid-derived HSR by FISH, we frequently observed bridges composed of plasmid sequences between two mitotic or postmitotic daughter cells. Such bridges were observed among the conventional chromosomal spread (Fig. 4D)
. Because COLO 320DM cells did not adhere to the substratum, we used human HeLa cells to address this issue. pSFVdhfr also generated HSR in these cells. These transformants were grown on chamber slides, fixed in situ by PFA, and analyzed by FISH with the plasmid probe. The bridge composed of plasmid sequences was detected at anaphase (Fig. 4A)
and telophase/cytokinesis (Fig. 4B)
but in early G1 phase was disassembled (Fig. 4C)
. In some G1-phase cells, a portion of the broken chromatin remained in the cytoplasm, thereby generating a micronucleus. On the contrary, among the metaphase spread, most HSR plasmid sequences were detected at the end of the chromosome arm (for examples, see Fig. 1
). On the basis of these two observations, we concluded that the BFB cycle is involved in the generation and expansion of plasmid-derived HSR. For the BFB cycle to initiate at the plasmid sequence, frequent strand breakage at the plasmid repeat should be necessary.
|
. pSFVdhfr has two drug selection marker genes, i.e., Bsr and Hyg. These plasmids were transfected into COLO 320DM cells, and the cells were selected with blasticidine. The plasmid locations in the pooled transformants were analyzed by the FISH as described in Fig. 1
. At least three independent transfections and analyses of these plasmids were performed. The fraction of metaphases exhibiting signals at DMs tended to fluctuate between different transfections, but the fraction exhibiting signals at HSR was very reproducible. A representative example is shown in Fig. 6
. In this experiment, we did the analysis on 2, 3, and 4 weeks after the transfection.
|
, however, pSFV-V did not produce any HSR (Fig. 6B)
. This indicated that DHFR IR is necessary for the generation of HSR and suggested that the replication from IR might induce strand breakage. On the contrary the Bsr transcription unit did not have a downstream poly(A) addition sequence. Because poly(A) addition is coupled to transcription termination (for a recent review, see Ref. 33
), a head-on collision between transcription and replication may be anticipated if the replication is initiated from the DHFR IR. It is important to note that the amplified plasmid sequences are arranged as a direct repeat in which mutual orientation between each sequence is the same as the one in the original circular construct. Therefore, we placed poly(A) sequences downstream of Bsr (pBN.polyA and pB.polyA). Transfection of these plasmids barely produced HSR, whereas the parent plasmids (pBN and pB) did produce many HSR (compare Fig. 6, F and I, to, D and G
, respectively). This indicated that if placed at the point where transcription and replication are predicted to meet, the poly(A) signal almost completely suppressed the generation of HSR. pB produced a relatively lower frequency of HSR than pSFVdhfr or pBN. The 202-bp sequence at the end of HSV poly(A), which was deleted during the construction of pB, might have some effect on transcription from Hyg, although Hyg was not used for selection. Furthermore, a striking effect was obtained using RFB derived from a human rDNA repeat. RFB arrests DNA replication fork progression in an orientation-dependent manner, and its action makes rDNA transcription possible, even during its replication (27)
. In our experiments, RFB inhibited the generation of HSR in an orientation-dependent manner (Fig. 6, J and K)
. Specifically, if it was placed in an orientation that blocks the replication from the DHFR IR (pB.RFB Dir.), cells having HSR did not appear, whereas they appeared frequently if the RFB was placed in a reverse orientation (pB.RFB Rev.). In the latter case, the replication directly meets transcription from Bsr. Taken together, these results strongly suggested that if the replication from the DHFR IR met transcription from Bsr, DNA strand breakage was induced. This conflict was resolved by inserting the poly(A) addition signal or the RFB in an orientation that blocked replication from the IR.
We next addressed the effect of MAR sequences. pBN.AR1 and pB.AR1 have a sequence showing strong MAR activity (AR1) downstream of Bsr. By transfecting these plasmids, we found that placing MAR at this position increased the generation of HSR (compare Fig. 6, E and H–D and G
). From this result, we hypothesized that the presence of MAR between replication and transcription may lead to strand breakage. On the contrary, transfection of pINV.NotI resulted in a very low frequency of HSR formation. Because pINV.NotI differed from the original pSFVdhfr with respect to the orientation of the DHFR insert, the order of Oriß and the intrinsic MAR was inverted with respect to Bsr transcription. Although the intrinsic MAR was relatively removed from Bsr in the original construct, it may contribute to strand breakage. Inversion of the DHFR insert resulted in no MAR being present between replication and transcription, which might explain how very little HSR was generated by pINV.NotI. Furthermore, the results reported in our previous study (17)
support this hypothesis. Namely, deletion of the SV40 3'-processing signal, which functions as a MAR in addition to the poly(A) addition signal from downstream neomycin-resistance gene, led to the complete suppression of HSR formation. In that construct, transcription from the neomycin-resistance gene directly met replication from the c-myc IR in the absence of MAR. Insertion of another MAR sequence (AR1) in place of the SV40 sequence led to a high level HSR-formation. Thus, all of the data available are consistent with the hypothesis that the presence of MAR between replication and transcription leads to strand breakage.
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, based on our experimental results shown here, and we will discuss each of the steps below.
|
. Similar direct repeats are now known to be targets of repeat-induced gene silencing (34)
. Actually, our preliminary experiment, using reverse transcriptase-mediated PCR, indicated that low expression arises from amplified plasmid sequences (data not shown). Furthermore, expression from only one copy sequence was sufficient for growth under 5 µg/ml blasticidine, which was used to select the transformant. Therefore, the amplification of a plasmid containing IR to a very large copy number inside the cells was not attributable to the selective pressure applied but rather to the intrinsic nature of the plasmid. On the contrary to our plasmid model, the amplified oncogenes or drug-resistance genes are usually overexpressed in patient’s cancer cells. This may most plausibly be explained that the amplicon size in natural cancer cells is usually far larger than our plasmid (i.e., >100 kbp; Refs. 16
, 35
). Furthermore, the amplicon of different genomic loci might be frequently intermixed (2)
. The larger amplicon size and/or its complexity may increase the chance to incorporate the sequences, for example, the insulator, which may release the amplified sequence from the repeat-induced gene silencing.
Mutual recombination of the plasmid repeat circles may result in a structure of a size indistinguishable from DMs. Alternatively, recombination of these circles with preexisting DMs may lead to the sequence being integrated at DMs (Fig. 7A)
. Intermolecular recombination between extrachromosomal molecules appears to be very common. This might be related to their colocalization inside the nucleus. Extrachromosomal genetic elements, including DMs or viral plasmids, are segregated by chromosomal tethering, and these are localized to the nuclear periphery after the completion of mitosis (11
, 21)
. Recombination between extrachromosomal elements has been reported repeatedly in human cancer. Specifically, the natural amplified structure found in cancer cells in vivo is known to be complex and discontinuous, involving several coamplified genes (for a review, see Ref. 2
). Furthermore, during the course of MDR gene amplification induced by colchicin, an 890-kbp submicroscopic episome generates DMs by dimerization (35)
. Mutual recombination between DMs has also been demonstrated, and this might reflect the lower methylation level of DMs (36)
. We hypothesized that the autonomous replication and stable maintenance of the circular plasmid repeat were required for integration into DMs, at least for a limited period of time. Actually, the plasmid should have both mammalian IR and MAR to be integrated into DMs (17)
. Furthermore, the transfection of an EBV-based plasmid, which was known to exhibit extrachromosomal autonomous replication and segregation, led to the appearance of the plasmid signal at DMs, although none appeared at HSR (17
, 21)
. Although the frequency was very low, pSFV-V, which had neither IR nor MAR, might be integrated into DMs (Fig. 6B)
. This may reflect the well-known argument that any sequence has the potential to be a replication initiation site in mammalian cells (37)
.
Our results showed that the frequency of signals at DMs decreased during culture, whereas the frequency of signals at HSR increased (Fig. 6)
. This bears some similarity to the observation that after their isolation from human patients, cancer cells bearing DMs are generally substituted by cells bearing HSR during in vitro passage (38)
. It is possible that DMs with the plasmid repeat recombine with the chromosome, thereby initiating the BFB cycle and generating HSR. Alternatively, the plasmid repeat at DMs could also induce strand breakage by the above mechanism. As previously suggested by us, breakage at DMs might lead to the detachment of DMs from the mitotic chromosome in the next M phase, followed by their elimination from cells (11)
. We suggested that sequences on DMs are very stable, which was evidenced by the fact that the direct repeat was surprisingly ordered in DMs, despite the very high copy number, and that the recombined sequences were stably maintained (Fig. 2)
. This may be because DMs would be eliminated if they had a breakage.
If the large circle of a plasmid direct repeat was recombined with the chromosome arm, it induced BFB cycles that led to the generation of HSR (Fig. 7A)
. The BFB cycle should be initiated by breakage at the plasmid sequence. Actually, a large ring chromosome composed of plasmid sequences was observed in HSR-bearing cells (Fig. 1E)
, which may be explained by the simultaneous breakage at two sites inside the HSR, followed by the rejoining of each end. Furthermore, we identified a recombined sequence inside the amplified plasmid array at DMs (Fig. 2D)
. This was a product of nonhomologous end joining between two sequences in a direct orientation, i.e., two cuts inside the direct repeat followed by the rejoining of each end. It is important to note that recombination, rather than breakage, between the sister chromatid may also generate a dicentric chromosome that initiates the BFB cycle. For simplicity, we have used the term breakage, but whether this is actually the case will require clarification in a future study.
Our results suggested that conflict between transcription and replication resulted in breakage inside the plasmid repeat, which triggered BFB cycles. This conflict was avoided by placing a poly(A) addition sequence or a RFB in an orientation that blocks the replication from IR (Fig. 7B)
. Many studies using bacterial or lower eukaryotic cells have described recombination or strand breakage induced by collision between transcription and replication (for a review, see Ref. 39
). For mammals, an article reported that transcription inhibited the autonomous replication of plasmids in human cells (40)
. Many studies have demonstrated that the arrest of replication fork progression by inhibitors induced strand breakage (for example, Ref. 41
). However, it has never been previously suggested that collision actually resulted in breakage and genomic instability in mammalian cells.
Our results suggested that the MAR, if placed between transcription and replication, strongly induced strand breakage as evidenced by HSR formation (Fig. 7B)
. This argument was consistent with the finding that the recombination identified in clone 12 involved breakage at the MAR sequence (Fig. 2)
. Furthermore, in CHO cells, MAR was positioned at the interamplicon junction in the amplified DHFR domain (42)
. In that case, amplification of the DHFR domain in the parental CHO cells was induced by treating cells with increasing concentrations of methotrexate. The 273-kbp amplicon, which contained the DHFR transcription unit and replication IR, was organized in a head-to-tail array. Therefore, it seems to be consistent with our plasmid model.
Analysis of mammalian origins of replication has been hampered by the inability to construct a plasmid that exhibits autonomous replication. Indeed, the presence of a defined origin of replication itself still remains to be debated. The results provided here, using DHFR Oriß, suggest that the replication fork actually arises from the IR. Furthermore, these findings highlight the importance of the arrangement of the transcription unit relative to the origin, as well as the appropriate placement of poly(A) addition, RFB and MAR sequences, for plasmid sequence stability.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported in part by a grant-in-aid for Scientific Research (B) Grant 14340238 and a grant-in-aid for Exploratory Research Grant 14658232, both from the Japan Society for the Promotion of Science (to N. S.). 
2 To whom requests for reprints should be addressed, at Faculty of Integrated Arts and Sciences, Hiroshima University, 739-8521, Japan. Phone: 81-824-24-6528; Fax: 81-824-24-0759; E-mail: shimizu{at}hiroshima-u.ac.jp
3 The abbreviations used are: DM, double minute; HSR, homogeneously staining region; BFB, breakage-fusion-bridge; IR, initiation region; MAR, matrix attachment region; DHFR, dihydrofolate reductase; CHO, Chinese hamster ovary; FISH, fluorescence in situ hybridization; RFB, replication fork barrier; Bsr, blasticidin resistance gene; Hyg, hygromycin resistance gene; PFA, paraformaldehyde; DIG, digoxigenin.
Received 3/15/03. Revised 6/ 3/03. Accepted 6/10/03.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Diefenbacher, S. Sekula, C. Heilbock, J. V. Maier, M. Litfin, H. van Dam, M. Castellazzi, P. Herrlich, and O. Kassel Restriction to Fos Family Members of Trip6-Dependent Coactivation and Glucocorticoid Receptor-Dependent Trans-Repression of Activator Protein-1 Mol. Endocrinol., August 1, 2008; 22(8): 1767 - 1780. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Zeppernick, R. Ahmadi, B. Campos, C. Dictus, B. M. Helmke, N. Becker, P. Lichter, A. Unterberg, B. Radlwimmer, and C. C. Herold-Mende Stem Cell Marker CD133 Affects Clinical Outcome in Glioma Patients Clin. Cancer Res., January 1, 2008; 14(1): 123 - 129. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K.-i. Utani, J.-k. Kawamoto, and N. Shimizu Micronuclei Bearing Acentric Extrachromosomal Chromatin Are Transcriptionally Competent and May Perturb the Cancer Cell Phenotype Mol. Cancer Res., July 1, 2007; 5(7): 695 - 704. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. S. Yakisich and G. M. Kapler Deletion of the Tetrahymena thermophila rDNA replication fork barrier region disrupts macronuclear rDNA excision and creates a fragile site in the micronuclear genome Nucleic Acids Res., January 30, 2006; 34(2): 620 - 634. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Shimizu, F. Kamezaki, and S. Shigematsu Tracking of microinjected DNA in live cells reveals the intracellular behavior and elimination of extrachromosomal genetic material Nucleic Acids Res., November 3, 2005; 33(19): 6296 - 6307. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Shimizu and K. Shingaki Macroscopic folding and replication of the homogeneously staining region in late S phase leads to the appearance of replication bands in mitotic chromosomes J. Cell Sci., October 15, 2004; 117(22): 5303 - 5312. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||
