Genetic stocks:

Alleles of the p1 gene are identified by a two-letter suffix that indicates their expression pattern in pericarp and cob: e.g., P1-rr (red pericarp and red cob); P1-wr (white pericarp, red cob); and p1-ww (white pericarp and white cob). The standard p1-vv (variegated pericarp and variegated cob) allele described by Emerson (1917) contains an Ac insertion in the second intron of a P1-rr gene. From p1-vv, we obtained a spontaneous derivative termed P1-ovov1114 (orange-variegated pericarp and orange-variegated cob), in which the Ac element had undergone an intragenic transposition to a site 153 bp upstream in p1 gene intron 2 and inserted in the opposite orientation (Peterson 1990). From P1-ovov1114, we obtained a spontaneous derivative termed p1-vv9D9A (Zhang and Peterson 1999); this allele contains an Ac element, a 112-bp rearranged p1 gene fragment (rP), and the terminally deleted Ac element fAc. From p1-vv9D9A, we obtained the allele P1-rr11 described here; this allele was generated from p1-vv9D9A by excision and transposition of the full-length Ac element from its location in intron 2 of p1 to a site ∼9 kb upstream of the p1 transcription start site (Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.—

Gene structure and phenotypes of P1-rr11 and derivative alleles. (A) Restriction map of the p1 gene and surrounding sequences in the P1-rr11 allele. The solid boxes are p1 gene exons 1, 2, and 3 (left to right). Red triangles indicate Ac or fAc insertions, and the open and solid arrowheads indicate the 3′ and 5′ ends, respectively, of Ac/fAc. The positions of fragments 8B and 15 used as probes are indicated by the numbered boxes. Additional sequences homologous to 15 are indicated as hatched boxes. The short horizontal arrows indicate the orientations and approximate position of PCR primers. Primers are identified by numbers above or below the arrows; primer prefixes used in the text (p1, Ac) are omitted due to space constraints. K, KpnI; H, HindIII. Lines below the map indicate the restriction fragments produced by digestion with KpnI or HindIII and hybridizing with the indicated probe; asterisks indicate HindIII restriction sites located within Ac or fAc sequences. (B) The kernel pigmentation phenotype of P1-rr11 and its derivatives. Typical representative kernels are shown for P1-rr11 and p1-ww21. For p1-vv22 and p1-vv30, most kernels have colorless pericarp; only a small fraction of kernels have red sectors as shown here. Kernels shown are from plants in which the indicated allele is heterozygous with a standard recessive p1-ww allele for colorless pericarp and cob.

Genomic DNA extractions and Southern blot hybridization:

Total genomic DNA was prepared from husk using a modified cetyltrimethylammonium bromide extraction protocol (Saghai-Maroof et al. 1984). Agarose gel electrophoresis and Southern hybridizations were performed as described (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.

PCR amplifications:

PCR amplifications were performed as described (Saiki 1989) using the oligonucleotide primers shown in Table 1.

Reactions were heated at 94° for 3 min and then cycled 35 times at 94° for 20 sec, 60° for 30 sec, and 72° for 1 min/1 kb length of expected PCR product and then at 72° for 8 min. The band amplified was purified from an agarose gel and sequenced directly. Sequencing was done by the DNA Synthesis and Sequencing Facility, Iowa State University.

The P1-rr11 allele and its derivatives:

We isolated a p1 allele designated P1-rr11 (red pericarp, red cob), which contains the truncated Ac element fAc, inserted in the second intron of p1, and a full-length Ac element, inserted in the p1 5′ flanking sequences, 13,175 bp upstream of the fAc element (materials and methods). In P1-rr11, the 5′ end of Ac and the 3′ end of fAc are oriented toward each other (Figure 1A). P1-rr11 exhibits unstable kernel pigmentation, as evidenced by frequent colorless pericarp sectors and colorless kernels (Figure 1B). Because the kernel pericarp and egg cell are derived from a common cell lineage, sporophytic mutations that give rise to sectors of colorless pericarp can be germinally transmitted through the kernel embryo (Greenblatt 1985). Colorless and light-colored kernels were selected from 20 ears of genotype P1-rr11/P1-wr, and from these we obtained 10 new p1-ww (white pericarp, white cob) alleles and two new p1-vv (variegated pericarp, variegated cob) alleles (Figure 1B).

The structures of the alleles derived from P1-rr11 were analyzed by Southern blot hybridization using p1 genomic fragments 15 and 8B as probes (Figure 2). Probe 15 detects an enhancer-associated fragment and related sequences located at several sites upstream and downstream of the p1 gene (Figure 1A), whereas probe 8B detects the p1 coding sequence and also cross-hybridizes with the coding sequence of a linked paralagous gene (p2) that is located >20 kb upstream of the p1 gene (Zhang et al. 2000, 2003). The results of the KpnI-15 blot shows that significant rearrangements occurred in all the alleles: while the progenitor allele P1-rr11 gives a doublet band of 10.9 and 10.7 kb, the 12 derivative alleles show a different size band or no hybridization with the probe. Moreover, the KpnI-8B, HindIII-15, and HindIII-8B blots show that p1 sequences located between the Ac and fAc termini have been deleted in the 10 new p1-ww alleles, while in the 2 p1-vv alleles (p1-vv22 and p1-vv30) the probe detects bands of altered size (see Generation of local rearrangements). In contrast, bands derived from the linked p2 gene are not altered in the derivative alleles with the exception of the allele p1-ww25 (see Generation of deletions).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.—

Genomic Southern blot analysis of P1-rr11 and its derivatives. Genomic DNA was prepared from plants homozygous for the indicated alleles, cut with KpnI and HindIII, and hybridized with probes 15 and 8B from the p1 gene. In the KpnI-15 blot, the band in P1-rr11 is a doublet that contains fragments derived from the probe 15-homologous sequences located 5′ and 3′ of the p1 gene (Figure 1). The common band in KpnI-8B and HindIII-8B blots arises from cross-hybridization with a 8B-homologous sequence in the p2 gene (Zhang et al. 2000).

Reversed Ac end transposition model:

The high frequency of p1 sequence deletion observed is consistent with a model for transposition reactions involving Ac ends in reversed orientation (Figure 3; see animation available at http://www.genetics.org/supplemental/). In the case of P1-rr11, the 5′ end of the intact Ac element and the 3′ end of the fAc element are transposition substrates (Figure 3A). Transposon excision generally is accompanied by rejoining of the flanking host DNA; in P1-rr11, joining of the sites flanking the 5′ end of the intact Ac element and the 3′ end of the fAc element would form a 13-kb circle (Figure 3B). This type of small extrachromosomal circle would ordinarily be lost, resulting in deletion of p1 sequences including fragment 8B, as was observed for the 10 p1-ww alleles by genomic Southern blot analysis (Figure 2). However, if the transposon ends inserted into a site in the 13-kb circle, a linear chromosome would be regenerated and the p1 sequences between the reversed Ac termini would be rearranged (Figure 3C), whereas insertion of the transposon ends into sites in the flanking DNA would generate either inversions (Figure 3D) or deletions (Figure 3E), depending on the orientation with which insertion into the target site occurs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.—

Reversed-Ac-ends transposition model (see also animation at http://www.genetics.org/supplemental/). Solid boxes indicate p1 gene exons 1, 2, and 3 (left to right). Red arrows indicate Ac or fAc; the open and solid arrowheads indicate the 3′ and 5′ ends, respectively, of Ac/fAc. Ovals indicate Ac transposase. Short vertical lines indicate transposon reinsertion sites, and solid triangles indicate TSD. (A) Ac transposase cleaves at the 5′ end of Ac and the 3′ end of fAc. (B) Following transposase cleavage at the junctions of Ac/p1 and fAc/p1, the internal p1 genomic sequences are joined to form a 13-kb circle. The “×” on the circle indicates the site where the joining occurred, marked by a transposon footprint. The Ac 5′ and fAc 3′ ends are competent for insertion anywhere in the genome. (C–E) The structures expected from insertion into two possible target sites. (C) The transposon ends insert into the 13-kb circle. The Ac 5′ end joins to the end adjacent to exon 1 and fAc 3′ end joins to the other end. The 13-kb sequence is rearranged (segment b–c and segment d exchanged positions). (D) The transposon ends insert distal to sequence e: the Ac 5′ end joins to the end adjacent to e, and the fAc 3′ end joins to the other end. Segment e is inverted, and the 13-kb circle is lost. The resulting chromosome contains an inversion of sequences from the fAc 3′ end to the insertion site distal to e. (E) The transposon ends insert distal to sequence e: the fAc 3′ end joins to the end adjacent to e, and the Ac 5′ end joins to the other end. Segment e is circularized and presumed lost; the 13-kb circle is also lost. The resulting chromosome contains an interstitial deletion from the Ac 5′ end to the insertion site distal to e.

It is important to distinguish whether chromosomal rearrangements originate from transposition per se or from other host-encoded repair and/or recombination mechanisms. The reversed-ends transposition model predicts that rearrangement breakpoints should be bounded by transposon footprint and target site duplications (TSD) that are characteristic of transposase-mediated events. Ideally, the sequences of the footprint and TSD should be identified for the products of a single transposition reaction. The deletions are not useful for this purpose because they carry only one TSD sequence [the other TSD sequence is associated with the circularized segment that would most likely be lost (Figure 3E)], whereas the inversions contain both TSD sequences, but no footprints (Figure 3D). In contrast, a local rearrangement produced by insertion of the transposon ends into the 13-kb excised circle would contain both the TSDs and the transposon footprint generated in a single transposition event (Figure 3C). In the following sections we present the evidence for the formation of each of these products through transposition of reversed Ac termini.

Generation of local rearrangements:

Genomic Southern blot analysis of the p1-vv22 and p1-vv30 alleles derived from P1-rr11 indicates that the p1 probe sequences are rearranged, but not deleted, in these alleles (Figure 2). According to the reverse-ends transposition model, excision of transposon ends would be accompanied by joining together the ∼13 kb of genomic sequences between the reversed 5′ and 3′ Ac termini (Figure 3B). The new junction should contain a characteristic Ac footprint. Therefore, we PCR amplified genomic DNA from P1-rr11 and derivative alleles using primers p1-6 and p1-13, which are complementary to p1 sequences flanking the Ac 5′ and the fAc 3′ termini (see Figure 1A for the orientations and approximate positions of the PCR primers). Importantly, the primers are in divergent orientations in the progenitor allele P1-rr11. As expected, no PCR products were obtained from P1-rr11 or p1-ww alleles containing p1 deletions; however, p1-vv22 and p1-vv30 produced PCR products of the correct size (510 bp). Direct sequencing of the PCR products showed that the genomic sequences flanking the Ac 5′ and 3′ termini in P1-rr11 were indeed joined together as predicted and that both junctions contained typical Ac footprints: a C nucleotide from each side of the junction was changed to G (Figure 4, B and C). Next, we identified the target site duplications from these two alleles. The results of Southern blot analysis (Figure 2) indicated the approximate transposon insertion sites for each allele; we then performed PCR using primers complementary to the Ac termini and the p1 genomic sequences around the suspected integration sites to amplify the Ac/p1 junction fragments in each mutant. For p1-vv22, the junction with the Ac 5′ end was PCR amplified using primer pair p1-7 + Ac-4. The PCR product was sequenced and found to contain the Ac 5′ end joined to p1 genomic sequences located 4552 bp upstream of the transcription start site. The p1 sequences are in an inverted orientation relative to their original genomic context (Figure 4B). The junction with the 3′ fAc end was amplified using primers p1-8 + Ac-2 and found to contain the 3′ end of fAc joined to p1 sequences beginning immediately downstream of the site in the p1 genomic DNA where the Ac 5′ end had inserted. Moreover, both junctions of the 5′ Ac and 3′ fAc termini are bounded by an 8-bp sequence (CAGAGTAT), which is found once at the original insertion site (Figure 4, A and B). This represents the 8-bp TSD characteristic of Ac insertions. For p1-vv30, we used nested primer pairs, p1-12 + Ac-4 and p1-11 + Ac-5, to amplify the new p1-Ac 5′ junction and p1-9 + Ac-3 and p1-10 + Ac-2 to amplify the new p1-fAc 3′ junction. Sequences of the PCR products show that the Ac 5′ end and fAc 3′ end are inserted into a site located 22 bp upstream of the p1 transcription start site. Additionally, both Ac/p1 junctions are bounded by the sequence CCGGCCGT, which represents an 8-bp TSD (Figure 4C). These results demonstrate that the p1-vv22 and p1-vv30 alleles were indeed generated by transposition reactions involving reversed Ac ends. The results are consistent with an intramolecular transposition model in which transposon excision produced a covalently closed circle composed of the 13-kb DNA between the reversed transposon ends, followed by insertion of the Ac 5′ and fAc 3′ ends into the circle. The resulting chromosome carries a rearrangement in which the 13-kb sequences between the Ac 5′ and fAc 3′ ends are circularly permuted. Depending upon the direction in which the transposon ends insert into the circle, the intervening DNA can be in the inverted or direct orientation, as for p1-vv22 and p1-vv30, respectively. It is interesting to note that in both the p1-vv22 and p1-vv30 alleles, the 5′ end of Ac and the 3′ end of fAc are still competent for transposition. It is possible that in further reversed-ends transposition events, the transposon ends could reinsert into the p1 intron 2 sequences in the correct orientation so as to restore p1 function. Consistent with this idea, the p1-vv22 and p1-vv30 alleles are distinctive in exhibiting occasional red sectors that are not observed with the p1-ww alleles (Figure 1B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.—

Schematic structures of the p1 region in P1-rr11 and derivative alleles. (A) P1-rr11. Arrows in a–f indicate the various genomic segments of the p1 gene. Boxes indicate the positions of the given sequences. (B) p1-vv22. Genomic segments b, c, and d have changed position and/or orientation. (C) p1-vv30. Genomic segments b, c, and d have changed position. (D) p1-ww21. Genomic segments b, c, and d were deleted; fAc and genomic segment e are inverted.

Generation of inversions:

The reversed Ac ends transposition model (Figure 3) predicts that insertion of the transposon ends into target sites in the flanking chromosomal regions would produce deletions and inversions. Southern blot analysis (Figure 2) suggested that the p1-ww21 allele could be a small inversion caused by insertion of the transposon ends into a site a short distance downstream of the p1 gene. To test this prediction, primer pairs p1-14 + Ac-4 and p1-15 + Ac-2 were used to amplify the new Ac-p1 and fAc-p1 junctions, respectively. Direct sequencing of the PCR products demonstrated that the insertion site is 496 bp downstream of p1 exon 3 (Figure 4D); both insertion junctions are bounded by an 8-bp TSD (CATATCTT). Moreover, both the fAc and the 3′ flanking p1 sequences (segment e in Figure 4D) are inverted in orientation with respect to the Ac 5′ end and sequences downstream of the insertion site (segment f in Figure 4D). These data, together with the Southern blot data indicating that p1 fragment 8B is deleted (Figure 2, KpnI-8B, and HindIII-8B), indicate that p1-ww21 has a deletion and inversion as shown in Figure 4D. The rearrangement disrupted p1 fragment 15, consistent with the HindIII-15 Southern blot results (Figure 2) of a 4-kb band (contains 281-bp p1 fragment 15 and the downstream sequence homologous to fragment 15) and a faint 5.5-kb band (contains only 136 bp of fragment 15 sequence).

We isolated six additional p1-ww alleles (p1-ww23, p1-ww24, p1-ww26, p1-ww31, p1-ww33, and p1-ww35), which are deleted for the sequences between Ac and fAc (Figure 2, KpnI-8B). Each of these alleles exhibits a new, different-sized band derived from the junction of 3′ fAc with the flanking DNA (Figure 2, KpnI-15). However, all of these alleles retain the 6.5-kb HindIII fragment derived from HindIII sites in fAc and the p1 3′ flanking sequence (HindIII-15 blot in Figure 2). Additionally, PCR analysis using primer pair p1-1 + Ac-2 shows that the junction of the 3′ end of Ac and the p1 5′ flanking sequence is intact (not shown). The simplest explanation of these results is that these alleles carry inversions with endpoints outside of the p1 locus to either side. As shown in Figure 3, inversions of this type would retain the original junctions of the Ac 3′ end and fAc with the outside flanking DNA, while the Ac 5′ end and the fAc 3′ end would be joined to new sequences. Similar structural data could also be obtained from reciprocal translocations; however, these alleles do not exhibit a semisterile phenotype as would be expected for translocation heterozygotes.

Generation of deletions:

As shown in Figure 3, the reversed-ends transposition model predicts that flanking deletions would be generated as a consequence of insertion into a flanking target site. We identified three cases of deletions that extend to sites outside of the Ac or fAc insertion sites. In p1-ww2052 and p1-ww2053, p1 fragment 15 and p1 fragment 8B are deleted, but the 8B-homologous sequence in the upstream paralagous gene p2 is intact (Figure 2; Zhang et al. 2000). PCR analysis using primer pair p1-1 + Ac-2 shows that the junction of the 3′ end of Ac and the p1 5′ flanking sequence is intact (not shown). Therefore, these alleles most likely contain deletions caused by insertion of the 5′ Ac end at a site downstream of the p1 fragment 15 homologous sequence. In p1-ww25, both p1 fragment 8B and the p2 sequence homologous to 8B are deleted (Figure 2), whereas sequences on the 3′ side of p1 are intact, as evidenced by the 6.5-kb HindIII fragment detected by p1 fragment 15. Therefore, this allele most likely contains a deletion caused by insertion of the 3′ fAc end into a site upstream of the p2 gene. This result is consistent with our observation that p1-ww25 lacks the silk-browning phenotype that is conferred by the presence of either p1 or p2 genes (Zhang et al. 2003).

Frequency and possible mechanism of reversed-ends transposition events:

The cases we isolated were derived from an allele (P1-rr11) that contains a fractured Ac element (fAc) inserted into the p1 gene intron 2 and an intact Ac element located ∼9 kb upstream of the p1 transcription start site. Neither the Ac nor the fAc elements in P1-rr11 disrupt p1 function; however, transposition reactions involving the 5′ end of the intact Ac and the 3′ end of the fAc will result in deletion or rearrangement of the p1 gene promoter, exon 1, and exon 2 and thus generate colorless loss-of-function sectors on the otherwise red kernel pericarp. These sectors appear at a frequency of ∼1 per kernel, depending on genetic background. The size of each sector reflects the time during development at which mutation occurs and ranges from very small (<1 mm) to large multikernel sectors. In contrast, simple transposition of the upstream intact Ac element is expected to occur more frequently (∼10 sectors/kernel if similar to the transposition frequency of the p1-vv allele); however, excision of the upstream Ac element would not affect p1 expression and thus would be undetectable in our screen. Further, the fAc element in p1 intron 2 containing only the 3′ end of Ac is not competent to transpose, nor does it induce any instability in p1 expression either alone or in the presence of an Ac element on another chromosome (not shown).

The 5′ end of the upstream intact Ac element could participate in transposition reactions with either the 3′ end of the same element or the 3′ end contributed by the downstream fAc element. It is not clear what determines which 3′ end is used in the transposition reaction. It has been proposed that assembly of Ac/Ds transposition complexes occurs during chromosomal DNA replication such that assembly is initiated with the first transposon end to be replicated and then completed using the next-replicated transposition-competent end (Xiao and Peterson 2002). According to this model, replication of P1-rr11 in the distal to proximal direction (from 3′ to 5′ with respect to the p1 gene) would replicate the 3′ end of fAc first, followed by the 5′ end of the intact upstream Ac element; these two ends would then interact and undergo reversed-ends transposition. However, replication in the proximal-to-distal direction (5′ to 3′ with respect to the p1 gene) would result in assembly of a transposition complex using the 3′ and 5′ ends of the intact upstream Ac element, leading to standard transposition of the upstream Ac element. This model is consistent with recent evidence that transposition is greatly enhanced by DNA replication (McElroy et al. 1997; Wirtz et al. 1997) and that competence for transposition is determined by differential binding of transposase to specific hemi-methylated strands of the Ac termini (Ros and Kunze 2001). However, this model is not consistent with the observation that a pair of closely linked Ds insertions at the maize Waxy locus can undergo alternative transposition reactions resulting in chromosome breakage (Weil and Wessler 1993).

It is notable that 2 of the 12 cases studied here contained insertions of the Ac and fAc termini into the 13-kb sequence located between the transposon endpoints. This seems to be an unusually high frequency, even considering the well-known tendency of Ac to transpose to linked sites (Greenblatt 1984). Possibly, the excised circle formed upon reversed-ends transposition is topologically linked with the chromatid from which it was excised and thus may be held in close proximity to the transposon ends, making it a preferred target for reinsertion. Circular extrachromosomal Ac and Ds molecules have been described previously (Gorbunova and Levy 1997, 2000), although their origin and possible relation to the circular molecules discussed here is unclear.

Potential role of unconventional transposition in genome evolution:

Numerous examples have shown that transposition can be accompanied by a variety of structural rearrangements. For example, McClintock described two cases in which transposition of Ds was associated with duplication of large segments of the short arm of maize chromosome 9 (McClintock 1951). Chromosome-breaking structures in maize typically contain multiple Ac and/or Ds termini; chromosome breakage has been proposed to occur through the transposition of a partially replicated macrotransposon (Ralston et al. 1989) or following fusion of sister chromatids in alternative transposition reactions (Weil and Wessler 1993). Support for the latter model has been obtained from analysis of transgenic tobacco showing that a pair of Ac/Ds 5′ and 3′ ends in direct orientation is capable of chromosome breakage (English et al. 1993, 1995) as a consequence of transposition reactions that act upon 5′ or 3′ Ac ends on different sister chromatids, leading to sister chromatid fusion. We previously presented evidence from twin sector analysis for a similar type of mechanism, termed sister chromatid transposition (previously termed nonlinear transposition; Zhang and Peterson 1999). In Drosophila, transposition reactions involving termini of P elements on different sister chromatids or homologous chromosomes can generate deletions and duplications (Gray 2000; Gray et al. 1996; Preston et al. 1996). Additionally, transposition involving termini of different Tc1-mariner elements in direct orientation generates deletions and inversions in Fusarium (Hua-Van et al. 2002). In contrast, the mechanism described here involves transposition of Ac ends in reversed orientation. To our knowledge, this configuration of transposon ends has not been directly tested for a eukaryotic transposon.

The rearrangements we describe here were generated through unconventional transposition reactions utilizing as substrates a pair of reversed Ac ends located 13 kb apart. Whenever two intact Ac/Ds elements are located on the same molecule in direct orientation, their terminal sequences adjacent to each other are in reversed orientation. This configuration can occur naturally when Ac transposes during DNA replication from a site that is already replicated to a closely linked but not-yet-replicated site. This mode of transposition generates genetically distinct daughter chromosomes, leading to the formation of twinned sectors, and is commonly observed for Ac transposition from the p1-vv allele (Greenblatt 1984). Similarly, Ac/Ds transpositions from the bz1 locus on maize chromosome 9s can generate pairs of linked elements, and these have been shown to induce chromosome breakage (Dooner and Belachew 1991). However, it is not yet known whether the chromosome 9 breaks are caused by reversed-ends transposition or an alternative model involving transposition of a partially replicated macrotransposon (Ralston et al. 1989). In addition, transformation of plants with Ac/Ds-containing transgenes via particle-bombardment or Agrobacterium-mediated methods often results in multiple copies integrated in close proximity (Frame et al. 2002). It is possible that mutations derived from these types of structures may include rearrangements of the type described here.

The Ac/Ds elements are members of the hAT family of eukaryotic transposons, which are widespread in the animal and plant kingdoms (Kunze and Weil 2002). Due to the similarity in the basic biochemical features of the hAT transposition reaction, it seems likely that reversed termini of other hAT family members, and possibly other transposons, are able to transpose in maize and other organisms. This activity could, during the course of evolution, generate a variety of large-scale chromosomal rearrangements. Some of these, such as inversions and translocations, commonly result in partial sterility and may thereby represent an initial step in reproductive isolation and speciation (Lewis 1966). Thus, our results suggest that transposon-induced chromosomal rearrangements could play an important role in macroevolution and the origin of new species. Recent reports indicating that rearrangement breakpoints are often associated with repetitive sequences are consistent with this hypothesis (Evgen'ev et al. 2000; Coghlan and Wolfe 2002; Dunham et al. 2002).