Multiple fates of non-mature lumenal proteins in thylakoids


For correspondence (e-mail


Most proteins found in the thylakoid lumen are synthesized in the cytosol with an N–terminal extension consisting of transient signals for chloroplast import and thylakoid transfer in tandem. The thylakoid-transfer signal is required for protein sorting from the stroma to thylakoids, mainly via the cpSEC or cpTAT pathway, and is removed by the thylakoidal processing peptidase in the lumen. An Arabidopsis mutant lacking one of the thylakoidal processing peptidase homologs, Plsp1, contains plastids with anomalous thylakoids and is seedling-lethal. Furthermore, the mutant plastids accumulate two cpSEC substrates (PsbO and PetE) and one cpTAT substrate (PsbP) as intermediate forms. These properties of plsp1-null plastids suggest that complete maturation of lumenal proteins is a critical step for proper thylakoid assembly. Here we tested the effects of inhibition of thylakoid-transfer signal removal on protein targeting and accumulation by examining the localization of non-mature lumenal proteins in the Arabidopsis plsp1-null mutant and performing a protein import assay using pea chloroplasts. In plsp1-null plastids, the two cpSEC substrates were shown to be tightly associated with the membrane, while non-mature PsbP was found in the stroma. The import assay revealed that inhibition of thylakoid-transfer signal removal did not disrupt cpSEC- and cpTAT-dependent translocation, but prevented release of proteins from the membrane. Interestingly, non-mature PetE2 was quickly degraded under light, and unprocessed PsbO1 and PsbP1 were found in a 440-kDa complex and as a monomer, respectively. These results indicate that the cpTAT pathway may be disrupted in the plsp1-null mutant, and that there are multiple mechanisms to control unprocessed lumenal proteins in thylakoids.


The viability of life on the surface of the Earth largely depends on oxygenic photosynthesis. In photosynthetic eukaryotes, this reaction occurs in the chloroplast. Within the chloroplast, four major oligomeric protein complexes in the thylakoid catalyze photosynthetic electron transport, generating proton motive force (PMF) across the membrane and producing reductants and ATP for carbon fixation and various other processes in the stroma. Photosynthetic electron transport components in thylakoids comprise both integral membrane proteins and soluble proteins in the lumen. The latter group includes oxygen-evolving complex subunits within photosystem II (PSII) and the copper-containing soluble electron carrier plastocyanin (Pesaresi et al., 2009; Bricker et al., 2011). The thylakoid lumen also contains thioredoxin-like proteins and proteases, which are included in a total of approximately 80 soluble or membrane-anchored proteins in this compartment, all of which, except cytochrome f, are encoded by the nuclear genome (Peltier et al., 2002; Schubert et al., 2002; Zybailov et al., 2008).

Nuclear-encoded thylakoid lumenal proteins are synthesized with an extension in their N–termini. This extension may be divided into two parts. The N–terminal portion is called a transit peptide. This is necessary for the translocon at the outer-/inner-envelope-membrane of chloroplasts (TOC/TIC)-dependent envelope transport and is cleaved by soluble stromal processing peptidase (SPP) (von Heijne et al., 1989; Richter and Lamppa, 1998; Bhushan et al., 2006; Li and Chiu, 2010). The C–terminal portion of the extension acts as a thylakoid-transfer signal (TTS). TTS is required for protein transport from the stroma to the thylakoid, and is removed by the membrane-bound thylakoidal processing peptidase (TPP) in the lumen (Figure 1a) (Paetzel et al., 2002; Teixeira and Glaser, 2012). TTS-dependent protein transport is homologous to bacterial protein export (Schatz and Dobberstein, 1996). It is catalyzed mainly by the cpSEC (chloroplastic SEC) and cpTAT (chloroplastic twin-arginine transfer) pathways, which utilize ATP hydrolysis in the stroma and PMF across the membrane, respectively (Mori and Cline, 2001; Müller and Klösgen, 2005). The TTS resembles a bacterial export signal in that it consists of an N–terminal positive domain followed by a hydrophobic core of 12–18 residues and a C–terminal region ending with the Ala-Xaa-Ala motif (Figure 1) (von Heijne et al., 1989; Mori and Cline, 2001).

Figure 1.

Bipartite targeting sequences of thylakoid lumenal proteins. (a) The import signal (transit peptide dashed line) and the thylakoid-transfer signal (TTS), which consists of an N–terminal positive region (N), a hydrophobic core (H, open box) and a C–terminal domain (C), are indicated. The conserved Ala-Xaa-Ala motif at −3 to −1 relative to the TPP cleavage site is shown as AxA. The SPP and TPP cleavage sites are indicated by arrows. (b) TTS sequences of three lumenal proteins used in this study. The SPP cleavage sites were predicted by comparison with those of wheat PsbO and PsbP and Silene pratensis PetE (Bassham et al., 1991). The TPP cleavage sites are based on the results of a previous proteomics study (Zybailov et al., 2008). The numbers of N-terminal residues of TTS are indicated. Basic residues in the N domains are indicated by asterisks. The H domains, defined based on Robinson and Mant (2005), are underlined. The residues introduced in ARAF and XP mutants are indicated below the sequences.

TPP belongs to the type I signal peptidase (SPase I) family (Hageman et al., 1986; Smeekens et al., 1986; Kirwin et al., 1987, 1988; Shackleton and Robinson, 1991). This group includes bacterial leader peptidases, which cleave export signals at the plasma membrane. The SPase I is anchored to the membrane via one or two α–helical transmembrane domains in its N–terminus, and its soluble C–terminal portion, containing the catalytic site, is located on the trans side of the membrane (San Millán et al., 1989). Disruption of leader peptidase activity in bacteria by gene knockout caused cell death (Date, 1983; Cregg et al., 1996; Zhbanko et al., 2005). This is attributed to accumulation of proteins with uncleaved export signals in the membrane (Dalbey and Wickner, 1985; Kuhn and Wickner, 1985). Similarly, inhibition of TTS removal by mutating residues around the TPP cleavage site of the 33 kDa subunit of the oxygen-evolving complex (also called OE33 or PsbO) and the 23 kDa subunit of the oxygen-evolving complex (also called OE23 or PsbP) resulted in tight association of non-mature proteins to the thylakoid membrane (Shackleton and Robinson, 1991; Di Cola and Robinson, 2005; Frielingsdorf and Klösgen, 2007).

In land plants, there are two TPP isoforms, called plastidic type I signal peptidase Plsp1 and Plsp2 (Chaal et al., 1998; Inoue et al., 2005; Hsu et al., 2011). Disruption of the PLSP1 gene in Arabidopsis impeded photoautotrophic growth and caused a pale seedling phenotype (Inoue et al., 2005; Shipman-Roston et al., 2010). Immunoblotting showed that plsp1-null plastids accumulated non-mature forms of two cpSEC substrates (PsbO and plastocyanin, also called PetE), one cpTAT substrate (PsbP) and an envelope protein (Toc75). The direct involvement of Plsp1 in processing of thylakoid and envelope proteins is supported by localization of Plsp1 to the two distinct membranes by the immunolocalization assay (Shipman and Inoue, 2009). Thylakoids of plsp1-null plastids were severely inflated and found to accumulate unprocessed PsbO in the membrane (Shipman-Roston et al., 2010). A genetic complementation assay using the toc75-null mutant revealed that complete maturation of Toc75 was dispensable for proper chloroplast development. These data suggest the importance of thylakoid-localized Plsp1 for photoautotrophic growth of plants (Endow et al., 2010; Shipman-Roston et al., 2010). However, whether and how accumulation of non-mature lumenal proteins disrupts proper assembly of photosynthetic membranes remains unexplored.

In this study, we used the Arabidopsis plsp1-null plastid and an in vitro import assay with pea (Pisum sativum) chloroplasts to obtain insights into the significance of TTS removal for protein targeting and stability. Our results suggest that the cpTAT pathway may not be active in the plsp1-null plastid, and that thylakoids have several mechanisms to control non-mature lumenal proteins.


Distribution of non-mature thylakoid lumenal proteins in plsp1-null plastids

A previous study showed that non-mature PsbO was present in the peripheral area of inflated thylakoids in plsp1-null plastids (Shipman-Roston et al., 2010). This result suggests that the mutant plastids have an active cpSEC pathway, which depends on ATP hydrolysis in the stroma, and that removal of TTS is necessary for release of this protein from the membrane, as shown by a previous in vitro study (Shackleton and Robinson, 1991). However, it was unknown whether there is enough PMF across the thylakoid membrane of plsp1-null plastids, which lacks sufficient photosynthetic electron transport activity to sustain photoautotrophic growth, to catalyze the cpTAT transport. Furthermore, it was not known whether all the non-mature lumenal proteins found in the mutant are anchored to the membrane. To address these issues, we isolated plastids from plsp1-null plants and examined membrane association of three non-mature lumenal proteins, PsbO, PsbP and PetE, by immunoblotting. We also included two integral membrane proteins, Toc75 and FtsH2/8, which are present in plsp1-null plastids (Shipman-Roston et al., 2010), in the analysis. Toc75 is the envelope protein that accumulates as unprocessed forms in plsp1-null plastids (Inoue et al., 2005; Shipman-Roston et al., 2010). FtsH2 was recently shown to use the cpTAT pathway for targeting to the thylakoid membrane (Rodrigues et al., 2011). FtsH2/8 in wild-type and plsp1-null plastids migrated at comparable rates in 12% SDS–PAGE (Figure 2, αFtsH2/8). However, under 7.5% SDS–PAGE, the mobility of FtsH2/8 in the mutant plastids was similar to that of the stromal processing intermediate, which migrated slightly more slowly than mature FtsH2/8 in wild-type (Figure S2). Thus, FtsH2/8 appear to accumulate as the non-mature form in the plsp1-null mutant. As the amount of proteins per plastid was comparable between wildtype and plsp1-null mutants (Shipman-Roston et al., 2010), the sample loading was normalized based on the amount of plastid proteins (Figure 2). The distribution of the two sets of the integral membrane proteins (Toc75 and FtsH2/8) validated the intactness of the membranes (Figure 2, lanes 3, 5, 8 and 10). In wildtype, all three lumenal proteins were found in the membrane fraction after hypotonic lysis (Figure 2, lane 3). After sonication treatment, which completely released PetE into the soluble fraction, approximately two-thirds of PsbO and PsbP remained in the pellet (Figure 2, lanes 4 and 5). These results are consistent with previous data showing that a portion of oxygen-evolving complex subunits was associated with PSII and was non-extractable by sonication (Peng et al., 2006) and that PetE exists free in the lumen (Kirwin et al., 1988). By contrast, in plsp1-null plastids, non-mature PsbO and PetE isoforms were mainly recovered in the membrane after sonication (Figure 2, lane 10, and Figure S1), and, interestingly, non-mature PsbP was found in the soluble fraction after hypotonic lysis (Figure 2, lane 7). These results indicate that accumulation of non-mature PsbO and PetE in plsp1-null plastids is due to the absence of processing peptidase. By contrast, accumulation of unprocessed PsbP may be due to either disruption of its transport from the stroma to the thylakoid or reverse translocation of the non-mature protein from the thylakoid to the stroma. The second scenario is less likely because previous studies showed that, once transported, non-mature cpTAT substrates remain in the membrane unless their TTS is removed (Di Cola and Robinson, 2005; Frielingsdorf and Klösgen, 2007).

Figure 2.

Membrane association of endogenous TPP substrates in wild-type and plsp1-null plastids. Isolated chloroplasts from wild-type and plsp1-1 plants (C) were lysed hypotonically and separated into soluble (S) and pellet (P) fractions. The resulting chloroplast membrane fractions were resuspended in HM buffer to 0.1 μg chlorophyll μl−1, and subjected to sonication for 1 min, followed by centrifugation into soluble (S) and pellet (P) fractions. Samples corresponding to chloroplasts containing 3.4 μg proteins were analyzed by immunoblotting using the antisera indicated on the left. The intermediate and mature forms are indicated on the right. Black vertical lines indicate grouping of images from different parts of the same gel. Enhanced chemiluminescence analysis with a longer exposure time did not reveal the presence of the soluble form of non-mature PsbO in plsp1-null plastids after sonication (Figure S1). Migration of FtsH2/8 was indistinguishable between wild-type and the mutant under these conditions (12% SDS–PAGE) (asterisk). However, the results of analysis by 7.5% SDS–PAGE suggest that FtsH2/8 in plsp1-null plastids corresponded to the intermediate form (Figure S2).

Mutation around the TPP cleavage site does not affect chloroplast import but prevents complete maturation of lumenal proteins

In order to obtain further insights into the fates of non-mature lumenal proteins, we performed an in vitro import assay with radiolabeled proteins and chloroplasts isolated from pea seedlings. This assay system has been used to address various questions about protein targeting to and within chloroplasts (Keegstra and Cline, 1999). We inhibited TTS removal by amino acid substitution, and examined the fates of non-mature forms of three lumenal proteins from Arabidopsis, PsbO1 (At5g66570), PsbP1 (At1g06680) and PetE2 (At1g20340) (Figure 1b). This selection of substrates was based on the presence of their non-mature forms in plsp1-null plastids, allowing us to compare the properties of non-mature proteins both in vivo and in vitro. For PetE2, we used PetE2MV, in which Met67 was replaced by Val. This mutation, which did not affect proper targeting and processing (Figure S3), was used in order to avoid interference from the [35S]Met-labeled translation product starting from the AUG corresponding to Met67, whose mobility was indistinguishable from that of the mature form starting at Ile69. To inhibit complete maturation, we generated two types of substitution (ARAF and XP, Figure 1b) that were previously shown to prevent maturation of a Plsp1 substrate, Toc75, both in vitro and in vivo (Shipman-Roston et al., 2010). Performing the assay using recombinant Escherichia coli LepB confirmed that these mutations prevented SPase I processing in vitro (Figure S4).

We then tested whether the mutations affect removal of import signals by SPP and chloroplast import in vitro. As shown in Figure 3(a), PsbO1 and PetE2 variants were processed by SPP to a single stromal processing intermediate (Si), and PsbP1 variants were processed to two processing intermediates (Si1 and Si2); the mobility of the intermediates derived from mutated forms was indistinguishable from those derived from non-mutated forms (lanes 2, 5 and 8).

Figure 3.

Substitutions around TPP cleavage sites disrupt proper maturation of lumenal proteins. (a) Radiolabeled precursor proteins were incubated with the stromal extract at room temperature for 90 min (SPP) or with intact chloroplasts and 3 mm ATP at room temperature for 20 min under light (for PsbO1 and PsbP1) or dark (for PetE2MV), and chloroplasts were re-isolated through a 40% Percoll (GE Healthcare Bio-Sciences, cushion (imp), and subjected to SDS–PAGE. ‘tl’ indicates 1 μl of translation product, which was the same amount used for the SPP assay and 10% of the amount used for the import assay. [35S]Met-labeled proteins were visualized using a phosphorimager. (b) For phosphorimager analysis (PI), radiolabeled precursor proteins were treated and separated by SDS–PAGE as described in (a) except that the incubation time for import was 10 min. For immunoblot analysis (IB), 12 μg of proteins from the wild-type (WT) or plsp1–1 were analyzed using antibodies against PsbO, PsbP or PetE. Precursor (pr), stromal processing intermediates (Si, Si1 and Si2), import intermediates (Ii1 and Ii2) and mature (m) forms are indicated on the right in (a) and (b). In (b), immunoreactive bands derived from isoforms of PsbO and PetE and an unknown protein are indicated by numbers (1, iPsbO2; 2, mPsbO2; 3, unknown; 4, iPetE1; 5, mPetE1). Black vertical lines indicate grouping of images from different parts of the same gel.

For the import assay in the presence of ATP, we incubated PsbO1 and PsbP1 variants in the light and PetE2 variants in dark as they were rapidly degraded in light, as discussed below. Under these conditions, all the lumenal protein variants were imported into chloroplasts (Figure 3a, lanes 3, 6 and 9) and were resistant to post-import treatment by trypsin, which degrades proteins outside the inner membrane (Figure S5) (Jackson et al., 1998). Interestingly, the migration patterns of imported proteins on SDS–PAGE varied between substrates. Imported PsbO1-ARAF and -EP exist as two bands (Figure 3a, PsbO1, lanes 6 and 9): the larger band corresponded to the Si form and the smaller band (designated as the import intermediate, Ii) migrated between the Si and mature forms (m). A previous in vitro import assay using pea chloroplasts demonstrated that substitution of Ala at the –1 position of wheat (Triticum aestivum) PsbO by various other residues resulted in alternative processing to yield a 36 kDa protein, which migrated between the Si and m forms, but did not yield Si (Shackleton and Robinson, 1991). The Arabidopsis PsbO1 import intermediate (Ii) generated by our assay may be equivalent to the 36 kDa protein from wheat PsbO, although it remains unknown why Si formation was detected for Arabidopsis variants but not for the wheat proteins. For PsbP1, both the ARAF and AP mutants yielded mainly Si1 after import assay: a smaller band corresponding to the mature form (m) was produced from ARAF but not from AP (Figure 3a, PsbP1, lanes 6 and 9). Finally, import of PetE mutants in dark resulted in three major bands, all of which were larger than Si, corresponding to the precursor (pr) and other two import intermediates (Ii1 and Ii2), which migrated between the pr and Si forms (Figure 3a, PetE2MV, lanes 6 and 9). This is quite different from the case of PsbO1 and PsbP1 mutants, which yielded intermediates comparable to or smaller than Si. We also noted that a small amount of a protein whose mobility was comparable to that of the mature form was produced from PsbO variants (Figure 3a, PsbO1, lanes 6 and 9). This may indicate that the mutation around the cleavage site did not completely inhibit processing in this assay, as was in the case of an E. coli leader peptidase substrate, maltose-binding protein (Fikes et al., 1990).

The data obtained suggest that inhibition of the TPP cleavage causes slower migration of imported proteins, including the Si and Ii forms, which were equivalent to and different from the SPP product, respectively. In order to further test this idea, we compared imported radiolabeled proteins and immunoreactive non-mature proteins in the plsp1-null mutant with respect to their mobility on SDS–PAGE. For the in vitro import assay, we chose XP mutants of PsbO1 and PsbP1 (PsbO1-EP and PsbP1-AP) and the ARAF mutant of PetE2 as the substrates. As shown in Figure 3(b), the major imported proteins derived from PsbO1-EP and PsbP1-AP migrated at the same rate as the major bands recognized by the anti-PsbO and anti-PsbP antibodies, respectively, in the plsp1-null mutant (lanes 4 and 6, Ii for PsbO, Si for PsbP). Similarly, two smaller bands derived from PetE2MV-ARAF after dark import corresponded well with two bands recognized by the anti-PetE antibody in the plsp1-null mutant (PetE2MV, lanes 4 and 6, labeled Ii1 and Ii2). These results indicate that the aberrant mobility of imported proteins derived from ARAF and XP mutants in vitro was due to disruption of TPP cleavage.

Inhibition of TPP cleavage does not affect thylakoid targeting but inhibits release from the membrane

A fractionation assay following in vitro import revealed that all imported non-mature proteins were localized to thylakoids (Figure S6). To test their membrane association, samples were treated by sonication, and the distribution of proteins after centrifugation was examined. As shown in Figure 4, all of the imported proteins were recovered mainly in the pellet fraction after hypotonic lysis (lane 4), similar to the endogenous proteins shown in Figure 2 except for non-mature PsbP in plsp1-null plastids, which was found in the soluble fraction. After sonication, non-mutated PsbO1 was distributed almost evenly to the soluble and membrane fractions, while 80% or more of non-mutated PsbP and PetE2MV were released from the membrane (Figure 4, WT, lane 5). The results for PsbO1 and PetE2MV were similar to those for immunoblotting, but imported PsbP1 showed a pattern different from that of the endogenous protein, being present in both the soluble and pellet fractions (Figure 2). This indicates that, under the conditions used for the assay, PsbO1 may have been efficiently integrated into PSII, as shown previously (Hashimoto et al., 1997), but PsbP was not. The inefficient integration of imported PsbP into PSII may be due to the different sources of proteins (Arabidopsis) and chloroplasts (pea). In the case of mutated proteins, the majority of those derived from PsbO1-EP and PetE2MV-ARAF, and approximately half of those derived from PsbP1-AP were recovered in the pellet after sonication (Figure 4, lane 6).

Figure 4.

Membrane association of imported TPP substrates. After import reactions, intact chloroplasts were re-isolated and used to prepare the chloroplast membrane fraction for sonication as described in Figure 2. Each lane contained 12.5 μg chlorophyll equivalent chloroplasts, and radiolabeled proteins were visualized by using a phosphorimager. The number indicates the relative intensity of protein signals in each lane relative to that in the total import fraction (imp). Signal intensities were measured using ImageJ ( Black vertical lines indicate grouping of images from different parts of the same gel.

Previously, an in vitro thylakoid transport assay yielded two types of membrane-bound cpTAT transport intermediates: one exposing its large portion to the stroma and another with its membrane translocation completed (Figure 5a, Ti1 and Ti2) (Frielingsdorf and Klösgen, 2007). We performed protease treatment to test whether this was also the case for the non-mature cpSEC and cpTAT substrates generated in the in vitro import assay. As a control, we showed that Plsp1, whose N–terminal soluble portion is located in the stroma (Figure 5a), was digested to a smaller fragment (Figure 5b, lane 4). The result of this control assay agreed with the results of a previous report (Shipman and Inoue, 2009), and also confirmed that the thylakoids maintained their right side-out orientation. Under these conditions, some of the mutated substrates, but none of the non-mutated variants, were partially degraded (Figure 5c). More specifically, among two imported proteins derived from PsbO1-EP, the larger band (Si) appeared to be degraded to a smaller fragment (SiD), but the Ii form appeared intact. Among the three bands derived from PetE2MV-ARAF, two larger bands (pr and Ii1) appeared to be digested to form a smaller band (Ii2). These results suggest that the majority of imported proteins were completely translocated (e.g. Ti2 in Figure 5a), but portions of some but not all PsbO1-EP and PetE2MV-ARAF intermediates were exposed to the stroma.

Figure 5.

Membrane topology of TPP substrates. (a) Membrane topology of non-mature lumenal proteins. The N- and C–termini are indicated. Thick bars represent the transmembrane domain in the mature portion of Plsp1 or the hydrophobic core within TTS (for Ti1 and Ti2). For Plsp1, the numbers indicate the amino acid residues at N- and C–termini of the mature protein as well as those of the transmembrane domain based on previous studies (Zybailov et al., 2008; Hsu et al., 2011). (b, c) After the import reaction, chloroplasts were re-isolated and lysed hypotonically. The pellet fractions after centrifugation were resuspended to 0.5 μg chlorophyll μl−1 in import buffer containing 10 mm CaCl2 without (mock) or with 0.2 μg thermolysin μl−1, and incubated in the dark on ice for 30 min. Protease activity was controlled by performing the assay in the presence of 2% v/v Triton X–100 (+TX). Arrowheads indicate digested products for Plsp1, PsbO1-EP and PetE2MV.

Taken together, these results indicate that the XP and ARAF mutations did not affect protein transport but prevented protein release from the thylakoid membrane in vitro.

Light-induced degradation of non-mature PetE2 requires ΔpH and ATP hydrolysis

As noted above, during the import assay, recovery of PetE2MV-ARAF but not that of PetE2MV was extremely low in the light but not in the dark (Figure 6a, lanes 2 and 5). As shown in Figure 6(b), imported PetE2 variants were stable after 20 min incubation in the dark (lane 3). However, in the light, under which the mature form was stable, all three imported bands derived from PetE2MV-ARAF disappeared gradually without yielding any detectable smaller fragments (lanes 4–6). Light illumination drives photosynthetic electron transport, generating a PMF across the thylakoid membrane and ATP in the stroma. We wished to determine whether ATP induces degradation of the protein in the absence of light, and whether the light-dependent degradation requires electron transport itself, PMF or ATP hydrolysis, or a combination of some or all of them. As shown in Figure 6(c), non-mature PetE2MV-ARAF was stable in the presence of ATP in the dark (lane 2), and its light-induced degradation was inhibited by the electron transport inhibitors 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) and methyl viologen (Mould and Robinson, 1991; Samuilov et al., 1995) (lane 4), the ionophores nigericin and valinomycin (Vredenberg and Bulychev, 1976) (lane 5), and a non-hydrolyzable ATP analog (adenosine 5′–[β,γ-imido]triphosphate, AMP–PNP; Sabbert et al., 1996) (lane 6). These results indicate that the degradation activity requires light-driven PMF and ATP hydrolysis. We next determined at which compartment, thylakoid or stroma, the PetE2MV-ARAF degradation activity is located. Interestingly, stromal extracts, unless they were boiled, enhanced the light-dependent degradation (Figure 6d, lanes 5–7) but did not affect protein stability in the dark (Figure 6d, lanes 3 and 4). This indicates involvement of proteinaceous components in the stroma during light-dependent degradation of non-mature PetE2.

Figure 6.

Light-dependent degradation of non-mature PetE2MV-ARAF. (a) Import reactions were performed as described in Figure 3(a) under dark or light conditions at room temperature for 20 min. After the reaction, chloroplasts were re-isolated, lysed, and fractionated into supernatant (S) and pellet (P) by centrifugation. The ‘tl’ lane contains 10% of the translation product used for the import assay. Each lane contained chloroplasts equivalent to 3.5 μg chlorophyll, and the radiolabeled proteins were visualized using a phosphorimager. (b) After import reactions in the dark for 20 min (imp), chloroplasts were re-isolated, resuspended in import butter and incubated under dark (D) or light (L) conditions for 2–20 min. Chloroplasts were re-isolated and subjected to SDS–PAGE separation followed by phosphorimager analysis. The ‘tl’ lane contains 10% of the translation product used for the import assay. Each lane contains chloroplasts equivalent to 5 μg chlorophyll. (c) Import and chase (20 min) assays were performed as described in (b) except that the buffer for chase contained ATP (3 mm), DCMU (15 μm) and methyl viologen (MV, 50 μm), nigericin (Nig, 0.75 μm) and valinomycin (Val, 1.5 μm), or AMP-PNP (adenosine 5′–[β,γ-imido]triphosphate, 3 mm) as well as 1% v/v ethanol, 3 mm MgCl2 and 12 mm LiCl. Chloroplasts were then re-isolated and subjected to SDS–PAGE separation followed by phosphorimager analysis. Each lane contained chloroplasts equivalent to 6 μg chlorophyll. (d) After import reaction in the dark for 20 min, chloroplasts were re-isolated and directly analyzed (imp) or hypotonically lysed. The resultant membranes were resuspended in import buffer to 0.5 μg chlorophyll μl−1 without (−S) or with (+S) stromal extracts prepared from chloroplasts equivalent to 4 μg chlorophyll μl−1, and incubated under dark or light conditions at room temperature for 20 min. +Δ, stromal extracts incubated at 98°C for 2 min.

Non-mature PsbO1-EP exists in a 440-kDa complex

A previous in vitro thylakoid transport assay demonstrated that non-mature cpTAT substrates exist either as monomeric or oligomeric forms (Frielingsdorf and Klösgen, 2007). To test whether this was also the case for non-mature lumenal proteins generated by the chloroplast import assay, we examined the mobility of the imported proteins on blue native PAGE (BN–PAGE). Chloroplast membranes were solubilized using 1% n–dodecyl-β–d–maltoside (DDM), which had previously been used to analyze the cpSEC complex (Zhang et al., 2001). Under these conditions, all transported proteins except those derived from PsbO1-EP were found near the dye front (Figure 7a, lanes 1, 3 and 4), suggesting that they exist as monomers. The majority of endogenous PsbO, which was shown to bind to PSII by the sonication assay (Figure 2), was also found to be monomeric (Figure S7), indicating that 1% DDM caused dissociation of the oxygen-evolving complex of PSII. In contrast to all other proteins tested, imported non-mature PsbO1-EP migrated at approximately 440-kDa (Figure 7a, lane 2). This complex was also found after dark import, in which imported PetE2MV-ARAF existed as a monomer (Figure S8). The level of the 440-kDa complex increased during the import time course (Figure 7b, BN, lanes 3 and 4). The two-dimensional BN/SDS–PAGE analysis confirmed that the 440-kDa complex contained the two non-mature forms (Si and Ii) but not the mature form (m) of PsbO1-EP (Figure 7c).

Figure 7.

Oligomeric formation of non-mature PsbO1-EP. (a) Import reactions were performed under light conditions for 20 min as described in Figure 3(a), and the products were analyzed by BN-PAGE (top, BN) and SDS–PAGE (bottom, SDS). For BN-PAGE analysis, chloroplasts were solubilized with 1% DDM for 20 min on ice in the dark. After centrifugation to remove insoluble materials, samples containing 3 μg chlorophyll were loaded onto a 7–16.5% BN-PAGE gel. Radiolabeled proteins were detected by phosphorimager analysis. Black vertical lines within the images indicate grouping of images from different parts of the same gel. (b) Chloroplasts containing imported proteins were separated by 4–14% BN-PAGE (BN) or 12% SDS–PAGE (SDS) after import for 10 or 20 min as described in (a). Each lane contained chloroplasts equivalent to 5 μg chlorophyll, and radiolabeled proteins were detected by phosphorimager analysis. Black vertical lines above the images indicate grouping of images from different parts of the same gel. (c) After BN-PAGE in (b), a lane was cut out, treated with 3.3% SDS and 4% β–mercaptoethanol, and subjected to 12% SDS–PAGE. The large arrow indicates the direction of the first electrophoresis, and the bar above the secondary dimension gel indicates the place of the BN-PAGE gel with the same scale as in (b). Intact chloroplasts containing 5 μg chlorophyll were loaded on the far left side of the secondary dimension SDS–PAGE gel. Radiolabeled proteins were detected by phosphorimager analysis. (d) Chloroplasts containing imported proteins were solubilized with 1% DDM, 3% digitonin (DIG), 1% Triton X–100 (TX100), or 6 m urea and 1% DDM. Samples containing 5 μg chlorophyll were separated by 5–12% BN-PAGE, and radiolabeled proteins were visualized by phosphorimager analysis. (e) After hypotonic lysis, chloroplast membranes containing imported proteins were treated with thermolysin as described in Figure 5 and solubilized with 1% DDM. Samples containing 5 μg chlorophyll were analyzed by 5–12% BN-PAGE. Samples without solubilization were also analyzed by SDS–PAGE. The susceptibility of proteins to the protease was confirmed by incubation with 1% v/v Triton X–100 (+TX100). Radiolabeled proteins were visualized by phosphorimager analysis analyzed by immunoblotting with the indicated antibodies. Black vertical lines above the images indicate grouping of images from different parts of same gel. (f, g) After import and solubilization as described in (a), samples were incubated with or without 10 or 100 mm DTT on ice for 1 h before separation by (f) BN-PAGE and SDS–PAGE separately, or (g) 2D BN/SDS–PAGE. Each lane contained chloroplasts equivalent to 5 μg chlorophyll, and radiolabeled proteins were detected by phosphorimager analysis.

As shown in Figure 7(d), the 440-kDa complex was also detected in chloroplasts solubilized using 1% Triton X–100 and 3% digitonin (lanes 2 and 3), but not in the sample solubilized using 6 m urea (lane 4), which disrupted most endogenous photosynthetic complexes (Figure S9). Interestingly, the level of the 440-kDa complex in digitonin-solubilized chloroplasts was approximately 10% of that in samples solubilized by other detergents, and the digitonin-solubilized sample contained an additional band at approximately 669 kDa (Figure 7d, lane 2). Digitonin is known to selectively solubilize proteins embedded in the stroma lamellae, but not those in the grana core (Järvi et al., 2011). Hence, low recovery of the 440-kDa complex may indicate grana localization of non-mature PsbO1-EP, and the larger band may contain an additional component that was dissociated by DDM and Triton X–100, but not by digitonin.

The 440-kDa complex remained largely intact after thermolysin treatment, under which endogenous cpSecY was degraded (Figure 7e, lane 3). The mobility of the 440-kDa complex did not match that of any of the major photosynthetic complexes (Figure S10). These results indicate that non-mature PsbO is not incorporated into the cpSEC translocon, which was shown to migrate at approximately 100 kDa on BN-PAGE (Zhang et al., 2001), or any of the PSII complexes.

Interestingly, when the solubilized chloroplasts were treated with dithiothreitol (DTT), the signal intensity of the 440-kDa band decreased (Figure 7f, BN, lanes 2 and 3) and that of the monomeric form increased (Figure 7g). The results of SDS–PAGE analysis confirmed that the decreased level of the 440-kDa complex was not due to degradation of PsbO by DTT treatment (Figure 7f, SDS). Under these conditions, the major photosynthetic complexes remained intact (Figure S10). These results suggest that disulfide bond formation is necessary for assembly of the 440-kDa complex containing non-mature PsbO.


Analysis of plsp1-null plastids and the in vitro import assay revealed the significance of protein maturation for thylakoid assembly

Establishment of the photosynthetic machinery requires proper synthesis, targeting, assembly and degradation of proteins in the thylakoid lumen. The TTS is essential for protein targeting, and its removal by TPP is critical for protein assembly. Previously, several studies had demonstrated that TTS removal is necessary for release of a subset of lumenal proteins from the membrane after translocation (Shackleton and Robinson, 1991; Di Cola and Robinson, 2005; Frielingsdorf and Klösgen, 2007). In this study, we used genetic and biochemical assays to provide further insights into the significance of TTS removal for turnover of lumenal proteins. First, analysis of plsp1-null plastids led us to conclude that the available data do not necessarily indicate direct involvement of Plsp1 in complete maturation of PsbP. Second, our results suggest that inhibition of complete maturation prevents release of soluble lumenal proteins from the membrane after their translocation. This is presumably due to the hydrophobic domain within TTS (Figure 1a), which may also act as a signal anchor (Hageman et al., 1990). Our results further demonstrate that membrane-bound non-mature lumenal proteins may have different fates, including light-dependent degradation (PetE2) and incorporation into a large oligomeric complex, which may require oxidative assembly (PsbO1).

These findings suggest a potential mechanistic scenario explaining the formation of anomalous thylakoids in plsp1-null plastids. During the early stage of chloroplast development, assembly of thylakoid membranes, including formation of the active cpSEC translocon, may be initiated without Plsp1, as evidenced by the presence of premature thylakoids in chloroplasts of newly developed pale green leaves (Shipman-Roston et al., 2010). However, lack of Plsp1 prevents TTS removal, thus inhibiting release of cpSEC substrates, such as PsbO and PetE, into the lumen after their translocation. Non-mature PsbO may be incorporated into a large oligomeric complex, and unprocessed PetE may be degraded. Consequently, assembly of the electron transport chain and thus generation of PMF may be disrupted. This is consistent with the severe effects of psbo1 psbo2 double knockdown (Yi et al., 2005) and the pete1 pete2 double knockout (Weigel et al., 2003) on photoautotrophic growth of plants. Without sufficient PMF, cpTAT substrates (e.g. PsbP) may remain in the stroma as non-mature proteins. However, the mutant plastids may contain a sufficient amount of ATP, either generated by oxidative pathways or imported from the cytosol, which may support active cpSEC transport. Newly transported PetE2 may be stably associated with the membrane because the membrane now lacks PMF, which is required for degradation of non-mature PetE2. Non-mature PsbO1 and PetE2 cannot support assembly of photosynthetic complexes due to their tight association with the membrane, causing accumulation of the anomalous thylakoids in the mutant. Interestingly, other cpTAT substrates FtsH2/8 (Rodrigues et al., 2011) were found in the membrane of plsp1-null plastids (Figure 2, αFtsH2/8), although their size was comparable to that of the stromal processing intermediate (Figure S2). These results suggest that mutant plastids may still contain an active cpTAT pathway, which cannot support PsbP transport, and that FtsH2/8 depend on Plsp1 for removal of the TTS. Alternatively, membrane sorting of FtsH2/8 in plsp1-null plastids may be independent of the cpTAT pathway.

Inhibition of TTS removal causes unusual processing and light-dependent degradation of PetE

Our in vitro import assay revealed aberrant processing and stability of non-mature PetE2. In the dark, at least three forms of non-mature PetE2 were found in the thylakoid membrane. Their slower migration compared to the stromal processing intermediate indicates that they were tar-geted to the thylakoid without removal of the transit peptide. This finding is consistent with the previous result showing that SPP cleavage was not essential for thylakoid transport and complete maturation of PetE, and that removal of the import signal inhibited thylakoid transport of this protein (Bauerle and Keegstra, 1991). As for light-dependent degradation of non-mature PetE2, it was shown previously that complete removal of TTS is necessary for copper binding to form holoPetE2 (Li et al., 1990), and that apoPetE in Chlamydomonas reinhardtii was quickly degraded when copper was deficient (Li and Merchant, 1995). Based on these data, we hypothesize that membrane-bound non-mature PetE cannot bind to copper and thus is prone to degradation. Our results indicate that the degradation activity requires PMF across the membrane and ATP hydrolysis. Among known proteases in thylakoids is FtsH, which was shown to degrade unassembled Rieske FeS protein, a subunit of the cytochrome b6f complex, under light conditions (Ostersetzer and Adam, 1997). In fact, FtsH uses ATP hydrolysis and its activity is stimulated by PMF (Akiyama, 2002). However, available data suggest that FtsH is most likely not responsible for degradation of non-mature PetE. FtsH activity was shown to be sensitive to EDTA but independent of the presence of stromal extracts (Ostersetzer and Adam, 1997; Yoshioka et al., 2006). By contrast, light-dependent degradation of non-mature PetE2 was not affected by 25 mm EDTA (Figure S11) but was stimulated by the presence of stromal extracts (Figure 6d). We are interested in testing the involvement of other thylakoidal proteases, such as Lon, DegP and EGY1 (Itzhaki et al., 1998; Chen et al., 2005; Ostersetzer et al., 2007), in degradation of non-mature PetE2.

Inhibition of TTS removal results in DTT-sensitive incorporation of PsbO into a 440-kDa complex

Non-mature PsbO1 was incorporated into a 440-kDa complex, which was distinct from PSII complexes or the cpSEC translocon. Its resistance to an added protease suggests that the complex may be embedded in the membrane or may be located mostly on the lumen side. Notably, DTT caused disassembly of the complex without producing any smaller complexes containing non-mature PsbO (Figure 7f,g). This suggests that a disulfide bond between PsbO and another protein in the lumen may play a role in formation of the 440-kDa complex. In addition to two Cys residues in TTS, PsbO1 contains two highly conserved Cys residues in its mature portion, which form an intramolecular disulfide bond that is necessary for stable protein folding (Tanaka et al., 1989). The intramolecular disulfide bond of PsbO was also shown to be necessary for its water solubility when produced in bacteria (Nikitina et al., 2008), but was found to be dispensable for PSII-binding and oxygen-evolving activity (Betts et al., 1996; Wyman and Yocum, 2005). Recently, Karamoko et al. (2011) showed that lumen thiol oxidoreductase1 (LTO1), a polytopic membrane protein with a thioredoxin-like domain, associated with PsbO in a yeast two-hybrid assay and catalyzed formation of a PsbO disulfide bond via its soluble thioredoxin domain in vitro. LTO1 was also found to be necessary for proper accumulation of PSII components in vivo (Karamoko et al., 2011). Hence, the 440-kDa complex may contain LTO1. We are currently preparing a transgenic plant that produces PsbO1-EP to test this possibility.

Protein maturation may play a regulatory role in protein quality control

Di Cola and Robinson (2005) used a transient expression assay to demonstrate the occurrence of reverse translocation of a completely processed lumenal protein back to the stroma, and suggested that TTS removal acts as a quality control point. Recently, we found that Plsp1 forms a complex with PGRL1 in Arabidopsis thylakoids (Endow and Inoue, 2013). PGRL1 is one of the key players in antimycin A-sensitive cyclic electron flow (Hertle et al., 2012). The pgrl1-null mutant does not show as severe a phenotype as the plsp1-null mutant (DalCorso et al., 2008), suggesting that complex formation with PGRL1 is not essential for Plsp1 activity. This leads to the possibility that PGRL1 may act as a negative regulator of Plsp1. Suppression of Plsp1 activity may result in accumulation of non-mature lumenal proteins, and thus prevent reverse transport of lumenal proteins. This scenario, although highly speculative, supports the biological relevance of our findings regarding the fates of non-mature lumenal proteins.

Experimental procedures

Preparation of cDNA constructs encoding TPP-unprocessable forms

Plasmids carrying coding sequences for PsbO1 (At5g66570), PsbP1 (At1g06680) and PetE2 (At1g20340) were obtained from the Arabidopsis Biological Resource Center ( (Yamada et al., 2003). Substitution of amino acid residues were performed by site-directed mutagenesis using these plasmids, primer sets listed in Table S1 and iProof high-fidelity DNA polymerase (Bio–Rad, After the reaction, template plasmids were digested using DpnI (New England Biolabs,, and the resultant materials were used to transform E. coli PirPlus DH10bpir116 cells (Thermo Scientific, to propagate plasmids. The identities of the plasmids were confirmed by sequencing the entire coding regions.

Chloroplast preparations from Arabidopsis

Chloroplasts were prepared from Arabidopsis wild-type (Col–0) or plsp1–1 (Inoue et al., 2005) grown on MS medium supplemented with 3% sucrose for 5 (wild-type) or 7 weeks (mutant) using the protoplast method as described previously (Fitzpatrick and Keegstra, 2001; Shipman-Roston et al., 2010). Chlorophyll concentration was determined as described previously (Arnon, 1949). For normalization, the protein concentration was determined as described previously (Bradford, 1976) using BSA as a standard. For hypotonic lysis, intact chloroplasts were resuspended in HM buffer (10 mm HEPES/KOH pH 8, 10 mm MgCl2) to 0.1 μg chlorophyll μl−1, followed by centrifugation at 16 000 g at 4°C for 20 min for separation into supernatant and pellet fractions. For sonication treatment, the pellet fraction after hypotonic lysis was resuspended in HM buffer to 0.1 μg chlorophyll μl−1, and sonicated at room temperature for 1 min in the dark using an Aquasonic 75T ultrasonic cleaner (VWR, before separation by centrifugation as described above.

Sources of antibodies and immunoblotting

Antibodies against FtsH2/8, PetE, PsbA, PsbO and PsbP were obtained from Professors W. Sakamoto (Research Institute for Bioresources, Okayama University, Japan), M. Pilon (Biology Department, Colorado State University, CO), E.–M. Aro (Department of Biochemistry and Food Chemistry, University of Turku, Finland), H.–M. Li (Institute of Molecular Biology, Academia Sinica, Taiwan) and S.M. Theg (Department of Plant Biology, University of California, Davis, CA), respectively. Antibodies against Toc75 and Tic110 were obtained from Professor K. Keegstra (MSU-DOE Plant Research Laboratory, Michigan State University, MI). Visualization of immunoreaction was performed by using alkaline phosphatase-conjugated antibodies against rabbit IgG from goat (Enzo Life Sciences, and 5-bromo-4-chloro-3′-indolyphosphate p-toluidine salt/nitro-blue tetrazolium chloride (Bio-Rad).

In vitro stromal processing and chloroplast import assays

For various in vitro assays, radiolabeled precursor proteins were prepared from cDNA constructs using a TNT® coupled reticulocyte lysate system (Promega, with [35S]methionine and SP6 RNA polymerase (for the Plsp1-coding plasmid; Inoue et al., 2005) or T3 RNA polymerase (all others), and diluted with an equal volume of 50 mm non-labeled methionine, 660 mm sorbitol, 100 mm HEPES/KOH, pH 8.0. Thus, each translation product contained the [35S]-labeled precursor protein, 25 mm non-labeled methionine and 25% v/v rabbit reticulocyte lysate in 1× import buffer (50 mm HEPES/KOH, pH 8.0, 330 mm sorbitol).

For stromal processing and import assays, chloroplasts were prepared from 11 to 15-day-old soil-grown pea (Pisum sativum ‘Little Marvel’) seedlings as described previously (Inoue and Keegstra, 2003). For the stromal processing assay, the pellet of isolated chloroplasts was resuspended in HM buffer to 1 μg chlorophyll μl−1, followed by centrifugation at 16 000 g at 4°C for 20 min. Ten microliters of the supernatant were mixed with 1 μl translation products, and incubated at room temperature for 90 min. The chloroplast import assay was performed as described previously (Inoue and Keegstra, 2003). For thermolysin treatment of the chloroplast membrane, the pellet after hypotonic lysis was resuspended into import buffer containing 10 mm CaCl2 at a final concentration of 0.5 μg chlorophyll μl−1, and incubated in dark on ice without or with thermolysin (0.2 μg μl−1) for 30 min. The protease reaction was terminated by addition of EDTA to 10 mm, and the chloroplast membranes were recovered by centrifugation at 3000 g for 8 min at 4°C.

Blue native PAGE analysis

BN-PAGE was performed as described previously (Hsu et al., 2012). Total chloroplasts were recovered by centrifugation at 2000 g for 5 min, resuspended in solubilization buffer [50 mm Tris/HCl, 500 mm 6–amino-n–caproic acid, 10% v/v glycerol, 10 μl protease inhibitor cocktail (P9599; Sigma-Aldrich, and 1% w/v DDM] at a final concentration of 0.5 μg chlorophyll μl−1, and incubated in dark on ice for 20 min. For DTT treatment, DTT was added to a final concentration of 10 or 100 mm in solubilization buffer, and the resuspended samples were incubated for 1 h instead of 20 min. After incubation, insoluble materials were removed by centrifugation at 16 000 g for 20 min at 4°C, and the chlorophyll concentration of the supernatant was measured and used for normalization.


We thank Professor R. Dalbey (Department of Chemistry, The Ohio State University) for the LepB construct and helpful advice on the assay, Professors E.–M. Aro (Department of Biochemistry and Food Chemistry, University of Turku), K. Keegstra (MSU-DOE Plant Research Laboratory, Michigan State University), H.-M. Li (Institute of Molecular Biology, Academia Sinica), M. Pilon (Department of Biology, Colorado State University), W. Sakamoto (Research Institute for Bioresources, Okayama University) and S.M. Theg (Department of Plant Biology, University of California at Davis) for antibodies, Professors D. Potter (Department of Plant Sciences, University of California at Davis) and J. K. Endow (Department of Plant Sciences, University of California at Davis) for critical comments on the manuscript, and Professor K. Cline (Horticultural Sciences Department and Plant Molecular and Cellular Biology, University of Florida), M. Mccaffery (Horticultural Sciences Department and Plant Molecular and Cellular Biology, University of Florida) and the Arabidopsis Biological Resource Center for cDNA clones. This work was funded by the Division of Chemical Sciences, Geosciences and Biosciences, Office of Basic Energy Sciences of the US Department of Energy through grant number DE-FG02-08ER15963.