Rmr6 Maintains Meiotic Inheritance of Paramutant States in Zea mays

Abstract

Paramutation generates heritable changes affecting regulation of specific alleles found at several Zea mays (maize) loci that encode transcriptional regulators of anthocyanin biosynthetic genes. Although the direction and extent of paramutation is influenced by poorly understood allelic interactions occurring in diploid sporophytes, two required to maintain repression loci (rmr1 and rmr2), as well as mediator of paramutation1 (mop1), affect this process at the purple plant1 (pl1) locus. Here we show that the rmr6 locus is required for faithful transmission of weakly expressed paramutant states previously established at both pl1 and red1 (r1) loci. Transcriptional repression occurring at both pl1 and booster1 (b1) loci as a result of paramutation also requires Rmr6 action. Reversions to highly expressed, nonparamutant states at both r1 and pl1 occur in plants homozygous for rmr6 mutations. Pedigree analysis of reverted pl1 alleles reveals variable latent susceptibilities to spontaneous paramutation in future generations, suggesting a quantitative nature of Rmr6-based alterations. Genetic tests demonstrate that Rmr6 encodes a common component required for establishing paramutations at diverse maize loci. Our analyses at pl1 and r1 suggest that this establishment requires Rmr6-dependent somatic maintenance of meiotically heritable epigenetic marks.

MEIOTICALLY heritable alterations in gene regulation conditioned by specific allelic interactions are known as paramutations (Brink 1958). At least four distinct examples of paramutations occur in Zea mays (maize) (Hollick et al. 1997; Chandler et al. 2000; Chandler and Stam 2004). Specific alleles exhibit this unusual inheritance behavior at the red1 (r1), booster1 (b1), pericarp1 (p1), and purple plant1 (pl1) loci (Brink 1956; Coe 1966; Das and Messing 1994; Hollick et al. 1995), all of which encode transcriptional regulatory proteins of flavonoid biosynthetic pathways. In general, these unique alleles can exist in a dynamic range of regulatory states manifest as heritable differences in visual patterns and levels of colorful red or purple pigments. Both the heritable switching and the stability of these states are influenced by the homologous allele in diploid sporophytes (Coe 1966; Styles and Brink 1966; Das and Messing 1994; Hollick and Chandler 1998). Such allelic interactions affecting inheritance of regulatory information has important implications for breeding efforts and represents a potential novel resource of heritable variation in plants (Kermicle and Alleman 1990; Hollick and Chandler 1998).

The pl1 locus encodes a R2R3 MYB domain protein (Cone et al. 1993) that, in combination with basic helix-loop-helix domain proteins encoded by either b1 or r1 loci, promotes transcription of genes encoding enzymes required for anthocyanin biosynthesis (Goff et al. 1992). The Pl1-Rhoades allele confers the highest pigment levels of any previously described pl1 allele. The high-expression reference state selected in culture (denoted Pl-Rh) is, however, unstable and can spontaneously change to weaker expression states referred to as Pl′ (Figure 1A; Hollick et al. 1995). A continuum of Pl1-Rhoades expression states can be achieved and is quantified using a visual 1–7 graded anther color score (ACS; Figure 1B; Hollick et al. 1995). Different pl1 RNA levels are directly correlated with these anther phenotypes and are considered to reflect differences in transcription rates as measured from isolated husk nuclei (Hollick et al. 2000).

Figure 1.

Dynamic states of the Pl1-Rhoades allele. (A) Isogenic siblings displaying Pl-Rh and Pl′ anther phenotypes conferred by Pl-Rh and Pl′ states of the Pl1-Rhoades allele, respectively. (B) Anther pigment phenotype examples corresponding to 1–7 graded ACS. (C) Generalized behaviors of Pl-Rh and Pl′ regulatory states (Hollick et al. 1995; Hollick and Chandler 1998). Solid arrow indicates invariant changes while dashed arrows indicate changes occurring at ∼10% frequency.

Pigment levels produced from various Pl1-Rhoades expression states are inversely correlated with levels of an experimentally defined activity—termed paramutagenicity—that facilitates heritable changes of the homologous Pl1-Rhoades expression state in trans (Hollick et al. 1995). Pl′ states representing ACS 1–4 classes are highly paramutagenic to the Pl-Rh (ACS 7) reference state: when Pl1-Rhoades alleles of both Pl′ and Pl-Rh states are combined in a diploid sporophyte, only Pl1-Rhoades alleles of ACS 1–4 Pl′ states are sexually transmitted (Figure 1C; Hollick et al. 1995). This is the general behavior of paramutation; certain allelic interactions lead to directed, and heritable, changes in gene action. A similar example at the sulfurea locus of Lycopersicum esculentum affecting chlorophyll levels (Hagemann 1969) indicates that paramutation is not confined to maize or to regulators of flavonoid biosynthesis. Parallels noted among diverse examples of non-Mendelian inheritance patterns suggest that paramutation may be a general feature of many eukaryotic genomes (Chandler and Stam 2004).

Several trans-acting components of the paramutation process have been identified by mutational analysis. Recessive mutations in mediator of paramutation1 (mop1; Dorweiler et al. 2000) and required to maintain repression1 and -2 (rmr1, rmr2; Hollick and Chandler 2001) lead to elevated pl1 RNA levels and corresponding increases in pigmentation from Pl1-Rhoades alleles of Pl′ state, suggesting involvement of Mop1, Rmr1, and Rmr2 functions in somatic maintenance of transcriptional repression. Indeed, Mop1 aids somatic maintenance of transcriptional repression of paramutant b1 alleles (B1-Intense in the B′ state) and is required to establish meiotically heritable paramutant states at both b1 and r1 loci (Dorweiler et al. 2000). One distinction of b1 paramutation is the remarkable stability of B′ states; regardless of somatic phenotype, B′ states are faithfully transmitted, even from plants homozygous for mop1 mutations (reviewed in Chandler et al. 2002). In contrast, Pl′ and paramutant R-r:standard (R-r′) are reversible (Styles and Brink 1966; Hollick and Chandler 1998). Pl1-Rhoades alleles of Pl′ state can heritably revert to fully active Pl-Rh in hemizygous condition (Hollick and Chandler 1998; J. Hollick, unpublished results) in heterozygous combination with structurally distinct pl1 alleles (Figure 1C; Hollick and Chandler 1998) or after passage through mop1 (Dorweiler et al. 2000), rmr1, or rmr2 homozygous mutant plants (Hollick and Chandler 2001). Mutant analyses of pl1 paramutation can therefore distinguish between functions required to maintain somatic repression and functions affecting meiotically transmissible information.

Although paramutation is often described as an example of homology-dependent gene silencing (Martienssen 1996; Matzke et al. 1996, 2002; Chandler and Stam 2004), it is important to recognize that paramutation at pl1 and r1 alters specific regulatory properties. The Pl1-Rhoades allele transitions from light-independent (Pl-Rh) to light-dependent (Pl′) regulatory modes, but transcription is still detectable from Pl′ states (Hollick et al. 2000). Similarly, the seed-specific component of R-r haplotypes is conditioned by paternal imprinting to produce a variable and irregular mottled pigment pattern to the kernel aleurone, and paramutation to the R-r′ state intensifies this effect (Kermicle 1970). In contrast, uniform seed pigmentation characterizes both R-r and R-r′ states following female transmission. The ability of both R-r:standard and Pl1-Rhoades to adopt a range of quantitatively discrete regulatory states combined with the highly reversible nature of these states strongly suggests that genetic mechanisms operate on R-r:standard and Pl1-Rhoades to define meiotically heritable epigenetic states. Herein we present a mutational analysis detailing functions of a novel maize locus, required to maintain repression6 (rmr6), encoding a key component of this genetic mechanism.

MATERIALS AND METHODS

Maize nomenclature:

Following established guidelines (http://www.agron.missouri.edu/maize_nomenclature.html), chromosomes, loci, alleles, and allelic states are all designated by italic type. Phenotypes conferred by specific allelic state genotypes are designated by non-italic type. Chromosome translocation breakpoints are designated “T,” while the same regions in structurally normal chromosomes are designated “+.” When known, specific alleles are designated with hyphenated extensions (i.e., Pl1-Rhoades). Dominant alleles either are abbreviated as “+” or begin with uppercase letters while recessive alleles are all lowercase. The prime symbol, “′,” refers to a paramutant allelic state. Sporophyte genotypes are written with pistillate (female)-derived chromosomes, alleles, and allelic states preceding those derived from staminate (male) parents. In previous publications, Pl-Rh and Pl′-mahogany (Pl′-mah or Pl′) denoted distinct pl1 alleles. Given the apparent lack of nucleotide polymorphisms, the continuum of quantitative expression states, and the high incidence of reversibility, the two extreme forms are discussed as representing alternative regulatory states of a single allele. Here, Pl1-Rh denotes the single pl1 allele that is able to exist in either Pl-Rh or Pl′ states of regulatory control.

Genetic stocks:

All stocks contain dominant and functional alleles required for anthocyanin pigment production in anther tissues unless otherwise noted. Specific b1, pl1, and r1 regulatory loci genotypes are noted since these directly affect pigment patterns used in stock syntheses and analyses. The Pl1-Rhoades and B1-Intense alleles used here derive from a full-color W23 line developed by Ed Coe, Jr. (USDA-ARS, University of Missouri, Columbia, MO) and maintained in several lines obtained from the Maize Genetics Cooperation Stock Center (USDA-ARS, University of Illinois, Urbana, IL) and Vicki Chandler (University of Arizona, Tucson, AZ). Elite inbred lines A619 and A632 were provided by the North Central Plant Introduction Station (USDA-ARS, Ames, IA). Pl1-Rhoades alleles of Pl-Rh state were introgressed into A619 and A632 inbred lines through recurrent backcrosses to the inbred lines as pistillate parents. Lines homozygous for Pl1-Rhoades alleles of Pl-Rh state, Pl-Rh/Pl-Rh (A619) and Pl-Rh/Pl-Rh (A632), were established with the following color factor genotypes: b1-A619, Pl1-Rhoades (Pl-Rh), R-r, (>98% A619); B1-A632, Pl1-Rhoades (Pl-Rh), R1-A632 (>93% A632). Spontaneous paramutation of Pl1-Rhoades from Pl-Rh to Pl′ states occurs in the A632-converted line at modest frequency and provided isogenic Pl′/Pl′ individuals used in experiments represented in Table 2. Similar isogenic siblings were crossed to Coe's full-color W23 line [B1-Intense (B-I); Pl1-Rhoades (Pl-Rh); R-r] to generate Pl-Rh/Pl-Rh and Pl′/Pl′ (W23/A632) hybrids used in one experiment described in Table 2. The Pl-Rh/Pl-Rh (W23/CO159) stock is a recombinant inbred line derived using the CO159 inbred and has the following color factor genotype: b1-CO159, Pl1-Rhoades (Pl-Rh), R-r. The source of the mutant alleles ems97406 and ems98225 is identical to that described in Hollick and Chandler (2001). The Pl′/Pl′ T6-9 translocation stock was derived by isolating recombination events between the Pl1-Rhoades allele and the 6L breakpoint present in the T6-9 (043-1; Longley 1961) reciprocal translocation. Recombination frequencies between breakpoint and pl1 translated to 1.5 cM with a 95% confidence interval of 1.1 cM. A B-I/B-I stock and isogenic B′/B′ derivative used for the b1 paramutation tests are of mixed parentage with plant color genotypes B1-I; Pl1-Rhoades (Pl-Rh); r-r. The rmr1-1, rmr2-1 lines (Hollick and Chandler 2001) and mop1-1, mop1-2 lines (Dorweiler et al. 2000) are as previously described. R-r:standard (R-r), R-stippled (R-st), and r-g haplotypes used for r1 paramutation tests derive from color-converted W22 inbred lines as previously described (Dorweiler et al. 2000).

TABLE 2.

Paramutagenicity testcross results: Pl1-Rhoades tester × Pl′/Pl′; rmr6-1/rmr6-1

		No. of progeny with specific anther color score
Testcrossa	Tester Pl1-Rhoades stateb	1	2	3	4	5	6	7
99-277 × 99-250-7	Pl-Rh/Pl-Rh (A632)	0	10	3	1	0	0	3
99-277 × 99-250-16	Pl-Rh/Pl-Rh (A632)	0	9	4	2	1	1	2
00-1028 × 00-1166-1	Pl-Rh/Pl-Rh (W23/A632)	1	1	4	1	0	0	2
99-277 × 99-250-7	Pl′/Pl′ (A632)	5	13	1	0	0	0	0
99-277 × 99-250-16	Pl′/Pl′ (A632)	0	17	1	0	0	0	0
00-1029 × 00-1166-1	Pl′/Pl′ (W23/A632)	6	7	2	1	0	0	0
Totals	Pl-Rh/Pl-Rh	1	20	11	4	1	1	7
	Pl′/Pl′	11	37	4	1	0	0	0

^aItalic numbers identify specific rmr6-1 homozygotes (ACS 7 anthers) from F₂ (99-250) and F₃ (00-1166) families.

^bInbred genetic backgrounds of individual Pl1-Rhoades stocks are indicated in parentheses (see materials and methods for details).

Stock constructions and complementation tests:

Hand pollinations were used for all genetic crosses and detailed pedigree information is available upon request.

Genetic complementation tests between the ems97406 mutation and other mutations known to affect Pl′ were conducted using +/ems97406 plants as pistillate parent. Staminate parent genotypes, number of crosses, and number of progeny plants with specific anther phenotypes are as follows: rmr1-1/rmr1-1, 4, 87 ACS 1–4, 2 ACS 5–6; rmr2-1/rmr2-1, 4, 88 ACS 1–4; mop1-1/mop1-1, 3, 56 ACS 1–4. Tests with the ems98225 mutation were conducted using ems98225 homozygotes as staminate parents. Pistillate parent genotypes, number of crosses, and number of progeny plants with specific anther phenotypes are as follows: rmr1-1/+, 3, 38 ACS 1–4; rmr2-1/+, 2, 25 ACS 1–4; mop1-2/+, 2, 31 ACS 1–4.

For both RNase protection analyses and in vitro transcription reactions, progeny sets were generated segregating 1:1 for +/rmr6-1 and rmr6-1/rmr6-1 siblings. Plants of these two genotypes were clearly identified by contrasting Pl′-like and Pl-Rh-like anther phenotypes, respectively. For in vitro transcription reactions, progeny sets also segregated B1-I alleles of B′ state as each parent was heterozygous B′/b1. In these families, +/rmr6-1 and rmr6-1/rmr6-1 plants were further distinguished by contrasting light and dark plant colors, respectively. Materials used for the rmr6-2 RNase protection assay came from a single F₂ family segregating 1:2:1 for rmr6-2/rmr6-2, +/rmr6-2, and +/+ siblings. Plants of these two latter genotypes have Pl′-like color phenotypes and are indistinguishable.

Reversion of Pl′ to Pl-Rh in rmr6-1/rmr6-1 plants was shown by following paramutagenic activity of Pl′ states carried on the T6-9 (043-1) translocation chromosome following segregation from a single rmr6-1 homozygote (Figure 3; Table 4). A family homozygous for the T Pl′ chromosome and segregating 1:1 for +/rmr6-1 and rmr6-1/rmr6-1 genotypes provided the individual for these reversion tests. Pollen from this ACS 7 individual was shared between isogenic Pl-Rh/Pl-Rh and Pl′/Pl′ testers (W23/A632). Progeny of the Pl-Rh/Pl-Rh testcross that had Pl-Rh-like anther phenotypes (ACS 7) were subsequently crossed to another Pl-Rh/Pl-Rh tester (W23/CO159). Construction of the T Pl′ stock segregating the rmr6-1 mutation was similar to that described for the rmr6-2 stock used in the pl1 paramutation tests: the T Pl′ line was crossed with an rmr6-1/rmr6-1; Pl′/Pl′ individual derived from a family segregating 1:1 for +/rmr6-1 (Pl′-like anthers) and rmr6-1/rmr6-1 (Pl-Rh-like anthers) genotypes and subsequent T Pl′/T Pl′; +/rmr6-1 (Pl′-like anthers) and T Pl′/T Pl′ ; rmr6-1/rmr6-1 (Pl-Rh-like anthers) F₂ individuals were intercrossed to establish the segregating families. Reversion of Pl′ to Pl-Rh in rmr6-2/rmr6-2 plants (Table 5) was shown using a similar strategy to that listed above for rmr6-1. Details of this construction are included as part of the following description for pl1-induced paramutation tests.

Figure 3.

Crossing scheme testing reversions of Pl′ to Pl-Rh states. (A) Cross used to examine paramutagenic properties of Pl1-Rhoades alleles transmitted from rmr6-1 homozygotes. Detailed information of parents and F₁ anther phenotypes is found in Table 2 for testcross 00-1028 × 00-1166-1. Parentheses denote uncertainty regarding paramutagenic properties of a particular Pl1-Rhoades allele. Boldface type indicates the genotype having an ACS 7 phenotype used in the subsequent testcrosses. (B) Individual F₁ plants (boldface type) having fully colored anthers were testcrossed to determine if the apparent loss of paramutagenicity was heritable. Segregating genotypes and associated pollen fertility phenotypes are listed. Anther phenotype data are presented in Table 4.

TABLE 4.

Testcross results measuring heritable loss of Pl′ paramutagenicity following transmission from rmr6-1/rmr6-1 plants

	Progeny structural genotype	No. of progeny with specific anther color scores
Testcrossa		1–4	5–6	7
01-698-1 × 01-532	T/+	5	0	0
01-698-7 × 01-532	T/+	5	1	0
01-532 × 01-698-1	+/T	3	1	3
01-532 × 01-698-7	+/T	7	5	3
01-698-1 × 01-532	+/+	5	0	1
01-698-7 × 01-532	+/+	1	0	8
01-532 × 01-698-1	+/+	0	0	5
01-532 × 01-698-7	+/+	0	0	7
Totals	T/+	10	1	0
+/T	10	6	6
	+/+	6	0	21

See Figure 3 for experimental design.

^aReciprocal crosses with a W23/CO159, Pl-Rh/Pl-Rh stock (01-532) in which the pistillate parent is listed first are indicated. Italic numbers represent individual F₁ plants (see Figure 3A) having an ACS 7 anther phenotype and ∼50% aborted pollen grains.

TABLE 5.

Testcross results measuring heritable loss of Pl′ paramutagenicity following transmission from rmr6-2/rmr6-2 plants

	Progeny structural genotype	No. of progeny with specific anther color scores
Testcrossa		1–4	5–6	7
02-490-8 × 02-470	T/+	1	2	5
02-470 × 02-490-8	+/T	0	0	6
02-313 × 02-490-8	+/T	0	0	9
02-234 × 02-490-3	+/T	0	0	11
02-470 × 02-490-10	+/T	2	0	8
02-313 × 02-490-13	+/T	1	0	7
02-490-8 × 02-470	+/+	0	0	10
02-470 × 02-490-8	+/+	1	0	4
02-313 × 02-490-8	+/+	0	0	9
02-234 × 02-490-3	+/+	0	0	5
02-470 × 02-490-10	+/+	1	0	7
02-313 × 02-490-13	+/+	0	0	8
Totals	T/+	1	2	5
+/T	3	0	41
	+/+	2	0	43

Experimental design is similar to that described in Figure 3 for rmr6-1.

^aReciprocal crosses with a W23 Pl-Rh/Pl-Rh stock (02-234, 02-313, and 02-470) are indicated in which the pistillate parent is listed first. Italic numbers represent individual F₁ plants from family 02-490 (materials and methods) having an ACS 7 anther phenotype (rmr6-2/rmr6-2) and ∼50% aborted pollen grains (+/T).

For pl1-induced paramutation tests using rmr6-2 alleles, Pl-Rh and Pl′ states were combined in rmr6-2 homozygotes through crosses between +/rmr6-2 individuals: Pl-Rh/T Pl-Rh; +/rmr6-2 crossed with T Pl′/T Pl′; +/rmr6-2 (Figure 4A). Progeny plants with fully colored anthers (ACS 7; rmr6-2/rmr6-2) that display ∼50% pollen abortion (Pl-Rh/T Pl′) were used to pollinate color-converted A619 Pl-Rh testers (Figure 4B). Both rmr6-2 parental lineages derive from an initial cross between our T Pl′ stock and a single rmr6-2 homozygote produced by sib mating two original M₂ rmr6-2 homozygotes. Homozygous T Pl′ individuals were selected from a single F₂ family [individual kernels homozygous for the translocation chromosome have a diagnostic mutant endosperm phenotype due to a 2.3-cM genetic linkage between the T6-9 (043-1) breakpoint and a mutant waxy1 allele] and sib crosses were made between individuals with a Pl′ anther phenotype (either +/rmr6-2 or +/+) and those with a Pl-Rh-like anther phenotype (rmr6-2/rmr6-2) to establish families segregating 1:1 for rmr6-2/rmr6-2 and +/rmr6-2 genotypes. This segregating line, preserved for three generations using sib matings, provided the T Pl′/T Pl′; +/rmr6-2 parent used in the acquisition of the paramutagenicity test (Figure 4A). From the same segregating line, a nonparamutagenic Pl-Rh revertant was isolated following outcross of a single rmr6-2/rmr6-2 plant to Coe's W23 full-color stock (B-I/B-I; Pl-Rh/Pl-Rh; R-r/R-r); several progeny with ACS 7 anther phenotypes were recovered. Reversion of Pl′ to a nonparamutagenic Pl-Rh state in individual 02-490-8 was confirmed by analyzing progeny derived from reciprocal backcrosses to Coe's W23 full-color stock (Table 5); 43/47 progeny, both fully fertile and semisterile types, had ACS 7 anther phenotypes. These ACS 7 progeny provided the Pl-Rh/T Pl-Rh; +/rmr6-2 parental plants used in the acquisition of paramutagenicity tests (Figure 4A).

Figure 4.

Crossing scheme testing acquisition of paramutagenicity in rmr6-2 homozygotes. (A) Cross used to introduce Pl′ and Pl-Rh states in rmr6-2 homozygotes. Segregant F₁ genotypes along with associated pollen and anther phenotypes are listed. The particular genotype used for subsequent testcrosses is in boldface type. (B) Testcross used to measure paramutagenicity of Pl1-Rhoades alleles following exposure of Pl-Rh to Pl′ states in rmr6-2 homozygotes. Boldface parental genotype corresponds to boldface segregant genotype found in A. Parentheses indicate uncertainty regarding paramutagenic properties of a given Pl1-Rhoades allele. Segregant genotypes and associated pollen fertility phenotypes are listed. Anther phenotype data are presented in Table 6.

For pl1-induced paramutation tests using rmr6-1 alleles, Pl-Rh and Pl′ states were combined in rmr6-1 homozygotes through crosses between individuals heterozygous for rmr6-1: T Pl-Rh/pl1-A619; +/rmr6-1 crossed with Pl′/Pl′; +/rmr6-1 (Figure 5A). Progeny plants with fully colored anthers (ACS 7; rmr6-1/rmr6-1) that display ∼50% pollen abortion (T Pl-Rh/Pl′) were used to pollinate color-converted A619 Pl-Rh/Pl-Rh testers (Figure 5B). A plant of pl1-A619/pl1-A619; rmr6-1/rmr6-1 genotype was crossed to a T Pl-Rh/T Pl-Rh stock derived from a T Pl-Rh reversion event described above to generate the T Pl-Rh/pl1-A619; +/rmr6-1 parent, and the Pl′/Pl′; +/rmr6-1 parent was provided from a family segregating 1:1 for +/rmr6-1 and rmr6-1/rmr6-1 plants. The pl1-A619/pl1-A619; rmr6-1/rmr6-1 plant derived from selfing an F₂ plant molecularly genotyped as being homozygous for a pl1-A619 RFLP polymorphism and shown by testcrosses to +/rmr6-1 plants to be homozygous for rmr6-1.

Figure 5.

Crossing scheme testing acquisition of paramutagenicity in rmr6-1 homozygotes. (A) Cross used to introduce Pl′ and Pl-Rh states in rmr6-1 homozygotes. Segregant F₁ genotypes along with associated pollen and anther phenotypes are listed. The particular genotype used for subsequent testcrosses is in boldface type. (B) Testcross used to measure paramutagenicity of Pl1-Rhoades alleles following exposure of Pl-Rh to Pl′ states in rmr6-1 homozygotes. Boldface parental genotype corresponds to boldface segregant genotype found in A. Parentheses indicate uncertainty regarding paramutagenic properties of a given Pl1-Rhoades allele. Segregant genotypes and associated pollen fertility phenotypes are listed. Anther phenotype data are presented in Table 7.

To test the role of Rmr6 in b1 paramutation, isogenic B-I and B′ stocks (B1-I; Pl1-Rh) were first pollinated by a single rmr6-1/rmr6-1; b1-W23/b1-W23; Pl1-Rh/Pl1-Rh plant. Intercrossing the resulting F₁ plants gave rise to progenies in which there was segregation of Pl-Rh-like and Pl′-like plants (Figure 6A). Among these Pl-Rh-like plants, the following b1 genotypes occurred at a 1:1:1:1 frequency: b1-W23/b1-W23; b1-W23/B′; b1-W23/B-I; B-I/B′. Blind testcrosses of plants displaying dark plant colors (rmr6-1/rmr6-1) were made to b1-CO159/b1-CO159 pistillate testers (Pl-Rh/Pl-Rh; CO159/W23 stock). Actual b1 genotypes of the staminate parents were assigned ex post facto on the basis of the segregation of plant color types among resulting testcross progenies.

Figure 6.

b1 paramutation analysis pedigree. (A) Plants with the indicated b1 and rmr6 genotypes were crossed together to obtain the given set of F₁ progeny genotypes. Two progeny sets (12873 and 12874) were used in the analysis. Three general plant-color phenotypes [colorless, light (B′-like), and dark (similar to B-I)] characterize the indicated genotypes. Genotypes giving rise to a dark plant phenotypes are in boldface type with the relevant B′/B-I genotype in parentheses. (B) Progeny plants listed in A displaying dark plant phenotypes (boldface type) were testcrossed to b1-CO159/b1-CO159 plants and assigned the actual b1 genotypes on the basis of segregation of plant phenotypes among testcross progenies. Testcross progeny genotypes are listed only for the relevant B′/B-I parent (boldface type in parentheses). Corresponding progeny phenotypes for the three relevant testcrossed plants are listed in Table 8 and are shown in Figure 7.

To test the role of Rmr6 in r1 paramutation, near isogenic R-r/R-r and R-st/R-st W22 lines were first pollinated by a single +/rmr6-1 plant homozygous for recessive r-r. F₁ progeny were backcrossed using a single recurrent staminate parent (+/rmr6-1; r-r/r-r) and subsequent progeny sets were screened for segregation of rmr6-1/rmr6-1 types. BC₁ plants were molecularly genotyped for Pl1-Rhoades vs. Pl1-W22 alleles using Southern blot hybridization analysis as previously described (Hollick et al. 1995) and only those with Pl1-Rhoades/Pl1-Rhoades genotypes were used in subsequent crosses. Crosses were made between R-st/r-r BC₁ plants with Pl′ anther phenotypes (either +/rmr6-1 or +/+) and R-r/r-r BC₁ plants with a Pl-Rh-like anther phenotype (rmr6-1/rmr6-1). Two progeny sets of these crosses segregating 1:1 for Pl′-like (+/rmr6-1) and Pl-Rh-like (rmr6-1/rmr6-1) types were evaluated for R-r pigmenting activity through individual testcrosses to recessive r1 W23 r-g/r-g pistillate stocks.

Rmr6 mapping:

B-A mapping of rmr6-1 and rmr6-2 was performed as outlined by Beckett (1978). Specifically for mapping mutations affecting Pl′, the TB-1La translocation chromosome was introgressed into stocks homozygous for the Pl1-Rhoades allele and maintained in heterozygous state. For mapping, Pl′/Pl′; +/rmr6-1 and Pl′/Pl′; +/rmr6-2 plants were pollinated by Pl′/Pl-Rh; TB-1La heterozygotes. B centromere nondisjunction occurring at the mitotic division giving rise to the sperm cells provides the opportunity to uncover recessive mutations transmitted through female meiosis (Beckett 1978). Resulting progeny were screened for 1L hypoploid individuals (characterized by early maturing, short, and tiny plants; Lee 1997) displaying an ACS 7 phenotype, indicative of a segmental monosomy uncovering mutant rmr6 alleles. Both +/rmr6-1 and +/rmr6-2 individuals were used as pistillate parents and, among their respective progeny, 3/9 and 4/7 hypoploid plants had ACS 7 phenotypes.

Cosegregation mapping of rmr6-1 with molecular markers was initiated by crossing an rmr6-1/rmr6-1; Pl′/Pl′ plant to a color-converted A632 line (Pl′/Pl′, >93% A632). F₁ progeny were self-pollinated, and 2-week-old F₂ seedlings were screened for Pl-Rh-like pigmentation. Leaf tissue was collected from mutant seedlings, mapping parents, and F₁ plants, and genomic DNA was extracted using the DNeasy 96 plant kit (QIAGEN, Valencia, CA). Undiluted genomic DNA was used as template for PCR amplification (AmpliTaq Gold DNA Polymerase, Applied Biosystems, Foster City, CA) of microsatellite marker umc1035, located in bin 1.06 of chromosome 1L (5′-CTGGCATGATCACGCTATGTATG-3′; 5′-TAACATCAGCAGGTTTGCTCATTC-3′; http://www.maizegdb.org/cgi-bin/displayssrrecord.cgi?id=174092). PCR products were electrophoresed in 3.5% agarose gels (Agarose MS; Roche, Mannheim, Germany), visualized by ethidium bromide staining, and imaged using an AlphaInnotech Image Capture system (Alpha Innotech, San Leandro, CA). A umc1035 size polymorphism between the mapping parents was used for cosegregation analysis. Among 62 F₂ rmr6-1/rmr6-1 seedlings, six recombination events were detected between rmr6 and umc1035. These results place the rmr6 locus ∼3 cM from umc1035 and confine the position of rmr6 to bin 1.06.

r1 activity assay:

Pigmentation of testcross r-g/r-g/R-r kernel samples was measured with an Agtron reflectometer as described (Alleman and Kermicle 1993) and presented as relative pigmentation values (Kermicle et al. 1995).

Pollen scoring:

Pollen samples from freshly extruded anthers were examined in the field with the aid of a 50× pocket microscope (Edmund Scientific, Tonawanda, NY). Plants either homozygous for the T6-9 (043-1) reciprocal translocation or not carrying the translocation pair have ∼95–100% plump and opaque pollen grains whereas plants heterozygous for the translocation display ∼50% collapsed or translucent pollen grains due to segmental chromosome deficiencies (Patterson 1994).

Anther pigmentation:

ACS refers to specific pigment patterns (Hollick et al. 1995). For each plant, a single ACS value was assigned on the basis of visual examination of freshly extruded anthers.

Molecular expression analyses:

Materials and methods used for nucleic acid purifications and RNase protection analyses are described in Hollick et al. (2000) except that three entire florets from the leading edge of anthesis on the tassel were used as sources of RNA for each sample. All tissues were harvested at ∼12:30 pm. Husk tissues used as sources of nuclei for in vitro transcription reactions were prepared as described in Hollick et al. (2000). Nuclei isolations and transcription reactions (Dorweiler et al. 2000) were carried out with 50 μCi of radiolabeled UTP per reaction. Slot blots were prepared using a slot-blot apparatus (Bio-Rad, Hercules, CA). Four micrograms of linearized and denatured plasmids per slot were placed on prewetted Magna nylon membranes (GE Osmonics, Minnetonka, MN) and UV treated to covalently attach. The pl1 plasmid (pJH7; Hollick et al. 2000), b1 plasmid (315-bp cDNA of b1; Selinger and Chandler 1999), a1 clone (Patterson et al. 1993), and both actin1 and ubiquitin2 clones (Dorweiler et al. 2000) have been previously described. pBS is a pBluescript II (KS+) cloning plasmid (Stratagene, La Jolla, CA). Slot blots were prehybridized, hybridized with heat-denatured nuclear RNA at 42°, and washed at a final stringency of 0.01% SSC, 0.05% SDS at 65° (Chandler et al. 1989). Hybridizations were visualized, quantified, and normalized as described in Dorweiler et al. (2000) using ImageQuant software (Amersham Biosciences, Piscataway, NJ).

pl1 molecular genotyping:

The phi031 primer set: 5′-GCAACAGGTTACATGAGCTGACGA-3′; 5′-CCAGCGTGCTGTTCCAGTAGTT-3′ (http://www.maizegdb.org/cgi-bin/displayssrrecord.cgi?id=111049; Chin et al. 1996) was used to distinguish Pl1-Rhoades and pl1-A632 alleles (187- and 223-bp products, respectively). Genomic DNA was isolated from seedling leaves (Voelker et al. 1997) with the following modifications: following the first isopropanol precipitation step, nucleic acids were resuspended in 400 μl of TE, extracted with a 1:1 volume of phenol/chloroform (1:1) and a 1:1 volume of chloroform, and reprecipitated with 0.3 m NaOAc and 95% EtOH. Precipitated DNA was rinsed with 70% EtOH and resuspended in 100 μl TE and quantified using Hoechst 33258 dye binding measured with a TD-360 fluorometer (Turner Designs, Sunnyvale, CA). Quantified DNA was diluted to 50 ng/μl prior to PCR analysis as described above. Southern blot analyses (Hollick et al. 1995) were used to genotype Pl1-Rhoades and Pl1-W22 alleles in individuals used for the r1 paramutation stock constructions.

RESULTS

EMS mutations define a novel trans-acting rmr locus:

Because Pl′/Pl′ seedlings are weakly pigmented (Hollick et al. 1995), recessive mutations that enhance pigmentation are readily identified in M₂ progenies derived from EMS-treated pollen (Dorweiler et al. 2000; Hollick and Chandler 2001). Two mutations, ems97406 and ems98225, were found in separate M₂ progenies segregating 1 of 38 and 3 of 26 dark-pigmented Pl-Rh-like seedlings, respectively. Correspondingly, anthers of mature plants had dark-pigmented Pl-Rh-like phenotypes (Figure 1A). The two mutations appeared allelic as they failed to genetically complement each other (16 of 43 progeny from two crosses of +/ems97406 plants by ems98225/ems98225 plants had Pl-Rh-like, ACS 7, anther phenotypes). Subsequent mapping experiments, using B-A translocation stocks to create segmental monosomics (Beckett 1978), confirmed that both mutations are found on the long arm of chromosome 1 (see materials and methods). Cosegregation analysis with molecular markers further positions ems97406 to bin 1.06 (see materials and methods). Both ems97406 and ems98225 genetically complement previously described mutations at rmr1, rmr2, and mop1 (see materials and methods), demonstrating these two mutations define the novel maize locus rmr6. The two mutant alleles, ems97406 and ems98225, are designated rmr6-1 and rmr6-2, respectively.

rmr6 mutations specifically affect Pl1-Rhoades alleles:

Pigment intensification seen in rmr6-1 and rmr6-2 homozygotes could be due to modifications of a rate-limiting step for either regulatory or biosynthetic pathways for anthocyanin production. The rmr6-1 mutation was found to act specifically on Pl1-Rhoades-type pl1 alleles in recessive Mendelian fashion by inspecting an F₂ family for segregation of distinct pl1 alleles and mutant anthocyanin phenotypes (Table 1). A single rmr6-1 homozygote was crossed to an A632 inbred individual and a single F₁ plant was self-pollinated to generate this F₂ family. The pl1 genotype of each F₂ individual was determined using a set of simple sequence repeat primers (phi031; Chin et al. 1996) that amplify allele-specific tetranucleotide repeat sequence polymorphisms from the first introns of Pl1-Rhoades (187 nt) and pl1-A632 (223 nt). The mutant anthocyanin phenotype (ACS 7) was exclusively found among plants carrying a Pl1-Rhoades allele and appeared to segregate in an ∼1:3 manner within the combined Pl1-Rhoades/Pl1-Rhoades and Pl1-Rhoades/pl1-A632 classes (27:76; 26%; χ² = 0.043; for H_o that difference from 1:3 segregation is due to random chance; no statistical difference), suggesting that rmr6-1 acts as a recessive mutation. Among F₂ Pl1-Rhoades homozygotes, the ACS 7 phenotype also segregated in an ∼1:3 fashion (9:22; 29%; χ² = 0.202; for H_o that difference from 1:3 segregation is due to random chance; no statistical difference), confirming that the rmr6 locus is genetically unlinked to pl1. None of the 23 pl1-A632 homozygotes had increased pigmentation (0:23; 0%; χ² = 5.75; for H_o that difference from 1:3 segregation is due to random chance; P < 0.05), indicating that Rmr6 functions specifically maintain the repressed state of Pl′ and are not involved in general suppression of either the genetic regulation or biosynthetic components of the anthocyanin pathway.

TABLE 1.

F₂ segregation analysis (F₁ genotype: pl1-A632/Pl′; Rmr6-A632/rmr6-1)

	No. of F2 progeny with specific anther phenotypesb
Pl1-Rhoades anther color scores
F2 progeny pl1 genotypea	A632	1–4	5–6	7	Fraction of F2 progenyc
Pl1-Rh/Pl1-Rh	0	22	0	9	31/126 (25)
pl1-A632/Pl1-Rh	0	53	1	18	72/126 (57)
pl1-A632/pl1-A632	23	0	0	0	23/126 (18)

^apl1 genotype determined with codominant simple sequence repeat polymorphisms within pl1 intron 1 (materials and methods).

^bA632 anther phenotype described in Hollick and Chandler (2001).

^cThe percentage of F₂ progeny is given in parentheses.

Rmr6 maintains transcriptional repression of paramutant Pl′ and B′ states:

Both paramutant Pl1-Rhoades and B1-Intense allelic states (Pl′ and B′, respectively) are associated with reduced transcriptional activity at the corresponding loci (Patterson et al. 1993; Hollick et al. 2000). Comparative RNase protection and nuclear run-on transcription analyses of rmr6-1 homozygotes and +/rmr6-1 siblings show that Rmr6 function maintains low levels of pl1 RNA largely by, if not exclusively, transcriptional repression (Figure 2). Using total floret RNA, we found that rmr6-1 homozygotes had ∼15-fold higher pl1 RNA levels than +/rmr6-1 siblings whereas rmr6-2 homozygotes had ∼18-fold higher levels than normal siblings with either +/rmr6-2 or +/+ genotypes (Figure 2, A and B). These RNA abundance differences are significantly higher than those observed between standard Pl-Rh/Pl-Rh and Pl′/Pl′ genotypes (an ∼9-fold difference; Figure 2B). Rmr6 thus maintains low pl1 RNA levels associated with the Pl′ state. To discern whether Rmr6-dependent effects on pl1 RNA levels involved transcription itself, we compared levels of radiolabeled RNA species transcribed in nuclei isolated from husks of rmr6-1/rmr6-1 and +/rmr6-1 genotypes (Figure 2, C and D). Relative to levels of labeled ubiquitin2 RNA, pl1 RNA levels were higher in nuclei from rmr6-1 homozygotes, indicating that synthesis of pl1 transcripts within this tissue was, on average, ∼4-fold greater than in +/rmr6-1 siblings. Although additional effects on post-transcriptional regulation of pl1 RNA levels are not excluded, it is clear that Rmr6 is required to maintain transcriptional repression associated with the Pl′ state. Because the tested nuclei came from plants also containing a paramutant B′ allele, we found that b1 transcription was also increased ∼8-fold in rmr6-1/rmr6-1 genotypes (Figure 2, C and D), a difference similar in magnitude to prior comparisons between standard B-I/B-I and B′/B′ genotypes (Patterson et al. 1993). Consistent with the fact that B1 and PL1 proteins are required for transcriptional induction of the biosynthetic anthocyaninless1 (a1) gene (encoding a dihydroflavonol reductase; O'Reilly et al. 1985), a1 transcription was increased ∼12-fold in rmr6-1/rmr6-1 genotypes (Figure 2, C and D). These molecular expression assays confirm that Rmr6 is genetically required to maintain transcriptional repression associated with paramutant Pl′ and B′ states.

Figure 2.

pl1 expression analyses. (A) RNase protection comparisons of pl1 and actin1 RNA levels from floret tissues of the indicated genotypes. (B) Histogram representation of average pl1 RNA levels relative to actin1 RNA levels (±SE) in the indicated genotypes (n = 2, 2, 3, 4, 2, 2 for genotypes Pl′/Pl′, Pl-Rh/Pl-Rh, +/rmr6-1, rmr6-1/rmr6-1, +/rmr6-2 or +/+, and rmr6-2/rmr6-2, respectively). Measurements are normalized with data for nonmutant types set at unit value. (C) Slot-blot hybridization comparison of in vitro radiolabeled RNA species produced using nuclei isolated from husk tissues of the indicated genotypes. Plasmid DNA (pBS) is included as a negative control. (D) Histogram representation of average transcription rates of the indicated genes measured relative to ubiquitin2 (±SE). Measurements for rmr6-1/rmr6-1 genotypes (solid bars; n = 5) are displayed relative to data of +/rmr6-1 genotypes (open bars; n = 5) set at unit value.

Rmr6 maintains the paramutagenic Pl′ state:

Although previous tests established a strong inverse correlation between Pl1-Rhoades RNA levels and paramutagenic strength (Hollick et al. 1995), analyses of mop1, rmr1, and rmr2 mutations illustrate that reacquisition of Pl-Rh-like RNA expression levels does not necessarily lead to loss of heritable paramutagenic activity. Most Pl1-Rhoades alleles inherited in the Pl′ state and subsequently transmitted from mop1-1, rmr1-1, or rmr2-1 homozygotes possess strong paramutagenic activity (Dorweiler et al. 2000; Hollick and Chandler 2001). Thus the type of repression mechanism acting to limit somatic Pl1-Rhoades expression is not necessarily the same as the mechanism responsible for defining meiotically heritable paramutagenicity. Given that Rmr6 acts to maintain transcription-based repression, we could now ask whether heritable paramutagenic activity was necessarily or inseparably related to somatic reduction of transcription.

A series of genetic crosses and progeny analyses helped clarify the functional role of Rmr6 in pl1 paramutation. Heritable maintenance of paramutagenic activities was the focus of initial tests. Passage of Pl′ states through plants homozygous for rmr6 mutations was predicted to have one of the following possible outcomes: (1) paramutagenicity would be maintained so all transmitted Pl1-Rhoades alleles remained in the Pl′ state, (2) paramutagenicity would be lost so all transmitted alleles were indistinguishable from Pl-Rh, (3) paramutagenicity would be partially affected such that transmitted alleles collectively represented a continuum of activities between fully paramutagenic Pl′ and nonparamutagenic Pl-Rh (similar to the behavior of Pl1-Rhoades alleles transmitted from plants displaying ACS 5 and ACS 6 phenotypes; Hollick et al. 1995), or (4) transmitted Pl1-Rhoades alleles would not only have a Pl-Rh-like identity but also be immune to subsequent changes to Pl′ states. To address these possibilities, pollen from individual rmr6-1/rmr6-1 plants was distributed to pairs of isogenic pistillate parents with contrasting Pl1-Rhoades states. We ensured that the rmr6-1 homozygotes used had received at least one Pl1-Rhoades allele in Pl′ state from the previous generation either by self-pollinating a +/rmr6-1 heterozygote having a Pl′ phenotype or by sib crossing a +/rmr6-1 plant having a Pl′ phenotype with a rmr6-1 homozygote having a Pl-Rh-like phenotype. Results of the Pl′/Pl′ testcrosses (Table 2) show that Pl1-Rhoades alleles transmitted from rmr6-1 homozygotes are fully susceptible to subsequent paramutation as all progeny from such crosses have clear Pl′ phenotypes (ACS 1–4). Results of Pl-Rh/Pl-Rh testcrosses represent a continuum of paramutagenic activity among Pl1-Rhoades alleles transmitted from rmr6-1 homozygotes, in accord with prediction 3 above. While most Pl1-Rhoades alleles remained fully paramutagenic (testcross progeny with ACS 1–4), others displayed evidence of weaker paramutagenic action following transmission from rmr6-1 homozygotes (testcross progeny with ACS 5–6). Furthermore, the maintenance of paramutagenicity is completely impaired in some rmr6-1/rmr6-1 plants as progeny with ACS 7 phenotypes are obtained (Table 2). Thus, passage of Pl′ states through rmr6-1/rmr6-1 sporophytes affected paramutagenic activity such that some transmitted Pl1-Rhoades alleles were now indistinguishable from a nonparamutagenic Pl-Rh state.

The frequency of Pl′-to-Pl-Rh reversions transmitted from Pl′/Pl′; rmr6-1/rmr6-1 plants was estimated from 25 separate testcrosses using two standard Pl-Rh/Pl-Rh stocks and 17 rmr6-1/rmr6-1 individuals. All but one of these rmr6-1 homozygotes gave individual progeny with an ACS 7 phenotype, indicating that a certain percentage (ranging from 7–100%) of Pl1-Rhoades alleles were transmitted in nonparamutagenic states indistinguishable from Pl-Rh (Table 3). In total, average reversion frequency was 36%. Reversion events were independent of prior allelic history as similar frequencies were obtained from rmr6-1 homozygotes generated in M₂, F₂, F₃, or F₄ families. In every case, at least one Pl1-Rhoades allele in rmr6-1/rmr6-1 plants was inherited from the previous generation in Pl′ condition. The fact that nonparamutagenic Pl-Rh states were transmitted from rmr6-1/rmr6-1 plants present in both M₂ and F₂ families in which both parental alleles were contributed in Pl′ states demonstrates the requirement of Rmr6 to faithfully maintain meiotically heritable information specifying paramutagenic behavior. Consistent with this interpretation, nine of the rmr6-1 homozygotes also transmitted Pl1-Rhoades alleles displaying intermediate levels of paramutagenic action as evidenced by testcross progeny having ACS 5 and ACS 6 phenotypes.

TABLE 3.

Loss of Pl′ paramutagenicity seen in Pl-Rh/Pl-Rh × Pl′/Pl′; rmr6-1/rmr6-1 testcrosses

Staminate parenta	Pistillate tester	No. of testcrosses	No. of progeny with specific anther color scores
			1	2	3	4	5	6	7b
98-273-17 (F2)	W23	2	0	8	3	1	4	2	20 (53)
98-354 (F2)	W23	2	2	17	10	4	1	1	9 (20)
98-1339-4 (F3)	W23	1	1	12	6	1	0	0	6 (23)
98-1339-8 (F3)	W23	1	2	6	4	0	0	0	12 (50)
98-1339-12 (F3)	W23	1	0	6	8	0	0	0	1 (7)
98-1340-1 (F3)	W23	1	0	2	4	0	0	3	15 (62)
98-1342-9 (F3)	W23	1	0	5	4	0	0	0	14 (61)
99-141-4 (F4)	W23	1	0	8	2	1	1	0	10 (45)
99-142-24 (F4)	W23	1	0	0	1	1	2	0	11 (73)
99-315-1 (F4)	W23	1	0	0	0	0	0	0	14 (100)
97-740-7 (M2)	W23/CO159	1	0	1	0	0	0	0	6 (86)
98-151-13 (F2)	W23/CO159	2	3	13	5	3	4	0	5 (15)
98-354 (F2)	W23/CO159	4	19	34	10	9	5	0	9 (10)
99-142-24 (F4)	W23/CO159	1	0	5	1	0	0	2	10 (56)
99-217-7 (F4)	W23/CO159	2	8	17	0	1	0	1	0 (0)
99-250-7 (F2)	W23/CO159	1	0	3	0	0	0	0	17 (85)
99-250-24 (F2)	W23/CO159	2	7	11	1	4	0	0	10 (30)
Totals		25	42	148	59	25	17	9	169 (36)

^aPedigree origins of rmr6-1 homozygotes are indicated in parentheses: M₂ is the family in which rmr6-1 homozygotes were first found, F₂ indicates families in which Pl-Rh phenotypes were recovered following self-pollination of Rmr6/rmr6-1 plants with Pl′-like phenotypes, F₃ and F₄ indicate families generated by intercrossing heterozygous (Rmr6/rmr6-1) individuals with homozygous rmr6-1 F₂ or F₃ siblings, respectively.

^bThe percentage of testcross progeny with ACS 7 is given in parentheses.

Although Pl1-Rhoades alleles of Pl′ state could revert to nonparamutagenic Pl-Rh states in rmr6-1 homozygotes, this reversion was not always stable. Using genetic linkage (∼1.5 cM; materials and methods) to a T6-9 translocation breakpoint, specific Pl1-Rhoades alleles in reference Pl′ state were followed through rmr6-1 homozygotes and two subsequent testcrosses to Pl-Rh/Pl-Rh plants. It is important to point out that Pl1-Rhoades alleles carried on all translocation chromosomes tested to date display normal paramutation properties (J. Hollick, unpublished results). Linkage between the translocation breakpoint “T” and reference Pl1-Rhoades allele is displayed as T Pl′. Figure 3 describes the crossing scheme in detail. When T Pl′ passed through a rmr6-1 homozygote, two of nine testcross progeny had ACS 7 phenotypes, suggesting that, as before, Pl′ had reverted to nonparamutagenic Pl-Rh state (Figure 3A). These testcross progeny (Pl-Rh/T Pl-Rh) display significant pollen abortion (∼50%) because chromosome segregations from translocation heterozygotes lead to a high percentage of aneuploid spores unable to develop functional male gametophytes (Patterson 1994). Results of a subsequent testcross assessed the status of Pl1-Rhoades alleles segregating from these initial ACS 7 plants (Figure 3B). To our surprise, we found that many Pl1-Rhoades alleles that had apparently reverted to the Pl-Rh state in the previous generation were strongly paramutagenic, while alleles found on normal homologs mostly retained nonparamutagenic Pl-Rh states (Table 4). Exceptions were found in both cases indicating that (1) Pl′ could revert to a meiotically stable nonparamutagenic Pl-Rh state on the T6-9 chromosome and (2) Pl-Rh could undergo heritable changes to paramutagenic Pl′ state at high frequency despite all genetic and phenotypic indications that these Pl1-Rhoades alleles were in Pl-Rh state during the preceding generation. Because these last testcrosses were carried out reciprocally, we observed that both reestablishment of Pl′ state on the translocation chromosome and spontaneous establishment of Pl′ on the normal chromosome occurred more frequently through female (pistillate), rather than male (staminate), transmission.

Reversions of Pl′ to Pl-Rh taking place in parallel lineages utilizing rmr6-2 homozygotes appeared more stable (Table 5). This finding, together with pl1 expression data (Figure 2, A and B), suggests that the rmr6-2 allele represents a more severe functional defect relative to rmr6-1. Virtually all revertant Pl-Rh states obtained from T Pl′/T Pl′; rmr6-2/rmr6-2 plants remained nonparamutagenic (Table 5). In this experiment, plants carrying the translocation chromosome were used as the pistillate parent in only one cross; therefore, it was not possible to assess whether there was parental transmission bias of T Pl′ vs. T Pl-Rh types. Although the frequency of spontaneous paramutation observed among the two different Pl-Rh/Pl-Rh testers (W23/CO159 vs. W23 backgrounds) used in the rmr6-1 and rmr6-2 tests is similar (<0.001; J. Hollick, unpublished results), it is possible that Pl1-Rhoades alleles maintained in these two lineages are differentially susceptible to low levels of paramutagenicity.

Rmr6 is needed to heritably acquire a Pl′ paramutagenic state:

The fact that Pl′ states can revert to a stable Pl-Rh state in plants homozygous for rmr6 mutant alleles obscures the role of Rmr6 in inducing paramutation in trans (Figure 1C). To address this issue, we tracked paramutagenic behaviors of two Pl1-Rhoades alleles of contrasting Pl′ and Pl-Rh states following segregation from T Pl′/Pl-Rh; rmr6-2/rmr6-2 plants. Figure 4 describes the series of crosses. The T Pl′ parents of these T Pl′/Pl-Rh ; rmr6-2/rmr6-2 plants were heterozygous for a normal Rmr6 allele to ensure transmission of a Pl′ state. As in previous experiments, Pl′ reverted to nonparamutagenic Pl-Rh state at high frequency following transmission from rmr6-2 homozygotes (Table 6). Overall, Pl-Rh states carried by normal chromosomes appeared unchanged. Three of 86 Pl′-like testcross progeny plants having no significant pollen abortion either could reflect rare cases in which Pl-Rh had changed to paramutagenic Pl′ state in rmr6-2 homozygotes or, alternatively, could represent the expected ∼1.5% of cases in which recombination placed the reference Pl1-Rhoades allele of Pl′ state onto a structurally normal chromosome 6. In either case, Rmr6 is clearly required for efficient acquisition of a heritable paramutagenic state in T Pl′/Pl-Rh heterozygotes.

TABLE 6.

Testcross results measuring acquisition of paramutagenicity by Pl-Rh in rmr6-2/rmr6-2; + Pl-Rh/T Pl′ plants

Staminate testcross parent	Progeny structural genotype	No. of progeny with specific anther color scores
		1	2	3	4	5	6	7
03-343-6	+/T	1	2	0	0	0	0	0
03-343-7	+/T	0	4	2	2	0	0	3
03-344-9	+/T	0	0	3	2	0	0	2
03-344-14	+/T	0	1	0	2	1	4	3
03-345-9	+/T	1	2	3	0	1	0	2
03-346-15	+/T	0	0	0	0	0	2	8
03-346-11	+/T	0	3	0	0	3	1	0
03-347-1	+/T	0	1	1	0	2	0	1
03-343-6	+/+	0	0	0	0	0	0	15
03-343-7	+/+	1	1	0	0	0	0	6
03-344-9	+/+	0	0	1	0	0	0	11
03-344-14	+/+	0	0	0	0	0	0	8
03-345-9	+/+	0	0	0	0	0	0	8
03-346-15	+/+	0	0	0	0	0	0	10
03-346-11	+/+	0	0	0	0	0	0	11
03-347-1	+/+	0	0	0	0	0	0	14
Totals	+/T	2	13	9	6	7	7	19
	+/+	1	1	1	0	0	0	83

Crossing scheme is presented in Figure 4. A619 Pl-Rh/Pl-Rh stocks were used for all testcrosses.

Similar tests performed with the rmr6-1 allele had slightly different results (Figure 5 and Table 7). In contrast to the rmr6-2 test, the reference Pl1-Rhoades allele of the Pl′ state was carried on a structurally normal chromosome. As before, Pl′ reverted back to Pl-Rh at lower frequency in rmr6-1 homozygotes (Table 7) relative to rmr6-2 homozygotes (Table 6); however, in this comparison, an identical Pl-Rh/Pl-Rh tester (A619 background) was used. While most Pl-Rh states on T6-9 chromosomes remained unchanged in T Pl-Rh/Pl′; rmr6-1/rmr6-1 plants, ∼40% acquired a paramutagenic Pl′ state (12 of 31 +/T testcross progeny genotypes had ACS 1–4 phenotypes; Table 7). While acquisition of paramutagenicity in trans appears to be prohibited in rmr6-2 homozygotes, in rmr6-1 homozygotes it is impaired but not prevented.

TABLE 7.

Testcross results measuring acquisition of paramutagenicity by Pl-Rh in rmr6-1/rmr6-1; T Pl-Rh/+ Pl′ plants

Staminate testcross parent	Progeny structural genotype	No. of progeny with specific anther color scores
		1	2	3	4	5	6	7
03-1202-3	+/T	0	0	0	1	0	2	0
03-1202-9	+/T	2	2	0	1	0	0	5
03-1202-12	+/T	0	1	0	1	0	0	8
03-1202-17	+/T	0	2	0	2	1	0	3
03-1202-3	+/+	0	4	0	0	1	0	1
03-1202-9	+/+	6	0	0	0	0	1	1
03-1202-12	+/+	0	0	0	1	1	2	3
03-1202-17	+/+	5	5	0	0	0	0	0
Totals	+/T	2	5	0	5	1	2	16
	+/+	11	9	0	1	2	3	5

Crossing scheme is presented in Figure 5. A619 Pl-Rh/Pl-Rh stocks were used for all testcrosses.

Rmr6 is needed to acquire a heritable paramutant B′ state:

In B-I/B′ heterozygotes, the strongly expressed B-I state invariably changes to a transcriptionally repressed B′ state (Patterson et al. 1993). To ask whether B-I would acquire a B′ state in the absence of Rmr6 function, B-I and B′ states were combined in rmr6-1 homozygotes and then evaluated for effects on plant pigmentation following transmission to recessive b1 allele testers (Figure 6). Dark plant pigment phenotypes of the three B′/B-I; rmr6-1/rmr6-1 plants tested were similar to that displayed by B-I/B-I genotypes. Although the two B1-I states in our experiment (B-I and B′) were on unmarked chromosomes, the near perfect 1:1 segregation of B-I-like and B′-like plant pigment types among testcross progeny (Table 8; Figure 7) strongly suggests that B-I is unable to attain a meiotically heritable B′ state in the absence of full Rmr6 function. On the basis of our experimental design (Figure 6; materials and methods), sibling rmr6-1 homozygotes with either B′/b1-W23 or B-I/b1-W23 genotypes were also testcrossed. On the basis of the segregation of either colorless (b1-CO159/b1-W23) and light-colored progeny (b1-CO159/B′) or colorless and dark-colored progeny (b1-CO159/B-I), the b1 genotypes of these plants were assigned ex post facto. Although plant pigment phenotypes of rmr6-1/rmr6-1; B′/b1-W23 plants were similar to rmr6-1/rmr6-1; B-I/b1-W23 plants (dark), B′ never reverted to a heritable B-I state in 26 examples; all 26 b1-CO159/B1-I testcross progeny from three B′/b1-W23 plants had a light plant pigment phenotype typical of B′/B′ plants. The occurrence of B-I-like types segregating from B-I/B′; rmr6-1/rmr6-1 plants is therefore highly unlikely to be due to reversion of B′ to B-I states. Thus Rmr6 is required for heritable acquisition of a B′ paramutant state and the maintenance of transcriptionally repressed B′ states, yet it appears not to be required for meiotic maintenance of a previously established paramutagenic B′ state.

TABLE 8.

Evaluation of b1 paramutation in rmr6-1/rmr6-1; B-I/B′ plants through crosses to Rmr6 b1 testers

Staminate parent	No. of progeny with specific plant phenotypes
	B-I	B′	χ2a
02-513-4	32	38	0.17
02-513-52	40	34	0.49
02-719-33	7	13	0.9
Totals	79	85	0.22

A stock homozygous for a null b1 allele (b1-CO159) was used as pistillate parents for testcrosses of rmr6-1 homozygotes in which B-I and B′ states were combined (see Figure 6 and materials and methods for details).

^aA χ² test was applied to the null hypothesis that the difference from an expected 1:1 segregation of B-I and B′ types is due to random chance. None of the χ² values were statistically significant (P ≫ 0.05).

Figure 7.

Transmission of B-I and B′ states in the absence of Rmr6 function. The two sibling plants represent color phenotypes displayed by the indicated b1 progeny genotypes resulting from Rmr6/Rmr6; b1-CO159/b1-CO159 × rmr6-1/rmr6-1; B-I/B′ testcrosses (Figure 6B; Table 8).

Rmr6 is required to acquire a heritable paramutant R-r′ state:

The seed component of r1 haplotype R-r:standard also undergoes heritable changes in activity states (reviewed in Chandler et al. 2000). R-r changes to weaker expression states (R-r′) following exposure to structurally dissimilar, yet strongly paramutagenic, r1 haplotypes like R-stippled (R-st) (Brink 1956). Standard measure of r1 action in these cases is relative pigmentation of the triploid kernel aleurone tissue following male transmission to recessive r1 testers. To address whether R-r could adopt a heritable R-r′ state in the absence of Rmr6 function, R-r action was measured following exposure to R-st in rmr6-1 homozygotes and in +/rmr6-1 heterozygous siblings. Figure 8 lists genotypes and corresponding kernel phenotypes of individuals used in the analysis. As control, R-r action was also measured following exposure to nonparamutagenic r-r haplotypes in sibling plants with similar rmr6 genotypes. Two r1 paramutation crosses (Figure 8A) were made and at least two individuals from each of the four relevant F₁ genotypes (genotypes in boldface type in Figure 8A) were testcrossed to recessive r-g/r-g testers (Figure 8B). Mottled kernels (r-g/r-g/R-r) from testcross ears were cleaned of chaff and bulk samples evaluated for pigmentation in a reflectometer. Averages of mean reflectance measurements from each kernel sample are presented in Table 9. From these data it is clear that R-r does not change to a R-r′-like expression state following exposure to R-st in rmr6-1 homozygotes. In contrast, R-r adopts a significantly reduced expression state following exposure to R-st in +/rmr6-1 heterozygous siblings. As is characteristic of R-r paramutation by R-st, aleurone pigmentation was reduced to a lower level in the lineage having the lower nonparamutant value (Styles and Brink 1966; H_o that the two average values from the 11,734 and 11,738 progenies differ from each other due to random chance; t = 5.74; P < 0.01; two sample t-test with 5 d.f.). In both cases, however, reduction of R-r action is consistent with heritable acquisition of a paramutant R-r′ state.

Figure 8.

r1 paramutation analysis pedigree. (A) Plants with indicated r1 and rmr6 genotypes were crossed together to obtain the given set of F₁ progeny genotypes. Two progeny sets (11734 and 11738) were used in this analysis. Diagnostic kernel phenotypes are indicated for given genotypic classes. Genotypes used as staminate parents for subsequent testcrosses are in boldface type. (B) r1 genotypes of testcross parents are given along with segregant triploid endosperm genotypes and corresponding phenotypes of resulting kernels. Kernel classes used to measure R-r expression (Table 9) are in boldface type.

TABLE 9.

r1 paramutation analysis

	Reflectance measurements for testcrossed r-g/r-g/R-r kernels from specific staminate genotypesa
r-r/R-r		R-st/R-r		Statistical significanceb
Progeny set	+/rmr6-1	rmr6-1/rmr6-1	+/rmr6-1		rmr6-1/rmr6-1
11734	80.6 ± 3.0 (2)	79.6 ± 2.5 (3)	58.5 ± 7.7 (6)	79.8 ± 3.5 (4)	P < 0.01
11738	76.3 ± 4.0 (6)	78.0 ± 4.6 (4)	34.5 ± 7.8 (8)	73.9 ± 2.1 (5)	P ≪ 0.01

^aReflectance measurements indicate relative R-r expression through its effects on kernel pigmentation. Average measurement values ±SD⁺ are presented for the total number of testcross individuals shown in parentheses. Measurements indicating absence of r1 paramutation are in italics.

^bA two-sample t-test was applied to the null hypothesis that differences between average measurements derived from R-st/R-r; +/rmr6-1 and R-st/R-r; rmr6-1/rmr6-1 genotypes are due to random chance.

Rmr6 maintains the R-r′ paramutant state:

When Rmr6 is present, R-r′ reverts partially toward the nonparamutagenic R-r state in R-r′/R-r homozygotes (Brink 1964). Self-mating an F₁ R-st/R-r; +/rmr6-1 plant (Figure 8A) generated R-r′-carrying plants in which maintenance of R-r′ in the absence of normal Rmr6 function could be evaluated (Table 10). Using the same testcross protocol as before, it is clear that R-r:standard remained in paramutant R-r′ state when transmitted from plants with normal Rmr6 function but regained meiotically heritable pigmenting actions similar to nonparamutant R-r when transmitted from rmr6-1 homozygotes. Remarkably, the extent of recovery in R-r′/R-st heterozygotes matched that of R-r′/R-r′ homozygotes. Both stand in contrast to sibling R-r′/R-st Rmr6-containing plants, which show low levels of pigmentation characteristic of F₂ R-r′/R-st plants (Mikula 1961). Thus Rmr6 is required for maintenance of paramutant R-r:standard as well as Pl′ states.

TABLE 10.

Reflectance measurements for testcrossed r-g/r-g/R-r kernels from specific staminate genotypes

Staminate r1 genotypes	Staminate rmr6 genotypesa		Statistical significanceb
	Rmr6/−	rmr6-1/rmr6-1
R-r′/R-r′	48.5 ± 8.8 (8)	70.2 ± 1.4 (3)	P < 0.05
R-st/R-r′	35.6 ± 4.2 (7)	68.2 ± 7.8 (3)	P < 0.05

Reflectance measurements indicate relative R-r expression through its effects on kernel pigmentation. Average measurement values ± SD⁺ are presented for the total number of testcross individuals shown in parentheses.

^aRmr6/− designation represents both Rmr6/rmr6-1 and Rmr6/Rmr6 genotypes.

^bA two-sample t-test was applied to the null hypothesis that the differences between average measurements derived from Rmr6 and rmr6-1/rmr6-1 genotypes are due to random chance.

DISCUSSION

Molecular expression and genetic segregation analyses with rmr6 mutations show that normal Rmr6 function maintains transcriptional repression of Pl′ and B′ relative to Pl-Rh and B-I reference states and is required for acquisition of meiotically heritable paramutant states at the pl1, b1, and r1 loci. Pedigree analyses following specific Pl1-Rhoades alleles further illustrate that Rmr6 is required to stably maintain meiotically heritable Pl′ states; Pl′ fully reverts to a transmissible nonparamutagenic Pl-Rh state in over one-third of gametes derived from rmr6-1 homozygotes and nearly two-thirds of gametes transmitted from rmr6-2 homozygotes. R-r′ similarly reverts to meiotically heritable nonparamutant R-r states. Among Pl-Rh revertants arising in rmr6-1 homozygotes, however, a Pl′ state is often restored in the next generation, suggesting that persistence of heritable information confers instability to Pl-Rh states. The Rmr6-encoded factor thus defines heritable patterns of gene regulation as a common component of the maize paramutation mechanism by maintaining meiotically heritable epigenetic alterations affecting transcriptional regulation.

Rmr6 function:

Only specific alleles are affected by Rmr6 action. Conceptually analogous to classical position effects (Muller 1930; Shultz 1936) in which the chromosomal environment can expose genes to heterochromatin-based influences, only certain alleles or transgene insertions exhibit paramutation or paramutation-like behaviors (reviewed in Chandler and Stam 2004). Because Mop1 (Dorweiler et al. 2000), Rmr1, Rmr2 (Hollick and Chandler 2001), and Rmr6 functions act specifically on alleles subject to paramutation, it will be interesting to know if paramutation is a general characteristic of other genomic targets of Rmr6 action.

Rmr6 is most similar to Mop1 (Dorweiler et al. 2000) in its requirement for induced paramutation at r1, b1, and pl1 loci, somatic maintenance of the B′ transcription state, and meiotic maintenance of Pl′. The fact that pl1 RNA levels are elevated in Pl′/Pl′; mop1-1/mop1-1 plants (Dorweiler et al. 2000) suggests that Pl′ transcription is also affected by Mop1 action. Mop1 and Rmr6 therefore appear to play parallel or epistatic, rather than redundant, roles in maintaining B′ transcriptional states. However, the observations that Pl′ can revert to the Pl-Rh state within a single generation of being homozygous for rmr6 mutations while similar reversions occur only after Pl′ states are carried through at least two successive sporophyte generations of being homozygous for mop1 mutations (Dorweiler et al. 2000) highlight an intriguing distinction. Both Rmr1 and Rmr2 also act specifically on Pl1-Rhoades alleles to maintain repressed pl1 RNA levels of Pl′ states and are required to maintain its meiotically heritable paramutagenicity (Hollick and Chandler 2001). It remains unclear whether loss of paramutagenicity in plants homozygous for rmr1 and rmr2 mutations is similarly correlated with loss of transcriptional repression. Ongoing tests of Rmr1 and Rmr2 functions in b1 and r1 paramutation will determine whether these, like Mop1 and Rmr6, also represent common components of the paramutation mechanism.

Functional nonequivalence in several respects suggests that rmr6-1 is a partial loss-of-function allele. The rmr6-2 mutation represents a more severe loss-of-function allele possibly equivalent to a null situation. Both r1- and b1-induced paramutations, however, were severely or totally impaired in rmr6-1 homozygotes whereas pl1-induced paramutation was only partially inhibited. Similarly, reversions of Pl′ to Pl-Rh states occurring in rmr6-1 homozygotes were variable from one plant to another and were, in general, inherently less stable than those occurring in rmr6-2 homozygotes. A parent-of-origin effect regarding Pl-Rh stability (Table 4) reveals additional variability manifest by partial activity of the rmr6-1 allele. Collectively, these observations point to a quantitative nature of Rmr6 action occurring during somatic maintenance of Pl′ states.

The fact that Rmr6-dependent maintenance functions are required for meiotic inheritance of both Pl′ and R-r′ states illustrates its involvement in maintaining heritable epigenetic alterations affecting transcriptional control. In this regard, Rmr6-encoded activity resembles the function of Swi6p—the presumed Drosophila melanogaster heterochromatin protein 1 ortholog in Schizosaccharomyces pombe—which is required to maintain specific transcription states through meiosis (Nakayama et al. 2000). However, Rmr6 effects, directly or indirectly, on cytosine methylation patterns are also considered. Although no regional cytosine methylation differences distinguish Pl′ from Pl-Rh states (Hollick et al. 2000), it is possible that distant regulatory elements, similar to those found reversibly modified by cytosine methylation upstream of B1-I (Stam et al. 2002a,b), may have functional consequences on Pl1-Rhoades transcription. DNA methylation and/or chromatin alterations might determine differential interactions between enhancers, silencers, or insulators affecting loop-domain organizations affecting subsequent transcriptional activities similar to those recently documented in murine Igf2 H19 parent-of-origin imprinting (Murrell et al. 2004).

Genetic nature of paramutant states:

Paramutant states share two general properties: they reflect a meiotically inherited muted expression pattern relative to reference nonparamutant states, and they facilitate paramutations in trans (paramutagenicity). Paramutation simply describes the process of adopting a paramutant state. This process, by definition, takes place during one sporophytic generation and is manifest in the following. In this study, paramutant R-r:standard and B1-I states were evaluated only in regards to pigmenting function following sexual transmission whereas paramutant Pl1-Rhoades states were defined by heritable paramutagenic properties. When paramutagenicity is measured by reduced pl1 expression patterns following testcrosses to plants homozygous for nonparamutant states, the pigmenting function and paramutagenicity properties are inseparable. In plants homozygous for mop1, rmr1, rmr2, or rmr6 mutations, however, it is obviously useful to discriminate between phenotypes suggestive of nonparamutant states vs. experimental evidence derived from inheritance tests. Spontaneous examples emphasize that the paramutation behavior is an endogenous property of these unique alleles and invites the concept that paramutagenicity reflects a permissive environment in which spontaneous paramutation is favored (Styles and Brink 1969). In previous studies of b1 and pl1 paramutation, somatic sectors of muted plant color in B-I/B-I; Pl-Rh/Pl-Rh plants presage the appearance of paramutant states in the immediately subsequent generation (Coe 1966; Walbot 2001). Hence paramutation is an inherent property of Pl1-Rhoades alleles that appears sensitive to allelic interactions potentiated by Mop1, Rmr1, Rmr2, and Rmr6 activities.

Maintenance vs. establishment of paramutation:

Given that Rmr6 somatically maintains Pl′ and R-r′ paramutant states, the role of Rmr6 in establishment of a paramutant state is in question. Without somatic maintenance, establishment cannot be detected. Somatic maintenance is required for meiotic transmission of a somatically adopted paramutant state and for paramutagenicity. The fact that B′ does not revert to the B-I state in rmr6-1 homozygotes indicates that transmission of ∼50% B-I states from B-I/B′ individuals reflects failure of B-I to acquire a heritable B′ state. This is surprising, given that ∼40% of Pl1-Rhoades alleles in Pl-Rh states changed to Pl′ in rmr6-1/rmr6-1; Pl′/T Pl-Rh plants. If Rmr6 acted similarly at b1 and pl1, we would predict that some b1 paramutation would occur in rmr6-1/rmr6-1; B-I/B′ plants. Similarly, why should B′ not revert to B-I in rmr6-1 homozygotes? One possibility is that new B′ states induced in B-I/B′ heterozygotes are not somatically maintained and hence not transmitted in B′ state from rmr6-1/rmr6-1 plants. Patterson and Chandler (1995) showed that newly formed B′ states from B-I/B′ plants are distinguished from the inducing B′ state in terms of pigmenting potential and Stam et al. (2002a) showed that cytosine demethylation of the upstream repeats associated with the B′ state lags approximately one generation from the original B-I to B′ induction event occurring in B-I/B′ plants. Thus previously established paramutant B′ states may persist in rmr6-1 and mop1-1 homozygotes, while somatically formed B′ states may not. In light of this possibility, it remains unclear as to whether Rmr6 or Mop1 are required for the actual inductive event necessary for establishment or more simply are required to somatically maintain labile paramutant states. These two functions, however, need not be mutually exclusive. Regardless, this apparent discrepancy of Rmr6 function in B-I/B′ vs. R-r/R-st and Pl-Rh/Pl′ genotypes underscores the unique features of these alleles that differentially utilize common components of the basic paramutation mechanism.

Transcriptional repression and paramutagenic activity:

Rmr6 action maintains the relatively low Pl′ transcription rates in husk tissues. The approximately fourfold increase in average transcription rate found in rmr6-1 homozygotes is slightly greater than differences between Pl′ and Pl-Rh states in Rmr6/Rmr6 materials (approximately threefold; Hollick et al. 2000). Consistent with this greater transcription rate measured in husk tissues, there is an approximately twofold greater level of pl1 RNA in mutant floret tissues (Pl′/Pl′; rmr6-1/rmr6-1) relative to Pl-Rh reference states (Pl-Rh/Pl-Rh; Rmr6/Rmr6). Despite this hyperexpression of pl1, a significant fraction of Pl1-Rhoades alleles are transmitted in paramutagenic Pl′ states. Thus neither pigment nor pl1 molecular expression phenotypes found in plants homozygous for mutant rmr6 alleles are a reliable indicator of subsequent paramutagenic activity. While heritable loss of paramutagenic activity occurs only in plants with Pl-Rh-like levels of pl1 expression (Hollick and Chandler 1998, 2001; Dorweiler et al. 2000), such RNA levels or transcriptional activities are not sufficient to stabilize a Pl-Rh state through meiosis nor fully prevent induced changes of Pl-Rh to Pl′ occurring in Pl-Rh/Pl′ genotypes. These observations point to regulatory features upstream of transcriptional activation, such as those affecting chromatin or nuclear architecture, as the target of Rmr6 action as well as the source of heritable paramutagenic activity.

The nature of paramutagenic action:

While our results clearly implicate that Rmr6 functions in the acquisition of paramutant states at multiple loci, its specific role is unknown. For example, in a Pl-Rh/Pl′ plant, Rmr6 could be required to adopt a meiotically heritable paramutagenic Pl′ state, to mediate paramutagenicity from the Pl′ state, to maintain paramutagenicity of the Pl′ state, or some combination thereof. It is difficult to discriminate among these conceptually distinct functions since the assays for stable paramutational changes are spatially and temporally separated from site and timing of Rmr6 function. On the basis of the observed reversions of Pl′ to Pl-Rh in Pl-Rh/Pl′; rmr6-2/rmr6-2 plants and R-r′ to R-r in R-st/R-r′; rmr6-1/rmr6-1 plants, the most parsimonious inference is that Rmr6 action maintains paramutagenicity in somatic lineages. Our results with the rmr6-1 mutation indicate that paramutagenicity, a property measured after meiosis, can be uncoupled to some extent from transcriptional regulatory activities measured in the preceding sporophytic generation. Even in cases where experimental evidence suggested heritable loss of paramutagenicity, our ability to follow specific Pl1-Rhoades alleles through subsequent generations showed that many of these revertant Pl-Rh-like states retained a latent mark allowing them to regain full, or partial, paramutagenic activity following the next meiotic segregation. Taken together, these results suggest that paramutagenic activity transmitted at meiosis is based on preexisting epigenetic marks that may, or may not, have altered transcriptional activities affecting changes in pigmentation. Rmr6 function serves to define and maintain this epigenetic mark.

An emerging genetic mechanism for meiotically heritable change:

So far, four functions—encoded by Mop1, Rmr1, Rmr2, and Rmr6—that are required to maintain reduced RNA expression typical of the Pl′ paramutant state have been described. Five other loci are currently under investigation (J. Hollick, unpublished results). Thus, at least nine genetic components defined by mutations are required for some aspect of pl1 paramutation. As specific sequences responsible for p1 and b1 parmutation are now identified (Sidorenko and Peterson 2001; Stam et al. 2002a,b), these mutations should help illuminate functionally relevant molecular alterations associated with paramutational change. Identification of sequences responsible for pl1 paramutation will be similarly important in understanding functional changes responsible for meiotically heritable alterations of transcriptional regulation. Discovery of other genomic targets of Mop1 and Rmr actions promises to clarify additional roles of paramutation in higher eukaryotes.