Transcriptionally Active MuDR, the Regulatory Element of the Mutator Transposable Element Family of Zea mays, Is Present in Some Accessions of the Mexican land race Zapalote chico

Corresponding author: Virginia Walbot, Department of Biological Sciences, Stanford University, Stanford, CA 94305-5020, walbot{at}leland.stanford.edu (E-mail).

Communicating editor: J. A. BIRCHLER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

To date, mobile Mu transposons and their autonomous regulator MuDR have been found only in the two known Mutator lines of maize and their immediate descendants. To gain insight into the origin, organization, and regulation of Mutator elements, we surveyed exotic maize and related species for cross-hybridization to MuDR. Some accessions of the mexican land race Zapalote chico contain one to several copies of full-length, unmethylated, and transcriptionally active MuDR-like elements plus non-autonomous Mu elements. The sequenced 5.0-kb MuDR-Zc element is 94.6% identical to MuDR, with only 20 amino acid changes in the 93-kD predicted protein encoded by mudrA and ten amino acid changes in the 23-kD predicted protein of mudrB. The terminal inverted repeat (TIR) A of MuDR-Zc is identical to standard MuDR; TIRB is 11.2% divergent from TIRA. In Zapalote chico, mudrA transcripts are very rare, while mudrB transcripts are as abundant as in Mutator lines with a few copies of MuDR. Zapalote chico lines with MuDR-like elements can trans-activate reporter alleles in inactive Mutator backgrounds; they match the characteristic increased forward mutation frequency of standard Mutator lines, but only after outcrossing to another line. Zapalote chico accessions that lack MuDR-like elements and the single copy MuDR a1-mum2 line produce few mutations. New mutants recovered from Zapalote chico are somatically stable.

MOLECULAR, genetic, and anthropological data indicate that maize arose in what is now Mexico through domestication of teosinte, a grass closely related to present-day maize (GOODMAN and BROWN 1988 ). Domestication is proposed to have occurred once, not more than 10,000 years ago (DOEBLEY 1990 ). Because this is such a recent event on an evolutionary time scale, corn should have a very homogeneous genome. However, both the allelic diversity (SHATTUCK-EIDENS et al. 1990 ) and the range of genome size (from 9.82 to 12.12 pg; RAYBURN et al. 1985 ) are among the highest for any eukaryotic species. In the short time span since domestication, transposable elements and selection for growth in many habitats are proposed to have played important roles in first generating and then maintaining this diversity (SCHWARZ-SOMMER et al. 1985 ; KLOECKENER-GRUISSEM and FREELING 1995 ; WALBOT 1996 ).

Characteristically, active transposable elements are found in a few populations within a species, and both the origin and maintenance of transposable elements remain enigmas (CAPY et al. 1994 ). Within a lineage, transposons are transmitted in predictable, albeit non-Mendelian, patterns and often exhibit mechanisms to increase copy number. In addition, analysis of P and Mariner elements in Drosophila spp. and insertion sequences in bacteria implicate horizontal transmission as a possible explanation of the punctate distribution of active transposons, at least for some cases (CAPY et al. 1994 ).

The Mutator transposons of maize are one of the most active transposable element families described in any organism. Standard Mutator activity is defined by a suite of characters: high forward mutation frequency (10^-3 to 10^-5 per gene per generation), frequent somatic excision late in development, and infrequent germinal excision (<10^-4 to 10^-5 per gene per generation) (reviewed in WALBOT 1992 ). These features have only been observed in plants derived from a single line of maize, first described as Mutator by ROBERTSON 1978 . These standard Mutator lines have multiple copies of a diverse family of transposable elements. Mu elements share ~210-bp Terminal Inverted Repeats (TIRs) and create 9-bp host sequence duplications at the site of insertion. The Mu transposon family is composed of at least six distinct subfamilies that are distinguished by unique internal sequences between the TIRs (reviewed in BENNETZEN 1996 ). In standard Mutator lines, the 4942-bp MuDR regulatory elements are present in 5–50 copies (HERSHBERGER et al. 1991 ), and non-autonomous Mu element copy numbers are even higher (ALLEMAN and FREELING 1986 ; TAYLOR and WALBOT 1987 ; reviewed in WALBOT 1991 ). All of the Mu elements exhibit non-Mendelian inheritance, with copy number maintained on outcrossing by replicative transposition late in the sporophytic or during the gametophytic phase of the life cycle (reviewed in WALBOT 1991 ; BENNETZEN 1996 ). Collectively, the multi copy Mu elements increase the mutation frequency 20–100-fold or more above the spontaneous level (reviewed in WALBOT 1992 ; BENNETZEN et al. 1993 ). In these standard Mutator lines, loss of Mutator activity is an epigenetic phenomenon rather than the result of segregation of MuDR; loss of activity is correlated with increased methylation of the regulatory MuDR and non-autonomous Mu elements (reviewed in CHANDLER and HARDEMAN 1992 ; BENNETZEN 1996 ).

A few exceptional Mutator lines contain only a single MuDR that segregates as a near-Mendelian factor; these lines were derived from standard, high copy number Robertson lines with the a1-mum2 or a1-mum3 reporter alleles (ROBERTSON and STINARD 1989 , ROBERTSON and STINARD 1992 ; CHOMET et al. 1991 ; LISCH et al. 1995 ). In the most thoroughly analyzed example, the MuDR element is located on chromosome 2L (ROBERTSON and STINARD 1992 ; LISCH et al. 1995 ). At this location, it programs the standard pattern of high frequency somatic excision. However, few element insertions occur and both MuDR and Mu element copy numbers typically remain low (LISCH et al. 1995 ). Independently, SCHNABLE and PETERSON 1986 SCHNABLE and PETERSON 1988 SCHNABLE and PETERSON 1989A (SCHNABLE and PETERSON) SCHNABLE and PETERSON 1989B described Cy/r-cy, a two-element transposable element system that has turned out to be a low activity Mutator line; Cy lines often contain a single, segregating regulatory element, now known to be a MuDR (HSIA and SCHNABLE 1996 ). The sequence of MuDR (HERSHBERGER et al. 1991 ) is identical to Cy (HSIA and SCHNABLE 1996 ) and, with the exception of a single, inconsequential base change, identical to the single MuDR in the a1-mum2 lines (JAMES et al. 1993 ). Consequently, the differences between high- and low- activity Mutator lines cannot be explained by differences in the primary sequence of MuDR.

Based on Southern hybridization, all Zea species tested contain multiple, dispersed sequences homologous to segments of Mu elements; some of these widely shared sequences appear to be parts of genes that have become incorporated into a Mu element (TALBERT et al. 1989 ). By Southern hybridization, TIR-homologous sequences are not found beyond the genus Zea and the maize X Tripsacum hybrid species Tripsacum andersonii (TALBERT et al. 1990 ). However, genomic clones with sequences similar to Mu TIRs and limited regions of similarity to MuDR have been reported in rice (EISEN et al. 1994 ; ISHIKAWA et al. 1994 ). Within maize, MuDR is not widely distributed (HERSHBERGER et al. 1991 ), nor is Mutator activity. The largest survey to date tested maize lines for a Cy capable of activating somatic instability of bz1-rcy. SCHNABLE and PETERSON 1986 found that active Cy elements were nearly restricted to the original Cy line and Robertson's Mutator lines. Weak Cy activity was found sporadically in a few plants in 7 of 47 other lines surveyed, but this rare occurrence could represent activation of cryptic regulatory elements from the Cy parent; this possibility could not be tested as MuDR had not yet been cloned. Because standard Mutator activity creates so many mutations, it is not surprising that MuDR is apparently missing from most strains of corn.

We are interested in whether MuDR can be maintained in the maize genome. To address this question, we surveyed for MuDR in American inbreds, exotic maize lines, and Zea spp. by Southern blot hybridization. While nearly all lines had cross-hybridizing fragments, only the Mexican land race Zapalote chico had a multicopy cross-hybridizing band of approximately the correct size. Analysis of different accessions of Zapalote chico demonstrated that only a subset of the population contains MuDR-like elements; only the Zapalote chico lines with MuDR-like elements exhibited a high forward frequency.

The sequenced example of a MuDR-like element of Zapalote chico (MuDR-Zc) is highly similar to MuDR and encodes similar transcripts. Unlike standard Mutator lines, which generate new mutants during selfing and outcrossing, Zapalote chico exhibits hybrid dysgenesis. Self-pollinated lines produce few mutations, but outcrosses to non-Mutator lines activate a high forward mutation frequency. In addition, new mutants are somatically stable, at least in seedlings. Zapalote chico is cultivated by the Zapotecs, a Native American people of Oaxaca, Mexico. This line is their economic staple and by their oral history has been cultivated for more than 5000 years, tracing to their cultural origin in the highlands of Central Oaxaca. We discuss the possibility that selection for a high-yielding stable crop has resulted in the novel properties of the Mutator system in Zapalote chico.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Mutator terminology:
MuDR is the regulatory Mutator element, and this name replaces prior nomenclature: MuA2 (QIN et al. 1991 ), MuR1 (CHOMET et al. 1991 ), Mu9 (HERSHBERGER et al. 1991 ), and Cy (HSIA and SCHNABLE 1996 ). Mu elements share ~200-bp TIRs with MuDR and require transcriptionally active MuDR to be mobilized.

Plant material:
Active Mutator individuals with the bz2-mu4::MuDR allele (family M87) were used as the Mutator stock for DNA and RNA analysis; this is a multicopy, standard Mutator line with a MuDR transposon inserted in the second exon of the bz2 gene (HERSHBERGER et al. 1991 ). The bz2 tester line in inbred W23 was used as the non-Mutator stock for most molecular analyses and for the mutagenesis tests; an a2 tester constructed in W23 by this lab, and an a2 tester obtained from the Maize Genetics Cooperation Stock Center (Urbana, IL) in a mixed nuclear background, were used for the Southern blot survey. For the original survey, all of the exotic lines of maize and the Zea spp. (listed in 1982 inventory as Zea mexicana luxurians, mexicana, nobogame, parviglumis, and peruviensis, Z. perennis, and Z. diploperennis) were obtained from the Stock Center. The Zea spp. collections now reside at the USDA Plant Introduction Station (Ames, IA). For subsequent experiments, existing Zapalote chico accessions were obtained from Pioneer Hi-Bred (Johnston, IA) and an overlapping set from CIMMYT (International Maize and Wheat Improvement Center, Texcoco, Mexico). Both RONALD PHILLIPS and RICHARD KOWLES provided several generations of crosses between Zapalote chico (cytogenetically many knobs) and Wilbur's Knobless Flint. Thirty-five new accessions were collected as individual ears in Juchitán, (Oaxaca, Mexico) (16.15N, 95.00W) directly from Zapotec farmers; two Tuxpeño X Zapalote chico F₁ hybrids and the F₂ backcross ears were donated by M. C. ARREDONDO, a retired Mexican corn breeder living near Juchitán. Tuxpeño is a widely adapted inbred line developed in Mexico and used as the foundation for breeding experiments.

Forward mutation test:
Individuals were self-pollinated to assess the phenotypes of any pre-existing mutations and crossed as pollen to the bz2 tester. The outcross seed were planted, and the F₁ plants were self-pollinated, yielding F₂ ears. Thirty progeny kernels of the F₂ and selfed parental ears were planted side-by-side in the summer field; mutants were counted in the F₂ only if they were clearly distinguishable from any segregating phenotype in the parent. All novel phenotypes recorded appeared to be recessive, present in ~one-quarter of the progeny. In ambiguous cases, i.e., in which there was low germination or only one or a few mutant plants were present, a second sample of 30 kernels was planted and evaluated. Somatic mutability was scored by eye and by observation through a stereozoom microscope (x20).

Ten kernels of each Zapalote chico accession were grown in summer, 1993 (M designations), at Stanford. Seed were planted in late June to promote flowering, because maturation of neotropical maize is inhibited by long days in the temperate zone. In most accessions, only a few individuals reached maturity within 75–90 days and could be both self-pollinated and crossed as pollen parent to bz2 tester. In the 1994 winter nursery in Molokai, Hawaii, Zapalote chico lines matured within 50–55 days, and additional representatives of some lines were self-pollinated and crossed to bz2 tester. Of all the Zapalote chico samples examined, one line M59 = N234 was the most consistent in flowering at Stanford, and additional individuals of the original accession were tested for forward mutations during 1994–1995. To assess spontaneous mutation frequency in a non-Mutator line, the crossing scheme was used with the bz2 tester. Four standard Mutator lines were used for comparison; two lines were selfed and crossed to bz2 in 1993 (M88, bz2-mu2::Mu1 reporter allele; M121, bz2-mu1::Mu1 reporter allele), and two lines in 1994 (N190, bz1-mu1::Mu1 early somatic excision line; N285, Mutator with Bz1-revertant alleles from the early excision of the bz1-mu1::Mu1 reporter allele). Two a1-mum2::Mu1, single MuDR lines were obtained from D. ROBERTSON and evaluated in 1994 to compare the low MuDR copy number lines to Zapalote chico.

RNA blot analysis:
Immature ears were collected from field-grown material of each line during the summer of 1994. Tissue was immediately frozen in liquid nitrogen and stored at -80° until RNA isolation. RNA was isolated by grinding the samples in liquid nitrogen, then extracting with Tri-Reagent (Molecular Research Center, Cincinnati, OH). Poly(A)⁺ RNA was purified from total RNA using a Mini-oligo(dT) cellulose spin column kit (5 Prime3 Prime, Boulder, CO).

For the RNA blots, 16–20 µg of poly(A)⁺ RNA was electrophoresed through an agarose formaldehyde gel for 6 hr and transferred to Hybond-N (Amersham, Arlington Heights, IL) using standard techniques (SAMBROOK et al. 1989 ). Two probes were generated by PCR amplification from pMuDR. This plasmid was constructed and sequenced by R. J. HERSHBERGER; it contains a full-length MuDR element, recovered from the bz2-mu4::MuDR allele (HERSHBERGER et al. 1991 ), with a one base frameshift mutation in mudrA that allows maintenance in Escherichia coli. Probe PA contains 927 nucleotides of mudrA (positions 183–1100), and PB contains 978 nucleotides of mudrB (positions 3774–4752). A third probe, BX1.0 (HERSHBERGER et al. 1995 ), was recovered from a BamHI (nucleotide position 2865) to XbaI (nucleotide position 3945) digest of pMuDR; this probe recognizes both mudrA and mudrB (Figure 1). Probes were labeled by the random primer method, using the DECAprime II Kit from Ambion, Inc. (Austin, TX) (FEINBERG and VOGELSTEIN 1983 ) and purified on push columns (Stratagene, La Jolla, CA). Prehybridization and hybridization were performed according to the protocol published for Gene-Screen (Du Pont, Wilmington, DE) using 10% dextran sulfate. Filters were washed once in 2x SSPE, 1% SDS at room temperature for 10 min, once in 1x SSPE, 1% SDS at 65° for 15 min, and once in 0.1x SSPE, 0.1% SDS at 65° for 15 min. Autoradiography was performed for 12–72 hr at -80° using two intensifying screens.

View larger version (20K): In this window In a new window Download PPT slide   Figure 1. —Diagram of MuDR. DNA probes (PA, PB, BX1.0) used in Southern and Northern hybridizations are shown above the transposon structure. Below the transposon, the mudrA and mudrB convergent transcription units are illustrated; the first intron of each transcript is spliced ~100%, the second introns are spliced ~80%, as is the third intron of mudrA. A 120-bp in-frame intron of mudrB that is spliced in about 5% of transcripts of MuDR is not shown. (A) Regions of MuDR for which PCR amplification was attempted in various accessions of Zapalote chico. (B) PCR fragments generated in the cloning of the MuDR-like element. Nucleotide sequence numbering according to HERSHBERGER et al. 1991 . B, BamHI; D, DraI; E, EcoRI; H, HindIII; S, SstI; X, XbaI.

DNA blot analysis:
Maize genomic DNA was prepared from immature ears of selfed Zapalote chico accessions grown in 1994 (N designations in Tables) and purified as described by STAPLETON and WALBOT 1994 . For Southern analysis, three µg of DNA were digested with restriction enzymes (BRL, Gaithersburg, MD) according to the manufacturer's instructions, electrophoresed through agarose gels, and blotted onto Hybond-N (Amersham). Probes were prepared as described above. The blots were prehybridized, hybridized, and washed as recommended by the membrane manufacturer. To quantify MuDR copy number, a plasmid containing MuDR was digested with SstI or SstI/DraI, diluted to the proper concentration equivalent to a specific copy number in the maize genome, and electrophoresed next to restriction digests of maize genomic DNA. Blots were probed with BX1.0. In some cases, stripped blots were reprobed with a 380-bp fragment of Adh1 as a loading control. To check for the presence of Mu1 and the related Mu2 elements, probe pA/B5 was used (TAYLOR and WALBOT 1987 ).

DNA amplification by polymerase chain reactions (PCR):
DNA amplification reactions were performed in volumes of 25–100 µl overlaid with 50–100 µl paraffin oil. Each reaction contained 0.2 mM of each of the four deoxyribonucleotides, 100 ng of each oligonucleotide primer, a buffer (15 mM Tris pH 8.3, 50 mM KCl, 1% Gelatin, 1.8 mM MgCl₂), Taq DNA polymerase (Perkin-Elmer, Norwalk, CT), and 50–100 ng of DNA. PCR reactions were carried out for 30 cycles of 1 min at 94°, followed by 1 min at 55°, and 1 min at 72°. The following DNA primers were used for mudrA: primer #183 5'-CGCCGTCTGGCAGGGCCTCTTGTCACCGTCTC-3' with primer #1996 5'-GAATGTCATAGGTTGCATAG-3' or primer #2017 5'-GATACGTTGGATACTGTAAG-3' with primer #2282 5'-TATGGATGTAGAGACCTTAG-3'. For the mudrA/intergenic region primer #2281 5'-GATTCCAGAGATGTAGGTAT-3' was used with primer #813 5'-CCAACCAAAGTAAGACCACA-3'. For the intergenic region to mudrB region, primer #2019 5'-GCCATTAGTTCTTACAACCT-3' was used with primer #2109 5'-ACAATACGCGTTAACCAAACA-3'. To amplify mudrB primer #3773 5'-CTTGTACAGATCTTGTGACCAGTCGCA-3' was used with primer #4752 5'-GTCCACAAATCGATGTTACGGTCGTT-3'. For TIRA primer #2466 5'-GCTGAGCCTCCTGCAGGGAGATAATTGCC-3' was used with primer #2467 5'-CCATGGTACCAAAATCAGAG-3'. The resulting fragment contains all of TIRA, plus the region of the 5' untranslated region that contains a transcriptional start site. To amplify TIRB, primer #2468 5'-TGAACGCCTCCTGCAGGAGAGATAATTGC-3' was used with primer #2470 5'-CAATCGGTACCCACAGGAGCAAGAG-3'. The resulting fragment contains TIRB plus the 5' untranslated region of mudrB.

Plasmids:
Seven plasmids were constructed by amplifying genomic DNA of Zapalote chico line N215 by PCR with the primers listed above. Amplified fragments were cloned into the pCR2.1 vector (Invitrogen, Carlsbad, CA). The MuDR-zc element was cloned in overlapping fragments, because the full-length MuDR is toxic to E. coli. Bracketed numbers to the right of each plasmid correspond to the base pairs of the standard MuDR sequence (numbering according to HERSHBERGER et al. 1991 ). Plasmid pTIRAzc [1–455] has a 455-bp insert; pA1zch [183–1110] has a 927-bp insert; pA2zch [1091–2423] has a 1332-bp insert; pA3zch [2404–3705] has a 1301-bp insert; pA4zch [3554–4580] has a 1026-bp insert; pB1zch [3773–4752] has a 979-bp insert; and pTIRBzc [4476–4944] has a 468-bp insert.

DNA sequencing:
MuDR-Zc regions were obtained as restriction fragments from the pAzch plasmid series and were subcloned into the M13mp19 vector for single-stranded sequencing (NORRANDER et al. 1983 ) with the Sequenase 2.0 kit (United States Biochemical, Cleveland, OH) or with the ABI 310 fluorometric automated sequencer. Both strands of all fragments were fully sequenced. To eliminate compression of bands that occurred when sequencing GC-rich regions, mixes containing deoxyinosine provided with the kit were used. Primers for sequencing were commercially available M13 primers; a few custom internal primers were used on long fragments. The MuDR-Zc sequence is registered in GenBank as accession number U75360.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Properties of MuDR:
MuDR encodes two, convergently transcribed genes (Figure 1). The major transcription initiation sites are in the TIRs, and the most abundant transcripts are 2.8 kb (mudrA) and 1.0 kb (mudrB). Intron skipping, multiple polyadenylation sites, and a second transcription initiation site in the 5' UTR of mudrA result in four distinct transcript types for each gene (HERSHBERGER et al. 1995 ). Although the exact roles of the proteins encoded by mudrA and mudrB are unknown, mudrA encodes a polypeptide with homology, over an extended motif of ~150 amino acids, to a suite of bacterial transposons (EISEN et al. 1994 ). In addition, it has recently been demonstrated that mudrA encodes a DNA-binding protein which binds to specific sequences within the highly conserved Mu TIR, leading to the proposal that MURA is a transposase (BENITO and WALBOT 1997 ). Deletions within mudrA in lines with a single MuDR abolish somatic instability of the a1-mum2 reporter allele; this evidence demonstrates that mudrA is essential for somatic excision (HSIA and SCHNABLE 1996 ; LISCH and FREELING 1994 ).

Distribution of MuDR elements:
Previously, we reported that the 4.7-kb SstI fragment characteristic of an intact, unmethylated MuDR element (Figure 1) was multicopy in Mutator lines and was not present in standard inbreds of maize (HERSHBERGER et al. 1991 ). Most non-Mutator lines did contain various sized fragments that hybridized to one or more internal MuDR probes, but there was no evidence for intact MuDR elements. To expand the analysis of distribution of MuDR-like elements, genomic Southern blotting was used to screen additional inbreds, exotic lines, and Zea spp. for intact MuDR elements. Genomic DNA was digested with SstI, which recognizes sites in unmethylated TIRs of MuDR, a Southern blot was prepared, and then hybridized with the BX1.0 fragment, which contains the 3' portions of both MudrA and MudrB and the intergenic region (Figure 1).

Of the lines examined by this Southern blot survey, a Co-op accession of Zapalote chico had a fragment about the size expected for MuDR (~5.1 kb). This fragment was slightly larger than MuDR and was present in ~3–5 copies per genome (data not shown).

All of the other exotic lines examined, including Argentine popcorn, Tama flint, Strawberry popcorn, Papago flint, gourd seed, Northern flint, Z. perennis, Z. diploperennis, and five teosinte types (see MATERIALS AND METHODS), hybridized weakly to the central MuDR probe. Similar cross-hybridization has been found in some (W23, K55, and A188) standard maize inbred lines (data not shown).

Screening for MuDR in Zapalote chico lines by PCR and Southern analysis:
The first Zapalote chico sample examined was collected in the 1950s from Oaxaca Mexico; it has been maintained by the Maize Genetics Stock Center, by periodically growing and selfing the line. Using eight sets of PCR primers that spanned most of MuDR, we determined that all regions of this putative regulatory element in Zapalote chico could be amplified from an immature ear DNA sample of one individual. Furthermore, seven of the fragments were the expected size, and each of these contained one or two restriction sites at the same positions as in MuDR. There were no polymorphisms for the 12 enzyme sites examined (data not shown). With the primer pair that spanned the intergenic region, however, several size variants were detected, ranging from 100 to 300 bp larger than the comparable region of MuDR (data not shown). As the intergenic region is composed of a complex set of repetitive elements (HERSHBERGER et al. 1995 ), we hypothesized that there had been an expansion of these motifs. Collectively, the results suggested that Zapalote chico contained elements that were very similar to MuDR.

To assess the distribution of MuDR-like elements in Zapalote chico populations, existing accessions were obtained from three other sources: seven examples of old accessions were obtained from CIMMYT, and a mostly overlapping set was obtained from Pioneer Hi-Bred. The CIMMYT materials were collected in the 1950s and 1960s, but then maintained under different growth conditions in central Mexico and in Iowa. RONALD PHILLIPS and RICHARD KOWLES contributed Zapalote chico X Wilbur's Knobless Flint hybrids, derived from a CIMMYT accession. Zapalote chico is classified as a land race, but it is also an economically important line. It is the only corn variety grown by the 300,000 Zapotecs living in southwestern Mexico. To obtain a current representation of Zapalote chico, 35 new accessions were collected in 1993 from farmers and from a corn-breeding program located in the main Zapotec population center, near Juchitán in the state of Oaxaca, Mexico. Most of these lines were successfully propagated at Stanford University in summer 1993. We conducted a more extensive investigation of the distribution of MuDR elements using Southern blot analysis and PCR experiments of genomic DNA samples from accessions that could be self-pollinated.

For Southern blot analysis, genomic DNA samples most likely to contain full-length elements (based on PCR screening; see Table 1 below) were digested with SstI and probed with BX1.0. Figure 2 shows the ~4.7, kb SstI fragment characteristic of an intact MuDR element was conserved in lines N215, N234, and N237, although the Zapalote chico hybridizing bands were always slightly larger (~50–100 bp) than MuDR from a standard Mutator line. Zapalote chico line N216 contained a fragment that is ~250 bp larger than MuDR; it may be similar to the larger MuDR-like element originally identified in the Maize Stock Center material.

View larger version (60K): In this window In a new window Download PPT slide   Figure 2. —Screening for MuDR in Zapalote chico lines by Southern analysis. 6–8 µg DNA samples were digested with SstI and probed with BX1.0.

To extend our analysis to additional lines, PCR experiments were carried out on samples from individual selfed progeny, using four sets of primer pairs that span MuDR (Figure 1). As shown in Table 1, eight samples (24%) yielded PCR products of the expected size with all four primer pairs; these samples represent seven distinct accessions, with the Oaxaca 50 accession represented from two distinct sources (N201 and Oaxaca 50 directly from CIMMYT). The majority (6/8) of the accessions positive for MuDR-like elements represented the most recently collected material, indicating that MuDR-like elements exist in the current Zapotec crops. Twenty accessions (61%) yielded products from a subset of the primer pairs, and five accessions (15%) did not seem to contain any amplifiable fragments. In this analysis, the particular individual sampled from the Maize Stock Center lineage (N240) gave a positive PCR result with only two primer pairs, suggesting that this individual did not contain an intact MuDR element. Also, two of the CIMMYT lines, Oaxaca 48 and Chiapas 224, yielded different PCR products in the two versions sampled. We conclude that there is heterogeneity within some accessions, reflecting either heterogeneity in the original material or changes during propagation at stock centers.

For a more detailed analysis of these MuDR-like elements, the PCR fragments generated from lines N234 and N215 were digested with enzymes for restriction sites present in the transcribed region of authentic MuDR elements. All ten of these enzymes produced fragments of identical size in digests from active Mutator lines and both N234 and N215 Zapalote chico accessions (data not shown). These data suggest that the differences in length between Zapalote chico MuDR-like elements and MuDR will be found in the TIRs and/or in the intergenic region.

Inheritance of the MuDR-like element:
To examine the propagation of MuDR-like elements through outcrosses with non-Mutator lines, Southern blots were perfomed in two lineages: (N234 = P56) and CIMMYT accession Oaxaca 2 (N264 and N265). The founder individual (M59) of the N234 lineage appears to have one copy of a MuDR-like sequence per haploid genome; it is 100 bp larger than MuDR. The M59 founder was crossed to bz2 tester, a non-Mutator source, and the progeny (individuals of family P56) contain ~1 copy of MuDR (Figure 3A, and as discussed below a PCR survey of 28 P56 individuals were all positive for MuDR). All P56 individuals could contain a single copy of MuDR if the founder had been homozygous for one MuDR locus. When individuals of P56 were outcrossed a second time to bz2 tester, the copy number of ~1 is maintained (compare parent P56-12 and outcross progeny, lanes 1 and 3; parent P56-17 and outcross progeny, lanes 2 and 4; Figure 3A). Figure 3B provides more evidence for transmission of the element through two outcrosses. M3-4 (lane 1) contains the MuDR-like element, and this element is maintained when outcrossed to bz2 tester (MH3, lane 2) and when MH3 was selfed to produce line N265 (lane 3). Siblings of line N265 are shown to contain the element (O70, O70-1, and O70-4, lanes 4–6), which is again maintained on selfing (OH59, P61, lanes 7 and 9), as well as after a second outcross (O70-4 x bz2, lane 8).

View larger version (193K): In this window In a new window Download PPT slide   Figure 3. —Inheritance of the MuDR-like element on outcrossing of MuDR-like lines to bz2 tester. Three µg samples were digested with SstI and probed with BX1.0. (A) Line P56, the F1 cross of M59 x bz2, and its progeny after a second outcross to bz2 tester. (B) CIMMYT accession Oaxaca 2, line M3-4 and its derivatives. (C) Line M3-1, a sibling of M3-4 and its derivatives. All Southern blots contain the non-Mutator line Tuxpeño, a Mutator line, and a copy reconstruction of MuDR plasmid. See text for details of each lineage.

Figure 3C shows a lineage of progeny of M3-1 (a sibling of M3-4), which was outcrossed once, repeatedly selfed, and outcrossed once more. Selfed progeny of M3-1 do not show the MuDR-like element (lane 1). MH2 (lane 2), selfed F₁ progeny from an M3-1 outcross to bz2 tester, do not show the element. Selfing of MH2 progeny produced line N264-1 (lane 3), which also lacks the MuDR-like element. However, after selfing of N264-1, bands of the correct MuDR-like element size appear in the siblings O69-5 and O59-9 (lanes 4 and 5) and persist in the F₁ of an outcross of O69-9 to bz2 tester (lane 6). These results demonstrate that cryptic, presumably methylated copies of MuDR-like elements exist in Zapalote chico accessions, and that these elements can appear during a crossing program.

Larger hybridizing bands are present in all of the above Southerns, including Tuxpeño, a non-Mutator line. The relationship of these larger fragments to MuDR cannot be ascertained from the available data, although it is interesting that at least some part of MuDR is widely distributed in the genus. The fragments could represent disrupted copies of MuDR, sequence similarity to either mudrA or mudrB, or methylated intact MuDR element as we suggest for the M3-1 individual. MuA, for example, is a larger MuDR-like element recovered from a Mutator line; it is disrupted by several insertions (QIN and ELLINGBOE 1990 ). Internal deletions within MuDR that retain the TIRs produce SstI fragments smaller than 4.7 kb (HERSHBERGER et al. 1995 ), but deletions missing the SstI site of one TIR could yield fragments larger than MuDR. As epigenetic loss of Mutator activity is correlated with DNA methylation, the larger fragments could also represent modified MuDR-like elements. The SstI (=SacI) sites (GAGCTC) in the TIRs are not followed by either G or NG, consequently, methylation of the "canonical" substrates CpG and CpNpG cannot explain the inability of these enzymes to digest methylated (epigenetic loss) MuDR elements (MARTIENSSEN and BARON 1994 ). Maize DNA can be methylated at other C residues (WANG et al. 1996 ), and it is possible that methylation at one or both of the internal C residues prevents digestion.

Distribution of MuDR-like elements in Zapalote chico families:
Given the diversity between and within accessions of Zapalote chico, we wished to determine the inheritance of MuDR-like elements in individual lines in which a founding individual was demonstrated to contain one or a few copies of the MuDR-like element. Our strategy was to PCR amplify the MuDR-like element in two halves (positions 113–2423, yielding a 2310-bp fragment, and from positions 2404–4829, yielding a 2425-bp fragment) that cover nearly the entire element. PCR analysis of 28 individuals of line P56 (progeny of bz2 x M59) indicated that all were positive for both halves of the MuDR-like element (data not shown); these data indicate either homozygosity of the M59 parent (although it was estimated to contain only a single MuDR-like element by Southern blotting) or copy number maintenance. Line P57 are progeny of the original Zapalote chico accession M62 (Tuxpeño X Zapalote chico) crossed to bz2 tester. In the 28 second outcross progeny examined, all tested positive for both halves of MuDR (data not shown). As the original individual had only 1–2 copies of MuDR, it seems likely that copy number is maintained in the stock either by transposition or by recruitment of formerly cryptic elements.

We conclude from the combination of PCR analysis and Southern blot hybridization tests that the parental Zapalote chico lines, which had only 1–2 copies of the MuDR-like element, transmitted the element to all progeny examined. This is circumstantial evidence that replicative transposition of the MuDR-like element occurs in Zapalote chico as is proposed for all Mu elements in standard lines (BENNETZEN 1996 ). The analysis is compromised, however, by possible recruitment of cryptic MuDR-like elements during the crossing scheme.

DNA sequence analysis:
Because MuDR and the gene, cDNA, and exon3 of mudrA are unstable in E. coli, the element and large subclones of it cannot be stably maintained on bacterial plasmids. Stable derivatives inevitably contain frameshift and deletion mutations that destroy the large open reading frame within exon3 (HERSHBERGER et al. 1991 , HERSHBERGER et al. 1995 ). As we wished to obtain the sequence of the MuDR-like element of Zapalote chico without selecting for mutations during cloning, segments of the MuDR-like element(s) of line N215 were cloned in five overlapping fragments. The cloning strategy is shown in Figure 1. Given that Southern analysis indicated only that ~3 copies of a MuDR-like element are present in N215 (Figure 2) and that the 10-enzyme-restriction survey demonstrated that the reading frames had the expected enzyme sites, we reasoned that an element sequence assembled from these pieces would represent a single, full-length element. It is possible, however, that the individual pieces sequenced are from different, but very closely related, elements.

The complete sequence for the deduced element assembled from the overlapping clones is shown in Figure 4, and the sequence comparisions to MuDR are summarized in Table 2A. The SstI/SstI internal fragments of MuDR and the cloned MuDR-like element from line N215 share 94.6% DNA sequence identity. In comparison to the known sequence of MuDR, it is possible to identify two putative coding regions in the MuDR-like element which correspond to the MudrA and MudrB genes of MuDR. The greatest divergence between the two elements is found in the intergenic region and in the sequence of the first intron of the MudrA gene. The intergenic region of the MuDR-like element contains a number of nucleotide insertions, including an additional copy of a direct repeat sequence; these insertions likely account for the slightly higher apparent molecular weight of MuDR-like SstI framents on genomic Southern blots (Figure 2). However, the putative coding regions and intron locations are highly conserved. As shown in Table 2B, there are only 10 nonsynonymous codon changes in MURB, and 20 nonsynonymous changes in the much larger MURA. Based on the high degree of conservation of the MURA and MURB proteins, we will designate the MuDR-like element of N215 Zapalote chico as MuDR-Zc.

View larger version (110K): In this window In a new window Download PPT slide   Figure 4. —Complete nucleotide sequence of the 4998-bp MuDR-zc element of accession N215. The TIRs are in italics. Bold bases are the ATG of mudrA (position 450), stop codon of mudrA (position 3126), start codon of mudrB (position 4531, in antisense orientation), and the stop codon of mudrB (position 3838; in antisense orientation).

Comparison of the TIR sequences of MuDR and N215 showed that TIRA of MuDR-Zc is 100% identical to TIRA of MuDR, whereas TIRB of MuDR-Zc is 91% identical to the TIRB of MuDR. The two TIRs of MuDR-Zc are only 88.1% identical to each other. Although the left and right TIRs of other Mu elements are rarely identical, the extent of divergence between the TIRs of MuDR-Zc is much higher than in Mu1–Mu8 (WALBOT 1991 ). In MuDR, there are only two base changes in the first 180 bp of the TIRs, and overall the 215-bp TIRs are 96% identical (HERSHBERGER et al. 1991 ). These nearly identical TIRs of MuDR contain the promoter regions and major transcription start sites for the two genes (BENITO and WALBOT 1994 ; HERSHBERGER et al. 1995 ).

Transcription of MuDR-Zc:
The biological significance of MuDR-Zc is best addressed by determining whether MuDR-Zc is an active element. Active and inactive Mutator lines can be distinguished by the presence or absence, respectively, of MuDR-hybridizing transcripts. The expression of the MuDR-Zc was examined by Northern blot hybridization. Figure 5 shows the analysis of a standard Mutator line and several Zapalote chico lines that yielded PCR (N201), or both PCR and Southern hybridization results (N215, N234, and N237), consistent with full-length MuDR-like elements. In poly(A)⁺ RNA samples, the active Mutator plants of standard lines have abundant transcripts for both mudrA (Figure 5A, lane 5) and mudrB (Figure 5B, lane 6). The MuDR transcripts are relatively abundant as they are readily observed in total RNA (data not shown; see HERSHBERGER et al. 1995 ). In the Zapalote chico samples, however, it was technically difficult to visualize mudrA transcripts using total RNA. With poly(A)⁺ RNA, very low levels of mudrA could be detected as a faint band of ~2.8 kb (Figure 5A, lanes 1–3). These transcripts are similar in size to those produced by the standard Mutator plants. The Zapalote chico mudrB transcripts were easily detected with poly(A)⁺ RNA (Figure 5B, lanes 1–4). Surprisingly, the mudrB probe identified two different sized transcripts, one slightly larger (1.05 kb) and one slightly smaller (0.95 kb) than the 1.0-kb transcript from the standard Mutator line.

View larger version (18K): In this window In a new window Download PPT slide   Figure 5. —Northern hybridization analysis of Mutator lines and Zapalote chico. Each sample contains 16–20 µg of poly(A)+ RNA. (A) The probe is PA corresponding to internal mudrA sequence. (B) The probe is PB corresponding to internal mudrB sequence.

Unexpectedly, we also observed novel-sized RNAs in the non-Mutator bz2 tester line that hybridized with the MuDR probe (Figure 5A, lane 4, and Figure 5B, lane 5). Similar size transcripts are also present at very low, comparable levels in the standard Mutator sample and in some of the Zapalote chico samples. The mudrA and mudrB gene probes may fortuitously recognize ubiquitous maize transcripts. In standard Mutator lines, transcripts as long as the entire element (4.9 kb) and truncated transcripts from internally deleted MuDR elements have been reported (HERSHBERGER et al. 1995 ). However, the cross-hybridizing material in the poly(A)+ sample from bz2 tester is the first report of any cross-hybridization with a non-Mutator line.

Non-autonomous Mu elements in Zapalote chico accessions:
One hallmark of active Mutator lines shared by both standard and low copy number lines is the presence of unmethylated Mu elements. For the Mu1 and related Mu2 elements, methylation status is conveniently assessed by Southern blot analysis after digestion with HinfI. There is a recognition site for this enzyme near the end of each TIR of the 1.4-kb Mu1 and 1.7-kb Mu2 elements. As shown in Figure 6, derivatives of accessions N201, N215, N234, and N237, four accessions with full-length MuDR-like elements, yield both the 1.3-and 1.6-kb expected fragments that hybridize to a probe that can detect both Mu1 and Mu2. Considering the Mu1 and Mu2 elements together, it appears that the Zapalote chico accessions examined contain ~3–10 copies of these non-autonomous elements in an unmethylated form. Although Mu1 elements typically predominate in standard Mutator individuals (TAYLOR and WALBOT 1987 ), the Mu2 elements are more abundant in the Zapalote chico accessions examined. In addition, the probe detects additional size classes that may represent one of the common deleted forms of these Mu elements (reviewed in WALBOT 1991 ), novel types of Mu1-derivatives or methylated copies of Mu1 or Mu2. Other known or novel Mu elements may also be present.

View larger version (50K): In this window In a new window Download PPT slide   Figure 6. —Southern blot analysis of non-autonomous Mu elements in four accessions with full-length MuDR-Zc elements. DNA was digested with HinfI and probed with pA/B5, which hybridizes to both Mu1 and Mu2.

Non-Mutator lines contain from zero to three Mu1 and Mu2 elements (BENNETZEN 1984 ; CHANDLER et al. 1986 ; CHANDLER and WALBOT 1986 ). These elements are completely stable in position and copy number, and they remain methylated in a non-Mutator line. On crossing with a standard, active Mutator line, the HinfI sites in the termini of the Mu1 element in inbred line B37 lost methylation and could be digested with methylation-sensitive enzymes, such as HinfI (CHANDLER et al. 1988 ). Thus, the moderate copy number and presence of unmethylated non-autonomous Mu elements suggest that these accessions of Zapalote chico are active Mutator lines.

Elevated forward mutation frequency in some Zapalote chico accessions:
The multiple copies of unmethylated Mu regulatory and non-autonomous elements in some accessions of Zapalote chico are similar to what is found in standard, active Mutator lines. On the other hand, the low abundance of MuDR-related transcripts is more similar to the single copy MuDR a1-mum2 lines (QIN and ELLINGBOE 1990 ) in which MuDR transcripts are only reliably detected with poly(A)+ RNA. The standard and single-copy MuDR lines both program the same pattern of high frequency somatic excision of Mu elements from reporter alleles, but the lines differ in the frequency of new mutants recovered (ROBERTSON and STINARD 1992 ). Given that the Zapalote chico lines share specific properties with each of the two characterized types of Mutator lines, we were interested in defining the forward mutation frequency.

We used the test devised by ROBERTSON 1978 . Each individual is self-pollinated to score pre-existing recessive mutants; each individual is also crossed to a non-Mutator line, and multiple F₁ plants are grown and self-pollinated to score for new mutants generated in the gametes of the presumptive Mutator parent. As Mu insertions occur late in development, new mutants are almost always recovered in only a single gamete (ROBERTSON 1981 ; reviewed in WALBOT 1991 ). One or a few individuals in most accessions of Zapalote chico were self-pollinated and crossed as pollen to bz2 tester. Mutants recovered in the parental selfed ear and the F₂ selfed ear were scored as visible seedling traits 10 and 28 days after germination in the summer field. For comparison, we also assessed forward mutation frequency with several standard Mutator lines, two single-copy MuDR lines with the a1-mum2 reporter allele, and the bz2 tester in the W23 inbred line.

As expected, the four standard Mutator lines generated many mutants (Table 3D). On selfing, nearly half (28/62) of the parental plants segregated 3:1 wild-type:mutant for a pre-existing, visible seedling mutation. Common recessive phenotypes included albinos, zebra-striped leaves, pale green, pale yellow, and developmental mutants with twisted, shredded or midrib-only leaves. In the outcross to bz2 tester, followed by selfing, 240 new mutants were observed in 835 families (29% mutation frequency) for the four standard Mutator lines. In the control for spontaneous mutation, no seedling mutants were observed in the selfed bz2 tester (Table 3F). We can estimate spontaneous mutation frequency in the bz2 tester if we also consider defective kernel (dek) mutations; one new dek mutation was recovered from the bz2 tester ( = 0.8%), a value similar to the spontaneous mutation frequency found in other non-Mutator lines examined in this test (ROBERTSON 1981 ). The dek phenotype is among the most common recessive class in standard Mutator lines, representing failure of the embryo, endosperm or both (reviewed in WALBOT 1991 ). There were 131 new dek mutations ( = 16%) in the standard Mutator sample, ~20-fold more than in bz2.

In contrast to standard Mutator lines, the single MuDR a1-mum2 line had a low forward mutation frequency (one mutant/344 families, Table 3E). Thus, the forward mutation frequency characteristic of standard Mutator may require multiple copies of MuDR and, most likely, a large population of non-autonomous elements.

The forward mutation test was completed before we classified the Zapalote chico accessions for MuDR-like elements and was therefore unbiased in selecting individuals for analysis. For simplicity, however, Table 3 groups lines on the basis of their MuDR phenotype. The four Zapalote chico accessions shown by PCR to contain all four segments of a MuDR-like element (Table 1) generated many new mutants after outcrossing as pollen parent to bz2; we observed 79 new mutations in 187 families (42%). The frequency of new mutations is equal to the most active standard Mutator line (M121), which gave 106 new mutants in 255 families (42%). In contrast, for the 11 Zapalote chico accessions in which PCR failed to detect all four segments of MuDR-like elements, the forward mutation frequency was low ( = 1.2%, Table 3B); this is within the range of the spontaneous mutation frequency in other non-Mutator lines analyzed by this test (ROBERTSON 1978 ), and similar to our results with bz2 and with the two single-copy MuDR lines. A third group of 10 Zapalote chico accessions was analyzed for mutation frequency but not tested by PCR (Table 3C). These lines yield an intermediate value, with 31 new mutants in 217 families (14.2%). This group is clearly heterogeneous, with some accessions generating multiple mutants and some none.

Because both plastids and mitochondria are maternally transmitted in maize, reciprocal crosses between disparate lines often lead to defective kernels as a result of nuclear-cytoplasmic incompatibility. We observed many dek mutants in the self-pollinated F₂ ears of the Zapalote chico outcrosses, particularly among accessions that also gave rise to seedling mutants (data not shown). This class was excluded from analysis, however, because it is unknown whether mutations or incompatibility are responsible for the small or defective kernel phenotype (ALLEN et al. 1989 ).

Unusual features of mutant induction in Zapalote chico:
It seems likely that both Mutator activity and MuDR-like elements are unevenly distributed in Zapalote chico populations. Only a subset of the Zapalote chico accessions qualify as Mutator lines by the forward mutation assay. This assay does not pinpoint what types of elements cause mutations. It is possible that some Zapalote chico lines contain several types of transposable elements.

One curious feature of the analysis is that Zapalote chico lines yielded few mutants on selfing. In the subset of lines with MuDR-like elements, we identified only two visible seedling mutations among the 19 parents (Table 3A). In subsequent years, continuous selfing of these lines, and tests with more individuals from the original accessions, have produced few new mutants (data not shown). The number of visible mutants was similar to what we found in the "no MuDR group" (seven visible mutations in 24 parents) and the unclassified group (four mutants in 16 parents). The Zapalote chico accessions contain more "mutants" than bz2, but one plausible explanation is that temperature-sensitive alleles were recognized as mutant at Stanford that have no mutant phenotype in the much warmer conditions of Oaxaca.

The low incidence of visible mutants in Zapalote chico lines containing MuDR (2/19) is particularly striking considering the incidence of such pre-existing mutants in standard Mutator lines (28/62). In contrast, the F₂ ears from the outcross part of the forward mutation test exhibit similar frequencies of newly induced mutants. In its native habitat, only selfing or crosses within Zapalote chico germplasm occur, because Zapotec farmers grow only this type of corn. The activation of a high forward mutation frequency on crossing with a heterologous line suggests that hybrid dysgenesis occurs. We completed too few exact reciprocal crosses between Zapalote chico and bz2 to determine whether the elevated mutation frequency results when an active Zapalote chico individual is the female parent as well as the pollen donor.

A second curious feature of the many new seedling mutants produced by the various Zapalote chico accessions is that none displayed somatic variegation. Typically, small wild-type sectors indicative of late somatic excision are visible in at least half of all new mutants produced by a standard Mutator line (ROBERTSON 1981 ; reviewed in WALBOT 1991 ). In the collection of mutants produced for this study, we also found that about half of the new albino, pale green and yellow mutants recovered from standard Mutator lines had visible dots of green on the first leaf (data not shown). The absence of somatic reversion is a novel property of new mutants produced in Zapalote chico.

Zapalote chico lines with MuDR-Zc can restore somatic mutability to cryptic bz2 mutable reporter alleles:
The lack of somatic instability of newly induced mutations in unknown genes made it difficult to analyze whether Mu elements were involved. To gain more direct evidence that MuDR-Zc elements were genetically active, we used a trans-activation test for Mutator activity. Lines derived from N237 and N264 with full-length MuDR-Zc (based on Southern blot hybridization) were crossed twice with bz2, in effect creating bz2 tester lines after selection for individuals without the dominant C-I allele. This allele prevents anthocyanin accumulation and was present in most Zapalote chico accessions. For the activation test, inactive Mutator lines homozygous for one of three well-characterized bronze2 alleles with precisely mapped Mu element insertions were selected from our collection; these lines contain multiple, methylated copies of MuDR and somatically stable Mu elements. As shown in Table 4, each inactive individual was self-pollinated and crossed to bz2 tester to score spontaneous reactivation of somatic mutability at the cryptic reporter allele; no instance of spontaneous reactivation was observed in the 55 individuals tested. On crossing to Zapalote chico or standard Mutator bz2 lines, fine purple spotting indicative of late, frequent somatic excision was restored in from zero to 85% of the test crosses. Such wide variation in reactivation is typical of Mutator reactivation tests (WALBOT 1986 ).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A high forward mutation frequency is a defining characteristic of standard Mutator lines; mutation frequency is elevated 20–100-fold above spontaneous or above what is observed in active Ac or Spm lines (reviewed in WALBOT 1992 ). Mutations in Mutator lines are caused by a diverse family of Mu elements, which share ~200-bp TIRs. Germinal insertion and somatic excision activities are controlled by the regulatory element MuDR. To date, MuDR has been found only in standard Mutator lines, in their immediate derivatives, and in the Cy germplasm (BENNETZEN 1996 ). In the standard U.S. germplasm, land races, and Zea spp. we have examined, we find evidence for unmethylated MuDR-like elements and Mutator activity in Zapalote chico. Even within this land race, only a subset of accessions appear to contain full-length elements.

For a neotropical maize, Zapalote chico is relatively tolerant of long daylength. It can be grown to maturity in the temperate zone and crossed with U.S. germplasm. Because it is so adaptable and contains many traits of potential agronomic importance, Zapalote chico has been used in breeding for disease, insect, and wind-damage resistance (MUNOZ et al. 1992 ). Zapalote chico contains large numbers of prominent heterochromatic knobs, and this line has been used in maize cytogenetic research (GOODMAN and BROWN 1988 ).

Several lines of evidence indicate that some accessions of Zapalote chico qualify as Mutator lines. First, they exhibit a high forward mutation frequency, similar to standard Mutator lines. Second, they contain multiple, unmethylated copies of non-autonomous Mu elements. Mu elements are methylated in inactive or non-Mutator lines (CHANDLER et al. 1988 ). Third, they contain multi-copy unmethylated and transcriptionally active MuDR-like elements, which to date have been found only in standard Mutator lines (BENNETZEN 1996 ). Fourth, MuDR-like element copy number is maintained through several outcrosses to non-Mutator lines. Approximately one-fourth of the Zapalote chico accessions examined appear to have Mutator activity by one or more of these criteria.

Molecular analysis of Mu elements in Zapalote chico accessions:
The three sequenced examples of MuDR are nearly identical, and it was expected that a search for additional Mutator sources would identify only this element. We have cloned the MuDR-Zc element in several fragments from one accession of Zapalote chico (N215) that contains several copies of the putative regulatory transposon. The MuDR-Zc sequence assembled from the fragments is highly similar, but clearly diverged, from the MuDR present in standard and the derived low-copy MuDR Mutator lines. MuDR-Zc is 4998 bp, 56 bp larger than the 4942-bp MuDR. Overall, MuDR and MuDR-Zc exhibit 94.6% DNA sequence identity. Identity is highest in TIRA and in the coding regions, with the intergenic region being the most divergent part of the element. At the amino acid level, the mudrA-like gene (mudrAzc) is more similar to that of MuDR, 97%, than the mudrB-like gene (mudrBzc), 95.2%. A portion of MuDR-Zc was also cloned and sequenced from N234; in the region 4398–4524, this sequence is identical to MuDR-Zc of line N215.

Southern blot analysis clearly demonstrates the presence of intact ~5.0-kb MuDR-Zc elements in N215. Because MuDR-Zc was cloned in fragments by PCR amplification, however, we do not have proof that all of the polymorphisms exist in the same element. It is also possible that a few of the nucleotide polymorphisms are from PCR mutation. Because MuDR is toxic to E. coli, point mutations are common during attempts to clone the intact element; for this reason we cloned MuDR-Zc in pieces that appear to be tolerated in E. coli. However, we were also able to amplify the fragment in two, large overlapping PCR fragments (position 113–2423 yielded a 2310-bp fragment; positions 2404–4829 yielded a 2425-bp fragment). Future recovery of overlapping genomic clones of MuDR-Zc and cDNA clones will confirm the distribution of sequence differences within individual MuDR-Zc elements in line N215. To gain a better understanding of the diversity of MuDR-like elements, full sequencing of elements from additional Zapalote chico accessions could be informative as well.

Evidence for Mutator activity in some Zapalote chico accessions:
Several approaches were taken to demonstrate that some Zapalote chico lines not only carry intact MuDR-like elements but may also have an actively transposing population of Mu elements. The first measure of Mutator activity was by Northern analysis, because it has been shown that only active Mutator lines express MuDR transcripts (HERSHBERGER et al. 1995 ). Second, we examined the methylation status and copy number of Mu elements and the transmission of MuDR-like elements to progeny. The third measure was a forward mutation test to determine if any Zapalote chico accessions had an elevated mutation frequency, and whether mutation frequency correlated with MuDR-like elements. Fourth, we examined the ability of Zapalote chico to activate somatic instability in inactive Mutator lines.

Northern analysis demonstrated that MuDR-Zc is actively transcribed; however, the levels and patterns of expression are different from standard Mutator lines (HERSHBERGER et al. 1995 ). mudrA and mudrB transcripts are easily detected in total RNA of standard Mutator lines and are approximately equally abundant, although in immature (prefertilization) ears there is an ~1:4 ratio of mudrA:mudrB transcripts (HERSHBERGER et al. 1995 ). Low transcript abundance is characteristic of the single-copy MuDR lines, but these lines have approximately equal amounts of transcript from genes A and B (QIN et al. 1991 ; JAMES et al. 1993 ). In Zapalote chico, however, mudrAzc transcript levels are extremely low, while those of mudrBzc are relatively more abundant. Both transcripts are only readily detected from poly(A)⁺ RNA. As the mudrAzc and mudrBzc transcripts are approximately the size of standard Mutator transcripts, we infer that the TIRs also act as the promoter elements in Zapalote chico, as well as constituting part of the 5' UTR of each transcript type. In the sequenced example of MuDR-Zc, TIRA is identical to TIRA of MuDR but TIRB is only 91% identical. The differences in TIRB may allow a higher level of mudrBzc transcription or increased transcript stability. It is possible that mudrA and mudrB differ in transcript abundance because there is Zapalote chico-specific host regulation or new forms of autoregulation by the MuDR-like elements.

It is not clear why there are two mudrBzc transcripts in the Zapalote chico accessions examined. It is possible that these transcripts are produced by two different, but related, MuDR-Zc elements. It is also possible that they are produced by alternative transcription start sites, differential splicing or different polyadenylation events from a single transcription unit. As mentioned earlier, both alternative splicing and multiple polyadenylation sites exist in mudrB transcripts in standard Mutator lines (HERSHBERGER et al. 1995 ), and such post-transcriptional events may explain the two transcripts found in Zapalote chico. We also observed novel-sized RNAs in the bz2 tester line that hybridized with the MuDR probe. These cross-hybridizing RNAs may result from fortuitous similarity or regions of similarity to MuDR in this non-Mutator line.

Methylation has previously been shown to be correlated with the loss of activity of MuDR (MARTIENSSEN and BARON 1994 ). In inactive Mutator lines with methylated Mu elements, Mu copy number decreases by approximately half with each successive outcross to a non-Mutator line (WALBOT and WARREN 1988 ). Neither MuDR-Zc elements nor Mu1 and Mu2 are methylated at the enzyme sites examined. Furthermore, the MuDR-Zc copy number is maintained on outcrossing: parents with just 1–3 elements transmit them to all progeny and through at least two outcrosses. Maintenance of Mu-element copy number is a key property of active Mutator lines (ALLEMAN and FREELING 1986 ; WALBOT and WARREN 1988 ), although we are uncertain whether transposition or demethylation of cryptic elements is responsible for copy number maintenance in Zapalote chico.

In the test for forward mutation frequency, we established that standard Mutator lines have a high forward mutation frequency (29% of families contain a new visible seedling mutation) compared to the low frequency of a standard inbred line of maize (bz2 tester in W23 background) or a single MuDR line (a1-mum2). In Zapalote chico, an elevated mutation frequency correlates with the presence of MuDR-like elements, but transposable elements of additional families and Mu elements of several types may contribute to the observed mutation frequency. The accessions for which there is molecular evidence of regulatory elements had a 42% forward mutation frequency, matching the level of the most active standard Mutator line. All of the new mutations recovered appear to be recessive, based on segregation data (data not shown).

Several properties of Mutator activity in Zapalote chico are distinct from both standard and single-copy MuDR Mutator lines. In contrast to standard Mutator lines, we found few mutations segregating in the original Zapalote chico parents. New mutants occur after outcrossing Zapalote chico as pollen donor onto a non-Mutator line. Consequently, new mutations occur as a result of hybrid dysgenesis and must be induced during or after fertilization. In standard Mutator lines, many new mutations are recovered as single-kernel events, indicative of Mu insertions that affect single gametophytes (reviewed in WALBOT 1991 ; CHANDLER and HARDEMAN 1992 ). When an active Mutator plant is used as a pollen donor, nonconcordant embryo and endosperm mutations occur in ~20% of the new mutants selected at Y1 (ROBERTSON and STINARD 1993 ). The lack of correspondence between the embryo and endosperm genotypes indicates that Mu insertions can occur after the mitosis that separates the two sperm in each pollen grain (ROBERTSON and STINARD 1993 ). Our data provide evidence that mutations in Zapalote chico sperm can be induced even later, after fertilization, provided the sperm interact with a non-Mutator egg.

A second unusual feature of new mutations induced in Zapalote chico is that they are not somatically mutable. Frequent late somatic excision is characteristic of both standard and low-copy Mutator lines (reviewed in WALBOT 1991 ; LISCH et al. 1995 ). If the Zapalote chico mutations are caused by Mu insertions, then the lack of somatic instability suggests that there is novel developmental regulation of element excision behavior. In Mutator lines losing activity, often assessed by a loss of somatic excision at a reporter allele, the levels of MuDR transcripts decline precipitously (JOANIN et al. 1996 ). The low abundance of MuDR-Zc transcripts may similarly be below the threshold required to program somatic excision.

A mutation screen is currently in progress to isolate mutations in anthocyanin reporter genes, using Zapalote chico accessions with MuDR-Zc elements. The isolation of a Mu element inserted into a known gene will provide the opportunity to analyze the type of insert and its excision behavior more precisely. It is