A Halt in Poly(a) Shortening during S-adenosyl-l- Methionine-induced Translation Arrest in Cgs1 Mrna of Arabidopsis Thaliana

Cystathionine γ-synthase (CGS) catalyzes the first committed step of methionine (Met) biosynthesis in plants. Expression of the Arabidopsis thaliana CGS1 gene is negatively feedback-regulated in response to the direct Met metabolite S-adenosyl-L-methionine (AdoMet). This regulation occurs at the step of mRNA stability during translation and is coupled with AdoMet-induced CGS1-specific translation arrest. In general, mRNA decay is initiated by a shortening of the poly(A) tail. However, this process has not been studied in detail in cases where regulatory events, such as programmed translation arrest, are involved. Here, we report that the poly(A) tail of the full-length CGS1 mRNA showed an apparent increase from 50–80 nucleotides (nt) to 140–150 nt after the induction of CGS1 mRNA degradation. This finding was unexpected because mRNAs that are destined for degradation harbored longer poly(A) tail than mRNAs that were not targeted for degradation. The results suggest that poly(A) shortening is inhibited or delayed during AdoMet-induced translation arrest of CGS1 mRNA. We propose an explanation for this phenomenon that remains consistent with the recent model of actively translating mRNA. We also found that CGS1 mRNA degradation intermediates, which are 5'-truncated forms of CGS1 mRNA, had a short poly(A) tail of 10–30 nt. This suggests that poly(A) shortening occurs rapidly on the degradation intermediates. The present study highlights CGS1 mRNA degradation as a useful system to understand the dynamic features of poly(A) shortening.


INTRODUCTION
Studies on the control of gene expression have focused on transcriptional regulation; however posttranscriptional regulation is also important for modulating gene expression (Moore, 2005;Wiederhold and Passmore, 2010).In particular, control of mRNA stability has recently been recognized as an important regulatory process in a number of genetic systems, especially in response to changes in external and intra-cellular conditions.A growing number of reports have shown the involvement of mRNA stability control in plants in response to environmental changes such as temperature, nutrient conditions, osmotic stress, and pathogen infection (Molin et al., 2009;Romero-Santacreu et al., 2009;Castells-Roca et al., 2011;Miller et al., 2011;Rabani et al., 2011;Tanaka et al., 2011;Chiba et al., 2013).
Cystathionine γ-synthase (CGS; EC 2.5.1.48)catalyzes the first committed step of methionine (Met) biosynthesis in plants (Matthews, 1999), and is encoded by the CGS1 gene in Arabidopsis thaliana (Kim and Leustek, 1996;gene ID At3g01120).Unlike many of the regulatory step enzymes in metabolic pathways, CGS is not an allosteric enzyme (Thompson et al., 1982) and its activity is negatively feedback-regulated at the level of mRNA stability in response to S-adenosyl-L-methionine (AdoMet), a direct metabolite of Met.Upon the induction of CGS1 mRNA degradation by AdoMet, 5'-truncated RNA species of CGS1 are formed as degradation intermediates (Chiba et al., 1999(Chiba et al., , 2003;;Haraguchi et al., 2008).
A short amino acid sequence, termed the MTO1 region, 77-RRNCSNIGVAQ-87, encoded in the N-terminal region of CGS1, acts as the cis-element for this regulation (Ominato et al., 2002;Chiba et al., 2003).A. thaliana mto1 mutants that bear single amino acid sequence alterations within the MTO1 region abolish this feedback regulation, resulting in an over-accumulation of soluble Met (Ominato et al., 2002; for review, see Goto et al., 2005).
Translation is required for the AdoMet-induced degradation of CGS1 mRNA (Lambein et al., 2003), which can be recapitulated in the wheat germ extract (WGE) in vitro translation system (Chiba et al., 2003).Studies using WGE revealed that, prior to mRNA degradation, AdoMet induces a temporal arrest of translation elongation at the Ser-94 codon located immediately downstream of the MTO1 region, and that an mRNA degradation event occurs near the 5'-edge of the arrested ribosome (Onouchi et al., 2005).Primer extension studies detected several 5'-ends for the degradation intermediates, which are separated by ~30 nucleotides (nt) from each other (Haraguchi et al., 2008).The two shortest intermediates are also detectable in vivo (Chiba et al., 2003;Fig. 1A).In vitro experiments showed that upon translation arrest, ribosomes are stacked behind the initially arrested ribosome, and suggested that the 5'-ends of the degradation intermediates correspond to these stalled ribosomes (Haraguchi et al., 2008).However, the molecular mechanism of AdoMetinduced CGS1 mRNA degradation remains to be elucidated.
mRNA degradation in eukaryotes has been studied extensively in yeast both genetically and biochemically (for review, see Parker, 2012).It is generally accepted that higher eukaryotes, including plants, share the basic mechanisms of mRNA degradation with yeast.According to the current understanding in higher eukaryotes, mRNA is predominantly degraded by two alternative pathways; a deadenylation-dependent decapping pathway and a 3' to 5' decay pathway.Poly(A) shortening is the first, and usually the rate-limiting, step in both pathways.Deadenylated mRNA is either degraded by 5'-3' exoribonuclease following decapping in the former pathway, or is degraded in a 3' to 5' direction in the latter pathway (for reviews, see Parker and Song, 2004;Chiba and Green, 2009;Houseley and Tollervey, 2009).Poly(A) length plays a critical role in determining the fate of mRNAs due to its effects on basal mRNA turnover as well as on the specific regulation of mRNA degradation and translation (Wiederhold and Passmore, 2010;Weill et al., 2012).Computational simulation of the mRNA degradation process has also suggested that deadenylation is the most effective step for the control of mRNA stability (Cao and Parker, 2001).Therefore, characterizations of changes in the poly(A) length are crucial to understanding the control of mRNA stability.
During the course of in vivo studies on CGS1 mRNA degradation, we noticed an unusual shift in the size of full-length CGS1 mRNA when CGS1 mRNA degradation  Sakurai et al., 2005) and 178 nt 3'-UTR (Kim et al., 1999).The coding region of 563 amino acids from AUG to UGA (Kim and Leustek, 1996) is shown as an open box.AdoMet-induced translation arrest occurs at the Ser-94 codon located immediately downstream of the MTO1 region (filled box).Open arrowheads indicate the 5'-ends of the degradation intermediates that we previously reported in wild-type calli after Met application, which differ in size by ~30 nt (Chiba et al., 2003).(B) Total RNA was extracted from wild-type calli at the indicated time points after application of 1 mM Met.Samples were size-fractionated on a 1.2% (w/v) agarose-formaldehyde gel and subjected to Northern hybridization using a 32 P-labeled CGSΔEx1 probe.The peak positions of the full-length mRNA (FL) before (green) and 4 h after (blue) Met application are marked with circles.The position of degradation intermediates (DI) is marked with an open arrowhead.Similar results were obtained in duplicate experiments.(C) Scanned image of (B) is presented.Positions of mRNA peaks are marked as in (B).Obvious noise peaks due to dirt spots were eliminated (dashed lines).
was induced.In the present study, we specifically focused on changes in poly(A) length, and we found that CGS1 mRNA harbored a longer poly(A) tail after the induction of CGS1 mRNA degradation than before the induction.The finding suggests that CGS1 mRNA is protected from poly(A) shortening during AdoMetinduced translation arrest.The mechanism of this phenomenon is discussed in the light of the recent understanding of actively translating mRNA.

A. thaliana callus cultures and Met application
Liquid callus cultures of wild-type or mto1-1 mutant A. thaliana (ecotype Col-0) were prepared as described (Chiba et al., 1999).Treatment with 1 mM Met (Wako Pure Chemicals) was carried out as described previously (Lambein et al., 2003).Samples were briefly dried on filter paper, frozen in liquid nitrogen and stored at -80°C until RNA extraction.
Total RNA extraction and Northern hybridization Total RNA was extracted from the calli using an RNeasy Plant Mini kit (Qiagen) after homogenization of samples in a Mixer Mill MM300 (Qiagen).Total RNA (10 μg) was separated on agarose-formaldehyde gels prior to transfer to nylon membranes (Gene Screen Plus, NEN Life Science Products).Northern hybridization was carried out as described previously (Chiba et al., 1999).The CGS ΔEx1 probe has been described (Suzuki et al., 2001).Poly(A) selection was not applied to the RNA preparations used in this study, so that RNA species with short poly(A) tails would be retained.

Determination of poly(A) tail length by RNase H digestion
Total RNA (30 μg) was mixed with 125 ng oligodeoxyribonucleotide(s) RnH1500 (5'-CTGACCACAC-CTCCAAAACCAG-3') and/or oligo(dT) 12-18 (Invitrogen) and denatured at 95°C for 5 min, followed by rapid cooling on ice for 5 min.For detection of the poly(A) tail without trimming the 5' region, the RnH1500 oligodeoxyribonucleotide was omitted.In a reaction volume of 75 μl, 7.5 μl 10× RnH buffer (500 mM Tris .HCl pH 8.0, 100 mM MgCl 2 , 1 M NaCl) was added and the samples were incubated at 42°C for 15 min to anneal the oligodeoxyribonucleotide(s) to RNA.After cooling on ice, 12.5 U of E. coli RNase H (Takara) was added and incubated at 37°C for 30 min.RNA samples (10 μg) were separated on agaroseformaldehyde gels and transferred to nylon membranes (Gene Screen Plus, NEN Life Science Products).Northern hybridization was carried out as described (Chiba et al., 1999).

Northern data acquisition and poly(A) length estimation
Hybridization signals were detected with a BAS1000 BioImage Analyzer (Fuji Photo Film) or a Fluoro Imager FLA-7000 (Fuji Photo Film).Digital scanning data of the hybridization signals were obtained using the NIH Image and MultiGauge softwares.The data were normalized to the highest value in each lane to facilitate comparison of peak positions.The lengths of the RNAs were estimated by fitting the size marker positions to a linear line in a semi-log plot by the least-square method, and the poly(A) lengths were estimated by subtracting the mRNA length without poly(A) from that with poly(A).To detect the peak position, a moving average method was applied to smooth the data, when necessary.The "gel smiling" effect was corrected after comparing the peak positions at the right and left ends of each lane.The hybridization image contrast was adjusted using Photoshop software to enhance visibility.

CGS1 mRNA of increased length accumulates when
AdoMet-dependent CGS1 mRNA degradation is induced Although AdoMet is the direct effector of the feedback regulation of CGS1 expression, Met application was used in this study to induce CGS1 mRNA degradation because it is metabolized to AdoMet within the cell and is more efficient than AdoMet for in vivo experiments (Chiba et al., 2003).Met was applied to the liquid callus cultures of A. thaliana, and time-course samples of total RNA were subjected to Northern blot analysis to detect CGS1 mRNA.Effects of Met application was followed for 4 h, because we have previously determined that repression of the full-length CGS1 mRNA accumulation, as well as appearance of CGS1 mRNA degradation intermediates, are evident by 2 h after Met application (Chiba et al., 1999;Lambein et al., 2003).Accumulation of fulllength CGS1 mRNA decreased gradually after Met application.In parallel with this decrease, the fulllength CGS1 mRNA showed an increase in length 2-4 h after Met application (Fig. 1, B and C).
Accumulation of CGS1 mRNA degradation intermediates was evident as early as 0.5 h after Met application, and the mobility of the degradation intermediates did not show any appreciable change during the 4-h experimental period.A faint degradation intermediate signal was also detected at 0 h (immediately prior to Met application), implying that a small fraction of CGS1 mRNA is subject to the feedback regulation even under normal conditions.We have previously reported that, in Met-treated calli, two 5' ends are detected for the degradation intermediates by primer extension experiments, and are thought to be formed at the 5'-edge of the arrested and stacked ribosomes (Chiba et al., 2003;Onouchi et al., 2005;Haraguchi et al., 2008; Fig. 1A).These two degradation intermediates are detected as one band in Northern blots that are designed to detect full-length CGS1 mRNA (~2,000 nt), as they differ in size by only ~30 nt.
The change in the length of full-length CGS1 mRNA after Met application was unexpected.A number of possible mechanisms could lead to an increase in mRNA length including changes in the transcription start site and/or poly(A) addition site, alternative splicing, or changes in poly(A) length.In our previous primer extension experiments with CGS1 mRNA prepared from Mettreated calli, we observed no detectable shift in the position of the full-length mRNA signal (Chiba et al., 2003), which eliminates the possibility of a change in the transcription start site.

Full-length CGS1 mRNA harbors a longer poly(A) tail after Met application
To evaluate the possible contribution of changes in the poly(A) tail length to the increase in full-length CGS1 mRNA size, poly(A) tail lengths were compared before and after Met application.The poly(A) tail was removed by annealing with oligo(dT) followed by RNase H digestion.RNase H makes an endoribonucleolytic cleavage in the RNA moiety of a DNA:RNA duplex and then digests the RNA part exoribonucleolytically in the 3' to 5' direction (Schatz et al., 1990;Rizzo et al., 2002).The poly(A) lengths can then be compared by running samples with and without oligo(dT) on a gel for Northern hybridization (Fig. 2A).
As shown in Fig. 2B, total RNA from wild-type calli before (0 h) and 4 h after Met application showed that the size of the full-length CGS1 mRNA was visibly reduced by the removal of poly(A) tail, suggesting that the full-length CGS1 mRNA harbored a relatively long poly(A) tail.On the other hand, the poly(A) tail removal showed a minimal effect on the mobility of the degradation intermediate band, suggesting that the degradation intermediates harbor short poly(A) tails (Fig. 2B, lanes 3 and 4).
The data also showed that the size of the body of fulllength CGS1 mRNA after removal of the poly(A) tail was not appreciably different between the samples before and after Met application (Fig. 2B, lanes 2 and 4).This result is consistent with a negligible contribution from changes in the transcription start site, poly(A) addition  To make the poly(A) length difference readily detectable, total RNA was annealed with the RnH1500 deoxyoligonucleotide (black bar) that anneals to the 3' region of CGS1 mRNA at nucleotide positions 1,486-1,507, or 449-428 nt upstream of the poly(A) addition site.A portion of the sample was also annealed with oligo(dT) as in Fig. 2A.RNase H digestion followed by Northern hybridization using a probe that hybridizes to the 3'-region will detect a band with a ~430 nt 3'-region plus the poly(A) tail in the "oligo(dT) -" sample, while the ~430-nt 3'-region alone will be detected in the "oligo(dT) +" sample.(B) Poly(A) lengths of CGS1 mRNA in wildtype (lanes 1-4) and mto1-1 mutant (lanes 5-8) calli were determined by RNase H digestion.Total RNA was extracted from the wild-type and mto1-1 mutant calli before (lanes 1, 2, 5, 6) and 4 h after (lanes 3, 4, 7, 8) application of 1 mM Met.Each sample was mixed with the RnH1500 oligodeoxyribonucleotide, divided into two portions with (lanes 2, 4, 6, 8) or without (lanes 1, 3, 5, 7) oligo(dT), followed by RNase H digestion.The samples were sizefractionated on a 3% (w/v) agarose-formaldehyde gel and the CGS1 mRNA 3'-region was detected by Northern hybridization using a 32 P-labeled 3'-UTR probe.Peaks of the bands in the samples from wild-type calli without oligo(dT) before (green) and after (magenta or blue) Met application are marked by brackets.The peaks of the mto1-1 mutant CGS1 mRNA are marked by green brackets.Similar results were obtained in triplicate experiments.Lanes M are size markers.(C, D) Scanned image of (A) is presented after grouping the data with (dashed line) and without (solid line) oligo(dT) for wild-type (C) and mto1-1 mutant (D).Brackets mark the peaks as in (A).3A).Total RNA was extracted from the wild-type calli at the indicated time points after application of 1 mM Met.Each sample was mixed with the RnH1500 oligodeoxyribonucleotide and divided into two portions with (lanes 2, 4, 6, 8, 10) or without (lanes 1, 3, 5, 7, 9) oligo(dT), followed by RNase H digestion.The samples were sizefractionated on a 3% (w/v) agarose-formaldehyde gel and the 3'region of CGS1 mRNA was detected by Northern hybridization using a 32 P-labeled 3'-RnH probe.Brackets mark the peaks of the bands in those samples without oligo(dT) before (green) and after (magenta and blue) Met application.(B) Poly(A) length of UBQ5 mRNA was determined by rehybridizing the membrane used in (A) with a 32 P-labeled UBQ5 probe.Trimming of the 5'-region was omitted because the full-length UBQ5 mRNA is ~700 nt, which is sufficiently short for poly(A) length measurements.(C and D) Scanned images of (A) and (B), respectively, are presented after grouping the data with (dashed line) and without (solid line) oligo(dT) for each time point.In (C), the brackets mark the peaks as in (A).
site and alternative splicing.

Poly(A) length determination of the full-length and degradation intermediates of CGS1 mRNA
To evaluate the poly(A) lengths more precisely, the body of the CGS1 mRNA was shortened by RNase H digestion with a CGS1 mRNA-specific deoxyoligonucleotide that anneals ~430 nt upstream of the poly(A) addition site (Fig. 3A).The resulting bands were then detected by Northern hybridization using a probe that hybridizes to the 3' region of the CGS1 mRNA.
Total RNA prepared from wild-type calli before (0 h) and 4 h after Met application were subjected to poly(A) length determination.For the samples from which the poly(A) tail was removed, Northern hybridization produced a band with a single peak of ~430 nt (Fig. 3B, lanes 2 and 4; Fig. 3C).On the other hand, for the samples from which the poly(A) tail was not removed, broad bands were observed (Fig. 3B, lanes 1 and 3; Fig. 3C).The poly(A) lengths of CGS1 mRNA prior to Met application distributed from 20 to 150 nt with a peak at 50-80 nt (Fig. 3B, lanes 1 and 2; Fig. 3C), whereas two peaks of poly(A) lengths, 10-30 nt and 140-150 nt, were observed 4 h after Met application (Fig. 3B, lanes 3 and 4; Fig. 3C).
The poly(A) tail of 50-80 nt before Met application can be attributed to that of full-length CGS1 mRNA because the degradation intermediates were present only as a minor fraction (Fig. 1B).Four hours after Met application, the full-length CGS1 mRNA was shown to carry a long poly(A) tail while the degradation intermediates carried a short poly(A) tail (Fig. 2B).Therefore, the poly(A) tails of 140-150 nt and 10-30 nt can be attributed to those of full-length and degradation intermediates, respectively, of CGS1 mRNA.mto1 mutation abolishes the changes in poly(A) length after Met application mto1 mutations bear single amino acid sequence alterations within the MTO1 region, and abolish AdoMet-induced CGS1 mRNA degradation (Chiba et al., 1999(Chiba et al., , 2003)).In order to assess the involvement of MTO1 function in poly(A) length shift, total RNA prepared from mto1-1 mutant calli before and after Met application was analyzed (Fig. 3B, lanes 5-8; Fig. 3D).mto1-1 mutation is a Gly-84 to serine substitution and is one of the most severe alleles of mto1 (Ominato et al., 2002).
The distribution of poly(A) lengths in mto1-1 mutant CGS1 mRNA before (0 h) Met application was essentially the same as that observed in wild-type calli before Met application with a peak at 50-80 nt.The distribution did not change appreciably 4 h after Met application.These results indicate that the changes in poly(A) length observed in wild-type CGS1 mRNA after Met application was abolished in mto1-1 mutant CGS1 mRNA, and that the changes in poly(A) length are tightly linked to the negative feedback regulation of CGS1 expression in response to AdoMet, for which the MTO1 region acts as the cis-element.

Time course of poly(A) length shift after Met application
The poly(A) tail of the full-length CGS1 mRNA shifted its length from 50-80 nt to 140-150 nt during the 4 h experimental period following Met application (Fig. 3).To clarify the temporal changes in poly(A) length after Met application, a time course analysis was conducted (Fig. 4).Following Met application, the 50-80 nt peak diminished by 0.5 h, and a poly(A) tail of 10-30 nt emerged.A poly(A) tail of 140-150 nt then gradually accumulated until 2 h after Met application (Fig. 4A, oddnumbered lanes; B).These results can be interpreted as demonstrating that the steady-state poly(A) tail of fulllength CGS1 mRNA under normal conditions is 50-80 nt, which is gradually replaced by the full-length CGS1 mRNA with a poly(A) tail of 140-150 nt following Met application.On the other hand, no change in the length of the 3'-region of the CGS1 mRNA body was detected in the samples from which the poly(A) tail was removed (Fig. 4A even-numbered lanes; C), providing further evidence that the position of the poly(A) addition site was not altered by Met application.
To evaluate the possibility that Met application may have had a general non-specific influence on poly(A) length, the temporal change in UBQ5 mRNA poly(A) was analyzed after Met application (Fig. 4, B and D).The results show that UBQ5 mRNA harbors a poly(A) tail of 30-40 nt, which did not appreciably change in response to Met application.

DISCUSSION
In this study, we showed that the poly(A) tail of fulllength CGS1 mRNA shifted in length from 50-80 nt to 140-150 nt after the induction of CGS1 mRNA degradation by Met application.This result explains the increase in the total length of full-length CGS1 mRNA after Met application (Fig. 1B).The potential contributions of other mechanisms, including alternate transcription start sites and/or poly(A) addition sites, or alternative splicing, were eliminated at least as major factor(s) (Figs. 2 and 4;Chiba et al., 2003).
We have reported previously that the half-life of CGS1 mRNA in wild-type calli shifted from ~150 min to ~80 min after Met application (Lambein et al., 2003).This indicated that CGS1 mRNA degradation is enhanced approximately twofold by feedback regulation.Nevertheless, the poly(A) tail of full-length CGS1 mRNA increased its length in response to Met application.This appeared to be contradictory to the current understanding of poly(A) shortening, as mRNA that is destined for degradation is expected to have a shorter poly(A) tail (for reviews, see Parker and Song, 2004;Chiba and Green, 2009;Houseley and Tollervey, 2009).This raises the question as to how this poly(A) tail length shift in full-length CGS1 mRNA can be explained when feedback regulation is induced and the mRNA is destined to be degraded.
The steady-state length of the poly(A) tail is determined by both the extent of polyadenylation in the nucleus, and the rate of poly(A) shortening in the cytosol.Polyadenylation also occurs in the cytosol (for review, see Richter, 2007).The fact that the poly(A) length of UBQ5 mRNA did not change appreciably in response to Met application (Fig. 4, B and D) argues against the possibility that Met application has a general effect on nuclear and/or cytoso-lic polyadenylation.In addition, the fact that the poly(A) length of mto1-1 mutant CGS1 mRNA did not respond to Met application (Fig. 3) renders a CGS1-specific effect on nuclear and/or cytosolic polyadenylation unlikely.
An alternative scenario is that poly(A) shortening is inhibited or delayed in CGS1 mRNA when it is targeted for AdoMet-induced degradation that is mediated by the MTO1 region.The above-mentioned results on mto1-1 mutant CGS1 mRNA (Fig. 3) provides strong support for this idea.We propose a model to account for the unusual poly(A) length behavior that takes into account the recent understanding of actively translating mRNA (Fig. 5).According to this understanding, mRNAs undergoing  (Tarun and Sachs, 1996;Jacobson, 1996;Kahvejian et al., 2001).(B and C) Poly(A) shortening during normal translation.This mechanism applies to wild-type CGS1 mRNA under a normal AdoMet condition, as well as to the mto1 mutant CGS1 mRNA, irrespective of the AdoMet conditions.(B) Upon termination of translation, eRF1-eRF3 complex releases the completed polypeptide, and the ribosomal subunits are recruited to a new round of translation through the interaction of eRF3 and PABP (Frolova et al., 1994;Zhouravleva et al., 1995;Uchida et al., 2002).(C) PABP occasionally recruits deadenylase to the poly(A) tail to shorten it (Hosoda et al., 2003;Funakoshi et al., 2007), resulting in a steady-state poly(A) tail of 50-80 nt in CGS1 mRNA.(D-F) The fate of wild-type CGS1 mRNA translated under excessive AdoMet levels.(D) Temporal translation arrest occurs at the Ser-94 codon immediately downstream of the MTO1 region.The ribosome is arrested at the step of translocation and the CGS1 nascent peptide adopts a compact conformation (Onouchi et al., 2005;Onoue et al., 2011).Stacking of ribosome(s) behind the arrested ribosome (Haraguchi et al., 2008) is not shown here for simplicity.During translation arrest, the poly(A)-PABP interaction is expected to be static and hence the poly(A) tail is protected from deadenylation, resulting in a long poly(A) tail of 140-150 nt.(E) Translation arrest induces the CGS1 mRNA degradation event near the 5'-edge of the arrested ribosome, mediated presumably by an endoribonuclease (magenta pacman) (Onouchi et al., 2005;Haraguchi et al., 2008), which is yet to be identified.The 5'-part of the degradation product(s) may be degraded rapidly and we have to date been unable to detect this product.(F) The 3'-part of the degradation product(s) has been identified as degradation intermediate(s) both in vivo and in vitro (Chiba et al., 2003).Rapid poly(A) shortening of the degradation intermediate occurs to give a 10-30 nt poly(A) tail.We have not determined whether or not the degradation intermediates bear PABP on it (indicated by a hatched PABP).
active translation form a closed-loop structure in which the poly(A) tail and the 5'-cap are brought into close proximity by specific interactions between the poly(A)-binding protein (PABP) and the cap-binding protein complex, namely eukaryotic initiation factor (eIF) 4G and eIF4E (Tarun and Sachs, 1996;Jacobson, 1996;Kahvejian et al., 2001;Fig. 5A).During normal translation termination, eukaryotic release factor (eRF) 1 recognizes a termination codon and induces the release of a completed polypeptide in a way that is stimulated by eRF3 (Frolova et al., 1994;Zhouravleva et al., 1995).Poly(A) shortening is coupled with the termination of translation through the competitive interaction of eRF3 and deadenylase with PABP in yeast and mammalian cells.PABP exerts a pivotal role during the termination of translation in that PABP interacts, on the one hand, with the eRF1/eRF3 complex to facilitate another round of translation (Uchida et al., 2002), but, on the other hand, it can interact with the deadenylase complex to induce poly(A) shortening (Hosoda et al., 2003;Funakoshi et al., 2007) to give a steady-state poly(A) tail of 50-80 nt in CGS1 mRNA under normal condition (Fig. 5, B and C).
In CGS1 mRNA translated under conditions of excessive AdoMet, translation arrest at the Ser-94 codon prevents ribosomes from reaching the termination codon, thereby preventing poly(A) shortening to occur.Therefore, CGS1 mRNA, with stalled ribosomes on it, is likely to be protected from poly(A) shortening, accounting for the appearance of a long poly(A) tail of 140-150 nt after Met application (Fig. 5D).
The degradation intermediates of CGS1 mRNA had a poly(A) tail of 10-30 nt, suggesting that rapid poly(A) shortening occurs on degradation intermediates (Fig. 5, E and F).We do not yet know the mechanism for this rapid poly(A) shortening; however, we surmise that a disruption of the closed-loop structure is involved in this process.The degradation intermediates of CGS1 mRNA are truncated at the 5'-region and are therefore in a form of decapped RNA.One might expect that decapped RNA would have a very short half-life since it would be readily degraded by 5'-3' exoribonuclease(s).However, these degradation intermediates are fairly stable and we have previously estimated their half-life to be 25-30 min in wild-type calli (Lambein et al., 2003).Since the 5'-ends of each of the CGS1 mRNA degradation intermediates are positioned near the 5'-edge of the stalled ribosome (Onouchi et al., 2005;Haraguchi et al., 2008), it is possible that the stalled ribosome protects the degradation intermediates from 5'-3' exoribonuclease digestion (Fig. 5F).A similar situation is thought to occur in the Bacillus subtilis ermCL gene, which undergoes mRNA degradation coupled with antibiotic-induced translation arrest (Drider et al., 2002).Following a resumption of translation, the degradation intermediates may then enter the general mRNA degradation pathway as decapped mRNAs.
Changes in the poly(A) tail length in response to environmental or intra-cellular stimuli is not well studied in plants, with an exception of heat-shock protein mRNA (Osteryoung et al., 1993).As evidenced by the unique property of poly(A) shortening reported here, studies of the AdoMet-induced MTO1-dependent CGS1 mRNA degradation will advance our understanding of the mechanisms that control mRNA stability, much of which is yet to be unveiled in plants.

Fig. 1 .
Fig. 1.Changes in the length of CGS1 mRNA after Met application.(A) Schematic representation of CGS1 mRNA.The body of the full-length CGS1 mRNA is 1,935 nt and includes a 65 nt 5'untranslated region (UTR) (the RIKEN Arabidopsis full-length cDNA clone, RAFL09-59-O05;Sakurai et al., 2005) and 178 nt 3'-UTR(Kim et al., 1999).The coding region of 563 amino acids from AUG to UGA(Kim and Leustek, 1996) is shown as an open box.AdoMet-induced translation arrest occurs at the Ser-94 codon located immediately downstream of the MTO1 region (filled box).Open arrowheads indicate the 5'-ends of the degradation intermediates that we previously reported in wild-type calli after Met application, which differ in size by ~30 nt(Chiba et al., 2003).(B) Total RNA was extracted from wild-type calli at the indicated time points after application of 1 mM Met.Samples were size-fractionated on a 1.2% (w/v) agarose-formaldehyde gel and subjected to Northern hybridization using a 32 P-labeled CGSΔEx1 probe.The peak positions of the full-length mRNA (FL) before (green) and 4 h after (blue) Met application are marked with circles.The position of degradation intermediates (DI) is marked with an open arrowhead.Similar results were obtained in duplicate experiments.(C) Scanned image of (B) is presented.Positions of mRNA peaks are marked as in (B).Obvious noise peaks due to dirt spots were eliminated (dashed lines).

Fig. 2 .
Fig. 2. Comparison of the poly(A) length of full-length CGS1 mRNA by RNase H digestion. (A) Schematic presentation of a comparison of poly(A) lengths.Total RNA was annealed with oligo(dT) and subjected to RNase H digestion to remove the poly(A) tail.Poly(A) length can be determined by comparing the RNA size with or without the poly(A) tail.In this cartoon, differing poly(A) tail lengths are shown for the full-length mRNA and degradation intermediates to corroborate the experimental data in (B).(B)Total RNA was extracted from wild-type calli before (lanes 1, 2) and 4 h after 1 mM Met application (lanes 3, 4).Each sample was divided into two portions with (lanes 2, 4) or without (lanes 1, 3) oligo(dT), followed by RNase H digestion.The samples were size-fractionated on a 1.2% (w/v) agarose-formaldehyde gel and subjected to Northern hybridization analysis using a 32 Plabeled CGSΔEx1 probe.Brackets mark the full-length (FL) and degradation intermediates (DI) of CGS1 mRNA.Filled circles mark the peak positions of the full-length mRNA without eliminating the poly(A) tail before (green) and 4 h after (blue) Met application.Similar results were obtained in triplicate experiments.(C) Scanned image of (B) is presented after grouping the data with (dashed line) or without (solid line) oligo(dT) for each time point.Marks are the same as in (B).

Fig. 3 .
Fig.3.Poly(A) tail length of CGS1 mRNA before and after Met application in wild-type and mto1-1 mutant calli.(A) Schematic presentation of poly(A) length determination after trimming the 5' region of CGS1 mRNA by RNase H digestion.To make the poly(A) length difference readily detectable, total RNA was annealed with the RnH1500 deoxyoligonucleotide (black bar) that anneals to the 3' region of CGS1 mRNA at nucleotide positions 1,486-1,507, or 449-428 nt upstream of the poly(A) addition site.A portion of the sample was also annealed with oligo(dT) as in Fig.2A.RNase H digestion followed by Northern hybridization using a probe that hybridizes to the 3'-region will detect a band with a ~430 nt 3'-region plus the poly(A) tail in the "oligo(dT) -" sample, while the ~430-nt 3'-region alone will be detected in the "oligo(dT) +" sample.(B) Poly(A) lengths of CGS1 mRNA in wildtype (lanes 1-4) and mto1-1 mutant (lanes 5-8) calli were determined by RNase H digestion.Total RNA was extracted from the wild-type and mto1-1 mutant calli before (lanes 1, 2, 5, 6) and 4 h after (lanes 3, 4, 7, 8) application of 1 mM Met.Each sample was mixed with the RnH1500 oligodeoxyribonucleotide, divided into two portions with (lanes 2, 4, 6, 8) or without (lanes 1, 3, 5, 7) oligo(dT), followed by RNase H digestion.The samples were sizefractionated on a 3% (w/v) agarose-formaldehyde gel and the CGS1 mRNA 3'-region was detected by Northern hybridization using a

Fig. 4 .
Fig. 4. Temporal changes in poly(A) tail lengths after Met application in CGS1 and UBQ5 mRNAs.(A) Poly(A) lengths of CGS1 mRNA were determined by RNase H digestion (Fig.3A).Total RNA was extracted from the wild-type calli at the indicated time points after application of 1 mM Met.Each sample was mixed with the RnH1500 oligodeoxyribonucleotide and divided into two portions with (lanes 2, 4, 6, 8, 10) or without (lanes 1, 3, 5, 7, 9) oligo(dT), followed by RNase H digestion.The samples were sizefractionated on a 3% (w/v) agarose-formaldehyde gel and the 3'region of CGS1 mRNA was detected by Northern hybridization using a 32 P-labeled 3'-RnH probe.Brackets mark the peaks of the bands in those samples without oligo(dT) before (green) and after (magenta and blue) Met application.(B) Poly(A) length of UBQ5 mRNA was determined by rehybridizing the membrane used in (A) with a 32 P-labeled UBQ5 probe.Trimming of the 5'-region was omitted because the full-length UBQ5 mRNA is ~700 nt, which is sufficiently short for poly(A) length measurements.(C and D) Scanned images of (A) and (B), respectively, are presented after grouping the data with (dashed line) and without (solid line) oligo(dT) for each time point.In (C), the brackets mark the peaks as in (A).

Fig. 5 .
Fig. 5. Incorporation of differential poly(A) shortening in CGS1 mRNA into the recent model for mRNA translation.The poly(A) lengths that were estimated in this study are indicated.(A) Actively translating mRNA exists in the form of a closed-loop structure through interactions of PABP and eIF4E-eIF4G cap-binding protein complex(Tarun and Sachs, 1996;Jacobson, 1996;Kahvejian et al., 2001).(B and C) Poly(A) shortening during normal translation.This mechanism applies to wild-type CGS1 mRNA under a normal AdoMet condition, as well as to the mto1 mutant CGS1 mRNA, irrespective of the AdoMet conditions.(B) Upon termination of translation, eRF1-eRF3 complex releases the completed polypeptide, and the ribosomal subunits are recruited to a new round of translation through the interaction of eRF3 and PABP(Frolova et al., 1994;Zhouravleva et al., 1995;Uchida et al., 2002).(C) PABP occasionally recruits deadenylase to the poly(A) tail to shorten it(Hosoda et al., 2003;Funakoshi et al., 2007), resulting in a steady-state poly(A) tail of 50-80 nt in CGS1 mRNA.(D-F) The fate of wild-type CGS1 mRNA translated under excessive AdoMet levels.(D) Temporal translation arrest occurs at the Ser-94 codon immediately downstream of the MTO1 region.The ribosome is arrested at the step of translocation and the CGS1 nascent peptide adopts a compact conformation(Onouchi et al., 2005;Onoue et al., 2011).Stacking of ribosome(s) behind the arrested ribosome(Haraguchi et al., 2008) is not shown here for simplicity.During translation arrest, the poly(A)-PABP interaction is expected to be static and hence the poly(A) tail is protected from deadenylation, resulting in a long poly(A) tail of 140-150 nt.(E) Translation arrest induces the CGS1 mRNA degradation event near the 5'-edge of the arrested ribosome, mediated presumably by an endoribonuclease (magenta pacman)(Onouchi et al., 2005;Haraguchi et al., 2008), which is yet to be identified.The 5'-part of the degradation product(s) may be degraded rapidly and we have to date been unable to detect this product.(F) The 3'-part of the degradation product(s) has been identified as degradation intermediate(s) both in vivo and in vitro(Chiba et al., 2003).Rapid poly(A) shortening of the degradation intermediate occurs to give a 10-30 nt poly(A) tail.We have not determined whether or not the degradation intermediates bear PABP on it (indicated by a hatched PABP).