We have done genetic and molecular studies on several derivatives of McClintock's a-m2 Spm insertion allele. These alleles have the unique property that the a gene is either co-expressed with, or expressed under the control of, the Spm element, while most Spm insertions inhibit gene expression. We have analysed 9 derivative alleles isolated by McClintock from the original a-m2 allele (now lost). Each derivative has a different mutation affecting the inserted Spm element, expression of the a gene, or both. The genetic relationship among the derivatives and their designations are shown in Figure 1. Our interest in these alleles stems from the observation that the manner in which a gene is expressed in these derivatives reflects the Spm's own regulatory mechanisms, suggesting that the gene has come under the control of the element and that the behavior of the derivatives therefore offers insight into Spm regulation. Moreover, there are several mutant Spm elements among the derivatives, the analysis of which extends our understanding of the element's genetic organization.
We have found the insertion site to be identical in all of the a-m2 derivatives; recent analyses of the a locus by Schwarz-Sommer et al. (EMBO J., in press) indicate that the insertion site is 99 bp upstream of the a gene's transcription start site. The derivatives differ from each other by mutations either within or very near the element.
Based on genetic analyses, two of the derivatives have standard Spm (Spm-s) elements (a-m2-7991A1 and -8010A), two have late-acting weak Spm (Spm-w) elements (a-m2-8011 and -8745) and five have transposition-defective Spm (dSpm) elements (a-m2-7995,-7977B, -8004, -8167B, and -8417). The two derivatives with Spm-s elements have 8.3-kb insertions, while most of the derivatives with Spm-w or dSpm elements have shorter insertions which differ from the 8.3 kb Spm-s element by internal deletions. The type and length of the Spm element in each derivative is given in Figure 1. Most derivatives have a single insertion at the a locus, but the a-m2-8745 allele has two insertions, the 6.6-kb Spm-w insertion of the a-m2-8011 allele from which it was derived and a second nearby insertion (or duplication) that has no effect on the element, but inactivates the a gene. Most derivatives have a single large deletion within the
Spm element, while the a-m2-8417 derivative has two deletions, consistent with its derivation in two steps from the original a-m2 allele.
The 8.3-kb Spm-s element of the a-m2-7991A1 allele has been cloned and sequenced; its structure is shown in Figure 2. The Spm-s element is virtually identical with the Enhancer (En) element sequenced by Pereira et al. (EMBO J. 5:835, 1986); it is shorter by 4 bp and differs at 6 additional nucleotides, only one of which affects the structure of the putative protein encoded by the element's major transcript. The two Spm-s alleles differ from each other and from the original a-m2 allele primarily in the respective levels of a gene expression and the phenotypes of stable alleles resulting from excision of the element. We believe these differences may be attributable to small changes at the insertion site immediately adjacent to the element because we have found a 3-bp insertion adjacent to the sequenced Spm-s element that is not present in any of the other alleles analysed.
The 6.6-kb Spm-w element cloned from the a-m2-8011 allele has an internal deletion, the location of which is shown in Figure 2. The abundance of the 2.3-kb major Spm transcript homologous to sequences in the element's right half (Figure 2) is lower by a factor of 5-9 in plants with an Spm-w element than in plants with an Spm-s element. The Spm-w element transposes less frequently and later in development than an Spm-s element, but retains the element's 'suppressor' function, as judged by its ability to inhibit expression of an Spm-suppressible insertion allele, such as the commonly used a-m1-5719A1 allele of McClintock. Since the deletion in the Spm-w-8011 element eliminates parts of both large open reading frames (ORFS) located in the element's left half, we conclude that the integrity of neither ORF is essential for expression of either known element-encoded function. The low transposition frequency of the Spm-w element is correlated with a reduction in the element's major transcript, suggesting that the transcript encodes the element's transposase. The observation that the element in the a-m2-7995 allele is completely transposition defective and has an overlapping deletion that eliminates sequences hybridizing to the 2.3-kb Spm transcript supports this inference (Figure 2). The structure of the closely related En element's transcript is reproduced from Pereira et al. (op. cit.) in Figure 2. Since there are no differences in structure between the En and Spm elements that are likely to affect transcription, it is likely that the Spm transcription unit is the same as that of the En element. The Spm-w deletion is confined to the large intron that occupies most of the element's left half. We do not know whether the deletion's quantitative effect on Spm mRNA abundance is the result of decreased RNA stability or processing or the consequence of decreased transcription initiation.
Several lines of evidence suggest that the Spm's suppressor' and 'mutator' functions are not encoded by separate, complementing genes. All of the dSpm elements in the present series were tested genetically for each function and all lack both functions. The a-m2-8167B allele, which has a full-length element, was also tested for its ability to complement an Spm-w element and it could not. These observations, taken together with the observation that the Spm-w-8011 deletion eliminates neither function, suggest that the element's two genetically defined functions either reside in the same protein or in proteins that share a subunit or domain and therefore coding sequences.
The frequency with which an internally deleted Spm element can be trans-activated to excise depends on its structure. Among the deleted elements whose structure is depicted in Figure 2, the Spm-w-8011 element excises at a very high frequency in the presence of an Spm-s, the dSpm-7995 and -7977B elements excise at an intermediate frequency, and the dSpm-8004 element excises at a very low frequency. It has been shown in other studies (Schiefelbein et al., PNAS 82:4783,1985; Schwarz-Sommer et al., EMBO J. 4:2439, 1985) that deletions that leave 1-1.5 kb of each element end do not affect excision frequency, while deletions that extend to within a few hundred nucleotides of an element end reduce excision frequency. The Spm element has 13-bp terminal inverted repeats (CACTACAAGAAAA) and a subterminal repetitive region at each end comprising several copies of the 12-bp sequence CCGACACTCTTA repeated in both orientations. It has been suggested by Schwarz-Sommer et al. (EMBO J. 4:2439, 1985) that the subterminal repeats form intramolecular duplexes in transposition and that deletions extending into the subterminal repeats reduce transposition frequency by disrupting secondary structure. Sequence analysis of cloned copies of the dSpm-7995 and -7977B elements has revealed that the deletions extend into the subterminal repetitive region at the element's right end, eliminating 5 and 4 of the repeats, respectively. The excision frequency of these elements is lower than that of the longer Spm-w-8011 element, but not much lower. We also find that the endpoints of intra-element deletions often occur within or at one end of the subterminal 12-bp repeats or a homologous sequence elsewhere within the element. Since it has been known for some time that intra-element deletions are element-catalysed, the implication is that an element-encoded protein capable of promoting the cleavage and religation of DNA (the transposase?) recognizes the 12-bp repeats. Thus we offer the alternative suggestion that the 12-bp sequence repeated near element termini is a recognition sequence for binding of an element-encoded protein, possibly the transposase itself.
Sequence analysis of the virtually immobile dSpm-8004 element suggests the existence of an additional, nonrepetitive determinant of excision (and, presumably, transposition) frequency. The dSpm-8004 element retains all of the repeats at both element ends and has normal termini. The sequence present in the more frequently excising 1.3-kb dSpm-7977B allele that is missing from the 1.1-kb dSpm-8004 allele comprises most of the element's first exon and a few hundred nucleotides of its first intron. Thus it appears that a non-repetitive sequence near, but not including, the element's transcription initiation site influences the element's mobility.
Perhaps the most interesting aspect of the a-m2 alleles is that the Spm element controls expression of the a gene. The original a-m2 allele exhibited an intermediate level of a gene expression, giving a palely pigmented kernel phenotype with small, deeply pigmented spots of normal a gene expression resulting from excision of the element. Expression of the a gene was affected by reversible changes in expression of the element, as well as mutations within the element. McClintock (CIW Yrbk. 61:265,1962) reported that when the Spm-s element of the original allele became inactive, so did the a gene. The a-m2-7991A1 allele resembles the original allele in a gene expression and we have isolated a derivative in which both element and locus are reversibly inactivated. We have observed that the inactive Spm element can be readily reactivated by the introduction of an Spm-w element, although we do not yet know whether the reactivation is heritable. This behavior suggests that 1) the a gene has come under the control of a mechanism that inactivates the Spm element and 2) that an element-encoded gene product can overcome inactivation of the element.
These inferences are strengthened by the behavior of the dSpm alleles. The phenotype of all of the dSpm derivatives studied here is colorless in the absence of an Spm-s. However, in the presence of an Spm, all except the a-m2-8417 allele exhibit intermediate a gene expression. Hence intra-element deletions that inactivate the element also inactivate the gene. Yet expression of the a gene in dSpm derivatives remains under Spm control: the gene is expressed when an Spm element is present elsewhere in the genome. It follows that an Spm-encoded gene product can reactivate expression of the gene in a-m2 derivatives with dSpm elements, just as it can reactivate an inactive Spm element. Thus it appears that expression of the a gene of the a-m2 alleles reflects the operation of the Spm's own regulatory mechanisms and suggests that the element encodes a positive regulatory function. Most of the derivatives used in the present study can be and have been tested for their ability to activate expression of the a gene of dSpm derivatives. We find that the Spm-w element can trans-activate a gene expression, while none of the dSpm elements can, suggesting that the element's positive regulatory function is not encoded by a separate, complementing gene.
Because dSpm elements as short as 1.3 and 1.1 kb are sufficient to mediate a gene activation, it appears that the target sequences for the element's regulatory protein are near its left end, which is located near the a gene's transcription start site. Differences among the dSpm derivatives in the capacity for Spm-dependent expression suggest that the minimal requisite element sequence is probably less than 300 bp of the element's left end, comprising the terminal inverted repeat, the subterminal repetitive region and the element's transcription initiation site. However, elements with at least an additional 1 kb of the element's left end express the a gene at a substantially higher level, suggesting that an internal element sequence can serve as an enhancer of a gene expression. The observation that Spm control of a gene expression is mediated by sequences around the element's own site of transcription initiation suggests that the element's positive regulatory mechanism functions at a transcriptional level. Thus we propose that an element-encoded protein interacts with sequences around the element's transcription start site to activate transcription of the element. We also suggest that the element's positive regulatory function can overcome the negative mechanism that reversibly inactivates the Spm element (as well as the a gene). The regulatory scheme that we propose is depicted in Figure 3.
Although the nature of the negative mechanism that inactivates the Spm element has not yet been elucidated, there is growing evidence that maize transposable elements are inactivated by methylation (Chandler and Walbot, PNAS 83:1767, 1986; Cone, Burr and Burr, PNAS, in press). The results of the present study suggest that the inactivation system is probably not element encoded. This follows from the observation that the a gene is inactive in derivatives with extensive intra-element deletions. However, the possibility cannot be excluded that defective Spm elements in the genome participate in the inactivation process.
There is some evidence that Spm's regulatory system not only overrides the negative mechanism, but can also interfere with it heritably. McClintock (CIW Yrbk. 63:592, 1964) reported that the a gene of the a-m2-7995 and -7977B derivatives could be 'preset' for expression after segregation of the resident Spm element away from the allele at meiosis. Under such circumstances, kernels appear to commence development with an actively expressed a gene that returns to the inactive state during development, yielding kernels with an irregular and distinctive pigmentation pattern. Assuming that the same mechanism effects a gene inactivation in the a-m2 dSpm derivatives and reversible inactivation of the Spm element, it follows that an Spm-encoded gene product can both directly overcome the inactivating mechanism and interfere with its propagation. This interpretation is supported by our previous report (MNL 60:18) that an Spm-w element substantially enhances the frequency of activation of cryptic elements. We suggest, therefore, that the Spm element's positive autoregulatory circuit functions to maintain the element in an active state both by promoting expression of the element directly and by interfering heritably with a negative, probably non-element-encoded, inactivating mechanism (Figure 3).
J. Kingsbury, P. Masson, R. Surosky and N. Fedoroff
Return to the MNL 61 On-Line Index
Return to the Maize Newsletter Index
Return to the Maize Genome Database Page