A Kunitz-type protease inhibitor regulates programmed cell death during flower development in Arabidopsis thaliana

Abstract

Flower development and fertilization are tightly controlled in Arabidopsis thaliana. In order to permit the fertilization of a maximum amount of ovules as well as proper embryo and seed development, a subtle balance between pollen tube growth inside the transmitting tract and pollen tube exit from the septum is needed. Both processes depend on a type of programmed cell death that is still poorly understood. Here, it is shown that a Kunitz protease inhibitor related to water-soluble chlorophyll proteins of Brassicaceae (AtWSCP, encoded by At1g72290) is involved in controlling cell death during flower development in A. thaliana. Genetic, biochemical, and cell biology approaches revealed that WSCP physically interacts with RD21 (RESPONSIVE TO DESICCATION) and that this interaction in turn inhibits the activity of RD21 as a pro-death protein. The regulatory circuit identified depends on the restricted expression of WSCP in the transmitting tract and the septum epidermis. In a respective Atwscp knock-out mutant, flowers exhibited precocious cell death in the transmitting tract and unnatural death of septum epidermis cells. As a consequence, apical–basal pollen tube growth, fertilization of ovules, as well as embryo development and seed formation were perturbed. Together, the data identify a unique mechanism of cell death regulation that fine-tunes pollen tube growth.

Cell death, flower development, protease inhibitors, RD21 (RESPONSIVE TO DESICCATION 21), transmitting tract, WSCP (water-soluble chlorophyll-binding protein).

Introduction

Brassicaceae contain an interesting group of chlorophyll-binding proteins with putative protease inhibitor activity. The members of this small protein family, termed water-soluble chlorophyll-binding proteins, WSCPs, contain a Kunitz trypsin inhibitor motif and a chlorophyll-binding motif, 39-PFCPLGI-45, that is homologous to the sequence [F/Y]DPLGL of the major light-harvesting chlorophyll a/b-binding protein of photosystem II and related proteins (for a review, see Satoh et al., 2001). A member of the WSCP family was recently discovered in Arabidopsis thaliana, termed AtWSCP (encoded by At1g72290) (Bektas et al. 2012).

WSCPs are not constitutively expressed. Instead, their expression was shown to be induced under various stress conditions such as drought (Downing et al., 1992; Reviron et al., 1992), heat shock (Annamalai and Yanagihara, 1999), leaf detachment (Nishio and Satoh, 1997), and nitrogen starvation, as well as in response to methyl jasmonate treatment (Desclos et al., 2008).

Based on the unique structure and expression, functions for WSCPs as chlorophyll carriers and/or protease inhibitors have been proposed (Satoh et al., 2001). Takahasi et al. (2012) explored the intracellular localization and biochemical properties of WSCP from cauliflower (Brassica oleracea), termed BoWSCP, and found the protein to accumulate in endoplasmic reticulum (ER) bodies. The authors proposed a role for BoWSCP in the sequestration and detoxification of chlorophyll released from chloroplasts broken during stress (Takahasi et al., 2012). The alternative, not mutually exclusive, model suggests a function for WSCPs as proteinase inhibitors, taking into account that some WSCPs inhibited trypsin, a serine protease (Ilami et al., 1997; Nishio and Satoh, 1997; Desclos et al., 2008). In contrast, AtWSCP blocked the activity of two cysteine proteases containing a granulin domain, the proaleurain maturation protease and papain (Halls et al., 2006).

All WSCPs are known to date bind chlorophyll, and some of them also sequester chlorophyll precursors such as protochlorophyllide and chlorophyllide (Schmidt et al., 2003; Reinbothe et al., 2004; Damaraju et al., 2011; Bektas et al., 2012). Bacterially expressed AtWSCP lacking the presumed signal peptide for intracellular targeting was able to form chlorophyll–protein complexes in vitro with features similar to those of recombinant and native BoWSCP (Bektas et al., 2012). WSCPs from Brassicaceae including AtWSCP are called class-II WSCPs because they do not change their absorption properties upon illumination. In contrast, class-I WSCPs from Amaranthaceae, Chenopodiaceae, and Polygonaceae undergo profound spectral changes when illuminated. WSCP from Lepidium virginicum (LvWSCP) forms pigment–complex tetramers in which four chlorophylls are tightly packed in a hydrophobic cavity and thereby shielded from the interaction with molecular oxygen. This architecture explains why LvWSCP operates in photoprotection in planta (Horigome et al., 2003, 2007).

Quite unexpected for a Chl binding protein is the expression of AtWSCP in flowers (Scutt et al., 2003; Tung et al., 2005). In situ hybridization showed strong WSCP transcript accumulation in the transmitting tracts of the style and vertical septum (Scutt et al., 2003). In situ localization revealed high AtWSCP protein levels in the gynoecium of open flowers and in the transmitting tract of developing siliques (Bektas et al., 2012).

Crawford and Yanofsky (2008) proposed the transmitting tract to represent a highway for pollen tubes growing from the apex to the base of the ovary after pollination. Hereby, cell death and degeneration of transmitting tract tissue seem indispensable to permit efficient pollen tube growth. Cell death in the transmitting tract is a type of programmed cell death (PCD) that is initiated even without pollination (Crawford and Yanofsky, 2008). Extracellular matrix (ECM), representing a complex mixture of polysaccharides, glycoproteins, and glycolipids (Lord and Russell, 2002), has been implicated in cell death in the transmitting tract (Crawford et al., 2007).

The transcription factor genes SPATULA (SPT; (Alvarez and Smyth, 1999, 2002; Penfield et al., 2005), NO TRANSMITTING TRACT (NTT; Crawford et al., 2007), HECATE 1, HECATE 2, and HECATE 3 (HEC1, HEC2, and HEC3; Gremski et al., 2007), and HALF FILLED (HAF; Crawford and Yanofsky, 2011) regulate transmitting tract development. In A. thaliana mutants, such as ntt (no transmitting tract), the transmitting tract was not formed. As a consequence, pollen tube growth was impaired in the carpel, resulting in reduced fertilization rates of basal ovules and lower seed yields (Crawford et al., 2007).

It is shown in this study that AtWSCP expression is differentially regulated by the NTT and HEC (HECATE) transcription factors. An involvement of AtWSCP in the regulation of cell death in the transmitting tract and septum epidermis is also reported. Lack of AtWSCP in a respective knock-out mutant led to changes in pollen tube movement, fertilization of ovules, as well as embryo development, seed set, and germination rates.

Materials and methods

Plant materials

For flower analyses, seeds of A. thaliana ecotype Columbia (Col-0) and the Atwscp mutant were sown on autoclaved soil, kept for 2 d at 4 °C in the dark, and then placed into a culture room under 16h light/8h dark cycles with a light intensity of 70 μM s^–1 cm^–2. The Atwscp mutant (SALK_009681), containing a T-DNA insertion in the single exon of the gene, was obtained from the SALK T-DNA collection (Alonso et al., 2003). Genotyping to identify homozygous knock-out plants was carried out using PCR-based techniques and standard procedures (Innis et al., 1990).

For genetic complementation of the Atwscp mutant, a cDNA encoding AtWSCP (At1g72290) comprising the ATG and stop codons was amplified using the following primers: 5′-GGG GACAAGTTTGTACAAAAAAGCAGGCTTCATGAAGAATC CTTCAGTGATCTCTTTT-3′ and 5′-GGGGACCACTTTGTAC AAGAAAGCTGGGTCCTAACCCGGGAAGTATAAGTTGC T-3′. The PCR product was introduced into pDONR221, sequenced, and introduced into pB7FWG2 (Plant System Biology, VIB-Ghent University) using Gateway technology (Invitrogen). Stable transformation of the construct into Arabidopsis (Col-0 ecotype) was achieved via Agrobacterium tumefaciens C58 pGV3121, using the floral dip method described by Clough and Bent (1998). Transgenic plants constitutively expressing Flag-tagged RD21 in the wild-type or Atwscp mutant background were generated by a similar approach. To produce 355::AtWSCP::GFP lines, cDNA encoding the AtWSCP precursor without a stop codon was cloned into pDONR221 with the primers 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCA TGAAGAATCCTTCAGTGATCTCTTTTCTCATCATTCTCC TGTTTGCT-3′ and 5′-GGGGACCACTTTGTACAAGAAA GCTGGGTCTGAACCCGGGAAGTATAAG TTGCT-3′, sequenced, and then cloned into the pK7FWG2 destination vector (Plant System Biology, VIB-Ghent University). Construction of ferredoxin (FD)–green fluorescent protein (GFP) lines was carried out as described previously (Samol et al., 2011; Rossig et al., 2013). Transgenic lines of the T₃ generation were used for confocal laser scanning microscopy.

Cloning of AtWSCP promoter–GUS fusions and transformation of A. thaliana wild-type plants

The AtWSCP promoter sequence (beginning after the stop codon of the AERO1 gene and ending in the AtWSCP gene before the ATG) was cloned into pDONR221 with the primers 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTATGAC AAATTACAAAAATGG-3′ and 5′-GGGGACCACTTTGTA CAAGAAAGCTGGGTTGATTGTTATGTGTGTTTTAGGG-3′, sequenced, and then introduced into the pKGWFS7 destination vector (Plant System Biology, VIB-Ghent University) using Gateway technology (Invitrogen). Arabidopsis transformants were obtained with A. tumefaciens as described above.

Yeast two-hybrid screens

Yeast two-hybrid screens were performed using the MatchMaker Two-hybrid System 3 (Clontech). pGADT7 and pGBKT7 vectors were modified by inserting in-frame a Gateway cassette in the SmaI site of the multiple cloning site using the Gateway Conversion System (Invitrogen). The orientation was checked by sequencing. These two vectors were kindly provided by Gilles Vachon (unpublished). cDNAs of cysteine proteases with a granulin domain, such as At3g19390, At5g43060, At1g47128 (RD21), and At1g09850 (XBCP3), were purchased from the Arabidopsis Biological Resource Center (stock nos U16700, U16840, U11707, and U83706). cDNAs lacking their predicted target peptide were inserted as described above in pDONR221 (http://cbs.dtu.dk/services/TargetP/), sequenced, and then transferred into the destination vectors. Primers to PCR-amplify (Innis et al., 1990) the cDNAs encoding the different cysteine proteases, AtWSCP and NADPH:protochlorophyllide oxidoreductase B from Arabidopsis (PORB, used for comparison), are described in Supplementary Table S1 available at JXB online. Yeast strain AH109 was co-transformed according to the manufacturer’s instructions and selected on synthetic drop-out medium without Leu and Trp (permissive medium). Individual colonies were picked, propagated in liquid culture, and aliquots were tested on synthetic drop-out medium lacking Leu, Trp, and adenine, and His-selective medium supplemented with 1mM 3-amino-1,2,4-triazole for 4 d at 30 °C.

Histochemistry

For embedding, inflorescences were infiltrated for 10min in a solution containing 5% (v/v) glacial acetic acid, 3.7% (v/v) formaldehyde and 50% (v/v) ethanol, and incubated for 24h at 4 °C. The samples then were dehydrated by successive incubations in ethanol (70, 90, and 100%, 1h each) and Histoclear (25, 50, 75, and 100%, 1h each). Embedding was done in paraffin (Paraplast X-TRA, Tyco Healthcare). Cross-sections (10 μm) were prepared with a Microm HM 355S, Zeiss, mounted on classical slides and incubated for 3h at 45 °C. The samples were de-paraffinized by two incubations with Histoclear, 2×10min each, and rehydrated through a series of solutions comprising an ethanol gradient.

To perform immunolocalization studies, the tissue sections were first treated as described in http://www.ihcworld.com/_protocols/epitope_retrieval/citrate_buffer.htm in order to inactivate endogenous alkaline phosphatase activity. This step is important because a phosphatase-conjugated antibody was used to reveal the presence of the AtWSCP or RD21 protein. Briefly, the sections were incubated for 40min in pre-heated citrate buffer (10mM citric acid, 0.05% Tween 20, pH 6.0) at 95 °C. The sections were then placed at room temperature to allow the slides to cool for 20min. Non-specific binding sites were blocked with 0.05% Tween-20, 5% (w/v) low-fat milk in TRIS-buffered saline (TBS) for 1h. The AtWSCP or RD21 antisera were diluted 1:300 in the same solution and used to incubate the tissue sections for 2h at room temperature. Excess antibodies were removed by washing in 0.05% Tween-20 in TBS several times. Thereafter, incubation was carried out with secondary antibodies (goat anti-rabbit IgG conjugated with alkaline phosphatase, Sigma) under the same conditions used for primary antibodies. After numerous washes, the antigen–antibody complexes were visualized with a solution containing 100mM NaCl, 100mM TRIS-HCl, 50mM MgCl₂, 0.5mM 4-nitroblue tetrazolium chloride (NBT), and 0.5mM 5-bromo-4-chloro-3-indolylphosphate (BCIP), pH 9. Slides were rinsed, mounted in water, and viewed under a light microscope (Eclipse E-600, Nikon). To perform histological analysis, slides were counterstained for 5min with 0.2% Fast Green FCF (Sigma) in 0.1% glacial acid acetic, rinsed in water, and then stained for 30 s with 0.1% alcian blue 8GS in 3% glacial acetic acid. The slides were rinsed and mounted in water, and were viewed under a light microscope (Eclipse E-600, Nikon).

Maximal and minimal pollination experiments combined with aniline blue staining

For maximal pollination experiments, emasculated flowers at stage 12 were left for 24h. Then maximal pollen was added with a paintbrush. The pistils were fixed and stained as described by Sumie et al. (2001). The pollen tubes formed after maximal pollination were stained with aniline blue after 2, 4, and 6h. First, pollinated pistils were fixed briefly in a solution of ethanol:acetic acid (3:1) for 2h at room temperature. After three washes with distilled water, the fixed pistils were treated with 8M NaOH overnight, washed again with distilled water, and stained in a solution of aniline blue (0.1% in 0.1M K₂HPO₄-KOH buffer, pH 11) for 3–5h in the dark. The stained pistils were mounted in water and observed by fluorescence microscopy under UV light excitation (Eclipse E-600, Nikon).

For the minimal pollination experiments, 1–5 pollen grains were taken with a paintbrush and were then transferred under a dissecting microscope (SZX12, Olympus) to the stigma of a 24h emasculated stage 12 flower. After 1 week, the pistils were cleared twice for 3h in 70% (v/v) ethanol solution. The cleared pistils were placed in a drop of glycerol on a microscope slide, fixed, and the position of the seeds within the carpel was determined using the ImageJ software (http://rsb.info.nih.gov./ij).

Preparation of WSCP antiserum

cDNA encoding AtWSCP was amplified by PCR with the primers 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCAAGA ATCCTTCAGTGATCTCTTTT-3′ and 5′-GGGGACCACTTTGT ACAAGAAAGCTGGGTCCTAACCCGGGAAGTATAA GTTGCT-3′, which include the recombination sequence for Gateway cloning (Invitrogen). The PCR product was cloned into pDONR221 and sequenced. To allow protein expression, the cDNA was introduced into pDEST17 and then used to transform Escherichia coli strain BL21. The N-terminal His-tagged protein was expressed after induction with arabinose (0.2% final concentration) and growth at 37 °C for 3h. Proteins were purified by Ni-NTA chromatography under denaturing conditions according to the manufacturer’s instructions (Qiagen) and used to raise rabbit polyclonal antibodies (Interchim, France) (Supplementary Fig. S1 at JXB online).

Protein analyses

Total protein extracts were prepared as described by Kush et al. (1990). Equal amounts of protein that had been determined according to Esen (1978) were separated by 12% SDS–PAGE and transferred to a nitrocellulose membrane (0.45 μm, Whatman) according to Towbin et al. (1979). After blocking the membrane in 1× TBS with 5% skim milk and 0.1% (w/v) Tween-20, immunodetection was carried out with either the AtWSCP (1:3000) or RD21 (1:1000) antiserum (Gu et al., 2012) diluted in blocking solution. Washing of the membranes was carried out in blocking solution followed by incubating the membranes in alkaline phosphatase-conjugated goat anti-rabbit antibodies (diluted 1:3000) or horseradish peroxidase-conjugated goat anti-rabbit antibodies (diluted 1:5000). Cross-reacting bands were detected with BCIP and NBT or via enhanced chemiluminescence (ECL Western Blotting Analysis system, Amersham).

Co-immunoprecipitation and isolation of HM_r complexes

Co-immunoprecpitation experiments were carried out according to Wiedmann et al. (1987), using total flower protein extracts and WSCP and RD21 antisera. Briefly, 20 μl of total flower protein extracts were diluted to 200 μl with buffer A (10mM TRIS-HCl, pH 7.5, 0.15M NaCl, 1mM EDTA, 1% Triton X-100, 0.1% SDS). After incubation at 4 °C on ice for 10min and centrifugation, the supernatant was subjected to two subsequent incubation steps with 5 μl of diluted anti-WSCP antiserum (1:10 in buffer A), one at room temperature for 2h and the other at 4 °C overnight. Then, 50 μl of a 30% suspension of protein A–Sepharose were added and the assays were incubated at 4 °C for 20min with shaking. The Sepharose was subjected to five washing steps with buffer A. Sepharose-bound proteins were liberated by boiling in SDS-sample buffer for 5min.

Isolation of high molecular mass (HM_r) complexes containing RD21 was achieved by immune adsorption chromatography of floral extracts of transgenic seedlings expressing a Flag-tagged AtRD21 protein. Protein complexes were run on non-denaturing PAGE and detected by either Coomassie staining or western blotting using the RD21 and AtWSCP antisera.

Activity measurements

Activity measurements were done using cDNA-encoded Flag-tagged RD21 containing the granulin domain (iRD21) or lacking the granulin domain (mRD21) and His-tagged mature AtWSCP lacking its targeting signal. Proteins were expressed in E. coli and purified to apparent homogeneity by affinity chromatography. Reactions contained 1mM BApNa (Nα-benzoyl-dl-arginine p-nitroanilide) as substrate (Novillo et al., 1997). Reactions were started by the addition of the substrate solution to the reaction buffer (10mM phosphate buffer, 27mM KCl, 137mM NaCl, pH 7.4) containing either iRD21 or mRD21 (Novillo et al., 1997). Catalytic activity towards BApNa was measured with and without l-3-carboxy-[(trans-2,3-epoxyprolyl) leucyl]amido(4-guanidino)butane (E-64) for 0.5, 1, 2, 3, 5, 10, and 20min. Thereafter, the reaction was stopped by the addition of 500 μl of 30% acetic acid. The solution was centrifuged at 10 000 g for 5min and the supernatant monitored spectrometrically at 410nm. Measurement of inhibitor action was carried out according to Tian and Tsou (1982) using the progress curve method. In this method, the inhibitor (E-64 or AtWSCP), substrate (BApNa), and enzyme (iRD21 or mRD21) were incubated together, and the rate of substrate hydrolysis was followed continuously. Plots of substrate hydrolysis product versus time provided hyperbolic curves, as the amount of active enzyme was reduced in either case. In Lineweaver–Burk analysis, the reciprocal of the initial rate (V₀) that was obtained in the absence or presence of E-64 (3, 10, and 100 μM final concentration) and AtWSCP (3, 10, and 100nM) was plotted against the inverse of the final substrate concentration (0.05, 0.1, 0.2, and 0.3mM BApNa). This allowed identification of the inhibitor type.

In planta localization of AtWSCP

Confocal laser scanning microscopy was carried out with a Leica TCS SP5 microscope with argon laser excitation at 488nm (GFP) and 561nm [red fluorescent protein (RFP)]. GFP, RFP, and chlorophyll were detected at emission wavelengths of 510–525, 575–605, and 650–750 m, respectively (Samol et al., 2011; Rossig et al., 2013). Leica confocal software LAS AF and Adobe Photoshop 7 (Adobe Systems, San Jose, CA, USA) were used for image acquisition and processing.

Results

Expression pattern of AtWSCP during different stages of flower development

AtWSCP promoter activity was fiirst investigated at different stages of flower and fruit development. Transgenic A. thaliana plants were generated expressing the AtWSCP promoter region in front of the coding region of bacterial β-glucuronidase (GUS). GUS staining revealed AtWSCP promoter activity in the vascular bundles of the carpels (Fig. 1A) and especially in the transmitting tract (Fig. 1B). Western blotting confirmed AtWSCP protein to accumulate during flower development (Fig. 1C).

Fig. 1.

Expression of AtWSCP during flower development. (A and B) AtWSCP promoter activity as revealed by GUS staining in flowers of a transgenic line expressing GUS under the control of the AtWSCP (At1g72290) promoter. Note the strong GUS signal in the carpels (A) and especially in the septum (B). (C) Protein gel blot analysis of AtWSCP expression in different organs. The upper panel shows a Coomassie-stained gel, whereas the lower panel shows a respective western blot probed with AtWSCP antiserum. (This figure is available in colour at JXB online.)

Next AtWSCP expression in the flower was examined by performing immune localization studies with an AtWSCP antiserum that had been raised against the bacterially expressed protein. Time course analyses during flower development demonstrated AtWSCP accumulation in the transmitting tract of the style, beginning at stage 12 (stages refer to Smyth et al., 1990) (Supplementary Fig. S2C, D at JXB online) and increasing steadily in amount until stage 14–15 (Supplementary Fig. S2E, F). Furthermore, AtWSCP protein started to accumulate in the transmitting tract of the septum at stage 11 (Fig. 2C, D) and remained detectable throughout subsequent stages of flower development (Fig. 2E–K). AtWSCP protein also accumulated in the septum epidermis, surrounding the almost empty septum lumen, at late stages of flower development (referred to as stage 14 in Fig. 2L).

Fig. 2.

AtWSCP accumulates in the transmitting tract of the septum (tt) and the septum epidermis (se) but not in the replum (re). (A–K) Immunohistochemical analysis of AtWSCP accumulation at stages 10 (A and B), 11 (C and D), 12 (E and F), 13 (G and H), and 14 (I and K) of flower development. (A), (C), (E), (G), and (I) show signals obtained after incubation with AtWSCP antiserum, whereas (B), (D), (F), (H), and (K) show background signals obtained with pre-immune serum. (L) AtWSCP accumulation in the septum epidermis. Scales bars: 100 μm in A–K, 25 μm in L. ‘va’ designates the vacuole. (This figure is available in colour at JXB online.)

Expression of AtWSCP is differentially regulated by the transcription factor genes NTT, HEC1, and HEC3

As stated in the Introduction, transmitting tract development is under the control of several transcription factor genes, namely NTT, HEC1, HEC2, and HEC3, as well as SPT and HAF (Alvarez and Smyth, 1999, 2002; Penfield et al., 2005; Crawford et al., 2007; Gremski et al., 2007; Crawford and Yanofsky, 2011; see the Introduction). To ask whether or not transmitting tract development and the expression of the AtWSCP gene are governed by the same transcription factors, the transcription factor mutants ntt, hec1, and hec1::hec3 were selected and the following determinations were carried out (i) real-time PCR (RT-PCR) and protein gel blot analyses on total flower extracts and (ii) immune localization studies on floral sections. ntt and hec1::hec3 were chosen because of their documented impairment in transmitting tract development, production of ECM, and PCD in the septum (Crawford et al., 2007; Gremski et al., 2007). hec1 was used as a control because it contains transmitting tracts like the wild type with regard to size, staining intensity (indicative of ECM production), and cytology (Gremski et al., 2007).

Both RT-PCR and western blot analyses (Fig. 3A; see also Supplementary Fig. S3 at JXB online) demonstrated faint amounts of AtWSCP transcript and protein in the ntt mutant. In hec1 flowers, the AtWSCP transcript and protein levels were slightly reduced, as compared with the wild type. In contrast, flowers of hec1::hec3 double mutants contained vanishingly low, in most cases undetectable, AtWSCP transcript and protein (Fig. 3A; see also Supplementary Fig. S3).

Fig. 3.

AtWSCP gene expression is controlled by the transcription factor genes NTT and HEC3. (A) RT-PCR (a) and western blot (b) data to quantify AtWSCP transcript and protein accumulation in flowers of Arabidopsis wild-type (WT), ntt, hec1, and hec1::hec3 plants. In (a), numbers define relative expression levels (see also Supplementary Fig. S3 at JXB online). Error bars referring to three independent analyses are indicated. In (b), PSBD defines the D subunit of chloroplast photosystem II used as control. (B) Immunohistochemistry of AtWSCP accumulation in ovaries taken from flowers at stages 10 (a, e, i, m), 11 (b, f, j, n), 12 (c, g, k, o), and 13 (d, h, l, p), and styles after fertilization (α, β, γ, δ) from the wild type (a, b, c, d, α), ntt (e, f, g, h, β), hec1 (i, j, k, l, γ), and hec1::hec3 (m, n, o, p, δ). (This figure is available in colour at JXB online.)

When immune-stained cross-sections of gynoecia were compared microscopically, significant differences became apparent for wild-type, ntt, hec1, and hec1::hec3 plants. AtWSCP signals were found in the transmitting tract of the style in wild-type, ntt, and hec1 mutant plants (Fig. 3B, α, β, and γ, respectively), but no AtWSCP signal was detectable for hec1::hec3 flowers (Fig. 3B, δ). In time course analyses during flower development, AtWSCP signals were present in the transmitting tract and the septum epidermis at stages 11, 12, and 13 (Fig. 3B, b, c, d). In contrast, ntt flowers of the same developmental stages contained either no AtWSCP protein (stage 11 and 13; see Fig. 3B, f, h) or tiny amounts of AtWSCP protein that were confined to the central part of the septum (stage 12, Fig. 3B, g). In the hec1 mutant, AtWSCP protein accumulation in the septum was comparable with that in wild-type flowers (Fig. 3B, j–l). In the hec1::hec3 double mutant, AtWSCP protein levels were below the level of detection at all stages of flower development (Fig. 3B, m–p).

Cell death in the transmitting tract and septum epidermis of Atwscp flowers

Cell death in the female reproductive organs is restricted to the transmitting tract and does not occur in the septum epidermis (Crawford and Yanofsky, 2008). Compared with the open style of lily where pollen tubes have a relative freedom to move, cell death and tissue degeneration are required to permit pollen tube movement in the closed transmitting tract of Arabidopsis (Crawford et al., 2007). In order to determine whether AtWSCP may play a role in cell death regulation in the transmitting tract and septum epidermis, an AtWSCP-deficient mutant (SALK _009681; Atwscp) and respective, genetically complemented Atwscp line re-expressing the AtWSCP coding region under the control of the 35S Cauliflower mosaic virus promoter (Atwscp::35SAtWSCP) were used for further analysis (see Supplementary Figs S4 and S5 at JXB online for their characterization).

Cell death and local tissue degeneration were studied in emasculated flowers of wild-type, AtWSCP mutant, and Atwscp::35SAtWSCP plants. Observations were made on at least three different emasculated flowers taken from two independent lines of each genotype. In the transmitting tract of Atwscp mutant flowers, the number of intact cells was lower (~30%) than in wild-type and Atwscp::35SAtWSCP flowers after 1 d of emasculation (Fig. 4A, first lane, and Fig. 4B). Two days after emasculation, the transmitting tract of Atwscp flowers still contained fewer intact cells than the wild type (~30%) and accumulated a diffuse mass presumably representing cell debris (Fig. 4A, second lane, and Fig. 4B). The difference in transmitting tract structure between wild-type and Atwscp flowers was less pronounced after 3 d of emasculation (Fig. 4A, third lane). It was tentatively concluded that cell death in the transmitting tract occurred earlier in the Atwscp mutant than in the wild type. When cell death was scored after toluidine staining, in fact a strikingly higher percentage of dead septum epidermis cells was seen for Atwscp versus wild-type flowers and Atwscp::35SAtWSCP flowers (Fig. 4C, lower panel, compare b with a and c, respectively, corresponding to flowers of stage 17). Thus, a distortion of cell death regulation was observed in the transmitting tract and septum epidermis of the Atwscp mutant that could be restored by re-expressing AtWSCP constitutively.

Fig. 4.

Transmitting tract and septum epidermis structure in emasculated wild-type, Atwscp, and Atwscp::35SAtWSCP flowers. (A) Gynoecia structures in wild-type (WT), Atwscp, and Atwscp 35S::AtWSCP plants after 1, 2, and 3 d of emasculation. Staining was performed with fast green. (B) Statistical assessment of the number of intact cells in transmitting tracts of WT, Atwscp, and Atwscp 35S::AtWSCP flowers after 1 d and 2 d of emasculation. Error bars refer to three independent analyses. (C) Structures of the septum (se) and replum (re) at stage 17 of flower development in WT (a), Atwscp (b), and Atwscp 35S::AtWSCP (c) flowers, as revealed by toluidine blue staining. Note the atypical collapse of septum epidermis cells in Atwscp flowers and that this effect is reversed in the Atwscp 35S::AtWSCP line. (This figure is available in colour at JXB online.)

Production of ECM in Atwscp flowers

ECM, representing a mixture of polysaccharides, glycoproteins, and glycolipids, has been proposed to function in the adhesion, guidance, and nutrition of pollen tubes in the transmitting tract (Lennon et al., 1998). ECMs also contain acidic polysaccharides operating as cell death inducers (Lennon et al., 1998; Gao and Showalter, 1999). Experiments were thus carried out to determine whether or not ECM production might be altered in the Atwscp mutant.

When alcian blue staining of acidic polysaccharides was used to trace ECMs, no major difference was observed for wild-type and Atwscp flowers (Fig. 5, compare B and A). In some Atwscp flowers, a smaller area of stained cells reflecting a reduced transmitting tract size was observed compared with wild-type and Atwscp::35SAtWSCP flowers (Supplementary Fig. S6 at JXB online, compare B with A and C). However, this reduction in transmitting tract size was not observed in all Atwscp flowers and might be the result of natural variation rather than a consequence of the absence of AtWSCP.

Fig. 5.

Production of ECM in the transmitting tract in late-stage 12 gynoecia of wild-type and Atwscp flowers. (A and B) Transverse sections through late-stage 12 gynoecia of wild-type (A) and Atwscp (B) flowers were stained with alcian blue plus fast green (for counterstaining) to detect acidic polysaccharides as an ECM marker tt, transmitting tract. Scale bar=20 μm. (This figure is available in colour at JXB online.)

Pollen tube growth and fertilization in Atwscp mutant flowers

Crawford et al. (2007) have introduced two experimental strategies referred to as maximal and minimal pollination, that differ by the number of pollen grains used for fertilization, to study pollen tube growth. In maximal pollination experiments, a few hundred pollen grains are used, whereas in minimal pollination experiments only 1–5 pollen grains per carpel are used. Experiments were carried out to determine whether precocious cell death in the transmitting tract and unnatural cell death in the septum epidermis in Atwscp flowers affected pollen tube growth and morphology. To score pollen tube growth and morphology at different stages, aniline blue staining was used (Crawford et al., 2007). For wild-type and Atwscp flowers, sets of at least 3–5 pollinated carpels were analysed from a representative number of different plants (n=3).

In maximal pollination experiments, pollen tube growth from the apex to the basal parts of the carpel was increased in the Atwscp mutant, as compared with the wild type (Fig. 6A). This effect became visible already 2h after pollination of emasculated pistils, with significantly more pollen tubes passing a longer distance from the apex to the centre of the ovary in the Atwscp mutant than in the wild type (Fig. 6A, a). The difference between the wild type and Atwscp became even more pronounced after 4h and 6h of pollination (Fig. 6A, b, c). This is apparent from the ~3.5-fold larger diameter the pollen tubes were occupying in the carpel of Atwscp versus wild-type carpels (white triangles in Fig. 6A and B). At the same time, lateral pollen tube growth was significantly reduced in Atwscp versus wild-type flowers (compare high-resolution enlargements designated c1 and c2 in Fig. 6A). Quantitative estimates revealed ~2.5-fold reductions in the number and length of pollen tubes sorting sideways from the transmitting tract.

Fig. 6.

Pollen tube growth and morphology in Atwscp mutant and wild-type siliques. (A) Pollen tube growth visualized by staining with aniline blue at different stages of development following pollination. Emasculated carpels of the wild type (WT) and Atwscp were left for 24h, pollinated, and fixed 2h (a), 4h (b), and 6h (c) after pollination. Arrows mark pollen tubes wandering sideways in wild-type and Atwscp carpels. White triangles mark regions of the carpel used for quantitative pollen number measurements. The enlargements designated c1 and c2 show larger images of the regions boxed in c and were used to illustrate the reduced lateral pollen tube growth in Atwscp versus wild-type flowers. (B) Morphological pollen tube diversity in WT (d and e), Atwscp (g), ntt (h), and genetically complemented Atwscp (i and j) plants after minimal pollination. Scale bars=0.1mm in A; 5 μm in B. (This figure is available in colour at JXB online.)

Crawford et al. (2007) have shown that pollen tubes growing inside the transmitting tract are usually thin and move ahead in a straighter fashion than those on the septum epidermis, where they are thick, intensely stained, and crooked. In the ntt mutant, only the crooked-type pollen tubes were seen (Crawford et al., 2007). When minimal pollination experiments were carried out, both thin and straight versus thick, intensely stained, crooked pollen tubes were seen for wild-type plants. In ntt plants, however, only the thick, intensely stained, crooked pollen tubes were observed (Fig. 6B, h) (cf. Crawford et al., 2007). In marked contrast, only thin and straight pollen tubes were visible in Atwscp flowers and they accounted for 90–95% of all pollen tubes examined (Fig. 6B, g). Thus, it is concluded that most of the pollen tubes remained inside and did not exit from the transmitting tract in Atwscp plants. In line with this view, genetic complementation of Atwscp plants gave rise to the same pollen tube diversity as found for the wild type (Fig. 6B, i, j).

Embryo and seed set distribution in Atwscp mutant and wild-type siliques

Since fertilization of ovules along the entire length of the ovary depends on both pollen tube growth from the apex to the ovary and lateral pollen tube exit from the septum, both aspects were quantified by carrying out minimal pollination studies (1–5 grains per carpel) and counting the number and distribution of embryos within siliques for a representative number of samples (n=100) from wild-type and Atwscp plants. For ease of interpretation, the entire fruit length was divided into 10 percentiles beginning at the most apical (1–10%) parts and terminating at the the most basal (90–100%) parts of the fruit. As shown in Fig. 7, there were significant differences in the number of embryos per percentile between Atwscp and wild-type plants. Atwscp plants indeed displayed a different embryo distribution from the wild type, with reduced embryo numbers in the apical parts and a higher proportion in the most basal parts of the fruit.

Fig. 7.

Seed set distribution in wild-type and Atwscp mutant fruits after minimal pollination. The entire fruit length was divided into 10 percentiles from the most apical (1–10%) to the most basal (90–100%) region of the fruit. The number of seeds is given for each percentile of wild-type and Atwscp mutant fruits.

Seed germination rates in Atwscp mutant versus the wild type

Atwscp mutant and wild-type siliques were allowed to mature, and the appearance and number of seeds were analysed. Although the total number of seeds per silique was similar for both types of samples, seeds of Atwscp plants appeared darker (Fig. 8A), slightly smaller (Fig. 8C), and were also slightly lighter (Fig. 8D). When germination rates were determined for mature seeds, a reduction was noticed for the Atwscp mutant as compared with the wild type (Fig. 8B). This effect was first visible after 24–30h of germination and remained detectable for the next 10–16h (Fig. 8B). At later stages of germination, the difference in the number of germinated seeds between Atwscp and the wild type was negligible, suggesting some redundant gene functions/mechanisms had become operational. In genetically complemented Atwscp plants, seed germination rates were similar to those seen for the wild type. Together, these findings suggested that the defects in fertilization and seed development were causally related to the deficiency in AtWSCP protein.

Fig. 8.

Seed set distribution, seed germination rates, and seed characteristics of the wild type (WT) and the Atwscp mutant. (A) Seed distribution in siliques of WT and Atwscp mutant plants. (B) Seed germination rates of the WT, the Atwscp mutant, and the complemented Atwscp 35S::AtWSCP line after 24, 30, and 41h. (C) and (D) Seed size (length and width) (C) and seed weight (D) of the WT and Atwscp mutant (3 years old). The length and width of 100 seeds was measured and the average value calculated. Seed mass was determined for 100 seeds (this measurement was repeated five times). Error bars are indicated. (This figure is available in colour at JXB online.)

AtWSCP interacts with the cysteine protease RD21 in yeast two-hybrid screens

Halls et al. (2006) characterized AtWSCP as a potent inhibitor of the proaleurain maturation protease containing a granulin domain (encoded by At3g19390) and papain. Based on this observation, experiments were conducted to determine whether AtWSCP may also interact with other cysteine proteases containing a granulin domain. To this end, four genes encoding this type of cysteine protease were identified in the Arabidopsis genome: At3g19390, At5g43060, At1g47120 (RESPONSIVE TO DESICCATION 21, RD21), and At4g34460 (XYLEM BARK CYSTEINE PEPTIDASE 3, XBCP3). cDNAs for these granulin domain-containing proteases as well as mature AtWSCP lacking its putative targeting signal were cloned into vectors appropriate for yeast two-hybrid screens. AtWSCP was linked to the DNA-binding domain and the protease to the activation domain, and vice versa. Successful interactions of AtWSCP with one of the proteases were assessed by the growth of leucine and tryptophan auxotrophs on non-permissive medium. Interactions were obtained when AtWSCP was used as bait and RD21 as prey. No interaction was found with any of the other proteases and PORB as prey (Fig. 9A, B). However, when AtWSCP was used as prey and the cysteine proteases as bait, RD21 and the cysteine proteases encoded by At3g19390 and At5g43060, but not XBCP3 or PORB, provided positive results and thus were identified as interacting partners of AtWSCP (Fig. 9C, D).

Fig. 9.

AtWSCP interacts with the granulin domain-containing cysteine protease RD21. Yeast two-hybrid screening was performed with the coding sequence for mature AtWSCP, NADPH:protochlorophyllide oxidoreductase B (PORB), or empty vector, used as bait, and those for the cysteine proteases encoded by At3g19390, At5g43060, as well as RD21, XBCP3, or empty vector, used as prey (A and B). As control, AtWSCP and empty vector were used as prey and the cysteine protease coding sequences were used as bait (C and D). The permissive medium (A and C) lacks l-leucine and l-tryptophan, whereas the selective medium lacks l-leucine, l-tryptophan, l-histidine, and adenine (B and D). Growth of colonies indicative of a protein–protein interaction between AtWSCP and RD21 was obtained in either combination (bait, AtWSCP; prey, RD21, and vice versa). (This figure is available in colour at JXB online.)

AtWSCP inhibits RD21’s protease activity

RD21A is synthesized as a pre-proprotease carrying a C-terminal granulin domain. Its maturation involves several steps and results in (i) a 35.38kDa intermediate carrying the C-terminal granulin domain (iRD21) and (ii) the 23.73kDa mature RD21 (mRD21) lacking the granulin domain (Yamada et al., 2001). Both iRD21 and mRD21 are active proteases. cDNA-encoded Flag-tagged RD21 containing or lacking the granulin domain and His-tagged mature AtWSCP lacking its targeting signal were produced by coupled in vitro transcription/translation of the respective cDNA clones, purified, and used to carry out protease inhibitor tests. Given that RD21 is a cysteine protease, activity measurements were carried out in the presence or absence of E-64 irreversibly blocking the enzyme activity (summarized in Powers et al., 2002).

Both iRD21 and mRD21 were catalytically active and converted the model substrate used into its product. Substrate conversion in either case obeyed Michaelis–Menten kinetics. In the presence of increasing amounts of E-64 in the assays, substrate conversion was progressively blocked. Lineweaver–Burk plots (Supplementary Fig. S7 at JXB online) revealed a non-competitive, irreversible type of inhibition of iRD21 and mRD21 by E-64. This result was consistent with data reported for other cysteine and cysteine-like proteases (Hanada et al., 1978; Hashida et al., 1980; Barrett et al., 1982; Powers et al., 2002).

Experiments that were carried out in the presence of AtWSCP revealed a dose-dependent inhibition of iRD21 and mRD21 activity towards the model substrate. Lineweaver–Burk analyses suggested a competitive type of inhibition of iRD21 and mRD21 in which only the binding of the model substrate used (reflected by the K_D) but not substrate conversion (reflected by the V_max) was lowered in the presence of the inhibitor (Fig. 10A, B). For both, iRD21 and mRD21 the apparent K_D of ~2×10^–8 M lay in a range similar to that reported for the interaction between recombinant human cystatin C and actinidin (EC 3.4.22.14), forming tight equimolar complexes (Björk et al., 1994, 1995). Other examples are compiled in Supplementary Table S2 at JXB online.

Fig. 10.

AtWSCP inhibits the cysteine protease activity of RD21 in vitro. (A and B) Lineweaver–Burk plots showing 1/V₀ versus 1/[S] characteristics in assay mixtures containing iRD21 (A) and mRD21 (B) incubated in the absence of AtWSCP [dotted line and open squares, referred to as ‘control’ (Ctr)] or in the presence of 3nM (short dashes and triangles), 10nM (long dashes and circles), and 100nM AtWSCP (solid line and dots).

AtWSCP forms larger complexes with RD21 in vivo

If iRD21 and/or mRD21 were to interact with AtWSCP in vivo, they should form stable complexes in the transmitting tract and septum epidermis. Given the difficulty in accessing these tissues specifically in biochemical experiments, (i) western blotting on total flower extracts and (ii) immune localization on floral sections as before were used to detect RD21. Western blots were performed with inflorescence extracts from the wild type, Atwscp, ntt, hec1, and hec1::hec3, and incubated with an RD21 antiserum (Fig. 11A). Two immune-reactive bands of different M_r were obtained. These bands most probably represent iRD21 and mRD21 described previously (Yamada et al., 2001). Notably, no differences in the amount or ratio of the two immunoreactive RD21 bands were visible between the different samples (Fig. 11A). Interestingly, RD21 was expressed in inflorescences irrespective of NTT, HEC1, and HEC3, and accumulated in both its intermediate and mature form. When immune localization studies were conducted with wild-type inflorescences, RD21 was present in the transmitting tract and septum epidermis as well as other parts of the ovary (Fig. 11B: compare panels a and b). Co-immunoprecipitation experiments revealed that only mature RD21 interacted with AtWSCP in soluble flower extracts (Fig. 11C).

Fig. 11.

Expression of RD21 during Arabidopsis flower development. (A) Western blot to detect RD21 in flower extracts of wild-type (WT), Atwscp, ntt, hec1, and hec1::hec3 plants. The two RD21-related proteins represent the granulin domain-containing processing intermediate of RD21 of 35.38kDa (iRD21) and the mature form of RD21 of 23.73kDa (mRD21). (B) Immunolocalization of RD21 in flowers of wild-type plants using RD21 antibodies (a) or pre-immune serum (b). Scale bars=50 μm. (C) Pull-down assay to detect AtWSCP–RD21 interactions in flower extracts of wild-type and Atwscp plants. Total flower extracts were subjected to co-immunoprecipitation using the AtWSCP and RD21 antisera, and replicate protein gel blots separating the recovered complexes then were probed for the presence of the respective interacting partner using the other antibody. (This figure is available in colour at JXB online.)

To add further evidence for the interaction between AtWSCP and RD21, transgenic plants constitutively expressing Flag-tagged RD21 in the wild-type or Atwscp mutant background were used to isolate RD21-containing protein complexes. The protein complexes in turn were run on non-denaturing PAGE and detected by Coomassie staining or western blotting. Figure 12A shows that higher molecular mass (HM_r) complexes containing RD21 were present in flower extracts of wild-type plants. In contrast, no such complexes were detectable in flower extracts of Atwscp plants (Fig. 12A). Western blotting confirmed that the recovered HM_r complexes contained AtWSCP (Fig. 12B). The lack of iRD21 in these experiments suggests that only mRD21 formed complexes with AtWSCP. iRD21 has been reported to aggregate easily and irreversibly in the vacuole (Yamada et al., 2001) and was presumably for this reason not recovered in the soluble flower extract used for complex isolation.

Fig. 12.

Interaction of AtWSCP with RD21 in vivo. (A and B) Non-denaturing PAGE of protein complexes isolated from transgenic plants expressing Flag-tagged RD21. Proteins were purified from floral extracts through immune adsorption chromatography and stained with Coomassie brilliant blue (A) or subjected to western blotting (B). Replicate blots were probed with antisera against RD21 (a) and AtWSCP (b), and cross-reactive protein bands were detected using ECL. Positions of free RD21-Flag and HM_r complexes containing AtWSCP and RD21 are indicated. (This figure is available in colour at JXB online.)

AtWSCP is not localized to plastids in planta

AtWSCP is predicted to bear a signal peptide and thus to be targeted to the secretory pathway (TargetP; www.cbs.dtu.dk/services/TargetP). On the other hand, the documented chlorophyll-binding properties (Damajaru et al., 2011; Bektas et al., 2012) imply that the protein should be localized to the plastid compartment. Transmitting tract cells are green and contain chloroplasts. When the chlorophyllous flower parts of transgenic A. thaliana plants expressing a fusion protein consisting of AtWSCP and jellyfish GFP were analysed, reporter protein fluorescence was not found in plastids but appeared in the cytosol (Fig. 13) and, in some experiments, it was detectable in cell walls/extracellular spaces (data not shown). Along with the results presented previously, these data disproved a role for AtWSCP as a chloroplast chlorophyll carrier.

Fig. 13.

Intracellular localization of AtWSCP–GFP fusion protein. (A–H) Representative images of cells expressing AtWSCP–GFP fusion protein. Fluorescence signals of GFP (A, E, I) and chlorophyll (B, F, K) were collected simultaneously. White arrows mark regions in which cytosolic but not plastid localization of AtWSCP–GFP is most obvious. In this case, plastids are recognized by their red autofluorescence (B, F) but are visible as black holes in the green-fluorescing cytosolic background (A, E). (I–M) pFD–GFP fusion protein localization in plastids, as evidenced by the overlap of green GFP fluorescence and red pigment fluorescence (I).

Discussion

During flower development in Arabidopsis, a type of PCD occurs in the female reproductive organ that permits pollen tube growth inside the transmitting tract and pollen tube exit from the septum. It is reported here that a Kunitz protease inhibitor related to water-soluble chlorophyll proteins of Brassicaceae (AtWSCP, encoded by At1g72290) controls both processes. The regulatory circuit identified depends on the restricted expression of WSCP in the transmitting tract and the septum epidermis. In this respect, AtWSCP resembles WSCP from Japanese wild radish, which showed a similar expression pattern. Transcripts of this WSCP were detected in buds and flowers but not in leaves, stems, and roots (Takahashi et al., 2013). AtWSCP expression is temporally controlled and occurred in (i) the transmitting tract of the style in flowers from stage 12 to 15 (Supplementary Fig. S2 at JXB online); (ii) the transmitting tract of the septum; and (iii) the septum epidermis of flowers from stage 12 to >14 (Figs 1, 2).

AtWSCP expression is under control of the transcription factor NTT as well as HEC1 and HEC3 that collectively regulate transmitting tract development, ECM production, and cell death (Crawford et al., 2007; Crawford and Yanofsky, 2011). In the hec1 mutant, exhibiting transmitting tracts of wild-type size and structure, unimpaired levels of ECMs, and PCD, AtWSCP transcript and protein accumulated to almost the same extent as in wild-type flowers (Fig. 3A, B). This suggests redundant roles for HEC1 and HEC3 in planta and that HEC3 functionally replaced HEC1 in the hec1 mutant. In contrast, reduced amounts or even no AtWSCP transcript and protein were detectable in the ntt and hec1::hec3 mutants lacking transmitting tracts, respectively (Fig. 3A, B). One might argue that this effect was indirect and reflected the lack of transmitting tract cells for AtWSCP deposition. However, ntt mutant flowers entirely lacking a transmitting tract still contained low amounts of AtWSCP transcript and protein in the septum at stage 12 of flower development (Fig. 3B, g). Therefore, a combined effect of NTT, HEC1, and HEC3 seems more likely. As shown here, NTT regulated AtWSCP expression differentially in the transmitting tract of the style and the septum. In the style, WSCP protein accumulated irrespective of NTT and was therefore detectable in ntt flowers (Fig. 3B, β). In contrast, almost no WSCP protein was found in the transmitting tract of the septum in ntt plants (Fig. 3B, e–h).

In contrast to ntt and hec1::hec3 flowers, Atwscp plants showed no impairment in gynoecia development but instead displayed accelerated cell death rates in the transmitting tract (Fig. 4A, B). Pollen tube movement from the apex to the base of the transmitting tract occurred faster in Atwscp than in wild-type flowers (Fig. 6). As a result, more pollen tubes were present in the basal parts of the transmitting tract and presumably due to their tight packaging seemed to hamper each other physically from sorting laterally, as depicted in the working model in Fig. 14. Siliques from AtWSCP plants indeed differed from siliques of wild-type plants by a higher percentage of seeds in the basal parts. The opposite effect was noted for siliques from ntt plants where seed formation was not observed in the basal parts of the fruits. In the hec1::hec3 double mutant, a tremendous reduction in overall fertility (17% of that of the wild type) was observed. The observed changes in pollen tube movement and exit in Atwscp flowers (Fig. 6A) were accompanied by/correlated with changes in seed set (Fig. 7). It is likely that the detected differences in seed colour and weight (Fig. 8) were caused by nutrient deprivation in the more basal parts of the transmitting tract. However, the measured delay in seed germination was only transient and easily overcome at later stages of seedling development.

Fig. 14.

Working model for the role of AtWSCP during flower development. (A–C) During fertilization, pollen (yellow spots) land on the stigma, germinate, and begin to grow in the transmitting tract. Pollen tube growth is mechanically unimpaired in minimal pollination experiments (A) but may be hampered by the large number of pollen grains that compete for growth in maximal pollination experiments (B). Pollen tube growth and exit from the transmitting tract for fertilization of ovules requires a type of PCD that is deregulated in the Atwscp mutant studied here, leading to heterogenous distribution of fertilized ovules, embryos, and, in turn, seeds in the developing fruit (C). Sometimes, embryos are even aborted. (D) How AtWSCP may regulate cell death by binding to and inhibiting the cysteine protease RD21 and related target proteases containing a granulin domain. (This figure is available in colour at JXB online.)

AtWSCP appears to contribute to cell death regulation in the female reproductive organ by binding to and inhibiting RD21. It is attractive to hypothesize that this interaction blocked RD21’s activity as a cysteine protease and pro-death factor. In line with this view, Atwscp knock-out mutant flowers exhibited precocious cell death in the transmitting tract and unnatural death of septum epidermis cells. Both effects were reversed by ectopic expression of AtWSCP.

RD21 is synthesized as a 57kDa pre-proprotein containing an N-terminal propeptide and a C-terminal extension that is composed of a proline-rich domain and the granulin domain. The N-terminal pre-domain is removed, giving rise to an intermediate form (iRD21) of 35.38kDa that is slowly processed to the mature, 23.73kDa protein (mRD21). iRD21 still contains the granulin domain and was shown to be easily aggregated under the acidic conditions of the vacuolar interior where most of the protein is thought to accumulate (Yamada et al., 2001).

Sequential processing of the pre-proprotein was proposed to be a mechanism to regulate RD21’s proteolytic activity. Similar to human cathepsin B and S, the N-terminal propeptides of RD21 functions as an autoinhibitory domain (Turk et al., 1996; Maubach et al., 1997). Activation is achieved by removing this propeptide. On the other hand, aggregation of the intermediate form in the vacuole was proposed to sequester RD21 in an inactive form (Yamada et al., 2001). A third way to regulate RD21’s activity was described in endothelium cells of developing seeds. Here, protein disulphide isomerase-5 (PDI5) trafficks together with RD21 from the ER to protein storage bodies and lytic vacuoles. PDI5 was able to inhibit the activity of recombinant cysteine protease in vitro and was preferentially expressed in developing seeds and floral tissues. In both tissues, RD21 appears to be a mediator of PCD. Mutants unable to express functional PDI5 due to a T-DNA insertion in the 5′-untranslated region of the PDI5 gene were characterized by premature initiation of PCD in endothelial cells during embryo development and by fewer, often non-viable seeds (Ondzighi et al., 2008).

A final way to regulate RD21’s proteolytic activity may be binding to specific protease inhibitors. In Arabidopsis, the cytoplasmic Arabidopsis serpin AtSerpin1 is one such component (Lampl et al., 2010, 2013). AtSerpin1 is expressed ubiquitously throughout the plant and irreversibly inhibits RD21 in leaf extracts (Lampl et al., 2010). Another example of a cysteine protease inhibitor is cystatin in soybean. Ectopic expression of cystatin inhibited cysteine protease activity and blocked PCD (Solomon et al., 1999). The present observations demonstrate that RD21’s activity is inhibited by AtWSCP during flower development. Kinetic measurements showed that both iRD21 and mRD21 interact with AtWSCP, leading to the reversible inhibition of their cysteine protease activity. Lineweaver–Burk plots allowed identification of (i) non-competitive, irreversible inhibition of RD21’s cysteine protease activity by E-64; and (ii) competitive inhibition of RD21 activity by AtWSCP. Kinetic analyses showed that AtWSCP obviously affected only substrate binding to iRD21 and mRD21. This suggests that the Kunitz protease inhibitor motif in AtWSCP protrudes into the active site of RD21 and thereby blocks the enzyme activity. Similar observations have been made for other proteins containing Kunitz protease inhibitor domains (e.g. Patil et al., 2012; Toubarro et al., 2013).

Based on the yeast two-hybrid screens, co-immunolocalization data, and biochemical studies tissue-specific interactions between RD21 and AtWSCP are proposed to control RD21’s proteolytic and pro-death activity. How this unique pathway of PCD proceeds molecularly is unresolved and will be characterized in future work. It will also be interesting to see how the other RD21-related granulin domain-containing cysteine proteases contribute to cell death regulation during flower development or if they have other functions.

Acknowledgements

We thank Brian Crawford, Gary Ditta, and Martin Yanofsky (University of California San Diego, La Jolla, CA, USA) for generous gifts of the Arabidopsis ntt and hec1 and hec3 mutant seeds, Ikuko Hara-Nishimura (Kyoto University, Japan) for a gift of the antibody directed against RD21 (as described in Yamada et al., 2001), Gilles Vachon, CEA, iRTSV, Laboratoire Physiologie Cellulaire et Végétale, Grenoble, France, for his advice on performing the yeast to hybrid screens, and Fayçal Ounnas, Laboratoire de Bioénergétique Fondamentale et Appliquée (LBFA), Université Joseph Fourier, Grenoble, for his help with the RT-PCR analyses. This work was supported by a grant of the German Research Foundation (RE 1465/3-1) and a grant of the Chaire d’Excellence Program of the French Ministry of Research and Education to CR.

Author notes

Editor: Peter Bozhkov