Alternative Transcription Initiation and the AUG Context ... - Cell Press

RESEARCH ARTICLE

Molecular Plant 7, 989–1005, June 2014

Alternative Transcription Initiation and the AUG Context Configuration Control Dual-Organellar Targeting and Functional Competence of Arabidopsis Lon1 Protease Gerasimos Darasa,2, Stamatis Rigasa,2, Dikran Tsitsekiana, Hadas Zurb, Tamir Tullerc, and Polydefkis Hatzopoulosa,1 a Department of Biotechnology, Agricultural University of Athens, Iera Odos 75, Athens 118 55, Greece b School of Computer Science, Tel Aviv University, Ramat Aviv 69978, Israel c Department of Biomedical Engineering, Faculty of Engineering, Tel Aviv University, Ramat Aviv 69978, Israel

ABSTRACT Cellular homeostasis relies on components of protein quality control including chaperones and proteases. In bacteria and eukaryotic organelles, Lon proteases play a critical role in removing irreparably damaged proteins and thereby preventing the accumulation of deleterious degradation-resistant aggregates. Gene expression, live-cell imaging, immunobiochemical, and functional complementation approaches provide conclusive evidence for Lon1 dual-targeting to chloroplasts and mitochondria. Dual-organellar deposition of Lon1 isoforms depends on both transcriptional regulation and alternative translation initiation via leaky ribosome scanning from the first AUG sequence context that deviates extensively from the optimum Kozak consensus. Organelle-specific Lon1 targeting results in partial complementation of Arabidopsis lon1-1 mutants, whereas full complementation is solely accomplished by dual-organellar targeting. Both the optimal and non-optimal AUG sequence contexts are functional in yeast and facilitate leaky ribosome scanning complementing the pim1 phenotype when the mitochondrial presequence is used. Bioinformatic search identified a limited number of Arabidopsis genes with Lon1-type dual-targeting sequence organization. Lon4, the paralog of Lon1, has an ambiguous presequence likely evolved from the twin presequences of an ancestral Lon1-like gene, generating a single dual-targeted protein isoform. We postulate that Lon1 and its subfunctional paralog Lon4 evolved complementary subsets of transcriptional and posttranscriptional regulatory components responsive to environmental cues for dual-organellar targeting. Key words: alternative transcription; alternative translation; ambiguous and twin presequences; chloroplasts; protein dual-targeting; Lon; mitochondria. Daras G., Rigas S., Tsitsekian D., Zur D., Tuller T., and Hatzopoulos P. (2014). Alternative transcription initiation and the AUG context configuration control dual-organellar targeting and functional competence of Arabidopsis Lon1 protease. Mol. Plant. 7, 989–1005.

Introduction Control of protein function involves an equilibrium between protein synthesis and degradation. Both are crucial components in regulating developmental processes and hormonally triggered morphogenetic events in multicellular organisms. Protein processing machines are present within the cell as a universal protein quality control mechanism to maintain constant protein cycling and homeostasis, necessary for vital biological processes. The 26S proteasome is the major proteolytic mechanism in the cytoplasm and nucleus modulating several aspects of plant development (Smalle and Vierstra, 2004). The ATP-dependent AAA+ proteases that belong to the Lon,

Clp, and FtsH families accomplish protein quality control in plant organelles (Adam et al., 2001; Sinvany-Villalobo et al., 2004; Sakamoto, 2006; Janska et al., 2010). Lon is a ubiquitous proteolytic machine present in unicellular and multicellular organisms (Rigas et al., 2012). To whom correspondence should be addressed. E-mail phat@aua. gr, tel./fax +30-210-5294321. 2 These authors contributed equally to this work. © The Author 2014. Published by the Molecular Plant Shanghai Editorial Office in association with Oxford University Press on behalf of CSPB and IPPE, SIBS, CAS. doi:10.1093/mp/ssu030, Advance Access publication 19 March 2014 Received 10 December 2013; accepted 5 March 2014 1

990

Molecular Plant

Arabidopsis Lon1 dual-organellar targeting

Mutants of PIM1, the yeast Lon gene homolog, are respiratory-deficient due to lesions in mitochondrial genome (Suzuki et al., 1994; van Dyck et al., 1994). Mitochondria of Lon-deficient yeast (Suzuki et al., 1994) and human cells (Bota et al., 2005) have aberrant morphology with electron-dense inclusion bodies in the matrix that most likely represent oxidatively modified and aggregated proteins. In plants, Lon-like proteolysis is induced under oxidative stress to increase mitochondrial durability (Sweetlove et al., 2002; Lister et al., 2004). Lon protein isoforms are encoded by small nuclear gene families predicted to be targeted to distinct organelles. Four genes (Lon1–4) have been identified encoding members of the Arabidopsis Lon family (Sinvany-Villalobo et al., 2004; Janska et al., 2010; Rigas et al., 2012). The degradation of a cytoplasmic male sterility-associated mitochondrial peptide in common bean was designated as the first molecular function of Lon protease in plants (Sarria et al., 1998). Genetic analysis of lon1 mutants revealed that Lon1 is predominantly involved in mitochondrial biogenesis and maintenance of function, critical processes for post-germinative growth leading to seedling establishment (Rigas et al., 2009a, 2009b). Biochemical analysis and proteomics revealed mitochondrial enzyme activity defects disturbing cellular energy metabolism (Rigas et al., 2009a; Solheim et al., 2012). The import of nuclear-encoded proteins translated on cytosolic ribosomes to mitochondria or chloroplasts is organelle-specific, determined by the N-terminal targeting peptide domain. The targeting domain is referred to as transit peptide and mitochondrial presequence when the precursor protein is recognized by the chloroplastic and mitochondrial import apparatus, respectively. Apart from specific protein import, dual-targeting of proteins to mitochondria and chloroplasts has been surprisingly frequent due to their post-endosymbiotic evolution (Millar et al., 2006). Two types of dual-targeting presequences have been described in plants (Peeters and Small, 2001; Silva-Filho, 2003; Mackenzie, 2005). The ambiguous presequence generates a single protein isoform with a targeting peptide domain recognized by the import apparatus of both mitochondria and chloroplasts. The ambiguous presequence can be organized in domains determining targeting specificity to an individual organelle (Berglund et al., 2009). The twin presequences include two distinct targeting domains arranged in tandem at the N-terminus. In eukaryotes, twin presequences can confer dual-targeting to distinct subcellular compartments by employing two alternative in-frame translation initiation codons (Danpure, 1995; Silva-Filho, 2003; Carrie et al., 2009; Carrie and Small, 2013). Both ambiguous and twin presequences amplify the number of protein isoforms in subcellular compartments without affecting the genome size. The majority of dual-targeted proteins in plants contain an ambiguous presequence showing an overall prevalence over twin presequences (Carrie et al., 2009).

Herein, we report that dual-targeting of Arabidopsis Lon1 protease to mitochondria and chloroplasts is determined by twin N-terminal presequences, being primarily regulated at the transcriptional level. The predominant transcript encompassing an AUG residing in an optimal context encodes for the mitochondrial protein isoform. Alternative initiation of transcription under specific conditions generates an additional transcript that through alternative translation initiation and leaky ribosome scanning encodes both chloroplastic and mitochondrial Lon1 isoforms. These two isoforms are necessary for Lon1 functional competence and canonical growth. Our results suggest that post-endosymbiotic gene evolution reveals a trend in protein isoform duplication by dual-targeting in response to developmental or environmental stimuli.

RESULTS Alternative Transcription Initiation of Lon1 Generates Two Differentially Expressed Transcripts In a previous study, the mutants of Arabidopsis Lon1 protease showed growth retardation and impaired seedling establishment (Rigas et al., 2009a). The mutant phenotype was complemented by Lon1 (At5g26860) genomic locus containing also a promoter region of approximately 2000 bp. Translation fusion of the existed cDNA clone U19013 with YFP (Figure 1A), under the control of the CaMV35S promoter, showed exclusive protein targeting to mitochondria. However, meticulous examination of Lon1 5′ genomic sequence revealed an upstream in-frame translation initiation codon (Figure 1A). This N-terminal extension bears 45 amino acid residues corresponding to a putative chloroplast transit peptide followed by the mitochondrial presequence (Figure 1B). An updated search of Arabidopsis cDNA databases revealed the presence of BP826885 EST at position +3 without the two in-frame codons (Figure 1A). To identify Lon1 transcripts containing both translation initiation codons, a comparative qualitative expression analysis was performed. Using F1 forward primer annealing upstream the first translation initiation codon, a long Lon1 transcript, hereinafter referred to as Lon1L, was profound in leaves and in young seedlings grown in dark following seed germination under normal conditions (Figure 2A and Supplemental Table 1). A cross-check analysis using the F2 forward primer, annealing to the sequence context of the second AUG, was performed to monitor total Lon1 mRNA expression pattern. Contrary to Lon1L, gene expression analysis revealed a transcript that was predominantly abundant under a set of combined stress hypoxic-like conditions when seedlings were grown devoid of light in liquid media and under elevated temperature, hereinafter referred to as Lon1S (Figure 2A and Supplemental Table 1). These results strongly suggest the presence of two differentially

Molecular Plant expressed Lon1 transcripts—the long Lon1L and the short Lon1S—which are generated by conditional selection of transcription start sites (TSS). To map dual TSS, one for Lon1L and the other for Lon1S transcripts, the SMART 5′ RACE cDNA amplification method was applied. As Lon1L transcript is highly abundant in response to light regime changes (Figure 2A), the 5′ amplification of Lon1L cDNA ends was performed using template RNA isolated from seedlings grown in darkness


991

after being exposed to light until germination. The analysis using the reverse primer R2 resulted in three major PCR products ranging from approximately 75 to 100 bp in length (Figure 2B). Sequencing analysis showed that the longest in size product corresponded to the most distal TSS of Lon1L located at position –38, whereas the intermediate in size product mapped at –13 most likely representing a truncated 5′-UTR transcript (Figure 1A). The TSS of the lowest in size band mapped at position +3 similar to BP826885

Figure 1 Structure and 5′-Upstream Sequence of Arabidopsis Lon1. (A) The context sequences of the translation initiation codons are indicated in bold italics and the AUGs are underlined. The ORF is depicted with lowercase bold letters. The twin presequences for chloroplast and mitochondrial targeting are colored in green and red, respectively. Putative TATA-elements are enclosed in boxes, whereas potential cis-sequences for transcriptional regulation are indicated. The blue flags depict the experimentally determined 5′-ends of Lon1 cDNAs, whereas the annotation numbers present the identified Lon1 cDNAs. F1 and F2 are the forward primers for gene expression analysis (see Supplemental Table 1). Nucleotide positions are referred to the first translation initiation codon. (B) Computational predicted subcellular targeting of Lon1 twin presequences. The first 45 residues are predicted to represent a chloroplast transit peptide, while the N-terminal portion of residues 46–90 is determined as a mitochondrial presequence. The values range from 0 (no prediction score) to 1 (highest score probability).

992


EST. The 5′-end of Lon1S that is predominantly expressed under hypoxic-like conditions (Figure 2A) was mapped by using the reverse primer R1 and revealed a single PCR product of approximately 320 bp in length with the same TSS as BP826885 EST at position +3 (Figures 1A and 2B). Interestingly, in silico analysis and manual search indicated potential TATA boxes approximately 30 bp upstream from

Molecular Plant the experimentally defined TSSs (Figure 1A). Together, the mapping of TSSs confirms that Lon1 gene expression responses differentially to exogenous cues resulting in distinct transcripts with different 5′ ends. Intriguingly, Lon4 expression resembled the cumulative up-regulation pattern of both Lon1L and Lon1S transcripts (Figure 2A). Nevertheless, the accumulation of Lon4

Figure 2 Differential Selection of Arabidopsis Lon1 Transcription Start Sites Depends on Tissue Specificity and Stress Response. (A) Qualitative comparative expression profile analysis of Lon1L, Lon1, and Lon4 transcripts by RT–PCR. Right panel depicts the position of gene-specific primers. RNA was extracted from: 1, 7-day-old seedlings; 2, stems; 3, leaves; 4, inflorescences; 5, cotyledons; 6, primary roots; 7, seedlings grown in liquid culture at 22°C without agitation for 2 d; 8, seedlings grown in liquid culture at 37°C without agitation for 2 d in dark; 9, seedlings grown in liquid culture at 22°C with 10 μg ml–1 chloramphenicol; 10, dark-grown seedlings on solidified media at 22°C; 11, seedlings grown on solidified media at 22°C recovering for 2 d after UV irradiation. GAPDH gene is the internal control for normalization, while gDNA indicates the genomic DNA amplification product. Reactions with no template RNA reverse transcription (no-RT) were performed to ensure that genomic DNA contamination is eliminated. (B) Mapping of Lon1L and Lon1S TSS. The arrows depict the position of primers utilized for SMART 5′ RACE cDNA amplification. The arrowheads present the PCR products. (C) Quantitative real-time PCR assay to detect the levels of Lon1L versus total Lon1 expression. The RNA was isolated from juvenile plants grown for 3 weeks under normal photoperiod (i), for 2 weeks under normal photoperiod followed by 1 week in darkness (ii), and grown as in (ii) followed by 6 h under constant light (iii). Error bars represent standard deviations of the means. * Indicates significant differences (t-test) of condition (ii) compared with (i) and (iii) (P ≤ 0.05). Further information for primers and techniques is provided (see Supplemental Table 1 and the ‘Methods’ section).

Molecular Plant transcript was similar to Lon1L and substantially different from Lon1S. Consequently, the Lon1L transcript corresponds to a small fraction of Lon1 gene expression supporting its functional role under specific conditions (Supplemental Table 1). These findings raised the issue of quantitative Lon1 gene expression analysis. To experimentally address this issue, the level of Lon1L transcript was assessed in comparison to total Lon1 expression (Supplemental Table 1). The analysis was performed on leaves isolated from juvenile plants grown upon different light–dark regimes and showed that the level of Lon1L transcript was significantly lower compared to Lon1S (Figure 2C). Moreover, the expression of Lon1L was 2.5-fold higher when plants were grown in continuous darkness for a week compared to these grown under normal photoperiod. Upon transition from dark to light, the abundance of Lon1L transcript significantly decreased, whereas Lon1S expression was marginally altered showing conditional regulation of Lon1 transcriptional activity. In silico analysis revealed two CCAAT-boxes in the promoter region at positions –375 and –271, respectively. Recent results support the role of CCAAT-binding heterotrimeric Hap complex in regulating gene expression for adequate response to oxidative stress (Thön et al., 2010). In dark-grown plants, a phytochromeregulated complex could bind to REa (AACCAA) or GATA (AGATAA) elements (Degenhardt and Tobin, 1996), thereby enhancing the expression of Lon1L transcript that encompasses the two in-frame AUGs.

Differential Selection of Lon1 TSS Results in Protein Dual-Targeting to Chloroplasts and Mitochondria As Lon1 expression generates distinct transcripts with initiation codons that, upon translation, could generate discrete N-terminal targeting domains, we tested the dual-organellar targeting of Lon1. Live-cell imaging of tobacco and Arabidopsis cells expressing under the control of the native promoter the YFP-tagged Lon1 gene showed simultaneous protein targeting to both chloroplasts and mitochondria (Figure 3A–3E). However, few chloroplasts accumulated the Lon1 protein in contrast to mitochondria. Given the critical role of Lon1 in mitochondria (Rigas et al., 2009a; Solheim et al., 2012), Lon1 accumulation in certain, but not all, chloroplasts is most likely associated with a conditionally elegant and fine-tuning function. The results corroborate the fundamental role of the first AUG in alternative initiation of translation for the synthesis of the chloroplast isoform. The longer chloroplast protein isoform designated as Lon1L begins from the first AUG codon at position +1, while the shorter mitochondrial isoform designated as Lon1S is translated from the second downstream initiation codon at position +46.


993

Complementary experiments were performed to confirm Lon1 protein accumulation in plant organelles including YFP immunobiochemical analysis of protein subfractions isolated from chloroplasts and mitochondria. To validate that each organelle preparation was free of contaminants from the respective other type, antibodies raised against the isocitrate dehydrogenase (IDH) located in mitochondrial matrix and CYTB6 that is a component of the photosynthetic electron transport chain were used. Immunoblot analysis of the YFP protein tag verified that Lon1 is actually targeted in vivo to both mitochondria and chloroplasts (Figure 3F). Taken together, these approaches provide confidence for Lon1 dual-targeting and accumulation.

Weak-to-Strong Kozak Context Configuration Is a Prerequisite for Lon1 Dual-Targeting To gain further insights into Lon1 dual-targeting, it is important to first understand the mechanism that controls selective translation initiation by the two in-frame codons. To experimentally elucidate this mechanism, a series of N-terminal deletion variants of Lon1 protein fused to YFP and expressed under the control of the CaMV35S promoter were assessed for organelle targeting (Supplemental Table 2). The Lon1LS–YFP construct encompassing the two in-frame AUG codons was simultaneously targeted to both chloroplasts and mitochondria (Figure 4A) in agreement with previously shown results using the native promoter (Figure 3). Therefore, regardless of the promoter sequences, two protein isoforms are synthesized, each one targeting to a different organelle. Leaky ribosome scanning could lead to the synthesis of two distinct protein isoforms when the first AUG context contains a weak Kozak consensus (Kozak, 2005; Wamboldt et al., 2009). In silico analysis showed that the sequence context of the first AUG deviates from the Kozak consensus (Joshi et al., 1997; Kozak, 2005; Kim et al., 2014) potentially allowing ribosomes to sporadically bypass the first and reach the second AUG context, which contains a strong Kozak consensus. To verify this, the first AUG context was substituted with the second optimum Kozak consensus while still retaining in tandem and intact the second wild-type context. The targeting of Lon1SS–YFP construct was restricted to chloroplasts (Figure 4B). Likewise, the AtLon1S[ΔS] and AtLon1L[ΔS] constructs that carry a methionine substitution in the second in-frame AUG initiation codon, irrespectively of the first AUG sequence context, led to exclusive Lon1 targeting to chloroplasts (Supplemental Figure 1). Fidelity of translation initiation depends on both the AUG sequence context and its distance from the 5′ transcript end. These structural features of eukaryotic mRNAs may affect the mechanism of leaky ribosome scanning for alternative translation initiation (Kozak, 2005). To validate

994


Molecular Plant

Figure 3 Lon1 Protein Is Targeted In Vivo to Chloroplasts and Mitochondria under the Control of the Native Promoter. (A) Agrobacterium-mediated transient expression analysis of PLon1:Lon1–YFP in tobacco epidermal cells shows dual-targeting to chloroplasts and mitochondria. Red fluorescence of the images in the upper panel corresponds to chloroplasts due to chlorophyll auto-fluorescence. In images of the lower panel, the mito-Cherry marker is additionally used to identify signal co-localization to mitochondria. (B) Dual-targeting of Lon1 to chloroplasts and mitochondria in stomata cells. (C) Dual-targeting of Lon1 to plastids and mitochondria in root-hair cells. Red fluorescence is due to the mitochondrial-specific dye MitoTracker. (D) Dual-targeting of Lon1 to plastids and mitochondria in mature embryo; boxed area in (D) is magnified in (E). Chloroplasts (or plastids) and mitochondria in images (A) to (E) are depicted by purple and white arrows, respectively. Images shown in (B) to (E) were obtained from stably transformed Arabidopsis lines expressing the PLon1:Lon1–YFP transgene. Bars: 10 μm. (F) YFP immunobiochemical analysis on protein subfractions isolated from chloroplasts (100 μg) and mitochondria (20 μg) reveals accumulation of Lon1 protein in both organelles.

the translation initiation capacity of the second AUG context, the 5′-end sequence of the Lon1L transcript was preserved in the Lon1[ΔL]S–YFP construct except that the first

AUG codon was replaced by an AAA triplet. Numerous observations revealed that fluorescence was restricted to mitochondria (Figure 4C) due to exclusive translation

Molecular Plant


995

Figure 4 Alternative Translation Initiation and Leaky Ribosome Scanning Generate Lon1 Protein Isoforms Targeted to Chloroplasts and Mitochondria. (A) The Lon1LS–YFP is dual-targeted to chloroplasts and mitochondria. (B) The second AUG context functions as a strong translation initiation codon restricting Lon1SS–YFP targeting exclusively to chloroplasts. (C) The Lon1[ΔL]S–YFP construct reveals that translation for subcellular localization to chloroplasts is not initiated by a non-AUG codon residing within the first 45 amino acid residues of the N-terminal extension. Chloroplasts (or plastids) and mitochondria are depicted by purple and white arrows, respectively. (i) Root-hair cells of Arabidopsis stably transformed lines carrying the Lon1 transgenes that are expressed under the control of the CaMV35S promoter. (ii, iii) Agrobacteriummediated transient expression analysis of Lon1 transgenes in tobacco epidermal cells. Red fluorescence in root-hair cells (i) is due to the mitochondrial-specific dye MitoTracker. Red fluorescence in (ii) images corresponds to chloroplasts due to chlorophyll auto-fluorescence, whereas, in (iii) images, the mito-Cherry marker is additionally used to identify signal co-localization to mitochondria. Bars: 10 μm.

initiation from the second AUG context and not from an upstream non-AUG that could lead to dual protein targeting (Christensen et al., 2005; Kozak, 2005; Wamboldt et al., 2009; Simpson et al., 2010). Given that this AUG context

also initiated translation successfully from the short 5′ UTR of Lon1SS, the results support that the second AUG context represents a bona fide strong Kozak consensus sequence regardless of the 5′ UTR length of the Lon1 transcript. Taken

996

Molecular Plant


together, Lon1 dual-targeting specificity is determined via twin presequences and hence via weak-to-strong Kozak context configuration.

Both Lon1 Isoforms Are Required for Functional Complementation of lon1-1 Mutant To elucidate the biological significance of Lon1 dualtargeting, lon1-1 mutant plants were transformed with constructs targeting the Lon1 protein to either or both organelles. At least three independent transgenic lines of the T3 generation were analyzed for each construct. Mitochondria deposition established by 35S:Lon1S resulted in partial complementation of lon1-1 mutant, as recorded by primary root and shoot length (Figure 5A and 5B), albeit full complementation has been accomplished by the Lon1 genomic locus (Rigas et al., 2009a). Comparatively, chloroplast deposition by 35S:Lon1SS resulted in noticeable complementation. This observation reveals the subtle and fine-tuning function of Lon1 in chloroplasts under normal conditions. Dual-targeting of Lon1 to both organelles by 35S:Lon1LS fully complemented the developmental defects of lon1-1 mutant in line with the previously reported successful complementation by the Lon1 genomic locus (Rigas et al., 2009a). Together, the results demonstrate that the two Lon1 protein isoforms targeted to both mitochondria and chloroplasts are necessary for functional complementation of lon1-1 mutant. To confirm that lon1-1 phenotype restoration is attributed to adequate expression and protein accumulation in different compartments of the transgenes, the level of Lon1 expression was analyzed in the hemi- and full-complemented revertant lines. The expression analysis revealed that the levels of Lon1 transcripts were similar between all independent transgenic lines (Figure 5C). Even though Lon1 expression in transgenic lines harboring the 35S:Lon1LS construct was higher compared to the wild-type plants, intriguingly the plants were identical. Complementary to the transgene expression, the accumulation of Lon1 in single or both organelles was investigated by FLAG epitope tag immunodetection analysis. The FLAG-tag was introduced to the C-terminal of the constructs (Supplemental Table 2) to clarify the organellar-specific location by Western blot analysis of proteins isolated from percoll-purified intact chloroplast and mitochondria subfractions (Figure 5D) showing exclusive Lon1 accumulation in mitochondria (35S:Lon1S) or chloroplasts (35S:Lon1SS). The 35S:Lon1LS construct, however, led to Lon1 accumulation to both organelles. The results show that, regardless of the mRNA level or protein isoform abundance in each organelle, the targeting specificity of Lon1 is the most critical parameter that determines the efficiency level of lon1 mutant complementation.

Arabidopsis Lon1 AUG Sequence Contexts Cause Leaky Ribosome Scanning in Yeast As previously shown by complementation assays on nonfermentable carbon sources, such as glycerol, Lon1 targeting to mitochondria restored the respiratory-deficient phenotype of Δpim1 yeast cells (Rigas et al., 2009a). To assess whether Arabidopsis Lon1 twin presequences could complement the yeast mitochondrial respiratory incompetence, several constructs were tested. Both AtLon1LS construct containing the wild-type configuration of two AUGs and AtLon1SS having in tandem the second wild-type context successfully complemented the yeast phenotype (Figure 6A). However, non-complementary results were obtained using AtLon1L[ΔS] and AtLon1S[ΔS] constructs in which the AUG initiating the synthesis of mitochondrial Lon1 isoform was deleted (Figure 6A). Transient expression analysis in tobacco epidermal cells revealed that these two constructs led to exclusive Lon1 targeting to chloroplasts (Supplemental Figure 1). The heterologous expression analysis in yeast transformants showed that the AtLon1 gene was expressed in Saccharomyces cerevisiae strains (Figure 6B). These results demonstrate that, in yeast, leaky ribosome scanning occurs around the non-optimal and the optimal AUG sequence contexts derived from Arabidopsis Lon1 gene. The two plant AUG sequence contexts deviate extensively from the consensus sequence around the AUG initiation codons in S. cerevisiae mRNAs (Hamilton et al., 1987). Taken together, the Arabidopsis Lon1 chloroplast transit peptide is not recognized by the yeast mitochondrial import apparatus corroborating the results of exclusive chloroplast targeting (Supplemental Figure 1).

Twin Presequences for Dual-Targeting Are Evident but Rare in Arabidopsis The information gained from Lon1 gene configuration was impetus for bioinformatics analysis of the Arabidopsis genome to identify candidate genes with twin presequences for dual-organellar targeting. The computer-based algorithm for the analysis was designed relying on the subsequent criteria obtained from Arabidopsis Lon1 experimental data (Supplemental Methods). First, the candidate gene should contain two AUGs in-frame. Second, the sequence context of the first AUG should be weak compared to the strong context of the downstream AUG initiation codon. Third, the deduced translated sequence extending from the first AUG until the downstream AUG and that downstream of the second AUG should be a potential chloroplast transit peptide and mitochondrial presequence or vice versa, respectively. This analysis identified a small set of 19 genes with twin-presequence configuration (Supplemental Table 3). Taken together, a small number of genes within the Arabidopsis genome contain twin presequences for

Molecular Plant


997

Figure 5 Biological Significance of Lon1 Organellar-Specific Localization. (A) Developmental profile of 6-day-old lon1-1 seedlings (upper panel) and of 45-day-old lon1-1 plants at vegetative stage (lower panel) overexpressing organellar-specific Lon1 transgenes. Scale bar: 0.5 cm. (B) Comparison of primary root elongation (6 d old) and shoot length (1 month old) between lon1-1 background plant lines overexpressing organellar-specific Lon1 transgenes; n = 40. Error bars represent standard deviations of the means. Single asterisk indicates significant difference between pairs of transgenic lines (P