| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
Epigenetic Interactions among Three dTph1 Transposons in Two Homologous Chromosomes Activate a New Excision–Repair Mechanism in PetuniaAdèle van Houwelingen1,a, Erik Souera, Jos Mola, and Ronald Koesaa Department of Genetics, Institute for Molecular Biological Sciences, Vrije Universiteit, BioCentrum Amsterdam, de Boelelaan 1087, 1081 HV, Amsterdam, The Netherlands Correspondence to: Ronald Koes, koes{at}bio.vu.nl (E-mail), 31-20-4447155 (fax)
Unstable anthocyanin3 (an3) alleles of petunia with insertions of the Activator/Dissociation–like transposon dTph1 fall into two classes that differ in their genetic behavior. Excision of the (single) dTph1 insertion from class 1 an3 alleles results in the formation of a footprint, similar to the "classical" mechanism observed for excisions of maize and snapdragon transposons. By contrast, dTph1 excision and gap repair in class 2 an3 alleles occurs via a newly discovered mechanism that does not generate a footprint at the empty donor site. This novel mechanism depends on the presence of two additional dTph1 elements: one located in cis, 30 bp upstream of the an3 translation start in the same an3 allele, and a homologous copy, which is located in trans in the homologous an3 allele. Absence of the latter dTph1 element causes a heritable suppression of dTph1 excision–repair from the homologous an3 allele by the novel mechanism, which to some extent resembles paramutation. Thus, an epigenetic interaction among three dTph1 copies activates a novel recombination mechanism that eliminates a transposon insertion.
The best characterized class of transposons comprises elements with short terminal inverted repeats that transpose via a "cut-and-paste" mechanism. This class includes, for example, Tn7 and Tn10 of Escherichia coli, Activator (Ac), Dissociation (Ds), and Enhancer/Suppressor-Mutator (En/Spm) elements from maize, various Tam elements from snapdragon, P and Hobo elements from Drosophila, and various families of Tc elements from Caenorhabditis elegans. In addition, an increasing amount of evidence indicates that the intervening sequences, which are excised from genes encoding immunoglobulins and T-cell receptors to join variable (V), diversity (D), and joining (J) segments, originate from a cut-and-paste transposon (
The mobility of these transposable elements is controlled by a variety of mechanisms, all of which regulate expression of the recombinase enzyme(s) involved. For instance, in mammals, V(D)J joining is restricted to certain lymphoid cells, because other cell types do not express the proteins RAG1 and RAG2 that initiate the joining reaction (
Recent evidence shows that transposition and V(D)J joining, as well as replication of certain bacteriophages and retroviral integration, rely on the same basic chemistry for the breaking and joining of DNA segments. Hydrolytic nicking of DNA strands (in which H2O acts as the nucleophile) generates free 3' hydroxyl groups, which by nucleophilic attack engage in transesterification reactions and joining to the target DNA (reviewed in
By analyzing (phenotypic) reversions of unstable transposon insertion alleles, one can monitor the excisions of these transposons. Strikingly, P, Ac, and Tam insertion alleles appear to revert by different mechanisms in their cognate hosts (reviewed in
The excision of a plant transposon usually results in the complete removal of transposon sequences and the formation of a so-called footprint (reviewed in
We recently generated by random transposon mutagenesis a series of mutant alleles for the anthocyanin3 (an3) locus of petunia ( In this study, we show that an3 alleles with a single transposon insertion (dTph1 or dTph2) behave like unstable transposon insertion alleles in maize and snapdragon. Surprisingly, we found that the four unstable an3 alleles harboring two dTph1 elements revert preferentially via a novel mechanism. This mechanism relies on (epigenetic) interactions among three dTph1 copies on the two homologous chromosomes and results in elimination of a dTph1 insertion without leaving a footprint.
Analysis of Excision Alleles
For one set of an3 alleles (termed class 1), we found that the wild-type sequence was rarely restored after excision of dTph1 or dTph2 and that a so-called transposon footprint was usually generated (Figure 2A). A partial reversion allele (called W29FT) in which dTph1 had excised from an3-W29 was identified by phenotype in progeny from self-pollinated W29/W29 plants. Sequence analysis showed that W29FT contained a footprint that consisted of remainders of the duplicated 8-bp target site separated by inverted repetitions of part of the target site (Figure 2A). However, phenotypic selection of excision alleles may favor the recovery of certain excision alleles (those that partially or fully restore gene activity) over others. To avoid such a bias, we analyzed PCR-amplified somatic excision products from plants homozygous for the dTph1 insertion alleles W29, W2018, W2303, W2140, W2298, V2221, and W138A. In all somatic excision products, transposon footprints were found that in structure resembled that in W29FT (Figure 2A).
Excisions of dTph2 from the an3-GSm allele created footprints that were very similar to the dTph1-induced footprints (Figure 2A). In one case, we recovered a perfectly restored wild-type sequence. However, we believe that this is a rare excision product, because it was isolated from a somatically reverted branch (phenotypic selection for wild-type activity), whereas crosses showed that (germinal) excision of dTph2 from an3-GSm virtually always results in stably inactivated an3 alleles and rarely in An3+ revertant alleles (data not shown). The an3 alleles S205, R134, S857, and X2092 behaved quite differently and were therefore collectively designated as class 2 alleles. These four alleles all contain two dTph1 elements at the an3 locus (Figure 1). In general, self-pollinating homozygotes of any of these class 2 alleles yields ~50% of progeny with an an3mut phenotype similar to the parental line and ~50% with an An3+ revertant phenotype (data not shown). In progenies obtained by selfing R134/R134 or S205/S205 plants, we found at a low frequency (1 to 5%) plants with a very low spotting frequency (weakly an3mut) that were often difficult to distinguish from a stable an3- phenotype. PCR analysis showed that such plants were always homozygous for the presence of the downstream dTph1 element but could be either heterozygous or homozygous for the presence of the upstream element (Figure 1A; see hereafter for a detailed analysis). Because too few progeny were grown, we do not know if X2092/X2092 or S857/S857 also produce weak an3mut progeny after self-pollination. PCR experiments showed that in the An3+ reversion alleles, the dTph1 element upstream of the ATG remained present, whereas the dTph1 copy downstream of the ATG was invariably lost. Sequencing of four S205-derived, two R134-derived, and one S857-derived (independent) germinal revertant alleles showed that the wild-type sequence had been perfectly restored at the empty donor sites in all seven cases. In another 24 somatic excision products recovered without phenotypic selection by PCR amplification from homozygous S205/S205, R134/R134, S857/S857, or X2092/X2092 plants, we again found a perfect restoration of the wild-type sequence (Figure 2B). By contrast, sequencing of somatic and germinal excision products in which the upstream dTph1 element in the an3 alleles S205, R134, or S857 had excised showed that these excision events had created footprints (Figure 2C). These data indicate that elimination of the single dTph1 element in a class 1 an3 allele or the upstream dTph1 element in a class 2 an3 allele involves a ("classical") mechanism similar to that observed for transposons in snapdragon and maize. The downstream element of a class 2 an3 allele, however, appears to be preferentially eliminated by a different ("novel") mechanism that yields products without a footprint.
Reversion of Class 2 Alleles Is Suppressed in Hemizygotes Crosses between homozygotes for an unstable class 1 an3 allele and the W62 allele yielded predominantly progeny with spotted (an3mut) flowers (Table 1). The spotting frequency of such flowers always corresponded with that of the parental class 1 allele in the homozygous condition (Figure 3A and Figure 3C). PCR analysis showed these an3mut progeny plants contained the parental class 1 an3 allele, including the dTph1 element. an3-W62 is a complete deletion, which implies that reversion of a class 1 allele is independent of an3 sequences on the homologous chromosome. Other plants in the same progenies had evenly colored (An3+) or nearly white flowers without spots (an3-) (Figure 3B). PCR analysis showed that such plants always lacked the dTph1 insertion in the class 1 allele, apparently due to an excision event in germinal cells of the parent. Depending on the sequence of the footprint and its position in the an3 gene, such excision alleles have either a wild-type (An3+) or a stable mutant (an3-) phenotype.
Surprisingly, crosses between homozygotes for the three class 2 alleles, S857, R134, or S205, and W62 yielded only An3+ and an3- progeny (Figure 3F); the expected an3mut phenotype was missing (Table 1). PCR analyses showed that the An3+ plants lacked the downstream dTph1 element of the class 2 allele, presumably as a result of germinal excision, whereas the an3- plants contained the parental class 2 allele, including both dTph1 copies. We did not find an3- progeny containing an excision derivative of the class 2 allele with a footprint (Table 1), consistent with the finding that these excisions preferentially result in restoration of the wild-type sequence (Figure 2B). The absence of revertant spots in hemizygotes for an3-W62 and a class 2 an3 allele indicates that in this F1 background, reversions of the class 2 allele are somehow suppressed (cf. the flowers in Figure 3E and Figure 3F). The suppressed somatic reversion frequencies might be explained by (1) a decreased frequency of dTph1 excision or (2) an alteration in the spectrum of excision products (to products that do not restore gene activity). To discriminate between these possibilities, we separated the complete array of PCR-amplified somatic excision products from class 1 and class 2 homozygotes and hemizygotes on high (1-bp) resolution gels. Using controls, we performed similar PCRs with plants harboring (stable) wild-type (Figure 2D, lanes 1, 5, 9, 13, 17, and 21) or germinal excision alleles (lanes 4, 8, 12, 16, and 20). Figure 2D shows that the class 1 an3 allele W2018 generates predominantly excision products that are 7 bp longer than is the wild-type sequence (cf. lanes 1 and 2), which is consistent with the presence of footprints (cf. Figure 2A and Figure 2D). When we used a cloned W2018 DNA fragment as a template, these products were not detected, confirming that they result from in vivo transposon excisions and not from PCR artifacts (data not shown). In W2018/W62 hemizygotes, no gross alterations in the spectrum of W2018-derived excision products were observed (Figure 2D, lane 3; note that W62 is a complete deletion that does not generate PCR products). An an3- sibling from the same cross yielded only the an3 fragment with a 7-bp footprint, whereas the dTph1-containing an3 fragment was not detectable (Figure 2D, lane 4), indicating that dTph1 had excised in germinal tissue of the an3-W2018 parental plant. Similar results were obtained for other class 1 an3 alleles (W29, W2258, W2303, GSm, and W138A; data not shown). In homozygotes of the class 2 alleles S857, R134, and S205, excisions of the downstream element preferentially yielded products of a wild-type size (consistent with the absence of a footprint), although a small amount of slightly larger sized products was detectable (Figure 2D, lanes 6, 10, and 14). However, when the same alleles were hemizygous over the W62 deletion, the wild-type-sized products were no longer detectable, whereas the larger excision products remained (Figure 2D, lane 7) or slightly increased in abundance (lane 11). The absence of wild-type excision products in these hemizygotes made it possible to clone and sequence some of the larger excision products. For R134/W62 and S857/W62 plants, this revealed that they contain a footprint similar to those in excision products of class 1 alleles (Figure 4). Because the downstream dTph1 element in S205, R134, and S857 is in the protein coding sequence, such footprints do not allow restoration of an3 activity, explaining the absence of revertant spots in the flower.
At first sight, the fourth class 2 an3 allele, X2092, seemed to behave differently from the three other class 2 alleles, because X2092/W62 heterozygotes had spotted (an3mut) flowers (Table 1) that were by phenotype (spotting frequency) indistinguishable from X2092/X2092 homozygous flowers. Analysis of excision products showed that in X2092/X2092 homozygotes, elimination of the downstream dTph1 copy preferentially resulted in restoration of the wild-type an3 sequence (Figure 2D, lane 18), although some slightly larger excision products were visible in longer gel exposures (data not shown). However, in X2092/W62 hemizygotes, the wild-type-sized excision products were almost completely missing, and the 6- to 7-bp larger products increased in relative abundance (Figure 2D, lane 19). Thus, at the level of excision products, X2092 behaves similarly to the other three class 2 alleles. The differences between the intron insertion (X2092) and the exon insertions (S205, R134, and S857) concern the phenotypic consequences of footprint formation. First, footprints in introns (as produced by hemizygous X2092) allow reversion of an3 gene function and the formation of a colored spot, but footprints in exons (as produced by hemizygous S205, S857, or R134) in general do not. Second, the reduction of wild-type excision products in X2092 hemizygotes is accompanied by a relatively strong increase in (reversion) products with a footprint, which can account for the relatively high reversion frequency seen in X2092 hemizygous flowers. We also examined excisions produced by the upstream element of the class 2 allele R134. Figure 2D shows that excisions of this dTph1 copy yielded products that were longer than the wild-type sequence both in R134/R134 homozygotes (lane 22) and in R134/W62 hemizygotes (lane 23). Apparently, excisions of this dTph1 element preferentially result in the formation of footprints independent of the homologous an3 allele, similar to excision of the dTph1 elements in class 1 alleles. We conclude from these data that in homozygotes, the downstream dTph1 element of a class 2 allele can be eliminated by either of two mechanisms. One mechanism, which is used infrequently, resembles the classical mechanism by which class 1 alleles revert, because it yields typical footprints and is not suppressed in a cross with the line W62. However, class 2 alleles revert preferentially by way of a second, novel mechanism that restores the wild-type sequence and becomes suppressed after a cross with W62.
Instability of Class 2 an3 Alleles Is Homolog Dependent Table 1 shows that plants harboring a class 1 an3 allele heterozygous with either S206 or W1006 have spotted flowers (an3mut). By contrast, flowers that harbored one of the class 2 alleles S857, R134, or S205 heterozygous with either S206 or W1006 were white (an3-), indicating that reversions of the class 2 alleles were suppressed. Analysis of the PCR-amplified excision products produced by the downstream dTph1 copy of S857 and R134 on high-resolution gels showed that excision products with a wild-type size were strongly reduced in heterozygotes with S206 (data not shown), similar to the results described above for S857 and R134 hemizygotes. The fourth class 2 allele, X2092, specified an an3mut phenotype when heterozygous with W1006 or S206, similar to the X2092/W62 hemizygotes described above. Next, we crossed class 2 and class 1 alleles with plants harboring the W2018FT or W29FT alleles of an3. The two latter alleles were isolated as germinal excision derivatives of W2018 and W29, respectively, and only contain small sequence alterations (footprints) in the otherwise intact an3 gene (Figure 1 and Figure 2A). The results of these crosses (Table 1) showed that W29FT and W2018FT also were capable of suppressing reversions of S857, R134, and S205 in heterozygotes (see the example in Figure 3G) but not of X2092 or any of the class 1 alleles (see the example in Figure 3D). These suppressed reversion frequencies were extended to germinal tissues because no An3+ revertants could be recovered in subsequent progeny of these plants (see below).
Because many of the S205/S206 and S857/S206 progeny were homozygous for the unstable an1-W138 allele, we were able to examine whether reversions at another locus also were suppressed in the same plants. The an1 locus controls transcription of a set of anthocyanin biosynthetic genes ( In summary, these data show that any stable recessive an3 allele can suppress reversions of a class 2 allele in transheterozygotes, making it unlikely that this suppression is due to the absence of an3 sequences alone. Furthermore, these data show that the observed suppression is specific for class 2 an3 alleles, because instability of an1, rt, and class 1 an3 alleles is not suppressed in the same crosses.
Loss of the Upstream dTph1 from a Class 2 Allele Results in a Novel Allele That Behaves as a Class 1 Allele
Family V2307, obtained from a cross between S857/S857 and W29FT/W29FT parents, consisted of 17 An3+ plants (germinal revertants that had lost the downstream dTph1 element of S857; data not shown), 15 an3- plants (suppression of S857 reversions; see Figure 3G), and one plant with an3mut flowers (Figure 3H). Sequence analysis of PCR products amplified from the single an3mut plant showed that the upstream dTph1 element of S857 had germinally excised, creating a 7-bp footprint (see Figure 2C, allele V2307), whereas in the transposon sequences and the insertion site of the downstream element, no changes were observed (data not shown). A similar phenomenon was seen in family X2330, originating from a cross between R134/R134 and W2018FT/W62 plants. This family consisted of 13 wild-type revertants (An3+; resulting from excision of the downstream element of an3-R134), 22 plants with white flowers (an3-, due to suppression of R134 instability by the W2018FT or W62 allele), and one plant with spotted flowers (an3mut). Sequence analysis of the PCR-amplified an3 allele in this plant showed that the upstream dTph1 element of R134 had excised, creating a 7-bp footprint (see Figure 2C, X2330). No alterations were found in the transposon sequences or the insertion site of the downstream element (data not shown). Family V1519 was generated by crossing S205/S205 and W62/W62 parents at a time when we were still unaware of the upstream dTph1 copy in the S205 allele. PCR analysis showed that the six An3+ plants contained a germinal excision derivative of S205 that lacked the downstream dTph1 copy, whereas this element was still present in the 21 progeny plants that were scored as an3-. However, hindsight shows that one of these 21 must have displayed a low number of very small revertant spots (similar to the V512/W62 and T3463/W62 plants described below) that we did not notice at that time. After we had discovered the upstream dTph1 copy, we reanalyzed the DNA samples from V1519 and found this element to be present in all progeny, except for one plant that had initially been (mis)scored as an3-. We then germinated the seeds of this plant and obtained an3mut progeny (data not shown), whereas self-pollinating S205/W62 siblings from the V1519 family gave only an3- progeny (see below). In two progenies obtained by self-pollination of S205/S205 plants, heterozygous plants were identified (by PCR) that harbored S205 and a derived excision allele lacking the upstream dTph1 element (alleles T3463 and V512; see Figure 1 and Figure 5). Sequence analysis showed that T3463 and V512 contained a footprint at the former insertion site of this dTph1 element (Figure 2C, an3-T3462 and an3-V512), whereas no changes were observed in the sequence or insertion site of the downstream element (data not shown). When homozygous, the reversion frequency of the T3463 and V512 alleles was so low that it was barely detectable in the W138 (an1mut) genetic background in which they were maintained. Therefore, V512 and T3463 are depicted as nearly stable recessive alleles in Figure 1 and Figure 5. Electrophoretic analysis of the complete spectrum of excision products showed that excision of the downstream element from S205 yielded, in the large majority of cases, a product of wild-type size and, at a much lower frequency, products with a footprint (Figure 6A, lanes 2 and 4). In V512/V512 homozygotes, the wild-type-sized excision products were strongly diminished, but the excision products with a footprint were still detectable (Figure 6A, lane 5). The fraction of footprints (+3, +6, or +9 bp) that restore the reading frame and potentially allow reversion makes up a small fraction of the complete collection of footprints generated, thus explaining the low reversion frequency in V512/V512 homozygotes. This indicates that V512 now behaves as a typical class 1 allele.
To test whether T3463 and V512 displayed other features associated with class 1 alleles, we crossed T3463/S205 and V512/S205 heterozygotes to W62/W62 homozygotes. In both cases, half of the progeny had weakly unstable (an3mut) flowers (one to five small spots per flower, representing a low number of late reversions), whereas the flowers of the other half had no revertant spots at all and were scored an3- (Table 2, families Z2345 and Z2046, respectively). PCR analysis showed that all S205/W62 progeny had an3- flowers, indicating suppression of S205 reversions. The progeny with weakly unstable an3mut flowers were genotypically V512/W62 or T3463/W62. Similar results were obtained when the V512/S205 plant was crossed to an an3-W29FT homozygote (Table 2, family Z2045). Apparently, the low reversion frequencies of V512 and T3463 (which are relatively easily detected in this An1+ genetic background) are not suppressed by the W62 or W29FP allele, whereas the high reversion frequency of S205 is completely suppressed. However, when we compared the complete spectrum of PCR-amplified excision products in S205, V512, and T3463 hemizygotes, we did not detect consistent differences that could account for the differing an3- and weak an3mut phenotypes (Figure 6A, lanes 6 to 9), most likely because the V512- or T3463-derived reversion products made up too small a fraction of the total amount of PCR products that were recovered in these assays. Taken together, these data show that loss of the upstream dTph1 element of a class 2 allele results in a new allele that genetically behaves as a class 1 allele, because (1) the fraction of excision derivatives with wild-type sequence is strongly diminished and (2) it specifies spotted (an3mut) flowers when heterozygous over a stable recessive an3 allele.
Reactivation of Suppressed Class 2 Alleles
Only 24 of these 777 progeny plants displayed some somatic An3+ reversions (i.e., spots). In all of these cases, an3 instability appeared to be irregular, because we found on the same plant flowers with an3mut and an3- phenotypes. Furthermore, the number of revertant spots on an3mut flowers was usually lower than the number on flowers of the parental class 2 homozygous lines. PCR analysis confirmed that all the plants bearing an3mut flowers were indeed homozygous for the class 2 allele. However, the large majority of class 2 homozygotes (identified by PCR) remained phenotypically an3- and did not show any An3+ reversions. This suggested that some unknown factor prevented the reactivation of instability in most (~85%) of the class 2 allele homozygotes.
Suppression of an3-S205 Reversions Has Epigenetic Features The V512 allele was identified in family V512 in a heterozygous (V512/S205) plant. This heterozygote had white flowers (an3-), indicating that V512 suppressed reversions of the S205 allele on the homologous chromosome (cf. Figure 3I and Figure 3J). Hereafter, we mark such a suppressed allele with an asterisk (S205*) to differentiate it from a frequently reverting S205 allele, even though we have not been able to detect structural differences between S205 and S205* (epi)alleles by using PCR or sequencing. The V512/V512 and V512/S205* progeny obtained by self-pollinating this plant (family Z2043) had an3- flowers, as expected. Also, the S205*/S205* homozygote was an3- (Figure 3K), indicating that reversions were not reactivated after the S205* allele was made homozygous again. When the same V512/S205* parent was outcrossed to an S205/S205 sibling with an an3mut phenotype, the progeny segregated for An3+, an3mut, and an3- phenotypes (family Z2044). PCR analysis showed that the An3+ plants were heterozygous for the downstream dTph1 insertion, presumably due to a germinal reversion in the S205/S205 parent. The an3- progeny were genotypically V512/S205*, indicating that the V512 allele again prevented reversions of the newly introduced S205 allele (thus marked with an asterisk in Figure 5). The an3mut progeny (see Figure 3L for an example) all appeared to be genotypically S205*/S205. Crosses with a V512/S205* plant from family Z2043 gave the same result. After self-pollination, the S205*/S205* homozygotes were an3- (family Z2355), whereas outcrosses to an S205/S205 plant yielded S205*/S205 progeny with an an3mut phenotype (family Z2356). We also selected a V512/S205* plant from family Z2044, in which suppression of S205 reversion was established by an independent event. Again, self-pollinations yielded only an3- progeny (family Z2357), whereas outcrosses to an S205/S205 plant with an an3mut phenotype yielded an3mut and An3+ progeny (Z2358). The An3+ progeny were all heterozygous for the downstream dTph1 insertion, presumably due to a germinal reversion in the S205/S205 parent. The T3463 allele has the same structure as V512 (Figure 1 and Figure 2C) but arose independently. In a heterozygote, T3463 also suppressed reversions of S205 (thus marked with an asterisk in Figure 5, family T3463), and this suppressed state was maintained when the S205* allele was made homozygous again by selfing (family Z2051). Interestingly, two other plants in the T3463 family displayed an an3- phenotype. PCR analysis showed that they were both homozygous for an an3 allele that was indistinguishable in structure from S205, except that somatic excision products were markedly decreased (data not shown). These suppressed S205* alleles behaved in further crosses in a manner identical to the suppressed S205* alleles described above. Thus, repeated self-fertilizations yielded only an3- progeny, even after three generations (families V2308, V2310, and progenies thereof), whereas outcrosses to homozygotes for an unstable S205 allele produced an3mut progeny (families V2309 and V2311). The simplest explanation for these results is that the T3463 allele arose in germinal cells of the parental plant, leading to the suppression of reversions of the S205 allele on the homologous chromosome and maintenance of this state in the gametes and progeny originating therefrom (family T3463). It is possible that in the S205*/S205 heterozygous plants produced by the various outcrosses, both alleles contributed to the spotted an3mut phenotype (implying that S205* had converted back to S205) or that only S205 contributed to the revertant spots, whereas S205* remained in its stabilized state. Molecular analyses of the S205/S205* heterozygotes themselves cannot distinguish between these possibilities, because excision products of either of the two alleles would have the same structure. Instead, we self-pollinated S205/S205* plants from three different families to examine whether the stabilized S205* could be recovered in the progeny. All resulting families (A2433 and data not shown) contained a high number of an3- plants, suggesting that the instability of the S205* allele had not been reactivated in the S205*/S205 parent. Similar to the crosses described above, 1 to 5% of the progeny resulting from self-fertilization of frequently reverting R134/R134 plants exhibited a reversion frequency that was so low (i.e., weakly an3mut) that it was difficult to distinguish from an an3- phenotype. PCR analysis of two such plants showed they were both homozygous for an an3 allele that was indistinguishable from R134, except that the amounts of somatic excision products were markedly decreased (data not shown). This suggests that, in addition to S205, R134 also could change into a "stabilized" (epi)allele, which we designated R134*, analogous to the (epi)allele S205*. To analyze why the reversion frequencies became so strongly suppressed in S205* and R134* heterozygotes and homozygotes, we analyzed the spectrum of excision products produced by the downstream element. Figure 6A shows that in S205*/V512, the fraction of wild-type-sized excision products was markedly reduced when compared with S205/S205 homozygotes (cf. lanes 10 and 11 with lanes 2 and 4) and was of an essentially similar composition as that produced by V512/V512 homozygotes (lane 5) or V512/W62 hemizygotes (lane 7). Because in S205*/S205* homozygotes again a similar pattern of excision products was found (Figure 6A, lanes 12 and 13), we assume that S205* and V512 contribute similarly to the collection of excision products found in heterozygotes. These data indicate that the essential difference between S205 and its derived epiallele S205* is that the latter produces many fewer wild-type-sized excision products, indicating that it has lost its capability to revert by the novel mechanism. Figure 6B shows that a frequently reverting R134 allele in the homozygous condition produced mainly wild-type-sized excision products, whereas only a small fraction of the excision products contained an 8-bp footprint (lanes 3 to 5). However, in a homozygous R134*/R134* plant, the wild-type-sized excision products were markedly reduced, whereas the +8-bp footprint products remained detectable (Figure 6B, lane 6), similar to R134(*)/W62 hemizygotes in which reversions were also suppressed (lane 2). In summary, these data show that a frequently reverting class 2 allele such as S205 undergoes heritable transition to a different "state" or epiallele (indicated as S205*) in which reversions by the novel mechanism are suppressed if it is exposed to a derivative allele such as V512 or T3463 in a heterozygous cell. An S205* allele, however, does not confer this state to an unstable S205 allele in heterozygotes.
It is by now widely accepted that the activity of a gene is not determined by its sequence alone and that additional, epigenetic factors (e.g., changes in methylation or chromatin structure) can heritably modulate its expression (discussed in
Class 1 an3 Alleles Revert by a Classical Mechanism
The simple class 1 an3 alleles, harboring a single dTph1 or dTph2 insertion, behaved very similarly. In the majority of cases, transposon excision resulted in the formation of a footprint, whereas at a low frequency, excision alleles without a footprint could be found, especially when excision products were phenotypically selected (Figure 2A). Similar results were obtained with dTph1 insertions at four other petunia loci ( Our results indicate that reversion of class 1 alleles is largely insensitive to the structure of the an3 allele on the homologous chromosome. The absence of an3 sequences on the homologous chromosome (as in hemizygotes with an3-W62) did not reduce the spotting frequency or cause alterations in the spectrum of excision products (Figure 2D). We also did not observe upregulation of the spotting frequency when intact an3 sequences were present opposite the insertion sites (as in heterozygotes harboring the S206 or W29FT allele). Thus, gene conversion mechanisms, like DSB repair, play at most a minor role in reversion of class 1 an3 alleles.
Although a single dTph1 insertion may in principle generate an array of different footprints, our data indicate that certain footprints are preferred over others. For instance, excision of dTph1 from W2018 preferentially results in a 7-bp footprint (Figure 2A and Figure 2D) that causes a frameshift, consistent with the observation that most excisions result in a stable an3- allele (Table 1; data not shown). Yet, the revertant spots on W2018 flowers show that different footprints, which restore an3 function, also can occur but at a lower frequency. Analysis of a large number of Ac and Ds excisions from different positions at the maize waxy locus or Arabidopsis transgene(s) revealed a similar preference for certain footprints over others, which is determined at least in part by DNA sequences immediately flanking the transposon insertion (
Class 2 an3 Alleles Revert by a Novel Mechanism
A homolog-dependent reversion frequency is the hallmark of the DSB repair mechanism by which transposon-excision gaps are repaired in the germ line of Drosophila and C. elegans (
Analysis of Ds excisions from six different positions in the maize waxy (wx) gene showed that one allele (wx-C28) produced predominantly excision alleles with a wild-type sequence (~66% of the excisions) rather than a footprint (~33% of the excisions), and this was taken as an indication that the two types of product were generated by two distinct mechanisms (
Reversion of a Class 2 an3 Allele Requires an Interaction among at Least Three dTph1 Elements The key feature that distinguishes class 2 an3 alleles from class 1 alleles is the presence of a second dTph1 insertion (copy I; Figure 7) that by itself does not affect the flower phenotype (although it does reduce the amount of an3 transcripts; A. van Houwelingen and R. Koes, unpublished data). Several observations indicate that the elimination of dTph1 copy II by the novel mechanism indeed depends on the presence of copy I in the same allele. First, excision of dTph1 copy I from S205 yields novel alleles (V512 and T3463) with a barely detectable reversion frequency because they preferentially generate excision products with a footprint (Figure 6). Second, the spectrum of excision products does not show any major alterations when the V512 allele is in a homozygous or hemizygous condition (Figure 6). Third, excision of dTph1 copy I from three different class 2 alleles results in new alleles that specify an unstable (an3mut) phenotype when heterozygous over a stable recessive an3 allele (Figure 3H and Table 2). These data indicate that upon loss of dTph1 element Ia (Figure 7), the class 2 allele changes its reversion behavior into that of a class 1 allele, which implies that dTph1 copy I is required for the element downstream in the same allele (dTph1 copy II) to be eliminated by the novel mechanism. Strikingly, the interaction between dTph1 copies I and II is not symmetrical. dTph1 element I appears to be preferentially removed from a class II allele by the classical mechanism, because excisions result independently in footprints from the structure of the an3 allele on the other chromosome (Figure 2C and Figure 2D). Possibly, the directionality of this interaction is specified by the relative orientations of both dTph1 copies, but because the relative orientation of dTph1 copies is the same in all four class 2 an3 alleles (Figure 1), we could not verify that experimentally. Although the presence of dTph1 copy I is necessary, it is clearly not sufficient to eliminate the downstream copy II by the novel mechanism, because sequences in the homologous an3 allele also are required. The one common feature shared by all an3 alleles that suppress reversions of a homologous class 2 allele in trans is the absence of dTph1 element Ia, whereas the only an3 allele that does not provoke the suppression, S205*, does contain element Ia. The effect of dTph1 copy Ia on the reversion frequency of the homologous allele is most clearly seen in plants in which the S205(*) allele is heterozygous over V512 or T3463. In these heterozygotes, both the reversion frequency and the amount of wild-type-sized excision products are markedly decreased when compared with those in S205/S205 homozygotes, indicating that neither V512 (or T3463) nor S205(*) can revert by the novel mechanism. This implies that the absence of dTph1 copy Ia (as in V512 and T3463) does not affect only the elimination of the element downstream in the same allele (copy IIa) but also that of the downstream element (II) in the homologous S205(*) allele. Thus, elimination of dTph1 element II somehow requires the presence of dTph1 copy Ia on the homologous chromosome. However, we cannot conclude from our data whether this interaction is direct or indirect, for instance, via dTph1 copy I (Figure 7). To test whether dTph1 copy II also senses dTph1 copy IIa on the homologous chromosome, one needs to examine the instability of a class 2 allele in a heterozygote harboring an allele like an3-W138A or a reversion allele. The difficulty is that in such a plant, the reversion products produced by the class 2 alleles cannot be discerned (either by PCR or by phenotype) from the an3 allele (W138A or R27) on the other chromosome.
Suppression of Class 2 Allele Reversions Fits the Genetic Definition of Paramutation Our data indicate that the interaction between dTph1 copies II and Ia has an epigenetic character and that the inactivation of the novel excision repair mechanism by which a class 2 predominantly reverts fits the (genetic) definition of paramutation. When dTph1 copy Ia is absent, the elimination of dTph1 copy II from the homologous allele by the novel mechanism is suppressed and the class 2 allele undergoes a transition to a different "state" (essentially becoming a class 1 allele) that is heritable over at least three generations (Figure 2, Figure 4, Figure 5, and Figure 6). How (methylation or otherwise?) and where (in dTph1 and/or an3 sequences?) the suppressed state of a class 2 allele is remembered remain to be determined. The epigenetic suppression of the novel transposon excision–repair mechanism differs in several important aspects from the previously described examples of paramutation or epigenetic inactivation of transposons. First, the target process is not transcription of a transposase gene or an immobile gene but one or more steps in a DNA recombination process. Second, the epigenetic suppression specifically blocks elimination of the transposon by the novel mechanism, without blocking elimination by a second independent mechanism (the classical mechanism). Third, unlike paramutated r, b, or pl (epi)alleles, a paramutated (epi)allele such as S205* does not change the behavior of a frequently reverting |
TOP
ABSTRACT
INTRODUCTION
