Functional Specialization amongst the Arabidopsis Toc159 Family of Chloroplast Protein Import Receptors^W⃞

Phylogenetic Analysis of the Arabidopsis Toc159 Gene Family

In Arabidopsis, as mentioned above, four Toc159-related proteins are present (Bauer et al., 2000; Hiltbrunner et al., 2001a). All four proteins exhibit a characteristic tripartite structure, consisting of an N-terminal acidic domain (A-domain), a central GTP binding domain (G-domain), and a C-terminal membrane-anchor domain (M-domain) (Chen et al., 2000), although the A-domain is greatly reduced in atToc90. Sequence similarities vary between the domains, with the G- and M-domains displaying significantly higher sequence conservation than the A-domain. The two most similar proteins, atToc132 and atToc120, share 93.4% identity within the G-domains and 68.9% identity over their entire length. Amongst the other proteins, G-domain sequence identities range from 44.3% (between atToc159 and atToc90) to 58.1% (between atToc159 and atToc120); identities between the full-length proteins range from 30.5% (between atToc159 and atToc90) to 36.7% (between atToc159 and atToc120).

To look at the relatedness and evolution of the different Toc-GTPases, we constructed a phylogenetic tree using only the G-domain sequences of the different proteins because the G-domain is present and of a similar length in all proteins (Figure 1). In addition to the previously described Arabidopsis and pea proteins, Toc159- and Toc34-related proteins from the monocotyledonous species rice (Oryza sativa) and maize (Zea mays) were included as well as H-Ras p21, a GTPase from human, which was used as an outgroup. Although a similar phylogenetic study was conducted by Hiltbrunner et al. (2001a), the reliability of the predicted clades was not tested by bootstrap analysis, and monocotyledonous sequences were not included. In both studies, two major branches were observed, which correspond to the two main groups of Toc-GTPases: Toc159-related proteins (atToc159, atToc132, atToc120, atToc90, psToc159, osToc86-like_1, and osToc86-like_2 in Figure 1) and Toc34-related proteins (atToc33, atToc34, psToc34, zmToc34-1, and zmToc34-2 in Figure 1).

The phylogenetic tree reflects the sequence identities observed within the Arabidopsis sequences. Arabidopsis atToc132 and atToc120 are within the same clade and, together with one of the rice Toc86-like proteins, form one subgroup of Toc159-related proteins. The other subgroup is comprised of atToc159, psToc159, and the other rice Toc86-like protein. All branches are supported by very high bootstrap values as indicated. The groupings imply that the Toc159 family diverged into two distinct subfamilies (atToc159-like and atToc132/atToc120-like) before species divergence occurred in a common ancestor of monocotyledonous and dicotyledonous species. This implies that there are at least two, fundamentally different subtypes of Toc159-related proteins in plants. By contrast, the data suggest that the Arabidopsis Toc34-related protein family formed later, after species divergence. Although atToc90 shows the lowest sequence identity to the other three Arabidopsis Toc159-related proteins, it clearly belongs to the Toc159 family.

Expression Profiles of the Arabidopsis Toc159 Homologous Genes

To gain an insight into the possible different functions of the four Arabidopsis Toc159-related proteins, we investigated their mRNA expression patterns in different tissues and at different developmental stages. Previous studies only compared three of the four genes (atTOC159, atTOC132, and atTOC120) in whole seedlings (Bauer et al., 2000) or mature leaves and roots (Yu and Li, 2001) by RT-PCR. We examined the expression patterns by comparative RNA gel blot analysis (Figure 2A). Identical RNA gel blots were probed in parallel using identically labeled cDNA probes for the four genes (see Methods). Quantification was performed using equivalent exposures of each blot, and the data were normalized for 28S rRNA (Figure 2B).

The atTOC159 gene is the most regulated of the four: it is expressed highly in young, rapidly dividing photosynthetic tissues and at much lower levels in mature tissues and nonphotosynthetic tissues (Figure 2B). By contrast, the other three genes, atTOC132, atTOC120, and atTOC90, show relatively uniform levels of expression in all tested tissues (Figure 2B). Expression of atTOC132 is much lower than atTOC159 expression in most tissues but always higher (5- to 10-fold) than the expression of atTOC120. In 10-d-old seedlings, expression of atTOC159 is approximately eightfold higher than atTOC132 expression, which is in agreement with the data shown by Bauer et al. (2000). In roots, however, atTOC159 is downregulated, and expression of atTOC132 is actually higher than that of atTOC159, suggesting that atToc132 (and its close relative, atToc120) might be relatively more important than atToc159 in this tissue. These results are broadly consistent with those presented by Ivanova et al. (2004).

Overall, the expression profiles of atTOC159 and atTOC132/atTOC120 parallel those of atTOC33 and atTOC34, respectively (Kubis et al., 2003), which is consistent with the hypothesis that atToc159 is involved preferentially in the import of highly abundant, photosynthetic proteins and that atToc132 and atToc120 are involved preferentially in the import of nonphotosynthetic proteins (Bauer et al., 2000). Interestingly, atTOC90 is expressed at a uniformly high level throughout development (Figure 2B), suggesting that atToc90 may not exhibit similar substrate specificity.

Visible Phenotypes of Toc159 Homolog Knockout Mutants

To directly address the hypothesis that different Toc159 isoforms are involved preferentially in import pathways with different preprotein recognition specificities, we identified Arabidopsis knockout mutants lacking each Toc159 isoform; the atToc159 knockout mutant ppi2 has been described previously (Bauer et al., 2000) and was kindly provided by Felix Kessler. Previously identified mutants affecting the import apparatus were referred to as ppi (Jarvis et al., 1998; Bauer et al., 2000; Constan et al., 2004) because the toc name had already been used to describe circadian clock mutants (Millar et al., 1995). However, the use of different names for corresponding mutants and proteins would become confusing when dealing with multiple components, and so the mutants described in this report are referred to using the names of the corresponding proteins (Schnell et al., 1997; Jackson-Constan and Keegstra, 2001); for the sake of consistency, the ppi2 mutant is referred to here as toc159. Because atToc132 and atToc120 knockout mutants (termed attoc132-1 and attoc120-1) were recently described by Ivanova et al. (2004), we refer to the primary, equivalent mutants described in this report as toc132-2 and toc120-2. The toc159, toc132-2, toc120-2, and toc90-1 mutants each contain a single-locus T-DNA insertion (Figure 3A, Table 1) and are null for the corresponding full-length mRNA as revealed by RNA gel blot analysis (Figure 2A). A shorter transcript was observed in the toc132 mutant (data not shown), which presumably corresponds to transcription of atTOC132 up to the T-DNA insertion site. A truncated protein encoded by this short transcript would lack part of its GTP binding domain and its entire membrane domain, and it is unlikely that it would be functional (Bauer et al., 2002; Lee et al., 2003).

The visible phenotypes of the original four mutants are shown in Figure 3. The strong albino phenotype of the toc159 mutant has been described previously and is consistent with the proposed role of atToc159 as the major receptor for photosynthetic protein import (Bauer et al., 2000). The toc132 mutant has a very slight pale phenotype in young plants (Figure 3B), which develops into a clear yellow-green, somewhat reticulate phenotype in mature plants (Figures 3D and 3E). By contrast, the toc120 and toc90 mutants were indistinguishable from the wild type throughout development (Figures 3B and 3D). To confirm these visible phenotypes, we identified additional, independent alleles of toc132, toc120, and toc90 (Figure 3A). The toc132-3 mutant had exactly the same yellow-green, reticulate phenotype as toc132-2 (see Supplemental Figure 1 online), and the toc120-3, toc90-2, and toc90-3 mutants all appeared exactly the same as wild-type plants (data not shown). Unless stated otherwise, all of the data presented in this article were derived using the originally identified mutant alleles (i.e., toc132-2, toc120-2, and toc90-1).

The phenotypes of the four original mutants were quantified by making chlorophyll measurements (Figure 4A). As expected, wild-type amounts of chlorophyll per unit of fresh weight were observed in the toc120 and toc90 mutants. By contrast, significantly reduced chlorophyll levels were present in 24-d-old toc132 plants (∼70% of the wild-type concentration), and only trace amounts of chlorophyll were present in toc159 plants of the same age (<1% of the wild-type concentration) (Figure 4A).

Our observation that atToc120 knockout mutants display neither a visible phenotype nor a chlorophyll deficiency phenotype is entirely consistent with the results of Ivanova et al. (2004). However, whereas these authors reported that an atToc132 knockout mutant was indistinguishable from the wild type (Ivanova et al., 2004), we observed a clear, visible phenotype in two independent atToc132 knockout mutants (Figure 3; see Supplemental Figure 1 online) and a chlorophyll deficiency phenotype in the toc132-2 mutant (Figure 4). The reason for this apparent discrepancy is not clear. One possibility is that the phenotypic differences reflect the different genetic backgrounds of the mutants. Whereas the toc132-2 and toc132-3 mutants are of the Columbia-0 (Col-0) ecotype, the attoc132-1 mutant described by Ivanova et al. (2004) is of the Wassilewskija (Ws) ecotype, and it is well documented that Col-0 is genetically divergent from many other Arabidopsis ecotypes, including Ws (Barth et al., 2002). Another possibility is that the different observations reflect different plant growth conditions because we repeatedly observed that the toc132 mutant phenotype becomes clear only after a significant period of growth on soil.

Genetic Interactions between the Knockout Mutations

To assess the functional relationships amongst the Arabidopsis Toc159-related proteins, the four knockout mutants were crossed to each other in all pairwise combinations. For these experiments, a 7× Col-0 introgressed toc159 line was used (Table 1) because this mutant was originally of the Ws ecotype and all of the other mutants were of the Col-0 ecotype. All crosses were analyzed in the F2 and F3 generations for novel phenotypes and using appropriate single or double antibiotic selection media. To confirm the genotypes of putative double mutants, PCR analysis was performed using appropriate gene-specific and T-DNA–specific primers. The results of these various double mutant studies are discussed below and in the following sections.

In F2 families derived from crosses between toc90 and toc132 or toc120, individuals that were homozygous for both mutations could be identified in each case. The toc120 toc90 double mutant did not exhibit any new visible phenotypes and appeared exactly the same as wild-type plants (Figure 3D). This phenotypic normality was reflected in the unchanged chlorophyll levels of the double mutant (Figure 4A). Similarly, the toc132 toc90 double mutant displayed the same visible phenotype and chlorophyll concentration as the toc132 single mutant (Figures 3D and 4A). Thus, our data indicate that there is no significant functional overlap or redundancy between atToc90 and atToc120 or atToc132.

The toc132 toc120 Double Homozygote Appears Similar to toc159

When the F2 progeny of crosses between toc132 and toc120 were scored after 10 d of growth in vitro, all individuals could be placed into one of three discrete, phenotypic classes, termed green, pale, and bleached (Table 2); at this early stage of development, the visible phenotype of toc132 is slight and difficult to score (Figure 3B), and plants classified as green here also included toc132 homozygotes. No individuals falling outside of these three phenotypic classes were observed. All 18 pale individuals tested by PCR were found to be homozygous for toc132 and heterozygous for toc120 (genotype: toc132/toc132; +/toc120), whereas all five bleached individuals tested were found to be homozygous for both mutations (genotype: toc132/toc132; toc120/toc120). The ratio observed between these three phenotypes in the F2 generation (Table 2) supports the hypothesis that all pale and bleached individuals have these genotypes (the expected ratio is 13:2:1). The segregation ratios of the pale, bleached, and antibiotic resistance phenotypes in the F3 generation further confirmed the genotypes associated with these novel phenotypes (Table 2).

Our identification of a viable toc132 toc120 double homozygote conflicts with the results of Ivanova et al. (2004) because these authors reported that this genotype causes lethality during embryogenesis or early seedling growth. To further confirm our results, we crossed the toc132-3 and toc120-3 mutants with the original toc120-2 and toc132-2 mutants, respectively, and analyzed the resultant F2 populations. In both cases, green, pale, and bleached plants were once again observed at the expected frequencies (13:2:1; Table 2). Furthermore, when 20 pale plants and 25 bleached plants from the toc132-2 × toc120-3 F2 population were genotyped by PCR, all of the pale plants were found to be homozygous for the toc132 mutation and heterozygous for toc120, and all of the bleached plants were found to be doubly homozygous. We therefore conclude that the toc132 toc120 double homozygous genotype is not lethal in the Col-0 background. The reason for the discrepancy between our data and those of Ivanova et al. (2004) is not clear, but it is once again possible that the phenotypic differences reflect genetic dissimilarities between the Col-0 and Ws ecotypes (Barth et al., 2002) or different plant growth conditions.

The pale phenotype of toc132 toc120 heterozygotes (genotype: toc132/toc132; +/toc120) is clearly visible at the seedling stage, especially in the cotyledons (Figure 3C), and remains so throughout development (Figure 3D). This phenotype is significantly more severe than that of the toc132 single mutant, and this is reflected in the reduced chlorophyll levels of the double mutant (Figures 4A to 4C): the toc132 toc120 heterozygote contains ∼34 to 37% of the wild-type chlorophyll concentration throughout development. Although we occasionally observed some very subtle variegation of the leaves of toc132 toc120 heterozygous plants (data not shown), we never observed the strongly variegated phenotype reported by Ivanova et al. (2004). Again, this phenotypic difference may be because of the use of different plant growth conditions or genetic differences between the Col-0 and Ws ecotypes.

The toc132 toc120 double homozygote has a very strong visible phenotype that is almost as severe as that of the toc159 mutant (Figures 3B to 3D). Its chlorophyll concentration is only ∼10% of that present in the toc132 toc120 heterozygote (genotype: toc132/toc132; +/toc120) but still threefold to fourfold higher than that present in toc159 (Figures 4A to 4C). Thus, unlike toc159, the toc132 toc120 double homozygote is able to survive to maturity on soil (Figure 3F) if first allowed to establish itself on medium containing 3% sucrose.

The progression of phenotypic severity in homozygous toc132 mutants lacking either one (genotype: toc132/toc132; +/toc120) or both (genotype: toc132/toc132; toc120/toc120) atTOC120 alleles clearly demonstrates the existence of substantial functional overlap between atToc132 and atToc120.

The toc159 toc132 Double Homozygote Is Embryo Lethal

To investigate the functional relationships between atToc159 and the other three proteins, we crossed the toc159 (Col-0) mutant to toc132, toc120, and toc90. In the F2 generation of each cross, plants that were homozygous for toc132, toc120, or toc90 and heterozygous for toc159 were identified by scoring for the appropriate antibiotic resistance phenotypes and by conducting PCR analysis. The F3 progeny of these plants were then examined carefully for evidence of genetic interactions. Relatively normal frequencies of albino toc159 homozygotes were observed in the progenies of the toc120 and toc90 crosses (Table 3), and, when these double homozygotes were grown to maturity alongside toc159 single mutants (on medium containing 3% sucrose), no obvious effects of the toc120 or toc90 mutations on toc159 phenotypic severity could be observed (data not shown).

Interestingly, when the F3 progeny of the toc159 × toc132 crosses were examined, no albino plants were observed (Table 3). Because toc159 homozygotes would normally be expected to occur at a frequency of 25% in the F3 generation, the chances of observing no albinos in the total scored population of 2108 plants (Table 3), because of random chance alone, is essentially zero (χ² = 702.667, df = 1, P value = 0.000). Thus, the data suggest that the toc159 toc132 double homozygous mutation is lethal during some early stage of development. Because plastids are known to play an important role during embryo development (Uwer et al., 1998; Apuya et al., 2001; Constan et al., 2004), we suspected that the defect might be occurring during embryogenesis. Consistent with this hypothesis, aborted seeds were observed at a frequency of exactly 25% in the siliques of F3 plants having the genotype +/toc159; toc132/toc132 (Figure 5). Taken together, these data provide strong evidence for the embryo lethality of the toc159 toc132 double homozygous genotype.

In summary, our double mutant studies provided clear evidence for functional redundancy between the close relatives atToc132 and atToc120. No evidence for functional overlap between atToc159 and atToc120, or between atToc90 and any other protein, was observed. The lethality of the toc159 toc132 genotype can be interpreted in two different ways: (1) it reflects the disruption of import of a broad range of precursor types through two different import pathways and indicates minimal functional overlap between atToc159 and atToc132; (2) it reflects a severe or complete block in a single import pathway that is only partially disrupted in either single mutant and indicates significant functional redundancy between atToc159 and atToc132. Considering the sequence similarities between the atToc159, atToc132, and atToc120 proteins, and their relative levels of expression, the former explanation seems the most likely to be correct.

Transgenic Complementation Studies

To corroborate our findings on the functional relationships between atToc132, atToc120, and atToc90 and to differentiate between the two possible explanations for the embryo lethality of the toc159 toc132 genotype, we conducted transgenic complementation studies. All four Arabidopsis Toc159-related proteins were expressed under the control of the strong, constitutive, Cauliflower mosaic virus 35S promoter in two different mutant backgrounds: (1) the toc159 single mutant background; (2) the toc132 toc120 double mutant background. The atTOC90 cDNA used in these experiments (accession number AV548084) was recently shown to be truncated at the 5′ end by the identification of a slightly longer clone (accession number NM_122037). However, the former encodes a protein that is only 14 residues shorter (779 residues) than that encoded by the full-length clone (793 residues). Because this small, N-terminal deletion does not impinge upon the G-domain and because atToc159 mutants completely lacking the N-terminal A- or A+G-domains are active in vivo (Bauer et al., 2002; Lee et al., 2003), it seems unlikely that this mutation would strongly affect protein activity.

At least 10 independent transgenic lines were identified for each transgene/mutant combination. From these 10 lines, between four and six lines were selected for detailed analyses in each case. Lines were selected at random and were phenotypically representative of the entire set. In the selected lines, the degree of complementation of the mutant phenotypes was quantified by making chlorophyll measurements, and the extent of transgene overexpression, relative to the wild type, was assessed using semiquantitative RT-PCR (Figure 6).

The results obtained from the toc159 complementation experiments were clear. All five selected 35S-atTOC159 lines displayed essentially wild-type chlorophyll concentrations (Figure 6A). These complemented plants were homozygous for the toc159 mutation (as revealed by segregation analysis and by PCR; data not shown) and were either heterozygous or homozygous for the 35S-atTOC159 transgene. Wild-type chlorophyll concentrations were achieved despite the fact that the transgene could only drive atTOC159 expression at ∼40% of the wild-type level (Figure 6A). By contrast, lines that were homozygous for the 35S-atTOC132, 35S-atTOC120, or 35S-atTOC90 transgenes, and the toc159 mutation, and which displayed significant levels of transgene overexpression, did not display significantly higher chlorophyll concentrations than untransformed toc159 homozygotes. Taking into account the data shown in Figure 2, the levels of transgene expression achieved with 35S-atTOC132 and 35S-atTOC120 are roughly equivalent to those achieved with 35S-atTOC159. Furthermore, when 35S-atTOC132 and 35S-atTOC120 lines were analyzed by immunoblotting using atToc132- and atToc120-specific antibodies (Ivanova et al., 2004), significant levels of protein overexpression were observed, indicating that both constructs were intact and fully active (see Supplemental Figure 2 online). Although the 35S-atTOC132 transformants did appear slightly larger than control toc159 plants (data not shown), the chlorophyll data clearly indicate that overexpression of atToc132 cannot compensate for the absence of atToc159 to any significant degree (Figure 6A).

The results obtained from the toc132 toc120 complementation experiments were also clear. All five selected 35S-atTOC132 lines and all four selected 35S-atTOC120 lines contained wild-type chlorophyll concentrations (Figure 6B). These complemented plants were homozygous for the toc132 mutation, either heterozygous or homozygous for the toc120 mutation, and either heterozygous or homozygous for the appropriate 35S transgene. These data confirm the previously drawn conclusion that atToc132 and atToc120 exhibit a high level of functional similarity. By contrast, toc132 toc120 heterozygotes (genotype: toc132/toc132; +/toc120) that were either heterozygous or homozygous for the 35S-atTOC159 or 35S-atTOC90 transgenes did not display significantly higher chlorophyll concentrations than untransformed toc132 toc120 heterozygotes (Figure 6B). Because the level of atTOC90 overexpression observed in these transgenic lines was good (5.3-fold on average; Figure 6B), these data strongly support the conclusion that atToc90 and atToc132/atToc120 do not share significant functional redundancy. However, because high levels of atTOC159 overexpression were not observed (presumably because atTOC159 is already expressed at very high levels in wild-type plants; Figure 2), the functional relationship between atToc159 and atToc132/atToc120 is more difficult to assess (data shown in Figure 6A are more informative in this regard). Nevertheless, individual lines that displayed 47 and 30% atTOC159 overexpression, relative to the wild type, did not contain significantly higher chlorophyll concentrations than untransformed toc132 toc120 heterozygotes, suggesting that the level of redundancy is minimal.

Taken together, these transgenic overexpression data confirm the conclusions drawn from the double mutant studies and indicate that the first hypothesis put forward to account for the lethality of the toc159 toc132 genotype is correct; that is, the severity of the phenotype of the double homozygote results from the disrupted import of a broad range of precursor types because of the fact that atToc159 and atToc132 operate preferentially in different import pathways.

Ultrastructure of Plastids in the Single and Double Mutants

To gain insight into the functional specialization of Toc159-related proteins implied by the phylogenetic, genetic, and transgenic data, we used transmission electron microscopy to characterize leaf and root plastid ultrastructure in the mutants (Figure 7). The severity of chloroplast ultrastructural defects broadly paralleled the severity of the visible and chlorophyll phenotypes of the mutants (Figures 3 and 4). In young plants, toc132 chloroplasts were similar to those in the wild type (data not shown), but the chloroplasts of the pale and bleached toc132 toc120 double mutants (genotypes: toc132/toc132; +/toc120 and toc132/toc132; toc120/toc120, respectively) were smaller with less developed thylakoids (Figure 7A). Remarkably, chloroplasts in the toc132 toc120 double homozygote were much more developed, with distinct thylakoids and grana, than those in the toc159 mutant (Figure 7A), despite the fact that two mutants look very similar (Figure 3). This latter observation is consistent with the hypothesis that atToc159 is relatively more important for the biogenesis of photosynthetic proteins and that atToc132 and atToc120 are relatively more important for the biogenesis of nonphotosynthetic proteins (Bauer et al., 2000).

In mature plants, chloroplasts in the toc132 single mutant and the pale toc132 toc120 heterozygote (genotype: toc132/toc132; +/toc120) were small with less developed thylakoids (Figure 7B). Interestingly, in both of these genotypes, the chloroplasts in cells surrounding the vascular bundles were more highly developed than those in the interveinal tissues of the mesophyll (Figure 7B). This observation most likely explains the reticulate appearance of toc132 leaves (Figure 3E). Arabidopsis cue1 mutants, which have defects in the phosphoenolpyruvate/phosphate translocator of the plastid inner envelope membrane, have a similar, but more severe, reticulate phenotype (Streatfield et al., 1999). The biosynthesis of aromatics is compromised in cue1, and the reticulate phenotype of the mutant can be rescued by feeding aromatic amino acids (Streatfield et al., 1999). It is possible that deficiencies in amino acid biosynthesis, caused by the inefficient chloroplast import of the necessary biosynthetic enzymes, are also responsible for the reticulate appearance of the toc132 mutant.

Particularly interesting observations were made when the root plastids of toc132 toc120 double homozygotes were examined. Whereas toc159 root plastids appeared relatively normal (Figure 7C) (Yu and Li, 2001), a high proportion of toc132 toc120 double homozygote root plastids were found to contain large cytoplasmic inclusions (Figure 7C), each surrounded by a normal double envelope membrane (Figure 7C). Of the 75 double mutant plastids examined, 31% contained at least one inclusion. By contrast, of the 49 wild-type and 45 toc159 root plastids examined, only 4 and 9%, respectively, contained cytoplasmic inclusions, and these tended to be smaller than those in the double mutant. Similar inclusions were also observed in the chloroplasts of toc132 toc120 plants (Figures 7A and 7B), which also exhibited other structural irregularities (i.e., they appeared elongated and flattened) (Figures 7A and 7B). It is possible that these structural defects are the result of the failure of toc132 toc120 plastids to import certain proteins, for example, cytoskeletal or division apparatus components. Alternatively, they may be the result of a specific response designed to increase the surface area of metabolically compromised plastids to facilitate the exchange of substances with the cytosol.

Taken together, these electron microscopy data are consistent with the hypothesis that atToc132 and atToc120 play a significant role in the import of nonphotosynthetic proteins.

Protein Import in the toc132/toc132; +/toc120 Double Mutant

To investigate whether the phenotypic differences between the toc132 toc120 double mutants and toc159 could be attributed to differential effects of the mutations on the import of different preproteins, import studies were performed using isolated chloroplasts. The pale toc132 toc120 double mutant (genotype: toc132/toc132; +/toc120) was chosen for these studies because it has a stronger phenotype than either single mutant and yet is sufficiently healthy to allow isolation of a reasonable yield of chloroplasts, which would not be feasible using the double homozygote. The F3 progeny of pale toc132 toc120 double mutants were plated on hygromycin plates, and hygromycin sensitive seedlings (corresponding to toc132 single mutants) were picked out by hand after 7 to 8 d of growth. Both pale and bleached toc132 toc120 seedlings (genotypes: toc132/toc132; +/toc120 and toc132/toc132; toc120/toc120, respectively) were used for chloroplast isolation, but, because of the greatly reduced growth of the double homozygotes, the yield of chloroplasts resulting from these was estimated to be <5% of the total.

The photosynthetic preprotein, preSSU (precursor of the small subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase [Rubisco]), and the nonphotosynthetic preprotein, preL11 (precursor of a 50S ribosomal protein), were selected for analysis because their import efficiencies were previously shown to be affected differentially by the ppi1 mutation (Kubis et al., 2003). However, using the wheat germ translation system for precursor synthesis, no defect in import was observed for either precursor in the double mutant chloroplasts (see Supplemental Figure 3 online). To eliminate the possibility that soluble factors in the wheat germ lysate were responsible for this negative result (Hiltbrunner et al., 2001b; Schleiff et al., 2002), we tested the same precursors again using the rabbit reticulocyte system for translation (Figure 8). Once again, however, no import deficiencies were detected, and our data show that double mutant chloroplasts import both precursors at a similar rate to wild-type chloroplasts in vitro.

One possible explanation of these data is that the soluble forms of Toc159-related proteins act, in vivo, to increase the efficiency of import by facilitating the targeting of precursors from the cytosol to the translocation channel (Hiltbrunner et al., 2001b). Thus, in an in vitro import system that lacks these soluble components, base rates of import might still be observed, but the effects of the toc132 and toc120 mutations would no longer be apparent. Alternatively, it is possible that the atToc132 and atToc120 proteins are involved preferentially in the import of subsets of proteins that do not include either preSSU or preL11.

Analysis of the toc132 Chloroplast Proteome

In a further attempt to gain insight into the substrate preferences of the import pathway in which atToc132 (and atToc120) are involved, the chloroplast proteome of the toc132 mutant was analyzed. To do this, we employed the same procedure—involving difference gel electrophoresis (DIGE) and CyDye technology—that was used to analyze the ppi1 chloroplast proteome (Kubis et al., 2003). The degree of enrichment or depletion of toc132 chloroplast proteins, relative to the wild type, was determined by CyDyeDIGE analysis (see Supplemental Figure 4 online), and these data were then used to select proteins for identification by mass spectrometry (Table 4).

The results clearly show that the perturbations that exist in the toc132 chloroplast proteome are distinct from those that were observed previously in the ppi1 chloroplast proteome. For example, photosynthetic proteins such as SSU, OE33, and ribose-5-phosphate isomerase—all of which were deficient in ppi1 chloroplasts—are present at normal levels in toc132. Furthermore, some photosynthetic proteins (e.g., Rubisco activase and the Calvin cycle enzyme phosphoribulokinase) are even slightly enriched in toc132 chloroplasts. However, no clear trend indicating a selective effect of the toc132 mutation on the accumulation of nonphotosynthetic proteins was observed because proteins depleted in the mutant included the Calvin cycle enzyme fructose-1,6-bisphospate aldolase in addition to obviously nonphotosynthetic proteins (the branched-chain amino acid biosynthetic enzymes ketol-acid reductoisomerase and dihydroxy-acid dehydratase). What is more, many nonphotosynthetic proteins are present at normal levels (e.g., the 50S ribosomal protein L12-C and the molecular chaperones Hsp70 and Hsp90) or slightly elevated levels (e.g., the ammonium fixation enzyme glutamate-ammonia ligase and the translation elongation factor EF-Tu) in toc132 chloroplasts. These data may indicate that atToc132 is not exclusively involved in the import of nonphotosynthetic precursors or reflect the continued presence of atToc120 protein in the toc132 mutant. Nevertheless, it is clear that the toc132 chloroplast proteome is quite different from the ppi1 chloroplast proteome, in that it is not specifically deficient in photosynthetic proteins.

The toc132 Transcriptome Response Is Distinct from the ppi1 Response

The expression of nuclear genes encoding plastidic proteins is regulated in response to complex signals emitted by the plastids themselves (Jarvis, 2003; Richly et al., 2003). We previously demonstrated that specific defects in the import and accumulation of photosynthetic proteins in the ppi1 mutant (which lacks a receptor, atToc33, with putative specificity for photosynthetic proteins) are correlated with the selective downregulation of the corresponding nuclear photosynthetic genes (Kubis et al., 2003). Because the transcriptome response of ppi1 reflected the characteristics of the ppi1 import defect, we used exactly the same procedure to characterize the toc132 transcriptome response. Thus, nylon filter DNA array technology (Richly et al., 2003) was employed to compare the mRNA expression of 3292 nuclear genes, most of them encoding plastidic proteins, in the toc132 mutant and the wild type; RNA gel blot analysis was previously used to confirm the accuracy of gene expression data generated using this method (Kubis et al., 2003). A total of 686 genes (∼20% of those analyzed) were found to be significantly differentially expressed in toc132 compared with the wild type (see Supplemental Table 1 online). Of those genes showing differential expression in toc132, ∼70% (486) were upregulated and ∼30% (200) were downregulated. A similar number of downregulated genes (161 of a total of 1461 differentially expressed genes) was observed in the ppi1 mutant (Kubis et al., 2003).

When we compared the toc132 transcriptome response with the previously described ppi1 response (Kubis et al., 2003) by cluster analysis, we observed that the responses were only partially overlapping (Figure 9A). Although the gene expression profiles of the two mutants were quite similar with respect to upregulated genes (Figure 9A), major differences were observed when only downregulated genes were considered (Figure 9B). Of all the genes showing downregulated expression in toc132 (200 genes), only 3% (six genes) were also downregulated in ppi1. Similarly, of all the genes showing downregulated expression in ppi1 (161 genes), only 4% (six genes) were also downregulated in toc132. Thus, toc132 does not exhibit the same specific downregulation of photosynthetic genes as was found in ppi1. Indeed, very few of the 200 genes that show reduced expression in toc132 encode known photosynthetic proteins (see Supplemental Table 1 online). Although these data do not provide a direct assessment of protein import in the toc132 mutant, they are nevertheless consistent with the hypothesis that atToc132 is involved preferentially in an import pathway with some specificity for nonphotosynthetic proteins.

Sequence Analysis

Amino acid sequences were aligned using ClustalW within the BioEdit program (Hall, 1999). Phylogenetic trees were calculated using PAUP^* 4.0 b10 (Swofford, 2003). All analyses were performed using branch and bound searches, with the collapse option and furthest addition sequence selected. No weighting or ordering was imposed on the characters, and any gaps were treated as missing data. Indels were coded separately and appended to the sequence data matrix. Coding of indels was usually binary (deletions 0, insertions 1), but in places where more than one size occurred, the alternatives were coded as 2, 3, etc. Support for clades was estimated by means of nonparametric bootstrap analyses, as implemented in PAUP* 4.0, using 1000 replicates. The human c-H-Ras1 p21 proto-oncogene (accession number P01112) was used as an outgroup.

Plant Material and Growth Conditions

Arabidopsis thaliana plants, both wild type and mutants, used in this study were of the Col-0 ecotype. The ppi2 (toc159) mutant, which is of the Ws ecotype, was kindly provided by Felix Kessler (Bauer et al., 2000) and was subsequently introgressed into the Col-0 ecotype (see Results). Seeds were surface sterilized, and plants were grown as described previously (Aronsson and Jarvis, 2002). When necessary, the following antibiotics were included in the medium at the indicated concentrations: 50 μg/mL of kanamycin monosulphate (Melford Laboratories, Ipswich, UK) was used to select for toc159 and the Salk Institute Genomic Analysis Laboratory (SIGnAL) lines; 10 μg/mL of dl-phosphinothricin (Duchefa, Haarlem, The Netherlands) was used to select for toc132-2 and toc90-1; 15 μg/mL of hygromycin B (Duchefa) was used to select for toc120-2; and 110 μg/mL of gentamicin sulfate (Duchefa) was used to select for pCHF2 transformants. Plants grown on soil were kept under standard greenhouse conditions.

Identification of Knockout Mutants

The three new mutants that were the main focus of this study were obtained from the following sources: Syngenta, lines Garlic_667_D04 (toc132-2) and Garlic_1236_C11 (toc90-1); Csaba Koncz Laboratory, pool 286, line 28,567 (toc120-2). The additional, supportive mutants were obtained from SIGnAL, lines SALK_017374 (toc120-3), SALK_064469 (toc90-2), and SALK_119434 (toc90-3) and from Genomanalyse im Biologischen System Pflanze (GABI)-Kölner Arabidopsis T-DNA (Kat), line 394E01 (toc132-3). The mutants were isolated according to published procedures: Syngenta (Sessions et al., 2002); Csaba Koncz Laboratory (Ríos et al., 2002); SIGnAL (Alonso et al., 2003); GABI-Kat (Rosso et al., 2003).

Gene-specific and T-DNA–specific primers used to confirm the mutations were as follows: Toc132-F, 5′-GATGGGACTGAGTTTGTGGTTAGGTC-3′; Toc132-R, 5′-CAAAGCACATCAACGCTCAGCAAAATCCA-3′; Toc120-F, 5′-CTAACCAGGTTAGATTCTCGCCATCAC-3′; Toc120-R, 5′-GACGTGTTAAATTGACGAGCACTGTAGAG-3′; Toc90-F, 5′-TTCTTCTTTAGTCGGTTTATTGTGCGGTGGAG-3′; Toc90-R, 5′-AGGAGTGGAGAAATCAAAGAGAAAGCGAGGAG-3′; Syngenta T-DNA LB3, 5′-TAGCATCTGAATTTCATAACCAATCACGATACAC-3′; Syngenta T-DNA 104RB3, 5′-TAACAATTTCACACAGGAAACAGCTATGAC-3′; Koncz T-DNA LB FISH1, 5′-CTGGGAATGGCGAAATCAAGGCATC-3′; Koncz T-DNA RB HOOK2, 5′-TACTTTCTCGGCAGGAGCAAGGTGA-3′; SIGnAL T-DNA LBa1, 5′-TGGTTCACGTAGTGGGCCATCG-3′; GABI-Kat T-DNA LB, 5′-CCCATTTGGACGTGAATGTAGACAC-3′.

For the original three mutants, both T-DNA junctions were sequenced to verify the predicted location of the T-DNA with respect to each gene (Figure 3A); for the additional, supportive mutants, only one junction was sequenced in each case (Figure 3A). Complex T-DNA insertions were present in the toc132-2 and toc90-1 mutants because left border sequences were found at both ends of the insertion; however, no evidence for rearrangements of gene sequences or larger deletions was observed. Single-locus T-DNA insertion lines were identified for each of the original mutants (Table 1), and homozygotes derived from these lines were confirmed by PCR before further analysis.

Chlorophyll Quantification and Electron Microscopy

Chlorophyll determinations and transmission electron microscopy were performed as described previously (Porra et al., 1989; Constan et al., 2004). Transmission electron microscopy was performed at the Electron Microscope Laboratory, Faculty of Medicine and Biological Sciences, University of Leicester.

Identification of Double Mutants

Homozygous toc132, toc120, and toc90 plants and heterozygous toc159 (Col-0) plants were crossed to each other in all pairwise combinations, the resultant F1 plants were allowed to self-pollinate, and F2 seeds were obtained. After 10 d of growth in vitro, in the presence or absence of appropriate antibiotics, F2 plants were scored for antibiotic resistance and chlorotic phenotypes. After several more days of growth, genomic DNA was extracted (Edwards et al., 1991) from selected individuals and used for PCR genotyping. The PCR primers used were the gene-specific and T-DNA–specific primers given above.

Transgenic Complementation Experiments

The atTOC159, atTOC132, atTOC120, and atTOC90 open reading frames were expressed from a double-enhancer version of the 35S promoter, using plant transformation/expression vector, pCHF2 (Jarvis et al., 1998). For atTOC90, cDNA clone RZL46 g05 (accession number AV548084) was subcloned into pCHF2 as a SacI/end-filled Asp718 fragment into SacI/SmaI-cut pCHF2. The full-length atTOC120 open reading frame was amplified from Arabidopsis Col-0 genomic DNA using forward (5′-TCTCTGAGCTCTGTTTCGTGATTTTGGGTGAT-3′) and reverse (5′-TCAAAGTCGACAAACAAAAGTGAATTCCGACAAC-3′) primers and cloned into pCHF2 using SacI/SalI restriction sites introduced into the forward and reverse primers, respectively. The resulting atTOC120 complementation vector was confirmed by sequencing. The atTOC132 complementation vector was constructed from the truncated cDNA clone H3A2 (accession number W43632), which carries a 975-bp 5′ terminal deletion. Firstly, the truncated cDNA was subcloned as a SacI/XhoI fragment into a modified version of pCHF2 (carrying a single EcoRI site in the polylinker) cut with SacI/SalI. A 2-kb atTOC132 RT-PCR product, amplified using forward (5′-GTCTGAGCTCCTATCTCTAACTTCTGCGGTGGTG-3′) and reverse (5′-CAGGTGGATTCTTCTTGATG-3′) primers, was then cloned as a SacI/EcoRI fragment into the complementation vector to reconstitute the full-length atTOC132 cDNA. The PCR-amplified portion of the full-length atTOC132 cDNA was confirmed by sequencing. The full-length atTOC159 open reading frame, including the intron, was constructed from cDNA clone APZL25e08 (accession number AV523359), which carries a 5′ truncation, and a 3-kb PCR product amplified from Col-0 genomic DNA using forward (5′-CAAAGGTACCATGGACTCAAAGTCGGTTAC-3′) and reverse (5′-GTCTGAGCTCCTATCTCTAACTTCTGCGGTGGTG-3′) primers. The atTOC159 amplification product was ligated as an ApaI/HindIII fragment into the partial cDNA clone to reconstitute the full-length open reading frame, which was then confirmed by sequencing and cloned into pCHF2 as an Asp718/XbaI fragment.

The complementation constructs were introduced into Agrobacterium strain GV3101 (pMP90) and then used to transform toc159 heterozygotes or toc132 toc120 heterozygotes (genotype: toc132/toc132; +/toc120) by the floral dip method (Clough and Bent, 1998). Transgene overexpression was estimated by RT-PCR and, in some cases, by immunoblotting. RNA was extracted from 15-d-old T2 plants grown on medium containing gentamicin using the RNeasy plant mini kit (Qiagen, Valencia, CA) and treated with DNase I (DNA-free; Ambion, Austin, TX). RT-PCR was conducted using previously described procedures (Constan et al., 2004), using 18 cycles of amplification. PCR primers were selected to give ∼700-bp products and were as follows: atTOC159 forward, 5′-CAGTAGCAAAGCGGAAATGGACTCAAAG-3′; atTOC159 reverse, 5′-GCCACATCAACATGCACTGATTC-3′; atTOC132 forward, 5′-GATGGGACTGAGTTTGTGGTTAGGTC-3′; atTOC132 reverse, 5′-CTCTTGTTCTGTCTGTATGCC-3′; atTOC120 forward, 5′-AATTGGTTCGCAGGAGGGTC-3′; atTOC120 reverse, 5′-CTTTCAGTCTCTCCCTTCTC-3′; atTOC90 forward, 5′-TTCCGTGACCCTCATCAAGAAC-3′; atTOC90 reverse, 5′-GAGAAATCACTGTATCGCATGTCG-3′; eIF4E1 forward, 5′-AAACAATGGCGGTAGAAGACACTC-3′; eIF4E1 reverse, 5′-AAGATTTGAGAGGTTTCAAGCGGTGTAAG-3′. PCR products were resolved by agarose gel electrophoresis, stained with ethidium bromide, and quantified using Quantity One software (Bio-Rad, Hercules, CA); a λ DNA dilution series was used to confirm the accuracy of the quantification procedure. Data for the Toc159-related genes were normalized using equivalent data for the translation initiation factor gene eIF4E1 (Rodriguez et al., 1998). Total protein extracts were prepared from 10-d-old T3 plants as described previously (Aronsson et al., 2003) and were analyzed by staining with Coomassie Brilliant Blue R 250 (Fisher Scientific, Loughborough, UK) or by immunoblotting (Aronsson et al., 2003) using antibodies against atToc132 and atToc120 (Ivanova et al., 2004).

Complementation was estimated by making chlorophyll measurements using 10-d-old T3 plants. T3 families in the toc132 toc120 background were grown on medium containing phosphinothricin, hygromycin, and gentamicin, and only triply-antibiotic-resistant plants were used for chlorophyll quantifications. T3 families carrying the 35S-atTOC132, 35S-atTOC120, and 35S-atTOC90 constructs in the toc159 background were scored on medium containing gentamicin to identify 35S transgene homozygotes; chlorophyll measurements were then conducted using only 35S transgene/toc159 double homozygotes. T3 families carrying the 35S-atTOC159 construct in the toc159 background were scored phenotypically and by PCR to identify toc159 homozygotes; chlorophyll measurements were then conducted using only toc159 homozygotes carrying the 35S transgene.

Isolation of Arabidopsis Chloroplasts

Chloroplasts were isolated from 10-d-old wild-type and pale toc132 toc120 double mutant (genotype: toc132/toc132; +/toc120) plants grown in vitro as described previously (Aronsson and Jarvis, 2002; Kubis et al., 2003). Plant material was homogenized for 3 to 4 s (wild type) or 2 to 3 s (toc132 toc120) using a polytron. The yield and intactness of the chloroplasts were determined as described previously (Aronsson and Jarvis, 2002).

Protein Import into Chloroplasts

Template DNA for the in vitro transcription/translation of preproteins was amplified by PCR from cDNA clones using M13 primers. The preSSU and preL11 cDNA clones were as described previously (Aronsson and Jarvis, 2002). Transcription/translation was performed using a coupled reticulocyte lysate system (TNT T7 Quick for PCR DNA; Promega, Madison, WI) or a coupled wheat germ system (Promega), containing ³⁵S-Met and T7 RNA polymerase, according to the manufacturer's instructions. Import reactions and quantifications were performed as described previously (Aronsson and Jarvis, 2002; Kubis et al., 2003).

Chloroplast Proteomics

Preparation of protein samples from wild-type and toc132 mutant chloroplasts, their labeling with complementary CyDyeDIGE fluors, separation, quantification, and identification was performed as described by Kubis et al. (2003). The following changes were applied: proteins within the gel-excised spots were first reduced, carboxyamidomethylated, and then digested to peptides using trypsin on a MassPrepStation (Micromass, Manchester, UK). The resulting peptides were applied to either matrix-assisted laser-desorption ionization time of flight MS (TofSpec2E; Micromass), for peptide mass fingerprinting, or LC-MS/MS. For LC-MS/MS, the liquid chromatographic separation was achieved with a PepMap C18 reverse phase, 180-μm i.d., 15-cm column (LC Packings, Amsterdam, The Netherlands), and the mass spectrometer was a Qtof2 (Micromass). Fragmentation data was used to search the National Center for Biotechnology Information database using the MASCOT search engine (http://www.matrixscience.com). Probability-based MASCOT scores were used to evaluate identifications. Only matches with P < 0.05 for random occurrence were considered significant (further explanation of MASCOT scores can be found at http://www.matrixscience.com).

RNA Isolation and RNA Gel Blot Analysis

RNA was isolated from plant material grown in vitro (10 d old) or on soil (28 d old) as described by Kubis et al. (2003). RNA gel electrophoresis, transfer to Hybond NX membrane (Amersham Pharmacia Biotech, Uppsala, Sweden), labeling of DNA fragments, and hybridization and washing of membranes at 65°C were all performed as described previously (Kubis et al., 2003). Band quantification was performed using ImageQuant software (Molecular Dynamics, Sunnyvale, CA).

Identical filters were probed simultaneously with atTOC159, atTOC132, atTOC120, and atTOC90 probes. The four probes corresponded to cDNA fragments of the A-domain (A+G-domain in the case of atToc90) that were similar in length (786-bp atTOC132, 802-bp atTOC120, 874-bp atTOC90, and 910-bp atTOC159) and guanine/cytosine nucleotide content (39.15% atTOC120, 42.31% atTOC132, 43.36% atTOC90, and 46.37% atTOC159). The probes were labeled simultaneously under identical conditions using the same isotope and were shown to have identical specific activities by scintillation counting (<10% variation). Hybridization, washing, and exposure steps were performed simultaneously under identical conditions.

Cross-hybridization among the four genes can be excluded because probes were selected from the most variable region of the genes, where homology is low or restricted to short DNA stretches (in the case of atToc120 and atToc132), and the wash stringency used only allowed probe-target combinations with >83% homology to remain hybridized.

DNA Array Analysis

The 3292 GST array (Richly et al., 2003) was used for the transcriptome analysis of the toc132 mutant as described by Kubis et al. (2003). The data obtained for all significantly differentially regulated genes are given in Supplemental Table 1 online. Complete data are deposited at the GeneOmnibus Web site (http://www.ncbi.nlm.nih.gov/geo/) under the accession number GSM14832. Hierarchical clustering of the expression profiles of the 288 genes that show significant differential expression in both toc132 and ppi1 was performed using Genesis Software (version 1.1.3; Sturn et al., 2002).

Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession numbers AV523359, AV548084, NM_122037, P01112, and W43632.