Chromosome engineering strategy
The chromosome engineering strategy that we employed uses theCre/loxP site-specific recombinase system to generate ICTs (fora recent review, see 10). Because the Cre/loxP system is relativelyinefficient at generating ICTs (<1x10–3), we used reconstitutionof a selectable marker to isolate the cells that had undergoneCre-mediated recombination. A schematic representation of thisstrategy is shown in Figure 1 (9). Briefly, this strategy involvesreconstitution of a cloned murine Aprt gene in a human cellline, HTD114, which is deficient for APRT (11). This strategyinvolves the generation of a collection of parental cell clones(P-clones), each containing two independently inserted plasmidcassettes. One cassette contains the 5' portion of the Aprtgene linked to a neomycin resistance gene (5'-AP-Neo) and theother cassette contains the 3' portion of the Aprt gene linkedto a hygromycin resistance gene (Hyg-3'RT). Each cassette containsloxP sites in the second intron of the Aprt gene. Thus, followingCre-mediated homologous recombination the Aprt gene is reconstitutedat the loxP sites and a reciprocal translocation is generated.Thus, following selection in media containing azaserine andadenine (AA), Aprt+ cells can be isolated. In addition, workfrom Dr M. Jasin's laboratory has indicated that chromosometranslocations can be generated from two DNA double-strand breaks(DSBs) induced by the rare cutting restriction enzyme I-Sce1(12). Therefore, as an alternative approach to generate thetranslocations, we also introduced I-Sce1 recognition sequencesinto the second intron of the Aprt gene in the 5' and 3' cassettes(Fig. 1). Thus, this system allows the generation of chromosometranslocations at the same translocation breakpoints in multipleindependent events using two distinct mechanisms: (1) homologousrecombination mediated by Cre and (2) non-homologous end joiningof DSBs generated by I-Sce1.
Screen for DMC
We isolated 97 P-clones that contained independent random insertionsof the two Aprt cassettes. Subsequently, each P-clone was transientlytransfected with a Cre-expression plasmid and cultured in mediathat selects for Aprt-expressing cells (azaserine plus adenine).We were able to generate Aprt+ colonies from 83 different P-clones.During development of this system, we measured the efficiencyof Cre-mediated recombination for 33 different P-clones. Inthese clones, Aprt+ colonies were generated at frequencies between1x10–2 and 1x10–6 per transfected cell (SupplementaryMaterial, Table S1). The Aprt+ colonies were pooled from eachP-clone to generate 83 independent recombinant pools (R-pools).We pooled the colonies for three reasons. First, every cellfrom a given P-clone is expected to generate the same translocation(or limited number of translocations depending on the numberof plasmid cassette insertions) following any Cre-mediated event,and assaying multiple clones from 83 different P-clones wouldbe prohibitive. Second, if DRT/DMC occurs in a stochastic mannerand could occur at any translocation breakpoint, pooling thecolonies would allow us to detect chromosomes with DRT/DMC inevery pool at similar frequencies. Third, chromosomes that displayDRT/DMC are unstable (8) and pooling the Aprt+ colonies wouldallow for fewer generations prior to karyotypic analysis.
One critical aspect in the analysis of chromosomes with DMCis that a 2–3-h pre-treatment of the cultures with colcemidprior to mitotic harvest interferes with our ability to detectthe under-condensed phenotype (8). Therefore, to determine whetherany of the R-pools contained chromosomes with DMC, each R-poolwas harvested for mitotic cells in the absence of colcemid.Mitotic spreads from each R-pool were analyzed for the presenceof chromosomes with DMC. We scored a cell as DMC-positive ifit contained one or more chromosomes with at least two of thefollowing characteristics: (1) at least twice as long as anyother chromosome within the same spread, (2) less than halfas wide as any other chromosome within the same spread and/or(3) contained a bend of greater than 180° as previouslydescribed (8,9). Note that the parental HTD114 cells containthree pre-existing ICTs (Supplementary Material, Fig. S1),but display a relatively low frequency of spontaneous DRT/DMC,0.5% [Table 1 and (9)]. We found that the majority of the R-poolsdisplayed a relatively low frequency of DMC, ranging from 0to 4% (Table 1). In contrast, 13 pools contained chromosomeswith DMC in greater than 5% of mitotic cells, ranging from 6to 50%. This observation indicated that DMC did not occur inall R-pools with similar frequencies, suggesting that only certainCre-mediated chromosome rearrangements resulted in DMC. Examplesof mitotic spreads containing chromosomes with DMC from thisinitial screen are shown in Figure 2.
Because there is a low but detectable frequency of DMC in theparental HTD114 cells, each engineered translocation had tobe analyzed separately using chromosome-specific probes. Therefore,to determine which chromosomes were affected in each clone,we performed a series of karyotypic studies on nine P-clonesand their respective R-pools that contained DMC in >6% ofthe mitotic spreads, as well as on four P-clones and their respectiveR-pools that showed a low frequency of DMC. This initial karyotypicanalysis was carried out in the presence of colcemid to facilitatethe characterization of the chromosomes involved. Initially,we used fluorescence in situ hybridization (FISH), with theAprt plasmid cassettes as probes, followed by R-banding to identifythe sites of insertion. We next used whole chromosome paints(WCPs) to confirm the insertion sites in the P-clones and tovisualize the Cre-dependent translocations in the R-pools. Finally,the chromosome translocations present in the R-pools were confirmedusing G-banding. An example of this analysis is shown for P175and its Aprt+ pool R175 in Figure 3. P175 contains plasmid cassetteinsertions in 6q14–15 and 10q11.2 (Fig. 3A and B).As expected, R175 contains a new translocation involving chromosomes6 and 10 (Fig. 3C–E). The breakpoints for this newtranslocation are at the plasmid cassette insertion sites andgenerated a balanced translocation, t(6;10)(q14–15;q11.2).A schematic view of the plasmid cassette insertions and thet(6;10) is shown in Figure 3F. A summary of this karyotypicanalysis for all 13 clones is shown in Table 1. In addition,we have characterized the plasmid cassette insertions in thisset of P-lines and in their respective R-pools for copy numberand for reconstitution of the Aprt cassettes using southernblot hybridizations (Supplementary Material, Fig. S2).
One of the advantages of this approach is that the same translocationscan be generated in multiple, independent events. Therefore,to determine if a particular translocation displayed DMC followingindependent Cre events, additional R-pools were generated andanalyzed for DMC in mitotic spreads prepared in the absenceof colcemid. We analyzed two independently derived pools ofR175 (R175A and R175B) using FISH with chromosome 6 and 10 WCPs.Figure 4A–D shows examples of this analysis and indicatedthat chromosomes with DMC hybridized to both the chromosome6 and 10 probes in both isolates of R175. Analysis of P268 andtwo Aprt-expressing pools, R268A and R268B, indicated that abalanced translocation involving the long arm of chromosome15 and the long arm of chromosome 16, t(15q;16q)(q24;q12.1)also displays DMC (Table 1 and Supplementary Material, Fig. S3).These results indicate that DMC was induced in two independentlyderived pools of R175 and R268, and it occurred on the Cre-dependentt(6;10) and t(15;16), respectively. A similar analysis of R27,R186 and R276 showed that DMC also occurred on Cre-dependentbalanced translocations (Table 1). In contrast, DMC was notdetected on balanced translocations generated by Cre in twoindependent pools each of R38, R161, R244, R248 or R263, withover 200 metaphase spreads analyzed from each pool (Table 1,Supplementary Material, Fig. S4, data not shown). Theseresults indicate that DMC was detected in independent isolatesof certain Cre-dependent chromosome translocations and not inothers, indicating that this abnormal chromosomal phenotypeis not the consequence of a stochastic process that could occurat any translocation breakpoint.
Only one or both derivative chromosomes display DMC
Because the DMC phenotype occurred on balanced translocationsand every balanced translocation generates two derivative chromosomes,we next determined whether one or both derivative chromosomesdisplayed the DMC phenotype. To address this question, we analyzedmitotic spreads for chromosomes with DMC using FISH with chromosome-specificcentromeric probes. Figure 4E and F shows two examples of thisanalysis for R175 and indicated that the chromosomes with DMChybridized to the chromosome 10 centromeric probe and not tothe chromosome 6 centromeric probe. In total, we detected 10DMC chromosomes that hybridized to the chromosome 10 probe andno DMC chromosome that hybridized to the chromosome 6 probe.A similar analysis on R268 indicated that the chromosome 15derivative, not the chromosome 16 derivative, displayed DMC(Supplementary Material, Fig. S3). Analysis of a thirdtranslocation with DMC, a t(3;16)(p13;p13.3) present in R27,also indicated that only one product, the 16 derivative, displayedDMC (data not shown). Although it would be impossible to provethat any given chromosome ‘never’ displays DMC,these observations suggest that only one of the products fromthese three translocations displays the phenotype. In contrast,analysis of a fourth translocation, a t(3;13) present in R186,indicated that both derivative chromosomes displayed DMC (Fig. 5).Note that P186 contains three independent sites of plasmid cassetteinsertion, 3(q29), 11(p15) and 13(q14). However, the only newtranslocation detected in two independent R186 pools was a t(3;13)(q29;q14),indicating that the 11(p15) insertion site did not participatein the Cre events that generated the Aprt+ cells in the R186pools. The reason that the 11p cassette has not been detectedin a translocation in R186 is unknown, but it may contain adeleted or rearranged cassette that prevents reconstitutionof a functional Aprt gene, or perhaps the frequency at whichthe 11p15 cassette can participate in a Cre event with the othercassette may simply be much lower than the frequency for theother insertion sites (Supplementary Material, Table S1). Regardless,using FISH with chromosome 3 and 13 centromeric probes, we detectedDMC on both derivative chromosomes from t(3;13) (Fig. 5Fand G). These observations indicate that either one or bothof the products from certain balanced translocations displaythe DMC phenotype.
Engineered chromosomes with DMC also display DRT
The DMC phenotype is preceded by a DRT of the entire chromosome(8). Therefore, to confirm that the engineered chromosomes withDMC also display DRT, we analyzed replication timing of thechromosomes with DMC using a BrdU incorporation assay combinedwith FISH using chromosome-specific centromeric probes. Figure6A shows a schematic illustration of this replication timingassay. Analysis of mitotic spreads from R175B harvested forlate replication indicated that the chromosomes with DMC hybridizedto the chromosome 10 centromeric probe and incorporated BrdUalong their length at a time when the fully condensed chromosomesdid not show any detectable BrdU incorporation. An example ofthis analysis is shown in Figure 6B–D. This analysis indicatesthat the derivative chromosome 10 present in R175 displays bothDRT and DMC.
Plasmid insertions and small intra-chromosomal deletions are not sufficient to induce DRT/DMC
The results described earlier indicate that DRT/DMC occurs onbalanced translocations induced by Cre-mediated recombination,and that one or both of the derivative chromosomes display thephenotype. These observations suggest that DRT/DMC occurs followinginter-chromosomal exchanges at specific chromosome locations.However, it was also possible that simple plasmid cassette insertionor Cre-mediated recombination at these specific chromosome locationswas responsible for inducing DRT/DMC. Therefore, to determinewhether DRT/DMC was simply the result of plasmid cassette insertion,we analyzed mitotic spreads for DMC in P-clones that generatedDMC at a high frequency in their R-pools. None of the P-clonesassayed displayed DMC in greater than 4% of the spreads (Table1). In addition, P175 and P27 were assayed for DMC combinedwith FISH with centromeric probes. This analysis indicated thatP175 and P27 did not retain DMC chromosomes that hybridizedto the chromosome 6 and 10, or 3 and 16 centromeric probes,respectively (Table 2). Therefore, simple plasmid cassette insertionat these chromosomal locations is not sufficient to induce DRT/DMC.
Next, to determine whether DRT/DMC was simply the consequenceof Cre-mediated recombination at these specific chromosomallocations, we took advantage of the fact that our 5' and 3'cassettes contain Neo and Hyg genes flanked by loxP sites (i.e.floxed; Fig. 1). Previous studies indicate that intra-chromosomalCre events are markedly more efficient than inter-chromosomalCre events (13). Thus, the design of this set of experimentswas to transiently transfect P-clones with a Cre-expressionvector and isolate clones that experienced Cre-mediated intra-chromosomaldeletion of the Neo and Hyg genes but did not generate ICTs.To aid in the isolation of Cre-transfected cells, we co-transfecteda GFP-expression vector and isolated the GFP-positive cellsby fluorescence-activated cell sorting (FACS). The GFP-positivecells, and presumably Cre-positive cells, were plated at clonaldensities, allowed to form colonies in the absence of Aprt selectionand isolated as individual clones. For this analysis, we chosetwo clones, P175 and P27, that generated DRT/DMC on Cre-dependentICTs. Deletion of the Neo and Hyg genes occurred in >90%of these clones as assayed by southern blot hybridization (SupplementaryMaterial, Fig. S2C), and the absence of the translocationswas confirmed by southern blot hybridization for reconstitutionof the Aprt gene and FISH with WCPs (data not shown). This setof experiments resulted in the generation of cell lines thatexperienced Cre-mediated intra-chromosomal deletion of the Neoand Hyg genes at the same chromosomal locations involved inthe translocation events produced in R175 and R27. The cloneswith the Neo and Hyg genes removed were designated P175NH andP27NH, respectively. Three independent P175NH clones were assayedfor DMC after hybridization with chromosome 6 and 10 centromericprobes and two independent P27NH lines were assayed for DMCafter hybridization with chromosome 3 and 16 centromeric probes.Chromosomes with DMC were not detected in any of the P175NHor P27NH clones with any of the centromeric probes (Table 2).Taken together, these observations indicate that insertion ofthe plasmid cassettes and Cre-mediated intra-chromosomal deletionsat these specific chromosome locations is not sufficient toinduce DRT/DMC and suggest that DRT/DMC is the result of inter-chromosomalexchanges at specific chromosomal locations. However, it isimportant to point out that this analysis does not rule outthe possibility that other intra-chromosomal deletions or rearrangementscould result in DRT/DMC.
DRT/DMC occurs on translocations
The results described earlier indicate that DRT/DMC can occurfollowing Cre-mediated site-specific recombination, but onlyon specific translocation derivatives. To determine if a differentmechanism for generating the translocations could produce DMCon one of these derivatives, we took advantage of the fact thatour 5' and 3' cassettes have I-Sce1 sites in the second intronof the Aprt gene (Fig. 1). We transiently transfected P175cells with an I-Sce1-expression vector and selected for Aprt-expressingcells. Because DSBs induced by I-Sce1 are subject to exonucleaseactivity prior to NHEJ and could potentially result in translocationswith different breakpoints (12), we isolated Aprt+ clones, namedthe RS175 series. Karyotypic analysis of RS175-1 indicated thatthe expected balanced translocation, a t(6;10)(q12–13;q11.2),was generated (Fig. 7A–C). Although we have not sequencedthe breakpoints in RS175-1, t(6;10) is cytogenetically identicalto the translocation produced by Cre in the R175 pools. In addition,southern blot analysis indicated that the Aprt gene was reconstitutedin the second intron, and the Neo and Hyg genes were still intact(Fig. 1; data not shown). Therefore, the translocationbreakpoints in RS175-1 are located within the plasmid cassettes,indicating that the DSBs induced by I-Sce1 did not experiencesignificant exonuclease digestion prior to the generation ofthe t(6;10) in this clone. To determine if DMC was occurringon the chromosome 10 derivative, as it is in R175, we analyzedmitotic spreads from RS175-1 using FISH with the chromosome6 and 10 centromeric probes. An example of this analysis shownin Figure 7D indicated that the chromosome 10 derivative, notthe chromosome 6 derivative, displayed DMC. A similar analysison I-Sce1-induced translocations from P244 and P38, which didnot generate DMC following Cre, showed undetectable levels ofDMC on the expected translocations (data not shown). Therefore,this analysis indicates that ICTs produced by two distinct mechanisms,Cre-mediated homologous recombination or NHEJ of DSBs, can generatechromosomes with DRT/DMC.
Chromosomes with DRT/DMC are unstable
We previously found that chromosomes with DRT/DMC participatein frequent secondary rearrangements and translocations (8).During characterization of the R-pools described here, we alsoobserved unexpected rearrangements involving the chromosomeswith DRT/DMC (Fig. 5 and Supplementary Material, Fig. S3).To further characterize these secondary rearrangements and tocompare the stability of DRT/DMC chromosomes to non-DRT/DMCchromosomes, clones containing Cre-induced translocations wereisolated and expanded through a defined number of generations(20–22) and examined for unexpected chromosome rearrangements.For this analysis, we isolated Aprt+ clones from four differentP-clones: three that generated DRT/DMC following Cre-mediatedrecombination (P175, P186 and P268) and one that did not (P244);these clones were named the R175F, R186F, R268F and R244F series,respectively. In addition, to characterize the stability ofthe chromosomes prior to the generation of the Cre-induced translocations,the P-clones were transfected with an intact Aprt-expressionvector and stable Aprt-expressing clones isolated; these cloneswere named the P175F, P186F, P268F and P244F series. Analysisof these Aprt-transfected clones would allow us to determinethe stability of the chromosomes prior to creation of the translocationsand to control for any variability in chromosome stability betweendifferent P-clones subjected to transfection, Aprt selectionand clonal expansion in AA media. Because the cells constitutingeach clone were cultured through a similar number of generations(20–22) and were grown under identical conditions, thestability of the chromosomes could be directly compared bothbefore and after the translocation events and between DRT/DMCand non-DRT/DMC containing clones. In addition, we analyzeda series of P268F clones that had experienced Cre transienttransfection, but did not generate the translocation. Theseclones, designated P268NH, were generated by co-transfectionwith Cre and GFP-expression vectors, FACS for GFP-expressingcells, expansion in the absence of selection and screening fordeletion of the Hyg gene by southern blot hybridization (datanot shown). These clones were cultured through a similar numberof generations as the other clones (20–22) and servedas an additional control for non-specific affects of Cre transienttransfection.
Mitotic spreads from each clone were analyzed using FISH withWCPs that hybridized to the Cre-dependent translocations, and100 mitotic spreads were scored for each clone with each WCP.This set of experiments allowed us to determine the frequencyat which unexpected gross chromosomal rearrangements (GCRs:translocations, deletions, insertions and rearrangements) occurredon Cre-dependent translocations, with or without DRT/DMC.
These cell lines could clearly be segregated into two groupsbased on the frequency of new GCRs. All the control Aprt-transfectedclones (P175F1–4, P186F1–4, P268F1–4 and P244F1–4),the non-DRT/DMC translocation containing clones (R244F1–5)and the P268NH clones displayed low or undetectable frequenciesof new GCRs involving the appropriate chromosomes. For example,we did not detect any GCRs involving chromosomes 15 and 16 inthe P268F or P268NH clones, and we did not detect any GCRs involvingchromosomes 13 and 16 in the P244F clones (Table 3). In addition,we detected only one GCR involving chromosome 13 in one cellfrom one R244F clone (Table 3 and Supplementary Material, Fig. S5).In contrast, the DRT/DMC containing clones showed dramaticallyhigher frequencies of GCRs involving the DRT/DMC chromosomes.For example, every R268F clone contained GCRs that hybridizedto the chromosome 15 and/or 16 WCPs, and cells with GCRs werealso quite frequent within each clone, occurring in 19–100%(Table 3 and Supplementary Material, Fig. S6). Furthermore,each R268F clone retained a different array of GCRs and manyof these GCRs represented translocations to other, and presumablynon-DRT/DMC chromosomes (Fig. 8). Interestingly, and consistentwith our observation that only the chromosome 15 derivativedisplayed DRT/DMC in the R268 pools (Supplementary Material,Fig. S3), the chromosome 16 derivative was still presentand un-rearranged in all of the R268F clones (Fig. 8).This observation indicates that the chromosome with DRT/DMC,the 15 derivative, generated the GCRs in the R268F clones. Inaddition, to determine whether any of the Cre-treated cloneshad integrated the Cre-expression vector, which could potentiallycontribute to new GCRs via Cre-mediated recombination betweenendogenous or cryptic loxP sites (14,15), we carried out southernblot hybridizations on genomic DNA extracted from all of theCre-treated clones shown in Table 3. This analysis indicatedthat none of these clones retained the Cre-expression vector(data not shown).
Finally, using a fluctuation analysis (16), as modified by Linet al. (17) to allow for estimating high mutation rates,we estimate that the rate at which cells acquired new GCRs involvingthe DRT/DMC chromosomes was 7x10–1 (or one new GCR every14 cell divisions) for the t(6;10) in R175, 1.6x10–1 (orone new GCR every six cell divisions) for the t(3;13) in R186and 1.7x10–1 (or one new GCR every five cell divisions)for the t(15;16) in R268. We also point out that this is likelyto be an underestimate of the rate for GCR formation involvingthese translocations, as many of the R-clones retained GCRsin a large fraction of the cells, ranging from 2 to 100% ofthe cells (Table 3). In contrast, we estimate that the rateat which cells acquired new GCRs involving the appropriate non-rearrangedchromosomes (i.e. in P175F, P186F, P244F and P268F clones),non-DRT/DMC translocations (i.e. the t(13;16) in R244F) andin Cre-transfected but non-translocation carrying clones (i.e.P268NH) was 2.5x10–3 (or one new GCR every 400 cell divisions).Therefore, we estimate that chromosomes with DRT/DMC have an30–80-fold increase in the rate at which new GCRs occuron the affected chromosomes.