Paramutation of the r1 Locus of Maize Is Associated With Increased Cytosine Methylation

Corresponding author: Elsbeth L. Walker, Biology Department, Morrill Science Center, University of Massachusetts, Amherst, MA 01003, ewalker{at}bio.umass.edu (E-mail).

Communicating editor: J. A. BIRCHLER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In paramutation two alleles of a gene interact so that one of the alleles is epigenetically silenced. The silenced state is then genetically transmissible for many generations. The large (220 kbp) multigenic complex R-r is paramutable: its level of expression is changed during paramutation. R-r was found to exhibit increases in its level of cytosine methylation (C-methylation) following paramutation. These C-methylation changes are localized to the 5' portions of the two genes in the complex that are most sensitive to paramutation. These methylation changes flank a small region called that is thought to have been derived from a transposon named doppia. A mutant derivative of R-r that has a deletion of the region fails to become methylated under conditions in which R-r is heavily methylated. This suggests that the presence of sequences at the locus is required for the methylation changes that are observed following paramutation.

PARAMUTATION is a regularly occurring, directed, and heritable alteration of gene expression resulting from the interaction of two alleles. One allele is referred to as paramutable: its expression changes following paramutation. The other allele is paramutagenic: it causes the change in the paramutable allele. When the paramutable/paramutagenic heterozygote is crossed to allow segregation of the two alleles, virtually 100% of the paramutable alleles transmitted display a decrease in expression. This decreased expression level (the paramutant phenotype) persists through many generations. Paramutation was first described in 1956 (BRINK 1956 ), but a mechanism to explain how one allele can heritably affect the expression of another has remained obscure.

One of the best studied examples of paramutation is the r1 locus of Zea mays (BRINK 1973 ; KERMICLE 1996 ). r1 genes encode helix-loop-helix proteins that are capable of directing transcription of the structural genes in the anthocyanin biosynthetic pathway (GOFF et al. 1990 ; LUDWIG et al. 1989 ). Thus, dominant R1 genes confer red or purple anthocyanin pigmentation to the tissues in which they are expressed. Paramutation at r1 is illustrated in Figure 1. The crosses involve two r1 complexes, the paramutable R-r complex, and the paramutagenic R-marbled (R-mb) complex. In the aleurone layer of the seed, the R-r complex normally confers full coloration. The R-mb complex confers a marbled pattern of pigmentation to aleurone. Following outcrossing of the heterozygote to a recessive colorless, r, tester strain, kernels receiving the R-r complex show a mottled pattern of pigmentation in which not all aleurone cells are colored. The paramutant R-r complex is referred to as R-r ', to distinguish it from nonparamutant R-r. Interestingly, the effect on aleurone pigmentation is only observed if R-r ' is transmitted through male gametes (pollen). When R-r ' is transmitted through the female, the normal dark uniform pigmentation pattern is observed. This dependence on the mode of transmission is not due to a simple dosage effect; two paternally transmitted copies of R-r ' still confer a paramutant phenotype in aleurone (KERMICLE 1970 ). Nonparamutant R-r also exhibits mild silencing on male transmission. R-r ' is characterized as being metastable; under certain conditions, it reverts toward the standard full color phenotype (BRINK 1956 ; BRINK 1958 ; STYLES and BRINK 1966 ; STYLES and BRINK 1968 ).

View larger version (68K): In this window In a new window Download PPT slide   Figure 1. —Paramutation of r1. The series of crosses shown involves three alleles of r1, the R-r complex (paramutable allele), the R-mb complex (paramutagenic allele), and a recessive tester allele, r. R-r confers solid purple pigmentation to the aleurone layer of the seeds; R-mb confers a marbled pattern of pigmentation to the aleurone; the r tester allele confers no color to aleurone. Paramutation is observed following test crossing of the R-r/R-mb heterozygote. Half of the kernels on the test cross ear carry the R-mb allele, which, when transmitted through the female, confers a marbled pattern of pigmentation in aleurone, but when transmitted through the male confers extremely infrequent sectors rendering the aleurones mostly colorless. The R-r allele transmitted from the R-r/R-mb heterozygote is referred to as being "paramutant" and is designated as R-r ' to indicate this new epigenetic state. Male transmission of R-r ' results in a pale mottled pattern of pigmentation in the aleurone, while female transmission of R-r ' results in a normal solid color pattern of aleurone pigmentation.

The paramutable r1 complex, R-r, consists of four duplicated copies or partial copies of the r1 transcription unit (WALKER et al. 1995 ). These four copies, referred to as components, are designated P, q, S1, and S2. The S1 and S2 components are complete genes and are responsible for pigmentation of the aleurone layer of the seed. The q component is nonfunctional because it lacks downstream coding sequences. The P component is the third complete r1 gene, and is active in several tissues in the plant including the coleoptile, the roots, and the anther walls. The three complete genes of the complex are not equally subject to paramutation: the S genes (S1 and S2) show a large decrease in the amount of pigment they confer following paramutation, while the P gene shows a relatively small decrease following heterozygosity with a paramutagenic r1 complex (BRINK and MIKULA 1958 ; BROWN 1966 ).

The S genes are arranged in an inverted head-to-head orientation and are separated by a 387-bp region called . The region appears to be a rearranged remnant of a transposable element of the CACTA family (CONE et al. 1993 ; UPADHYAYA et al. 1985 ) called doppia (WALKER et al. 1995 ). At its ends, contains partial copies of the terminal inverted repeats of the doppia element. Internal to these inverted repeats are multiple copies of a subterminal repeated element of doppia, and a second region consisting of sequences that have undergone extensive rearrangement and that may or may not have been derived from doppia. The region functions as the promoter for both the S1 and S2 genes of R-r (WALKER et al. 1995 ). doppia sequences that include one terminal inverted repeat element and multiple subterminal repeats are also found distal to q (WALKER et al. 1995 ). No doppia sequences are found at the P component.

Recently, a variety of paramutation and paramutation-like phenomena has been observed that can involve either endogenous genes or transgenes. In maize, two other genes, b1 and pl1, are known to be subject to paramutation (COE 1966 ; HOLLICK et al. 1995 ; PATTERSON et al. 1995 ). In petunia, a transgenic copy of the maize A1 gene can spontaneously become paramutagenic (MEYER et al. 1993 ). Paramutation-like phenomena include a broad class of "homology-dependent gene silencing" events associated with meiotically stable reductions in gene expression (reviewed in MATZKE et al. 1996 ; MEYER and SAEDLER 1996 ). Certain examples are particularly similar to paramutation since homologous gene copies interact in a way that heritably alters the expression of one of the copies, although the silencing and silenced genes are nonallelic. Examples of this type of interaction include the transgenic H locus in which an inactive transgene locus has the ability to silence partially homologous transgenes introduced into the same genome by genetic crossing (MATZKE et al. 1994 ), and inactivation of endogenous genes that appears to occur among members of the phosphoribosylanthranilate synthase (PAI) gene family of the Wassilewskija strain of Arabidopsis (BENDER and FINK 1995 ).

Three features stand out as being held in common by several of these systems. One is that multiple copies of the genes are often present in the same genome (reviewed in FLAVELL 1994 ; MATZKE et al. 1996 ; MEYER and SAEDLER 1996 ). A second feature is that inactivation tends to be associated with increased cytosine methylation (C-methylation) of both silenced and silencing loci (BENDER and FINK 1995 ; EGGLESTON et al. 1995 ; MATZKE et al. 1994 ; MEYER et al. 1993 ; also reviewed in MATZKE et al. 1996 ; MEYER and SAEDLER 1996 ). A final feature noted by some reviewers (MARTIENSSEN 1996A ; MATZKE et al. 1996 ) is the possible involvement of transposable elements in some of these systems. Epigenetic inactivation of transposable elements is a well-known phenomenon (reviewed by FEDOROFF 1996 ; MARTIENSSEN 1996B ).

Some paramutation systems, e.g., b1 and pl1 in maize, do not share all of these features. Alleles of b1 and pl1 that are active in paramutation are simple, containing only a single gene copy (CHANDLER et al. 1996 ). Furthermore, no C-methylation differences have been found during either b1 or pl1 paramutation (CHANDLER et al. 1996 ). Alleles of pl1 that are active in paramutation do contain a doppia transposable element (CHANDLER et al. 1996 ; HOLLICK et al. 1995 ), but alleles of pl1 that are inactive in paramutation also contain this element (COCCIOLONE and CONE 1993 ). No transposable element has been reported near or within the b1 alleles that are active in paramutation. In contrast, the r1 paramutation system shares all three of these features. Both the paramutable R-r complex (ROBBINS et al. 1991 ; WALKER et al. 1995 ) and the paramutagenic R-mb (E. L. WALKER, unpublished results) and R-stippled (R-st) (EGGLESTON et al. 1995 ) complexes contain multiple r1 genes. Epigenetic silencing of r1 genes has been correlated with increased C-methylation (EGGLESTON et al. 1995 ; RONCHI et al. 1995 ). Finally, the R-r complex harbors at least two fragments of the transposable element, doppia (WALKER et al. 1995 ).

The question of whether the region, the multiple repeated genes, or both make R-r susceptible to paramutation has been partially addressed using a series of mutant derivative alleles of R-r that have lost the region and varying amounts of flanking S1 and S2 sequences (KERMICLE 1996 ). In this experiment, a test of secondary paramutation was used to assess the paramutability of these -deleted alleles. Secondary paramutation refers to the phenomenon that paramutant r1 alleles become weakly paramutagenic; prior to paramutation, these alleles are not paramutagenic. In this test, the potentially paramutable -deleted alleles were kept heterozygous with a paramutagenic allele for three generations, then outcrossed to a paramutationally neutral allele, then crossed to the paramutable R-r allele to determine if they have acquired paramutagenicity. KERMICLE found that all but one of the -deleted alleles failed to acquire paramutagenicity. Earlier tests on a similar, but molecularly undefined, set of alleles had found that none were able to acquire paramutagenicity (BROWN 1966 ). Only the r-r:N1-3-1 allele acquired any paramutagenicity, and it was severely compromised in its ability to become paramutagenic relative to the standard paramutable allele, R-r (KERMICLE 1996 ). The r-r:N1-3-1 allele has the smallest deletion of the alleles tested: it has a nearly precise deletion of with only 26 bp of the left end of retained (WALKER et al. 1995 ). The remainder of the complex is intact. The finding that r-r:N1-3-1 has lost most of its ability to acquire secondary paramutagenicity implies that the region has a role in causing paramutability of R-r. The next smallest deletion is found in the r-r:N1-3-2 allele which has a deletion that removes all but the right-most 18 bp of from the complex as well as an additional 1578 bp of S1 (WALKER et al. 1995 ). Removal of this region takes away even the minimal ability to acquire secondary paramutagenicity that was retained by r-r:N1-3-1. Presumably, in the r-r:N1-3-2 derivative, the remaining sequences necessary for acquisition of secondary paramutagenicity have been deleted.

The aim of the experiments presented here is to examine whether r1 paramutation is associated with changes in the level of C-methylation of the paramutable R-r complex. Changes were detected and were mapped to a region of the complex within the highly paramutable S1 and S2 genes flanking the region. To examine whether has a role in eliciting these C-methylation changes, the r-r:N1-3-1 allele was used. This allele has a deletion of most of the doppia-containing region, but retains the multiple homologous genic elements of the R-r complex (i.e., the P, q, S1, and S2 genes). The failure of r-r:N1-3-1 to become methylated in a context that elicits strong C-methylation of intact R-r suggests that the region is needed for induction of C-methylation during paramutation, and that the duplicated gene segments at R-r do not themselves trigger C-methylation in r1 paramutation.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic stocks:
All stocks were maintained in the W22 inbred background. All stocks are homozygous dominant for the a1, a2, c1, c2, bz1, and bz2 genes necessary for anthocyanin synthesis in aleurone and homozygous recessive for the pl1 and b1 genes (see DOONER et al. 1991 for descriptions). The R-r:standard allele used in this study confers strong pigmentation of the aleurone of the seed, the coleoptile, the roots and leaf tip of seedlings, and to the roots and anthers of mature plants, and has been described previously (STADLER 1948 ; WALKER et al. 1995 ). The R-mb allele confers strong pigmentation to the aleurone and embryo in a "marbled" pattern of dark blotches on a colorless background, and also weakly pigments the coleoptile (WEYERS 1961 ). The isolation, characterization, and structure of derivative allele r-r:N1-3-1 has been described previously (KERMICLE and AXTELL 1981 ; WALKER et al. 1995 ). The recessive (r) tester allele used was r-g:Stadler ; it confers no pigmentation to any part of the plant or seed.

Derivation of paramutant R-r :
First-generation heterozygotes were generated in 1990, 1991, and 1993 by crossing homozygous R-r plants with homozygous R-mb plants. Second-generation heterozygotes were generated in 1991, 1992, and 1994 by crossing each of these three lineages of first-generation heterozygotes with homozygous R-mb plants. Third-generation heterozygotes were generated in 1992 and 1995 by crossing two independent second-generation heterozygote lineages with homozygous R-mb plants. For the studies presented here, at least two plants from each of the three second-generation heterozygote lineages and each of the two third-generation heterozygote lineages were reciprocally crossed to a homozygous r tester strain. These crosses resulted in six families segregating r/R-r (second-generation paramutants) and four families segregating r/R-r ''' (third-generation paramutants). Homozygous R-r and heterozygous r/R-r stocks that serve as controls for basal levels of methylation were grown at the same time and under the same conditions (field or greenhouse) as the paramutant stocks to which they were compared.

Preparation of DNA and genomic blotting:
For analysis of methylation in first-generation heterozygotes, DNA was prepared from leaves of mature, flowering plants homozygous for R-r or heterozygous (first generation) for R-r/R-mb. For analysis of the standard pattern of methylation for R-r ', DNA samples were prepared from the third leaf of mature, flowering plants of genotype r/R-r (six families) and r/R-r ''' (four families) according to the protocol of DELLAPORTA 1994 . For methylation analysis of the r-r :N1-3-1 allele, DNA was prepared from leaves of three-week-old seedlings of second-generation r-r:N1-3-1 '/R-mb heterozygotes and from leaves of three-week-old seedlings of second-generation R-r '/R-mb heterozygotes. DNA samples were digested with a 5–10-fold excess of a non-methylation-sensitive restriction enzyme, and were then purified by organic extraction and ethanol precipitation. Six µg of DNA (or 3 µg from homozygous stocks) were then digested for 18–22 hr with a 5–10-fold excess of the methylation-sensitive restriction enzymes. Agarose gel electrophoresis and blotting were performed as described previously (ROBBINS et al. 1991 ). Quantitation of the percentage of sites methylated was accomplished using ImageQuant (Molecular Dynamics, Inc., Sunnyvale, CA) software analysis of phosphorimager scans.

Description of DNA probes:
Probes were used that detect the promoter regions and first seven exons of each gene at the R-r complex. The probes used were: pR-nj:1 (ROBBINS et al. 1991 ), which detects the promoter and/or first exon of P, q, S1 and S2; 1011 (WALKER et al. 1995 ), an S-specific probe that hybridizes to the region between the S1 and S2 genes; SAH (WALKER et al. 1995 ), which detects the 5' portion of the second intron of P, S1, and S2; and cDNA-B, a probe derived from position 812 to 1334 in the Sn cDNA (CONSONNI et al. 1992 ) that detects r1 exons 4 through 7. To detect specifically the r-r:N1-3-1 allele, the probe N131 was generated from primers (oN131-L: 5'-CAGGAAACGACTAAATTTGCC-3' and oN131-R: 5'-GCAACGAAGCATCAGTAT-3') flanking the filler DNA in the r-r:N1-3-1 derivative. The N131 probe was generated by amplification from a plasmid clone containing the relevant region of r-r:N1-3-1 (WALKER et al. 1995 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Paramutation results in increased C-methylation of R-r :
In many cases of epigenetic gene silencing, changes in the level of C-methylation of the affected gene are observed. To determine whether this correlation is observed in paramutation at r1, the methylation status of the paramutable allele R-r was examined both before and after making it heterozygous with the paramutagenic allele R-mb. DNA was extracted from R-r homozygotes and R-r/R-mb heterozygotes, digested with the non-methylation-sensitive enzyme, HindIII, and then with methylation-sensitive restriction enzymes known to cut near the region of R-r. Genomic blots were prepared and hybridized with the probe 1011 which detects the region of R-r but does not hybridize to R-mb DNA. Representative blots are shown in Figure 2. The level of C-methylation of the paramutable R-r allele, measured by resistance to cleavage by methylation-sensitive restriction enzymes, increases when R-r is heterozygous with the paramutagenic R-mb allele.

View larger version (39K): In this window In a new window Download PPT slide   Figure 2. —Methylation changes at R-r during paramutation. DNA samples from naive (nonparamutant) homozygous R-r (lanes 1 and 2 of each panel) and from first-generation heterozygous R-r/R-mb individuals (lanes 3 and 4 of each panel) were digested to completion with the non-methylation-sensitive restriction enzyme, HindIII followed by digestion with the methylation-sensitive restriction enzymes, BstUI (A), HaeII (B), and HpaII (C), which have recognition sites near the region of R-r. The DNA probe used was 1011, an S-specific probe that hybridizes to the region between the S1 and S2 genes and does not recognize R-mb.

Analysis of the methylation status of R-r in first-generation R-r/R-mb heterozygotes is limited by the ability to distinguish between the r1 genes of the R-mb complex and the r1 genes of the R-r complex. The only region of R-r that can be clearly distinguished from R-mb is the region. Thus, in order to determine the extent of methylation of R-r, it must be segregated from R-mb by sexual crossing of the heterozygote to a tester allele that is more readily distinguished from R-r on genomic blots. To do this, the R-r complex was kept heterozygous with the paramutagenic R-mb complex for either two or three generations, and was then outcrossed via pollen to a homozygous recessive (r) tester strain. DNA was made from leaves of the resulting r/R-r (second-generation paramutant), r/R-r ''' (third-generation paramutant) plants, and from R-r/R-r (nonparamutant), and r/R-r (nonparamutant) individuals. Southern blots were prepared and hybridized with probes that allow mapping of sites within particular genes of the R-r complex. At least two individuals from each of 10 families (see MATERIALS AND METHODS for derivation of these stocks) were examined with at least two methylation-sensitive enzymes known to cut within the genic portions of R-r.

A typical set of blots is shown in Figure 3. In A–C, the probe used hybridizes to the region of R-r. With each C-methylation-sensitive enzyme used, there is an increase in the level of C-methylation of sites close to the region in paramutant versus nonparamutant R-r. Thus, the C-methylation that occurred in the heterozygote was retained following outcrossing. Every r/R-r and r/R-r ''' individual tested in all 10 families (not shown) exhibited increases in its level of C-methylation in this region relative to naive (nonparamutant) R-r. D–F show the same three blots hybridized with a coding region probe (cDNA-B) that detects all three of the intact components of R-r : P, S1, and S2. No increase in the level of methylated sites is evident in the coding regions detected by cDNA-B in paramutant versus nonparamutant R-r. These results suggest that C-methylation changes in paramutation are confined either to the 5' regions of the genes of the complex and/or are restricted to the S genes. Further mapping experiments (see below) also support this idea.

View larger version (55K): In this window In a new window Download PPT slide   Figure 3. —Methylation analysis of paramutant R-r. (A–C) DNA samples were digested to completion with the non-methylation-sensitive restriction enzyme HindIII, followed by digestion with the methylation-sensitive restriction enzymes AvaI (A), PstI (B), and PvuII (C). Lanes 1 and 2 are homozygous R-r ; lanes 3 and 4 are r/R-r heterozygotes; lanes 5–8 (or 5–7 in B) are r/R-r heterozygotes; lanes 9 and 10 (or 8 and 9 in B) are r/R-r ''' heterozygotes. The DNA probe used was 1011, an S-specific probe that hybridizes to the region between the S1 and S2 genes. (D–F) The blots shown in A, B, and C were stripped of the original probe and rehybridized with the probe cDNA-B which detects the coding region of each intact gene at R-r (P, S1, and S2). This probe also detects the r tester allele used in these studies; the hybridizing r1 fragments may (E) or may not (D and F) comigrate with fragments from the R-r complex.

C-methylation changes map to the 5' portions of the S genes:
Because the R-r complex comprises both genes that are strongly affected by paramutation (the S genes) and a gene which is only weakly affected by paramutation (the P gene) (BRINK 1956 ; BRINK and MIKULA 1958 ; BROWN 1966 ), it provides an interesting opportunity to examine whether changes in C-methylation are localized to the strongly affected components. Three independently derived r/R-r individuals and two independently derived r/R-r ''' individuals were used for detailed mapping. The maps of each R-r component and the methylation status of the C-methylation-sensitive restriction sites on these maps are shown in Figure 4. The only region of R-r that exhibits changes in the level of C-methylation is a 3.4-kb region flanking and including approximately 1.5 kb of the transcribed portion of each gene. No changes in the level of methylation were observed in the P or q genes of the R-r complex, although it should be noted that sites within the 5' regions of these genes were methylated at all times. In order to detect changes in the P and q genes, those changes would have to involve decreases in C-methylation levels, or, in the case of the P gene, the changes would have to occur in the normally unmethylated downstream coding portions. Thus paramutation is correlated with a striking increase in C-methylation, but only in the 5' regions of the strongly affected S1 and S2 genes.

View larger version (13K): In this window In a new window Download PPT slide   Figure 4. —Methylation map of the R-r complex. The methylation status at sites for the methylation-sensitive restriction enzymes AvaI (A), PstI (P), PvuI (P1), PvuII (P2), MluI (M), and NruI (N) were measured from r/R-r and r/R-r ''' plants. The probes used were pR-nj:1 (ROBBINS et al. 1991 ), 1011 (WALKER et al. 1995 ), SAH (WALKER et al. 1995 ), and cDNA-B which together allow detection of the promoter regions and first seven exons (indicated by black boxes) of each gene at the R-r complex. The accumulated results are represented as restriction maps of each component, P, q, S1, and S2. Circles above the respective restriction sites summarize the methylation profile for each site in paramutant (upper row of circles) and nonparamutant (lower row of circles) R-r. Open circles indicate that fewer than 25% of the sites were methylated. Half-filled circles indicate that 25–50% of the sites were methylated. Closed circles indicate that greater than 50% of the sites were methylated.

Deletion of results in loss of paramutation associated C-methylation changes:
The observation that C-methylation changes in paramutation are localized to the area flanking the transposon-derived region suggests that this region may play a role in inducing C-methylation changes during paramutation. A spontaneous mutant derivative of R-r that has a nearly precise deletion of was used to address the question of whether the region is necessary for C-methylation to take place. This derivative, r-r:N1-3-1 (KERMICLE and AXTELL 1981 ) has a 361-bp deletion spanning from the right end proximally to a position 26 bp from the left end of (Figure 5) (WALKER et al. 1995 ). Thus, r-r:N1-3-1 retains 26 bp of the left end of but the remainder of has been deleted. The rest of the complex, i.e., the P gene, the q gene, and the coding portions of the S1 and S2 genes, remains intact (ROBBINS et al. 1991 ; WALKER et al. 1995 ). At the breakpoint within r-r:N1-3-1, an 84-bp stretch of filler DNA has been inserted. This filler DNA is derived from adjacent sequences within S1 and S2 (WALKER et al. 1995 ). The r-r:N1-3-1 derivative confers no pigmentation to the aleurone presumably due to the lack of its promoter, . This allele has been shown to be severely compromised in its ability to acquire secondary paramutagenicity (KERMICLE 1996 ). It is possible to examine whether r-r:N1-3-1 acquires C-methylation following heterozygosity with a paramutagenic allele, thus testing whether the repeated genic elements of R-r (the P, q, S1, and S2 genes) can trigger C-methylation in response to heterozygosity with a paramutagenic allele.

View larger version (14K): In this window In a new window Download PPT slide   Figure 5. —Structure of the r-r:N1-3-1 derivative. Above (and not to scale) are shown the genic components of R-r and its derivative, r-r:N1-3-1. Below is the restriction map of and its flanking DNA in the S1 and S2 genes. Restriction sites shown are AvaI (A), BstUI (B), PvuII (P), and HpaII (H). The r-r:N1-3-1 allele arose spontaneously from hemizygous R-r (KERMICLE and AXTELL 1981 ). This derivative has retained the P, q, S1, and S2 genes of the complex, but has suffered a deletion of the region (white box with arrows representing multiple subterminal repeat elements of doppia) between the S1 and S2 genes (ROBBINS et al. 1991 ). The deletion includes all but the left-most 26 base pairs of (white box). Between the deletion breakpoints, filler DNA (gray box) has been inserted. This filler DNA is derived from nearby S2 sequences (WALKER et al. 1995 ). Note that the r-r:N1-3-1 allele has gained a BstUI site close to the left end of the filler DNA.

The r-r:N1-3-1 derivative was maintained as a heterozygote with R-mb for two generations. Even though increased C-methylation can be detected (see Figure 1) in first generation heterozygotes, methylation is stronger following several generations of paramutation (E. L. WALKER, unpublished results, and see below). Thus, using two generations of heterozygosity gives a better chance of observing weak methylation changes at r-r:N1-3-1, if such changes occur. DNA was prepared from these second-generation r-r:N1-3-1'/R-mb heterozygotes. DNA from a second-generation R-r '/R-mb heterozygote was used as a control to ensure that C-methylation of R-r ' is readily detectable under these heterozygous conditions. All DNAs were subjected to Southern analysis as described above. The results are shown in Figure 6. The relevant portion of S1 and S2 in the control R-r '/R-mb heterozygotes is heavily methylated, as shown in Figure 5A. When the same sites were tested in r-r:N1-3-1'/R-mb heterozygotes (Figure 5B), no evidence of C-methylation was observed. Thus, removal of the region renders the complex ineffective in acquiring C-methylation during paramutation. The multiple repeated structure of R-r, which is maintained in the r-r:N1-3-1 derivative, does not, by itself, predispose the complex to methylation during paramutation.

View larger version (55K): In this window In a new window Download PPT slide   Figure 6. —C-methylation analysis of the -deletion mutant r-r:N1-3-1. (A and B) As a control for the ability to detect C-methylation in R-mb heterozygotes, DNA from second-generation R-r '/R-mb heterozygotes (lanes 1 and 2) and homozygous nonparamutant R-r (lanes 3 and 4), and was digested with HindIII and either BstUI (A) or PvuII (B). Blots were prepared and hybridized to the 1011 probe (WALKER et al. 1995 ), which does not hybridize to genomic DNA from R-mb homozygotes (not shown). (C–F) DNA was prepared from second-generation r-r:N1-3-1'/R-mb heterozygotes (lanes 1–4) and homozygous r-r:N1-3-1 (lane 5), and was digested with HindIII and one of the methylation-sensitive restriction enzymes, AvaI (C), BstUI (D), HpaII (E), or PvuII (F). Blots were prepared and hybridized to the N131 probe which specifically detects the filler DNA of the r-r:N1-3-1 derivative.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Paramutation of the R-r complex is correlated with increases in its level of C-methylation. The region of R-r was observed to have increased C-methylation while in its first generation of heterozygosity with the paramutagenic R-mb allele. Thus, methylation of R-r can be detected even before the paramutant phenotype of decreased pigmentation in aleurone, which is not observed without sexual transmission of R-r' into a subsequent generation. The methylation changes associated with paramutation of R-r are very consistent: in every second- and third-generation paramutant plant tested in over 10 different families, paramutation was always correlated with detectable increases in C-methylation levels. These changes, however, are not found throughout the R-r complex. They occur only in specific portions of the S1 and S2 genes. C-methylation was found in other parts of the complex, most notably in the promoter of the P gene, and the q gene fragment, but methylation at P and q was not observed to change following paramutation. Thus the correlation of paramutation and increased C-methylation exists even within the R-r complex, since the highly paramutable S genes show obvious methylation changes while the much less paramutable P gene was never found to show increased C-methylation under the conditions used in this study. With regard to changes in C-methylation levels, r1 paramutation is distinct from two other maize paramutation systems, b1 and pl1, in which methylation changes are not found following paramutation. The correlation with increased C-methylation is shared with other meiotically stable silencing phenomena, e.g., paramutation of A1 transgenes (MEYER et al. 1993 ), meiotically stable silencing associated with the transgenic H locus (MATZKE et al. 1994 ), and with silencing of PAI loci in Arabidopsis (BENDER and FINK 1995 ).

Paramutation has been compared to other epigenetic gene silencing phenomena, most notably cases that occur via transcriptional inactivation (FLAVELL 1994 ; JORGENSEN 1995 ; MATZKE and MATZKE 1995 ; MEYER and SAEDLER 1996 ). These silencing phenomena have been termed "homology-dependent gene silencing" events (MATZKE and MATZKE 1995 ) since all are correlated with the presence of multiple homologous copies of a gene or promoter. Because both the paramutable R-r complex (ROBBINS et al. 1991 ; WALKER et al. 1995 ) and the paramutagenic R-mb (E. L. WALKER, unpublished results) and R-st (EGGLESTON et al. 1995 ) complexes contain multiple r1 genes, paramutation at r1 has been regarded as being consistent with the idea that multiple gene copies can lead to epigenetic silencing (ASSAD et al. 1993 ; MATZKE et al. 1996 ).

Recently, an alternative explanation has been suggested: that transposable elements, which make up a large part of the moderately repetitive DNA in plant genomes and which are often located in the upstream portions of normal genes, could be responsible for making genes nearby susceptible to epigenetic silencing events (MARTIENSSEN 1996A ; MATZKE et al. 1996 ). In the case of silenced transgenes, the expectation is that they would have inserted near or within transposable elements, thus coming under their epigenetic influence. Stably active transgenes would have inserted at positions where there are no transposable elements nearby. This idea has also been applied to naturally occurring systems like r1 paramutation. In this case, residual transposon sequences at might be able to influence the epigenetic state of the R-r complex.

Mutant derivatives of R-r exist which have deletions of most or all of and, usually, portions of the S1 and/or S2 genes. These derivatives can be used to address the relative contribution of the gene repeats of R-r vs. region of R-r. The r-r:N1-3-1 allele is particularly informative in addressing the role of , since it has lost all but 26 bp of the left inverted repeat of (WALKER et al. 1995 ). Using a secondary paramutation test, KERMICLE 1996 has reported that an r-r:N1-3-1 allele that had been heterozygous for three generations with the strong paramutagenic allele, R-st, became very weakly paramutagenic, thus indicating that paramutation of r-r:N1-3-1 had occurred, but at greatly reduced efficiency compared to R-r. In the study presented here, r-r :N1-3-1 was used to examine whether C-methylation changes would occur when the region was absent. No C-methylation of the S genes in r-r:N1-3-1 was detected in second-generation r-r:N1-3-1'/R-mb heterozygotes. Second-generation R-r '/R-mb heterozygotes used as a positive control showed obvious methylation of the same sites that were unmethylated in r-r:N1-3-1. In comparison to KERMICLE's secondary paramutation test of r-r:N1-3-1, two important differences must be noted. KERMICLE's test was carried out using a paramutagenic allele, R-st, that is stronger than the R-mb allele used in the present study. Furthermore, in KERMICLE's experiment, three generations of heterozygosity were used before testing for secondary paramutation. It is quite possible that if C-methylation status of r-r:N1-3-1 were tested under these conditions, some C-methylation might be found. Clearly, though, the level of C-methylation observed in paramutagenized r-r:N1-3-1 is far smaller (in the present study undetectable) than what is routinely observed for R-r. Thus, just as deletion of most of severely compromises the ability of the complex to acquire secondary paramutagenicity, it also severely compromises the ability of the complex to become methylated.

Taken together, these results are inconsistent with models in which the highly duplicated structure of the R-r complex is the sole factor in targeting a gene to be silenced in paramutation. Instead, the sequences derived from the transposable element doppia appear to be responsible for the effects associated with paramutability. These results do not rule out the possibility that both repeated sequences and doppia sequences must be present to cause paramutability. Alleles with a simple S gene regulated by do not exist to allow this question to be addressed. Simple P-only alleles of r1 are not paramutable, but they contain neither repeats nor , and so do not satisfactorily address the question either.

The region is required for transcription of the S genes. The r-r:N1-3-1 derivative is rendered transcriptionally incompetent due to deletion of the region. Thus, one interpretation of these results is that transcriptional competence is a requirement for paramutation. While this possibility cannot be ruled out, it seems unlikely since the S genes of R-r are expressed only in aleurone, a part of the endosperm (WALKER et al. 1995 ). Methylation changes at R-r are readily observed in nonendosperm lineages where the S genes are normally silent. It seems more likely that the particular complement of factors normally bound at the region, even while it is transcriptionally inactive, would be critical for targeting de novo methylation in paramutation.

The region contains sequences derived from a doppia transposable element, some of which could potentially bind regulatory factors that normally regulate transposable elements. Epigenetic regulation of transposable elements in maize is well documented (see FEDOROFF 1996 and MARTIENSSEN 1996B for reviews). Furthermore, the doppia transposon whose remnants are found at , is a member of the same superfamily of transposable elements as Spm, an element with extremely well documented epigenetic behavior (FEDOROFF 1996 ). One of the hallmarks of epigenetic silencing of Spm is C-methylation of certain regions (FEDOROFF 1996 ). It is possible that 's role in the C-methylation of the S genes during paramutation reflects a function of the doppia transposon sequences in . Changes in Spm methylation can be elicited by an Spm-encoded protein, TNPA (SCHLAPPI et al. 1994 ), that binds to the subterminal repeat elements of Spm (GIERL et al. 1988 ; TRENTMANN et al. 1993 ). The Spm subterminal repeats are similar in both arrangement and sequence to the subterminal repeat elements from doppia (FEDOROFF 1996 ; WALKER et al. 1995 ). It is tantalizing to speculate that the doppia subterminal repeats found at could have a role in making R-r susceptible to paramutation.

	ACKNOWLEDGMENTS

I thank STEVE DELLAPORTA, ALISON DELONG, TOM BRUTNELL, RANDY PHILLIS, and ANNE SIMON for their comments on this manuscript. This work was supported by the National Science Foundation (NSF9406483).

Manuscript received September 25, 1997; Accepted for publication December 8, 1997.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ASSAD, F., K. L. TUCKER, and E. R. SIGNER, 1993  Epigenetic repeat-induced gene silencing (RIGS) in Arabidopsis.. Plant Molecular Biology 22:1067-1085[Medline].

BENDER, J. and G. R. FINK, 1995  Epigenetic control of an endogenous gene family is revealed by a novel blue fluorescent mutant of Arabidopsis. Cell 83:725-734[Medline].

BRINK, R. A., 1956  Change associated with the R locus in maize is directed and potentially reversible. Genetics 41:872-889[Free Full Text].

BRINK, R. A., 1958  Paramutation at the R locus in maize. CSH Symp. Soc. Quant. Biol. 23:379-391.

BRINK, R. A., 1973  Paramutation. Annu. Rev. Genet. 7:129-152[Medline].

BRINK, R. A. and B. MIKULA, 1958  Plant color effects of certain anomalous forms of the R-r allele in maize. Z. ind. Abstamm. Vererb-lehre 89:94-102.

BROWN, D. F., 1966  Paramutability of R-g and r-r mutant genes derived from an R-r allele in maize. Genetics 54:899-910[Free Full Text].

CHANDLER, V. L., K. M. KUBO and J. B. HOLLICK, 1996 b and pl paramutation in maize: heritable transcription states programmed during development, pp. 289–304 in Epigenetic Mechanisms of Gene Regulation, edited by V. E. A. RUSSO, R. A. MARTIENSSEN and A. D. RIGGS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

COCCIOLONE, S. M. and K. C. CONE, 1993  Pl-Bh, an anthocyanin regulatory gene of maize that leads to variegated pigmentation. Genetics 135:575-588[Abstract].

COE, E. H., 1966  The properties, origin, and mechanism of conversion-type inheritance at the B locus in maize. Genetics 53:1035-1063[Free Full Text].

CONE, K. C., S. M. COCCIOLONE, C. A. MOEHLENKAMP, T. WEBER, and B. J. DRUMMOND et al., 1993  Role of the regulatory gene pl in the photocontrol of maize anthocyanin pigmentation. Plant Cell 5:1807-1816[Abstract].

CONSONNI, G., A. VIOTTI, S. L. DELLAPORTA, and C. TONELLI, 1992  DNA nucleotide sequence of Sn, a regulatory gene in maize. Nucleic Acids Res. 20:373[Free Full Text].

DELLAPORTA, S., 1994 Plant DNA Miniprep and Microprep: Versions 2.1–2.3, pp. 522–525 in The Maize Handbook, edited by M. FREELING, and V. WALBOT. Springer-Verlag, NY.

DOONER, H. K., R. JORGENSON, and T. P. ROBBINS, 1991  Genetic and developmental control of anthocyanin biosynthesis. Annu. Rev. Genet. 25:173-199[Medline].

EGGLESTON, W. B., M. ALLEMAN, and J. L. KERMICLE, 1995  Molecular organization and germinal instability of R-stippled maize. Genetics 141:347-358[Abstract].

FEDOROFF, N., 1996 Epigenetic regulation of the maize Spm transposable element, pp. 575–592 in Epigenetic Mechanisms of Gene Regulation, edited by V. E. A. RUSSO, R. A. MARTIENSSEN and A. D. RIGGS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

FLAVELL, R. B., 1994  Inactivation of gene expression in plants as a consequence of specific sequence duplication. Proc. Natl. Acad. Sci. USA 91:3490-3496[Abstract/Free Full Text].

GIERL, A., S. LUTTICKE, and H. SAEDLER, 1988  TnpA product encoded by the transposable element En-1 of Zea mays is a DNA binding protein. EMBO J. 7:4045-4053[Medline].

GOFF, S. A., T. M. KLEIN, B. A. ROTH, M. E. FROMM, and K. C. CONE et al., 1990  Transactivation of anthocyanin biosynthetic genes following transfer of B regulatory genes into maize tissues. EMBO J. 9:2517-2522[Medline].

HOLLICK, J. B., G. I. PATTERSON, E. H. J. COE, K. C. CONE, and V. L. CHANDLER, 1995  Allelic interactions heritably alter the activity of a metastable maize pl allele. Genetics 141:709-719[Abstract].

JORGENSEN, R. A., 1995  Cosuppression, flower color patterns, and metastable gene expression states. Science 268:686-691[Abstract/Free Full Text].

KERMICLE, J. L., 1970  Dependence of the R-mottled phenotype in maize on the mode of sexual transmission. Genetics 66:69-85[Free Full Text].

KERMICLE, J., 1996 Epigenetic silencing and activation of a maize r gene, pp. 267–287 in Epigenetic Mechanisms of Gene Regulation, edited by V. E. A. RUSSO, R. A. MARTIENSSEN and A. D. RIGGS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

KERMICLE, J. L. and J. D. AXTELL, 1981  Modification of chlorophyll striping by the R region. Maydica 26:185-197.

LUDWIG, S. R., L. F. HABERA, S. L. DELLAPORTA, and S. R. WESSLER, 1989  Lc, a member of the maize R gene family responsible for tissue-specific anthocyanin production, encodes a protein similar to transcriptional activators and contains the myc-homology region. Proc. Natl. Acad. Sci. USA 86:7092-7096[Abstract/Free Full Text].

MARTIENSSEN, R., 1996a  Epigenetic phenomena: paramutation and gene silencing in plants. Curr. Biol. 6:810-813[Medline].

MARTIENSSEN, R. A., 1996b Epigenetic silencing of Mu transposable elements in maize, pp. 593–608 in Epigenetic Mechanisms of Gene Regulation, edited by V. E. A. RUSSO, R. A. MARTIENSSEN and A. D. RIGGS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

MATZKE, M. A. and A. J. M. MATZKE, 1995  How and why do plants inactivate homologous (trans)genes? Plant Physiol. 107:679-685[Medline].

MATZKE, A. J. M., F. NEUHUBER, Y.-D. PARK, P. F. AMBROS, and M. A. MATZKE, 1994  Homology-dependent gene silencing in transgenic plants: epistatic silencing loci contain multiple copies of methylated transgenes. Mol. Gen. Genet. 244:219-229[Medline].

MATZKE, M. A., A. J. M. MATZKE, and W. B. EGGLESTON, 1996  Paramutation and transgene silencing: a common response to invasive DNA? Trends Pl. Sci. 1:382-388.

MEYER, P. and H. SAEDLER, 1996  Homology dependent gene silencing in plants. Ann. Rev. Pl. Physiol. Pl. Mol. Biol. 47:23-48.

MEYER, P., I. HEIDMANN, and I. NIEDENHOF, 1993  Differences in DNA methylation are associated with a paramutation phenomenon in transgenic petunia. Plant J. 4:86-100.

PATTERSON, G. I., K. M. KUBO, T. SHROYER, and V. L. CHANDLER, 1995  Sequences required for paramutation of the maize b gene map to a region containing the promoter and upstream sequences. Genetics 140:1389-1406[Abstract].

ROBBINS, T. P., E. L. WALKER, J. L. KERMICLE, M. ALLEMAN, and S. L. DELLAPORTA, 1991  Meiotic instability of the R-r complex arising from displaced intragenic exchange and intrachromosomal rearrangement. Genetics 129:271-283[Abstract].

RONCHI, A., K. PETRONI, and C. TONELLI, 1995  The reduced expression of endogenous duplications (REED) in the maize R family is mediated by DNA methylation. EMBO J. 14:5318-5328[Medline].

SCHLAPPI, M., R. RAINA, and N. FEDOROFF, 1994  Epigenetic regulation of the maize Spm transposable element: Novel activation of a methylated promoter by TnpA. Cell 77:427-437[Medline].

STADLER, L. J., 1948  Spontaneous mutation at the R locus in maize. I. The aleurone-color and plant-color effects. Genetics 31:377-394.

STYLES, E. D. and R. A. BRINK, 1966  The metastable nature of paramutable R alleles in maize. I. Heritable enhancement in level of standard R-r action. Genetics 54:433-439[Free Full Text].

STYLES, E. D. and R. A. BRINK, 1968  The metastable nature of paramutable R alleles in maize. IV. Parallel enhancement of R action in heterozygotes with r and in hemizygotes. Genetics 61:801-811.

TRENTMANN, S. M., H. SAEDLER, and A. GIERL, 1993  The transposable element En/Spm-encoded TNPA protein contains a DNA binding and a dimerization domain. Mol. Gen. Genet. 238:201-208[Medline].

UPADHYAYA, K. C., H. SOMMER, E. KREBBERS, and H. SAEDLER, 1985  The paramutagenic line niv-44 has a 5 kb insert, Tam 2, in the chalcone synthase gene of Antirrhinum majus.. Mol. Gen. Genet. 199:201-207.

WALKER, E. L., T. P. ROBBINS, T. E. BUREAU, J. KERMICLE, and S. L. DELLAPORTA, 1995  Transposon mediated chromosomal rearrangements and gene duplications in the formation of the maize R-r complex. EMBO J. 14:2350-2363[Medline].

WEYERS, W. H., 1961  Expression and stability of the marbled allele in maize. Genetics 46:1061-1067[Free Full Text].