Alternative Transcription Initiation Sites and Polyadenylation Sites Are Recruited During Mu Suppression at the rf2a Locus of Maize

Corresponding author: Patrick S. Schnable, Iowa State University, Ames, IA 50011., schnable{at}iastate.edu (E-mail)

Communicating editor: J. A. BIRCHLER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Even in the absence of excisional loss of the associated Mu transposons, some Mu-induced mutant alleles of maize can lose their capacity to condition a mutant phenotype. Three of five Mu-derived rf2a alleles are susceptible to such Mu suppression. The suppressible rf2a-m9437 allele has a novel Mu transposon insertion (Mu10) in its 5' untranslated region (UTR). The suppressible rf2a-m9390 allele has a Mu1 insertion in its 5' UTR. During suppression, alternative transcription initiation sites flanking the Mu1 transposon yield functional transcripts. The suppressible rf2a-m8110 allele has an rcy/Mu7 insertion in its 3' UTR. Suppression of this allele occurs via a previously unreported mechanism; sequences in the terminal inverted repeats of rcy/Mu7 function as alternative polyadenylation sites such that the suppressed rf2a-m8110 allele yields functional rf2a transcripts. No significant differences were observed in the nucleotide compositions of these alternative polyadenylation sites as compared with 94 other polyadenylation sites from maize genes.

THERE are two broad categories of DNA transposons, autonomous and nonautonomous. Autonomous transposons encode all nonhost factors required for their own transposition. In contrast, the transposition of nonautonomous transposons is dependent upon factors encoded by autonomous transposons of the same family. Hence, only in the presence of factors encoded by the autonomous Spm/En, Ac, and MuDR maize transposons can the nonautonomous dSpm/I, Ds, and Mu transposons undergo excision and transposition (reviewed by MASSON et al. 1991 ; BENNETZEN et al. 1993 ; BENNETZEN 1996 ; KUNZE 1996 ; FEDOROFF 1999 ). However, both autonomous and nonautonomous transposons can cause mutations when they insert into genes. The Mutator (MuDR/Mu) transposon family (ROBERTSON 1978 ) has been widely used for gene mutagenesis and cloning (BENNETZEN et al. 1993 ; BENNETZEN 1996 ). Its 4.9-kb autonomous member, MuDR, mediates the transposition of the nonautonomous transposons, Mu1–Mu8 (SCHNABLE and PETERSON 1986 ; ROBERTSON and STINARD 1989 ; CHOMET et al. 1991 ; HERSHBERGER et al. 1991 ; QIN et al. 1991 ; HSIA and SCHNABLE 1996 ). MuDR contains two open reading frames (ORFs), mudrA and mudrB (HERSHBERGER et al. 1991 ; JAMES et al. 1993 ). The mudrA gene encodes a DNA-binding protein with a region of sequence similarity to bacterial transposases (EISEN et al. 1994 ; BENITO and WALBOT 1997 ). The mudrA gene is necessary and sufficient for somatic excision of Mu transposons (LISCH et al. 1999 ; RAIZADA and WALBOT 2000 ). The mudrB gene may be involved in suppression (DONLIN et al. 1995 ; LISCH et al. 1999 ).

Many transposon-induced mutants exhibit unstable phenotypes as a result of DNA rearrangements such as excision. In addition, some transposon-induced alleles also exhibit instabilities that occur in the absence of DNA rearrangements (MCCLINTOCK 1964 , MCCLINTOCK 1967 ; MARTIENSSEN et al. 1989 ). The mechanisms underlining these phenomena are not well understood.

The loss of a Mu-induced allele's capacity to condition a mutant phenotype is termed Mu suppression. Mu suppression was first reported at the hcf106::Mu1 mutant allele, which has a Mu1 insertion in its 5' untranslated region (UTR; MARTIENSSEN et al. 1989 ). In the presence of active MuDR transposons, this allele conditions a pale-green seedling. In the absence of MuDR, the hcf106:: Mu1 allele can condition a nonmutant phenotype that consists of dark-green "normal" leaf tissue (MARTIENSSEN et al. 1989 , MARTIENSSEN et al. 1990 ). In this suppressed state, alternative transcription initiation sites generate functional hcf106 transcripts that are not present in plants that carry active MuDR. These alternative transcription initiation sites are in the 5' terminal inverted repeat (TIR) of Mu1 and in regions of the 5' UTR of the hcf106 genes that are 3' of the Mu1 insertion site (BARKAN and MARTIENSSEN 1991 ).

Mu suppression has been observed in several other Mu-induced mutants. Two suppressible Knotted1 (Kn1) alleles arose via Mu1 and Mu8 insertions in the junction region of the Kn1-0 repeats. This junction region contains the promoter region of the downstream copy of Kn1-0 (LOWE et al. 1992 ). Two other suppressible Kn1 alleles, Kn1-mum2 and Kn1-mum7, have Mu8 and Mu1 insertions in the third intron of the Kn1 locus (GREENE et al. 1994 ). Mu suppression has also been observed with the Les22-7 mutant allele, which has a Mu1 insertion in its 5' UTR (HU et al. 1998 ), and with the a1-mum2 allele, which carries a Mu1 insertion 81 bp upstream of the transcription initiation site (CHOMET et al. 1991 ). Some Lg3 and Rs1 alleles, which have Mu insertions in 5' UTRs (Lg3-Or422, Lg3-Or102, and Lg3-Or331) or introns (Lg3-Or211 and Rs1-Or11) also exhibit suppression (GIRARD and FREELING 2000 ). Hence, Mu insertions in promoter regions, 5' UTRs, and introns can all generate suppressible alleles.

The hcf106::Mu1 mutation and the genetically unlinked Mu-induced suppressible mutation Les28 exhibit coordinated suppression and reactivation. The suppression of both mutant phenotypes is well correlated with hypermethylation of Mu transposons throughout the entire genome and also with hypermethylation of the region of the hcf106 locus flanking the Mu1 insertion (MARTIENSSEN and BARON 1994 ). As is true for hcf106:: Mu1 and Les28, Mu suppression of the four Kn1 alleles is correlated with genomewide hypermethylation of Mu transposons (LOWE et al. 1992 ; GREENE et al. 1994 ). At least two of these Kn1 alleles (Kn1-mum2 and Kn1-mum7) can be reactivated by crossing to Mu-active lines (GREENE et al. 1994 ). Because loss of Mu activity is correlated with hypermethylation of Mu transposons (reviewed by CHANDLER and HARDEMAN 1992 ; BENNETZEN et al. 1993 ) and the presence of MuDR (CHOMET et al. 1991 ), and because expression of the 823 amino acids of MURA in transgenic plants is sufficient to result in demethylation (RAIZADA and WALBOT 2000 ), it is likely that Mu suppression is caused by an absence of MURA.

Mu suppression has the potential to complicate efforts to clone genes via transposon tagging, because a critical step in such a project is to identify the particular Mu-containing restriction fragment length polymorphism or PCR fragment responsible for a Mu-induced mutation. This step is generally accomplished by identifying Mu-containing DNA fragments that cosegregate with the mutant phenotype through meioses. Because suppressed plants exhibit a wild-type phenotype even though they carry a Mu transposon in the target gene, suppression can mask the cosegregation between the mutant phenotype and the Mu-containing DNA fragment. Mu suppression can also lead to the loss of mutant phenotypes during backcrossing programs, which are becoming increasingly important with the adoption of new technologies such as RNA profiling and proteomics. These problems can be avoided if these analyses are conducted in an active Mu stock but this restriction limits the available genetic backgrounds.

Mu suppression also impacts technologies, such as Trait Utility System for Corn (BENSEN et al. 1995 ) and Mu rescue (WALBOT 1999 ), which are important components of maize functional genomics projects. Each of these projects utilizes Mu transposon sequence as bait to obtain Mu-insertion mutants through reverse genetics for functional analysis. This selection strategy does not exclude suppressible mutants, which can condition either mutant or normal phenotypes and thereby complicate the functional analysis of target genes. This is a particular concern because Mu transposons appear to exhibit a strong preference for insertion within the 5' UTRs of at least one gene (DIETRICH et al. 2002 ). On the other hand, Mu suppression during somatic development can produce clonal sectors on an unsuppressed background. Such chimeric plants provide a unique environment for analyzing mutants (FOWLER et al. 1996 ).

Therefore, understanding the frequency and mechanisms of Mu suppression is critical to using Mu transposons to understand maize biology. Here we report the identification and characterization of three Mu-suppressible rf2a alleles. We found that Mu suppression can occur not only at alleles caused by Mu insertions in a 5' UTR via the recruitment of alternative transcription initiation sites as reported previously, but also at an allele caused by a Mu insertion in the 3' UTR. Polyadenylation sites within the TIRs of the inserted Mu transposon are recruited during Mu suppression of this new class of Mu-suppressible alleles. In addition, these studies establish that insertions of two additional classes of Mu transposons can generate suppressible alleles.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Alleles of the rf2a gene:
The rf2a gene (previously designated rf2; SKIBBE et al. 2002 ) is one of the two complementary restorers of T-cytoplasm male sterility (reviewed by SCHNABLE and WISE 1998 ; WISE et al. 1999 ), which encodes a mitochondrial aldehyde dehydrogenase (CUI et al. 1996 ; LIU et al. 2001 ). Plants that carry T cytoplasm and that are homozygous for mutant alleles of either rf1 or rf2a are male sterile. All lines used in this study carry Rf1 and Rf2a alleles unless otherwise indicated. The reference allele, rf2a-R213, is a spontaneous mutant carried by the inbred lines Wf9 and R213. All other described rf2a mutant alleles were isolated via transposon tagging experiments (SCHNABLE and WISE 1994 ). Mutant alleles rf2a-m8110, rf2a-m8122, rf2a-m9323, rf2a-m9390, rf2a-m9385, and rf2a-m9437 were all obtained from a Mu-containing population. The wild-type progenitor of rf2a-m8110 and rf2a-m9390 is Rf2a-Q67. The progenitor of rf2a-m8122 and rf2a-m9323 is Rf2a-Q66. The progenitor of rf2a-m9437 is Rf2a-B79 (CUI et al. 1996 ). Each of the rf2a-m alleles used in these experiments was backcrossed to the inbred line Ky21 for at least three generations. The rf2a-m9385 allele was not included in this study because an adequate number of backcrosses had not been completed. The rf2a-m8904 allele was isolated from an Spm/En population. The inbred line B73 has the genotype rf1Rf2a.

Backcrosses of the rf2a-m alleles:
Plants heterozygous for each rf2a-m allele (rf2a-m/Rf2a) carrying T cytoplasm were crossed as females by the inbred line Ky21. Progeny that carried the rf2a-m alleles were identified by DNA gel blot or PCR analyses. The 1.2-kb partial or 2.2-kb full-length rf2a cDNA was used as a hybridization probe against EcoRV- (for rf2a-m9437) or HindIII- (for the rest of the rf2a-m alleles) digested genomic DNA in DNA gel blot analyses. Genotypes of plants in families segregating for rf2a-m alleles were often confirmed via testcrosses: (T) rf2a-R213/rf2a-R213 x rf2a-m/Rf2a-Ky21 (or Rf2a-Ky21/Rf2a-Ky21). Segregation of male-sterile and male-fertile plants (1:1) in the resulting progeny confirmed the presence of a nonsuppressed rf2a-m allele in a particular male parent.

Male-fertility ratings:
Phenotypes were scored in the morning during the period of pollen shedding for several days according to the rating system of SCHNABLE and WISE 1994 . Plants were scored as male fertile (F), full fertile, with >90% of the anthers exerted; semifertile ("F"), with <90% but more than half of the anthers exerted; semisterile ("S"), with <50% of the anthers exerted (usually only a few percent of anthers exerted); or male sterile (S), completely male sterile with no anthers exerted.

Mu-active lines:
Mu-active lines were derived either from the stocks used to generate the rf2a-m alleles (SCHNABLE and WISE 1994 ) or from very similar crosses that also involved Mu stocks and the inbred R213.

Rf2a genomic clones:
Two overlapping rf2a-hybridizing genomic clones were obtained by screening B73 libraries. Both libraries were constructed using the DASHII (Stratagene, La Jolla, CA) vector and were prepared by Pam Close and John Tossberg, respectively. Library screening conditions were as described by XU et al. 1997 . Phage inserts were subcloned into pBluescript SK or KS (Stratagene) vectors for further analysis or sequencing. Some of these fragments were sequenced by utilizing the TN1000 transposon system (Gold Biotechnology, St. Louis, adapted from STRATHMANN et al. 1991 ). Both DNA strands were completely sequenced unless otherwise indicated.

Clone rf2-DNA1 was obtained using a 900-bp probe (DD1, see Fig 1C of CUI et al. 1996 ) that includes the last two introns (9 and 10) and exons (10 and 11) of the rf2a gene. An alignment of the sequence of the entire 20,072-bp insert of rf2-DNA1 with the full-length rf2a cDNA (GenBank accession no. U43082) revealed that 353 bp from the 5' end of the cDNA clone were not included in clone rf2-DNA1. Hence, another rf2a genomic clone (rf2-DNA2-65) was isolated from the second B73 genomic library using the full-length rf2a cDNA as probe. Sequencing and restriction mapping experiments revealed that the two rf2a genomic clones differ in the region 5' of exon 2 but not in the region defined by exons 2 and 11. Furthermore, data from PCR amplification of B73 genomic DNA using various primer pair combinations and genomic mapping via DNA gel blot analyses indicate that the structure of rf2-DNA2-65, but not of rf2-DNA1, reflects the structure of the Rf2a-B73 allele. On the basis of these results we propose that clone rf2-DNA1 is chimeric and contains the region 3' of intron 1 derived from the rf2a gene, while the region 5' of intron 1 is of unknown origin. Hence, to generate the complete genomic sequence of the Rf2a-B73 allele, 5.2 kb of sequence from the 5' end of clone rf2-DNA2-65 was combined with 12.6 kb of sequence from the 3' end of clone rf2-DNA1 and deposited in GenBank (accession no. AF215823).

View larger version (10K): In this window In a new window Download PPT slide   Figure 1. The structure of the rf2a gene (GenBank accession no. AF215823). Solid and open boxes represent coding regions and 5' or 3' UTRs, respectively. Hatched box represents a copia-like retrotransposon, DON QUIXOTE. The positions of the transposon insertions responsible for the indicated alleles are indicated by triangles (not to scale). Solid triangles represent Mu insertions responsible for suppressible alleles. The arrows inside the triangles indicate the orientations of the transposon sequences in GenBank. The position of probe rf2-5m within exon 1 is shown as a bar underneath the 5' UTR. The positions of PCR primers are indicated by arrows. Bm, BamHI; Bg, BglII; Xb, XbaI.

Mapping transposon insertion sites and identifying Mu transposons:
The transposon insertion site in the rf2a-m8122 allele was established during the cloning of the rf2a gene (CUI et al. 1996 ). Genomic restriction mapping was conducted on rf2a-m8110, rf2a-m9323, rf2a-m9390, and rf2a-m9437 alleles using a variety of restriction enzymes. These DNA gel blots were hybridized with the rf2a-specific probes rf2-5m, C4-C6, B461-xq, and C1-C2 (see below). Using this method, it was possible to map the Mu insertions in three of the four alleles analyzed (rf2a-m8110, rf2a-m9390, and rf2a-m9437) to specific regions of the rf2a gene.

PCR reactions were also conducted on DNA from plants homozygous for each of the rf2a-m alleles to map and to identify Mu transposons. Each PCR reaction included a Mu-TIR primer and one of many primers specific to the exons of rf2a. Due to the large introns of this gene, most of these PCR reactions did not yield rf2a-specific products. After some of the transposon insertion sites were mapped via genomic restriction mapping, appropriate rf2a-specific primers were used (RF2C6 for rf2a-m9390 and rf2a-m9437 and RF2C1 for rf2a-m8110) to amplify these rf2a-m alleles (Fig 1). In these instances the transposon insertion sites were physically mapped via sequence comparisons between the resulting PCR products and the sequence of the rf2a gene (GenBank accession no. AF215823). The junction between the rf2a and Mu-TIR sequences in these PCR products defines the Mu insertion site. Because the TIRs of each class of Mu transposon contain diagnostic polymorphisms, it was possible to determine the identities of the Mu transposons inserted in rf2a-m alleles by comparing the Mu-TIR sequence contained in the allele-specific PCR products to the TIRs of all known Mu transposons. In addition, because the sequences of the two TIRs of most Mu transposons have one or more polymorphisms relative to each other, the orientations of the Mu insertions in rf2a-m alleles could also be determined.

To confirm the identity of the transposon inserted into the rf2a-m9390 allele, genomic DNA samples from plants homozygous for rf2a-m9390 and its progenitor Q67 were digested with EcoRV, HindIII, XbaI, EcoRV + HindIII, EcoRV + XbaI, and HindIII + XbaI and then hybridized with the rf2a-specific probe, rf2-5m. The same filter was stripped (AUSUBEL et al. 1999 ) and hybridized with a Mu1-specific probe. In each restriction digest the rf2a-hybridizing fragments were the same size as the Mu1-hybridizing fragments, which provided further support for the conclusion that the transposon inserted into rf2a-m9390 is Mu1.

The identity of the rcy/Mu7 transposon in rf2a-m8110 was confirmed similarly with the rf2a-specific probe C1-C2 and a rcy/Mu7-specific probe. To further confirm the identity and orientation of the rcy/Mu7 transposon insertion in rf2a-m8110, an rf2a-specific primer, RF2C2, which is 0.1 kb 3' of the rcy/Mu7 insertion site in this allele, and an rcy/Mu7 internal primer, Mu7-R, were used to amplify a 1.7-kb fragment from the 5' end (i.e., the rightmost end in Fig 1) of the rcy/Mu7 transposon in rf2a-m8110. This PCR product was subcloned into a pGEM-T vector (Promega, Madison, WI) and >80% was sequenced. Sequence comparisons between this PCR product and rcy/Mu7 (GenBank accession no. X15872) identified only a few nucleotide polymorphisms.

The transposon insertion in the rf2a-m8904 allele was identified by PCR amplification using primers rf2a-3320 and RF2C5UTRR, which flank the transposon insertion. Based on its sequence, this transposon is Ds1 (GenBank accession no. AF010445).

Probes:
DNA fragments used as probes in this study were obtained as follows. The 2.2-kb full-length rf2a cDNA fragment was obtained from plasmid prf273-11 with the restriction enzymes XhoI and EcoRI. The 1.2-kb partial rf2a cDNA was obtained from plasmid prf2a-1.2 with the restriction enzyme EcoRI. Plasmids prf273-11 and prf2a-1.2 contain the 1.2-kb and 2.2-kb rf2a cDNAs as described in CUI et al. 1996 . The rf2a 5' 0.15-kb probe, rf2-5m, was PCR amplified from plasmid prf273-11 using primers RF2C6 and RF2C8 (Fig 1). Another rf2a 5' probe, C4-C6, was PCR amplified from plasmid prf273-11 using primers RF2C4 and RF2C6. Probes B461-xq (from 461 to 927 bp in GenBank accession no. U43082) and C1-C2 (from 1374 to 2029 bp in GenBank accession no. U43082) were PCR amplified from plasmid prf273-11 using primers rf2-B461 and rf2-xq or primers RF2C1 and RF2C2 (Fig 1). A 0.96-kb Mu1-specific fragment was isolated from plasmid pRB1 (ROBERTSON et al. 1988 ) using the restriction enzyme MluI. The 0.16-kb rcy/Mu7 probe was isolated from plasmid pSB9-in (SCHNABLE et al. 1989 ) using restriction enzymes BamHI and EcoRI. Maize GAPDH cDNA (GenBank accession no. X07156) was isolated from pGAPDH with the restriction enzymes EcoRI and HindIII. All probes were labeled with dCTP³² using a random primer labeling protocol (AUSUBEL et al. 1999 ).

Genotyping rf2a-m alleles via PCR:
To determine which plants in a segregating family carry an rf2a-m allele, PCR amplifications were conducted using a Mu-TIR primer in combination with an appropriate rf2a-specific primer. Various primers that anneal to Mu-TIRs, such as XX153, were used in these PCR reactions. The rf2a-specific primers used for genotyping rf2a-m alleles include RF2C1 (for rf2a-m8110 and rf2a-m8122) and RF2C6 (for rf2a-m9390 and rf2a-m9437; Fig 1). PCR reactions were performed for 34 cycles as follows: denature at 94° for 40 sec; anneal primers for 40 sec at 55°–58° (depends upon the Mu-TIR primer used); extend at 72° for 2 min in the presence of 2.5 units of Taq polymerase (Promega) per reaction.

Primer sequences:

• Mu7-R: 5' TTCTCCGCCGTTGCCATCTC 3^'

• RF2C1: 5' GCGTCGTTGGTGATCCGTTC 3'

• RF2C2: 5' CCAGGCTAGGGCAAATCTTAT 3'

• RF2C4: 5' AGCGGGAGACGAGCGAGGAC 3'

• RF2C5: 5' ATGCTGCGATTCCGTTTGGTG 3'

• RF2C6: 5' TCCTCACTCCCACACCAACC 3'

• RF2C8: 5' GCAGCAGGAGAAGCGGCAGGCAG 3'

• RF2C9: 5' GTGATGGGCTCCTCTACT 3'

• XX153: 5' CGCCTCCATTTCGTCGAATCC 3'

• rf2-B461: 5' ACAGATCTAAAGCTCCTCATTAAT 3'

• rf2-xq: 5' CCAACTTTCCAGGCATACATCA 3'

• rf2a-3320: 5' GAGGAACCAGTAGCGGAGGC 3'

• RF2C5UTRR: 5' GCTCCCGTTCGCAGTCG 3'

DNA and RNA gel blot analyses:
Maize genomic DNA was isolated using a 1x CTAB procedure (SAGHAI-MAROOF et al. 1984 ). About 10 µg of DNA was digested with the indicated restricted enzyme in a 30-µl reaction volume for >3 hr and then separated via electrophoresis through a 0.8% agarose gel. DNA was transferred to nylon filters (MSI, Westboro, MA) and hybridized with probes labeled with dCTP³² (AUSUBEL et al. 1999 ).

Total RNA from immature tassels (still in the whorl) was isolated according to DEAN et al. 1985 . Approximately 10 µg of total RNA was subjected to electrophoresis using a MOPS buffer system (AUSUBEL et al. 1999 ) and transferred onto Gene-Screen filters (NEN Research Products, Boston). The filters were hybridized with probes labeled with dCTP³² (AUSUBEL et al. 1999 ). The hybridization density of each band was quantified using a GS-710 densitometer (Bio-Rad Laboratories, Hercules, CA).

3' and 5' rapid amplification of cDNA ends:
RNA was isolated from immature tassels (still in the whorl) using the Trizol reagent (GIBCO BRL, Rockville, MD) and treated with PCR-grade DNaseI (GIBCO BRL) according to the manufacturer's instructions. A total of 5 µg and 100 ng of RNA were used for 3' and 5' rapid amplification of cDNA ends (RACE) experiments, respectively, using kits obtained from GIBCO BRL. The rf2a-specific primers RF2C5 and RF2C9 (Fig 1) were used for 3' and 5' RACE, respectively. The position of RF2C9 in exon 2 allows the inadvertent amplification of genomic DNA to be detected. Primer RF2C 4 was used as a nested primer for the 5' RACE experiments. RACE products were subcloned into the pGEM-T vector (Promega) and sequenced.

Analysis of polyadenylation sites:
All GenBank records for which the organism was Zea mays (excluding chloroplast and mitochondrial genes) that had the feature of "polyA_site" were downloaded on March 22, 2000. This data set was then parsed for the feature of polyadenylation site. The few records that lacked sequence data downstream of the reported poly(A) site or that had only "A" bases downstream of this site were excluded. Only the most recent GenBank submission was used in those instances of multiple submissions of the same gene. Chi-square homogeneity tests were performed according to STEEL and TORRIE 1980 .

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Structure of the rf2a gene:
The rf2a cDNA encodes a mitochondrially localized aldehyde dehydrogenase (CUI et al. 1996 ; LIU et al. 2001 ). Two overlapping genomic clones (rf2-DNA1 and rf2-DNA2-65) were obtained by screening B73 genomic libraries (MATERIALS AND METHODS). Sequence comparisons of 17.8 kb derived from these genomic clones (GenBank accession no. AF215823) and the full-length rf2a cDNA clone (GenBank accession no. U43082) defined 11 exons and 10 introns (Fig 1). The extreme 5' end of the genomic sequence (base positions 102–740) is 81.5 and 80.7% identical to the 3' long terminal repeat (LTR; GenBank accession no. AF050449) and the 5' LTR (GenBank accession no. U68407) of the maize Milt retrotransposon. In addition, a member of a newly defined class of retrotransposons, DON QUIXOTE, which is present in intron 5 (base positions 6763–13,927 of GenBank accession no. AF215823), contains a characteristic nucleic acid binding site that serves as the primer site for reverse transcription and an uninterrupted ORF of 4 kb (positions 8335–12,129) that encodes predicted proteins with high degrees of sequence similarity to the reverse transcriptase, protease, integrase, and endonuclease typical of copia-like retrotransposons (KONIECZNY et al. 1991 ). A 0.6-kb region that is 140 bp 5' of exon 2 in rf2-DNA2-65 exhibits 91% sequence identity to the LTR of Grande1-4 retrotransposon (GenBank accession no. X97604). Two other regions of the rf2a gene (base positions 2065–3096, 5' of the transcription initiation site and positions 4875–5270 in intron 1) exhibit 88 and 85% sequence similarity, respectively, to the noncoding regions of various maize genes (GenBank accession nos. X06333, L40803, AF030385, AF031569, AF041044 and AJ005343, Z22760, AF043346, Z26824, AJ223471). These regions of sequence similarity have not been assigned any putative functions and do not exhibit characteristics of DNA transposons or retrotransposons.

Positions of transposons in rf2a-m alleles:
The positions of transposon insertions responsible for four of the five Mu-derived rf2a-m alleles (rf2a-m8110, rf2a-m8122, rf2a-m9390, and rf2a-m9437) were mapped via genomic restriction mapping and sequence comparisons between PCR products obtained using rf2a-m DNA templates in conjunction with Mu-TIR- and rf2a-specific primer pairs and the sequence of the rf2a gene (Fig 1). The specific class of Mu transposon responsible for each mutant and its orientation were determined via sequence analysis of the Mu-TIRs obtained from these PCR products. The rf2a-m9390 allele has a Mu1 transposon inserted into its 5' UTR, 105 bp 5' of the translation start codon. The identity of this Mu1 transposon was confirmed by DNA gel blot analysis (MATERIALS AND METHODS). The rf2a-m9437 allele contains a novel Mu transposon insertion in its 5' UTR, 35 bp 5' of the translation start codon. The TIR of this transposon differs from all other described Mu transposons. Therefore, this Mu transposon has been designated Mu10 (GenBank accession no. AF231940). The rf2a-m8110 allele arose via an rcy/Mu7 insertion in the 3' UTR, 30 bp 3' of the translation stop codon. The identity of this rcy/Mu7 transposon was confirmed by sequence analysis of a 1.7-kb fragment of it and DNA gel blot analyses. The rf2a-m8122 allele has a Mu1 insertion in exon 9 (CUI et al. 1996 ). It has not yet been possible to identify the molecular lesions associated with the rf2a-m9323 and rf2a-m9385 alleles. The rf2a-m8904 allele contains a Ds1 transposon a few base pairs downstream of the translation start codon. This 395-bp Ds1 has three single nucleotide polymorphisms relative to the Ds1 sequence in GenBank (accession no. AF010445).

Reanalysis of data from Schnable and Wise 1994 :
Although one-half of the progeny from the testcross, (T) rf2a-m8110/Rf2a x rf2a-R213/rf2a-R213, would be expected to be male sterile, Schnable and Wise found only 4 of 26 progeny from this cross to be male sterile. A high rate of nonconcordance between the male sterile phenotype and a nearby marker (wx1) was also obtained from a similar testcross but involving rf2a-m9390 allele (SCHNABLE and WISE 1994 ). After the rf2a gene was cloned and the positions of the Mu insertions responsible for these alleles were determined, genomic DNA samples from these testcross families were genotyped via PCR (MATERIALS AND METHODS). The results of these analyses are presented in Table 1. Families segregating for rf2a-m8110 (92 2123) and rf2a-m9390 (92 2148) contain three and six male-fertile plants, respectively, with the genotype of rf2a-m/rf2a-R213. Hence, in these families, these two alleles display only 70 and 40% penetrance, respectively. In contrast, a similar family but segregating for the rf2a-m8122 allele, which has a Mu1 insertion in the coding region, exhibited 100% penetrance (family 92 2126 in Table 1). The rf2a-m9323 allele was not included in these genotyping experiments, but it has not exhibited low penetrance in the backcrossing program. Another allele, rf2a-m9437 (family 92 2153), with a Mu10 insertion in the 5' UTR, also exhibited 100% penetrance in this generation. However, after three generations of backcrossing to the inbred line Ky21, the rf2a-m9437 allele also began to show evidence of low penetrance (data not shown). These genotyping results establish that the low penetrance observed by SCHNABLE and WISE 1994 does not involve excisional loss of the Mu transposons. Because the rf2a-m8122 allele (which has a Mu1 insertion in the coding region) did not exhibit low penetrance in otherwise identical crosses (e.g., family 92 2126, Table 1), it is not likely that the low penetrance of these alleles is conditioned by genetic suppressors (PRELICH 1999 ) of rf2a carried by the inbred line Ky21.

Reactivation of rf2a-m alleles:
Because these three alleles (rf2a-m8110, rf2a-m9390, and rf2a-m9437) all have a transposon insertion in an UTR, this low penetrance could reflect Mu suppression, which occurs when a genome lacks Mu activity (MARTIENSSEN et al. 1989 ). One of the hallmarks of Mu suppression is the possibility of reactivation, whereby suppressed Mu-induced alleles are reactivated by the reintroduction of Mu activity via genetic crosses. To determine whether rf2a-m8110 and/or rf2a-m9390 can be reactivated (and thereby again condition mutant phenotypes), stocks carrying suppressed versions of these alleles were crossed to Mu-active lines. To minimize the confounding effects of genetic backgrounds, the Mu-active lines used in these crosses were closely related to the lines used for generating the rf2a-m alleles (SCHNABLE and WISE 1994 ). As a control, some male-fertile plants that carried suppressed versions of these two alleles were self-pollinated. No male-sterile plants appeared in the resulting progenies (96g 6104-05 and 98 6336 in Table 2). In contrast, various rates of reactivation (as determined by the frequency of male-sterile progeny) were obtained in the crosses with Mu stocks (Table 2). Hence these rf2a-m alleles can be reactivated via crosses to active Mu lines. This result is consistent with the hypothesis that the low penetrance of these alleles is due to Mu suppression.

The stability of the reactivation of rf2a-m8110 was tested by crossing a reactivated (i.e., male sterile) heterozygous plant, (T) rf2a-m8110/rf2a-R213, by (N) rf2a-m8904/rf2a-m8904. The rf2a-m8904 allele was used for this cross instead of rf2a-R213 because unlike rf2a-R213 it does not accumulate a detectable amount of rf2a mRNA (CUI et al. 1996 ). About 100 of the resulting progeny were scored for male fertility (data not shown). Most were male sterile. Of 16 randomly selected male-sterile plants, 7 carried the rf2a-m8110 allele. Hence, rf2a-m8110 can be transmitted through meiosis in the reactivated state.

Because the rf2a-m9390 allele became suppressed early in the backcrossing program, male-sterile plants homozygous for rf2a-m9390 were not initially available for mRNA accumulation analyses. To generate unsuppressed stocks homozygous for this allele, some unsuppressed heterozygous plants, (T) rf2a-m9390/rf2a-R213 from an early generation, were crossed by suppressed (i.e., male fertile) plants with the genotype (T) rf2a-m9390/rf2a-m9390. About 300 progeny were screened for male-sterile (i.e., unsuppressed) plants. Only 19 were obtained. When these 19 plants were genotyped, it was found that they all had the genotype rf2a-m9390/rf2a-R213; i.e., no male-sterile rf2a-m9390 homozygous plants were obtained. The reason for this unbalanced result is not known.

Methylation and Mu suppression of rf2a-m9390:
There is a good correlation between the methylation of Mu-TIRs and Mu activity (reviewed by BENNETZEN 1996 ). In previously reported studies of Mu suppression, hypermethylation of Mu transposons throughout the genome was generally correlated with Mu suppression, although exceptions were found (MARTIENSSEN et al. 1990 ; LOWE et al. 1992 ). To further test whether the low penetrance of rf2a-m9390 is due to Mu suppression, the methylation status of Mu1 elements in the genome was determined via DNA gel blot hybridization following HinfI digestion (Fig 2C; CHANDLER and WALBOT 1986 ). Mu1 elements with hypomethylated HinfI sites (which are therefore susceptible to restriction digestion) in their TIRs will yield Mu1-hybridizing fragments of 1.34 kb. In contrast, Mu1 elements with methylated HinfI sites will yield larger hybridizing fragments (Fig 2C). Although the Mu1 probe can also cross-hybridize with Mu2 transposons, there were few Mu2 transposons in the families analyzed in this study. The results obtained using genomic DNA isolated from young leaves of plants carrying rf2a-m9390 are summarized in Table 3. All male-sterile (i.e., not suppressed) and semisterile ("S," partially suppressed) plants in families segregating for recently reactivated rf2a-m9390 (families 96 2209, 96 2214, and 96 2218) from the reactivation experiment contained mostly hypomethylated Mu1 elements (Table 3). Of the 13 analyzed fertile (i.e., suppressed) plants, 11 contained completely or mostly methylated Mu1 elements; the two exceptions contained both methylated and hypomethylated elements. Examples of these analyses are shown in Fig 2A. Generally, fertile (i.e., suppressed) plants exhibited a higher degree of methylation than did sterile (i.e., not suppressed) plants. A chi-square homogeneity test showed that the ratios of 0 methylated:4 hypomethylated male-sterile plants vs. the 11 methylated:2 hypomethylated fertile plants are significantly different (² = 9.59). This result is consistent with the hypothesis that the low penetrance of rf2a-m9390 is due to Mu suppression. Interestingly, plants from a family (family 92 2148) that had just begun to show suppression after outcrossing to inbred lines exhibited a higher level of Mu methylation in general (Table 3) than did plants from the reactivated families.

View larger version (56K): In this window In a new window Download PPT slide   Figure 2. Methylation status in suppressed and nonsuppressed plants with the genotype rf2a-m9390/rf2a-R213. DNA isolated from young leaves of plants from families 96 2209 and 96 2214 was digested with HinfI. (A) DNA gel blot was hybridized with the Mu1-specific probe illustrated in C. (B) The same blot was stripped and hybridized with the rf2a-specific probe rf2-5m (C). (C) The Mu1 insertion and HinfI sites in the rf2a-m9390 allele. All plants carry T cytoplasm and their phenotypes are indicated as F, fully male fertile, and S, completely male sterile with no anthers exerted. H, HinfI; distances between restriction sites are indicated in kilobases.

MARTIENSSEN et al. 1990 reported that suppression is also correlated with hypermethylation of the Mu1 transposon inserted in the hcf106::Mu1 allele. To detect the methylation status of the Mu1 insertion in rf2a-m9390, an rf2a probe (rf2-5m), was hybridized to HinfI-digested genomic DNA from the segregating progenies (the results are shown in Table 3). This probe hybridizes to a 1.1-kb fragment bounded by the HinfI site 5' of the probe in the rf2a promoter region and by the HinfI site in the 3' TIR (i.e., leftmost end in Fig 2C) of the Mu1 transposon. The methylation status of the Mu1 inserted in rf2a-m9390 was scored by determining the signal strength of the 1.1-kb rf2a-hybridizing fragment relative to the 2.4- and 2.8-kb rf2a-hybridizing fragments. If the 3' HinfI site in the Mu1 transposon is susceptible to HinfI digestion (i.e., if it is hypomethylated), a 1.1-kb rf2a-hybridizing fragment will be detected (e.g., the lanes of "S" in Fig 2B). If this site is not susceptible to HinfI digestion (i.e., if it is methylated), either the 2.4-kb (the 5' HinfI site in the Mu1 transposon is digested) or the 2.8-kb (neither of the two HinfI sites in the Mu1 transposon is digested) rf2a-hybridizing fragment will be detected (e.g., the "F" category in Fig 2B). A strong correlation was observed between the methylation status of the Mu1 transposon in rf2a-m9390 and the methylation status of other Mu1 transposons in the genome (compare A and B in Fig 2).

Accumulation of rf2a mRNA in plants homozygous for suppressed rf2a-m alleles:
Prior RNA gel blot analysis identified a plant homozygous for rf2a-m9390 that accumulated rf2a mRNA at the same level as wild-type plants (CUI et al. 1996 ). Because the entire young tassel of this plant had been collected for RNA in that study, the male fertility status of this plant could not be determined. To overcome this problem in subsequent experiments, only a few branches of each young tassel were collected from plants homozygous for the rf2a-m9390 or rf2a-m8110 alleles. An RNA gel blot analysis of these individuals is shown in Fig 3. Hybridization with a full-length rf2a cDNA detected an mRNA that accumulated to the same or only slightly lower levels in fertile plants (i.e., suppressed) that were homozygous for either rf2a-m8110 or rf2a-m9390 as accumulated in wild-type controls. In addition, at the level of resolution provided by RNA gel blot analyses, these mRNAs were indistinguishable in size from those that accumulated in the wild-type controls.

View larger version (59K): In this window In a new window Download PPT slide   Figure 3. The accumulation of rf2a mRNA in many suppressed plants is as high as or only slightly lower than that in wild type. An RNA gel blot containing RNA isolated from young tassels of the indicated genotypes was hybridized with a full-length rf2a cDNA probe (A) and the loading control maize GAPDH (B). The density of each band was quantified using a GS-710 densitometer (Bio-Rad Laboratories). The density ratio of each band in A to its corresponding band in B is listed below B. All plants were homozygous for the indicated rf2a alleles unless otherwise noted. The inbred line Q67 is homozygous for the wild-type progenitor (Rf2a-Q67) of rf2a-m8110 and rf2a-m9390. (T), T cytoplasm; (N), N cytoplasm. The male fertility phenotypes were denoted unless unavailable. F, fertile; "F," semifertile.

Native transcription initiation sites are not used in suppressed rf2a-m9390 plants:
Because the rf2a-m9390 allele contains a Mu1 insertion in its 5' UTR, the rf2a transcripts that accumulate in the suppressed rf2a-m9390 plants (Fig 3) could be derived from alternative transcription initiation sites downstream of the native sites (i.e., in Mu1 and/or in its flanking region) as has been reported for hcf106::Mu1 (BARKAN and MARTIENSSEN 1991 ). Alternatively, these transcripts could arise via readthrough transcription of the Mu1 transposon followed by the splicing of the Mu1 from the transcripts. To distinguish between these two possibilities, a probe (rf2-5m) upstream of the Mu1 transposon (Fig 1) was hybridized to RNA from suppressed plants homozygous for rf2a-m9390. The results of this experiment are shown in Fig 4. Because rf2a-m9390 transcripts could not be detected with the rf2-5m probe (Fig 4B), it can be concluded that transcription initiation in suppressed rf2a-m9390 does not occur from the native transcription initiation sites of rf2a. Instead, transcription of this allele (Fig 4A) must initiate downstream of the position of probe rf2-5m (Fig 1), likely from the 5' TIR (i.e., the rightmost TIR in Fig 1) of Mu1 and/or of an rf2a flanking region.

View larger version (31K): In this window In a new window Download PPT slide   Figure 4. RNA gel blot analysis reveals that the suppressed rf2a-m9390 allele does not utilize native transcription initiation sites. About 10 µg of RNA from young tassels was loaded in each lane. Full-length rf2a cDNA (A) and rf2-5m, a 5' fragment (see Fig 1) of the rf2a gene (B), were used as hybridization probes. All plants were homozygous for the indicated alleles.

Transcription initiation sites in rf2a-m9390:
Initially 5' RACE experiments were conducted on RNA extracted from the inbred line Q67, which is homozygous for the wild-type progenitor of rf2a-m9390 (CUI et al. 1996 ). The primer (RF2C9) used for the first-strand synthesis was located at position +182 relative to the start of translation and the nested primer (RF2C4) was located at position +37. DNA gel blot analysis revealed that the resulting RACE products were 290 bp, a result that is consistent with the size of the longest rf2a cDNA clone isolated to date (prf273-11), which begins at position -253. Cloned 5' RACE products obtained using RNA extracted from a plant homozygous for suppressed rf2a-m9390 were considerably smaller than those obtained from the inbred line Q67. Sequence analysis of nine of the RACE products from rf2a-m9390 revealed multiple transcription initiation sites, all of which were located between the position of the Mu1 insertion (-105) and the translation start codon (Fig 5). This result is similar to what has been observed during suppression of hcf106::Mu1 (BARKAN and MARTIENSSEN 1991 ).

View larger version (32K): In this window In a new window Download PPT slide   Figure 5. Alternative transcription initiation sites in the suppressed rf2a-m9390 allele. The open and solid boxes indicate the 5' UTR and coding regions of exons 1 and 2, respectively. Horizontal arrows indicate the primers used for the 5' RACE experiments. Sequence is provided for the rf2a 5' UTR and the beginning of the coding region. The translation start codon is boxed. The triangles indicate the position of the Mu1 insertion in rf2a-m9390. The vertical arrows indicate the transcription initiation sites revealed by 5' RACE using tassel RNA from a male-fertile plant homozygous for the rf2a-m9390 allele. The number of clones (if there are multiple clones) isolated with each initiation site is indicated above the arrows. Alternative transcription initiation sites are designated by roman numerals.

Alternative polyadenylation sites in the TIR of rcy/Mu7 are used during suppression of rf2a-m8110:
The rf2a-m8110 allele contains an rcy/Mu7 insertion in its 3' UTR. As is true for rf2a-m9390, suppressed plants homozygous for rf2a-m8110 accumulate rf2a mRNA that is indistinguishable from wild type by RNA gel blot analyses (Fig 3). On the basis of the analysis of cDNA clones, two Rf2a alleles (Rf2a-B73 and Rf2a-W22) use polyadenylation sites >140 bp 3' of the site of the rcy/Mu7 insertion in the rf2a-m8110 allele (Fig 6). In addition, 3' RACE experiments have established that this is also true for the Rf2a-Q67 allele, which is the wild-type progenitor of rf2a-m8110 (Fig 6). These results indicate that the rf2a-m8110 allele either uses the native polyadenylation sites but with the rcy/Mu7 transposon subsequently spliced out or uses alternative polyadenylation sites. 3' RACE was used to identify the polyadenylation sites used by suppressed rf2a-m8110 and to thereby distinguish between these two possibilities. As shown in Fig 6, five alternative polyadenylation sites were detected in two independent 3' RACE experiments. Some sites were recovered in both experiments. All identified sites are within the 3' TIR (i.e., leftmost TIR in Fig 6) of the rcy/Mu7 transposon insertion in rf2a-m8110.

View larger version (39K): In this window In a new window Download PPT slide   Figure 6. Polyadenylation sites used in suppressed rf2a-m8110 plants and in wild-type Rf2a plants. The solid and open boxes indicate the translated and 3' UTR regions of the last exon of the rf2a gene, respectively. The triangle represents the rcy/Mu7 transposon insertion responsible for the rf2a-m8110 mutation. The two hatched boxes containing horizontal arrows represent the TIRs of the rcy/Mu7 transposon. The vertical arrows above the TIRs indicate the polyadenylation sites used in the suppressed plants as revealed by 3' RACE experiments. Each arrow represents one event unless otherwise indicated. The solid and dashed arrows indicate the events revealed in the first and the second 3' RACE experiments, respectively. The primer used in the 3' RACE experiment is indicated as a horizontal arrow labeled as RF2C5. The rcy/Mu7 TIR sequence in which the polyadenylation sites were revealed is shown above the transposon. The 3' UTR sequence of Rf2a-B73 is shown below. The arrows below the gene structure represent the polyadenylation sites used by the Rf2a-B73 (solid arrows), Rf2a-W22 (shaded arrowhead), and Rf2a-Q67 (dashed arrows) alleles as determined by sequence analysis of cDNA clones (B73 and W22) or 3' RACE (Q67). Polyadenylation sites are designated by roman numerals. The translation stop codon, TAG, is boxed. The AAUAAA-like PE elements are shaded. U-rich regions flanking polyadenylation sites are underlined.

Polyadenylation sites are preferentially located at the 3' end of a YA (i.e., UA, CA, or GA) sequence in at least some eukaryotes (reviewed by ROTHNIE 1996 ; WAHLE and RUEGSEGGER 1999 ). To determine whether this pattern holds for maize, the polyadenylation sites associated with Z. mays records in GenBank were examined. Fifty-eight records with a total of 94 polyadenylation sites were obtained. Of these 94 sites, 73 fit the YA rule (Table 4). Similarly, 6 of the 7 polyadenylation sites within the 3' UTR of the rf2a gene are YA. In contrast, only 2 of the 5 alternative polyadenylation sites within rcy/Mu7 (GA, GA, UC, AC, and CC) fit this pattern. A ² homogeneity test (² = 3.67) indicated that the rate of YA polyadenylation sites in rcy/Mu7 at the 95% confidence level is not significantly different from that in other maize genes.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Mu suppression occurs at a high frequency:
Mu transposons are widely used for the functional analysis of the maize genome, including gene tagging, gene discovery, and reverse genetics. It has been known for over a decade that in the absence of Mu activity, some Mu-insertion alleles lose the capacity to condition a mutant phenotype (MARTIENSSEN et al. 1989 ). This phenomenon, Mu suppression, has important implications for the use of Mu transposons in understanding maize biology.

Of the thousands of mutant alleles derived from Mu stocks, only 12 have been described as suppressible (MARTIENSSEN et al. 1990 ; CHOMET et al. 1991 ; GREENE et al. 1994 ; HU et al. 1998 ; GIRARD and FREELING 2000 ). These suppressible alleles arose via Mu insertion into promoters (3 alleles), 5' UTRs (5), and introns (4). The low rate at which suppressible alleles have been described probably underestimates the rate at which they arise. This is because many suppressible alleles are likely to have been incorrectly characterized as not transmissible/heritable and were therefore not further analyzed. Consistent with this hypothesis, we found that 65 of 75 Mu insertions in the glossy8 gene were in the promoter or in the 5' UTR and are therefore potentially suppressible (DIETRICH et al. 2002 ).

In this study a collection of five Mu-derived rf2a mutant alleles originally obtained via a phenotypic screen was backcrossed to a non-Mu inbred line (Ky21) for at least three generations. Two alleles (rf2a-m9390 and rf2a-m8110) with Mu insertions in noncoding regions exhibited low penetrance early in this backcrossing program. A third allele, rf2a-m9437, exhibited low penetrance in later generations (data not shown). The low penetrance of these three alleles is not the result of DNA rearrangements and at least rf2a-m8110 and rf2a-m9390 can be reactivated by the introduction of Mu activity. The Mu1 transposon in rf2a-m9390 is methylated in fertile plants, further confirming the relationship between the low penetrance and the absence of Mu activity. Furthermore, on the basis of its low penetrance and Mu insertion site, rf2a-m9437 is also likely to be a suppressible allele. Thus three of the five Mu-insertion alleles analyzed appear to be Mu suppressible. Hence, it is clear that Mu-suppressible alleles arise frequently enough to seriously impact the outcome of genetic experiments. In retrospect, suppression probably accounts for our lack of success in identifying a Mu-containing DNA fragment that cosegregated with any of the five Mu-induced rf2a alleles except rf2a-m8122 (CUI et al. 1996 ). An even greater proportion of suppressible alleles are likely to be recovered by reverse genetic screens (e.g., DIETRICH et al. 2002 ), because such alleles would not have been prescreened by phenotypic selection as was true for this collection of rf2a-m alleles.

It has previously been established that the insertion of Mu transposons into promoter regions, 5' UTRs, and introns can generate suppressible alleles. By determining the structure of the Rf2a-B73 allele and the insertion sites of the Mu transposons responsible for four of five rf2a-m alleles, this study not only identified two suppressible alleles with Mu insertions in the 5' UTR of the rf2a gene (rf2a-m9390 and rf2a-m9437), but also for the first time demonstrated that a Mu insertion into a 3' UTR can result in a suppressible allele (rf2a-m8110). Only the apparently nonsuppressible rf2a-m8122 allele arose via a Mu insertion in an exon.

The Rf2a-B73 allele spans >17 kb (Fig 1) and contains many small exons and introns as well as two large introns (the largest of which is >7 kb). It is worth noting that if a reverse genetic screen that depends upon PCR-based detection of Mu insertions (BENSEN et al. 1995 ) were to be conducted on the rf2a gene, it is likely that most of the rf2a-m alleles described in this report would have been missed. This is because as a result of the presence of two large introns, only rf2a primers from a small portion of the gene can amplify the rf2a-m alleles. For example, Mu- and rf2a-specific (such as RF2C6, Fig 1) primers from the 5' end of the rf2a gene fail to amplify DNA templates from plants that carry rf2a-m8122 or rf2a-m8110 (data not shown). Similarly, under all PCR conditions tested, Mu and rf2a primers from the 3' end of the gene fail to amplify DNA from plants that carry rf2a-m9390 and rf2a-m9437. Instead, such a reverse genetic screen would most likely identify Mu insertions in introns that would be unlikely to confer a mutant phenotype. These results point to the importance of having a gene structure available before performing reverse genetics and/or using a large number of primer pairs when conducting reverse genetic screens in species such as maize, which may have large introns.

Suppressible alleles described previously arose via the insertion of a variety of Mu transposons: Mu1, Mu3, Mu8, and MuDR deletion derivatives (BARKAN and MARTIENSSEN 1991 ; LOWE et al. 1992 ; GREENE et al. 1994 ; HU et al. 1998 ; GIRARD and FREELING 2000 ). Three rf2a-m-suppressible alleles, rf2a-m9390, rf2a-m8110, and rf2a-m9437, are Mu1-, rcy/Mu7-, and Mu10-induced, respectively. These results indicate that insertions of at least six different classes of Mu transposons can generate suppressible alleles.

Mu suppression occurs to alleles that have a Mu insertion in the 5' UTR:
Three mutant alleles that arose via Mu transposon insertions in 5' UTRs recruit alternative transcription initiation sites during suppression. The hcf106::Mu1, Lg3-Or422, and rf2a-m9390 alleles have Mu1, Mu3, and Mu1 insertions at positions -34 (BARKAN and MARTIENSSEN 1991 ), -238 (GIRARD and FREELING 2000 ), and -105 relative to the presumed translation start codons, respectively. The alternative transcription sites recruited during suppression of hcf106::Mu1 and rf2a-m9390 exhibit a significant difference in relation to their Mu1 insertions. In the suppressed hcf106::Mu1 allele, alternative transcription sites are recruited both from the downstream TIR of the inserted Mu1 transposon and from the region of the 5' UTR that is 3' of this Mu1 insertion. In contrast, all of the transcription initiation sites recruited during suppression of rf2a-m9390 are located 3' of the Mu1 insertion; i.e., there is no evidence that this allele recruits transcription initiation sites from within Mu1. This difference is not due to the orientation of the transposons; the Mu1 transposons responsible for the hcf106::Mu1 allele and the rf2a-m9390 allele are inserted in the same relative orientation. However, the positions of all these alternative transcription initiation sites are similar relative to the presumed translation start sites: hcf106::Mu1 uses positions -7 to -80 and rf2a-m9390 uses positions -6 to -91. This observation suggests that there may be a downstream cis-acting signal that interacts with the promoter to determine the locations of transcription initiation sites.

The rf2a-m9437 allele also arose via a Mu insertion in the 5' UTR. Unlike rf2a-m9390, which exhibited evidence of suppression very early in the backcrossing program, rf2a-m9437 did not begin to exhibit evidence of suppression until after three generations of backcrossing to the inbred line Ky21 (Table 1 and data not shown). Interestingly, the position of the Mu10 insertion responsible for rf2a-m9437 (base pair -35 relative to the presumed translation start codon) precludes the use of all but one of the alternative transcription initiation sites used by rf2a-m9390.

Mu suppression affects an allele that has a Mu insertion in its 3' UTR:
Analysis of rf2a-m8110 revealed a novel mechanism of Mu suppression. This allele has an rcy/Mu7 transposon insertion 30 bp downstream of the stop codon in the 3' UTR. Sequence analysis of cloned 3' RACE products revealed that during suppression the rf2a-m8110 allele recruits novel polyadenylation sites from within the TIR of the rcy/Mu7 transposon. This is the first report that rcy/Mu7 contains polyadenylation sites. Because the TIRs of the 10 classes of Mu transposons are well conserved, analysis of the rf2a-m8110 allele suggests that the recruitment of alternative polyadenylation sites might be a general mechanism by which mutants containing other classes of Mu insertions in their 3' UTRs could be suppressed. This is similar to the finding that the long terminal repeats of retroviruses can provide polyadenylation sites for human cellular transcripts (MAGER et al. 1999 ; BAUST et al. 2000 ). Thus, the TIRs of Mu transposon may serve as polyadenylation sites for maize genes and thereby contribute to the generation of genetic diversity.

Polyadenylation sites in Mu1 were identified previously following the analysis of truncated transcripts produced by the adh1-s3034 allele, which contains a Mu1 insertion in its first intron (ORTIZ and STROMMER 1990 ). However, these alternative polyadenylation sites differ from those in the current report in that they are located in the central (i.e., non-TIR) region and the 3' TIR of the Mu1 transposon.

In eukaryotes, polyadenylation occurs following the cleavage of the hnRNA, which produces a 3' -OH. The positions at which cleavage, and hence polyadenylation, occur are controlled by cis-acting elements that are usually located between the stop codon and the cleavage site. In mammals, the polyadenylation signals consist of the AAUAAA positioning element (PE) and the U- or UG-rich downstream element (DE; WAHLE and RUEGSEGGER 1999 ; ZHAO et al. 1999 ). Plant and yeast genes contain AAUAAA-like PEs and U- or UG-rich upstream elements (UEs), which resemble the DEs of mammals (reviewed by ROTHNIE 1996 ; HUNT and MESSING 1998 ; WAHLE and RUEGSEGGER 1999 ). In addition, the polyadenylation sites of rice and Arabidopsis genes are often located in U-rich regions (GRABER et al. 1999 ).

Seven polyadenylation sites were detected in Rf2a alleles (VI–XII in Fig 6). Each of these sites is 20–50 bases downstream of an AAUAAA-like PE and in the vicinity of a U-rich flanking region (each 50% U). No obvious UE was detected in the rf2a gene. Similarly, all five polyadenylation sites in the TIR of rcy/Mu7 (I–V in Fig 6) are located in the vicinity of U-rich regions and 20–50 bp downstream of one of the three AAUAAA-like PE elements that exist in this TIR.

The fact that only the TIR polyadenylation sites are used in suppressed rf2a-m8110 plants indicates either that transcription does not proceed through the rcy/Mu7 transposon to the native polyadenylation sites or that the TIR sites are preferred over the native sites by the polyadenylation machinery. One hypothesis to explain such a preference would be the presence of an as-yet-unidentified polyadenylation signal upstream of the position of the rcy/Mu7 insertion. The spatial separation between this upstream signal and the native sites caused by the 2.2-kb rcy/Mu7 insertion in the rf2a-m8110 allele might cause such an upstream signal to interact more efficiently with the nearby TIR sites than with the native sites.

In vitro and in vivo polyadenylation experiments demonstrated that cleavage occurs preferentially at the 3' end of a YA (i.e., UA, CA, or GA) sequence (reviewed by ROTHNIE 1996 ; WAHLE and RUEGSEGGER 1999 ). In mammals, this YA consensus was observed in 70 of 100 polyadenylation sites (SHEETS et al. 1990 ). YA was also found to be the most frequent sequence at the polyadenylation sites in the yeast cyc1 gene (GUO and SHERMAN 1995 ) and several plant genes (WU et al. 1995 ). Our analysis of 94 polyadenylation sites from 58 maize genes confirmed that this rule also holds for maize; 73 of 94 polyadenylation sites fit the YA pattern (Table 4). In contrast, only 2 of 5 polyadenylation sites in the TIR of the rcy/Mu7 transposon inserted into rf2a-m8110 follow this rule. In mammals, the preferred bases at the first position of the YA consensus are C >> U > G, where C accounts for 60% of sites (SHEETS et al. 1990 ). In contrast, UA is most common (45%) among maize polyadenylation sites.

	FOOTNOTES

¹ Present address: The Jackson Laboratory, 600 Main St., Bar Harbor, ME 40609.
² Present address: The University of Texas Health Science Center San Antonio, 7703 Floyd Curl Dr., San Antonio, TX 78229.

	ACKNOWLEDGMENTS

We thank Ms. Weijun Chen for assistance in sequencing the rf2-DNA1 genomic clone; Steve Briggs, who, while at Pioneer Hi-Bred International, provided the two B73 genomic libraries; and Alfons Gierl (Technische Universitat, Munchen, Germany) for the GAPDH clone. This research was partially supported by a USDA/NRI grant (9600804) to P.S.S. and R.P.W., USDA/NRI grants (98001805, 0001478, 0201414) to P.S.S., a Human Frontiers in Science program grant (RG0067) to Cris Kuhlemeier (Institute of Plant Physiology, University of Berne, Switzerland) and P.S.S., a National Science Foundation grant (DBI-9975868) to P.S.S. and D.A., and Hatch Act and State of Iowa funds. This is journal paper no. J-18707 of the Iowa Agriculture and Home Economics Experiment Station (Ames, IA); project numbers are 3368, 3390, 3485, and 3554.

Manuscript received March 1, 2002; Accepted for publication November 4, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AUSUBEL, F. M., R. BRENT, R. E. KINGSTON, D. D. MOORE, J. G. SEIDMAN et al., 1999 Current Protocols in Molecular Biology. John Wiley & Sons, New York.

BARKAN, A. and R. A. MARTIENSSEN, 1991  Inactivation of maize transposon Mu suppresses a mutant phenotype by activating an outward-reading promoter near the end of Mu1.. Proc. Natl. Acad. Sci. USA 88:3502-3506.[Abstract/Free Full Text]

BAUST, C., W. SEIFARTH, H. GERMAIER, R. HEHLMANN, and C. LEIB-MJOSCH, 2000  HERV-K-T47D-related long terminal repeats mediate polyadenylation of cellular transcripts. Genomics 66:98-103.[Medline]

BENITO, M. I. and V. WALBOT, 1997  Characterization of the maize Mutator transposable element MURA transposase as a DNA-binding protein. Mol. Cell. Biol. 17:5165-5175.[Abstract]

BENNETZEN, J. L., 1996  The Mutator transposable element system of maize. Curr. Top. Microbiol. Immunol. 204:195-229.[Medline]

BENNETZEN, J. L., P. S. SPRINGER, A. D. CRESSE, and M. HENDRICKX, 1993  Specificity and regulation of the Mutator transposable element system of maize. Crit. Rev. Plant Sci. 12:57-95.

BENSEN, R. J., G. S. JOHAL, V. C. CRANE, J. T. TOSSBERG, and P. S. SCHNABLE et al., 1995  Cloning and characterization of the maize An1 gene. Plant Cell 7:75-84.[Abstract]

CHANDLER, V. L. and K. J. HARDEMAN, 1992  The Mu elements of Zea mays.. Adv. Genet. 30:77-122.[Medline]

CHANDLER, V. L. and V. WALBOT, 1986  DNA modification of a maize transposable element correlates with loss of activity. Proc. Natl. Acad. Sci. USA 83:1767-1771.[Abstract/Free Full Text]

CHOMET, P., D. LISCH, K. J. HARDEMAN, V. L. CHANDLER, and M. FREELING, 1991  Identification of a regulatory transposon that controls the Mutator transposable element system in maize. Genetics 129:261-270.[Abstract]

CUI, X., R. P. WISE, and P. S. SCHNABLE, 1996  The rf2 nuclear restorer gene of male sterile T-cytoplasm maize. Science 272:1334-1336.[Abstract]

DEAN, C., P. VAN DEN ELZEN, S. TAMAKI, P. DUNSMUIR, and J. BEDBROOK, 1985  Differential expression of the eight genes of the petunia ribulose bisphosphate carboxylase small subunit multi-gene family. EMBO J. 4:3055-3061.[Medline]

DIETRICH, C. R., F. CUI, M. L. PACKILA, J. LI, and D. A. ASHLOCK et al., 2002  Maize Mu transposons are targeted to the 5' untranslated region of the gl8 gene and sequences flanking Mu target-site duplications exhibit nonrandom nucleotide composition throughout the genome. Genetics 160:697-716.[Abstract/Free Full Text]

DONLIN, M. J., D. LISCH, and M. FREELING, 1995  Tissue-specific accumulation of MURB, a protein encoded by MuDR, the autonomous regulator of the Mutator transposable element family. Plant Cell 7:1989-2000.[Abstract]

EISEN, J. A., M. I. BENITO, and V. WALBOT, 1994  Sequence similarity of putative transposases links the maize Mutator autonomous element and a group of bacterial insertion sequences. Nucleic Acids Res. 22:2634-2636.[Abstract/Free Full Text]

FEDOROFF, N. V., 1999  The suppressor-mutator element and the evolutionary riddle of transposons. Genes Cells 4:11-19.[Abstract]

FOWLER, J. E., G. J. MUEHLBAUER, and M. FREELING, 1996  Mosaic analysis of the liguleless3 mutant phenotype in maize by coordinate suppression of Mutator-insertion alleles. Genetics 143:489-503.[Abstract]

GIRARD, L. and M. FREELING, 2000  Mutator-suppression alleles of rough sheath1 and liguleless3 in maize reveal multiple mechanisms for suppression. Genetics 154:437-446.[Abstract/Free Full Text]

GRABER, J. H., C. R. CANTOR, S. C. MOHR, and T. F. SMITH, 1999  In silico detection of control signals: mRNA 3'-end-processing sequences in diverse species. Proc. Natl. Acad. Sci. USA 96:14055-14060.[Abstract/Free Full Text]

GREENE, B., R. WALKO, and S. HAKE, 1994  Mutator insertions in an intron of the maize knotted1 gene result in dominant suppressible mutations. Genetics 138:1275-1285.[Abstract]

GUO, Z. and F. SHERMAN, 1995  3'end-forming signals of yeast mRNA. Mol. Cell. Biol. 15:5983-5990.[Abstract]

HERSHBERGER, R. J., C. A. WARREN, and V. WALBOT, 1991  Mutator activity in maize correlates with the presence and expression of the Mu transposable element Mu9.. Proc. Natl. Acad. Sci. USA 88:10198-10202.[Abstract/Free Full Text]

HSIA, A.-P. and P. S. SCHNABLE, 1996  DNA sequence analyses support the role of interrupted gap repair in the origin of internal deletions of the maize transposon, MuDR.. Genetics 142:603-618.[Abstract]

HU, G., N. YALPANI, S. P. BRIGGS, and G. S. JOHAL, 1998  A porphyrin pathway impairment is responsible for the phenotype of a dominant disease lesion mimic mutant of maize. Plant Cell 10:1095-1105.[Abstract/Free Full Text]

HUNT, A. G., and J. MESSING, 1998 mRNA polyadenylation in plants, pp. 29–39 in A Look Beyond Transcription: Mechanisms Determining mRNA Stability and Translation in Plants, edited by J. BAILEY-SERRES and D. R. GALLIE. American Society of Plant Physiologists, Rockville, MD.

JAMES, M. G., M. J. SCANLON, M. QIN, D. S. ROBERTSON, and A. M. MYERS, 1993  DNA sequence and transcript analysis of transposon MuA2, a regulator of Mutator transposable element activity in maize. Plant Mol. Biol. 21:1181-1185.[Medline]

KONIECZNY, A., D. F. VOYTAS, M. P. CUMMINGS, and F. M. AUSUBEL, 1991  A superfamily of Arabidopsis thaliana retrotransposons. Genetics 127:801-809.[Abstract]

KUNZE, R., 1996  The maize transposable element activator (Ac). Curr. Top. Microbiol. Immunol. 204:161-194.[Medline]

LISCH, D., L. GIRARD, M. DONLIN, and M. FREELING, 1999  Functional analysis of deletion derivatives of the maize transposon MuDR delineates roles for the MURA and MURB proteins. Genetics 151:331-341.[Abstract/Free Full Text]

LIU, F., X. CUI, H. T. HOMER, H. WEINER, and P. S. SCHNABLE, 2001  Mitochondrial aldehyde dehydrogenase activity is required for male fertility in maize (Zea mays L.). Plant Cell 13:1063-1078.[Abstract/Free Full Text]

LOWE, B., J. MATHERN, and S. HAKE, 1992  Active Mutator elements suppress the Knotted phenotype and increase recombination at the Kn1-0 tandem duplication. Genetics 132:813-822.[Abstract]

MAGER, D. L., D. G. HUNTER, M. SCHERTZER, and J. D. FREEMAN, 1999  Endogenous retroviruses provide the primary polyadenylation signal for two new human genes (HHLA2 and HHLA3). Genomics 59:255-263.[Medline]

MARTIENSSEN, R. and A. BARON, 1994  Coordinate suppression of mutations caused by Robertson's Mutator transposons in maize. Genetics 136:1157-1170.[Abstract]

MARTIENSSEN, R. A., A. BARKAN, M. FREELING, and W. C. TAYLOR, 1989  Molecular cloning of a maize gene involved in photosynthetic membrane organization that is regulated by Robertson's Mutator.. EMBO J. 8:1633-1639.[Medline]

MARTIENSSEN, R., A. BARKAN, W. C. TAYLOR, and M. FREELING, 1990  Somatically heritable switches in the DNA modification of Mu transposable elements monitored with a suppressible mutant in maize. Genes Dev. 4:331-343.[Abstract/Free Full Text]

MASSON, P., J. A. BANKS, and N. FEDOROFF, 1991  Structure and function of the maize Spm transposable element. Biochimie 73:5-8.[Medline]

MCCLINTOCK, B., 1964  Aspects of gene regulation in maize. Carnegie Inst. Wash. Year Book 63:592-602.

MCCLINTOCK, B., 1967  Genetic systems regulating gene expression during development. Dev. Biol. Suppl. 1:84-112.

ORTIZ, D. F. and J. N. STROMMER, 1990  The Mu1 maize transposable element induces tissue-specific aberrant splicing and polyadenylation in two Adh1 mutants. Mol. Cell. Biol. 10:2090-2095.[Abstract/Free Full Text]

PRELICH, G., 1999  Suppression mechanisms: themes from variations. Trends Genet. 15:261-266.[Medline]

QIN, M., D. S. ROBERTSON, and A. H. ELLINGBOE, 1991  Cloning of the Mutator transposable element MuA2, a putative regulator of somatic mutability of the a1-Mum2 allele in maize. Genetics 129:845-854.[Abstract]

RAIZADA, M. N. and V. WALBOT, 2000  The late developmental pattern of Mu transposon excision is conferred by a cauliflower mosaic virus 35S-driven MURA cDNA in transgenic maize. Plant Cell 12:5-21.[Abstract/Free Full Text]

ROBERTSON, D. S., 1978  Characterization of a mutator system in maize. Mutat. Res. 5:21-28.

ROBERTSON, D. S. and P. S. STINARD, 1989  Genetic analyses of putative two-element system regulating somatic mutability in mutator-induced aleurone mutants of maize. Dev. Genet. 10:482-506.

ROBERTSON, D. S., D. W. MORRIS, P. S. STINARD and B. A. ROTH, 1988 Germline and somatic Mutator activity: Are they functionally related? pp. 17–42, in Plant Transposable Elements, edited by O. NELSON. Plenum Press, New York/London.

ROTHNIE, H. M., 1996  Plant mRNA 3'-end formation. Plant Mol. Biol. 32:43-61.[Medline]

SAGHAI-MAROOF, M. A., K. M. SOLIMAN, R. A. JORGENSEN, and R. W. ALLARD, 1984  Ribosomal DNA spacer-length polymorphisms in barley: Mendelian inheritance, chromosomal location, and population dynamics. Proc. Natl. Acad. Sci. USA 81:8014-8018.[Abstract/Free Full Text]

SCHNABLE, P. S. and P. A. PETERSON, 1986  Distribution of genetically active Cy elements among diverse maize lines. Maydica 31:59-82.

SCHNABLE, P. S. and R. P. WISE, 1994  Recovery of heritable, transposon-induced, mutant alleles of the rf2 nuclear restorer of T-cytoplasm maize. Genetics 136:1171-1185.[Abstract]

SCHNABLE, P. S. and R. P. WISE, 1998  The molecular basis of cytoplasmic male sterility and fertility restoration. Trends Plant Sci. 3:175-180.

SCHNABLE, P. S., P. A. PETERSON, and H. SAEDLER, 1989  The bz-rcy allele of the Cy transposable element system of Zea mays contains a Mu-like element insertion. Mol. Gen. Genet. 217:459-463.[Medline]

SHEETS, M. D., S. C. OGG, and M. P. WICKENS, 1990  Point mutations in AAUAAA and the poly (A) addition site: effects on the accuracy and efficiency of cleavage and polyadenylation in vitro.. Nucleic Acids Res. 18:5799-5805.[Abstract/Free Full Text]

SKIBBE, D. S., F. LIU, T. J. WEN, M. D. YANDEAU, and X. CUI et al., 2002  Characterization of the aldehyde dehydrogenase gene families of Zea mays and Arabidopsis. Plant. Mol. Biol. 48:751-764.[Medline]

STEEL, R. G., and J. H. TORRIE, 1980 Principles and Procedures of Statistics: A Biometrical Approach. McGraw-Hill, New York.

STRATHMANN, M., B. A. HAMILTON, C. A. MAYEDA, M. I. SIMON, and E. M. MEYEROWITZ et al., 1991  Transposon-facilitated DNA sequencing. Proc. Natl. Acad. Sci. USA 88:1247-1250.[Abstract/Free Full Text]

WAHLE, E. and U. RUEGSEGGER, 1999  3'-end processing of pre-mRNA in eukaryotes. FEMS Microbiol. Rev. 23:277-295.[Medline]

WALBOT, V., 1999  Genes, genomes, genomics. What can plant biologists expect from the 1998 National Science Foundation Plant Genome Research Program? Plant Physiol. 119:1151-1156.[Free Full Text]

WISE, R. P., C. R. BRONSON, P. S. SCHNABLE, and H. T. HORNER, 1999  The genetics, pathology, and molecular biology of T-cytoplasm male sterility in maize. Adv. Agronomy 65:79-130.

WU, L., T. UEDA, and J. MESSING, 1995  The formation of mRNA 3'-ends in plants. Plant J. 8:323-329.[Medline]

XU, X., C. R. DIETRICH, M. DELLEDONNE, Y. XIA, and T. J. WEN et al., 1997  Sequence analysis of the cloned glossy8 gene of maize suggests that it may code for a beta-ketoacyl reductase required for the biosynthesis of cuticular waxes. Plant Physiol. 115:501-510.[Abstract]

ZHAO, J., L. HYMAN, and C. MOORE, 1999  Formation of mRNA 3' ends in eukaryotes: mechanism, regulation, and interrelationships with other steps in mRNA synthesis. Microbiol. Mol. Biol. Rev. 63:405-445.[Abstract/Free Full Text]

This article has been cited by other articles:

				W. J. Haun, O. N. Danilevskaya, R. B. Meeley, and N. M. Springer Disruption of Imprinting by Mutator Transposon Insertions in the 5' Proximal Regions of the Zea mays Mez1 Locus Genetics, April 1, 2009; 181(4): 1229 - 1237. [Abstract] [Full Text] [PDF]

				M. L. Robbins, R. S. Sekhon, R. Meeley, and S. Chopra A Mutator Transposon Insertion Is Associated With Ectopic Expression of a Tandemly Repeated Multicopy Myb Gene pericarp color1 of Maize Genetics, April 1, 2008; 178(4): 1859 - 1874. [Abstract] [Full Text] [PDF]

				J. Li, T.-J. Wen, and P. S. Schnable Role of RAD51 in the Repair of MuDR-Induced Double-Strand Breaks in Maize (Zea mays L.) Genetics, January 1, 2008; 178(1): 57 - 66. [Abstract] [Full Text] [PDF]

				J. Li, L. C. Harper, I. Golubovskaya, C. R. Wang, D. Weber, R. B. Meeley, J. McElver, B. Bowen, W. Z. Cande, and P. S. Schnable Functional Analysis of Maize RAD51 in Meiosis and Double-Strand Break Repair Genetics, July 1, 2007; 176(3): 1469 - 1482. [Abstract] [Full Text] [PDF]

				B. P. May, H. Liu, E. Vollbrecht, L. Senior, P. D. Rabinowicz, D. Roh, X. Pan, L. Stein, M. Freeling, D. Alexander, et al. Maize-targeted mutagenesis: A knockout resource for maize PNAS, September 30, 2003; 100(20): 11541 - 11546. [Abstract] [Full Text] [PDF]