Reversion of the Arabidopsis rpn 12 a-1 Exon-Trap Mutation by an Intragenic Suppressor that Weakens the Chimeric 5 ' Splice Site

In the 26S proteasome mutant , an Background: Arabidopsis rpn12a-1 exon-trap T-DNA is inserted 531 base pairs downstream of the STOP RPN12a codon. We have previously shown that this insertion activates a STOP codon-associated latent 5' splice site that competes with the polyadenylation signal during processing of the pre-mRNA. As a result of this dual input from splicing and polyadenylation in the mutant, two transcripts rpn12a-1 RPN12a are produced and they encode the wild-type RPN12a and a chimeric RPN12a-NPTII protein. Both proteins form complexes with other proteasome subunits leading to the formation of wild-type and mutant proteasome versions. The net result of this heterogeneity of proteasome particles is a reduction of total cellular proteasome activity. One of the consequences of reduced proteasomal activity is decreased sensitivity to the major plant hormone cytokinin. We performed ethyl methanesulfonate mutagenesis of and Methods: rpn12a-1 isolated revertants with wild-type cytokinin sensitivity. We describe the isolation and analyses of suppressor of ( Results: rpn12a-1 ). The mutation is intragenic and located at the fifth position of the sor1 sor1 chimeric intron. This mutation weakens the activated 5' splice site associated with the STOP codon and tilts the processing of the mRNA back RPN12a towards polyadenylation. These results validate our earlier interpretation of the unusual Conclusions: nature of the mutation. Furthermore, the data show that optimal 26S rpn12a-1 proteasome activity requires RPN12a accumulation beyond a critical threshold. Finally, this finding reinforces our previous conclusion that proteasome function is critical for the cytokinin-dependent regulation of plant growth. Referee Status:


Abstract
In the 26S proteasome mutant , an Background: Arabidopsis rpn12a-1 exon-trap T-DNA is inserted 531 base pairs downstream of the STOP RPN12a codon. We have previously shown that this insertion activates a STOP codon-associated latent 5' splice site that competes with the polyadenylation signal during processing of the pre-mRNA. As a result of this dual input from splicing and polyadenylation in the mutant, two transcripts rpn12a-1 RPN12a are produced and they encode the wild-type RPN12a and a chimeric RPN12a-NPTII protein. Both proteins form complexes with other proteasome subunits leading to the formation of wild-type and mutant proteasome versions. The net result of this heterogeneity of proteasome particles is a reduction of total cellular proteasome activity. One of the consequences of reduced proteasomal activity is decreased sensitivity to the major plant hormone cytokinin.
We performed ethyl methanesulfonate mutagenesis of and Methods: rpn12a-1 isolated revertants with wild-type cytokinin sensitivity.
We describe the isolation and analyses of suppressor of ( Results: rpn12a-1 ). The mutation is intragenic and located at the fifth position of the sor1 sor1 chimeric intron. This mutation weakens the activated 5' splice site associated with the STOP codon and tilts the processing of the mRNA back RPN12a towards polyadenylation.
These results validate our earlier interpretation of the unusual Conclusions: nature of the mutation. Furthermore, the data show that optimal 26S rpn12a-1 proteasome activity requires RPN12a accumulation beyond a critical threshold. Finally, this finding reinforces our previous conclusion that proteasome function is critical for the cytokinin-dependent regulation of plant growth. The 26S proteasome (26SP) is a multisubunit protease responsible for the degradation of proteins that are covalently labeled with a polyubiquitin (Ub) chain via the combined action of Ub activating enzymes, Ub conjugating enzymes and Ub ligases 1 . The 26SP is localized in the cytosol and the nucleus, and it degrades proteins involved in many signaling and metabolic pathways 1,2 . The 26SP is also essential for the destruction of misfolded proteins that are generated by mistranslations and during stress 2-4 .
Studies with proteasome mutants in Arabidopsis have revealed that the 26SP is required for both male and female gametogenesis, confirming its essential role in plant growth and development 2,5,6 . Partial loss-of-function mutants, on the other hand, have been indispensable for uncovering pathways in which key components are regulated by proteasome-dependent degradation 7-13 .
The rpn12a-1 mutant, which carries an insertion in the RPN12a gene (At1g64520) encoding the regulatory particle non-ATPase subunit (RPN) 12a, was isolated from a collection of exon-trap lines 14,15 . These lines were generated by transforming Arabidopsis plants (C24 accession) with a T-DNA construct that contains a promoterless neomycin phosphotransferase gene (NPTII) without a starting methionine which is preceded by a 3´ splice site of the first intron of the apurinic endonuclease (APR) 14 . Kanamycin-resistant exon-trap lines are therefore predicted to have the APR-NPTII construct inserted downstream of an active promoter either in frame with the coding region or in a position that allows the formation of a novel, chimeric intron. The rpn12a-1 mutation is unusual because the T-DNA is inserted downstream of the RPN12a gene, and both the full-length RPN12a cDNA and a chimeric RPN12a-NPTII cDNA are produced 15 . This suggested that two types of cis signals involved in the pre-mRNA processing of RPN12a are competing. Because the wild-type transcript is produced in the mutant and is stable enough to be detected using routine RNA analytical procedures, the poly(A) signal of the RPN12a gene must be intact and active. On the other hand, since a chimeric RPN12a-NPTII transcript is also produced, the 3´ splice site of the inserted T-DNA must have recruited a latent 5´ splice site in the RPN12a gene. We have previously shown that this predicted latent 5´ splice site is STOP codon-associated, and that the pre-mRNA splicing of the chimeric intron leads to the production of the fusion mRNA 15 . As a result of the action of these two opposing pre-mRNA processing mechanisms, one part of the mRNA species transcribed from the mutant RPN12a gene is translated into a functional RPN12a protein, and the other is translated into a chimeric RPN12a-NPTII fusion protein. Because both RPN12a forms are incorporated into the 26SP, the total proteasome activity in these mutant seedlings is reduced, but not abolished 15 .
The reduction of 26SP activity in rpn12a-1 caused a pleiotropic phenotype, which included altered responses to cytokinins 15 . Cytokinins are plant hormones that are essential for every aspect of growth and development [16][17][18][19] . For example, cytokinins control the development of meristems and vasculature, and play an important role in senescence and nutrient allocation 19,20 . To gain better insight into the cytokinin insensitivity of rpn12a-1 seedlings, we screened for suppressor mutants that have a wild-type cytokinin growth response. Here we describe the intragenic suppressor of rpn12a-1 (sor1) that disrupts the latent 5´ STOP-associated splice site. Sor1 reduced the expression of the RPN12a-NPTII fusion mRNA with a concomitant increase in RPN12a transcript level. As a result, RPN12a accumulation in sor1 seedlings was identical to the wild-type and was accompanied by wild-type cytokinin sensitivity. These results validate our transcript processing interpretation of the rpn12a-1 exon-trap effect and accentuate the importance of optimal RPN12a expression for cytokinin signaling.

Plant material and growth conditions
The Arabidopsis thaliana rpn12a-1 mutant in the C24 background was described by us previously 15 . To grow plants on soil and in axenic cultures, seeds were surface-sterilized in 70% ethanol followed by 50% bleach and plated on MS/2 medium that contained half-strength MS salts (pH 5.7, Sigma, St. Louis, MO) and 1% (w/v) sucrose. The seeds were kept for 4 days in darkness at 4°C, and either plated on MS/2 or on soil (Miracle-Gro potting mix:Perlite at 1:1 ratio). Plants were grown in continuous light at 22°C.

EMS mutagenesis and screening for rpn12a-1 suppressors
The rpn12a-1 seeds were pre-incubated in 1.0% KCl for 12 hours, and then mutagenized for 5 hours in 100 mM sodium phosphate buffer (pH 5) containing 5% DMSO and 80 mM ethyl methanesulfonate (EMS; Sigma-Aldrich, St. Louis, MO). Seeds were washed twice in 100 mM sodium thiosulphate and then twice in distilled water. Seeds were incubated and chilled in 0.1% agar and sown directly to soil. All the seeds in the M2 generation were pooled upon harvest, surface-sterilized and plated on the MS/2 medium containing 0.1 µM kinetin (6-furfurylaminopurine; obtained from Duchefa Biochemie by Gold Biotechnology, St. Louis, MO, USA). The putative suppressor mutants were transferred from the selection medium onto MS/2 medium to allow recovery, and were then transferred to soil.

Phenotypic analyses of sor1
Cytokinin treatments were as previously described 15 . For fresh weight analyses, seedlings were germinated and grown on kanamycincontaining media, and their weight was measured in pools of 5 seedlings after 24 days of growth. Kanamycin monosulfate was obtained from Gold Biotechnology.

Changes from Version 1
We would like to thank the reviewers for their time and comments. In response to Dr. Citovsky's review, we did the qPCR analyses of the RPN12a-related and proteasome subunit gene transcript levels.
The results indeed improved our understanding of the strength of the suppressor mutation and also allowed us to provide a clear answer to the third question posted by Dr. Masson. As suggested, we also included more information about the suppressor screen. We did not specify the number of mutants isolated since many of these lines have not yet been thoroughly analyzed and await confirmation. However, the sor1 mutant described here was the only mutant with a near complete reversion to the wild-type phenotype and this was clarified in the methods section. In reply to the second question by Dr. Masson: although we see its merit, we did not do the suggested experiment with the suppressed RPN12a gene. A new author, Yan Li who conducted the qPCR analyses, has been added to this version.
For immunoblotting analyses, total proteins were isolated, separated and transferred to nitrocellulose membranes as described 15 . Rabbit polyclonal anti-RPN12a and anti-PBA1 antibodies (used at 1:1000 dilution) were purchased from Enzo Life Sciences (Plymouth Meeting, PA, USA). The rabbit, polyclonal anti-NPTII antibodies (used at 1:1000) were obtained from Abcam (Cambridge, MA, USA).
Analyses of the sor1 mutation Genomic DNA fragments from rpn12a-1 and sor1 were amplified using F1 and R2 primers and sequenced using dye-termination chemistry (Perkin-Elmer, Foster City, CA, USA) at Advanced Genetic Technologies Center (AGTC, KY, USA). Sequences were analyzed using Vector NTI Suite (Invitrogen, Carlsbad, CA).

Results and discussion
Isolation of an intragenic rpn12a-1 suppressor To obtain rpn12a-1 suppressors, we mutagenized seeds with EMS and plated ~50,000 M2 seeds on a medium with 0.1 µM of the cytokinin kinetin. Because wild-type plants grown on 0.1 µM kinetin are chlorotic and smaller compared to rpn12a-1 15 , we selected 14-day-old M2 seedlings which were pale green and small as putative suppressors. These putative suppressors were first transferred to cytokinin-free media to recover, and subsequently to soil for self-pollination. We isolated several classes of candidate mutants with varying degrees in rpn12a-1 suppression. However, only one of these mutant lines displayed a near-complete reversion to the wild-type phenotype. Here we describe the molecular analyses of this line that we named suppressor of rpn12a-1 1 (sor1).
Analyses of the M3 generation showed that in suppressor of rpn12a-1 1 (sor1), all visible phenotypes of rpn12a-1 were reverted back to the wild-type (Figure 1). For example, the rpn12a-1 mutant has a smaller rosette than the wild-type and a reduced leaf initiation rate 15 . The sor1 plants had a leaf number and rosette size similar to the C24 wild-type plants ( Figure 1). The sor1 mutant plants also displayed wild-type sensitivity to cytokinin. After three weeks of growth on a medium with 0.2 µM kinetin, both wild-type and sor1 seedlings were chlorotic and their growth was severely inhibited, while the rpn12a-1 seedlings were green and larger ( Figure 1).
Next, we analyzed the kanamycin (Km) resistance of the sor1 mutant line. The Km resistance of the rpn12a-1 mutant is completely linked to the proteasome-related phenotypes and thus, all the progeny of a plant homozygous for the rpn12a-1 mutation should be Km resistant. All sor1 seedlings were indeed resistant to Km, but the levels of resistance were significantly lower compared to rpn12a-1 ( Figure 2). While Km did not affect the growth of rpn12a-1 seedlings, both root and shoot growth of sor1 were partially inhibited ( Figure 2). We did not observe any attenuation of Km resistance over several generations, a phenomenon that has been documented for a number of T-DNA insertion mutant collections 23 (see also the Salk Institute Genomic Analysis Laboratory Arabidopsis sequence indexed T-DNA insertion Project FAQ). An explanation for the change in Km tolerance in sor1 is that the mutation affects the expression of the NPTII gene which is an integral part of the exon-trap (Figure 3a and Babiychuk et al. 1997 14 ). When the sor1 mutant was outcrossed to the C24 wild type, none of the plants of the F2 population displayed an rpn12a-1 phenotype, indicating that sor1 is intragenic and tightly linked with the rpn12a-1 mutation.
sor1 suppresses the accumulation of the RPN12a-NPTII fusion transcript To obtain further insight into the nature of the sor1 mutation, we analyzed the expression of the RPN12a gene and the accumulation of the RPN12a protein. RT-PCR analyses showed that in sor1, the RP-N12aNPTII fusion transcript was not detectable and that the RPN12a cDNA level was comparable to the wild type ( Figure 3b). Quantitative RT-PCR (qPCR) analyses confirmed that there was no statistically significant difference between RPN12a levels in sor1 and the wild Immunoblotting analyses using anti-RPN12a antibodies showed that the sor1 mutant does not accumulate the RPN12a-NPTII fusion protein (Figure 4). The RPN12a abundance in sor1 was increased compared to rpn12a-1 and similar to the wild-type. We were also unable to detect the RPN12a-NPTII fusion in sor1 by using type (Figure 3c). The fusion transcript, which was not detected in the C24 line, was present in the sor1 plants at a ratio of 1:15,000 compared to the rpn12a-1 mutant (relative transcript levels were calculated to be 1.0 ± 1.2 and 15, 856 ± 542 for sor1 and rpn12a-1, respectively).
Reductions in proteasome activity typically lead to the activation of a feedback mechanism that induces the transcription of proteasome subunit genes. This mechanism is operational in all eukaryotes, including yeasts, Drosophila, mammals and plants 7,24-28 . Due to this global feedback up-regulation of 26SP subunit genes, the 20S proteasome subunit β1 (PBA1) and 26SP regulatory particle subunit RPT2a transcripts were more abundant in rpn12a-1 compared to the wild type (Figure 3b and 3c). RT-PCR analyses suggested and qPCR analyses confirmed that the proteasome subunit transcript levels in sor1 were reduced compared to rpn12a-1, but still increased compared to the wild-type (Figure 3b and 3c), indicating that the sor1 mutation did not lead to a complete suppression of the rpn12a-1 mutation. Taking into account both the result of the Km resistance tests ( Figure 2) and the expression data ( Figure 3 and Figure 4), we concluded that the sor1 mutation strongly but incompletely suppresses the formation of the RPN12a-NPTII fusion transcript which was sufficient to restore 26SP function back to the wild-type level.
sor1 weakens the STOP codon-associated 5´ splice site in rpn12a-1 To find the mutation that causes the sor1 phenotype, we amplified and compared the sequences of the RPN12a-NPTII chimeric gene from sor1 and rpn12a-1. No mutations were found in NPTII, indicating that the loss of Km resistance and NPTII abundance was not caused by any disruption of the NPTII coding region. We also did not detect any changes in the RPN12a coding region, but did find a single nucleotide change immediately downstream of the RPN12a STOP codon ( Figure 5). Sequencing of the entire region between RPN12a and NPTII did not reveal any additional mutations, confirming that the RPN12a STOP codon-associated G to A mutation was indeed sor1.
To analyze how this G-to-A substitution leads to reversion of the rpn12a-1 phenotype, we manually compared the consensus sequence for 5´ splice sites in Arabidopsis 29 with the sequence of the exon/intron junction that precedes the RPN12a STOP codon in rpn12a-1 and sor1 (Figure 6a). The alignment revealed that both the intron and exon residues adjoining the splice junction of the mutants match the consensus well. Interestingly, the sor1 mutation changes a consensus G at the fifth position of the intron into an A, thus weakening the 5´ splice site of the chimeric intron. The G at the position +5 is thought to be required for efficient binding of U1snRPN 29 . Reduced splicing of the chimeric intron between the RPN12a and NPTII coding regions is predicted to lead to a reduced accumulation of the RPN12a-NPTII transcript and protein (Figure 6b and 6c). The combination of reduced intron splicing and unaffected 3´ end processing is therefore predicted to lead to a dramatic shift in favor of the formation of the wild-type RPN12a transcript, and thus to the accumulation of the RPN12a protein back to the wild-type level, which is what we observed in sor1 seedlings.

Conclusions
Collectively, the results shown here validate our earlier interpretation of the effects of the rpn12a-1 mutation on RPN12a expression and 26SP function 15 . In the original study, we proposed that the partial loss of 26SP function in rpn12a-1 seedlings is caused by the competition between RPN12a and RPN12a-NPT-II transcript Figure 4. The RPN12a-NPTII fusion protein is absent in the sor1 mutant. Total protein was isolated from two-week-old wild-type (C24), rpn12a-1 and sor1 seedlings and used for immunoblotting analyses with RPN12, NPT and PBA1 antisera. In addition to the RPN12a and RPN12a-NPTII fusion proteins, the anti-RPN12 sera also recognized two proteins (cross) that are not related to RPN12a. Ponceau S-stained membrane showing the large RuBisCO subunit (LSU) is presented as a loading control. The size of the proteins used as molecular mass standards is shown on the right-hand side.
anti-NPTII antisera (Figure 4). In the rpn12a-1 mutant, a fraction of the assembled 26SP contains the fusion protein leading to a decrease in total cellular 26SP activity and a compensatory increase in the expression of proteasome subunit genes 15,28 . In the sor1 mutant, with no or little fusion protein, 26SP function is expected to be restored back to the wild-type level. Indeed, immunoblotting analyses with the anti-PBA1 antibodies showed that the abundance of the 20S proteasome subunit PBA1 in sor1 seedlings was comparable to that of the wild-type, indicating that proteasome activity was restored to optimal levels and that feedback up-regulation of proteasome subunit genes was halted (Figure 4). Figure 5. Sequence alignment of the RPN12a gene (At1g64520) in rpn12a-1 and sor1. Genomic DNA fragment was amplified using F1 and R2 primers (presented in Figure 3), sequenced and the sequence was aligned using Vector NTI suite. Alignment of the region starting with base pair 1615 and ending with base pare 1804 of the annotated RPN12a gene is presented using BoxShade 3.2. The red arrowhead points to the sor1 mutation and the RPN12a STOP codon is boxed in red.
processing that leads to a decrease of RPN12a protein levels and thus, to a decrease in the abundance of wild-type 26SP particles 15 . Our finding that suppression of RPN12a-NPTII accumulation was sufficient to restore RPN12a accumulation and reverse the plant development and cytokinin sensitivity back to the wild-type level validates the proposed interpretation and accentuates the importance of optimal 26SP abundance for Arabidopsis growth and cytokinin regulation 1,2,15,22,30 .
Author contributions JK and JAS designed the experiments, performed all experiments except the qPCR analyses, analyzed the data, and wrote the manuscript. YL performed the qPCR analyses and critically revised the second version of the manuscript.

Competing interests
No relevant competing interests were disclosed. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Referee Report: This manuscript nicely documents the molecular basis for an intragenic suppressor of the rpn12a-1 exon-trap mutation of , which weakens a chimeric 5' splice site that fuses the open Arabidopsis RPN12 reading frame to the coding region in the original mutation. These authors had previously shown NPTII that the T-DNA insertion of results in a competition between 3' splicing of RNA (using a rpn12a-1 RPN12A donor splice site that overlaps with the stop codon and the acceptor splice site upstream of its NPTII coding region of the T-DNA) and its normal polyadenylation. They had suggested that a fraction of the mutant transcripts encoded a non-functional RPN12A-NPTII fusion protein that, upon insertion into the proteasome, altered its activity. Hence, in the original mutant, overall altered proteasome activity resulted in pleiotropic phenotypes associated with cytokinin resistance compared to wild type. In this suppressor line, a point mutation 5 nucleotides within the cryptic intron altered this competing splicing, thereby restoring more efficient polyadenylation and production of enough functional RPN12a protein to restore fully functional proteasome activity. Hence, this analysis confirms the initial interpretation of the source of phenotypes associated with , and documents an interesting example of alteration through rpn12a-1 mutation of a balance between 3' splicing and polyadenylation of a precursor RNA.

Grant information
The design of this work, protocols and results are well presented and justify the conclusions. However, I had a few minor questions on this work: How many suppressors were identified in this analysis? Were other intragenic suppressors identified?
Considering the information provided here, one would suspect that sor1 is a dominant mutation. Is it? If it is, has an experiment been carried out to show that a transgenic copy of the suppressed rpn12a-1 sor rescues the cytokinin-resistance phenotype of rpn12a-1?
Analysis of kanamycin resistance in wild type C24, rpn2a-1 and sor1 seedlings showed that sor1 retains a reasonably high level of resistance compared to wild type (Fig 2). Yet, the molecular characterization described in Figures 3 and 4 shows no evidence of NPTII transcript or protein being produced in this suppressor. Is this a problem of experimental sensitivity? A brief discussion of this observation should be included in the text.

I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed. This is a very nicely executed and clearly written work. The results are clear, and they support the authors' This is a very nicely executed and clearly written work. The results are clear, and they support the authors' conclusions and previously published hypotheses of proteasome involvement in cytokinin response.
One potential enhancement would be to use qPCR to quantify the amount of transcripts, especially since these data represent one of the major findings of the paper.
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed. Competing Interests: