The TL29 Protein is Lumen Located, Associated with PSII and Not an Ascorbate Peroxidase

Abstract

The TL29 protein is one of the more abundant proteins in the thylakoid lumen of plant chloroplasts. Based on its sequence homology to ascorbate peroxidases, but without any supporting biochemical evidence, TL29 was suggested to be involved in the plant defense system against reactive oxygen species and consequently renamed to APX4. Our in vivo and in vitro analyses failed to show any peroxidase activity associated with TL29; it bound neither heme nor ascorbate. Recombinant overexpressed TL29 had no ascorbate-dependent peroxidase activity, and various mutational analyses aiming to convert TL29 into an ascorbate peroxidase failed. Furthermore, in the thylakoid lumen no such activity could be associated with TL29 and, additionally, TL29 knock-out mutants did not show any decreased peroxidase activity or increased content of radical oxygen species when grown under light stress. Instead we could show that TL29 is a lumen-located component associated with PSII.

Introduction

Oxygenic photosynthesis in plants occurs in the thylakoid membranes of the chloroplast, that are folded into stacked regions, called grana, connected via stroma-exposed lamellae. The photosystems are heterogeneously inserted into the membrane, with an enrichment of PSII in the grana region while PSI is located at the margins of the grana regions and in the stromal lamellae (Albertsson 2001). Enclosed between the membranes is the continuous, aqueous space, called the thylakoid lumen. During photosynthetic electron transport a proton gradient is built up across the membrane, generating the proton motive force. Following the first isolation and characterization of the lumen content (Kieselbach et al. 1998), there has been growing interest in this cellular compartment with respect to regulatory and protective functions in photosynthetic processes. Proteomic studies have estimated that the Arabidopsis lumen fraction contains between 80 and 200 different proteins (Schubert et al. 2002, Peltier et al. 2002, Kieselbach and Schröder 2003). Several biochemical studies have indicated that distinct groups of proteins reside in the lumen, including proteases (Kirwin et al. 1988, Oelmuller et al. 1996) and immunophilins (Fulgosi et al. 1998, Gupta et al. 2002, Edvardsson et al. 2003). Together with recently acquired evidence of nucleotide metabolism in the lumen (Spetea et al. 2004), these findings are providing a more profound understanding of the functions of the lumen compartment. Proteomic and high-throughput genomic projects generate vast amounts of sequence data. One way to process this information is to group proteins based on homology, and assign them putative functions. However, it has to be kept in mind that of the approximately 27,000 gene products in Arabidopsis only roughly 10% (Arabidopsis Genome Initiative 2000) have been biochemically analyzed, and the putative functions of most of the proteins remain to be confirmed.

One of these uncharacterized proteins is the thylakoid lumen protein TL29. This 29 kDa protein was first identified in the thylakoid lumen isolated from spinach chloroplast by N-terminal sequencing (Kieselbach et al. 1998), and orthologs were subsequently found in tomato (Kieselbach et al. 2000), pea (Peltier et al. 2000) and Arabidopsis thaliana (gene locus At4g09010) (Kieselbach et al. 2000, Peltier et al. 2002, Schubert et al. 2002). The TL29 protein contains the characteristic bipartite signal peptide found in lumenal proteins. Protein import analysis has demonstrated that the twin arginine motif found in the pre-sequence directs the protein to the chloroplast lumen via the Tat pathway (Kieselbach et al. 2000). The protein’s sequence shows homology with ascorbate peroxidases (APXs) and the TL29 protein has therefore been designated a putative APX function (EC 1.11.1.11). APX proteins are found in various cellular compartments where they scavenge hydrogen peroxide using ascorbate as an electron donor. In Arabidopsis, two forms have been described to date that apparently reside in the chloroplast (Jespersen et al. 1997): the stromal sAPX and the thylakoid-bound tAPX. Based on sequence similarity to these APXs, the TL29 protein was recently renamed APX4 (Panchuk et al. 2002, Mittler et al. 2004, Panchuk et al. 2005). Expression patterns of the TL29 (APX4) gene have been analyzed in plants before and after heat stress treatments (Panchuk et al. 2002) and during senescence (Panchuk et al. 2005) to elucidate its putative function in oxygen radical scavenging; and its mRNA expression profiles have been compared with those of other APX gene transcripts. However, the peroxidase activity of the TL29 protein has never been characterized; therefore, the suggested function of the protein is entirely based on homology predictions.

The aim of the study reported here was to investigate whether the ‘putative APX’, the TL29 protein, indeed has peroxidase activity. Peroxidase activity was measured in lumen fractions isolated from wild-type and TL29 knock-out plants and from recombinant overexpressed TL29 protein. However, none of these analyses provided any evidence for a peroxidase activity associated with the TL29 protein and no ascorbate was found to be bound to the protein. TL29 knock-out plants did not show any significant phenotype, suggesting that the protein is not directly involved in photosynthetic electron transport. Instead here we show that the TL29 protein is a lumen-located subunit associated with PSII and not a peroxidase involved in ROS (reactive oxygen species) stress response.

Results

The TL29 homology to APX is low in important enzyme regions

Comparing the amino acid sequence of TL29 with that of seven known APXs of Arabidopsis by multiple sequence alignments showed that all important amino acids of the APX catalytic site were missing in TL29 (Fig. 1). These results were supported by the comparison of TL29 with the crystal structure of the ascorbate-binding peroxidase complex of recombinant APX1 from soyabean (rsAPX1) (Sharp et al. 2003). Neither the key catalytic residues in the distal heme cavity Arg38 and His42 (numbering referring to APX1), nor the proximal heme-binding His163 or His169 are conserved in the TL29 protein. Previously it has been shown by Macdonald and co-workers (2006) that two amino acids in APX are important for ascorbate binding; Arg172 is essential (Sharp et al. 2004) while Lys30 only plays a minor role. Furthermore, in all APXs there is another conserved residue, Cys32, which seems not to be directly involved in ascorbate binding. In TL29 only the less important Lys30, and not the Arg172, was found to be conserved. Despite the obvious lack of homology in important APX domains, TL29 has been assigned and regarded as a ‘putative ascorbate peroxidase’ (Panchuk et al. 2002, Mittler et al. 2004, Panchuk et al. 2005); peroxidase activity has been detected in the thylakoid lumen previously (Kieselbach et al., 1998). We therefore decided to investigate the peroxidase activity of TL29 in more detail as well as its ascorbate binding ability.

Fig. 1

Multiple sequence alignment of Arabidopsis ascorbate peroxidases and TL29. The proposed important key residues are marked (Sharp et al. 2003). The heme-binding residues, His163 and His169, are in red and the key catalytic residues, Arg38 and His42, are in cerise (numbering referring to APX1). The proposed ascorbate-binding residues, Lys30 and Arg172, are shown in blue, together with a conserved Cys32 which is not directly involved in ascorbate binding. Other important residues according to Sharp et al. (2003) are in gray. Clustal W was used for the multiple sequence alignment. The underlined part of the TL29 sequence represents the part that shows homology to APX. Yellow marked residues Lys38 and Asn42 were point mutated to arginine and histidine, respectively, to test the ability to initiate a heme binding or an ascorbate peroxidase activity in the TL29 protein.

The peroxidase activity in the chloroplast thylakoid lumen is not ascorbate dependent

To find out if the peroxidase activity detected in the thylakoid lumen (Kieselbach et al. 1998) was associated with the APX homolog, TL29, we purified the lumen content from wild-type Arabidopsis plants. The peroxidase activity in the lumen fractions was analyzed using two different substrates, ascorbate and 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS). As seen in Table 1, peroxidase activities were detected when either substrate was used. However, pre-treatment of the intact thylakoids with the protease thermolysin before isolation of the lumen fraction almost completely abolished the ascorbate-dependent peroxidase activity, but not the ABTS-mediated activity (Table 1). These results suggest that the detected activities are attributable to two different enzymes. The protein responsible for the ABTS peroxidase activity evaded proteolysis by thermolysin, due to its location in the lumen and the protection by the intact thylakoid membrane. The ascorbate-dependent activity, however, diminished after protease treatment, because the peroxidase was degraded by thermolysin. This enzyme therefore must be attached on the stromal side of the thylakoid membrane, accessible to thermolysin.

Table 1

Peroxidase activity in lumenal fractions isolated from wild-type Arabidopsis plants

	AsA (nmol ox ascorbate min−1 mg protein−1)	ABTS (nmol ox ABTS min−1 mg protein−1)
No. of treaments of thylakoids	861 ± 45 (100 %)	818 ± 17 (100%)
Thermolysin-treated thylakoids	71 ± 100 (8%)	822 ± 21 (100%)

The activity was measured with either ascorbate (AsA) or 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) as substrates.

To elaborate these results further, a gel-based ascorbate assay was performed (Mittler et al. 1993) on purified lumen samples. Purified rsAPX1, which has a cytosolic location in the cell (Sharp et al. 2004, Wong et al. 2004, Macdonald et al. 2006), was loaded as a control for the sensitivity of this experiment. While APX1 showed strong activity, at a protein amount of 0.1 μg (Fig. 2 left side), no APX activity was visible in the lumen sample, even if 400 times more (42 μg of protein) was loaded (Fig. 2, center). Thus, none of the different assays supported ascorbate-dependent peroxidase activity in the thylakoid lumen.

Fig. 2

In-gel ascorbate peroxidase activity measurements were performed in 12% native gels. Samples loaded are: rsAPX1 (left side), isolated lumen fraction (middle) and raTL29 (right side). The amounts of protein loaded are marked on each lane.

In vivo analysis of the proposed TL29 peroxidase activity

The predicted APX activity of TL29 in the thylakoid lumen was further investigated in vivo by generation of homozygotic TL29 knock-out Arabidopsis plants. The absence of the TL29 protein was confirmed using antibodies directed against both the N-terminus (Fig. 3A, inset) and the C-terminus of the protein (not shown). No obvious phenotypic deviations between wild-type plants and plants lacking the TL29 protein were detected (Fig. 3A) when grown on soil under either long (18 h) or short (8 h) day regimes and standard light conditions. Lumen fractions were isolated from the TL29 knock-out mutant and wild-type plants, and peroxidase activity was measured using the less specific, but more sensitive ABTS assay. As shown in Fig. 3B, no difference in peroxidase activity was observed in lumen extractions containing TL29 (wild type) or in its absence (knock-out mutant), supporting our previous finding that the lumen-located peroxidase activity is not associated with the TL29 protein.

Fig. 3

(A) Photographs of 7-week-old TL29 knock-out and wild-type plants. The inset image shows an immunoblot demonstrating the absence and presence of TL29, respectively, in these plants. (B) Peroxidase activities measured in lumen fractions isolated from wild-type Arabidopsis plants (solid bar) and TL29 knock-out plants (open bar). ABTS was used as substrate. Data represent means ± SD.

To eliminate the possibility that another, unknown peroxidase in the thylakoid lumen could compensate for the lack of TL29 in the knock-out mutant, keeping the total peroxidase activity unchanged, we performed a differential in-gel elecrophoresis (DIGE) analysis. In this type of experiment we were able to compare the relative protein content of lumen isolated from the TL29 knock-out mutant directly with that isolated from wild-type plants. After labeling the two samples with different CyDye™ Fluors, mixing and addition of an internal standard they were separated by 2D-PAGE. The gels were scanned and the differences in relative protein content could be evaluated. Performing a full DeCyder™ analysis with four replicate gels could not detect any significant (P < 0.05) down- or up-regulation of any proteins except TL29, which of course was missing in the lumen fraction of the knock-out mutant (Supplementary Fig. S1).

In vitro analysis of peroxidase activity of the purified TL29 protein

A more direct enzyme characterization was performed on overexpressed recombinant Arabidopis TL29 (raTL29), purified by its His tag (see Materials and Methods). Interestingly, when using the gel-based ascorbate assay (Mittler and Zilinska 1993) a weak activity was found associated with the protein (Fig. 2, right part), which was at least 10–20 times weaker than that observed in rsAPX1 (Fig. 2, left; note the much higher protein concentration for raTL29 than for rsAPX1). Interestingly we found that after removal of the N-terminal His tags from raTL29 (not shown), the very weak activity of TL29 was totally abolished; meanwhile APX1 still showed activity (not shown). The reason for this His tag- related activity is at present not clear to us, but as the Rz value (A₄₈₀/A₂₈₀) did not indicate any heme binding to the TL29 protein it might be due to unspecific association of metals or metal complexes with the His tag. In conclusion, again we could not support an ascorbate-dependent peroxidase activity of TL29.

The TL29 protein does not bind heme

APXs belong to the class I superfamily of heme peroxidases. Therefore, if TL29 was an APX it should be able to bind heme. In isolated and purified raTL29 no heme could be detected, neither after inspection by eye (appearance of red-orange color of the protein solution), nor by the Rz value. An Rz value of >2 indicates a fully native reconstituted protein with heme bound in a 1 : 1 ratio (Jones et al. 1998); the Rz value of purified raTL29 was 0–0.1. It should be noted that recombinant rsAPX1 had a large fraction of heme bound after its expression in Escherichia coli, which was clearly visible as a red-orange solution after breakage and clarification, in contrast to raTL29 which was light yellow (not shown).

In vitro reconstitution assays with hemin were performed on raTL29 and rsAPX1, and Rz values as well as peroxidase activities were compared. For rsAPX1 the Rz value increased from 0.7 to >2.1, which is in good agreement with the data reported earlier (Sharp et al. 2003). However, various reconstitution methods failed on the raTL29 protein, and its Rz value stayed close to zero and it did not gain any peroxidase activity (not shown).

Furthermore, point mutations were introduced into raTL29 to convert the enzyme into an APX by site-directed mutagenesis. Assuming structural resemblance to heme APXs (Fig. 1), a double mutant (K118R, N122H) was created to introduce the amino acids corresponding to Arg38 and His42 in APX1 (Fig. 1). After purification the mutated raTL29 protein was reconstituted with hemin as described before. However, the mutant did not show any tendency to be able to bind heme, and the Rz value never increased significantly from zero (not shown).

The TL29 protein does not bind ascorbate

As TL29, despite its homology to APXs, seems not to have any peroxidase activity, we wanted to investigate if, during evolution, it had lost this activity, but kept the ascorbate binding capacity and thus could be involved in ascorbate regulation. To investigate this hypothesis we used isothermal titration calorimetry (ITC), which enables the detection of small enthalpy changes upon molecular interactions (Heerklotz and Seelig 2000). When analyzing rsAPX1 a distinct negative enthalpy change (0.6 μcal min⁻¹) was observed after ascorbate injection, indicating the binding of ascorbate to the protein (Fig. 4A). In contrast, no enthalpy change was observed when ascorbate was added to raTL29 (Fig. 4B). The injections produced weak signals of similar size and the corresponding data could not be fitted, indicating that raTL29 is not an ascorbate binder. Also surface plasmon resonance (Biacore, Uppsala, Sweden) failed to detect any ascorbate binding by raTL29 (data not shown).

Fig. 4

Calorimetric investigation of TL29 binding to ascorbate. Ascorbate was injected (see Materials and Methods) at 25°C in 20 mM HEPES pH 7.0, 0.1 M NaCl, 10% glycerol containing rsAPX1 (A) and raTL29 (B). The final concentration of ascorbate was twice the protein concentration, assuming 1 : 1 stochiometry for ascorbate and protein. Negative μcal s⁻¹ represents an exothermic reaction and positive μcal s⁻¹ an endothermic reaction.

The TL29 seems not to be directly involved in ROS stress response

Excess light can lead to the formation of ROS, which can cause severe damage to cells; ascorbate as well as APXs have been suggested to be involved in cell responses to ROS. To investigate the role of TL29 in these processes we immunologically compared the amount of TL29 during the day and at different growth light intensities (high light, 290 μmol s⁻¹ m⁻²; normal light, 130 μmol s⁻¹ m⁻²; and low light, 40 μmol s⁻¹ m⁻²), but were not able to detect significant changes (not shown). We then compared the content of ascorbate in leaves of wild-type Arabidopsis plants and the TL29 knock-out mutant plants grown under low, normal and high light conditions. In the TL29 knock-out mutant grown at normal or low light intensity the total (reduced and oxidized) concentration of ascorbate did not differ from that of wild-type plants. A small, but not significant change was seen in mutant plants grown at high light (Fig. 5).

Fig. 5

Ascorbate content was measured at three different time points: morning before the lights were turned on (7.00 h), the middle of the day (11.30 h) and at the end of the day before the lights were turned off (16.30 h). Three different light conditions were also used: control light (CL) at 130 ± 20 μmol photons m⁻² s⁻¹, low light (LL) at 40 ± 10 μmol photons m⁻² s⁻¹ and high light (HL) at 290 ± 30 μmol photons m⁻² s⁻¹.

The TL29 is a lumen-located subunit associated with PSII

Our earlier work using biochemical methods (Kieselbach et al. 1998, Schubert et al. 2002) as well as in vitro import experiments had shown that the TL29 protein is located in the thylakoid lumen (Kieselbach et al. 2000). The microenvironment at the lumenal membrane surface is heterogeneous due to the organization of photosynthetic complexes in the thylakoid membrane, with PSII mainly being localized in the stacked thylakoid membrane regions (grana lamellae) and PSI in the stroma lamellae. To analyze the distribution of the TL29 protein between stacked and unstacked regions, stacked thylakoid membrane preparations were fractionated with digitonin into grana and stroma fractions, and the presence of the TL29 protein was analyzed by immunodetection. The results, shown in Fig. 6A, revealed a pronounced enrichment of the TL29 protein in the grana regions as compared with the stroma lamellae, where virtually no TL29 protein was detected. This distribution indicates that it specifically interacts with the PSII-enriched grana membranes since a more even distribution would be expected if the association had been unspecific. Two PSII proteins, the integral D1 and the extrinsic PsbO, were used as grana markers (Fig. 6A, middle and lower panels). These proteins showed the same pattern, with enrichment in the grana fraction although less exclusively for PsbO than for TL29 and D1. This was consistent with expectations since the thylakoid lumen has been shown to contain a pool of free PsbO proteins, amounting to ∼10% of the total thylakoid complement (Hashimoto et al. 1996).

Fig. 6

The association and distribution of TL29 were investigated in thylakoid membranes from A. thaliana. (A) Intact thylakoid membranes were treated with digitonin to yield grana fractions, intermediate fractions and stroma lamellae fractions. The Chl a/b ratios in the thylakoid, grana, intermediate and stroma lamellae fractions were 3.1, 2.8, 3.9 and 7.6, respectively. We used 5 μg of total chlorophyll per sample to analyse the distribution of TL29 by immunoblotting. (B) Thylakoids were fragmented in a Yedapress at pH 7.8, pH 6.4 and pH 5.4. Of the lumen sample, 0.5 μg of protein was loaded, whereas 5 μg of chlorophyll per membrane sample were loaded and separated by SDS–PAGE. After blotting, the TL29 protein was immunodetected in lumen fractions, thylakoid fragments and thylakoids.

To investigate the possible association of TL29 with PSII, we prepared PSII membrane fractions (BBY particles). The BBY particles were treated with 1 M NaCl, a treatment known to release PsbP and PsbQ proteins, and the collected supernatant was analyzed by SDS–PAGE. Fig. 7A shows the acrylamide gel, stained by silver nitrate. In BBY the TL29 protein was found to co-migrate with the bulk fraction of light-harvesting complex II (LHCII), and thus will not be detected by SDS–PAGE. However, after washing with NaCl, a protein band was observed that was identified immunologically to correspond to TL29 (Fig. 7B). Due to the low linearity using silver nitrate, this method is not appropriate to compare the amount of protein in different gel bands; however, it is worth noting that the band corresponding to TL29 was much weaker than the two extrinsic PSII subunits PsbP and PsbQ that are well known to be released by washing with NaCl. Immunoblotting of the different fractions further indicated that TL29 is present in BBY particles, but not in the salt-washed particles (Fig. 7B). These data show that at least part of the TL29 protein pool is associated with the PSII complex. The interaction seems to be of an electrostatic nature as high salt will remove the protein from the complex. Also clearly visible in Fig. 7B, where a longer SDS–PAGE was used resulting in higher resolution, is the detection of TL29 using the N-terminal-directed antibody. A double band is revealed, indicating that there are at least two forms of the protein with slightly different electrophoretic mobility. This observation is in agreement with the detection of different forms of the TL29 protein in previous studies of the Arabidopsis lumen proteome by 2D-PAGE analysis (Kieselbach et al. 2000, Peltier et al. 2002, Schubert et al. 2002); see also Supplementary Fig. S1. Using a C-terminal-directed antibody the same double band was detected (data not shown), discounting the possibility that the shorter form is due to proteolytic activity. It should be noted that if so-called ‘mini-gels’ with higher percentages of acrylamide were used, only one TL29 protein band could be seen (Fig. 6B). Thus, we estimate the maximum mass difference between the two forms to be <0.5 kDa. Currently, despite subjecting the protein to various mass spectrometric analyses or post-modification analyses, we do not know why two forms are observed in 1D-PAGE and four in 2D-PAGE. Most probably some post-translational modifications (such as phosphorylations and/or glycosylations) or, alternatively but less possible due to the small differences between them, some kind of splicing modes may be involved. However, the similar electrophoretic pattern occurs for recombinant raTL29, and therefore seems to be an intrinsic property of the polypeptide chain.

Fig. 7

(A) Silver-stained SDS–PAGE separation of isolated PSII particles (BBY) before and after washing with 1 M NaCl. The third lane shows the collected supernatant after washing (NaCl-wash). (B) The same samples as in A, imunodecorated with an antibody directed against the TL29 protein.

The acidification of the thylakoid lumen during exposure to light has been shown to affect the binding of lumenal proteins to the membrane. Violaxanthin de-epoxidase, for example, binds to the membrane only when the pH is below 6, reflecting its function in the xanthophyll cycle (Bratt et al. 1995, Hashimoto et al. 1996). Similarly, the TLP40 protein has been shown to interact more strongly with the membrane at low pH (Eshaghi 2001). To investigate further the nature of the TL29 binding, thylakoid membranes were equilibrated in buffers with pH 5.4, 6.4 and 7.8 before fragmentation and release of the lumen fraction. As shown in Fig. 6B, we did not observe any differences in the amounts of TL29 in these preparations, suggesting that the interaction with the membrane is not pH dependent.

The location of TL29 close to PSII prompted us to analyze PSII function in the TL29 knock-out plants. However, both PSII oxygen evolution and fluorescence measurements (PAM) showed no differences in comparison with wild-type plants, eliminating a direct involvement of TL29 in PSII electron transport. TL29 knock-out plants grown under different laboratory conditions but also under field conditions (H. Johansson-Jänkänpää, W. P. Schröder and S. Jansson, unpublished results) did not show any phenotypic differences measured in biomass, growth rate, flowering time or herbivore attack. Therefore, TL29 seems not to be directly involved in photosynthetic electron transport.

Discussion

For the majority of gene products found in various organisms, no functional data are available. Gene-based homology searches allow prediction of the putative function in case significant homologies exist to proteins with known roles. However, without—often tedious—biochemical proof one cannot be certain about a protein’s function.

The putative APX TL29 recently was renamed APX4 and suggested to be localized in the microsomal fraction of the cell by Panchuk and co-workers (Panchuk et al. 2002, Panchuk et al. 2005). However, in this report we provide biochemical evidence that TL29 (APX4) is localized in the thylakoid lumen, supporting earlier data based on proteomic studies of spinach (Kieselbach et al. 1998), tomato (Kieselbach et al. 2000), pea (Peltier et al. 2000) and Arabidopsis (Kieselbach et al. 2000, Schubert et al. 2002, Peltier et al. 2002) and on protein import analysis (Kieselbach et al. 2000), where we were able to show unambiguously that the TL29 protein is located in the thylakoid lumen of the Arabidopsis chloroplast. We also substantiate the location in the thylakoid lumen: a significant proportion of the TL29 protein pool interacts predominantly with grana membrane regions.

TL29 is not involved in the ROS protection system as assumed earlier (Panchuk et al. 2002, Panchuk et al. 2005). Under conditions where APX proteins usually are induced, i.e. high light stress, the amount of TL29 did not change. Additionally the level of ascorbate in the plant leaf did not change upon deletion of the TL29 protein. These data are consistent with the transcriptional analysis of TL29 showing that TL29 (APX4) is expressed differently compared with other APX genes (Panchuk et al. 2002, Panchuk et al. 2005). However, while Vanderauwera et al. (2005) reported a reduction in transcript levels for the TL29 (APX4) gene after 8 h of high light treatment in wild-type plants, we found that the protein level did not change during high light treatment (data not shown). Reductions in the transcript level of the TL29 gene (APX4) following high light treatment have also been reported by Mittler et al. (2004) although in the cited study the transcript in catalase-deficient plants was reported to be up-regulated (1993), in contrast to the findings of Vanderauwera et al. (2005). However, low correlations of transcription and translation have been reported for several proteins within the chloroplast; changes in protein level following various light treatments did not correspond to changes in mRNA amount of the same protein (Walters 2005).

The most important of our finding is that TL29 is not an APX. Despite large homologies between TL29 and other APX proteins, our alignment of TL29 and the other APX proteins revealed that all seven amino acids that are conserved in most APX proteins and are crucial for the function of the active site are absent in TL29. Only the least important of the two ascorbate-binding sites (Lys30, APX1 numbering) is present in TL29, being too weak for actual ascorbate binding. We used three different biochemical approaches to test whether TL29 is an APX: (i) the activity in ‘native’ wild-type systems was explored by isolating thylakoid lumen samples and analyzing their peroxidase activity with different assays; (ii) TL29 knock-out plants were obtained and their peroxidase activities were compared with those of wild-type plants; and (iii) overexpressed recombinant TL29 protein was in vitro reconstituted in the presence of hemin. However, none of these approaches supported APX activity of TL29. The fact that APX1 (rsAPX1) could be used as a positive control in our experiments strengthened our conclusions. rsAPX1 showed high activity and very good reconstitution with heme (Fig. 2), while reconstitution of TL29 failed. We further could show that no other unknown peroxidase compensates for loss of TL29’s putative peroxidase activity. The DIGE experiment, comparing protein abundance in the lumen isolated from wild-type plants and the knock-out mutant plants, did not reveal any changes. We also used site-directed mutagenesis trying to convert TL29 into an APX, without success. This result also did not give any support for TL29 being an APX.

However, it should be stressed that we and others detected peroxidase activity in the thylakoid lumen, as shown in this report and in earlier studies (Kieselbach et al. 1998). This activity was not abolished by protease treatment of the intact thylakoid membrane prior to obtaining the lumen fraction. The activity was not ascorbate dependent and it was not associated with the TL29 protein, since similar levels of activity were detected in the knock-out and wild-type plants. The origin of the lumenal peroxidase activity is presently under investigation and is not yet clear, but a similar activity has previously been detected in the thylakoid lumen of spinach (Kieselbach et al. 1998). Such an enzyme would be important to protect the photosynthetic machinery from oxidative damage produced by PSII (Andersson and Anderson 1980, Ljungberg et al. 1984, Arellano et al. 1994).

We have recently been able to obtain crystals of TL29, but initial efforts to solve the structure using the molecular replacement technique, which requires high structural similarity between model and target protein, have been unsuccessful (E. Lundberg, P. Storm, W. P. Schroder and C. Funk, unpublished results). Therefore, significant structural difference can be assumed between TL29 and the APX family members. Thus it seems that TL29 differs from APXs not only in function, but also in structure. We therefore suggest using its original name TL29 (Kieselbach et al. 1998, 2000) instead of APX4.

The data presented here clearly rule out that TL29 is an APX; however, the function of this protein remains unclear. As shown in Figs. 6 and 7, TL29 is at least in part associated with the lumenal side of the PSII complex and it seems almost exclusively located in the grana part of the thylakoids. TL29 could be washed off by NaCl, suggesting an electrostatic association. Furthermore, the amount of TL29 found seems to be substoichiometric compared with the extrinsic proteins of the oxygen-evolving complex. Therefore, TL29 protein might be an auxiliary protein of PSII, perhaps involved in stabilizing or assembling the lumenal side of PSII. Experiments to elucidate its function are in progress.

Materials and Methods

Plant material

Arabidopsis thaliana ecotype Columbia plants were grown either hydroponically in a nutrient solution (Noren et al. 1999) or in soil in 16/8 h dark/light cycles with a light intensity of 130 ± 20 μmol photons m⁻² s⁻¹ as standard conditions. To analyze TL29 expression under different light intensities, plants were grown in soil for 6 weeks under these conditions before being transferred to high (290 ± 30 μmol photons m⁻² s⁻¹) or low light (40 ± 10 μmol photons m⁻² s⁻¹) conditions for 3 d with the same photoperiod. A table fan placed behind an ice bag was used to keep the temperature down during high light treatment. With this arrangement the temperature rose to 27°C by the end of the light period, compared with 22°C for the control plants, for which a fan was also used, without the ice.

Isolation of the lumen fraction, thylakoid membrane fraction and BBY particles.

The lumen fractions were isolated as in Kieselbach et al. (1998) with modifications according to Schubert et al. (2002). When the lumen fraction was isolated to measure APX activity, 1 mM Na-ascorbate was added to all buffers. To analyze the possible pH dependence of TL29 with the membrane, the lumen fraction was prepared as above, but the final two wash steps before the Yedapress rupture were performed at pH 5.4 and pH 6.4 instead of pH 7.8 using 30 mM sodium phosphate, 5 mM magnesium chloride, 50 mM sodium chloride and 100 mM sucrose. Thylakoids used for TL29 expression analysis were prepared according to the same protocol omitting the steps following osmotic shock of the chloroplasts. Thylakoid membranes prepared to study oxygen evolution and fluorescence were washed once in 30 mM sodium phosphate pH 7.8, 5 mM magnesium chloride, 50 mM sodium chloride and 200 mM sucrose, and frozen in liquid nitrogen until use.

PSII particles (BBY) were prepared as described in Berthold et al. (1981) with modifications according to Ford and Evans (1983) and Arellano et al. (1994). Salt washes were performed as described in Ljungberg et al. (1984). The Chl a/b ratio of the BBY particles was 2.1 ± 0.1.

Fractionation of thylakoids with digitonin was performed on thylakoids as in Andersson and Anderson (1980). Preparations from the TL29 knock-out mutant and wild-type plants were performed at the same time to minimize sampling differences.

Chlorophyll contents in all fractions were determined according to Porra et al. (1989), and protein concentrations were determined using the Bradford assay (Bradford 1976) for lumen samples or according to Lowry et al. (1951) for thylakoid membrane samples.

SDS–PAGE, Western blotting and TL29 quantification.

SDS–PAGE was performed according to Laemmli (1970) using gels containing 17.5% polyacrylamide and 4 M urea for thylakoid membrane samples; 15% polyacrylamide gels were used for all other samples. Gels were either silver stained according to Heukeshoven and Dernick (1988) or electroblotted onto a polyvinylidene difluoride membrane using a semi-dry blotting system. Antisera were raised in rabbits against the N- or C-termini of TL29 using the N-terminal motif ADLIQRSEFQSDC (spinach sequence) and the C-terminal motif ACQKYQRSRETVSQTDC (Arabidopsis sequence), respectively, as antigens. The TL29 protein were detected using a horseradish peroxidase-conjugated goat anti-rabbit IgG. The result from the reaction was visualized using a CCD camera (Fujifilm, LAS-3000 V2.2, Stockholm, Sweden) and quantified using Science imager software (Fujifilm, Multi Gauge V3.1, Stockholm, Sweden). The silver-stained gels were used as loading controls and total protein levels were normalized using Image Master 1D-Prime software (Amersham Biosciences/GE Healthcare).

Construction, production and purification of recombinant TL29 and APX1

A full-length cDNA coding for TL29, clone U16014, was obtained from the Arabidopsis Biological Resource Center, DNA stock center, Ohio State University. The DNA sequence coding for the mature part of TL29 was amplified by PCR using the forward primer 5′-GACGACGACAAGATGGCTGACTTGAATCAAC-3′ and the reverse primer 5′-GAGGAGAAGCCCGGTTTATAGCTTGAGTTTG-3′, and inserted into a plasmid using the pET46 Ek/LIC cloning kit (Novagen, Whitehouse Station, NJ, USA). These code for a translation product with an N-terminal His6 tag to the mature TL29. Escherichia coli Fusion Blue (Novagen) was then transformed with the raTL29/pET46 plasmid by heat shock and an overnight culture was grown for preparation of glycerol stocks, and from which plasmid was purified (QIAprep Spin Miniprep kit, Qiagen) and sent for sequencing (MWG Biotech, Martiensried, Germany). Plasmid pQE30 encoding rsAPX1 with an N-terminal His6 tag was kindly supplied by Emma Raven, University of Leicester, UK (Sharp et al. 2004). The E. coli Rosetta2 (DE3) strain for overexpression was then transformed with the respective plasmid and grown overnight in 20 ml of LB medium at 37°C while shaking at 160 r.p.m. One liter of LB medium was inoculated with an overnight culture and grown under previous conditions until an OD₆₀₀ of ∼0.5, at which point the cultures were induced with 0.5 mM isopropyl-β-d-galactopyranoside (IPTG) and grown overnight (18 h) at 23°C with shaking at 160 r.p.m. Cultures were harvested by centrifugation at 3,000 × g, 4°C for 15 min. Bacterial pellets resuspended in 50 ml of breakage buffer [0.1 M HEPES pH 7.0, 0.5 M NaCl, 20 mM imidazole, 100 μM phenylmethylsulfonyl fluoride (PMSF), 10% glycerol) were broken on ice by 4 × 20 s pulsed sonications (0.5 s on, 0.5 s off, 60% amplitude) at 1 min intervals with a Branson Digital Sonifier 450 (Branson, USA). This solution was centrifuged at 40,000 × g for 20 min at 4°C. The supernatant was then loaded at room temperature onto a 1 ml HisGraviTrap Ni-IMAC column (GE Healthcare, Uppsala, Sweden) equilibrated with 10 ml of start buffer (20 mM HEPES pH 7.0, 0.5 M NaCl, 20 mM imidazole, 10% glycerol) and subsequently washed with start buffer containing 60 mM imidazole. The target protein was eluted in three 1 ml fractions using start buffer containing 0.5 M imidazole; the majority eluted in the second fraction, which was purified by the use of an S-100-HR size exclusion chromatography column coupled to an ÄktaPrime (GE Healthcare). About 25 mg of raTL29 and 29 mg of rsAPX1 was collected. Buffer was exchanged depending on downstream requirements at room temperature using a 5 ml HiTrap desalting column coupled to an ÄktaPrime. Point mutations of raTL29 were made with a QuikChange mutagenesis kit (Stratagene, La Jolla, CA, USA) according to the manufacturer’s protocol.

Reconstitution with hemin

Recombinant rsAPX1 and raTL29 were reconstituted with hemin (Frontier Scientific) according to Jones et al. (1998). In brief, 5 mg ml⁻¹ hemin in 0.1 M NaOH was added in 0.5 μl aliquots to 2 ml of protein solution, pH 6, while stirring at 4°C until the Rz value (A₄₀₈/A₂₈₀) was ∼2. Then a final aliquot of 1 μl was added before dialysis overnight at 4°C against 10 mM potassium phosphate buffer, pH 7.0. Precipitate was removed by centrifugation at 12 000 × g for 10 min at 4°C. To remove unbound hemin, rsAPX1 was purified by anion exchange chromatography (Jones et al. 1998) and raTL29 by size exclusion chromatography as above.

Isothermal titration calorimetry

The interaction between ascorbate and protein was determined with a VP-ITC MicroCalorimeter (MicroCal, Northampton, MA, USA) controlled by VPViewer 2000 software. The VP-ITC unit directly measures heat evolved (negative μcal s⁻¹) or absorbed (positive μcal s⁻¹) in liquid samples when reactants are mixed. Degassed sample buffer, also used for reference, consisted of 20 mM HEPES, pH 7.0, 0.1 M NaCl, 10% glycerol. After a 600 s initial delay, 1–2 mM freshly made ascorbate was injected in 40 μl aliquots (20 μl for rsAPX1) with 80 s duration (40 s for rsAPX1), 300 s spacing and a 2 s filter period into the sample chamber (1.4 ml) containing 0.1–0.2 mM protein in sample buffer at 25°C while stirring at 305 r.p.m. The final concentration of ascorbate, after injecting 280 μl, became twice the protein concentration, assuming a 1 : 1 stoichiometry for ascorbate and protein. Other settings were high feedback mode/gain and autoequilibration. The resulting data were fitted and plotted using MicroCal Origin 5.0 software. The sample for rsAPX1 was not reconstituted with hemin, but exchanged to the above buffer after purification.

Enzyme assays

Peroxidase activities were determined using ABTS as substrate according to Sigma enzyme assay protocol P6782, except that 1 ml reaction mixtures were used. Each 1 ml assay mixture contained 100 mM potassium phosphate, pH 5, 8.6 mM ABTS, 3 mM H₂O₂ and the enzyme fraction to be tested. The reaction was started by the addition of H₂O₂ and followed at 405 nm for 120 s in a Shimadzu spectrophotometer at room temperature. APX activity was measured by the decrease in absorbance at 290 nm due to ascorbate oxidation (2.8 mM⁻¹ cm⁻¹). Here, each 1 ml assay mixture contained 50 mM potassium phosphate, pH 7, 0.9 mM ascorbate, the test enzyme fraction and 1 mM H₂O_2. The reaction was started by the addition of H₂O₂ and followed for 120 s. In-gel APX activity in 12% native ready-made gels from BioRad was determineded following the procedure described by Mittler and Zilinska (1993).

Screening T-DNA mutants

A T-DNA insertion mutant (Garlic_519_E04. b.1a.Lb3Fa.) from Syngenta was used to screen for homozygotic insertion into the TL29 gene. Genomic DNA samples were isolated from wild-type and transgenic plants. From each plant, one leaf was cut, ground in liquid nitrogen and homogenized in 500 μl of buffer (200 mM Tris–HCl pH 7.5, 250 mM NaCl; 25 mM EDTA; 0.5% SDS). After centrifugation, 300 μl of the supernatant were transferred to a new tube, an equal volume of isopropanol was added and the samples were centrifuged again. The precipitated DNA was resuspended in 50 μl of water. A 1.5 μl aliquot was used for PCR with the following primers: forwardTL29 5′-CGAAGGGAGGTCCTA TTTCA-3′; reverseTL29 5′-CGGTTTCACGGCTTCTTTGA-3′; and left border T-DNA 5′-CAGAAATGGATAAATAGCCTT GCTTCC-3′. One leaf from each plant was also used to isolate a total cell extract and the presence of TL29 was additionally analyzed by immunoblotting after SDS–PAGE.

Oxygen evolution and room temperature fluorescence

Oxygen evolution was measured in thylakoid membranes prepared from wild-type plants and TL29 knock-out plants using a Clark-type oxygen electrode at 20°C. A 25 μg aliquot of chlorophyll was added to a 1 ml reaction mixture containing 0.4 mM phenyl-p-benzoquinone, 2 mM potassium hexacyanoferrate and 900 μl of buffer (0.2 M sucrose; 50 mM MES pH 6.5; 5 mM MgCl₂). Light intensities used in these oxygen measurements were 2,000 μmol photons m⁻² s⁻¹, and the rates were 200 ± 35 μmol oxygen mg⁻¹ h⁻¹ for both TL29 knock-out and wild-type thylakoids. Leaves of TL29 knock-out and wild-type Arabidopsis plants grown at 150 μmol photons m⁻² s⁻¹ were incubated in the dark for 1 h, and pulse-modulated chlorophyll fluorescence traces of PSII were obtained using the PAM-101 chlorophyll fluorescence measuring system (Heinz Walz GmbH, Effeltrich, Germany) (Baker 2008). The measurements were performed at room temperature with ambient CO₂ concentrations. PSII activity, monitored as the ratio of variable to maximum chlorophyll fluorescence [F_v/F_m =  (F_m − F_o)/F_m] of dark-adapted samples, was found to be 0.82 ± 0.02 in both types of samples. The effective quantum yield of photochemical energy conversion in PSII [ΔF/F_m′ =  (F_m′ − F)/F_m′], the photochemical quenching representing the fraction of open PSII reaction centers in the light-adapted state [qP =  (F_m′ − F)/(F_m′ − F_o′)] and the non-photochemical quenching as a measure of radiationless dissipation of absorbed light energy [NPQ =  (F_m − F_m′)/F_m′] were calculated from the traces (Rohacek 2002), but no change between TL29 knock-out and wild-type plants was found.

Determination of ascorbic acid

Arabidopsis leaves from wild-type and TL29 knock-out mutant plants were measured spectrophotometrically to determine the reduced ascorbate content as described in Barry et al. (1998). Each measurement was repeated four times.

Fluorescent labeling and DIGE analysis

Thylakoid lumen samples extracted from wild-type and TL29 knock-out plants from three biological replicates were compared using the fluorescent minimal CyDyes™ (Cy2, Cy3 and Cy5) developed for DIGE (GE Healthcare, Uppsala, Sweden). Following the minimal labeling procedure with a three-dye strategy, 50 μg of each protein sample was labeled with 200 pmol of amino-reactive cyanine dyes (Karp and Lilley 2005). Four samples (whereof three biological replicates) were separated with dye swap between wild-type Arabidopsis and TL29 knock-out mutant plant samples to avoid artifacts due to preferential labeling. The labeling reaction and solubilization procedures were performed according to the manufacturer’s instructions. Non-linear IPG strips, 24 cm long, pH 3–11 (GE Healthcare) were rehydrated overnight at 20°C, with 60 V applied. Isoelectric focusing (IEF) was performed using the Ettan IPGphor II apparatus (GE Healtcare) for a total 90,000 V h at 20°C and maximum current setting of 50 μA per strip. The IEF strip was equilibrated prior to SDS–PAGE for 15 min in 50 mM Tris pH 8.8, 30% glycerol, 6 M urea, 2% SDS and 2% dithiothreitol (DTT) and then for 15 min in the same equilibration buffer containing 4.5% iodoacetamide instead of DTT. SDS–PAGE was performed using the EttanDaltSix at 5 mA per gel for 1 h and then 14 mA per gel overnight. Gels were scanned using a Typhoon TM9400, Variable Mode Imager (GE Healthcare). Image gel analysis was carried out using DeCyder™ V 6.5 following the manufacturer’s instructions (GE Healthcare). Protein spots that showed a statistically significant change in abundance between control and treated samples using a Student’s t-test (P < 0.05) were considered as being differentially expressed in response to the TL29 knock-out mutant.

Thermolysin treatment of the lumen sample

The washed thylakoids (2 mg of Chl ml⁻¹) were incubated in the presence of 10 mM thermolysin (0.4 mg ml⁻¹) for 2 min on ice in 100 mM sucrose, 30 mM HEPES (pH 7.8), 50 mM NaCl, 5 mM MgCl₂ and 2 mM CaCl₂. The digestion was stopped by adding EDTA to a final concentration of 50 mM, and the thylakoids were washed twice with 100 mM sucrose, 30 mM HEPES (pH 7.8), 50 mM NaCl and 50 mM EDTA. These conditions were balanced to degrade the major part of peripheral proteins on the stromal side of the thylakoid membrane, without degrading too many of the integral membrane proteins. Harsher treatments were found to lead to leaky membranes followed by a loss of especially the lumenal part of the exposed stromal lamellae.

Supplementary data

Supplementary data are available at PCP online.

Funding

The Swedish Research Councils VR and FORMAS; Kempe Foundation (fellowship to P. Storm and financial support to C. Funk); the Wallenberg Foundation (financial support to C. Funk, the instruments and the bioinformatic infrastructure of the Umeå Protein Analysis Facility).

Acknowledgments

We would like to thank Dr. Emma Raven, University of Leicester, for the generous gift of the rsAPX clone, her help in expressing the protein and for discussing this manuscript. We also acknowledge Dr. Dmitry Sveshnikov for his help with the fluorescence measurements.

Abbreviations

Abbreviations
 
• APX

  ascorbate peroxidase

 
• ABTS

  2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid)

 
• DIGE

  differential in-gel electrophoresis

 
• DTT

  dithiothreiotol

 
• IEF

  isoelectric focusing

 
• IPTG

  isopropyl-β-d-thiogalactopyranoside

 
• ITC

  isothermal titration calorimetry

 
• PMSF

  phenylmethylsulfonyl flouride

 
• ra

  recombinant Arabidopsis

 
• ROS

  reactive oxygen species

 
• rs

  recombinant soyabean.