Abstract
Transposable elements have long been considered as potential agents of large-scale genome reorganization by virtue of their ability to induce chromosomal rearrangements such as deletions, duplications, inversions, and reciprocal translocations. Previous researchers have shown that particular configurations of transposon termini can induce chromosome rearrangements at high frequencies. Here, we have analyzed chromosomal rearrangements derived from an unstable allele of the maize P1 (pericarp color) gene. The progenitor allele contains both a full-length Ac (Activator) transposable element and an Ac terminal fragment termed fAc (fractured Ac) inserted in the second intron of the P1-rr gene. Two rearranged alleles were derived from a classical maize ear twinned sector and were found to contain a large inverted duplication and a corresponding deficiency. The sequences at the junctions of the rearrangement breakpoints indicate that the duplication and deletion structures were produced by a single transposition event involving Ac and fAc termini located on sister chromatids. Because the transposition process we describe involves transposon ends located on different DNA molecules, it is termed nonlinear transposition (NLT). NLT can rapidly break and rejoin chromosomes and thus could have played an important role in generating structural heterogeneity during genome evolution.
TRANSPOSITION is essentially a biochemical reaction. The enzyme that catalyzes the reaction is transposase, and the substrates of transposase are the 5′ and 3′ termini of the transposon. Theoretically, noncontiguous 5′ and 3′ transposon termini could serve as transposase substrates, and transposition could involve transposon termini located on different chromosomes. Such transposition events involving dispersed transposon ends could lead to major chromosomal rearrangements, whereas ordinary transposition of a contiguous element changes only the location of the transposon in the genome. However, genomes containing multiple copies of related transposons are generally quite stable; this suggests that transposition involving noncontiguous transposon termini is rare. It has been estimated, on the basis of genomic Southern blot hybridizations, that the maize genome contains ~30–50 copies of Ac/Ds-like transposons (Fedoroffet al. 1983). However, it is unclear how many of these copies are transposition competent, as a certain proportion may be immobile fragments (Kunze 1996) or inactivated by epigenetic modifications associated with DNA hypermethylation (Leuet al. 1992). In the case of maize Ac/Ds elements, the ability of dispersed transposons to participate in transposition reactions may be further restricted by differences in the timing of replication of individual transposon ends (Wirtzet al. 1997) and by the methylation state of the transposon ends (Wang and Kunze 1998).
Transposition events resulting in chromosome breakage or other rearrangements can be detected by the use of appropriate genetic markers. In the early phases of transposon discovery, McClintock observed that transposition of Ds (Dissociation) was occasionally accompanied by chromosomal rearrangements, such as deletions, duplications, inversions, and reciprocal translocations. Because these rearrangements occurred only in the presence of Ac (Activator), it was believed that they were produced by Ac/Ds transposition events (McClintock 1953a,b, 1978). Subsequently, transposon-related rearrangements were also observed in Antirrhinum and Drosophila (Martin and Lister 1989; Listeret al. 1993; Lim and Simmons 1994); in some cases, however, the rearrangements were attributed to recombination between dispersed copies of transposons (Lim and Simmons 1994).
Most of the chromosomal rearrangements isolated by McClintock have not been studied at the molecular level, with the exception of the sh-m5933 allele. This allele contains a >30-kbp inverted duplication (Burr and Burr 1982) flanked by “double Ds” elements (one Ds inserted into a second Ds in opposite orientation; Courage-Tebbeet al. 1983; Döringet al. 1989). Federoff (1989) proposed that the inverted duplication in the sh-m5933 allele was generated by a transposition reaction involving Ds termini located on sister chromatids. Later, English et al. (1993) and Weil and Wessler (1993) proposed similar models to account for the phenomenon of Ds-induced chromosome breakage. These models predict that transposition reactions involving Ds termini on sister chromatids should generate reciprocal deletions and duplications. Such rearrangements were indeed identified in transgenic tobacco (Englishet al. 1995). However, in no single case have all rearrangement junctions been sequenced to identify the characteristic nucleotide sequence changes predicted to arise from the transposition process. These include the so-called “footprint” at the site of transposon excision, and the target site duplication of host sequences at the site of transposon insertion.
Here we report that noncontiguous transposon termini can serve as substrates for unusual transposition events and thereby generate major genome rearrangements. We analyzed rearranged chromosomes derived from a classical maize twinned sector (Greenblatt and Brink 1962). One chromosome has a large (4.6 cM) deficiency, while the other chromosome carries the deleted segment as an inverted duplication. The rearrangement breakpoints contain the footprints and target site duplications typically generated by Ac transposition; these sequences prove that the rearranged chromosomes are the reciprocal products of a single nonlinear transposition (NLT) event.
MATERIALS AND METHODS
Genomic DNA extractions, Southern blot hybridization, and genomic cloning: Total genomic DNA was prepared using a modified cetyltrimethylammonium bromide (CTAB) extraction protocol (Porebskiet al. 1997). Agarose gel electrophoresis and Southern hybridizations were performed according to Sambrook et al. (1989), except hybridization buffers contained 250 mm NaHPO4, pH 7.2, 7% SDS, and wash buffers contained 20 mm NaHPO4, pH 7.2, 1% SDS. Genomic libraries were prepared using λFix II vector and in vitro packaging reactions (Stratagene, La Jolla, CA). Genomic fragments were subcloned in pBluescript (Stratagene).
PCR amplifications: PCR amplifications were performed as described by Saiki (1989) using the following oligonucleotide primers: Ac5, GGAATTCGTTTTTTACCTCGGGTTC; Ac6, GGAATTCTGCAACCCTTCCCCTCC; A13, ATTGTGGATCCGCCCCTG. Reactions were heated at 94° for 4 min, then cycled 35 times at 94° for 20 sec, 60° for 30 sec, and 72° for 1 min per 1-kb length of expected PCR product, then 72° for 10 min. The rearrangement junction containing the 5′ Ac end in the P-ww-def1 allele was isolated by ligation-mediated PCR (LM-PCR) (Prod'homet al. 1998) as follows. Genomic DNA from plants of genotype P-ww-def1/P-wr was digested with SalI, ligated with SalI adaptor oligonucleotides (TCGCACTTCATTCAAGCTACTA and TCGATAGTAGCTTGAATGAA), and used as template in PCR amplification using primer Ac5 and SalI adapter primer. A single band that matched the expected size was obtained. (Genomic Southern analysis indicated that the rearrangement breakpoint in P-ww-def1 is located on a 7.9-kbp SalI fragment, visible in Figure 4B, lane 1. Because the 3′ flanking SalI site is located 3.0 kbp 3′ of the 4.6-kbp Ac element, the 5′ flanking SalI site should be located ~0.3 kbp 5′ of Ac in the P-ww-def1 allele.) The band amplified by LM-PCR was purified from an agarose gel and sequenced directly.
Schematic representation of structure of the progenitor P1-vv-9D9A allele (top) and the P1-ww-12:27-3 allele. Green arrows indicate the 5′ portion of the P1 gene and its associated upstream sequence. Red lines with arrow(s) indicate Ac or fAc, and the open and solid arrows indicate the 3′ and 5′ ends, respectively, of Ac/fAc. The short black line between Ac and fAc in P1-vv-9D9A indicates a 112-bp rearranged P1 sequence (rP) that is duplicated in P1-ww-12:27-3 (not to scale). The hatched box containing rP, fAc, and the 5′ portion of the P1 gene and its upstream sequence indicates the large insertion in P1-ww-12:27-3. The arrows below the DNA structure indicate the positions and orientation of PCR primers. The X at the junction of rP and the green arrow indicates the position of the 2-bp sequence change in P1-ww-12:27-3.
RESULTS AND DISCUSSION
Origin of a novel P1-ww allele with a large inverted duplication: The P1 gene regulates the synthesis of a red phlobaphene pigment in maize floral organs, including the pericarp (outermost layer of the kernel derived from the ovary wall) and the cob (Grotewoldet al. 1994). The two-letter suffix of P1 indicates its expression in pericarp and cob; i.e., P1-rr specifies red pericarp and red cob, P1-wr specifies white pericarp and red cob, and P1-ww specifies white pericarp and white cob (Anderson 1924). The standard P1-vv allele described by Emerson (1917) conditions variegated pericarp and variegated cob. The P1-vv allele contains an Ac transposable element inserted in intron 2 of a P1-rr gene (Lecheltet al. 1989). The P1-vv allele gave rise to the P1-ovov-1114 allele (conditions orange-variegated pericarp and cob) by intragenic transposition of Ac (Peterson 1990; Athmaet al. 1992); numerals placed after the two-letter suffix indicate the culture number of origin of each allele and alleles with the same phenotype but different culture numbers may have different gene structures.) The P1-ovov-1114 allele in turn gave rise to P1-vv-9D9A, which contains an Ac element, a 112-bp rearranged P1 gene fragment (rP), and a terminally deleted Ac element termed fAc (fractured Ac) inserted in intron 2 of P1-rr (Figure 1). The fAc element in P1-vv-9D9A contains the 2039 bp 3′ portion of Ac; a similar fAc element was described previously (Ralstonet al. 1989; Dooner and Belachew 1991). The intact Ac element can excise from P1-vv-9D9A to give a revertant allele with the P1-rr phenotype of red pericarp and red cob; this indicates that the rP and fAc insertions in P1-rr intron 2 do not interfere with P1-rr expression (J. Zhang and T. Peterson, unpublished results). From a large multikernel white pericarp sector on a P1-vv-9D9A/P1-ww ear, we isolated a novel P1-ww allele termed P1-ww-12:27-3.
Genomic Southern analysis (not shown), genomic cloning, and DNA sequencing indicate that P1-ww-12:27-3 contains a very large (>30 kb) insertion in P1-vv-9D9A at the junction of the 3′ end of fAc and the 5′ portion of the P1 gene (hatched box in Figure 1). The insertion is an inverted duplication derived from the P1-vv-9D9A sequence: from rP, it extends upstream through fAc and beyond the 5′ end of the P1 gene. The sequences at the junctions of the Ac/fAc termini and the P1 gene in P1-vv-9D9A and P1-ww-12:27-3 are identical except for a 2-bp change in P1-ww-12:27-3 at the junction of rP and the 5′ portion of the P1 gene (Figure 1). The sequence changes, A to T and C to G, are similar to a typical footprint created by Ac transposition (Rinehartet al. 1997), suggesting that the complex structure of P1-ww-12:27-3 probably resulted from an unusual transposition event.
Nonlinear transposition: On the basis of its inverted duplication structure and putative Ac transposition footprint, we propose that the P1-ww-12:27-3 allele arose by NLT (Figure 2; compare to models by Englishet al. 1993; Weil and Wessler 1993; Garyet al. 1996). The central feature of the model is that transposon termini located on different DNA molecules (sister chromatids are shown) can be utilized as transposase substrates. The resulting transposon (i.e., DNA internal to the Ac termini) is nonlinear and very large, in this case comprising the terminal ~70 cM of the short arm of maize chromosome 1 (Figure 2B). Upon excision, the sequences originally flanking the Ac/fAc termini join to form a chromatid bridge, and some minor sequence changes occur at the junction to create a typical transposon excision footprint (Figure 2B). Insertion of the nonlinear transposon at a target site in the chromatid bridge will generate two unequal chromatids: one containing an inverted duplication (P1-ww-id) and the other with a corresponding deficiency (P1-ww-def; Figure 2C). P1-ww-id contains a transposition footprint and a target site duplication (TSD), while P1-ww-def contains the other TSD. The inverted duplication structure of P1-ww-12: 27-3 and the position of the 2-bp sequence change (footprint) is exactly what would be predicted for a P1-ww-id allele produced by the nonlinear transposition model (compare Figure 1 and Figure 2C).
We did not detect, however, the P1-ww-def allele predicted as the reciprocal product of the NLT reaction, possibly because the cell clone containing the P1-ww-def allele gave rise to the nonheritable internal portion of the cob (Greenblatt 1985). Therefore, we initiated a search for the reciprocal products of a single nonlinear transposition event. Such reciprocal products could be detected and recovered in maize due to the cell lineage relationship between kernel pericarp and embryo (Greenblatt 1985). If nonlinear transposition events were to occur during ear development, the rearranged sister chromatids should segregate at mitosis into two daughter cells. Subsequent mitotic divisions of the daughter cells would generate a twinned sector, in which one twin carries the inverted duplication chromosome (P-ww-id) and the other twin carries the corresponding deficiency (P-ww-def). Because both deletion and insertion would destroy P1 gene function, both twinned alleles should specify colorless pericarp instead of the variegated pericarp specified by the progenitor allele P1-vv-9D9A. Following meiosis, each kernel in the twinned sector has a 50% chance to carry either P1-ww-id or P1-ww-def.
Nonlinear transposition model. The two lines indicate sister chromatids joined at the centromere (oval). All the symbols have the same meaning as in Figure 1. (A) Ac transposase (small ovals) binds to the 5′ terminus of Ac in one sister chromatid and the 3′ terminus of fAc in the other sister chromatid. (B) Cuts are made at the Ac and fAc termini. The two nontransposon ends join together to generate a chromatid bridge, and minor sequence changes occur at the junction to form the transposon footprint. (C) The excised transposon termini insert at the target site (the junction between the black line and the green arrow) to generate one sister chromatid (P1-ww-id) with an inverted duplication (green arrows), and a second sister chromatid with a corresponding deficiency (P1-ww-def). The P1-ww-id and P1-ww-def sister chromatids will segregate to adjacent daughter cells at the subsequent mitotic division, forming a potential twinned sector. The small boxes labeled a to f indicate the rearrangement junction sequences shown in Figure 5, including footprint, target site, and TSD. Note that in A and B, interactions between transposase molecules are not shown for clarity. In B, a fully excised transposon is depicted, but the transposition reaction could proceed through sequential cut and ligation steps in which no free intermediate is formed. In B and C, the outcome of insertion of the excised transposon at a target site on the top chromatid is depicted. Alternatively, insertion at a target site on the bottom chromatid could occur; this would give the opposite orientation of fAc and rP in P1-ww-id (not shown).
Ear with twinned sector produced from the cross: P1-vv-9D9A/P1-wr × P1-wr, r-m3::Ds. Both sides of a single ear are shown. The left side has a sector of kernels (bottom) with variegated pericarp and purple spotted aleurone containing the progenitor P1-vv-9D9A allele. The right side has a sector of kernels (bottom) with colorless pericarp and large purple aleurone spots, containing the P1-ww-def1 allele. Encompassing the entire upper portion of the ear is a large sector of kernels with colorless pericarp and tiny, barely visible purple aleurone spots; this sector contains the P1-ww-id1 allele. Genotypes were confirmed by molecular analysis as described in the text. The unequal sizes of the twinned P1-ww-id1 and P1-ww-def1 sectors is most likely an indirect result of the very early formation of the original twinned daughter cells. In general, larger twinned sectors tend to be more irregular and unequal in size than smaller twins, probably because twinned sectors that are formed early in development are more likely to be affected by differences in the subsequent growth and development of the daughter cell clones. As stated by Greenblatt (1985), irregularities can also result from the fact that pericarp twinned sectors are formed in the three-dimensional structure of the ear, whereas they are visible only at the surface of the ear. Kernels with colorless aleurone are of P1-wr genotype and lack Ac due to meiotic segregation; hence the R gene required for aleurone pigmentation remains nonfunctional due to Ds insertion.
To screen for the reciprocal products of a nonlinear transposition event, we crossed P1-vv-9D9A/P1-wr with P1-wr, r-m3::Ds pollen. The r-m3::Ds is an Ac tester allele: Ac-induced excision of Ds from the r locus gives rise to purple anthocyanin pigment in aleurone cell clones (Kermicle 1980). Among approximately 1500 ears screened, one ear had a large colorless pericarp sector in which the kernels within the sector were phenotypically twinned for anthocyanin pigmentation: one twin had kernels with large purple aleurone sectors, while the other twin had kernels with small purple aleurone sectors (Figure 3). Two alleles, P1-ww-def1 and P1-ww-id1, were recovered from the twinned sector. Like P1-ww-12:27-3, P1-ww-id1 exhibits a dominant delay in the developmental timing of Ac-induced Ds excisions from r-m3::Ds as evidenced by small purple aleurone sectors. The relationship between the delayed timing of Ac-induced Ds excisions and the structures of P1-ww-12: 27-3 and P1-ww-id1 is under investigation. In contrast, P1-ww-def1 exhibits normal Ac-induced Ds excisions as evidenced by large purple aleurone sectors. P1-ww-def1 is transmitted at normal frequencies through both pollen and ovum, but no homozygous P1-ww-def1 plants could be obtained.
Molecular analysis of twinned alleles: As predicted by the NLT model, Southern blot analysis indicated that both a P1 locus probe (fragment 15) and a P1-linked probe (p1.5B22, 3.5 cM from P1 locus) are deleted in P1-ww-def1 and duplicated in P1-ww-id1 (Figure 4). To test whether the P1-ww-id1 allele has an inverted duplication, we screened a genomic P1-ww-id1 library with both a P1 gene probe (probe 10′) and an Ac probe (1.6-kb Ac internal HindIII fragment). Eight clones hybridizing with both probes were obtained, and these were grouped into two types: type I (five clones) contains fAc and the 5′ portion of the P1 gene, whereas type II (three clones) contains Ac, fAc, and the 3′ portion of the P1 gene. No clones contained both 5′ and 3′ portions of the P1 gene. This result is predicted by the nonlinear transposition model because the large insertion in P1-ww-id1 separates the 5′ and 3′ portions of the P1 gene. Southern blot and sequence analysis of the P1-ww-id1 clones indicate that P1-ww-id1, like P1-ww-12:27-3, contains an inverted duplication that begins at rP and extends beyond the 5′ end of the P1 gene. However, the P1-ww-id1 duplication extends beyond the distal endpoint of the P1-ww-12:27-3 duplication.
We tested several additional predictions of the NLT model. The P1-ww-id1 allele should contain an Ac-type footprint at the junction of rP and the 5′ portion of the P1 gene (Figure 2C); such a footprint was identified by PCR as follows. In P1-vv-9D9A, oligonucleotide primers A13 and Ac6 cannot produce a PCR product because the Ac6-homologous sequence in fAc has the same orientation as that of A13, and Ac6 in Ac is 5.2 kb from A13 in P1-vv-9D9A (Figure 1). However, following nonlinear transposition, Ac6 in the P1-ww-id insertion lies opposite to A13, and the distance between them is 1.4 kb (Figure 1). PCR amplification produces the predicted product from P1-ww-id1, but not from P1-vv-9D9A or P1-ww-def1/P1-wr (data not shown). Compared to the sequences of the P1-vv-9D9A progenitor allele (Figure 5, A and B), the P1-ww-id1 allele contains a typical Ac-type footprint precisely at the junction of rP and the 5′ portion of the P1 gene (Figure 5C); 2 bp are changed (A to T and C to G).
Structural analysis of the twinned sector alleles. (A) Restriction map of P1-vv-9D9A. The backbone is P1-rr, the solid boxes indicate exons, the numbered boxes indicate hybridization probes, and the hatched boxes indicate sequences homologous to probe 15. S, SalI; S*, methylated SalI site. Insertion of fractured Ac (fAc, smallest triangle), rP (small gray rectangle), and Ac (medium triangle) within intron 2 of P1-rr gives rise to P1-vv-9D9A; the 5′ and 3′ ends of Ac/fAc are indicated. (B) Genomic DNA was digested with SalI and hybridized with probe 15. Lane 1, P1-ww-def1/P1-wr; lane 2, P1-vv-9D9A/P1-wr; lane 3, P1-ww-id1/P1-wr. The 12.6- and 1-kb bands are derived from the P1-wr allele (Chopra et al. 1996, 1998), and the 1-kb band is a single copy sequence used as internal control for DNA loading. The similar intensity of the 1-kb band in lanes 1, 2, and 3 indicates that these lanes contain equivalent amounts of DNA. The location of the 3.0- and 1.2-kb bands is shown in A (see also Athma and Peterson 1991). In the P1-ww-def1/P1-wr genotype (lane 1), the 3.0-kb band is missing, and the 1.2-kb band is less intense than in the progenitor P1-vv-9D9A/P1-wr genotype (lane 2). In contrast, the same 1.2- and 3.0-kb bands are more intense in the P1-ww-id1/P1-wr genotype (lane 3). Lanes 1, 2, and 3 contain bands at 7.9, 10, and ~12 kb, respectively. These bands arise from cutting at the SalI site at the 3′ boundary of probe fragment 15 and at SalI sites upstream of the Ac and/or fAc insertion in the P1-ww-def1, P1-vv-9D9A, and P1-ww-id1 alleles. The SalI site in fragment 10′ is shown for the P1-vv-9D9A allele in A. This SalI site is removed by the rearrangements in the P1-ww-def1 and P1-ww-id1 alleles and replaced by other SalI sites that alter the sizes of the corresponding fragments. (C) Genomic DNA was digested with the indicated enzymes and hybridized with p1.5B22, a probe located 3.5 cM from the P1 locus. The loading is the same as in B (same preparation and amount of DNA). The bands in P1-ww-def1/P1-wr genotype (lane 1) are the weakest, and the bands in P1-ww-id1/P1-wr genotype (lane 3) are the strongest. The signal intensities reflect the copy number of the probe sequence: the P1-wr haplotype contains one copy of the probe sequence, whereas P1-ww-def1, P1-vv-9D9A, and P1-ww-id1 contain 0, 1, and 2 copies, respectively, of the probe sequence.
Upon transposition, Ac elements generate an 8-bp TSD of the host sequence at the insertion site. The NLT model predicts that an 8-bp TSD will likewise be formed at the transposon insertion site; however, because the NLT transposon consists of two sister chromatids, the TSD should be found at the rearrangement breakpoints of the chromosomes carrying the twinned alleles. In the P1-ww-id1 allele, the TSD should be adjacent to the fAc; in the P1-ww-def1 allele, the complementary 8-bp sequence should be found adjacent to the Ac 5′ end (Figure 2C). We subcloned and sequenced the DNA fragment containing the 3′ end of fAc from P1-ww-id1 type II clones. The sequence of the suspected TSD adjacent to the 3′ end of fAc is AGCGAGGC (Figure 5D). We cloned the DNA fragment containing the suspected TSD in P1-ww-def1 by modified LM-PCR (Prod'homet al. 1998). The sequence of the PCR product contains the expected TSD (GCCTCGCT, the complementary sequence of AGCGAGGC) at the junction of the Ac 5′ end and the rearrangement breakpoint in P1-ww-def1 (Figure 5E). The presence of the identical TSD at the Ac 5′ end in P1-ww-def1 and the fAc 3′ end in the P1-ww-id1 strongly supports the hypothesis that the rearrangements were generated by a single Ac transposition event.
Footprint and target site duplications of the P1-ww-def1 and P1-ww-id1 alleles created by nonlinear transposition. The sequences from A to F correspond to the sequences from boxes a to f in Figure 2; the color and orientation of the sequence letters match the color and orientation of lines in Figure 2. The footprint and target site duplication sequences are underlined. See text for further details. (A) Junction of rP and Ac 5′ terminus in P1-vv-9D9A. (B) Junction of 5′ portion of P1 gene and the 3′ terminus of fAc in P1-vv-9D9A. (C) Junction of 5′ portion of P1 gene and rP in P1-ww-id1 containing a 2-bp transposon footprint. (D) Junction of inverted duplication and fAc 3′ end in P1-ww-id1. (E) Junction of deletion endpoint and Ac 5′ end in P1-ww-def1. (F) Original target site sequence in P1-vv-9D9A.
According to the nonlinear transposition model, the endpoints of the rearrangements in P1-ww-def1 and P1-ww-id1 should be adjacent to each other in the progenitor allele P1-vv-9D9A. We designed primers near each endpoint and used this primer pair to PCR amplify the genomic sequence from the P1-vv-9D9A template. The size of the PCR product (240 bp) matched the size inferred from the primer positions (data not shown). The PCR product contains a single copy of the GCCTCGCT target site, and the sequences flanking GCCTCGCT are the same as those from the endpoints of P1-ww-id1 and P1-ww-def1 (Figure 5F). These results show that the rearrangement breakpoints identified in the P1-ww-def1 and P1-ww-id1 alleles are derived from insertion of Ac transposon ends into the GCCTCGCT target site in the progenitor chromosome.
The P1-ww-12:27-3 and P1-ww-id1 alleles both contain inverted duplications that begin at the rP in the P1-vv-9D9A allele and extend upstream beyond the 5′ end of the P1 gene. According to the NLT model, the duplications should end at the transposon insertion site located in the chromatid bridge (Figure 2B). Indeed, restriction fragment length polymorphism mapping shows that the endpoints of the inverted duplications in P1-ww-12:27-3 (p1.5B22) and P1-ww-id1 (pJZPX) map 3.5 and 4.6 cM, respectively, proximal to the P1 locus, in the order: P1 3′ end, P1 5′ end, p1.5B22, pJZPX, centromere (M. McMullen and T. Musket, personal communication).
Relative frequencies of normal and nonlinear transposition: McClintock (1949) described two alternative states of Ds elements: state I, which produces frequent chromosome breakage and rearrangement events, and state II, in which breakage events are rare, and reversions are frequent. McClintock's original chromosome-breaking state I Ds has since been associated with the compound double Ds, a structure in which one Ds element is inserted in reverse orientation into a second identical Ds copy (Döringet al. 1990; Martínez-Férez and Dooner 1997). The inverse relationship between the frequencies of chromosome breakage vs. normal excision has also been observed in tobacco plants containing engineered Ds constructs: double Ds elements promote chromosome breakage at high frequencies, but their rates of excision are much reduced compared to that of simple Ds elements. Thus, the presence of the double Ds configuration appears to inhibit simple excision of the individual Ds subunits (Englishet al. 1993). The reason for this is unclear, but it has been proposed that directly repeated 3′ and 5′ Ds ends are preferred to ends in normal orientation as substrates of Ac transposase when they are present together in double Ds elements (Englishet al. 1993). In the case of the P1-vv-9D9A allele, simple excision of the intact Ac element is easily recognized by the occurrence of red revertant sectors and germinal P1-rr revertants. The pattern of variegation (red stripes) given by the P1-vv-9D9A allele is very similar to that of the standard P1-vv allele that contains a single Ac insertion in the P1-rr gene. Thus, simple excision of Ac is not noticeably inhibited by a nearby fAc element in the P1-vv-9D9A allele, even though this allele contains directly repeated 3′ and 5′ termini in the same configuration as the natural (Döringet al. 1990) and engineered (Englishet al. 1993) double Ds elements.
In contrast, nonlinear transpositions of the Ac/fAc elements in P1-vv-9D9A will fragment the P1 gene and generate colorless pericarp sectors. Among 1500 ears carrying the P1-vv-9D9A allele, we obtained 15 ears with large multikernel colorless pericarp sectors, which gave rise to rearranged alleles characteristic of the nonlinear transposition reaction (J. Zhang, P. Zhang and T. Peterson, unpublished results). This is an underestimate of the actual frequency of NLT events for several reasons: first, we selected for study only large, easily recognized multikernel sectors. Colorless sectors smaller than one kernel in size are difficult to distinguish from the background of variegated pericarp, which itself is a mosaic of red stripes on a colorless pericarp background. Moreover, some of the NLT events would be predicted to be inviable and hence would not have been analyzed. Nevertheless, for the P1-vv-9D9A allele, the nonlinear transposition events appear to be much rarer than the frequency of simple Ac excision. Further research will be required to determine the parameters that influence the propensity of individual Ac/Ds termini to participate in normal or aberrant transposition reactions.
Significance of NLT-induced rearrangements: A number of physical and genetic agents can induce deletions in plants at random sites. In contrast, sister chromatid NLT has the unique property of producing deletions that extend from a single site in the genome (in this case the P1 locus) to multiple flanking sites. The deletion endpoints will represent the insertion sites of the nonlinear transposon; in the case of Ac-mediated transposition, these sites will likely be relatively close due to the tendency of Ac to transpose to nearby locations (Greenblatt 1984). The resulting nested deletions can be used to rapidly map molecular markers in a relatively small genetic interval (J. Zhang, P. Zhang and T. Peterson, unpublished results), a process that is difficult by standard meiotic mapping due to the limited resolution of most mapping populations. Generation of deletions by sister chromatid transposition could be extended to sites throughout the genome by transformation with transgene constructs containing Ac termini and a reporter gene whose loss is easily detected. Additionally, NLT events can generate duplications of varying sizes that may be useful for studying the effects of gene dosage on expression levels (Guoet al. 1996).
The NLT model predicts that insertion of a nonlinear transposon into a target site on another chromosome would generate other chromosomal rearrangements including translocations, acentric fragments, and dicentric chromosomes. Thus, nonlinear transposition could have contributed to the major genome rearrangements observed between related species (Bennetzen and Freeling 1993; Bennetzenet al. 1998). Moreover, because chromosome rearrangements can often lead to semisterility among progeny heterozygous with the progenitor genotype, NLT could lead to reproductive isolation and thus be an important first step in speciation (Lewis 1966). As we have shown here, NLT events generate characteristic chromosomal structures and sequences at the rearrangement junction. These molecular signs of NLT events may yet be recognizable in the genome sequences of closely related species.
Acknowledgments
We thank Suzy Cocciolone and Dan Voytas for helpful comments on the manuscript and T. Musket and M. McMullen for mapping of pJZPX and p1.5B22. This is journal paper no. J18497 of the Iowa Agriculture and Home Economics Experiment Station, Ames, Iowa, project no. 3297, supported by Hatch Act and State of Iowa funds.
Footnotes
-
Communicating editor: J. A. Birchler
- Received January 26, 1999.
- Accepted July 6, 1999.
- Copyright © 1999 by the Genetics Society of America
