The Activator/Dissociation Transposable Elements Comprise a Two-Component Gene Regulatory Switch That Controls Endogenous Gene Expression in Maize

Abstract

The maize Activator/Dissociation (Ac/Ds) elements are able to replicate and transpose throughout the maize genome. Both elements preferentially insert into gene-rich regions altering the maize genome by creating unstable insertion alleles, stable derivative or excision alleles, or by altering the spatial or temporal regulation of gene expression. Here, we characterize an Ac insertion in the 5′-UTR of the Pink Scutellum1 (Ps1) gene and five Ds derivatives generated through abortive transposition events. Characterization of Ps1 transcription initiation sites in this allelic series revealed several that began within the terminus of the Ac and Ds elements. Transcripts originating within Ds or Ac accumulated to lower levels than the wild-type Ps1 allele, but were often sufficient to rescue the seedling lethal phenotype associated with severe loss-of-function alleles. Transcription initiation sites were similar in Ac and Ds derivatives, suggesting that Ac transposase does not influence transcript initiation site selection. However, we show that Ac transposase can negatively regulate Ps1 transcript accumulation in a subset of Ds-insertion alleles resulting in a severe mutant phenotype. The role of maize transposons in gene evolution is discussed.

THE maize hAT family members Activator/Dissociation (Ac/Ds) are composed of the autonomous Ac and nonautonomous Ds transposable elements (Kunze and Weil 2002). Ac is 4565 bp and encodes a 3.5-kb open reading frame (ORFa) that directs the synthesis of an 807-amino-acid-transposase (TPase) protein (Fedoroff et al. 1983; Kunze et al. 1987). The TPase transcript initiates at several sites within a 100-bp interval at the 5′ end of the element and spans most of Ac (Kunze et al. 1987). The lack of canonical CAAT and TATA boxes in the promoter region may account for the multiple transcriptional initiation sites within Ac and the low levels of Ac transcript (Dynan 1986; Fusswinkel et al. 1991; Fridlender et al. 1998). Ac contains ∼240 bp subterminal repeats and 11 bp terminal inverted repeats at each end. The subterminal repeat regions at both Ac ends contain TPase binding sites that are essential for TPase recognition and subsequent transposition of both Ac and Ds (Muller-Neumann et al. 1984; Pohlman et al. 1984a,b; Coupland et al. 1989; Kunze and Starlinger 1989).

Ds elements are structurally diverse but share the feature that they are capable of nonautonomous transposition (reviewed in Doring and Starlinger 1986; Kunze 1996). Most Ds elements arise from Ac internal deletions that ablate TPase activity. Some Ds elements, such as Ds9, are derived from simple internal deletions of Ac sequence (Pohlman et al. 1984b). More complex structures can also be generated when one Ds element inserts into or near another Ds element (Courage-Tebbe et al. 1983; Klein et al. 1988; Doring et al. 1989; Ralston et al. 1989; Dooner and Belachew 1991). These complex elements are capable of inducing chromosome breakage (English et al. 1995), a phenomenon first discovered by McClintock in 1946 (McClintock 1946; Doring et al. 1989). Ac and Ds are also capable of inducing large-scale structural rearrangements of the genome such as chromosome translocations, inversions, and deletions (Dooner and Belachew 1991; Zhang et al. 2009).

Ac and Ds element insertions can regulate and modify flanking gene expression in a variety of ways (reviewed in Wessler 1988; Feschotte et al. 2002). Ac/Ds elements can alter the RNA splicing patterns (Simon and Starlinger 1987; Wessler et al. 1987; Grotewold et al. 1991) or affect the timing and tissue-specific accumulation of adjacent gene transcripts (Klein et al. 1988; Schiefelbein et al. 1988a; Sullivan et al. 1989; Dowe et al. 1990; Moreno et al. 1992). Vollbrecht et al. (2000) have also described a Ds-insertion allele of knotted1 that shows altered expressivity in the presence of Ac. The mutant phenotype is most enhanced (i.e., more knots on leaves) in the presence of Ac-st1, whereas in the absence of Ac, the mutant phenotype is relatively weak. Ac-st1 is an allele of Ac that accumulates higher levels of ORFa transcript (Brutnell 1995) but is identical in sequence to Ac (J. Messing, unpublished data). Although the precise mechanism by which Ac-st1 mediates the effect on knotted∷Ds is unknown, the interaction between Ds and Ac results in ectopic accumulation of Kn1 transcripts in leaves (Vollbrecht et al. 2000).

To explore the mechanisms by which Ac/Ds may control adjacent gene expression, we characterized an Ac insertion in the ps1 gene, ps1-m8∷Ac [hereafter referred to as ps1(Ac)]. The Ps1 gene encodes lycopene β-cyclase (LCYB), an essential enzyme in β-carotene biosynthesis (Singh et al. 2003). In previous studies, a series of Ac-induced alleles of ps1 were generated through a large-scale Ac regional mutagenesis (Singh et al. 2003; Bai et al. 2007). Among a collection of 24 independent ps1 alleles, all but 1 allele conditioned a seedling lethal phenotype when homozygous due to insufficient accumulation of xanthophylls in seedling tissues (Bai et al. 2007). The one viable insertion allele, ps1(Ac), carries a single Ac insertion in the 5′-untranslated region (UTR) 16 bp upstream of the predicted Ps1 start codon in a 3′ to 5′ orientation relative to Ps1 transcription (Bai et al. 2007). This single insertion allele and the nonautonomous Ds deletion derivatives generated from this allele (Conrad et al. 2007) are the focus of this report.

Quantitative real-time PCR and 5′ rapid amplification of cDNA ends (RACE) techniques were utilized to examine Ps1 transcript pools in mutant and wild-type individuals and to map Ps1 transcription start sites (TSS), respectively. A series of Ds derivatives was generated from ps1(Ac) to examine the contribution of Ac transposase on ps1 expression. The Ps1 TSS in these Ds derivatives were also mapped by 5′ RACE analysis and the ps1 transcript pools among the Ds derivatives quantified. We found that in some Ds alleles, the presence of Ac results in a significantly reduced accumulation of Ps1 product leading to seedling lethality. We discuss the implications of these results in elucidating the molecular mechanisms of transposon-mediated control of endogenous gene expression.

MATERIALS AND METHODS

Plant stocks:

All materials were maintained in the color-converted W22 inbred genetic background (Dooner and Kermicle 1971). Line ps1(Ac), originally defined as ps1-m8∷Ac, was discovered through an Ac regional mutagenesis of the ps1 locus from donor site bti97156∷Ac as previously described (Bai et al. 2007). Ds insertional alleles ps1(Ds1), ps1(Ds2), ps1(Ds3), ps1(Ds4), and ps1(Ds5) were derived from ps1(Ac) through a two-step genetic selection scheme as previously described (Conrad et al. 2007).

To monitor Ac copy number in the maize genome, we used a Ds tester line, r-sc:m3 as previously reported (Brutnell and Dellaporta 1994). This line contains a Ds insertion at the r1 locus that is responsible for anthocyanin production in the aleurone and scutellum tissues. In the absence of Ac, Ds insertions at r1 are stable, which results in a colorless aleurone. In the presence of Ac, Ds excises from r1 and restores gene function producing purple sectors in the aleurone. The timing of Ds excision is delayed with increasing Ac copy number in the genome, resulting in a “negative dosage effect” (McClintock 1951). The Ac-immobilized (Ac-im) element encodes the Ac transposase protein but is incapable of autonomous transposition (Conrad and Brutnell 2005). The Ac-stablized1 (Ac-st1) allele is a derivative Ac line that does not display a negative dosage effect (Chomet 1988).

Phenotypic screens:

Seeds were surface sterilized and placed in a seed germination pouch as described in Bai et al. (2007) or sown directly in pots without treatments. Seedlings were grown in growth chambers under the following conditions: 31° day/22° night, 50% humidity, 12 hr night and 12 hr light of intensity at 500 μmol/m²/sec. Photos were taken 7 days after planting.

Nucleotide sequences:

Sequences for the ps1(Ds) alleles have been deposited in GenBank under identifiers HQ641289–HQ641293.

RACE:

Total RNA was extracted from seedlings harvested at the three-leaf stage using TRIzol reagent following the manufacturer's recommendations (Invitrogen, Carlsbad, CA). Approximately 200 μg total RNA was used to isolate mRNA with the Oligotex mRNA mini kit (Qiagen, Valencia, CA). Approximately 200 ng mRNA was treated using the GeneRacer kit according to the manufacturer's recommendations (Invitrogen). Only capped full-length mRNA pools were captured and used to construct RACE libraries. The first-strand cDNA synthesis was performed using SuperScript III (Invitrogen) with GeneRacer Oligo dT primer (Invitrogen). An RNase H treatment was performed (Invitrogen) followed by two rounds of nested PCR to amplify the cDNA product. The first-round PCR reaction was performed using 0.6 μm GeneRacer 5′ Primer (Invitrogen) and 0.2 μm Ps1 gene-specific primer PS1-38 (5′-GACCTTGGCCTGGTGGAAGACGA-3′) from 1 μl first-strand cDNA template. The nested PCR reaction was performed using 0.2 μm GeneRacer 5′ Nested Primer (Invitrogen) and 0.2 μm Ps1 gene-specific primer PS1-36 (5′-AACTCGTCCACCCACACGCCGTA-3′) with 1 μl first-round PCR product. The PCR reactions consisted of 2.5 units Platinum Taq DNA Polymerase High Fidelity (Invitrogen), 1× Platinum Taq buffer, 0.2 mm dNTP's, 0.5 m betaine, 4% DMSO and 2 mm magnesium sulfate. A touchdown PCR program was used under the following conditions: denaturing at 94° for 2 min, 5 cycles at 94° for 30 sec and 72° for 2 min, 5 cycles of 94° for 30 sec and 70° for 2 min, 25 cycles of 94° for 30 sec, 65° for 30 sec, and 68° for 2 min, with a final extension at 68° for 10 min. RACE products were visualized by gel electrophoresis, subcloned into pCR4-TOPO cloning vector (Invitrogen) and sequenced.

Real-time PCR:

Total RNA was extracted as described above, purified using an RNeasy mini kit, and treated with RNase-free DNase (Qiagen). cDNA libaries were reverse transcribed from purified total RNA using a High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA). TaqMan gene primers and a 3′ minor groove binder (MGB) probe specific to ps1 were used in real-time PCR as previously described (Bai et al. 2009). For each allele, three biological replicates and for each replicate three technical replicates were performed. The relative ps1 gene expression was calculated as previously described (Bai et al. 2009).

RESULTS

Ac insertion in the 5′-UTR directs transcription at the ps1 locus:

To explore the mechanisms of Ac-regulated gene expression, we performed a detailed characterization of the ps1(Ac) allele. Unlike all other Ac-induced ps1 alleles described (Singh et al. 2003; Bai et al. 2007), seedlings homozygous for the ps1(Ac) grew to maturity and set seed. When seeds are planted and grown in seed germination pouches as shown in Figure 1, homozygous ps1(Ac) seedlings display a slightly pale green and pink leaf sheath, due to the accumulation of lycopene in seedling tissues (our unpublished data). When grown in a field nursery, plants homozygous for ps1(Ac) display a weak mutant phenotype including thinner stalks, slower growth, and delayed flowering time, indicating that the Ac insertion at the 5′-UTR disrupts but does not completely abolish PS1 function.

Figure 1.—

Phenotypic and transcriptional analysis of wild-type and ps1(Ac) alleles. (A) Wild-type W22 and (B) homozygous ps1(Ac) seedling phenotypes of 7-day old plants grown in germination pouches. (C) Relative ps1 transcript levels in wild-type and homozygous ps1(Ac) seedlings. Error bars (SE) represent the variation between three biological replicates. Mapped TSS of ps1 transcripts in seedling tissues are shown for (D) wild-type and (E) ps1(Ac). Each arrow denotes a TSS identified as an independent 5′ RACE clone. The 8-bp Ac target site duplication is highlighted in boldface type and underlined. The Ac insertion in ps1(Ac) is shown as a blue triangle with the insertion orientation indicated (5′ or 3′). The 5′ terminal sequences of Ac insertion are in blue. The PS1 start codon is represented in red. Nucleotide positions are relative to this ATG.

To characterize Ps1 transcript profiles in ps1(Ac) homozygous mutant plants, ps1 mRNA pools were quantified using real-time PCR (see materials and methods). Relative gene expression levels of ps1 alleles were calculated and are shown in Figure 1C. Transcripts encoded by ps1(Ac) were reduced to ∼14% of wild-type levels, suggesting that the Ac insertion interfered with Ps1 transcription initiation or transcript stability. Despite the greatly reduced transcript pools, these results indicate that sufficient PS1 protein is able to accumulate to produce viable offspring in self-pollinated homozygous ps1(Ac) mutants.

Rare transcripts initiate within terminus of Ac element:

To further characterize the transcriptional regulation of the ps1(Ac) allele, we performed 5′ rapid amplification of cDNA ends (RACE) (see materials and methods) to map the ps1 TSS. The 5′ RACE products were amplified from both wild-type and homozygous ps1(Ac) mutant tissues. A pool of PCR products was amplified, and 47 independent clones for both the wild-type and mutant allele were sequenced to define the TSS. It should be noted that the same primer sets were used to map the TSS for both Ps1 and ps1(Ac) alleles and were designed to sequence 321 bp downstream of the ATG start codon. Furthermore, 5′ RACE was performed on capped full-length mRNA templates as described in materials and methods. Thus, it is likely that all the TSS mapped reflect genuine TSS and the relative abundance of the products is directly comparable between Ps1 and ps1(Ac) alleles. Transcription initiation sites are shown in Figure 1, D and E.

The majority of wild-type Ps1 transcripts initiate within a window of −49 to −93 bp upstream of the start codon ATG (Figure 1D). The lack of a strong consensus TSS is consistent with the absence of a strong consensus TATA motif as seen in mammalian systems (Carninci et al. 2006) and suggests that other promoter elements are utilized to drive Ps1 transcription. In ps1(Ac), the predominant ps1 transcripts initiate within the Ac terminus 18 nt from the 5′ end of Ac and read out into the flanking ps1 sequence (Figure 1E). These initiation sequences are ∼300 bp from the predominant Ac transposase TSS (ORFa; Kunze et al. 1987), and Ac and ps1 transcription is directed in opposite orientation. Such 5′ Ac read-out promoter activity has not been previously reported in maize, though a similar activity has been described for a Ds insertion in transgenic tomato plants (Rudenko et al. 1994). In addition to the TSS initiating within Ac, several TSS were mapped within the coding regions of ps1(Ac). However, these transcripts are unlikely to encode functional protein products, since they would lack 31 amino acids of the mature protein, including plastid targeting sequences (Singh et al. 2003). In summary, multiple TSS are utilized in both the wild-type and Ac insertion allele, suggesting plasticity in the control of Ps1 expression.

Ds derivatives at ps1-m8:

The Ac-encoded TPase binds to subterminal repeats within Ac and Ds elements to catalyze the transposition reaction (Kunze and Starlinger 1989). Thus, we examined the possibility that the Ps1 promoter activity associated with ps1(Ac) is mediated through Ac binding at end sequences. To test this hypothesis, we examined promoter activity in the absence of TPase expression and in the presence of low and high levels of TPase expression.

To eliminate TPase expression from ps1(Ac), a genetic scheme was used to recover five Ds derivatives from the Ac insertion at ps1(Ac) (Conrad et al. 2007). Four of the five ps1(Ds) alleles were identified as internal deletion derivatives that do not produce functional TPase protein but retain the element at the original genomic location as the donor Ac. A schematic of the Ds allelic structures at the ps1(Ac) locus is shown in supporting information, Figure S1. The ps1(Ds1), ps1(Ds3), ps1(Ds4), and ps1(Ds5) alleles each suffered an internal deletion of varying length: 1177 bp, 775 bp, 455 bp, and 1805 bp, respectively, from a 4565-bp intact Ac. In each of these four ps1(Ds) alleles, the Ds element has intact terminal and subterminal repeats. The ps1(Ds1) allele contains 39 bp, and ps1(Ds3) contains 77 bp of “filler” DNA, respectively, inserted at the deletion junction as shown in Figure S1. The fifth ps1(Ds) allele, ps1(Ds2), is unique in that sequences at the 5′ end of the Ac element including the subterminal and terminal inverted repeats were deleted together with the internal sequence. One base pair of the target site duplication adjacent to this Ds insertion was also deleted. In addition to the 3563-bp deletion in ps1(Ds2), filler DNA was also present at the deletion junction, which originated from both Ac and ps1 5′-UTR sequences that flanked the insertion site (Conrad et al. 2007).

To dissect the function of Ac TPase in regulating ps1 expression, ps1 transcription initiation sites were mapped by 5′ RACE in ps1(Ds) derivatives. The ps1(Ds1) allele was not selected for this assay because it has a similar structure to ps1(Ds3), ps1(Ds4), and ps1(Ds5). Homozygous ps1(Ds2), ps1(Ds3), ps1(Ds4), and ps1(Ds5) seedlings were harvested and mRNA used in 5′ RACE experiments (see materials and methods). The 5′ RACE products were amplified from each cDNA library and a minimum of 12 PCR products were cloned and sequenced for each genotype to map TSS. As shown in Figure 2, multiple TSS were identified for each allele and the profile of each was similar among all Ds derivatives characterized. In particular, the predominant ps1 transcription initiation site in ps1(Ds3), ps1(Ds4), and ps1(Ds5) was within the Ac terminus 18 nt from the 5′ end. Additional TSS were also detected downstream of the ATG in plants carrying these ps1(Ds) alleles as was observed with ps1(Ac). However, it is unlikely that any of these downstream TSS restore functionality as the predicted encoded protein would lack a plastid transit peptide. The finding that the predominant TSS are conserved between these Ds alleles and the Ac allele indicates that Ac transposase is not necessary for TSS selection.

Figure 2.—

Results of 5′ RACE analysis for ps1(Ds) alleles. The Ds insertions are shown as a blue triangle in a 3′ to 5′ orientation. Sequences originating from Ac/Ds are in blue, whereas sequences derived from ps1 are in black. Parentheses represent deletions within Ds elements.

The ps1(Ds2) allele is unique in that it carries ∼1 kb of 3′ Ac end sequences and completely lacks the 5′ Ac end sequences that immediately flank the ps1 gene in ps1(Ac). As the derivative is missing the end of the element it is most similar to fAc elements described previously by Dooner (Ralston et al. 1989). However, we refer to it here as a Ds element as it was derived from Ac and is no longer capable of autonomous transposition. Interestingly, ps1 transcription still initiates within the Ds element in ps1(Ds2) but no consensus sequences were identified and TSS initiated further upstream at sites ∼47–115 bp from the Ps1 start codon. Thus, it appears that 5′ terminal sequences of Ac do influence the TSS selection but are not absolutely required for transcription initiation.

A comparison of the mutant phenotypes conditioned by the various Ds alleles (Figure 3) is consistent with the results of TSS site selection. The ps1(Ds1), ps1(Ds3), ps1(Ds4), and ps1(Ds5) alleles condition similar mutant seedling phenotypes that are not appreciably different from the parental ps1(Ac) allele. Seedlings are very slightly pale green, but grow to maturity in the field and set seed. However, ps1(Ds2) shows a severe mutant phenotype; seedlings are albescent and die soon after germination. In summary, these results suggest that Ac/Ds sequences at the 5′ end of the element promote ps1 gene expression independently of Ac TPase.

Figure 3.—

Seedling phenotypes of homozygous ps1(Ds) alleles. (A) ps1(Ds2), (B) ps1(Ds2) + Ac-im, (C) ps1(Ds2) + Ac-st1, (D) ps1(Ds3), (E) ps1(Ds3) + Ac-im, (F) ps1(Ds3) + Ac-st1, (G) ps1(Ds5), (H) ps1(Ds5) + Ac-im, and (I) ps1(Ds5) + Ac-st1.

High levels of TPase expression in ps1(Ds) alleles:

Although Ac TPase did not noticeably influence the mutant seedling phenotype, plants that carried the Ds insertion alleles [with the exception of ps1(Ds2)] displayed weaker mutant phenotypes in the field than the original ps1(Ac) allele (data not shown). This finding suggests that Ac transposase may negatively regulate gene expression from Ps1 through interactions with Ac or Ds sequences upstream of Ps1. Thus, we reasoned that very high levels of Ac TPase could negatively influence the phenotype of the Ds insertion alleles. To dissect the role that TPase may play in regulating Ps1 gene expression, we examined the influence of Ac on ps1(Ds) allelic expression using the Ac derivatives Ac-immobilized (Ac-im) (Conrad and Brutnell 2005) and Ac-stabilized1 (Ac-st1) (Chomet 1988). The Ac-im element encodes a functional TPase but is incapable of autonomous transposition due to a 10-bp deletion of 5′ end sequences (Conrad and Brutnell 2005). Ac-st1 shows a dosage insensitive Ds-mediated variegation pattern (B. McClintock, unpublished data) that is correlated to a high steady-state level of Ac expression (Vollbrecht et al. 2000) and hypomethylation of the 5′ sequences of the element (Brutnell 1995). The Ac-st1 allele likely produces high levels of TPase protein that delays the developmental timing of the transposition (Heinlein 1996). Thus, we compared the seedling phenotypes and expression of the Ps1 gene in the absence of TPase [ps1(Ds), +/+], with low levels of TPase [ps1(Ds), Ac-im/+] and with high levels of TPase [ps1(Ds), Ac-st1/+].

To introduce TPase into ps1(Ds) derivatives, a two-step genetic strategy was developed. Pollen collected from plants homozygous for ps1(Ds1), ps1(Ds3), ps1(Ds4), and ps1(Ds5) was used to pollinate plants homozygous for Ac-im or Ac-st1 in the field. To create stocks with the ps1(Ds2) allele, crosses could only be performed with heterozygous individuals because homozygous ps1(Ds2) are seedling lethal. Heterozygous progeny carrying the ps1(Ds) mutant alleles and Ac were then backcrossed to plants homozygous for their respective ps1(Ds) mutant alleles. Using this two-step genetic scheme, testcross ears were harvested and kernels homozygous for ps1(Ds1), ps1(Ds3), ps1(Ds4), or ps1(Ds5) alleles were phenotypically identified in approximately one-half of the kernels as they conditioned slightly pink endosperm tissue. Of these kernels, half carried one copy of Ac-im or Ac-st1 (spotted) and half lacked Ac (nonspotted). For the ps1(Ds2) allele, heterozygous plants were used to generate mutant progeny and similar kernel selections were made [e.g., ps1(Ds2) mutant kernels were selected with or without Ac-im and Ac-st1].

To visualize the effects of TPase on ps1(Ds) alleles, homozygous ps1(Ds) mutant kernels exhibiting three different levels of Ac expression (+/+, Ac-im/+, or Ac-st1/+) were grown in growth chambers (see materials and methods). Plants homozygous for ps1(Ds2) display a seedling lethal albescent phenotype (Figure 3A). The introduction of one copy of Ac-im (Figure 3B) or Ac-st1 (Figure 3C) did not noticeably affect the expressivity of the mutant phenotype in the homozygous ps1(Ds2) individuals.

The ps1(Ac) derivative alleles ps1(Ds1), ps1(Ds3), ps1(Ds4), and ps1(Ds5) are very weak mutants and condition similar phenotypes. Homozygous ps1(Ds3) and ps1(Ds5) seedlings were selected as representatives of these four derivative mutants (Figure 3, D and G, respectively). Introduction of a low level of TPase expression through the addition of a single copy of Ac-im did not noticeably influence the mutant seedling phenotypes in either ps1(Ds3) or ps1(Ds5) plants as shown in Figure 3, E and H, respectively. Interestingly, the introduction of high levels of TPase expression by adding a single copy of Ac-st1 resulted in a virescent seedling lethal phenotype [Figure 3, F and I for ps1(Ds3) and ps1(Ds5), respectively] that is similar to the seedling phenotype of lines homozygous for the ps1(Ds2) allele, with the exception that occasional revertant sectors are observed in the presence of Ac-st1 (Figure 3, F and I), which was confirmed by DNA blot analysis (data not shown).

ps1 transcript levels in ps1(Ds) derivatives:

To understand the phenotypic variation found in the ps1(Ds) derivatives, real-time PCR was performed to examine the relative ps1 transcript levels in ps1(Ds2) and ps1(Ds5) genotypes containing different Ac doses (Figure 4). The ps1(Ds5) allele was selected for this assay to represent the four weak ps1(Ds) derivatives. TaqMan real-time PCR primers and probe were used to compare ps1 transcript abundance in seedling tissues (see materials and methods). Relative to the transcript of the ps1(Ac) allele, the ps1(Ds2) +/+, ps1(Ds2) Ac-im/+, and ps1(Ds2) Ac-st1/+ genotypes all showed ps1 transcript levels that were approximately fivefold lower in abundance. The ps1(Ds5) +/+ and ps1(Ds5) Ac-im/+ genotypes accumulated ps1 transcripts at a level similar to that of ps1(Ac). However, ps1 transcript accumulation was substantially lower in ps1(Ds5) Ac-st1/+ mutant tissues relative to ps1(Ds5) +/+, ps1(Ds5) Ac-im/+, or ps1(Ac) seedlings. These results suggest that the presence of Ac-st1 negatively affects the accumulation of the ps1 transcripts.

Figure 4.—

Real-time PCR assay of ps1 relative expression in homozygous ps1(Ac), ps1(Ds2), and ps1(Ds5) seedling tissues. Gene expression is plotted as fold change relative to the ps1(Ac) allele. Error bars (SE) represent the variation between three biological replicates.

TSS in ps1(Ds) alleles containing Tpase:

To understand whether or not additional TPase expression regulates ps1 transcription, ps1 TSS were examined by 5′ RACE in ps1(Ds2) and ps1(Ds5) derivatives in the presence of TPase. Homozygous mutant seedlings of ps1(Ds2) +/+, ps1(Ds2) Ac-im/+, ps1(Ds2) Ac-st1/+, ps1(Ds5) +/+, ps1(Ds5) Ac-im/+, and ps1(Ds5) Ac-st1/+ were harvested and mRNA used in 5′ RACE experiments. TSS were mapped by cloning and sequencing a minimum of 30 PCR products from each genotype. As shown in Figure 5, multiple TSS were detected among all genotypes characterized. Importantly, transcription initiation sites were similar in the presence of TPase expressed from either Ac-im or Ac-st1. Interestingly, several TSS were mapped upstream of the Ds insertions in ps1(Ds5) lines carrying Ac-im and Ac-st1. These minor products represent Ds excision alleles and show that the preferred TSS in revertant alleles are similar to those in the original Ps1 allele, indicating the Ds element displaces but does not eliminate endogenous promoter sequences. Taken together, these findings strongly suggest that Ac-st1 inhibits the expression of the ps1(Ds5) allele by reducing the rates of transcription initiation rather than affecting transcript site selection.

Figure 5.—

Transcription initiation sites of ps1 in homozygous ps1(Ds) derivatives. TSS from at least 30 independent clones were mapped for each allele and are represented as arrows. Other features are as described in Figure 2.

DISCUSSION

Plant transposable elements are capable of regulating flanking gene expression through numerous mechanisms including providing novel TSS (Masson et al. 1987; Schiefelbein et al. 1988a; Barkan and Martienssen 1991; Rudenko et al. 1994; Kiger et al. 1999), introducing novel enhancer elements (Naito et al. 2009), altering the splicing of the genes in which the transposon resides (Wessler et al. 1987; Menssen et al. 1990; Weil and Wessler 1990), and subjecting flanking genes to epigenetic regulation (Chomet et al. 1987; Slotkin and Martienssen 2007).

Here we describe a two-component gene regulatory switch whereby Ds insertions in the promoter of the Ps1 gene bring the gene under the control of Ac. In the presence of an Ac-overexpression allele (i.e., Ac-st1) transcript pools from Ps1 are reduced significantly, resulting in a seedling lethal phenotype. In the absence of Ac or in the presence of a single standard Ac element, expression of the Ps1 gene is sufficient to result in a green seedling that reaches maturity and sets viable seeds. A model describing this interaction is presented in Figure 6. In a wild-type seedling, transcription is initiated from several sites in a window ∼50–90 bp upstream of the ATG (represented by a single arrow in Figure 6A). An Ac insertion in the 5′-UTR displaces the endogenous TSS 4.5 kb upstream of the translation start site. Nevertheless, ps1 transcripts accumulate to ∼14% of wild-type levels and are initiated from sequences within the ends of Ac [relative levels represented by poly(A) transcripts in Figure 6]. Transposase is not required to drive transcription from within the Ds element as insertion alleles can also provide TSS in the absence of Ac (Figure 6C). We propose that a transcription factor complex is recruited to the subterminal repeat region at the Ac/Ds terminus (Figure 6, B and C). The weak promoter activity results in a low abundance of ps1 transcript pools. However, these pools are sufficient for plant survival. In the presence of Ac-st1, excessive Ac transposase binds to the Ds subterminal repeat regions and inhibits recruitment of a transcription factor complex (Figure 6D). Thus, ps1 transcripts are greatly reduced resulting in a severe mutant phenotype. The absence of Ac binding sites at the 5′ end of the element in ps1(Ds2) renders this allele insensitive to this effect. Importantly, the ps1(Ac) and ps1(Ds) alleles represent new regulatory circuits that are controlled through both novel cis (i.e., Ac/Ds end sequence) and trans (i.e., transposase) regulatory components.

Figure 6.—

A model of Ac-mediated regulation of ps1 gene expression. (A) In wild-type tissues, a transcriptional complex binds to promoter sequences and initiates transcription of Ps1. Ps1 transcript abundance is represented by lines with poly(A) tails. (B) In ps1(Ac), a transcriptional complex binds to Ac terminal sequences and initiates transcription from sequences within the Ac element. The ps1 transcript pool is reduced to ∼14% of wild-type levels. Ac TPase is also present in the genome, and the resulting low levels of Ac TPase may interact with terminal Ac end sequences to slightly inhibit transcription. (C) In ps1(Ds) alleles with 5′ terminal Ac/Ds sequences, transcription initiates from the end sequences within the element. ps1 transcript pool abundance is reduced to ∼20%. There is no Ac TPase in the genome. (D) The presence of Ac-st1 in the genome produces high levels of Ac TPase that then compete with the transcriptional complex for binding at Ds terminal sequences. Transcription of the ps1 gene is inhibited. Transcription factor binding sites within ps1 are denoted in blue, whereas transcription factor binding sites from the Ac/Ds terminus are denoted in red. Triangles represent Ac or Ds insertions at the ps1 locus. TF, transcription factor.

Our findings suggest that Ac transposase is capable of suppressing gene expression when Ac or Ds inserts in 5′-UTR sequences of target genes. Similar “suppressible” gene expression systems have been described for the maize Mu and Spm families. For instance, a Mu transposon inserted in 5′-UTR of the hcf106 gene provides transcription initiation sites when Mu is inactive (Barkan and Martienssen 1991). However, transcription from the hcf106 gene is suppressed when an active source of Mu transposase is present elsewhere in the genome (Barkan and Martienssen 1991). A similar phenomenon has been described for Spm (McClintock 1954, 1965). In Spm-suppressible alleles, dSpm insertions reduce but do not completely eliminate the expression of the target gene in the absence of Spm. In the presence of Spm, gene expression is inhibited but dSpm excisions restore gene function (McClintock 1954). The molecular characterization of Spm-suppressible alleles at a2-m1 (Menssen et al. 1990), a1-m1 (Schwarz-Sommer et al. 1985), and bz-m13 (Schiefelbein et al. 1988b) revealed that these insertions were in coding regions of the gene and likely involved the misprocessing of transcripts in the presence of Spm (Gierl et al. 1988). McClintock (1962) also characterized Spm-dependent alleles in which flanking gene expression was dependent on the presence of a functional Spm element. In Spm-dependent alleles of a1, an insertion of the dSpm element in the promoter region drives flanking gene expression in the presence but not in the absence of Spm (Masson et al. 1987). The Kn1-2F11 allele described by Vollbrecht et al. (2000) could be considered an Ac-dependent allele; only in this instance Ac enhances ectopic accumulation of Kn1, resulting in a dominant mutant phenotype.

The Ac-suppressible alleles of ps1 are most similar to the Mu-suppressible alleles at hcf106, in which insertions in the 5′-UTR result in decreased transcript accumulation in the presence of high levels of transposase. In addition, our observations that transcripts are initiated from both within the 5′ end of the Ac/Ds elements and in the flanking regions of the ps1(Ac/Ds) alleles are consistent with the finding that TSS also initiate in the ends of Mutator in the hcf106 allele (Barkan and Martienssen 1991). In both instances flanking gene expression is inhibited in the presence, but not in the absence of the autonomous element. However, Spm-suppressible alleles characterized to date are associated with the insertion of elements in transcribed regions of the gene. Thus, the mechanism of suppression shared by Ac/Ds/Mu appears to differ from the mechanism used by the Spm system.

An interesting contrast between the Mu and Ac/Ds systems is that Mu active lines generally contain a high copy number of Mu insertions. Germinal insertions increase rapidly with each generation (May et al. 2003) and global silencing of all endogenous copies occurs rapidly and frequently (Martienssen and Baron 1994; May et al. 2003). In Ac active lines, copy numbers are generally maintained at low levels and silenced at a relatively low frequency (Brutnell and Dellaporta 1994). It is interesting to speculate that the stronger preference of Ds elements to insert into or near genes relative to Mu elements (Vollbrecht et al. 2010) may provide a mechanism to negatively regulate both Ds copy number and Ac copy number. As Ac dosage increases (i.e., copy number), expression of genes flanking Ds promoter insertions would be subject to negative regulation. In the case of Ps1, this results in a seedling lethal phenotype and the selection for lines that carry reduced copy numbers of Ac and Ds promoter insertions. To examine this potential for Ac suppression more globally, we mapped known Ds insertions relative to the 5′-UTR of annotated genes in the maize genome (Vollbrecht et al. 2010). As shown in Table S1, we identified 114 Ds insertions that are likely to act as Ac-suppressible alleles (3′ to 5′ orientation) and another 161 Ds insertion alleles that may serve as potential Ac-suppressible alleles (5′ to 3′ orientation). Thus, while Mu activity may be targeted directly through gene silencing mechanisms (Slotkin et al. 2005), more indirect regulation of Ac/Ds copy number may occur through the detrimental effect of insertion into or near essential maize genes.

The identification of additional Ac-suppressible alleles could be exploited for functional genomics studies. A program is currently under development to distribute thousands of Ds insertions throughout the maize genome as platforms for regional mutagenesis (Ahern et al. 2009; Vollbrecht et al. 2010). These insertions will be used to generate multiple Ds-insertion alleles at defined target genes. Those that insert in promoter regions could be potential targets for the “fine-tuning” of flanking gene expression by the modulation of Ac expression levels. We are currently testing this concept using the ps1(Ds5) allele to monitor expression of Ps1 in the presence of several derivatives of Ac-st1 that have altered transposase expression levels. The ability to influence flanking gene expression through Ac/Ds also has potential to regulate gene expression in transgenic systems. Several thousand Ds elements have been distributed throughout the rice and Arabidopsis thaliana genomes (Sundaresan et al. 1995; Kolesnik et al. 2004) and those that insert into 5′-UTR sequences in a 3′ to 5′ orientation relative to target gene transcription are good candidates for Ac-suppressible insertion alleles.

Although transcription is initiated from within the end sequences of both Ac/Ds (this study) and Mu (Barkan and Martienssen 1991), it is unlikely that Ac/Ds or Mu end sequences are sufficient to drive high level expression of flanking gene transcription. As noted by Barkan and Martienssen, the ends of Mu do not harbor a canonical TATA motif, but do have sequences similar to Initiator (Inr) elements described in mammalian genes (Smale and Baltimore 1989). TATA motifs are also absent from the ends of Ac/Ds and although a read-out promoter activity has been described for a Ds insertion in tomato (Rudenko et al. 1994), the TSS was ∼300 bp from the 5′ end compared to ∼18 bp from the 5′ end of the Ac/Ds element in this study. This suggests that sequences within the end of the elements function as transcription initiation sites if they are inserted in a suitable context (e.g., near an endogenous gene promoter). A more detailed characterization of additional 5′-UTR insertions should help further delimit the promoter activity of Ac/Ds in driving flanking gene expression.

To examine additional regulatory motifs in the ps1 gene, we scanned the regions surrounding the TSS for signatures of transcription initiation. The multiple transcriptional initiation sites associated with wild-type and mutant ps1 alleles are consistent with the absence of consensus CAAT and TATA boxes in the promoter region (Smale and Kadonaga 2003). In many promoters lacking a TATA element, the Inr functions at the site of transcription initiation (Smale and Baltimore 1989). The Inr motif is contained within the transcription start site and is a strong core promoter element that is functionally analogous to the TATA box (O'Shea-Greenfield and Smale 1992). In yeast, the Inr extends from nucleotides −6 to +11 and has the consensus sequence of Py Py A(+1) N T/A Py Py (Smale 1997). Within the consensus sequence, an A at +1, a T or A at +3, and a C or T at −1 are most critical for determining the strength of an Inr (Smale 1997). In maize, an Inr element has been identified as directing transcription from the TATA-less gene Abp1 (Elrouby and Bureau 2000). In our study, most of the transcripts identified initiated with an A at +1. The predominant transcripts have sequences similar to Inr element, but with mismatch(es).

The wild-type Ps1 TSS lie in a window of −22 to −93 bp upstream of the start codon ATG. The predominant transcript initiation sites within Ac and most Ds insertions are located from −34 bp to +10 relative to the ATG. In ps1(Ds2), the TSS fall in a window from 47 to 115 bp upstream of the ATG. These findings suggest that the proximity of TSS to the ATG is relatively conserved among the ps1 transcripts detected, within a window of −115 to +10 bp relative to the ATG. This may reflect the ability of sequences surrounding the start codon to aid in the recruitment of RNA polymerase II.

Previous studies in Drosophila have suggested that sequences downstream of the Inr help direct transcription initiation sites. One such element, the downstream promoter element (DPE), is highly conserved (A/G G A/T C/T G/A/C) among Drosophila TATA-less promoters and is located at +28 to +32 relative to the start of transcription (Burke and Kadonaga 1996; Kutach and Kadonaga 2000). The DPE is also found in human core promoters and is conserved but not identical to the one described in Drosophila (Burke and Kadonaga 1997). In our study, sequences at +28 to +32 relative to the predominant TSS in wild-type Ps1 allele are identical to the Drosophila DPE motif with a single base pair mismatch. The predominant transcript found within the Ac/Ds end starting at 18 bp from the 5′ end also contains a sequence from +37 to +41 that matches the Drosophila DPE at all but 1 bp. The sequence at +33 to +37 of the transcript with TSS at 105 bp upstream of ATG in ps1(Ds2) is identical to the DPE and sequence at +38 to +42 has all but 1 bp, same as the DPE. Collectively, these results suggest that an Inr-DPE core promoter functions to direct transcription in maize ps1 alleles. It appears that in ps1(Ac/Ds) alleles, the physical distance separating the Inr from the downstream DPE element is variable but similar to the spacing observed between the Inr and DPE elements in both Drosophila and human promoters. Thus, gene regulation in maize may involve similar Inr-DPE modules. Studies to investigate this possibility are now possible with the sequencing of the maize genome (Schnable et al. 2009).