A eukaryotic chromosome consists of a centromere, two telomeres and a number of replication origins, and ‘artificial chromosomes’ may be created in yeast and mammals when these three elements are artificially joined and introduced into cells. Plant artificial chromosomes (PACs) have been suggested as new vectors for the development of new crops and as tools for basic research on chromosomes. However, indisputable PAC formation has not yet been confirmed. Here, we present a method for generating PACs in the model plant Arabidopsis thaliana using the Cre/LoxP and Activator/Dissociation element systems. The successfully generated PAC, designated AtARC1 (A. thaliana artificial ring chromosome 1), originated from a centromeric edge of the long arm of chromosome 2, but its size (2.85 Mb) is much smaller than that of the original chromosome (26.3 Mb). Although AtARC1 contains only a short centromere domain consisting of 180 bp repeats approximately 250 kb in length, compared with the 3 Mb domain on the original chromosome 2, centromere-specific histone H3 (HTR12) was detected on the centromeric region. This result supported the observed stability of the PAC during mitosis in the absence of selection, and transmission of the PAC to the next generation through meiosis. Because AtARC1 contains a unique LoxP site driven by the CaMV 35S promoter, it is possible to introduce a selectable marker and desired transgenes into AtARC1 at the LoxP site using Cre recombinase. Therefore, AtARC1 meets the criteria for a PAC and is a promising vector.
The centromeres of most higher eukaryotes consist of tandem repeats, with typical unit sizes of 150–200 bp; however, the total size at each of the centromeres may range between several hundred kilobase pairs and megabase pairs (Henikoff et al., 2001). Thus, construction of artificial chromosomes was anticipated to be quite difficult in mammals; however, mammalian artificial chromosomes have been constructed by introducing 35 kb–1 Mb of α–satellite arrays into human cultured cells, together with linked or unlinked telomeres (Harrington et al., 1997; Ikeno et al., 1998). These successes encouraged plant researchers to generate plant artificial chromosomes (PACs) as new vectors for engineering plant genomes. However, because plants have rigid cell walls, making the introduction of long DNA molecules into cells difficult, it is not a simple procedure to construct PACs using the bottom-up approach used in yeast (Murray and Szostak, 1983) and mammals. Although there have been two reports on de novo formation of maize artificial chromosomes to date (Carlson et al., 2007; Ananiev et al., 2009), these chromosomes were generated by particle bombardment, which is damaging to long molecules of DNA. Indeed, the results are still controversial, and the mechanisms for forming mini-chromosomes remain unclear (Houben et al., 2008).
Telomere-associated chromosome fragmentation (Farr et al., 1992) has been used to create mini-chromosomes in maize (Yu et al., 2006). Although use of this type of top-down approach was also demonstrated in Arabidopsis (Nelson et al., 2011; Teo et al., 2011), it is difficult to control the telomere-directed chromosome truncations because Agrobacterium-mediated transformation results in random insertions of the cloned telomere sequences (Kim et al., 2007). In addition, T-DNA insertions around or within the centromeres, which possibly increase the efficiency of PAC generation, were found to be limited under selective conditions. Therefore, alternative approaches, which are applicable to a wide range of plant species, are needed to generate PACs.
Strategy for generating artificial ring chromosomes
As we recently found that ring mini-chromosomes are relatively stable in A. thaliana during mitosis and are transmissible to the next generation (Murata et al., 2008; Yokota et al., 2011), we attempted to generate a plant artificial ring chromosome using the Cre/LoxP system in combination with Activator (Ac)/Dissociation (Ds) transposons. An outline of the strategy used in this study is shown in Figure 1. To implement this strategy, it was necessary to generate plants carrying two LoxP sites, separated from each other on the same chromosome, with a portion of centromere sequence between the two LoxP sites. To identify plant lines that meet these criteria, we first searched the A. thaliana insertion datasets using the BLASTN program and centromeric 180 bp repeats (e.g. GenBank accession number S73532; Murata et al., 1994) as the query (http://www.arabidopsis.org/Blast/). In particular, we focused on the WiscDsLox T–DNA lines (Woody et al., 2007) because these lines were created by transformation with binary vector pDs–Lox, which contains two LoxP sites on the T–DNA, one of which is located within the maize Ds element (Figures 1 and 2d). Hence, the DS–LoxP–DS (DLD) cassette is able to mobilize to other sites when the maize Ac transposase (TPase) is expressed in the cells.
Characterization of the T–DNA insertion lines
Fourteen lines were selected as possible candidates by the BLAST search (Table S1), and seeds were obtained from the Arabidopsis Biological Resource Center (Columbus, OH). To investigate the number of T–DNA insertion sites and the number of copies per site, DNA fiber-fluorescence in situ hybridization (f–FISH; Shibata and Murata, 2004) was performed using a probe for the T–DNA sequence (Figure 2d) and the centromere-specific 180 bp repeats. The f–FISH analysis revealed that most of the lines investigated contain two or more T–DNA insertion sites. Most insertions consist of multi-copy T–DNA segments, with the exception of one line, CS859268 (Figure 2a). Because multi-copy and/or multi-locus T–DNAs result in complex segregation patterns in the progeny, the subsequent experiments used the single-copy T–DNA insertion line, CS859268, in the 180 bp repeat array.
To identify chromosomes with the T–DNA insertion in the CS859268 line, FISH was performed using 5S and 18S rDNA and T–DNA sequences as the probes, revealing a T–DNA insertion in the centromeric region adjacent to the long arm of chromosome 2 (Figure 2b). Although the PCR analysis also showed that the T–DNA is inserted in the 180 bp repeat array of chromosome 2 (CEN2; Figure S1), it was unclear how the T–DNA element inserted into that location. Because the size of CEN2 is estimated to be approximately 3 Mb (Hosouchi et al., 2002), it is important to determine the orientation and position of the T–DNA to ascertain whether there is an opportunity for ring chromosome formation. Unfortunately, the A. thaliana genome sequence is incomplete or unreliable for BAC clones around CEN2, such as T12J2 and T6C20, mainly due to the existence of the homogeneous 180 bp repeat arrays. To resolve this issue, we performed f–FISH using the 180 bp repeats and four fragments amplified from T–DNA as the probes, and visualized the resulting FISH signals using three pseudocolors (yellow, pink and light blue; Figure 2c,d). Consequently, the T–DNA insertion site was determined to be in the 180 bp repeat array in the reverse orientation at a distance of approximately 250 kb (251.6 ± 11.1 kb, n =5) from the edge of the long arm (Figures 2c and S2).
Launching the DLD cassette
To launch the DLD cassette from the T–DNA inserted in CEN2, the CS859268 line was crossed with plants expressing the Ac TPase (Ac–Rlc–62 or -63) linked to the neomycin phosphotransferase II gene (Zhang et al., 2003). In the progeny, 20 plants expressing resistance to both glufosinate and kanamycin were identified; these plants were allowed to self-pollinate, and their seeds were collected for subsequent experiments. When the DLD cassette is excised from the T–DNA insertion, the hygromycin-resistance gene is expressed as a result of fusion with the CaMV 35S promoter. Therefore, the F2 seeds were germinated on MS medium containing hygromycin B and glufosinate, and the seedlings containing no Ac TPase gene were identified by PCR. To map the DLD cassettes that had excised and re-inserted, TAIL–PCR (Liu and Whittier, 1995) or inverse PCR (Ochman et al., 1988) were performed using primers designed from the Ds sequences and/or arbitrary primers (Table S2). As a result, four plants (10B5, 10F7, 14D6 and H003) were found to have the DLD cassette at a position distinct from the original site but still on chromosome 2. The PCR sequences revealed that two of the four plants (10B5 and 14D6) had the DLD cassette on the short arm of chromosome 2, whereas the other two plants (10F7 and H003) had the DLD cassette on the long arm (Figure S3). However, the plant designated H003 withered and died before maturation.
Ring chromosome generation
Among the three remaining plants, only 10F7 had two LoxP sites oriented in the same direction (Figure 1, upstream transposition). Therefore, plant 10F7 was crossed with plants expressing Cre recombinase (Cre13; Marjanac et al., 2008) to induce site-specific recombination between the two LoxP sites on chromosome 2. A circular DNA molecule containing approximately 250 kb of the 180 bp repeat array and its flanking sequence was expected to be formed by Cre-mediated recombination. Twenty-eight F1 plants showed a PCR band (approximately 0.4 kb) using primers Ca35S and Luc, which target the CaMV 35S promoter and the luciferase gene, respectively (Figures 1 and S4). These plants were investigated by FISH using the 180 bp repeats as the probes, and mini-chromosomes were detected in two of the F1 plants (numbers 21 and 24; Figure 2e). Although the frequency of mini-chromosome appearance was low (approximately 5% of the cells investigated), the generation of mini-chromosomes by sequence-specific recombination between the two LoxP sites was clearly demonstrated by FISH using BAC clone F9B22, which has been mapped to the long arm of chromosome 2 (AGI map position 5 833 104–5 928 868), as a probe. FISH signals appeared on one of the chromosome 2 pair and on the mini-chromosome (Figure 2e, left), but no telomere sequences were detected on the mini-chromosome (Figure 2e, right), suggesting that the mini-chromosome has a ring-shaped structure. This structure was confirmed by observation of pachytene chromosomes (Figure S5).
As the DLD re-insertion site was mapped to the 3rd exon of At2 g14620 (AGI map position 6 245 200) on the long arm of chromosome 2, the size of the mini-chromosome was estimated to be 2.85 Mb (0.25 Mb of the 180 bp repeat array plus 2.6 Mb of the peri-centromeric region; Figure S3). However, the size is doubled to be 5.7 Mb if an exchange occurs between the sister chromatids after replication (Figure S6).
Stability and transmission of the mini-chromosome
To investigate the stability and transmission of the ring mini-chromosome, approximately 1000 F2 plants were investigated first by PCR and then by FISH using the 180 bp repeat as the probe. Mini-chromosomes were observed at high frequencies (>95%) in all of the mitotic cells investigated from all of the F2 plants that showed a 0.4 kb PCR band with the Ca35S and Luc primer pair (Figure S4). Although some F2 plants still carried the Cre recombinase gene(s), the stability of the mini-chromosome was confirmed by FISH in the pollen mother cells of plant numbers 21–231, 21–531 and 24–414 carrying no Cre gene (Figure 2f). The transmission of the mini-chromosome, designated AtARC1 (A. thaliana artificial ring chromosome 1), was investigated by PCR and FISH in both selfed and crossed progeny of plant number 21–231 (Table 1). The transmission rate was approximately 8% from the female parent and 18% from the male parent, suggesting that AtARC1 transmission occurs at a higher frequency from pollen. However, this low transmission rate from the male parent was highly dependent on the structure of chromosome 2, as described below. Thus, when AtARC1 was added to wild-type plants as a supernumerary chromosome (e.g. plant number 43–07, Figure 3a) by subsequent back-crossing, the transmission rate increased by 10% compared with that observed in plant number 21–231 (Table 1). However, for the transmission rate from the female parent there was no distinct difference between the two plant genotypes (7.8% versus 11.1%). Surprisingly, in both plant genotypes, AtARC1 was transmitted to over 40% of the selfed progeny.
Table 1. Transmission rates for an artificial ring chromosome (AtARC1) in plant numbers 21–231 and 43–07
Col, 21–231 and 43–07 represent a wild-type plant (Col–0, 2n = 2x = 10), a plant carrying D and L variants of chromosome 2 and AtARC1 (2n = 2x (D/L) + AtARC1 = 11), and a wild-type plant carrying AtARC1 (2n = 2x (W/W) + AtARC1 = 11), respectively.
One AtARC1 was present in all plants investigated.
Col (♀) × 21–231 (♂)
21–231 (♀) × Col (♂)
21–231 (♀) × 21–231 (♂)
Col (♀) × 43–07 (♂)
43–07 (♀) × Col (♂)
43–07 (♀) × 43–07 (♂)
In this experiment, four structurally distinct forms of chromosome 2 were expected to occur: wild-type (W), T–DNA insertion (T), Ds transposition (D) and deletion (L; Figure 3c). These four variants are distinguishable from each other by PCR using specific primer sets (Figure 3c and Table S2). All six combinations of AtARC1 with the three chromosome 2 variants (W, D and L) were identified in the selfed and crossed progeny of the F2 plants (Figure 3d). Variants T and D were present in CS859268 and 10F7, respectively. Recombination driven by Cre recombinase simultaneously produced the L variant and AtARC1. Although all of the plants with the six chromosomal combinations were fertile, the seed set of the L/L+AtARC1 plants was significantly lower than that of the others due to partial pollen sterility. This is mainly because AtARC1 is able to compensate for the 2.85 Mb deletion on chromosome 2. Intriguingly, AtARC1 appeared to be essential to pollen development but not for egg development: transmission of AtARC1 through the female parent was limited, and all of the selfed progeny of the L/L+AtARC1 plants were found to carry at least one copy of AtARC1. According to the TAIR10 genome release (http://www.arabidopsis.org/), 597 genes have been annotated between the two LoxP sites separated on chromosome 2 in the 10F7 line, corresponding to 2.6 Mb between BAC T6C20 and AT2G14620 in T6B13. Therefore, deleterious effects caused by gene duplication or deletion are expected, depending on the structure of chromosome 2 and the presence of AtARC1. Most of the annotated genes are classified as transposons and pseudogenes (438/597, 73.3%). The effects of the remaining 153 protein-coding and six RNA-coding genes, which were transferred into AtARC1 from chromosome 2, are crucial. However, there were no obvious phenotypic alterations in the W/W+AtARC1 plants compared with wild-type plants carrying no AtARC1 (Figure 3b).
The size of the 180 bp repeat cluster on AtARC1 was estimated to be approximately 250 kb by f–FISH, which is much shorter than on the normal chromosome 2 (approximately 3 Mb; Hosouchi et al., 2002). Nevertheless, the centromere function appeared to be almost normal: the AtARC1 chromosome was transmissible to the next generation through mitotic and meiotic divisions. To support this conclusion, localization of a centromere-specific protein, HTR12 (Arabidopsis centromeric histone H3 variant, CENH3; Talbert et al., 2002), was investigated by indirect immunostaining using anti-HTR12 and subsequently by FISH using a 180 bp repeat as the probe. Immunosignals from HTR12 on AtARC1 co-localized with the FISH signals of the 180 bp repeats in both monocentric and dicentric AtARC1 (Figure 4), indicating that the centromere of AtARC1 is able to recruit the kinetochore proteins normally. It was also noted in this experiment that AtARC1 maintained a monocentric form (85% in plant number 21–231 and 72.7% in plant number 43–07) rather than a dicentric form (Figures 2f and 4). Although AtARC1 contained part of the long arm of chromosome 2, no pairing between AtARC1 and chromosome 2 was found to occur at meiosis (Figures 2f and 4); this is important for maintenance of the integrity of the plant genome in the presence of an artificial chromosome.
The Cre protein was shown to be able to induce intra-chromosomal recombination between two LoxP sites separated by several megabases in mice (Ramirez-Solis et al., 1995), although the efficiency decreased as the distance increased (Zheng et al., 2000). However, no such events, including large rearrangements and ring chromosome formation, have been reported in plants. Up to 200 kb somatic deletions (Coppoolse et al., 2005) and a 130 kb somatic and germinal inversion (Stuurman et al., 1996) were induced in tomato (Solanum lycopersicum), but the efficiency decreased with increased distance between the two LoxP sites. In A. thaliana, a 547 kb deletion appeared to be generated, but plants carrying this deletion were not recovered (Zhang et al., 2003). The present study demonstrated that Cre recombinase induces recombination between two LoxP sites that are more than 2.5 Mb apart, resulting in circularization of long DNA molecules in A. thaliana, including the repetitive centromeric sequences. The resulting plant artificial ring chromosome, approximately 2.85 Mb in size, was successfully transferred to F2 progeny and to wild-type plants as a supernumerary chromosome (Figure 3a,d).
Ring chromosomes have been found in a wide range of organisms, from yeast to animals and plants (Yokota et al., 2011), but the majority are unstable during cell divisions and are rarely transmissible to the next generation through meiosis. Ring chromosomes tend to convert to a double-size ring chromosome with two centromeres as the result of an odd number of recombination(s), which results in chromosomal breakage (Figure S6; McClintock, 1938). Interestingly, however, the AtARC1 molecule generated in this study predominantly maintained a monocentric form (72–85%) rather than a dicentric form, which may reflect the low frequency of recombination around the centromeres, resulting in the increased stability of AtARC1. When AtARC1 was added to wild-type plants (2n = 2x (W/W) + AtARC1 = 11), the monocentric form was more stable than the dicentric form. However, it should be noted that the monocentric form does not appear to be essential for the stability of the ring chromosome: some dicentric ring mini-chromosomes were found to be stably transmissible to the next generation (Murata et al., 2008; Yokota et al., 2011).
AtARC1 contains approximately 250 kb of the 180 bp repeat arrays, the size of which is comparable to that of another ring mini-chromosome (miniδ1–1; Yokota et al., 2011). Although miniδ1–1 is less stable at mitosis than AtARC1, HTR12 (CENH3) was found on both centromeres, indicating that the centromere domains (approximately 250 kb) are sufficient for loading the kinetochore protein as well as for centromere functions. Interestingly, all stable ring mini-chromosomes including AtARC1 contain the peri-centromeric region in addition to the 180 bp repeats >250 kb from the edge of the short arm or long arm of chromosome 2. This suggests that the peri-centromeric region is also needed for cohesion and subsequent accurate separation of sister chromatids (Murata, 2013). A similar situation was also found in human X mini-chromosomes, with an active sub-domain anchored approximately 150 kb from the Xp edge (Spence et al., 2002).
In addition to its stability and independence from other chromosomes, AtARC1 has an additional advantage as a chromosome vector. The existence of a unique LoxP site under the control of the CaMV 35S promoter makes possible the introduction of a selectable marker (Louwerse et al., 2007). Therefore, AtARC1, which is stably transmissible for generations without selection, is a supernumerary PAC containing a LoxP site for delivery of exogenous genes or DNA sequences into A. thaliana. Furthermore, the plant lines identified and created in this study are useful for studying centromere function and for constructing other artificial chromosomes in Arabidopsis. Although de novo formation of maize artificial chromosomes by particle bombardment has been reported (Carlson et al., 2007; Ananiev et al., 2009), no ‘bottom-up’ approaches were successful in A. thaliana. There is a possibility that the long 180 bp repeat arrays introduced contained no unidentified sub-domain(s) that is essential for de novo kinetochore formation, like the CENP–B box of α–satellites in human artificial chromosome formation (Masumoto et al., 1998; Ohzeki et al., 2002). This issue may be solved using the top-down approach proposed here.
The seeds of the WiscDsLox T–DNA lines and AcRLC–62 and -63 of Arabidopsis thaliana ecotype Columbia used in this study were obtained from the Arabidopsis Biological Resource Center (Columbus, OH). Cre-expressing seeds from plant line Cre13 (ecotype C24; Marjanac et al., 2008) were a gift from A. Depicker (VIB Department of Plant Systems Biology, Ghent University, Belgium). The seeds from the WiscDsLox T–DNA lines and Cre13 were germinated on MS medium containing 10 μg ml−1 glufosinate (Basta®; Sigma-Aldrich, http://www.sigmaaldrich.com/), and the seeds from the AcRLC–62 and -63 lines were germinated on MS medium containing 50 μg ml−1 kanamycin (Sigma-Aldrich). One-month-old seedlings were transferred to small pots (8 cm diameter) containing soil.
All PCR experiments in this study were performed using Applied Biosystems (http://www.lifetechnologies.com/) thermal cyclers (GeneAmp PCR system 9700). To investigate the DLD re-insertion sites, TAIL–PCR or inverse PCR were performed as described previously (Ochman et al., 1988; Liu and Whittier, 1995), using nested primers and/or arbitrary degenerate (AD) primers (Table S2). Template DNA was purified from lyophilized leaves of the hygromycin-resistant F2 plants from crosses between CS859268 and AcRLC–62 or -63, using a DNeasy plant mini kit (Qiagen, http://www.qiagen.com). Primary TAIL–PCR was performed using one of the AD primers in combination with Ds1–I to amplify the left-side sequence of the DLD cassette, or with Ds2–I to amplify the right-side sequence of the DLD cassette. Secondary and tertiary TAIL–PCR were similarly performed using the same AD primer as used in the primary TAIL–PCR, in combination with Ds1–II and Ds1–III to amplify the left-side sequence, and Ds2–II and Ds2–III to amplify the right-side sequence, respectively. The PCR programs used were the same as described previously (Liu and Whittier, 1995).
To screen for transmission of AtARC1 or the AC TPase or Cre gene, PCR was performed using specific primer sets (Table S2) and the following conditions: 94°C for 2 min, followed by 30 cycles of 94°C for 30 sec, 55°C for 30 sec and 72°C for 45 sec, with final extension at 72°C for 3 min. Template DNA was extracted from leaves using a DNeasy plant mini kit (Qiagen) and/or a simple and rapid isolation method (Edwards et al., 1991), especially when a large number of samples were handled for PCR analyses.
FISH, fiber-FISH and immunolabeling
FISH and f–FISH were performed essentially as described previously (Fransz et al., 1996; Shibata and Murata, 2004; Murata et al., 2008). For detection of centromeres, the 180 bp repeats amplified by PCR were labeled using an FITC High Prime kit (Roche Applied Science, http://www.roche-applied-science.com/; Shibata and Murata, 2004), and other probes including BAC clones from chromosome 2, the 500 bp 5S rDNA repeat unit (Murata et al., 1997) and the 18S rDNA (Shibata and Murata, 2004) were labeled either with biotin-7–dATP (Life Technologies) or digoxigenin-11–dUTP (Roche Applied Science) by nick translation, and detected using streptavidin–Cy5 conjugate (GE Healthcare Japan, http://www.gelifesciences.co.jp/) and anti-digoxigenin rhodamine Fab fragments (Roche), respectively. Slides were stained using 0.1 μg ml−1 4',6–diamidino-2–phenylindole dihydrochloride (DAPI). Images were captured using a digital CCD camera (AxioCam HRm; Zeiss, http://microscopy.zeiss.com/microscopy/en_de/home.html) and overlaid using AxioVision 3.1 (Zeiss). For indirect immunolabeling of the centromere-specific histone H3 variant (CENH3), anti-HTR12 antibodies raised as previously described (Talbert et al., 2002) were applied to pachytene chromosomes prepared from pollen mother cells fixed in 4% v/v paraformaldehyde for 30 min at room temperature. Immunofluorescence signals were detected as described previously (Shibata and Murata, 2004).
This work was supported by the program for Promotion of Basic Research Activities for Innovative Biosciences (BRAIN), Japan.