Barley cysteine protease PAP14 plays a role in degradation of chloroplast proteins.

Chloroplast protein degradation is known to occur both inside chloroplasts and in the vacuole. Genes encoding cysteine proteases have been found to be highly expressed during leaf senescence. However, it remains unclear where they participate in chloroplast protein degradation. In this study HvPAP14, which belongs to the C1A family of cysteine proteases, was identified in senescing barley (Hordeum vulgare L.) leaves by affinity enrichment using the mechanism-based probe DCG-04 targeting cysteine proteases and subsequent mass spectrometry. Biochemical analyses and expression of a HvPAP14:RFP fusion construct in barley protoplasts was used to identify the subcellular localization and putative substrates of HvPAP14. The HvPAP14:RFP fusion protein was detected in the endoplasmic reticulum and in vesicular bodies. Immunological studies showed that HvPAP14 was mainly located in chloroplasts, where it was found in tight association with thylakoid membranes. The recombinant enzyme was activated by low pH, in accordance with the detection of HvPAP14 in the thylakoid lumen. Overexpression of HvPAP14 in barley revealed that the protease can cleave LHCB proteins and PSBO as well as the large subunit of Rubisco. HvPAP14 is involved in the normal turnover of chloroplast proteins and may have a function in bulk protein degradation during leaf senescence.

Introduction

Chloroplasts contain the highest proportion of the total leaf protein content. Approximately 70% of the nitrogen in chloroplasts is bound in stromal proteins, whereby the majority is bound in Rubisco (Peoples and Dalling, 1988; Feller ). Most of the remaining 30% of the nitrogen is bound to proteins of the thylakoid-located photosynthetic apparatus. The protein composition of chloroplasts is variable and changes in response to environmental cues and development (Wollman, 2001; Ruban, 2015). During leaf senescence the majority of chloroplast proteins are degraded, allowing a massive recycling of nitrogen (Krupinska, 2006; Avila-Ospina ; Havé ).

Chloroplasts possess about 20 proteolytic machineries, most of which are of prokaryotic origin. They are involved in protein biogenesis, quality control, and remodelling of the photosynthetic apparatus (van Wijk, 2015; Nishimura ; Nishimura ). Surprisingly, among the many protease genes found to be up-regulated in transcriptomic studies on senescing leaves, the genes encoding components of these proteolytic machineries are lacking. The plastid proteases involved in senescence-associated degradation of chloroplast proteins, such as CND41, are regulated at the post-transcriptional level (Kato ). In case of the plastidic CLP protease, even down-regulation during senescence of barley flag leaves has been reported (Humbeck and Krupinska, 1996). Indeed, neither in studies with barley (Parrott ; Jukanti ; Hollmann ) nor in similar gene expression studies with Arabidopsis thaliana (Breeze ) have senescence-associated genes encoding proteases with a plastid target peptide for import by the canonical TOC/TIC system been found. Instead, many genes encoding members of the cysteine peptidase family that are synthesized at the endoplasmic reticulum (ER) were found to be up-regulated during senescence (Ueda ; Gepstein ; Andersson ; Guo ), such as the senescence marker gene SAG12 (Lohmann ).

It is well known that cysteine proteases are active during leaf senescence, as shown by inhibition with E-64 (Martínez ). The acidic pH optima of cysteine proteases indicate that they are active in the vacuole (Callis, 1995). The detection of senescence-associated genes encoding cysteine proteases is in accordance with the idea that bulk degradation of chloroplast proteins during senescence occurs in vacuoles (Martínez ; Kato and Sakamoto, 2013). Rubisco, after oxidative modification (Mehta ; Desimone ), was proposed to be partially cleaved inside chloroplasts by different proteolytic activities and thereafter delivered to the vacuole for final degradation (Feller ; Martínez ; Ono ). The delivery of Rubisco and other stromal proteins to vacuolar compartments involves at least three pathways (Xie ). The first pathway entails the formation of Rubisco-containing bodies (Chiba ) and autophagy (Wada and Ishida, 2013). A second pathway is independent of autophagy and involves a small proteolytic vacuolar compartment accumulating during leaf senescence, that is, senescence-associated vacuoles (Otegui ; Martínez ). More recently, a third pathway dependent on the chloroplast vesiculation protein (CV) has been described (Wang and Blumwald, 2014). CV-containing vesicles induced by CV were shown to contain stromal proteins, envelope membrane proteins, and thylakoid membrane proteins (Xie ).

HvPAP14 is the cysteine protease found to have the highest gene expression in senescent barley leaves (Hollmann ). The aim of this study was to investigate whether HvPAP14 plays a role during senescence-associated degradation of plastid proteins. Immunological analyses revealed that the protease accumulates in chloroplasts, where the active form was found in association with thylakoid membranes. Overexpression of HvPAP14 in barley confirmed the working hypothesis that HvPAP14 acts on chloroplast proteins located in thylakoid membranes. Moreover, our studies revealed that HvPAP14 contributes to the degradation of the large subunit of Rubisco.

Materials and methods

Plant material

Hordeum vulgare cv. Golden Promise was grown in a climate chamber in a 16/8 h day/night cycle on soil (Einheitserde ED73, Einheitswerk Werner Tantau, Uetersen, Germany). The light intensity was adjusted to 150 µmol s−1 m−2. Primary foliage leaves were collected 7–12 d after sowing for protein extraction, cell fractionation, and isolation of protoplasts. For the identification of proteases using the DCG-04 method, plants cultivated in a glasshouse were sown in standard soil and provided with supplemental artificial light (120 µmol s−1 m−2) in a 15 h/9 h day/night cycle. Non-senescent samples were collected 7 d after sowing and senescent samples were collected 21 d after sowing. The field trials were conducted at the Experimental Station Hohenschulen of the University of Kiel, Germany. For the field trials, grains were sown at a density of 300 plants m–2. The field was divided into plots of 3×8 m, which were supplied with a standard amount of nitrogen (80 kg ha–1) (Hollmann ).

Transgenic barley plants overexpressing HvPAP14 were generated by transformation of immature embryos of H. vulgare cv. Golden Promise with Agrobacterium tumefaciens containing the binary vector p6d35S-Ubi-HvPAP14. To create the vector, first, the HvPAP14 encoding sequence of the clone NIASHv2079N05 was amplified by PCR with the primers HvPAP14_for and HvPAP14_rev (see Supplementary Table S2 at JXB online) to integrate under the control of the maize UBIQUITIN-1 promoter (Christensen and Quail, 1996). Subsequently, a SfiI fragment containing the whole expression cassette was cloned into p6d35S-TE9 (DNA-Cloning-Service, Hamburg, Germany). The transformation of barley was performed as described by Hensel . To verify the presence of the transgene, PCR with primers for the hygromycin resistance cassette (GH-Hyg-F1 and GH-Hyg-R2; Supplementary Table S2) was performed. Southern blot analysis with genomic DNA prepared from six lines revealed that lines 4, 16, and 25 had different insertion sites of the transgene (Supplementary Fig. S6A). Homozygosity was demonstrated by hygromycin resistance of all 20 plants of the T2 progeny of a 3:1 segregating T1 progeny. Protein analyses were performed with the first true leaves of the T3 generation.

Determination of chlorophyll content

The relative chlorophyll content of leaves was determined at a position 2 cm above the base of the leaves using a Minolta SPAD-502 instrument (Konica Minolta Sensing, Osaka, Japan). The total chlorophyll content was determined in 80% (v/v) acetone as described by Porra .

Activity-based identification of cysteine proteases

DCG-04, a biotinylated derivative of the specific cysteine protease inhibitor E-64, was used to label active members of all cysteine proteases (Greenbaum ; van der Hoorn ; Martínez ). The labelling was performed as described by Carrión . The biotin-labelled cysteine proteases were either detected using streptavidin conjugated to horseradish peroxidase (Sigma-Aldrich, St. Louis, MO, USA) after SDS-PAGE and western blot, or purified using streptavidin-coupled magnetic beads and then identified by electrospray ionization–liquid chromatography/mass spectrometry (ESI-LC/MS). For this purpose, the labelled proteins were solubilized in sodium phosphate buffer [25 mM sodium phosphate, 0.05% (v/v) Triton X-100, 0.4 mM PMSF, 4 µM EDTA, 20 µM E-64, pH 7.4] and incubated with streptavidin-coupled magnetic beads (Sigma-Aldrich) for 1 h at room temperature with gentle shaking. The beads were washed in PBS-T [8 mM potassium phosphate buffer, 150 mM NaCl, 0.02% (v/v) Tween-20, pH 7.4], resuspended in Laemmli buffer (Laemmli, 1970), and boiled for 1 min before being loaded on to a SDS-containing 12% (w/v) polyacrylamide gel. After staining of proteins with Coomassie Brilliant Blue, spots (non-senescent and senescent) from stained protein bands (Fig. 1B) were excised from the gel, in-gel digested with trypsin, and subsequently analysed by ESI-LC/MS. The resulting mass spectra were processed and subjected to a database search against the UniRef90 protein database for plants, supplemented by data for keratin and trypsin. Identified protein sequences were cross-checked by sequence similarities to further datasets (Kohl ; The International Barley Genome Sequencing Consortium, 2012).

Fig. 1.

Identification of cysteine proteases in non-senescent (NS) and senescent (S) primary barley leaves. SDS-PAGE of cysteine proteases labelled with DCG-04, a biotinylated derivative of the specific cysteine protease inhibitor E-64. (A) Detection of labelled cysteine proteases by horseradish peroxidase (HRP)-conjugated streptavidin. The arrow indicates a protein band detected only in senescent leaves. The specificity of the approach was determined by preincubation of the samples with E-64. (B) Coomassie Brilliant Blue (CBB) staining of the labelled cysteine proteases after affinity purification using streptavidin-coupled magnetic beads. Protein bands with a similar size to those in (A) were excised (NS and S) and analysed by ESI-LC/MS. Results are presented in Supplementary Table S1. (This figure is available in colour at JXB online.)

Protoplast isolation and transformation

Protoplasts were isolated from primary foliage leaves collected from barley plants 12 d after sowing. After removal of the lower epidermis, the leaves were incubated in digestion buffer [1% (w/v) Cellulase R-10, 0.3% (w/v) Macerozym R-10, 20 mM succinate, 0.5 M sorbitol, 0.5 mM KH2PO4, 1 mM KNO3, 1 mM MgSO4, 1 mM EDTA, 0.025% (w/v) BSA, pH 5.7] for 1 h at room temperature. The protoplast suspension was filtered through a nylon gauze (80 µm mesh size) and centrifuged at 63 g for 5 min. The pellet was underlaid with flotation buffer (1 mM CaCl2, 1 mM sucrose, 0.1 M sorbitol, 25 mM MES, pH 5.8) and overlaid with W5 solution (154 mM NaCl, 125 mM CaCl2, 5 mM KCl, 5 mM glucose, pH 5.8). After centrifugation at 80 g for 10 min (low acceleration and brake) the intact protoplasts were collected from the interphase. For polyethylene glycol (PEG)-mediated protoplast transformation, 4 × 106 protoplasts were sedimented by centrifugation at 80 g for 5 min and resuspended in 500 µl MaMg solution [0.8 M mannitol, 0.03 M MgCl2, 0.2% (w/v) MES, pH 6.5]. Plasmid DNA (40 µg) and 500 µl PEG solution [40% (w/v) PEG 1500 in MaMg solution] were added to the protoplasts before incubation for 30 min with gentle mixing. After stepwise addition of 5 ml W5 solution in 1× 1 ml and 2× 2 ml aliquots, the protoplasts were sedimented at 80 g for 10 min and resuspended in 1 ml W5 solution. The protoplasts were incubated at 22 °C overnight in darkness, microscopically analysed for intactness, frozen in liquid nitrogen, and used for protein extraction and immunoblot analyses.

For fluorescence microscopic analyses, the constructs used for transient transformations were pUbi:PtGFP, pUbi:ER-GFP (with the coding sequence of mGFP5ER4), and pUbi:HvPAP14:RFP, containing the maize UBIQUITIN-1 promoter. The HvPAP14 coding sequence of the NIASHv2079N05 clone was amplified by PCR. The red fluorescent protein (RFP) coding sequence was inserted into the HvPAP14 coding sequence. Thus, RFP was located in the resulting protein, between Val306 and Lys307 of HvPAP14 (Fig. 3A). The primers used were PAP14N_for and PAP14N_rev for the sequence encoding the N-terminal part, and PAP14C_for and PAP14C_rev for the sequence encoding the C-terminal part of HvPAP14 (see Supplementary Table S2 for details of the primers). The transformed protoplasts were analysed by confocal laser scanning microscopy using a Leica TCS SP5 microscope (Leica Microsystems GmbH, Wetzlar, Germany). Excitation and emission wavelengths differed for RFP (λ ex=543 nm, λ em=590–615 nm), green fluorescent protein (GFP) (λ ex=488 nm, λ em=505–530 nm), and chlorophyll (λ ex=633 nm, λ em=660–700 nm).

Fig. 3.

Microscopic analyses of the subcellular localization of HvPAP14. (A) Scheme of the HvPAP14:RFP fusion protein. RFP was inserted into the HvPAP14 sequence N-terminal to the HDEL sequence. (B) Fluorescence microscopic images of protoplasts from primary barley leaves that were transiently transformed with the HvPAP14:RFP construct and an ER-GFP or GFP construct. Scale bars=5 µm. (C) Immunogold labelling of leaf sections derived from a specimen fixed by high-pressure freezing followed by freeze substitution. Sections were prepared from non-senescent (I) or senescent (II, III) barley leaves collected on 2 June (non-senescent) or 23 June 2010 (senescent) from field plots. Immunogold labelling was performed with antibody raised to HvPAP14. (I, II) Overviews; (III) details showing thylakoids of broken plastids. Scale bars=500 nm.

Protein extraction and immunoblot analysis

Total proteins from barley leaves were extracted as described previously (Pötter and Kloppstech, 1993). The protein concentrations were determined using Bradford reagent (BioRad Protein Assay, Hercules, CA, USA).

The proteins were separated by SDS-PAGE and transferred on to a nitrocellulose membrane. Proteins were stained with colloidal Coomassie G-250 (Dyballa and Metzger, 2009). For the immunological detection of proteins, polyclonal antibodies to HvPAP14 and HvPAP14i (BioGenes, Berlin, Germany), SAG12, RbcL, LHCB1, LHCB5, PSBO, and PsaA (all from Agrisera, Vännäs, Sweden) were used. Immunoreactive proteins were visualized with a peroxidase-coupled secondary antibody employing chemiluminescence (ECL Select Amersham, Pierce Therma Scientific Waltham, MA, USA; Ultra TMA-6 Lumigen, Southfield, MI, USA).

Isolation of chloroplasts and subfractions

Chloroplasts were isolated from barley leaves as described by Gruissem . For the isolation of chloroplast membrane and stroma fractions, chloroplasts were incubated in hypotonic buffer (20 mM HEPES/KOH, pH 7.6, 2.5 mM EDTA, 5 mM MgCl2) for 5 min on ice. Membranes were collected by centrifugation at 10 000 g for 10 min at 4 °C. This fraction was further fractionated into grana thylakoids, margins, and stroma thylakoids as described by Anderson and Boardman (1966). The preparation of thylakoid membranes and thylakoid lumen from chloroplasts was performed as described by Sokolenko .

Recombinant HvPAP14

The sequences encoding mature HvPAP14 and maltose-binding protein (MBP) were fused and inserted into the vector pMAL-c5X. Cloning, expression in Escherichia coli, and protein purification were done by using the pMAL™ Protein Fusion and Purification System (NEB, Frankfurt am Main, Germany).

The activation of HvPAP14 was measured in pH buffer (200 mM NaCl, 1 mM EDTA, 20 mM HEPES, 20 mM MES, 20 mM Tris, 20 mM sodium acetate, pH 3.5–9) with the fluorescing substrate Z-FR-AMC (20 µM) (Bachem, Bubendorf, Switzerland). The substrate turnover was analysed ratiometrically using the emission at 440 nm and 393 nm after excitation at 345 nm. For the activity measurements, HvPAP14 was activated in a buffer with a pH of 4.5 and then incubated in a 10-fold volume of pH buffer (pH 3.5–9) with Z-FR-AMC (final concentration 20 µM).

Immunogold labelling and transmission electron microscopy

For immunogold analysis, leaf samples were fixed by high-pressure freezing followed by freeze-substitution. Leaf pieces (2 mm in diameter) were punched out of leaves at a position 1 cm below the leaf tip. The pieces were incubated in a solution of 8% (v/v) methanol in tap water containing 100 mM betaine monohydrate (B2754; Sigma-Aldrich, USA), degassed for 15 min at 8 kPa, and frozen in a Bal-Tec HPM010 high-pressure freezing machine (Bal-Tec, Leica Microsystems, Germany). Freeze-substitution was performed in an EMAFS (Leica Microsystems) at –90 °C in fixative (0.2% glutardialdehyde, 0.1% uranyl acetate in 100% acetone) for 47 h and at –60 °C for 10 h (slope +2 °C). The samples were washed in pure ethanol and infiltrated in a graded series of ethanol/LR white resin (London Resin Company, Reading, UK) at –4 °C (slope +3 °C). Embedding in LR white resin was at –4 °C and polymerization in gelatine capsules was at 50 °C. Ultrathin sections were cut with a diamond knife in an Ultracut UCT ultramicrotome (Leica Microsystems).

Ultrathin sections of the embedded leaf pieces were blocked in blocking buffer consisting of PBS supplemented with 0.2% (w/v) fish gelatine (Sigma-Aldrich) and 0.2% (w/v) BSA (Type V; Sigma-Aldrich). After incubation with the first antibody (α-HvPAP14), the sections were washed and then incubated with anti-rabbit IgG coupled to 10 nm gold particles (Aurion, Wageningen, The Netherlands). Finally, the sections were stained for 10 min with uranyl acetate (Watson, 1958) and lead citrate (Reynolds, 1963). Observations were made with a Philips CM10 TEM and Tecnai G2 Biotwin (Phillips Electron Optics, Eindhoven, The Netherlands).

Accession number

HvPAP14 refers to the protein sequence CAQ00109.1, available in NCBI GenBank (https://www.ncbi.nlm.nih.gov/genbank).

Results

Activity-based identification of senescence-associated cysteine proteases

To identify cysteine proteases in senescing barley leaves, an activity-based labelling approach (Greenbaum ; van der Hoorn ; Martínez ) was applied. DCG-04, a biotinylated derivate of the inhibitor E-64, binds to the active sites of cysteine proteases. After incubation of protein extracts from non-senescent and senescent leaves (which had ~70% of the chlorophyll content of non-senescent leaves) with DCG-04, proteins were separated by SDS-PAGE and the DCG-04-labelled peptidases were detected using streptavidin conjugated to horseradish peroxidase (Fig. 1A). Four bands, with molecular weights of 48, 44, 35, and 28 kDa, were detected in crude non-senescent leaf extracts (Fig. 1A), with the 35 kDa band being most abundant. When the labelled proteins from extracts of senescent leaves were analysed, an abundant band with a molecular weight of 32 kDa was also detected. The specificity of this approach was checked by preincubation of leaf extracts with E-64, reducing the binding of DCG-04 (Fig. 1A).

To isolate the proteins labelled by DCG-04, sepharose-coated magnetic beads were used. The areas representing proteins with molecular weights of 32–35 kDa from non-senescent and senescent leaves were excised from the gel (Fig. 1B). After digestion with trypsin, peptides were analysed by ESI-LC/MS. The resulting peptide spectra could be assigned to eight protein sequences by a database comparison (Supplementary Table S1). Five of the eight proteins were peptidases, comprising one metallopeptidase and four cysteine peptidases. Two of the cysteine proteases (HvPAP14 and HvSAG12) were exclusively found in the sample from senescent leaves. HvSAG12 is a well-known senescence marker (Lohman ). HvPAP14 was one of the most up-regulated genes in a microarray experiment analysing the senescence of field-grown barley (Hollmann ). Two other peptidases, HvPAP6 and the thiol protease aleurain, were detectable in both senescent and non-senescent leaf extracts.

HvPAP14 accumulates during leaf senescence

For immunological analyses, a polyclonal antibody was raised against the peptide C-AEDAYPYKARQASS located in a region shared by the mature protease and all other forms of HvPAP14 (Fig. 2A). The specificity of the antibody was shown by competitive immunoblot assays with extracts from barley primary foliage leaves and the specific peptide (Supplementary Fig. S2A). Furthermore, a second polyclonal antibody (HvPAP14i) raised against the peptide C-QHTVARDLGEKARR, found in the inhibitory pro-peptide of HvPAP14, recognizes the inactive pro-form. Recombinant HvPAP14:MBP was detectable with both antibodies (Supplementary Fig. S2B).Samples prepared from flag leaves comparable to those used for gene expression analyses (Hollmann ) and characterized by their relative chlorophyll contents (determined by SPAD measurements) (Fig. 2B) were used for immunoblot analysis. The two prominent proteins, with molecular weights of approximately 40 kDa and 32 kDa (Fig. 2C), had approximately the predicted protein masses of the potentially inactive pro-forms, either with or without the ER retention sequence (37 kDa and 33 kDa). While the 40 kDa protein accumulated during senescence, the abundance of the 32 kDa protein was rather stable (Fig. 2C). After longer exposure of the immunodecorated blot, a 26 kDa protein was detected, which is likely the potentially active mature form. Its abundance was found to be rather stable during leaf development (Fig. 2C).

Fig. 2.

Immunological analysis of HvPAP14 abundance at different stages of barley flag leaf senescence. (A) Predicted forms of HvPAP14. Calculated sizes of the pro-peptidase, its processed forms, and the mature peptidase are indicated. The positions of peptides chosen for antibody production are highlighted in grey. (B) Relative chlorophyll contents determined by SPAD measurements of flag leaves. Error bars indicate SD (n=100). (C) Immunodetection of HvPAP14 in protein extracts from flag leaves of field-grown barley. Samples contained the same amount of protein. Leaf material was collected in field plots from 4 to 28 June 2010. The putative forms of the HvPAP14 protein are indicated. Asterisks indicate the exposure times of the immunodecorated blots: * 4 min, ** 10 min. For comparison, an immunoblot of SAG12 is shown. The prominent band in the Coomassie Brilliant Blue (CBB) gel corresponds to the molecular weight of the large subunit of Rubisco (RbcL). (This figure is available in colour at JXB online.)

For comparison, the amount of the senescence-associated cysteine peptidase HvSAG12 was determined immunologically. Both the 40 kDa pro-enzyme and the 32 kDa processed form of HvSAG12 accumulated during senescence of the flag leaves, with similar kinetics to the 40 kDa pro-enzyme of HvPAP14. Coomassie Brilliant Blue staining revealed that at the same time as these proteins accumulated, the abundance of the large subunit of Rubisco (RbcL) decreased (Fig. 2C).

HvPAP14 is targeted to the endoplasmic reticulum and accumulates in chloroplasts as well as microbody-like structures

With regard to the N-terminal signal peptide and the C-terminal ER retention motif (HDEL), HvPAP14 is predicted to be synthesized at the ER and also to reside in this compartment. To investigate its subcellular localization in more detail, barley protoplasts were transiently transformed with the fusion construct of the full-length cDNA of HvPAP14 fused to the RFP gene (Fig. 3A).

The RFP signal of the HvPAP14:RFP fusion protein was observed in association with a net-like structure distributed all over the protoplasts as well as in small vesicular bodies. The structures showed a high degree of co-localization with ER-GFP, which marked the ER (Fig. 3B). In comparison, the fluorescence signal from free GFP was uniformly distributed in the cytoplasm and nucleus. The HvPAP14:RFP fluorescence did not overlap with either marker GFP signals of the Golgi apparatus or with GFP signals specific for peroxisomes (Supplementary Fig. S3).

To analyse the subcellular localization of HvPAP14 in more detail, immunogold labelling was performed on ultrathin sections of non-senescent (100% chlorophyll) and senescent (80% chlorophyll) barley flag leaves collected from the same field as was used for protein analyses (Fig. 2B, C). By immunogold labelling using the antibody to HvPAP14, gold particles were observed inside chloroplasts as well as in small bodies of ~500 nm diameter (Fig. 3C). Labelling of chloroplasts was similar in non-senescent and senescent tissue, where the organelles contained large plastoglobuli (Fig. 3C). The gold particles detected in chloroplasts by the HvPAP14 antibody appeared to be associated with thylakoids, as they stayed attached to the membranes of ruptured chloroplasts (Fig. 3C).

Different forms of HvPAP14 show different subcellular localization

To gain insight into the subcellular distribution of the different forms of HvPAP14, immunoblot analyses were performed with protein fractions prepared from non-senescent primary foliage leaves. Proteins were isolated from leaves, chloroplasts, and membrane and stroma fractions derived from chloroplasts. The thylakoid membrane fraction was further fractionated into grana thylakoids, margins, and stroma thylakoids (Fig. 4A).

Fig. 4.

Abundance of HvPAP14 in protein fractions from barley primary foliage leaves. (A) Immunoblot detection of HvPAP14 in total protein extracts (TP), chloroplasts (CP), plastid membranes (Me), and stroma (St). Thylakoids were further fractionated into grana thylakoids (G), margins (M), and stroma thylakoids (ST). The putative forms of HvPAP14 protein are indicated. As a control, Ponceau staining of the proteins is shown. (B) Detection of HvPAP14 in association with the thylakoid membrane (TM) and in the thylakoid lumen (Lu) prepared from chloroplasts. For comparison, immunoblots with antibodies to PSBO and PsaA are shown. (This figure is available in colour at JXB online.)

In the total protein extract, the 40 kDa pro-form of HvPAP14 (Fig. 2A) was detected. Furthermore, an abundant form, most likely representing the 32 kDa form lacking the C-terminal end with the ER retention motif (HDEL) (Fig. 2A), was observed in total protein extracts, isolated chloroplasts, and chloroplast membrane fractions (Fig. 4A). The 26 kDa mature form of the peptidase was identified in chloroplasts and in the chloroplast membrane fraction (Fig. 4A). None of the HvPAP14 forms could be detected in the stroma. Further analyses of the chloroplast membrane fraction showed that the 32 kDa and 26 kDa forms of HvPAP14 were enriched in grana thylakoids and margins but were hardly detected in stroma thylakoids (Fig. 4A).

These immunoblot analyses clearly indicated that both the predicted pro-enzyme lacking the HDEL motif and the mature form of HvPAP14 are associated with chloroplast membranes. To identify the location of the HvPAP14 forms more precisely, thylakoids were further fractionated. Both the 32 kDa and 26 kDa forms of HvPAP14 were detected in the thylakoid membrane fraction, but only the 32 kDa form was detected in the thylakoid lumen (Fig. 4B). The purity of the fractions was demonstrated by immunological detection of the lumen-located subunit of the oxygen-evolving complex PSBO and the thylakoid-membrane-located core protein of photosystem I (PsaA) (Fig. 4B). Although most of the PSBO protein was detected in the lumen fraction, a small amount was associated with the thylakoid membranes, as reported previously (Bricker and Frankel, 2011). PsaA was detected exclusively in the membrane fraction (Fig. 4B).

Processing of recombinant HvPAP14

To investigate its functional properties, the mature form of HvPAP14 was synthesized in E. coli in fusion with MBP. The recombinant protein was affinity purified by binding of the MBP to an amylose resin and detected by the specific antibody described in the Materials and methods (see Supplementary Fig. S2B). The structure of HvPAP14 (Fig. 2A) suggests that the N-terminal inhibitory pro-enzyme domain needs to be cleaved off for processing and activation of the enzyme. This step could be pH-dependent, as has been shown for other members of the C1A group of cysteine proteases (Taylor ; Schmid ; Wiederanders ). The activity of the recombinant protein was therefore tested at different pH values using the fluorescent substrate Z-FR-AMC. The results of these assays clearly showed that the enzyme was processed only at pH <6, with the highest activity around pH 4–4.5 (Fig. 5A). After processing, however, the protease activity was independent of pH in the range of pH 4.5–7 (Fig. 5B), as shown previously for the related Arabidopsis KDEL-tailed cysteine endopeptidase 2 (AtCEP2) (Hierl ).

Fig. 5.

pH dependence of the activation and activity of recombinant HvPAP14. (A) pH dependence of the activation of recombinant HvPAP14 was measured by the cleavage of Z-FR-AMC in buffer with pH 3.5–9. The ratios of the fluorescence intensities between cleaved (440 nm) and uncleaved (393 nm) substrate were calculated. (B) pH dependence of the activity of recombinant HvPAP14 was measured in the same way but using HvPAP14 previously activated at pH 4.5. Error bars indicate SD (n=3).

Chloroplast proteins identified as putative substrates of HvPAP14

To identify putative substrates of HvPAP14 in chloroplasts, HvPAP14 was transiently overexpressed in barley protoplasts. Twenty hours after transformation, the protoplasts were determined to be intact by microscopic analysis (Supplementary Fig. S4A). Proteins extracted from the protoplasts were separated by SDS-PAGE and blotted for the immunological detection of HvPAP14 and the putative substrates. Coomassie staining of the proteins extracted from transformed protoplasts revealed significant differences in protein composition compared with control protoplasts, which were confirmed by immunoblot analyses (Supplementary Fig. S4B). In contrast to the control protoplasts, in the HvPAP14-overexpressing protoplasts there was a high accumulation of the two inactive forms of HvPAP14 with molecular weights of 40 kDa and 32 kDa (Supplementary Fig. S4B).

In the transformed protoplasts, degradation fragments of RbcL in the range of 50–30 kDa were detectable. Additionally, the 30 kDa PSBO protein was degraded to several fragments with molecular weights of approximately 27, 26, and 21 kDa (Supplementary Fig. S4B). LHCB1 and LHCB5 were also degraded in the transformed protoplasts. While the 26 kDa LHCB1 protein only decreased in abundance compared with the control, the decrease in the abundance of the 27 kDa LHCB5 protein was accompanied by the appearance of a degradation product with a molecular weight of ~25 kDa (Supplementary Fig. S4B).

The same putative substrates were cleaved in vitro by purified recombinant HvPAP14. However, in the case of RbcL only the 50 kDa cleavage product was obtained (Supplementary Fig. S5).

Chloroplast protein degradation in HvPAP14-overexpressing barley lines

In vitro assays and transient transformation of protoplasts were used to identify the chloroplast proteins RbcL, PSBO, LHCB1, and LHCB5 as potential substrates of HvPAP14. To check whether these proteins are also putative substrates of HvPAP14 in plants, transgenic barley plants overexpressing HvPAP14 under the control of the maize UBIQUITIN-1 promoter were characterized (Fig. 6A; Supplementary Fig. 6). The transgenic barley plants had no obvious phenotype and their appearance did not differ from that of wild-type plants; however, protein extracts from the first true leaves of the transgenic plants had highly increased amounts of the 40 kDa pro-form and the 32 kDa processed form of HvPAP14 (Fig. 6B, C; Supplementary Fig. S6B).

Fig. 6.

Overexpression of HvPAP14 in barley plants. (A) The T-DNA based transformation cassette. d35S T, doubled-enhanced CaMV 35S promoter; E9 T, pea RBCS E9 terminator; HPT-int, HYGROMYCIN PHOSPHOTRANSFERASE gene with STLS1 intron; HvPAP14, barley PAP14 gene; LB, left border; NOS T, terminator of the A. tumefaciens NOPALINE SYNTHASE gene; RB, right border; ZmUBI-int, maize UBIQUITIN-1 promoter with first intron. (B) Immunological detection of HvPAP14 in total leaf extracts, chloroplasts, and plastid fractions. Leaf extracts and chloroplasts were obtained from the wild type (WT) and three HvPAP14-overexpressing lines, E4, E16, and E25. Membrane (M) and stroma (S) fractions were prepared from the WT and line E4. As a control, the proteins were stained with Coomassie Brilliant Blue (CBB). (C) For immunological analyses of putative substrates of HvPAP14 (see Supplementary Fig. S4), antibodies to RbcL, PSBO, and LHCB5 were used. (This figure is available in colour at JXB online.)

Immunological analyses with specific antibodies to selected chloroplast proteins revealed that RbcL, PSBO, and LHCB5 were partially degraded in the transgenic plants (Fig. 6C). Besides the 50 kDa form of RbcL, 45 kDa and 30 kDa degradation products were detected (Fig. 6C). LHCB1 was rather stable except for a lower level in transgenic line 4, which showed the highest abundance of HvPAP14 (Fig. 6C).

Degradation products of PSBO (30 kDa) with molecular weights of 27, 26, and 21 kDa were enriched in the HvPAP14-overexpressing plants relative to the wild type (Fig. 6C). In contrast to the findings of the protoplast assays, the 26 kDa fragment could be detected in the transgenic plants as well as in the wild-type plants. In the case of LHCB5, a protein of ~25 kDa accumulated in the overexpressing plants, whereas it was hardly detectable in the wild-type barley (Fig. 6C). This degradation product might be the same as the one detected in the protoplast assay (Supplementary Fig. S4B).

Discussion

HvPAP14 belongs to the plant-specific group of K(H)DEL cysteine endopeptidases (KDEL-CysEPs) reported to be involved in processes of programmed cell death (Hierl ). The level of HvPAP14 mRNA was highly up-regulated during senescence of flag leaves collected from barley fields (Hollmann ), and the HvPAP14 protein was isolated from senescing primary foliage leaves by cysteine protease affinity labelling. It was hence tempting to expect that HvPAP14 would be responsible at least in part for the increased activity of cysteine proteases in senescing leaves (Martínez ; Prins ). Nevertheless, the protein was also detectable in leaves before senescence. Three different forms were detected by immunological analyses: the 40 kDa pro-enzyme, a 32 kDa protein likely corresponding to a protein lacking the ER retention motif HDEL, and the 26 kDa mature enzyme (Fig. 2A, C). Intermediate forms indicating stepwise processing have been also detected with other KDEL-tailed cysteine endopeptidases, such as the tomato SlCysEP (Senatore et al., 2008; Trobacher ) and the well-characterized RcCysEP from castor bean (Schmid ), as well as AtCEP2 from Arabidopsis (Hierl ).

Immunoblot analyses with samples normalized to the same amount of protein showed that the 40 kDa pro-enzyme of HvPAP14 apparently accumulates during leaf senescence, while the abundances of the 32 kDa form and the active 26 kDa form did not change significantly (Fig. 2C). The senescence-associated increase in the ratio between the pro-enzyme and the processed forms could indicate that the abundance of the pro-form is at least partly regulated at the transcriptional level. Processing of the HDEL tail is seemingly independent of the developmental stage of leaves. In comparison with the inactive forms, the mature enzyme was detected at rather low abundance at all stages of leaf development. As is the case for other enzymes in this group, processing is tightly controlled because proteolysis at the wrong time and location may be lethal (Wiederanders ).

HvPAP14 is a typical member of the KDEL-CysEPs of the C1A family (Rawlings ), all of which have N-terminal signal peptides directing them to the ER (Supplementary Fig. S1A). The sequence of the C1A cysteine peptidases is highly conserved in monocot and dicot plants, with the enzymes of Aegilops tauschii and Triticum aestivum being the most similar to the barley enzyme (Supplementary Fig. S1B). Transient transformation of protoplasts from barley leaves with a HvPAP14:RFP fusion construct showed an accumulation of the fluorescing fusion protein in net-like structures resembling the ER, as well as in vesicular structures similar to those observed to harbour the Arabidopsis KDEL-endopeptidase AtCEP2 in epidermal cells (Hierl ). Usually, CysEPs are further transported to vacuolar compartments, where they are activated by low pH. Activation can also be achieved in the cytoplasm after rupture of the vacuole, as reported for the CysEP of Ricinus communis (Hierl ). Unexpectedly, HvPAP14 was also detected in chloroplasts as well as the ER. When the antibody directed towards all forms of HvPAP14 was used for immunogold labelling, gold particles were found to be mainly associated with chloroplasts, as well as vesicular structures likely to be linked to the ER (Fig. 3C).

Several chloroplast proteins were reported to be synthesized into the ER and delivered to chloroplasts by vesicles from the ER via the Golgi apparatus (Radhamony and Theg, 2006; Baslam ). Carbonic anhydrase was the first protein reported to be delivered to chloroplasts by the secretory system (Villarejo ). Furthermore, α-amylase Amyl-1 (Kitajima ) and nucleotide pyrophosphatase/phosphodiesterases were reported to be transported by the secretory system (Nanjo ; Kaneko ). These proteins share potential N-glycosylation occurring in the ER and modified in the Golgi apparatus. The HvPAP14 sequence has several predicted glycosylation sites. Whether HvPAP14 is indeed glycosylated and is transported to chloroplasts via the secretory pathway remains to be investigated. In order to examine whether the vesicular structures detected by RFP fluorescence belong to the Golgi apparatus, barley protoplasts were co-transformed with a Golgi-specific GFP construct (CD3-963, ABRC/Tair) and the HvPAP14:RFP construct (Supplementary Fig. S3). These assays revealed that the vesicular structures showing RFP fluorescence most likely are not derived from the Golgi apparatus. Immunogold labelling employing the HvPAP14 antibody showed the association of gold particles with bodies resembling microbodies in their size and ultrastructure (Fig. 3C). Protoplast transformation with a GFP construct directed to peroxisomes (p7-PtGFPpx) clearly showed that the HvPAP14:RFP fluorescence is not associated with peroxisomes (Supplementary Fig. S3). It is likely that the vesicular structures belong to the ER, as described for vesicular structures called ricinosomes in the endosperm of castor bean (Schmid ) or in germinating seeds of Vigna mungo containing a KDEL-tailed endopeptidase, where the KDEL tail was found to be required for transport to the vacuole (Okamoto, 2001, 2003). This suggests that the loss of the HDEL tail of HvPAP14 could be important for its localization to chloroplasts.

When HvPAP14 in fusion with MBP was synthesized in E. coli, a recombinant protein with protease activity was obtained. The purified protein was shown to be activated at low pH (pH 4–5) (Fig. 5A). As shown for the orthologous endopeptidase AtCEP2 of Arabidopsis (Hierl ) and the RcCysEP of R. communis (Schmid ), activity was highest at pH 4.5. After activation, the protease activity was shown to be pH independent (Fig. 5B), as was also shown for the orthologous protein AtCEP2 from A. thaliana (Hierl ).

In immunoblots decorated with the specific HvPAP14 antibody, the 40 kDa pro-enzyme was detected in the total protein fraction, but not in purified fractions of chloroplasts. Immunoblot analyses revealed that the 32 kDa pro-enzyme likely lacking the HDEL tail as well as the mature enzyme are associated with chloroplast membranes (Fig. 4A). When a purified thylakoid fraction was further fractionated into membranes and lumen, the 32 kDa form was detected in both parts, whereas the 26 kDa active enzyme was exclusively detected in association with membranes (Fig. 4B). Association of HvPAP14 with thylakoid membranes was further demonstrated by immunogold labelling. In chloroplasts ruptured by the preparation procedure, it was obvious that gold particles stayed attached to the thylakoid membranes (Fig. 3C).

Among the proteases identified in chloroplasts from A. thaliana, cysteine proteases were almost absent (Majsec ). However, in a proteomic study performed with pea chloroplasts, a cysteine protease belonging to the OTU-like cysteine protease family (Makarova ) was detected. Its targeting to chloroplasts was confirmed by transient transformation using a yellow fluorescent protein fusion construct (Bayer ). In an earlier search for thylakoid-associated proteases, cysteine protease activity was found to be enriched in the luminal fraction from spinach chloroplasts (LP27) (Sokolenko ). The detection of the 32 kDa form of HvPAP14 in the thylakoid lumen likely indicates that cleavage of the inhibitory pro-domain occurs in the lumen, where a low pH is established during photosynthesis (Kramer ; Järvi ), and that the activity of mature HvPAP14 occurs at the membrane (Fig. 7). Activation inside the lumen is in accordance with the low pH required for activation of the recombinant enzyme.

Fig. 7.

Model describing the activation and putative function of HvPAP14 in chloroplasts. After transport from the endoplasmic reticulum into the chloroplasts, HvPAP14 can be activated at low pH in the thylakoid lumen by cleaving off the inhibitory pro-peptide. The active peptidase binds to the thylakoid membrane, where the substrates PSBO, LHCB1, and LHCB5 are available on the luminal side. After activation at low pH, HvPAP14 activity is independent of pH over a broad range of physiological pH. The activated cysteine peptidase might therefore leave the thylakoid lumen and bind to the stromal side of the thylakoid membrane, where it is able to cleave stromal proteins such as RbcL.

Besides thylakoid membrane proteins, luminal and stromal proteins were also shown to be targeted by HvPAP14. Such a broad spectrum of substrates is typical for KDEL-CysEPs (Hierl ).The degradation of PSBO by HvPAP14 is in accordance with an earlier report on cysteine protease activity in the thylakoid lumen (Sokolenko ). HvPAP14-overexpressing plants contained PSBO fragments of 27, 26, and 21 kDa (Fig. 6C). Similar fragments in DEG1-deficient A. thaliana plants (Li ) suggest that HvPAP14 and DEG1 might work together in the degradation of PSBO.

In leaves of plants overexpressing HvPAP14, several degradation products of RbcL were detected (Fig. 6C). HvPAP14 might catalyse the initial degradation step, resulting in a 50 kDa fragment that was also detected in vitro (Supplementary Fig. S5), and afterwards RbcL could be cleaved by other peptidases to produce the smaller fragments. Degradation products of RbcL have also been observed during stress and senescence (Parrott ; Thoenen ; Feller ). The formation of RbcL degradation products induced by starvation in wheat leaves was shown to be inhibited by E-64, the inhibitor of cysteine proteases (Thoenen ). Further evidence for the degradation of RbcL by cysteine proteases has been provided by overexpression of the gene encoding the protease inhibitor cystatin in tobacco. In the transgenic tobacco lines, RbcL degradation was inhibited in the presence of E-64 (Prins ). Data obtained with these plants furthermore revealed that cysteine proteases are involved in the turnover of Rubisco in control conditions as well as in its degradation during stress (Prins ).

Rubisco degradation has been attributed to both proteases located inside chloroplasts and proteases located in vacuoles (Hörtensteiner and Feller, 2002; Feller ). The current view about Rubisco degradation is that it is initiated inside chloroplasts and finally occurs in vacuolar compartments containing cysteine proteases (Feller ). Rubisco is thought to be delivered from chloroplasts to vacuoles by Rubisco-containing bodies (Chiba ), also called Rubisco vesicular bodies (RVBs) (Prins ). With regard to the low pH required for cysteine protease activity, Rubisco degradation by cysteine proteases was considered to be located in the vacuole (Thoenen ). Recently, it has also been reported that the SAG12 cysteine protease, which is known to be located in acidic senescence-associated vacuoles (Otegui ), is involved in Rubisco degradation during leaf senescence (James ). However, cystatin in overexpressing tobacco plants was found to be enriched inside chloroplasts and in RVBs (Prins ). The detection of HvPAP14 in chloroplasts is hence in accordance with the data on cystatin-overexpressing tobacco. Partial degradation of Rubisco by HvPAP14 is likely to occur inside chloroplasts, preceding the formation of Rubisco-containing vesicles (Prins ) and autophagic processes (Xiong ).

HvPAP14 has not been found in the stroma. Its tight binding to membranes might prevent its transfer to Rubisco-containing vesicles that are known to contain stromal proteins only (Chiba ). Hence, HvPAP14 is unlikely to participate in chloroplast protein degradation outside chloroplasts.

The results of this study show that the putative substrates of HvPAP14 are chloroplast proteins. It is likely that the protein fragments produced in plants overexpressing HvPAP14 are produced by cooperative or sequential proteolysis also involving other proteases (Kato , Moreno ).

Most KDEL-CysEPs are involved in processes of programmed cell death (Hierl ; Höwing et al., 2014, 2017). The gene encoding HvPAP14 has been identified as a senescence-associated gene, in accordance with a role in cell death and stress, as has also been proposed for another member of the barley C1A family (Velasco-Arroyo ). However, the three forms of the protein (40, 32, and 26 kDa) have been detected in barley leaves at all stages of development, indicating that the enzyme functions in the normal turnover of chloroplast proteins. This does not exclude an additional function during bulk protein degradation occurring during leaf senescence.

Supplementary data

Supplementary data are available at JXB online.

Table S1. Mass spectrometric analysis of peptides.

Table S2. Primers used in this study.

Fig. S1. Sequence alignment and phylogenetic tree of papain-like cysteine peptidases.

Fig. S2. Specificity test of the HvPAP14 antibodies.

Fig. S3. Protoplast transformation: co-localization of GFP fusion proteins specific for the Golgi apparatus and peroxisomes, respectively, with HvPAP14:RFP.

Fig. S4. Analysis of HvPAP14 and its putative substrates in HvPAP14-overexpressing protoplasts.

Fig. S5. In vitro identification of putative substrates of HvPAP14 in barley chloroplasts.

Fig. S6. Characterization of transgenic barley lines overexpressing HvPAP14.

Click here for additional data file.