Duplication and Suppression of Chloroplast Protein Translocation Genes in Maize

Corresponding author: Robert A. Martienssen, PO Box 100, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 11724., martiens{at}cshl.org (E-mail)

Communicating editor: K. J. NEWTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The HCF106 (high chlorophyll fluorescence) gene of maize encodes a chloroplast membrane protein required for translocation of a subset of proteins across the thylakoid membrane. Mutations in HCF106 caused by the insertion of Robertson's Mutator transposable elements have been mapped to chromosome 2S. Here we show that there is a closely related homolog of HCF106 encoded elsewhere in the maize genome (HCF106c) that can partially compensate for these mutations. This homolog maps on chromosome 10L and is part of the most recent set of segmental duplications in the maize genome. Triple mutants that are disrupted in both the HCF106 and Sec-dependent protein translocation pathways provide evidence that they act independently. The HCF106c gene accounts for a previously reported exception to the correlation between epigenetic suppression of hcf106 and methylation of Mutator transposons. We also demonstrate that insertions of Robertson's Mutator elements into either introns or promoters can lead to mutations whose phenotypes are suppressed in the absence of Mu activity, while alleles with insertions in both positions are not suppressed. The implications of these observations are discussed.

IN higher plants, nuclear mutations that disrupt chloroplast thylakoid membrane biogenesis can be identified as pale green, nonphotosynthetic seedling lethal mutants that survive for a short period after germination by using seed reserves of starch (MILES 1980 ). These mutants are classically known as high chlorophyll fluorescence (hcf) mutants because they release harvested light energy as red fluorescence rather than using it for photosynthetic capture. One class of such mutations disrupt protein translocation during thylakoid biogenesis. "Export" protein translocation mechanisms that facilitate protein transfer through the chloroplast thylakoid, the bacterial plasma membrane, and the endoplasmic reticulum are highly conserved (for reviews see SCHATZ and DOBBERSTEIN 1996 ; SETTLES and MARTIENSSEN 1998 ; DALBEY and ROBINSON 1999 ). In chloroplasts, these pathways include Sec-dependent, SRP-dependent, pH-dependent, and spontaneous pathways. Substrates for the Sec pathway require chloroplast SecA (cpSecA) for efficient translocation, while substrates for the SRP pathway require a homolog of the bacterial SRP54 (cpSRP54), and pH substrates show a strict requirement for the thylakoid pH gradient. In plants, mutations in all but the spontaneous pathway have been identified by using insertional mutagens (BARKAN et al. 1986 ; MARTIENSSEN et al. 1989 ; VOELKER and BARKAN 1995 ; KLIMYUK et al. 1999 ; WALKER et al. 1999 ). These mutations have allowed the properties of mutant membranes to be studied biochemically and the genes to be characterized molecularly (MARTIENSSEN et al. 1989 ; SETTLES et al. 1997 ; VOELKER et al. 1997 ; ROY and BARKAN 1998 ; KLIMYUK et al. 1999 ; WALKER et al. 1999 ).

The hcf106 mutation disrupts the pH-targeting pathway (VOELKER and BARKAN 1995 ) and encodes one of two or more related genes found in higher plants (SETTLES et al. 1997 ). THA4 encodes a distant HCF106 homolog and results in a similarly weak phenotype, suggesting that these genes may be partly redundant (WALKER et al. 1999 ). In prokaryotes HCF106-related genes are found in a conserved operon, the tat operon, along with a gene encoding a multimembrane-spanning protein (tatC). These genes are required for an analogous export targeting pathway in bacteria (BOGSCH et al. 1998 ; SARGENT et al. 1998 ; SETTLES and MARTIENSSEN 1998 ; WEINER et al. 1998 ). Three HCF106/THA4-like genes have been found in Escherichia coli (TatA, B and E) and have redundant roles in targeting, while TatC homologs have been identified in plants and algae through genome sequencing. The tha1 and csy1 mutations disrupt the chloroplast SecA and SecY genes, respectively, in maize. tha1 specifically interferes with the targeting of Sec substrates (VOELKER and BARKAN 1995 ), while the csy1 mutant disrupts targeting through the Sec, SRP, and pH pathways (ROY and BARKAN 1998 ). By analogy with bacteria, this might indicate that all the targeting pathways converge on one translocation channel that includes the SecY protein. Alternatively, the csy1 mutation may indirectly affect the pH pathway due to the overall loss of membrane associated with this mutation, or because TatC, a multimembrane-spanning-domain protein, might require SecY for integration into the membrane (SETTLES and MARTIENSSEN 1998 ). To test potential interactions between the pH- and Sec-targeting machineries, we made triple mutants with lesions in both the pH and Sec pathways. Our results suggest that the Sec- and pH-targeting pathways are independent of each other, in agreement with previous biochemical and genetic studies (CLINE et al. 1993 ; ROY and BARKAN 1998 ; MORI et al. 1999 ).

Although both the tha1 and hcf106 mutants specifically interfere with the normal targeting of Sec and pH substrates, respectively, mutant thylakoids accumulate significant levels of properly targeted substrates in vivo, indicating that cpSecA and HCF106 functions may not be removed completely (VOELKER and BARKAN 1995 ; SETTLES et al. 1997 ). We show here that HCF106 has a "cognate" gene, HCF106C, which encodes a nearly identical protein that can largely rescue the phenotype of hcf106 mutants; this accounts for exceptions to data published previously (MARTIENSSEN et al. 1990 ). THA4, likewise, has a closely related gene (THA9) that exhibits overlapping expression patterns (WALKER et al. 1999 ). In contrast, cpSecA appears to be a single-copy gene in maize (VOELKER et al. 1997 ). Despite the fact that cpSecA accumulates to exceedingly low levels in tha1 mutants, mutant thylakoids accumulate significant levels of properly targeted cpSecA substrates, suggesting that these proteins may not be entirely dependent on cpSecA for their translocation.

The Robertson's Mutator family of transposable elements comprise six families of defective transposons (Mu1/Mu2, Mu3, Mu4, Mu6, Mu7, and Mu8), as well as a master regulatory transposon and its deletion derivatives (MuDR, dMuDR, and Mu5; reviewed in CHANDLER and HARDEMAN 1992 and BENNETZEN 1996 ). Transposition of all Mu elements depends on the transposase encoded by MuDR. When MuDR is lost, transposition ceases and the defective elements acquire DNA methylation. In a relatively large number of cases, mutant phenotypes caused by insertion of defective elements are suppressed when MuDR is lost. These are referred to as epigenetically suppressible mutations because MuDR activity can be lost epigenetically in a process that may involve methylation of the MuDR elements themselves (MARTIENSSEN and BARON 1994 ). The reference allele of tha1-m1 was caused by a Robertson's Mutator insertion into an intron; tha1-m1 mutant seedlings accumulate no detectable tha1 mRNA and have 50-fold reduction in cpSecA, suggesting that the reference allele is null or nearly so (VOELKER et al. 1997 ). Here, we show that this allele is epigenetically regulated by transposon activity. A similar insertion into the intron of hcf106-mum4 is also regulated epigenetically. Comparison between these and other examples of suppression leads us to propose a new model for the epigenetic regulation of genes by transposon insertions.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Cloning and mapping of HCF106C DNA:
The HCF106C cDNA was isolated from a cDNA library prepared from the leaves of 2-week-old greenhouse-grown maize seedlings (inbred B73) in a ZAPII vector (Stratagene, La Jolla, CA) made as recommended by the supplier (FISK et al. 1999 ). A 0.4-kb SstI, PstI fragment spanning the first two exons and first intron of the gene (MARTIENSSEN et al. 1989 ) was used as a probe for the cDNA library. Two classes of cDNAs were identified, corresponding to HCF106 and HCF106C. Sequencing revealed that they encoded closely related open reading frames (ORFs), but differed in their 5' and 3' untranslated regions (UTRs). At least four independent clones with differing 5' and 3' ends were sequenced from each class. The longest clones are represented in GenBank entries AF027808 and AF237945 and begin from 11 bp (HCF106c) and 23 bp (HCF106) from the start point of transcription (BARKAN and MARTIENSSEN 1991 ). The genomic locus of HCF106C was amplified from hcf106-mum3, hcf106-mum2, and B73 genomic DNA using the TaKaRa LA PCR kit and the primers HCFCOG1 and HCFCOG3 (Table 1), which are specific for the 5' and 3' UTR of HCF106C. The 3.6-kb PCR products were purified using the QIAquick PCR purification kit (QIAGEN) and sequenced directly using AmpliTaqFS Dye Terminator cycle sequencing (Perkin Elmer, Norwalk, CT). The entire product was sequenced with the primers listed in Table 1. The position of the third intron in the HCF106 genomic locus was confirmed by amplifying and sequencing a 171-bp product with HCFT1 and HCFB1 primers (Table 1) that are specific to the HCF106 cDNA.

The HCF106C locus was mapped by using restriction fragment length polymorphisms (RFLP) identified with the 3'-360-bp SphI XhoI fragment of the HCF106C cDNA as a probe. Digests of genomic DNA with BglII and HindIII had large differences in fragment sizes between the inbred lines Tx303 and CO159, while no polymorphisms were identified between the inbred lines T232 and CM37. Thirty-six recombinant inbreds from the Tx303 and CO159 population were scored by DNA gel blot for both the BglII and the HindIII RFLP. The genetic map position of HCF106C was calculated with MapMaker and is available in the Acemaz database (http://burr.bio.bnl.gov/acemaz.html).

Protein analysis:
The nearly full-length HCF106 and HCF106C cDNA clones obtained from the B73 library were transcribed in vitro with T3 RNA polymerase (Boehringer Mannheim, Indianapolis), capped, and translated in vitro with wheat germ extract (Promega, Madison, WI) in the presence of [³⁵S]methionine according to recommendations of the suppliers (SETTLES et al. 1997 ). Half of the translation products were then boiled in 2% SDS and immunoprecipitated in 10 vol of RIPA buffer (SAMBROOK et al. 1989 ) with 2 µl of unpurified antisera against HCF106 (SETTLES et al. 1997 ). After 1 hr incubation at room temperature, labeled proteins were recovered by binding to protein A Sepharose and washing extensively in RIPA buffer as well as 100 mM Tris, pH 7.5, supplemented with NP-40 (nonionic detergent). Samples were boiled in SDS sample buffer before resolving on 12% polyacrylamide minigels that were fixed and infiltrated with Enhance (New England Nuclear, Boston) before drying and autoradiography.

For the triple-mutant analysis, total leaf protein was extracted from 0.2 g of seedling leaf tissue into 0.5 ml protein extraction buffer (100 mM Tris-HCl, pH 7.2, 5 mM EGTA, 5 mM MgCl₂, 10% sucrose, 40 mM 2-mercaptoethanol, 2 mM phenylmethylsulfonyl fluoride) on ice. The samples were filtered through glass wool and then separated on 15% polyacrylamide gels using standard methods (SAMBROOK et al. 1989 ). Protein gel blots were performed as described by SETTLES et al. 1997 .

Genetic analysis:
Crosses between hcf106-mum1 and a full-colored stock carrying bz-mum9, which was of mixed genetic origin (DOSEFF et al. 1991 ), were self-pollinated and segregated a low frequency of lethal, hcf seedlings. Families homozygous for hcf106-mum1 were selected by DNA gel blot and scored for Mu1 methylation by HinfI digestion and DNA gel blot (MARTIENSSEN et al. 1990 ). The selected lines that continued to segregate hcf mutants were polymorphic for a BclI site close to the 3' end of the HCF106C gene when tested by DNA gel blot. The B73 inbred is not polymorphic at this site.

To test cosegregation of the hcf106C mutation with the hcf106 phenotype, hcf106-mum2 and hcf106-mum3 heterozygous plants were crossed to B73. The F₁ progeny were self-pollinated and the F₂ progeny were screened for hcf106 mutant seedlings. DNA was extracted from pools of four to seven mutants and seven wild-type siblings from individual families. The HCFCOG11 and HCFSST primers were used to amplify a 450-bp product from the HCF106C locus, which was then digested with MspI. Common 295-bp (not shown) and 65-bp fragments, as well as polymorphic 86- and 93-bp fragments, were resolved on a 3% TBE agarose gel. This marker was tested for codominance by mixing B73 and hcf106C mutant DNA (see Fig 4B). Genomic DNA was extracted as described previously (SETTLES et al. 1997 ).

View larger version (29K): In this window In a new window Download PPT slide   Figure 1. Structure of HCF106 and HCF106C. (A) Predicted amino acid sequences of HCF106 and HCF106C. The intron positions in these sequences are marked with filled triangles and the predicted chloroplast-targeting domain cleavage site is marked with an arrow. The conserved Hcf106 domain includes a hydrophobic membrane domain (underlined) and a predicted amphipathic -helix (boldface type). The cDNA sequence is available in GenBank (accession no. AF237945). (B) Schematic of the hcf106 and hcf106C loci. The structure of the hcf106 locus was determined by restriction mapping and sequencing the splice sites. Although both loci have the same intron/exon junctions, the hcf106 locus has a much larger second intron. The hcf106C genomic sequence is available in GenBank (accession no. AF237944).

View larger version (7K): In this window In a new window Download PPT slide   Figure 2. Relative map positions of hcf106/hcf106C and b1/r1. hcf106 and hcf106C were mapped by recombinant inbred mapping (see MATERIALS AND METHODS).

View larger version (65K): In this window In a new window Download PPT slide   Figure 3. Analysis of Hcf106 and Hcf106C protein. (A) Antibodies against HCF106 were used to immunoprecipitate in vitro translation products from HCF106 (lanes 1 and 2) and HCF106C (lanes 3 and 4) cDNA. Total products (lanes 1 and 3) and immunoprecipitates (lanes 2 and 4) are shown. Precursor protein migrates as a 35-kD protein, more slowly than its amino acid sequence predicts (SETTLES et al. 1997 ). (B) Western analysis of viable seedlings from an F2 family from hcf106-mum1/+ crossed to B73: circles are homozygous and squares are heterozygous for hcf106-mum1; solid symbols are homozygous and hatched symbols are heterozygous for HCF106C. This family also segregated lethal mutants (not shown), but all the viable plants had significant levels of HCF106 or HCF106C protein. Mature HCF106 protein (lower arrow) migrates more slowly than predicted, at 30 kD (SETTLES et al. 1997 ). A larger protein corresponding to HCF106C (middle arrow) can be observed in viable hcf106 seedlings with unmethylated Mutator elements (lane 7) as well as in wild type (lanes 4 and 6) at levels equivalent to HCF106 protein in suppressed mutants (lanes 1–3 and 5). Overloading of wild-type proteins (lane 8) revealed a 35-kD band that may correspond to unprocessed precursor protein (upper arrow). (C) Mutator methylation. DNA was isolated from the seedlings used in part B, digested with HinfI, and probed with Mu1. Unmethylated elements are marked with arrows. Samples in lanes 1–3 were from viable seedlings corresponding to lanes 3, 7, and 5 in B; lane 4 is from a lethal hcf seedling from the same family. (D) RFLP analysis. A pedigree from hcf106-mum1/+ crossed to a full color line (see MATERIALS AND METHODS) is shown. DNA samples from F2 progeny of self-pollinated plants were digested with BclI and analyzed by DNA gel blot with a hybridization probe from the 3' end of the hcf106C gene. All samples shown are homozygous for hcf106-mum1 but viable. Lanes 1–7 contain DNA from siblings that did not segregate any lethal mutants, while lethals segregated among siblings of the plants in lanes 8–11 (not shown).

View larger version (39K): In this window In a new window Download PPT slide   Figure 4. Molecular analysis of hcf106C alleles. (A) The hcf106C mutation is caused by a 7-bp direct repeat in the 5' end of the coding sequence. The MspI site is therefore polymorphic and can be detected as a cleavable amplified polymorphic sequence marker. (B) The hcf106C mutation cosegregates with the hcf106 mutant phenotype. The 5' end of the hcf106C locus was amplified and digested with MspI from the F2 progeny of a cross between either hcf106-mum2 or hcf106-mum3 and B73. The polymorphism can be seen between the parental lines (lanes 1 and 2), and the marker shows codominance when the parental DNA is mixed (lane 3). Pools of four to seven hcf mutants and seven normal siblings are compared in lanes 4–7. The hcf106C mutant allele cosegregates with the hcf phenotype in seven F2 families (not shown). Lanes 8 and 9 show progeny from a hcf106-mum1 homozygous plant that was Mu active and viable in a previous study (MARTIENSSEN et al. 1990 ). This plant was heterozygous for the hcf106C mutant allele.

To construct triple mutants, hcf106-mum3 heterozygous plants were crossed to tha1 heterozygous plants. The F₁ plants were self-pollinated and triple heterozygotes were identified by DNA gel blots for hcf106-mum3 and tha1 (all F₁ plants are heterozygous for hcf106C; see Fig 7, top). The F₂ progeny from triple heterozygotes were screened for pale green mutants. Protein and DNA were extracted from all pale green mutants. The protein extracts were analyzed by protein gel blots with antibodies directed against OE17, PC, and HCF106 (VOELKER and BARKAN 1995 ; SETTLES et al. 1997 ). hcf106, tha1, and triple-mutant genotypes were confirmed by DNA gel blot (SAMBROOK et al. 1989 ). Mu activity was monitored by HinfI digestion and probing with a Mu1 internal fragment (MARTIENSSEN et al. 1990 ).

View larger version (127K): In this window In a new window Download PPT slide   Figure 5. hcf106C partially complements hcf106. (A) hcf106-mum1 homozygote and Les28, hcf106C double heterozygote derived from a backcross to B73. The Mu activity state in leaf tissue is marked both by Les28 and hcf106-mum1. Dark green sectors (arrow) are seen coordinately with loss of lesions. (B) Rare pale green sectors (arrow) in the same genetic background are seen on only Mu-active portions of the leaf.

View larger version (22K): In this window In a new window Download PPT slide   Figure 6. hcf106, hcf106c, tha1 triple mutants are additive. (A) Crossing scheme to generate hcf106, hcf106C, tha1 triple mutants. Only plants homozygous for both hcf106 and hcf106C show the hcf106 mutant phenotype. Triple heterozygotes were selected by DNA gel blot and the F2 progeny from these lines were screened for pale green seedling mutants. DNA and protein were extracted from the F2 pale green seedlings. Only 1/64 of the F2 is expected to be triple mutant. (B) Protein gel blots of a hcf106, hcf106C, tha1 triple mutant and its single-mutant siblings. The blot was incubated with anitbodies against HCF106, OE23 (a substrate of the pH pathway), and PC (a substrate of the Sec pathway). DNA extracted from the same seedling was used to genotype each seedling and monitor Mu activity by DNA gel blot. The single mutants variably accumulate stromal intermediates specific for the pathway blocked (lanes 1 and 2). The triple mutant accumulates similar or higher levels of mature target proteins (lane 3) as the single mutants do, while a normal F2 sibling appears wild type (lane 4).

View larger version (142K): In this window In a new window Download PPT slide   Figure 7. The tha-m1 phenotype is suppressed in Mu-inactive tissue. Lanes 1–12 show DNA extracted from a single tha1 homozygote that was pale green as a seedling but survived through flowering. Lanes 13–17 show DNA extracted from seedlings of different genotypes. DNAs were digested with HinfI to monitor methylation of Mu elements. The blot was probed with a Mu1 probe (Mu1, top). The DNAs were also digested with EcoRI and probed with a tha1 clone to monitor the tha1 genotype (tha1, bottom). The pale green seedling that survived through flowering shows progressive increases in Mu1 methylation in the upper leaves of the plant. Viable and lethal tha1 siblings (lanes 13–15) show similar differences in Mu1 methylation as Mu-on and Mu-off hcf106 plants (lanes 16–17). A tha1 heterozygous plant shows the EcoRI polymorphism for tha1 (lane 18).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

HCF106C encodes a homolog of HCF106:
The hcf106-mum1 allele has a Mu1 transposable element inserted in the promoter of the HCF106 gene (MARTIENSSEN et al. 1989 ; BARKAN and MARTIENSSEN 1991 ). HCF106 cDNA was isolated from the maize inbred B73 by rapid amplification of cDNA ends (RACE) PCR using primers from the 5' untranslated region (SETTLES et al. 1997 ). At the same time, hybridization was used to identify clones in a maize leaf cDNA library. Two classes of cDNA clones were identified and sequenced: one corresponded to the RACE PCR product, but the second class did not and was termed HCF106C for "cognate" gene (see MATERIALS AND METHODS). The HCF106C cDNA encodes a predicted peptide of 238 amino acids that is highly homologous to the 243-amino-acid HCF106 polypeptide (Fig 1A).

To further characterize HCF106C, the genomic locus was amplified with specific primers from the 5' and 3' ends of the cDNA. Analysis revealed that the HCF106C coding region is 3.6 kb and has the same intron/exon structure as HCF106 (Fig 1B). A similarly close homolog (AtHCF106) has been identified on Arabidopsis thaliana chromosome 5 with identical intron/exon structure (NAKAMURA 2000 ; K. CLINE, personal communication), suggesting that both HCF106 and HCF106C are orthologous to the AtHCF106 locus and that the maize duplication arose after the divergence of monocots and dicots. The two classes of cDNA in maize are distinguished by unique 5' and 3' UTR and some nucleotide differences in the ORF (not shown). Using these sequences as probes, HCF106C was mapped with recombinant inbred populations (BURR et al. 1988 ) to chromosome 10L near the male gametophyte specific 1 (mgs1) gene (STINSON et al. 1987 ), while HCF106 mapped to chromosome 2S (see MATERIALS AND METHODS). The r and b loci of maize are also duplicated genes on chromosomes 2S and 10L (Fig 2).

The hcf106 phenotype is dependent upon two genetic factors:
Plants that are homozygous for hcf106-mum1 have the hcf106 phenotype only if they retain Robertson's Mutator activity (MARTIENSSEN et al. 1989 , MARTIENSSEN et al. 1990 ; MARTIENSSEN and BARON 1994 ; see Table 1). Mutator activity can be monitored by following the extent of DNA methylation in Mu1 elements: plants with active Mutator have unmethylated Mu1 elements, while plants with inactive Mutator have methylated elements (CHANDLER and WALBOT 1986 ). In general, F₂ families from Mu-active, heterozygous plants segregated hcf seedlings, while families from Mu-inactive heterozygotes never segregate hcf106 mutant seedlings. However, an exceptional family included a plant that was homozygous for hcf106-mum1 and had hypomethylated Mu1 elements, but the plant was viable and gave rise to wild-type progeny on self-pollination (MARTIENSSEN et al. 1990 ). As the exceptional plant had been derived from the F₂ of an outcross to B73, one possibility was that B73 carried a second locus that suppressed the mutant phenotype independently of methylation.

To determine if there was a second locus that influenced the mutant phenotype, derivative alleles of hcf106-mum1 were first generated by site-selected mutagenesis to eliminate Mutator suppression (DAS and MARTIENSSEN 1995 ). Three new alleles were generated from this screen. hcf106-mum2 contains an additional insertion of a Mu1 element in intron 1 of the HCF106 gene, as well as the original insertion in the upstream region. hcf106-mum3 retains the original upstream insertion, but has lost exon 1 and part of intron 1 to produce a null allele of HCF106. The third derivative, hcf106-mum4, has a new dMuDR insertion in intron 1 and no upstream insertion. It was derived from insertion into the wild-type (B73) allele at the same nucleotide as the insertion in hcf106-mum2. The hcf106-mum2 and hcf106-mum3 alleles are no longer epigenetically regulated so that plants heterozygous for these alleles segregate seedling lethal progeny in Mu-off lines (Table 2). In contrast, plants homozygous for hcf106-mum4 are fully viable in Mu-off backgrounds (DAS and MARTIENSSEN 1995 ).

Although the HCF106 locus was inherited as a single locus in the original background (MARTIENSSEN et al. 1990 ), in outcrosses to the B73 inbred line, F₁ heterozygotes typically have fewer than 1/16 lethal hcf mutant progeny when self-pollinated (not shown). This low frequency of mutants could be caused by epigenetic suppression or by a second gene that suppresses the hcf106 phenotype (MARTIENSSEN et al. 1990 ). We reasoned that if epigenetic factors alone were responsible, then the derivative alleles hcf106-mum2 and hcf106-mum3 should no longer show low penetrance in this way. However, outcrosses of these derivative alleles to B73, followed by self-pollination, resulted in F₂ progeny that segregated lethal hcf mutants at a frequency that approximates 1/16 (Table 2). ² analysis firmly rejects a single genetic factor being responsible for the lethal hcf phenotype and is consistent with two independent recessive mutations (see Table 2). A similar genetic phenomenon had been observed previously with the orange pericarp (orp) mutation, in which two unlinked factors are required to inhibit tryptophan biosynthesis (WRIGHT et al. 1992 ). These unlinked factors encode duplicate genes, one of which is defective in the original inbred line that was mutagenized to recover orp. By analogy, a good candidate for the second locus required for the hcf106 phenotype is hcf106C, due to its near identity in the predicted protein sequence in the B73 inbred line.

The predicted HCF106C protein (Fig 1A) is 87% identical in amino acid sequence to HCF106 and can be immunoprecipitated by the same antibody (Fig 3A). These antibodies detect neither protein in hcf106 mutant seedlings (SETTLES et al. 1997 ), suggesting that the cognate locus is mutated in the genetic background from which hcf106-mum1 was originally isolated. In support of this hypothesis, B73 seedlings express two proteins of very similar size that are recognized by HCF106 antibodies (Fig 3B), while wild-type siblings in the original hcf106 background express a single cross-reacting protein that is absent in mutants (SETTLES et al. 1997 ). The additional protein appears to rescue the lethal hcf phenotype in hcf106-mum1 families crossed into B73 (Fig 3B) even when Mutator elements are unmethylated (Fig 3C).

Additional backcrosses with hcf106-mum1, Bz1, r1 to bz1-mum9, R1 Mutator stocks resulted in a much reduced frequency of F₂ hcf seedlings when bz1-mum9 (bronze-mutable), R1 (colored) kernels were selected (not shown). These results suggested that a kernel color locus might be linked to the second-site suppressor in the hcf106-mum1 background. The R1 locus maps to chromosome 10L 32 map units from the HCF106C locus [ Fig 2B. Burr, Acemaz database (http:/burr.bio.bnl.gov)], consistent with the hypothesis that HCF106C is the second-site suppressor. The genetic factor that suppresses the mutant phenotype cosegregated with an HCF106C RFLP in these lines (Fig 3D).

The hcf106C mutant allele cosegregates with the hcf106 phenotype:
To determine if there is a mutation in the HCF106C locus in mutants that express the hcf106 mutant phenotype, the genomic locus from this background was amplified and sequenced. The hcf106C mutant allele was found to have a 7-bp insertion in the first exon of the predicted coding sequence (Fig 4A). The insertion causes a frameshift and a new predicted ORF of 112 amino acids. A truncated HCF106C protein could potentially be synthesized from a second methionine codon; however, this protein would be truncated for most of the chloroplast-targeting domain, making it very unlikely to be functional (not shown). The 7-bp insertion creates an MspI polymorphism with the normal B73 allele. We took advantage of this polymorphism to determine if hcf106C cosegregates with the lethal hcf106 mutant phenotype. We designed primers that amplified small PCR products, including the insertion, and digested with MspI to distinguish the normal and mutant hcf106C alleles. Bulk segregant analysis with pools of normal and mutant seedlings showed that the mutant hcf106C allele cosegregates with the lethal hcf106 mutant phenotype (Fig 4B). In addition, the hcf106C polymorphism also accounts for the exceptional B73 outcross family described in an earlier study (MARTIENSSEN et al. 1990 ). Pooled progeny from this family are heterozygous for hcf106C, indicating that the original plant had a wild-type copy of HCF106C, and that a single dose is sufficient to produce a viable plant (not shown). We conclude that the hcf106C mutation is required for the lethal hcf106 phenotype.

Interestingly, a second class of pale green seedlings were observed in our F₂ population from hcf106-mum2/B73 outcrosses (14 pale green seedlings/139 plants screened). These seedlings accumulated more chlorophyll than the lethal mutants described above and were viable. This class may include plants that have a single dose of normal HCF106C but are homozygous for hcf106-mum2. Consistent with this hypothesis the pale green seedlings segregate at a frequency that approximates one-eighth (² = 0.76, P = 0.38). This result suggests that HCF106C complements hcf106 in a dose-dependent fashion, with one dose complementing lethality and higher doses complementing the pale green pigmentation.

This dose dependence is further illustrated in plants that have a single dose of wild-type HCF106C and are homozygous for hcf106-mum1. These pale green plants frequently displayed clonal dark green sectors, suggestive of epigenetic suppression of the hcf106-mum1 mutation (Fig 5). In plants that also carried the dominant, suppressible, Lesion-mimic mutant Les28, the pale green sectors had lesions while dark green sectors did not (Fig 5B). This demonstrated that the dark green sectors were caused by epigenetic suppression of hcf106-mum1 and that this activity could be distinguished from the pale green pigmentation due to heterozygosity for HCF106C. This is because Les28 and hcf106-mum1 are coordinately regulated by Mu activity such that both mutant phenotypes are lost in Mu-off sectors (MARTIENSSEN and BARON 1994 ). In this way, we confirmed that epigenetic and second-site suppression of hcf106 could be separated somatically. Note that suppression of hcf106C cannot account for the stripes, as this mutation is caused by a frameshift and not by a transposon. Occasional paler green sectors were also observed in these plants (Fig 5B). As they only arose in Mu-active portions of the leaf, we believe these to be the result of the transposon mutation of the HCF106C wild-type gene, resulting in fully mutant hcf106; hcf106C tissue.

Triple-mutant analysis of hcf106 with tha1:
Contradictory lines of genetic evidence in bacteria and maize have suggested either that the Hcf106-dependent protein-targeting pathway is independent of the Sec-dependent pathway (SANTINI et al. 1998 ) or that there may be some interaction between these pathways (ROY and BARKAN 1998 ). Both of these conclusions were based on mutations in the secY gene, a key subunit of the Sec translocation channel. Due to the pleiotropic nature of SecY mutants, it is difficult to separate direct from indirect effects due to failure to target membrane-bound components of the translocation apparatus itself in secY mutants. SANTINI et al. 1998 showed that a substrate of the pH/Hcf106 pathway was unaffected by temperature-sensitive SecY mutations in E. coli. In contrast, ROY and BARKAN 1998 showed a defect in pH targeting in a null mutant of chloroplast SecY (cpSecY) in maize. However, defects in the targeting of pH pathway substrates in the cpSecY mutant could be a secondary effect of the severe loss of thylakoid membrane or could reflect a requirement for the Sec machinery to assemble components of the pH pathway.

To help resolve these issues we generated hcf106, hcf106C, tha1 triple mutants. In vitro, cpSecA is only required for translocation of Sec-dependent thylakoid proteins, and inhibition or depletion of cpSecA does not affect HCF106-targeted proteins (YUAN et al. 1994 ). In addition, the tha1 mutation does not disrupt the targeting of HCF106-dependent substrates despite the fact that cpSecA accumulates to exceedingly low levels in the mutant (VOELKER and BARKAN 1995 ; ROY and BARKAN 1998). Both tha1 and hcf106 mutant seedlings accumulate low levels of mature, properly targeted substrates of the pathway disrupted by the mutation (VOELKER and BARKAN 1995 ; SETTLES et al. 1997 ). We reasoned that the tha1, hcf106, hcf106C triple mutant should reveal any interaction between these two targeting pathways by an enhancement of both mutant phenotypes. With similar reasoning, ROY and BARKAN 1998 analyzed a tha4, tha1 double mutant with mutations in the pH and Sec pathways, respectively, that revealed an additive phenotype, suggesting that the two pathways are independent. However, the leakiness of the tha4 mutant phenotype precluded a firm conclusion (ROY and BARKAN 1998 ). hcf106 mutants have a more severe lesion of the pH pathway and the triple mutant should be more likely to reveal any interaction.

Outcrosses between tha1 and hcf106 heterozygotes result in triple heterozygous plants (Fig 6A). We screened the selfed progeny of triple heterozygotes molecularly to find the low-frequency (1/64) triple-mutant seedlings. Fig 6B shows protein gel blots with the triple mutant in comparison to tha1 and hcf106 mutant seedlings. The accumulation of both HCF106 and Sec substrates in the triple-mutant seedling was similar to that in the single mutants alone, indicating that there is an additive phenotype between Hcf106 and tha1 mutants. This is analogous to the phenotype of tha4, tha1 double mutants (ROY and BARKAN 1998 ) and provides further support for the notion that there is no direct interaction between the pH/HCF106 and Sec pathways.

Insertions of Mu elements in introns can cause mutant phenotypes that are suppressed in the absence of Mu activity:
While screening for triple mutants of hcf106 and tha1, we noted a novel class of pale green seedlings in the F₂ family. These seedlings were viable and survived through flowering. DNA gel blots showed that the viable pale green seedlings were homozygous for tha1-m1. Because tha1-m1 homozygotes are ordinarily seedling lethal, this indicates that some modification of the tha1 phenotype had occurred after crossing to the hcf106 background. Since the hcf106 background lacks active Mu transposons (i.e., the background is Mu-off) and the tha1 mutant background has active Mu transposons (Mu-on), we determined whether the state of Mu activity differed in the lethal and viable tha1 homozygotes. Southern blots were probed with tha1 sequences to genotype plants used for this analysis (Fig 7, bottom). Mu activity in the same samples was assessed by determining the susceptibility of Mu elements to digestion with the methylation-sensitive enzyme HinfI (Fig 7, top). The lethal class of tha1 homozygotes had hypomethylated Mu1 transposons, while the viable pale green seedlings had hypermethylated Mu1 elements (Fig 7, lanes 13–15). The lethal tha1 seedlings could usually be distinguished from the viable tha1 seedlings by their more severe chlorophyll deficiency at the seedling stage. However, occasionally, one of the more chlorophyll-deficient classes survived to maturity. Analysis of DNA extracted from the adult leaves of one of these viable plants revealed progressive increases in Mu1 methylation from the lowest leaves to the flag leaf (Fig 7, lanes 1–12, top). These results suggest that the severity of the tha1 mutant phenotype is diminished when it is in a Mu-off background, analogous to the epigenetic regulation of other Mu-induced alleles by Mu activity (MARTIENSSEN et al. 1990 ).

If the tha1-m1 mutant phenotype is suppressed in Mu-inactive lines, then the severe phenotype should be recovered when the Mu-off family is reactivated by crossing viable homozygous tha1-m1 plants to a Muactive background (MARTIENSSEN et al. 1990 ). We therefore crossed viable (suppressed), tha1/tha1 homozygous plants to Mu-on, tha1/+ heterozygotes. The tha1/+ heterozygotes were known to be Mu active because their homozygous siblings were seedling lethal and had small revertant sectors. The F₁ seedlings from crosses between the tha1/tha1 suppressed homozygotes and tha1/+ heterozygotes had a high frequency of tha1 mutant seedlings (Table 3). In contrast, self-pollinated progeny of the tha1/tha1 suppressed homozogotes did not show any tha1 mutant seedlings. These results confirm that Mu transposon activity regulates the tha1 locus.

Recessive suppressible Mutator alleles are usually caused by insertions in upstream regions, but interestingly, the tha1-m1 mutation is caused by a Mu1 insertion in an intron (VOELKER et al. 1997 ). Since the hcf106-mum4 allele has a dMuDR insertion in intron 1 (DAS and MARTIENSSEN 1995 ) and is fully viable in a Mu-off background, we suspected that the expression of the hcf106-mum4 mutant phenotype may respond to Mu activity in a similar fashion. To test this hypothesis, we crossed Mu-off, hcf106-mum4 homozygous plants to Mu-on, hcf106-mum1 heterozygotes. The F₁ progeny from some of these crosses segregated hcf seedling lethals at high frequencies (e.g., 12 hcf lethals: 35 normal siblings in one family). Additional families from these crosses also segregated hcfs at lower frequencies with some siblings showing pale green sectors suggestive of a partial reactivation (pooled results shown in Table 3). The hcf106-mum4 parents were marked for Mu activity with the unlinked, suppressible, Les28 mutation, and reactivation was confirmed by monitoring the lesion mimic phenotype (MARTIENSSEN and BARON 1994 ; data not shown). In contrast, self-pollinated hcf106-mum4 heterozygotes gave rise to no mutant seedlings in Mu-off families (DAS and MARTIENSSEN 1995 ). These results confirmed that hcf106-mum4 is also epigenetically regulated by Mu activity and suggest that Mu insertions in introns may frequently cause mutations whose phenotypes are suppressed in the absence of Mu activity.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

HCF106 and HCF106C encode duplicate factors:
The HCF106C gene is a close homolog of HCF106 and can partially complement a defect in HCF106, resulting in pale green plants that are viable and fertile. However, complete complementation was not observed. Plants segregating hcf106C and hcf106-mum1 illustrate the difference in hcf106 and hcf106C function. In hcf106-mum1 plants with a single dose of wild-type hcf106C, Mu-on sectors are pale green in comparison to the Mu-off sectors that have restored wild-type HCF106 function (Fig 6). In contrast, no such differences are observed between plants that are wild-type for HCF106 and mutant for HCF106C (data not shown). These results suggest that the HCF106C protein either does not target thylakoid proteins with the same efficiency as HCF106 or that its expression level in leaf tissue is lower than that of HCF106 and is insufficient to support the same degree of protein translocation. The correlation between the degree of chlorophyll deficiency and the dosage of HCF106C further indicates that HCF106C may be rate limiting for transport in the absence of HCF106 itself.

The HCF106C ORF is 90% identical to HCF106 in nucleotide sequence and 87% identical in amino acid sequence, suggesting that HCF106C and HCF106 diverged in a group of recently duplicated maize genes (J. DOEBLEY and J. HILL, personal communication). GAUT and DOEBLEY 1997 have estimated that this group of genes was duplicated 11.4 mya and contributed to the segmental allotetraploidy observed in Zea mays. Interestingly, the R and B genes, which are linked to HCF106C and HCF106, respectively, are also duplicate genes, but they are much more divergent from each other and are estimated to have duplicated 20.5 mya (GAUT and DOEBLEY 1997 ). The difference in divergence between the R/B loci on the one hand and the HCF106/HCF106C loci on the other could be explained by the segmental allotetraploid model of maize chromosome duplication. According to this model, maize derives from an interspecies hybridization such that two distinct sets of chromosomes were initially tetrasomically inherited but later inherited disomically (GAUT and DOEBLEY 1997 ). The HCF106/HCF106C loci may have become fixed for one of the two diploid ancestral alleles while they were inherited as a single locus in a tetraploid ancestor of maize, while the R/B loci may have maintained the two ancestral forms. After a switch to disomy, the R/B loci would have been divergent while the HCF106/HCF106C loci would have been identical and have since diverged. An alternative explanation is that the HCF106//HCF106C loci might have undergone gene conversion during recent evolution: this notion is supported by the curious observation that the nucleotide sequences of the HCF106 gene pair are more closely related to each other than are the protein sequences. This would not be expected if silent substitutions had accumulated during the tetrasomic phase.

hcf106 and hcf106C do not genetically interact with tha1:
Double mutants between hcf106 and other mutations affecting chloroplast thylakoid targeting were complicated by the presence of the duplicate gene. Furthermore, hcf106-mum1 (MARTIENSSEN et al. 1990 ), tha1-m1 (shown here), and tha4-m1 (M. WALKER and A. BARKAN, unpublished results) are all suppressible when Mu activity is lost. This greatly complicates the construction and interpretation of double mutants. Nonetheless, in the few triple mutants between hcf106, hcf106C, and tha1 that were recovered, the localization of Sec and pH substrates was affected to the same degree as in single mutants, with lesions in either hcf106 or tha1. This result is consistent with a study of tha4, tha1 double mutants (ROY and BARKAN 1998 ) and suggests an additive, nonoverlapping role for the Sec and pH pathways. This conclusion is further supported by competition studies in which substrates for one pathway failed to compete with substrates for the other pathway (CLINE et al. 1993 ) and with antibody inhibition studies using HCF106 and SecY antibodies (MORI et al. 1999 ). The defective transport of pH substrates in secY mutants (ROY and BARKAN 1998 ) is thus most likely explained as a secondary effect, either due to the lack of target membrane in these mutants or because components of the pH translocation machinery itself, such as plant homologs of the TatC gene, require SecY for their insertion into the membrane. These components are presumably taken up via a SecA-independent, but SecY-dependent pathway, accounting for the genetic results. Double mutants between hcf106 and tha4 would be especially interesting, but must await the construction of alleles of tha4 that are no longer subject to suppression.

A heterochromatic model for suppression of transposon insertions in maize:
We report the characterization of two mutations resulting in suppressible phenotypes, tha1 and hcf106-mum4, that have Mu elements inserted in their introns. There is at least one reported case of an intron insertion of Robertson's Mutator that results in a mutant phenotype that depends on Mutator activity, namely, Kn1-mum2 (GREENE et al. 1994 ). However, in this case the mutation was dominant due to the misexpression of the Knotted1 gene. It was postulated that this misexpression resulted from the insertion of Mutator into a negative regulatory sequence that was somehow restored when MuDR transposase protein was absent.

In the case of hcf106-mum4 and tha1-m1, as we show here, the mutant phenotype is also dependent on Mutator activity, but in these cases the mutations are recessive. In all other previously reported recessive suppressible mutations caused by Robertson's Mutator, and in the case of tha4-m1, the transposon is inserted upstream of the coding region (MARTIENSSEN 1996 ; WALKER et al. 1999 ). In one case, that of hcf106-mum1, transcripts from the suppressed allele have been mapped and shown to derive from an outward reading promoter in the end of Mu (BARKAN and MARTIENSSEN 1991 ). Curiously, however, regulation of these transcripts by tissue type and by light is identical to that of wild type (BARKAN and MARTIENSSEN 1991 ). The transposon separates the gene by 1.4 kb from the upstream promoter region of the gene, which also carries a G-box known to be required for light regulation. In two other well-characterized cases, that of Knotted1-r174 and a1-mum2, insertions were significantly upstream of the transcription start and unlikely to give rise to novel functional transcripts (CHOMET et al. 1990 ; LOWE et al. 1992 ). In the case of A1, however, the pattern of anthocyanin accumulation in different tissues was different from wild type as only very pale pigmentation was observed in the aleurone when the allele was epigenetically suppressed (CHOMET et al. 1990 ). Expression in leaves and other vegetative organs was apparently normal.

Insertion in introns by the transposable element Suppressor-mutator (Spm), also known as Enhancer (En), can also cause suppressible mutant phenotypes that change in response to transposon activity. In contrast, when Spm inserts into the promoter of a gene, it controls gene expression in the opposite way to Robertson's Mutator. This is well illustrated by comparing a1-m2 with a1-mum2. These are alleles of the a1 locus that result from insertion of Spm and Mu at precisely the same nucleotide upstream of the a1 gene, which controls kernel and plant pigmentation (MASSON et al. 1987 ; SCHWARZ-SOMMER et al. 1987 ). a1-m2 results in expression of the A1 gene in leaves and kernels only in the presence of fully active Spm elsewhere in the genome (MCCLINTOCK 1965 ). In contrast, expression of A1 in the leaves of a1-mum2 plants is abolished by the presence of MuDR transposase elsewhere in the genome (CHOMET et al. 1990 ). a1-m2 pigmentation is occasionally inherited independently of Spm activity for one or two generations, a phenomenon known as presetting (MCCLINTOCK 1965 ). Presetting has not been observed with Mutator insertions in the few cases that have been examined (MARTIENSSEN and BARON 1994 ).

These observations illustrate the fact that different transposons can have different effects when integrated in the same gene at the same location. However, these effects are difficult to predict. In many ways they are reminiscent of position effect variegation and other phenomena that involve the interaction of heterochromatin with adjacent genes (MCCLINTOCK 1965 ; MARTIENSSEN et al. 1990 ; MARTIENSSEN and RICHARDS 1995 ). One property shared by transposase and chromatin proteins such as those of the Polycomb group is the ability to form complexes that simultaneously bind to many sites (PIRROTTA and RASTELLI 1994 ; RAINA et al. 1998 ). In one case this facilitates transposition from donor to acceptor site; in the other it facilitates chromatin compaction. In both cases, alternate epigenetic states can be set up and propagated for multiple cell and occasionally germinal generations (MARTIENSSEN 1996 ). It is possible then that epigenetic effects of transposons on gene expression are caused by local heterochromatinization, which depends in part on the proximity of other elements that might facilitate complex formation. In the absence of transposase, chromatin remodeling might allow the transposon to avoid detection by the transcriptional apparatus, which can then access regulatory sequences. The point is well illustrated by hcf106-mum2 (Fig 8). In this allele, Mutator insertions in the promoter and the intron are combined (DAS and MARTIENSSEN 1995 ). The mutant phenotype caused by each insertion independently is suppressed in the absence of MuDR transposase (in hcf106-mum1 and hcf106-mum4, respectively). However, in hcf106-mum2 the two insertions together permanently inactivate the gene, even in the absence of transposase. This may result from chromatin interactions that sequester the promoter from flanking regulatory elements even in the absence of transposase (Fig 8). Ten years after she discovered Suppressor-mutator, McClintock made a parallel between transposon-mediated gene control and gene regulation in bacteria (MCCLINTOCK 1965 ). While certainly an example of gene regulation, the mechanism of transposon-regulated suppression may be more closely related to epigenetic phenomena involving chromatin assembly and inheritance. McClintock's earlier work, which drew parallels between transposon regulation and paramutation, X inactivation, position effect variegation, and transvection, may have been nearer to the mark, perhaps reflecting the origin of heterochromatin from transposable elements (MCCLINTOCK 1951 ).

View larger version (26K): In this window In a new window Download PPT slide   Figure 8. Model for Mutator suppression. In hcf106-mum1 and hcf106-mum4, transcription is blocked by transposase that binds to the Mutator transposon, but is relieved in its absence. In hcf106-mum2 the host chromatin complex binds both elements, mimicking the effect of transposase.

	FOOTNOTES

¹ Present address: Horticultural Sciences Department, University of Florida, Gainesville, FL 32611.
² Present address: Department of Psychiatry and Behavioral Sciences, Emory University School of Medicine, Atlanta, GA 30322.

	ACKNOWLEDGMENTS

We thank Ben Burr for mapping the hcf106 and hcf106C loci using recombinant inbred lines. We also thank Tim Mulligan for growing and taking care of the plants, Ann Yonetani for help with antibodies and early genetic experiments and Dick McCombie and Marja Timmermans for help with the manuscript. A.M.S. thanks the members of his thesis committee for their support and advice. This work was supported by grants to R.A.M. from the National Science Foundation (MCB-9220774) and the United States Department of Agriculture National Research Initiative (97-35304-4566) and to A.B. from the National Institutes of Health (R01 GM48179).

Manuscript received November 30, 1999; Accepted for publication September 7, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

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].

BARKAN, A., D. MILES, and W. C. TAYLOR, 1986  Chloroplast gene expression in nuclear photosynthetic mutants of maize. EMBO J. 5:1421-1427[Medline].

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

BOGSCH, E. G., F. SARGENT, N. R. STANLEY, B. C. BERKS, and C. ROBINSON et al., 1998  An essential component of a novel bacterial protein export system with homologues in plastids and mitochondria. J. Biol. Chem. 273:18003-18006[Abstract/Free Full Text].

BURR, B., F. A. BURR, K. H. THOMPSON, M. C. ALBERTSON, and C. W. STUBER, 1988  Gene mapping with recombinant inbreds in maize. Genetics 118:519-526[Abstract/Free Full Text].

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, 1990  Identification of a regulatory transposon that controls the Mutator transposable element system in maize. Genetics 129:261-270[Abstract].

CLINE, K., R. HENRY, C. LI, and J. YUAN, 1993  Multiple pathways for protein transport into or across the thylakoid membrane. EMBO J. 12:4105-4114[Medline].

DALBEY, R. E. and C. ROBINSON, 1999  Protein translocation into and across the bacterial plasma membrane and the plant thylakoid membrane. Trends Biochem. Sci. 24:17-22[Medline].

DAS, L. and R. MARTIENSSEN, 1995  Site-selected transposon mutagenesis at the hcf106 locus in maize. Plant Cell 7:287-294[Abstract].

DOSEFF, A., R. MARTIENSSEN, and V. SUNDARESAN, 1991  Somatic excision of the Mu1 transposable element of maize. Nucleic Acids Res. 19:579-584[Abstract/Free Full Text].

FISK, D. G., M. B. WALKER, and A. BARKAN, 1999  Molecular cloning of the maize gene crp1 reveals similarity between regulators of mitochondrial and chloroplast gene expression. EMBO J. 18:2621-2630[Medline].

GAUT, B. S. and J. F. DOEBLEY, 1997  DNA sequence evidence for the segmental allotetraploid origin of maize. Proc. Natl. Acad. Sci. USA 94:6809-6814[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].

KLIMYUK, V. I., F. PERSELLO-CARTIEAUX, M. HAVAUX, P. CONTARD-DAVID, and D. SCHUENEMANN et al., 1999  A chromodomain protein encoded by the Arabidopsis CAO gene is a plant-specific component of the chloroplast signal recognition particle pathway that is involved in LHCP targeting. Plant Cell 11:87-100[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-O tandem duplication. Genetics 132:813-822[Abstract].

MARTIENSSEN, R. A., 1996 Epigenetic silencing of Mu transposable elements in maize, pp. 593–610 in Epigenetic Mechanisms of Gene Regulation, edited by V. E. A. RUSSO, R. A. MARTIENSSEN and A. RIGGS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

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. and E. J. RICHARDS, 1995  DNA methylation in eukaryotes. Curr. Opin. Genet. Dev. 5:234-242[Medline].

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., R. SUROSKY, J. A. KINGSBURY, and N. V. FEDOROFF, 1987  Genetic and molecular analysis of the Spm-dependent a-m2 alleles of the maize a locus. Genetics 117:117-137[Abstract/Free Full Text].

MCCLINTOCK, B., 1951  Chromosome organization and genic expression. Cold Spring Harbor Symp. Quant. Biol. 16:13-47[Medline].

MCCLINTOCK, B., 1965  The control of gene action in maize. Brookhaven Symp. Biol. 18:162-184.

MILES, D., 1980  Mutants of higher plants: maize. Methods Enzymol. 69:3-23.

MORI, H., E. J. SUMMER, X. MA, and K. CLINE, 1999  Component specificity for the thylakoidal Sec and DeltapH chloroplast protein transport pathways. J. Cell Biol. 146:45-56[Abstract/Free Full Text].

NAKAMURA, Y., 2000  Structural analysis of Arabidopsis thaliana chromosome 5. X. Sequence features of the regions of 3,076,755 bp covered by sixty P¹ and TAC clones. DNA Res. 7:31-63[Abstract].

PIRROTTA, V. and L. RASTELLI, 1994  White gene expression, repressive chromatin domains and homeotic gene regulation in Drosophila. Bioessays 16:549-556[Medline].

RAINA, R., M. SCHLAPPI, B. KARUNANANDAA, A. ELHOFY, and N. FEDOROFF, 1998  Concerted formation of macromolecular Suppressor-mutator transposition complexes. Proc. Natl. Acad. Sci. USA 95:8526-8531[Abstract/Free Full Text].

ROY, L. M. and A. BARKAN, 1998  A SecY homologue is required for the elaboration of the chloroplast thylakoid membrane and for normal chloroplast gene expression. J. Cell Biol. 141:385-395[Abstract/Free Full Text].

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning: A Laboratory Manual, Ed. 2. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SANTINI, C.-L., B. IZE, A. CHANAL, M. MÜLLER, and G. GIODANO et al., 1998  A novel Sec-independent periplasmic protein translocation pathway in Escherichia coli.. EMBO J. 17:101-112[Medline].

SARGENT, F., E. G. BOGSCH, N. R. STANLEY, M. WEXLER, and C. ROBINSON et al., 1998  Overlapping functions of components of a bacterial Sec-independent protein export pathway. EMBO J. 17:3640-3650[Medline].

SCHATZ, G. and B. DOBBERSTEIN, 1996  Common principles of protein translocation across membranes. Science 271:1519-1526[Abstract].

SCHWARZ-SOMMER, Z., N. SHEPHERD, E. TACKE, A. GIERL, and W. ROHDE et al., 1987  Influence of transposable elements on the structure and function of the A1 gene of Z. Mays. EMBO J. 6:287-294[Medline].

SETTLES, A. M. and R. MARTIENSSEN, 1998  Old and new pathways of protein export in chloroplasts and bacteria. Trends Cell Biol. 8:494-501[Medline].

SETTLES, A. M., A. YONNETANI, A. BARON, D. R. BUSH, and K. CLINE et al., 1997  Sec-independent protein translocation by the maize Hcf106 protein. Science 278:1467-1470[Abstract/Free Full Text].

STINSON, J. R., A. J. EISENBERG, R. P. WILLING, D. D. HANSON, and J. P. MASCARENHAS, 1987  Genes expressed in the male gametophyte of flowering plants and their isolation. Plant Physiol. 83:442-447[Abstract/Free Full Text].

VALENT, Q. A., P. A. SCOTTI, S. HIGH, J.-W. L. DE GIER, and G. VON HEIJNE et al., 1998  The Escherichia coli SRP and SecB targeting pathways converge at the translocon. EMBO J. 17:2504-2512[Medline].

VOELKER, R. and A. BARKAN, 1995  Two nuclear mutations disrupt distinct pathways for targeting proteins to the chloroplast thylakoid. EMBO J. 14:3905-3914[Medline].

VOELKER, R., J. MENDEL-HARTVIG, and A. BARKAN, 1997  Transposon-disruption of a maize nuclear gene, tha1, encoding a chloroplast SecA homologue: in vivo role of cp-SecA in thylakoid protein targeting. Genetics 145:467-478[Abstract].

WALKER, M. B., L. M. ROY, E. COLEMAN, R. VOELKER, and A. BARKAN, 1999  The maize tha4 gene functions in sec-independent protein transport in chloroplasts and is related to hcf106, tatA, and tatB.. J. Cell Biol. 147:267-275[Abstract/Free Full Text].

WEINER, J. H., P. T. BILOUS, G. M. SHAW, S. P. LUBITZ, and L. FROST et al., 1998  A novel and ubiquitous system for membrane targeting and secretion of co-factor containing proteins. Cell 93:93-101[Medline].

WRIGHT, A. D., C. A. MOEHLENKAMP, G. H. PERROT, M. G. NEUFFER, and K. C. CONE, 1992  The maize auxotrophic mutant orange pericarp is defective in duplicate genes for tryptophan synthase ß. Plant Cell 4:711-719[Abstract/Free Full Text].

YUAN, J., R. HENRY, M. MCCAFFERY, and K. CLINE, 1994  SecA homolog in protein transport within chloroplasts: evidence for endosymbiont-derived sorting. Science 266:796-798[Abstract/Free Full Text].

This article has been cited by other articles:

				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]

				W. Pooma, C. Gersos, and E. Grotewold Transposon Insertions in the Promoter of the Zea mays a1 Gene Differentially Affect Transcription by the Myb Factors P and C1 Genetics, June 1, 2002; 161(2): 793 - 801. [Abstract] [Full Text] [PDF]

				R. A. Martienssen and V. Colot DNA Methylation and Epigenetic Inheritance in Plants and Filamentous Fungi Science, August 10, 2001; 293(5532): 1070 - 1074. [Abstract] [Full Text] [PDF]

				T. Singer, C. Yordan, and R. A. Martienssen Robertson's Mutator transposons in A. thaliana are regulated by the chromatin-remodeling gene Decrease in DNA Methylation (DDM1) Genes & Dev., March 1, 2001; 15(5): 591 - 602. [Abstract] [Full Text]