An in vivo expression system for the identification of cargo proteins of vacuolar sorting receptors in Arabidopsis culture cells


  • Jinbo Shen,

    1. School of Life Sciences, Centre for Cell and Developmental Biology and State Key Laboratory of Agrobiotechnology, The Chinese University of Hong Kong, Hong Kong, China
    Search for more papers by this author
  • Pui Kit Suen,

    1. School of Life Sciences, Centre for Cell and Developmental Biology and State Key Laboratory of Agrobiotechnology, The Chinese University of Hong Kong, Hong Kong, China
    Current affiliation:
    1. Department of Orthopaedics and Traumatology, The Chinese University of Hong Kong, Hong Kong, China
    Search for more papers by this author
  • Xiangfeng Wang,

    1. School of Life Sciences, Centre for Cell and Developmental Biology and State Key Laboratory of Agrobiotechnology, The Chinese University of Hong Kong, Hong Kong, China
    Search for more papers by this author
  • Youshun Lin,

    1. School of Life Sciences, Centre for Cell and Developmental Biology and State Key Laboratory of Agrobiotechnology, The Chinese University of Hong Kong, Hong Kong, China
    Search for more papers by this author
  • Sze Wan Lo,

    1. School of Life Sciences, Centre for Cell and Developmental Biology and State Key Laboratory of Agrobiotechnology, The Chinese University of Hong Kong, Hong Kong, China
    Search for more papers by this author
  • Enrique Rojo,

    1. Departamento de Genética Molecular de Plantas, Centro Nacional de Biotecnología, Consejo Superior de Investigaciones Científicas, Madrid, Spain
    Search for more papers by this author
  • Liwen Jiang

    Corresponding author
    1. School of Life Sciences, Centre for Cell and Developmental Biology and State Key Laboratory of Agrobiotechnology, The Chinese University of Hong Kong, Hong Kong, China
    2. CUHK Shenzhen Research Institute, The Chinese University of Hong Kong, Shenzhen, China
    Search for more papers by this author

  • Upon request, all novel materials described in this publication will be made available in a timely manner for non-commercial research purposes, subject to the requisite permission from any third-party owners of all or parts of the material. Obtaining any permission will be the responsibility of the requestor.


Vacuolar sorting receptors (VSRs) are type I integral membrane family proteins that in plant cells are thought to recognize cargo proteins at the late Golgi or trans-Golgi network (TGN) for vacuolar transport via the pre-vacuolar compartment (PVC). However, little is known about VSR cargo proteins in plants. Here we developed and tested an in vivo expression system for the identification of VSR cargos which is based on the premise that the expressed N-terminus of VSRs will be secreted into the culture medium along with their corresponding cargo proteins. Indeed, transgenic Arabidopsis culture cell lines expressing VSR N-terminal binding domains (VSRNTs) were shown to secrete truncated VSRs (BP80NT, AtVSR1NT and AtVSR4NT) with attached cargo molecules into the culture medium. Putative cargo proteins were identified through mass spectrometry. Several identified cargo proteins were confirmed by localization studies and interaction analysis with VSRs. The screening strategy described here should be applicable to all VSRs and will help identify and study cargo proteins for individual VSR proteins. This method should be useful for both cargo identification and protein–protein interaction in vivo.


In plant cells, soluble vacuolar proteins in the endomembrane system are believed to be mainly sorted by transmembrane receptors such as the vacuolar sorting receptor proteins (VSRs) or receptor homology-transmembrane-RING-H2 proteins (RMRs) (Neuhaus and Rogers, 1998; Jiang et al., 2000), because these receptors recognize the vacuolar sorting determinants (VSDs) present in the soluble cargo proteins. BP80, a type I integral membrane protein containing an N-terminal binding domain (NT), a single transmembrane domain (TMD) and a cytoplasmic tail (CT), was the first VSR isolated from peas (Kirsch et al., 1994, 1996; Paris et al., 1997; Neuhaus and Rogers, 1998; Paris and Neuhaus, 2002). Vacuolar sorting receptor proteins have been shown to localize to the pre-vacuolar compartments (PVCs) or multivesicular bodies (MVBs), the trans-Golgi network (TGN) and the plasma membrane (Sanderfoot et al., 1998; Li et al., 2002; Tse et al., 2004; Wang et al., 2011). The TMD and CT of a VSR are necessary and sufficient for its targeting of PVCs in plant cells (Jiang and Rogers, 1998; Tse et al., 2004; Miao et al., 2006). The N-terminus of the VSRs is believed to be responsible for sorting cargo proteins at the late Golgi or TGN via a specific receptor–cargo interaction (Cao et al., 2000; Suen et al., 2010), packing them in the clathrin-coated vesicles (CCVs) and delivering them to PVCs (Bethke and Jones, 2000) for subsequent cargo delivery to vacuoles while the VSRs are recycled back to the late Golgi (Oliviusson et al., 2006; daSilva et al., 2006; Saint-Jean et al., 2010). Nevertheless, recent studies using transient expression system in tobacco protoplasts suggested that such cargo–receptor interaction could initiate already in the endoplasmic reticulum (ER) and a retromer could play a role at the level of the TGN (Niemes et al., 2010a,b). Known Arabidopsis native VSR cargo proteins include the AsnProIleArg (NPIR)-containing vacuolar cysteine protease Ataleurain, Arabidopsis carboxypeptidase Y (AtCPY) and the Arabidopsis storage protein 2S albumin and 12S globulin (Shimada et al., 2003; Fuji et al., 2007; Zouhar et al., 2010; Lee et al., 2013).

In Arabidopsis, there are seven VSR proteins (AtVSR17) with highly conserved N-terminal domains, but little is known about the identity of their cargo proteins. Proteomic analysis of isolated vacuoles has identified hundreds of soluble vacuolar proteins in Arabidopsis (Carter et al., 2004) and many of these are likely to be sorted by AtVSRs. To identify and characterize the cargo proteins of AtVSRs in Arabidopsis we developed and tested an in vivo system, which is based on the premise that a truncated VSR lacking its TMD/CT when expressed in Arabidopsis suspension culture cells will be secreted along with its cargo proteins into the culture medium allowing their identification via mass spectrometry.

As a proof-of-principle, we expressed the functional VSR cargo-binding domains (VSRNTs) in Arabidopsis PSB-D culture cells. Therefore, cargo proteins, co-secreted along with VSRNTs into the culture medium, were further identified by mass spectrometry. The identities of some of these cargo proteins were confirmed by localization studies, interaction analysis with VSRs and VSD analysis. The expression of functional VSRNTs in Arabidopsis culture cells therefore provides a useful platform for identifying the cargo proteins and to better understand cargoreceptor interaction mechanisms in vivo. This method can also be extended to VSR proteins and their cargoes from other plant species.


Cargo proteins are co-secreted with the truncated VSRNT protein into the culture medium

Since the VSR N-terminus (NT) is responsible for cargo binding while its TMD/CT is responsible for targeting (Cao et al., 2000), we have hypothesized that over-expression of truncated VSR (VSRNT) in transgenic plant cells would result in secretion of VSRNT proteins together with their specific cargo proteins into the culture medium (Figure 1). To test this hypothesis, we first systematically compared the effects of increasing concentrations of BP80NT-T7 on trafficking of aleurain-GFP to the vacuole in a transient expression system (Figure 2a,b). In Arabidopsis protoplasts, aleurain-GFP trafficked to the central vacuole, forming two protein species at 33 and 28 kDa when detected with a GFP antibody. The 33-kDa species corresponds to the intact protein, and the 28-kDa species is the processed form in the central vacuole, termed the GFP core (Figure 2b). Thus, the amount of the 28-kDa species can be used to measure vacuolar trafficking efficiency. The secGFP (a secreted form of GFP)-expressing protoplasts, served as a positive control (daSilva et al., 2005). We then examined the dose dependence of cargo secretion, using constant amounts of aleurain-GFP and increasing amounts of BP80NT-T7 DNA. In this secretion assay, the protoplasts were split into equal portions for transfection, and equal amounts of extracted proteins were detected by Western blotting using the cis-Golgi marker ManI (anti-ManI). BP80NT-T7 protein expression increased with increasing amounts of BP80NT-T7 DNA as indicated using T7 antibody in the protoplast fraction. The amount of the 33-kDa processed form in the protoplasts gradually decreased with increasing amounts of BP80NT-T7, and concomitantly the amount of aleurain-GFP secreted into the medium was increased. Interestingly, aleurain-GFP was partially proteolytically processed to a 28-kDa form in the medium, which may be attributable to the presence of extracellular proteases (Matsuoka et al., 1995; Frigerio et al., 1998; Kim et al., 2005). The increase of VSRNT protein in the medium was detected by a VSR antibody (Figure 2b), which further confirmed that the secretion of aleurain-GFP was because of the secretion of BP80NT-T7. The cytoplasmic marker tubulin was not detected using Anti-TUB in the medium even when the film was exposed for a long time, which excludes the possibility that the presence of BP80NT-T7 and aleurain-GFP in the medium was due to leakage from damaged protoplasts. Taken together, these data demonstrate that the cargo proteins can be co-secreted together with VSRNT.

Figure 1.

Model of protein trafficking, structure and predicted subcellular localization of full-length, truncated vacuolar sorting receptor (VSR) and cargo proteins. The full-length VSR traffics from the late Golgi or trans-Golgi network (TGN) to the pre-vacuolar compartment (PVC) mediating vacuolar delivery of cargo proteins while the truncated VSR (VSRNT) will be secreted into the culture medium of transgenic cells, together with their corresponding cargo proteins.

Figure 2.

Cargo protein aleurain-GFP co-secreted with BP80NT-T7 in Arabidopsis protoplasts. (a) Constructs and predicted molecular weight of aleurain-GFP and BP80NT-T7. (b) Dosage-dependent secretion of BP80NT-T7 and aleurain-GFP. The asterisk and double asterisk indicated the intact aleurain-GFP and the processed GFP core, respectively.

Generation of Arabidopsis suspension cell culture lines of VSRNT

As a next step, we generated a stable Arabidopsis cell culture line expressing three truncated VSR proteins, BP80NT, AtVSR1NT and AtVSR4NT, having a T7 tag at their C-terminus. The expressed BP80NT-T7, AtVSR1NT-T7 and AtVSR4NT-T7 also accumulated in the culture medium of the transgenic PSB-D cells but not in wild type (WT) cells (Figure 3a). In fact, in Arabidopsis WT cells, anti-VSR antibodies only detected the endogenous VSR proteins in the intracellular fraction that were missing from the secreted medium fraction (Figure 3b, single asterisk). In contrast, in transgenic Arabidopsis PSB-D cells expressing the truncated BP80NT, a protein band corresponding to the truncated BP80NT-T7 (Figure 3b, double asterisk) in addition to the presence of the endogenous VSR proteins (Figure 3b, single asterisk), was detected in the intracellular fraction. The secreted form of BP80NT but not the endogenous VSR was detected in the secreted medium fraction (Figure 3b, double asterisk). Similar results were also obtained from transgenic PSB-D cells expressing the truncated AtVSR1NT-T7 or AtVSR4NT-T7. The T7 antibody further proved the detection in which only the truncated form of VSRNT was detected (Figure 3b). Western blotting analysis on intracellular and secreted proteins of both WT and transgenic VSRNT PSB-D cells further indicated that there is no leakage of the cellular components into the secreted fraction and the method of collecting the secreted medium fraction faithfully reflects the integrity of the culture cells without any cell contamination (Figure S1).

Figure 3.

Expression and characterization of truncated BP80 (BP80NT-T7), AtVSR1 (AtVSR1NT-T7) and AtVSR4 (AtVSR4NT-T7) in a transgenic Arabidopsis PSB-D cell. (a) Western blot analysis of the secreted proteins in the culture medium of wild-type (WT) or transgenic cells using vacuolar sorting receptor (VSR) antibodies. The asterisk indicates the secreted truncated VSRs (VSRNTs) in three transgenic cell lines (a, b and c). The highest expression lines (b in BP80NT, c in AtVSR1 and c in AtVSR4, respectively) were selected for further analysis. (b) Western blot analysis of intracellular (I) and secreted (S) proteins isolated from either WT or transgenic Arabidopsis PSB-D cells. The asterisk indicates the endogenous full length VSRs, while the double asterisk indicates the secreted VSRNTs.

Furthermore, the BP80NT-T7, AtVSR1NT-T7 and AtVSR4NT-T7 expressed in Arabidopsis PSB-D culture cells are functional proteins that are able to interact specifically against the synthetic VSD peptides but not their mutant forms (Figure S2). Therefore, based on the hypothesis that cargo proteins of individual VSRs will be co-secreted together along with the VSRNT into the medium allows for their detection and identification by mass spectrometry.

Identification of the secreted VSRNT cargo protein by mass spectrometry

In the WT cell culture medium, a total of 86 proteins were identified via nano liquid chromatography tandem mass spectrometry (NANO-LC-MS/MS) and Fourier transform (FT)-MS analysis. These proteins are mostly cell wall-related proteins, such as glycoside hydrolyses, esterases, lyases, expansins, lectin domain-containing proteins, leucine-rich repeat domain-containing proteins and enzyme inhibitors (Table S1). In the transgenic cell line medium we could identify the corresponding VSRNT protein (BP80NT, AtVSR1NT, and AtVSR4NT) in all samples. In addition, we identified 17 putative cargo proteins that were uniquely found in the medium of transgenic PSB-D cell lines (Table 1), including three vacuolar proteases (At3 g52500, At5 g67360 and At4 g16190), five glycosyl hydrolases (At4 g34480, At5 g10560, At5 g11720, At5 g45280 and At5 g58090), three oxireductases (At1 g30740, At5 g06720 and At5 g11540), two chitinases (At2 g43570 and At3 g12500), a redox enzyme glutaredoxin (GRX, At5 g20500), a curculin-like (mannose-binding) lectin family protein (At1 g78850), a cysteine protease inhibitor family protein (CPI, At4 g16500) and an unknown protein (At4 g34260). Out of the 17 identified proteins, none of them had been previously described as a VSR-interacting protein, and only four of them (At4 g16190, At5 g10560, At4 g16500 and At5 g11720) were previously reported as vacuolar proteins (Carter et al., 2004). Among the identified potential cargo proteins, each of the truncated VSR proteins (BP80NT, AtVSR1NT and AtVSR4NT) secreted different kinds of proteins, and yet shared common cargo proteins (Table 1).

Table 1. The identified putative vacuolar sorting receptor cargo proteins
Protein nameAccessionLocusMWpIBP80NT (total score/peptide no.)AtVSR1NT (total score/peptide no.)AtVSR4NT (total score/peptide no.)
  1. MW, molecular weight in Da; pI, isoelectric point. ‘Y’ indicates the presence of this protein in the medium of transgenic PSB-D cell lines. The total score and peptide number of mass spectrometry are in parentheses with Fourier transform mass spectrometry results underlined.

FAD-binding domain-containing proteingi|15221497At1 g3074059,869.18.9 Y (106/3)(97/2) 
Curculin-like (mannose-binding) lectin family proteingi|15219200At1 g7885049,050.97.7Y (166/3)(136/3)Y (152/2)(131/3)Y (177/3)(145/3)
Glutaredoxingi|98960865At5 g2050014,827.05.8Y (90/2)(85/2)Y (125/3)(98/2)Y (121/3)(102/2)
Putative chitinasegi|15224308At2 g4357029,775.15.8 Y (100/2)(137/3) 
Aspartyl protease family proteingi|18409620At3 g5250051,004.58.3  Y (98/2)(191/3)
Chitinasegi|332641682At3 g1250036,183.47.7Y (73/2)(107/2)Y (73/2)(107/2) 
Cysteine protease inhibitor family proteingi|15235771At4 g1650012,555.89.7Y (90/2)(175/5) Y (117/2)(215/5)
Unknown proteingi|30689979At4 g3426093,724.57.3 Y (64/3) 
Glycosyl hydrolase family 17 proteingi|30690053At4 g3448053,120.96.4 Y (92/2)(57/3) 
Putative peroxidasegi|21593262At5 g0672034,988.74.5Y (151/4)Y (147/3)Y (159/4)(127/2)
Glycosyl hydrolase family 3 proteingi|15238197At5 g1056087,184.36.3Y (74/4)(150/4) Y (136/3)(165/4)
FAD-binding domain-containing proteingi|15239081At5 g1154064,960.68.4Y (75/3)  
Alpha-glucosidase 1gi|2323344At5 g11720101,117.45.7Y (122/5)Y (145/6) 
Pectin acetylesterasegi|21593445At5 g4528040,789.110.3 Y (73/2) 
Glycosyl hydrolase family 17 proteingi|22327934At5 g5809052,211.05.7 Y (63/2)(115/3) 
Subtilasegi|18425181At5 g6736079,414.56.3Y (410/7)(216/4)Y (387/6)(217/4) 
Cysteine proteinase, putativegi|18414611At4 g1619041,263.67.0Y (65/2)(153/4)Y (98/3)(135/3)Y (98/2)(88/3)

Putative cargo protein localized in the PVC and finally transported into the vacuole

The VSR-sorted cargo proteins are believed to traffic through PVCs prior to reaching vacuoles in plant cells (Miao et al., 2006, 2008). Therefore, PVC localization of the cargo can be confirmed by co-localization with VSR marker proteins. Four putative cargo proteins, including cysteine proteinase (CP, At4 g16190), glycosyl hydrolase family 3 protein (GH3, At5 g10560), GRX and CPI that were well-expressed in Arabidopsis PSB-D protoplasts were further analyzed in the following experiments. As a positive control, aleurain-GFP showed both a punctate pattern co-localizing with the PVC marker mRFP-AtVSR2 (85.7 ± 1.0%) and a diffuse pattern in the vacuole (Figure 4a). Similarly, CP-GFP and GRX-GFP gave rise to punctate fluorescence within the cytoplasm, which co-localized with the PVC marker mRFP-AtVSR2 (83.9 ± 1.9 and 83.3 ± 0.9%, respectively), and a diffuse pattern in the vacuole after a 12 h incubation (Figure 4b,c). GH3-GFP and CPI-GFP showed a slight ER pattern and co-localized with the PVC marker mRFP-AtVSR2 as punctate PVC signals after a 12 h incubation (86.6 ± 0.8 and 82.3 ± 1.0%, respectively). Finally, GH3-GFP and CPI-GFP also showed a diffuse fluorescence pattern in the vacuole after incubation for 18 and 24 h, indicating their accumulation in the vacuole (Figure 4d,e). These cargo proteins partially co-localized with the cis-Golgi ManI-mRFP marker or the TGN marker mRFP-SYP61 (Figure S3), probably owing to saturation of the sorting system in the cis-Golgi or TGN, thus delaying the transport of these over-expressed cargoes through these organelles.

Figure 4.

The identified vacuolar sorting receptor (VSR) cargo proteins reached vacuoles via pre-vacuolar compartments (PVCs). (a) The cargo protein aleurain-GFP co-localized with the PVC marker (mRFP-AtVSR2) and transported into vacuoles in Arabidopsis protoplasts. (b–e) The identified VSR cargo proteins cysteine proteinase (CP)-GFP (b), glutaredoxin (GRX)-GFP (c), glycosyl hydrolase family 3 (GH3)-GFP (d), or cysteine protease inhibitor (CPI)-GFP (e) co-localized with the PVC marker (mRFP-AtVSR2) and accumulated in vacuoles in Arabidopsis protoplasts. DIC, differential interference contrast. Scale bar = 50 μm.

To confirm the trafficking of these putative cargo proteins, we also used AtRabF2b (Ara7) as a tool to study the cargo proteins. In Arabidopsis PSB-D protoplasts, when mRFP-Ara7 was co-expressed with aleurain-GFP, the mRFP-Ara7 labelled structures co-localized with the aleurain-GFP-positive PVCs (Figure S4a). Ara7(Q69L), which promotes the GTP-bound state (Kotzer et al., 2004), was also used because its expression induces PVC enlargement derived from homotypic PVC fusion (Jia et al., 2013). When mRFP-Ara7(Q69L) was co-expressed with aleurain-GFP, the bright, punctate, aleurain-containing PVCs assumed a ring-like structure and the soluble marker aleurain-GFP accumulated in the PVC lumen (Figure S4b). Therefore, the PVC-localized cargo proteins should have the same characteristics when co-expressed with Ara7(Q69L). Indeed, when the putative cargo proteins CP-GFP, GH3-GFP, CPI-GFP and GRX-GFP were co-expressed together with the mRFP-Ara7(Q69L) in Arabidopsis PSB-D protoplasts, the signals for the soluble GFP form were found mainly inside the vacuolated PVCs, with mRFP-Ara7(Q69L) signals being found on their limiting membrane (Figure S4c–f). These results confirm that GFP-tagged putative cargo proteins are targeted to the lytic vacuole via the PVCs.

If the GFP-tagged cargo proteins were to transport into the vacuole, the full-length precursor proteins would be cleaved into their mature form and a GFP core (Kim et al., 2010; Zouhar et al., 2010; Lee et al., 2013). To test this possibility and find out the nature of the protein of these identified cargos, Arabidopsis PSB-D protoplasts were transformed with individual cargo proteins (CP-GFP, GH3-GFP, CPI-GFP and GRX-GFP), followed by isolation of soluble proteins from intracellular fraction and subsequent Western blot analysis using anti-GFP antibodies. Indeed, as shown in Figure 5, in all tested samples, a large amount of the 28-kDa processed GFP core was detected, probably owing to processing of the cargo proteins from the precursor form into the mature form during vacuole trafficking. These results point to the vacuole trafficking of these cargo proteins.

Figure 5.

Western blot analysis of expressed GFP tagged cargo proteins. The asterisk and double asterisk indicate precursor and mature GFP-tagged cargo proteins, respectively. The arrow indicates the processed GFP core.

Co-immunoprecipitation of putative cargo proteins and VSRs

To explore the interaction specificity between identified cargo proteins and VSRs, in vitro co-immunoprecipitation studies were further carried out. The T7-tagged versions of BP80NT, AtVSR1NT and AtVSR4NT were transiently expressed together with GFP-tagged cargo proteins in protoplasts from PSB-D suspension cells. Subsequently, protein extracts from transformed protoplasts were subjected to immunoprecipitation using the GFP-Trap, which is based on recombinant single domain antibody fragments (VHHs) coupled to agarose beads. secGFP was used as a negative control for the co-immunoprecipitation. The immunoprecipitates were analyzed by Western blotting using anti-GFP or anti-T7 antibodies. As shown in Figure 6(a–c), secGFP did not interact with BP80NT-T7, AtVSR1NT-T7 or AtVSR4NT-T7 because there was no protein band detected in the immunoprecipitates using T7 antibodies, whereas aleurain-GFP was found to interact with BP80NT-T7, AtVSR1NT-T7 and AtVSR4NT-T7. In addition, BP80NT-T7 was detected in the immune complexes when co-transformed with CP-GFP or CPI-GFP individually but with low signal intensity with GH3-GFP and GRX-GFP. These results indicate that BP80 may interact with nearly all identified cargo proteins, although with different binding affinities. In contrast, AtVSR1NT-T7 was not detected in the immunoprecipitates when co-expressed with GH3-GFP or CPI-GFP, indicating no interaction between AtVSR1 and these two cargoes. Further quantification of the corresponding signal intensities (Figure 6d) indicated that GRX-GFP has a high binding affinity with AtVSR1NT-T7, reaching 98.3% binding affinity compared with that of CP-GFP (100%). Moreover, AtVSR4NT-T7 showed relatively low intensity in immunoprecipitates in protoplasts transformed with GH3-GFP or GRX-GFP, with 16.3 and 61.1% binding affinity, respectively. Taken together, these data point to specific interaction between various cargo proteins and different VSRs.z

Figure 6.

Co-immunoprecipitation analysis on specific vacuolar sorting receptor (VSR) and cargo interaction. (a–c) Soluble proteins were isolated from Arabidopsis protoplasts co-expressing truncated BP80NT-T7 (a), AtVSR1NT-T7 (b) or AtVSR4NT-T7 (c) with individual GFP-tagged cargo proteins as indicated, followed by immunoprecipitation (IP) using GFP-Trap agarose beads and subsequent immunoblot (IB) analysis on eluted proteins using GFP or T7 antibodies. P, immunoprecipitates; T, total protein extracts (10% of input). (d) Quantification of co-immunoprecipitation. The same amount of plasmids of VSRNT-T7 (15 μg) and GFP-tagged cargo proteins (25 μg) was used for electroporation in each experiment. The amount of cargo binding to individual truncated VSRs (VSRNTs) was compared with that of cysteine proteinase (CP)-GFP (as 100% binding affinity). The intensity of the co-immunoprecipitate was quantified using imagej software and normalized by the input. The images of the blot were captured at different exposure time points. For quantification, we selected images that yielded a linear relationship between the intensity and exposure time. To quantify the degree of co-immunoprecipitation, three independent immunoprecipitation experiments were performed. Error bars are the SD from three independent experiments. The t-test was performed for each experiment in comparison with the CP-GFP binding data. Asterisks indicate significant difference (P < 0.05).

The identified cargo proteins contain VSDs that are necessary and sufficient for PVC targeting

Proteins destined for vacuoles usually contain VSDs that are recognized by VSRs. Basically, three different types of VSDs have been studied: (i) sequence-specific VSDs (ssVSD, NTPP) (Matsuoka and Nakamura, 1991; Frigerio et al., 2001; Shimada et al., 2002), (ii) non-sequence-specific C-terminal VSDs (ctVSD, CTPP), and (iii) large internal stretches of polypeptide (physical structure VSD) in certain storage proteins for PSV sorting (Saalbach et al., 1991; Vonschaewen and Chrispeels, 1993). The VSDs of the putative cargo proteins were further investigated by software analysis and vacuolar targeting analysis.

The putative VSDs in these 17 identified cargoes were analyzed using meme (Bailey and Elkan, 1994) and are shown in Table 2. One protein (CP) contains a NPIR motif and 10 other proteins have NPIR variants following the predicted signal peptide cleavage site, in the middle or at the C-terminus. In addition, several other proteins, including subtilase (At5 g67360), chitinase (At3 g12500) and two glycosyl hydrolase family 17 proteins (At4 g34480 and At5 g58090), were identified to contain a highly hydrophobic C-terminal extension, probably representing ctVSD for vacuole transport in Arabidopsis. However, two proteins (GRX and At3 g52500) contain no known VSDs.

Table 2. The putative vacuolar sorting determinants (VSDs) in the identified cargo proteins
Protein nameLocusSequenceVSD type and location
  1. The putative VSDs are underlined. ssVSD, sequence-specific VSD; ctVSD, non-sequence-specific C-terminal VSD.

FAD-binding domain-containing proteinAt1 g3074096AKTLNLQLKIRSGGHDYD113ssVSD, middle
Curculin-like (mannose-binding) lectin family proteinAt1 g78850266PKHNATLSFIRLESDGNI283ssVSD, middle
GlutaredoxinAt5 g20500
Putative chitinaseAt2 g43570227SFKSGFGATIRAVNSREC244ssVSD, middle
Aspartyl protease family proteinAt3 g52500
ChitinaseAt3 g12500306YCNIFGVNPGGNLDCYNQRSFVNGLLEAAI-335ctVSD, C-terminus
Cysteine protease inhibitor family proteinAt4 g1650026GGGLGSRKPIKNVSDPDV43ssVSD, N-terminus
Unknown proteinAt4 g3426028VRNPVRPRSSERRALMDG45ssVSD, N-terminus
Glycosyl hydrolase family 17 proteinAt4 g34480475EKKNGATEPKVSSSLSFLLIFLSLIFHVYM-504ctVSD, C-terminus
Putative peroxidaseAt5 g0672058SDTRIGASLIRLHFHDCF75ssVSD, N-terminus
Glycosyl hydrolase family 3 proteinAt5 g10560648FEYKILSAPIRLSLSELL665ssVSD, middle
FAD-binding domain-containing proteinAt5 g1154025VPPQPPIRCDQTGCTVSN42ssVSD, N-terminus
Alpha-glucosidase 1At5 g1172078SLETSERLRIRITDSSQQ95ssVSD, N-terminus
Pectin acetylesteraseAt5 g45280107NPDFYNWNRIKVRYCDGS124ssVSD, middle
Glycosyl hydrolase family 17 proteinAt5 g58090448MIEPYYGGAAREHGFFFPLLMVAAIAVSIF-477ctVSD, C-terminus
SubtilaseAt5 g67360728SKPSGSNSFGSIEWSDGKHVVGSPVAISWT-757ctVSD, C-terminus
Cysteine proteinase, putativeAt4 g1619027EVTDGFVNPIRQVVPEEN44ssVSD, N-terminus

We next performed a localization study to confirm the predicted VSDs of CP (Figure 7a), CPI (Figure 7b) and GH3 (Figure 7c) using GFP fusions with predicted VSDs and alanine mutants (VSDM) in Arabidopsis PSB-D protoplasts. Upon transient co-expression, CP-VSD-GFP (Figure 7e), CPI-VSD-GFP (Figure 7f) and GFP-GH3-VSD (Figure 7g) co-localized with the PVC marker mRFP-AtVSR2 (84.0 ± 2.0, 83.3 ± 1.4 and 79.5 ± 0.8%, respectively), and transported into the vacuole after incubation for 12 or 24 h. The punctate dots of CP-VSD-GFP (Figure S5a,b), CPI-VSD-GFP (Figure S5c,d) and GFP-GH3-VSD (Figure S5e,f) were partially co-localized with the cis-Golgi marker ManI-mRFP or the TGN marker mRFP-SPY61. However, when their VSDs were subjected to alanine mutation (VSDM), all the resulting mutant fusions CP-VSDM-GFP, CPI-VSDM-GFP and GFP-GH3-VSDM were found to be separated from the PVC marker mRFP-AtVSR2, and did not reach the vacuole even after incubation for 24 h (Figure S6). In contract, the punctate dots of CP-VSDM-GFP (Figure S7a,b) and CPI-VSDM-GFP (Figure S7c,d) were co-localized with the cis-Golgi marker ManI-mRFP but separated from the TGN marker mRFP-SPY61. No co-localization was observed between GFP-GH3-VSDM and the cis-Golgi marker ManI-mRFP or TGN marker mRFP-SPY61 (Figure S7e,f).

Figure 7.

Analysis of vacuolar sorting determinants (VSDs) of putative cargo proteins. (a–d) Constructs for studying putative VSDs (highlighted in red) of cysteine proteinase (CP) (a), cysteine protease inhibitor (CPI) (b), glycosyl hydrolase family 3 (GH3) (c), and glutaredoxin (GRX) (d). (e–h) CP-VSD-GFP (e, green), CPI-VSD-GFP (f, green), GFP-GH3-VSD (g, green) or GRX-VSD-GFP (h, green) co-localized with the pre-vacuolar compartment (PVC) marker (mRFP-AtVSR2, red) and accumulated in vacuoles in Arabidopsis protoplasts.

The GRX does not contain classically known ssVSD and ctVSD. To study its putative VSD-containing region, we fused its first 24 amino acids, containing signal peptide, with GFP (Figure 7d). The resulting GRX-VSD-GFP showed both punctate dots and vacuole patterns (Figure 7h), whereas the punctate dots of GRX-VSD-GFP were found to be co-localized with the PVC marker mRFP-AtVSR2 (85.4 ± 3.3%), but separated from the cis-Golgi marker ManI-mRFP and occasionally co-localized with TGN marker mRFP-SPY61 (11.5 ± 1.1%; Figure S5g,h). These results indicated that the first 24 amino acids of GRX contain sufficient information for its PVC-vacuole targeting. Indeed, when its hydrophobic amino acid sequence VKKTI was substituted by alanine (Figure S6d) the resulting GRX-VSDM-GFP fusion was found to be retained in the ER when co-expressed with mRFP-AtVSR2 (Figure S6 h), ManI-mRFP (Figure S7 g) or mRFP-SPY61 (Figure S7 h). Thus, we have defined the VKKTI sequence as being a VSD for GRX.

Cargo proteins are sorted into vacuoles because their VSDs are recognized by VSRs. However, when a VSD is mutated, the cargo will be secreted into the culture medium (Koide et al., 1999; Castelli and Vitale, 2005; Claude et al., 2005; Maruyama et al., 2006). We thus next test if VSD mutation on the identified cargos will result in their secretion upon their expression. Arabidopsis protoplasts were first transformed with various GFP fusions with cargo VSDs or their alanine mutants, followed by protein isolation from intracellular (I) and extracellular medium (or secretion, S) for subsequent Western blot analysis using anti-GFP antibodies. As shown in Figure 8(a), 18 h after transformation, CP-VSD-GFP, GFP-GH3-VSD, CPI-VSD-GFP and GRX-VSD-GFP were detected as two protein bands of 31 and 28 kDa in the intracellular fraction (I) with no detectable signal in the extracellular medium (S). The 31-kDa band corresponds to the intact GFP fusion protein whereas the 28-kDa band is the processed GFP core form in the central vacuole. In contrast, identical experiments using their VSD alanine mutants (CP-VSDM-GFP, GFP-GH3-VSDM, CPI-VSDM-GFP and GRX-VSDM-GFP) resulted in the detection of the 31-kDa protein band in the extracellular medium (S), together with little GFP core signal (Figure 8a). The amount of secreted protein was highest for GRX-VSDM-GFP (44.2%) and lowest for GFP-GH3-VSDM (25.6%, Figure 8b). Taken together, these results indicate that VSD alanine mutants caused secretion of these cargo proteins.

Figure 8.

Secretion assay of vacuolar sorting determinant (VSD) and VSD alanine mutant (VSDM) GFP fusions. (a) Western blot analysis of VSD and VSDM GFP fusions trafficking in Arabidopsis PSB-D protoplasts. Protoplasts were transformed with indicated constructs, followed by protein isolation from intracellular (I) and extracellular medium (S), and subsequent Western blot analysis using GFP antibodies. The asterisk indicates the full-length GFP fusion protein (I or S). The double asterisk indicates the processed GFP core. (b) Measurement of protein secretion. To measure the efficiency of protein secretion, the amount of full-length GFP fusion (GFP-Fusion), processed GFP core (GFP-Core), and secreted proteins (Secretion) was quantified using imagej software and expressed as a value relative to the amount of total protein (as 100%). Tubulin was used as a control for both protein leakage from protoplasts and protein loading. Three independent experiments were performed to obtain an average targeting efficiency.


Cargo proteins can be identified by over-expressing VSRNT

Vacuolar sorting receptor proteins are important for correct and efficient transport of soluble proteins to vacuoles in the plant secretory pathway (Paris and Neuhaus, 2002; Laval et al., 2003; Shimada et al., 2003). Various methods have been used to elucidate the function of these receptors in sorting cargo proteins. For example, heterologous expression of truncated BP80 in Drosophila S2 cells has been used to study the BP80–proaleurain peptide interaction (Cao et al., 2000), the over-expression of GFP fusions with the TMD/CT of AtVSRs induces the secretion of cargo proteins (daSilva et al., 2005, 2006), and Arabidopsis mutants have been used for studying the function of AtVSRs in Arabidopsis for the transport of 2S albumin and 12S globulin to protein storage vacuoles and Arabidopsis carboxypeptidase Y (AtCPY) and Ataleurain to lytic vacuoles (Shimada et al., 2003; Craddock et al., 2008; Zouhar et al., 2010; Lee et al., 2013). However, plants contain multiple copies of VSR genes, for example the Arabidopsis genome contains seven VSRs with different spatial and temporal expression profiles (Laval et al., 2003; Shimada et al., 2003; Raikhel et al., 2008). This suggests a possible functional diversity and redundancy throughout various stages in plant development, even though all these AtVSRs are found in the same PVC populations in Arabidopsis cells as well as in TGN in Arabidopsis seeds. Therefore, it is difficult and time-consuming to study the functions of individual VSR proteins, and especially to identify the specific cargo proteins of individual VSRs.

To address this issue we have developed and tested an expression system using suspension cells to produce functional truncated VSR (BP80NT, AtVSR1NT and AtVSR4NT) proteins that are secreted into the culture medium (Suen et al., 2010). Hence cargo proteins should be co-secreted attached to the truncated VSR proteins, thus allowing the identification of cargo proteins through proteomic techniques providing clues to the function of individual VSR proteins (Figure 1). Here, by taking advantage of the Arabidopsis protein database, we have presented a pilot study of the functions of three truncated VSR proteins, BP80NT, AtVSR1NT and AtVSR4NT, in Arabidopsis. We successfully expressed these three VSRNT proteins in Arabidopsis PSB-D culture cells and followed their secretion into the culture medium. These VSRNTs proteins bind to synthetic peptides containing VSDs from various known cargo proteins, suggesting that they are functionally indistinguishably from those VSRNTs previously studied (Cao et al., 2000; Suen et al., 2010). Indeed, potential cargo proteins are uniquely found in the culture medium of transgenic PSB-D culture cells expressing BP80NT, AtVSR1NT and AtVSR4NT. Each truncated VSR secreted its own group of interacting cargo proteins while sharing common cargo proteins (Table 1), which suggests that they have overlapping functions yet possess diversity in the vacuolar transport of cargo proteins. In addition, none of these potential cargo proteins has ever been reported as being an interacting partner with VSR protein.

Over-expression of one VSR isoform might have a dominant negative effect on VSRs, resulting in non-specific secretion of cargo proteins. Our mass spectrometry analysis demonstrated that each truncated VSR protein caused the secretion of its own group of interacting cargo proteins that have specific interactions with different VSRs. Thus, overexpression of VSRNT in the suspension culture system had relatively little effect on normal VSR function for non-specific secretion; rather, such system can be used to identify cargos specific for individual VSRs.

The VSR cargo proteins in Arabidopsis

Major types of VSR cargo proteins identified in the PSB-D cells in this study are related to vacuolar proteases, glycosyl hydrolases, oxireductases, chitinases, redox enzyme, a curculin-like (mannose-binding) lectin family protein and a cysteine protease inhibitor family protein. These results are consistent with a proteomic study in which vacuolar proteases, glycosyl hydrolases and chitinases were identified in the Arabidopsis vegetative vacuole (Carter et al., 2004). Moreover, three proteases (CP, the aspartyl protease family protein and subtilase) are identified in VSRNT cell line medium, thus providing hints for functional characterization of these vacuolar proteins in the future.

AtVSR1 was previously shown to able to interact with the synthetic peptide of Ataleurain (Shimada et al., 2003); however, Ataleurain was not identified in the culture medium of the three transgenic cell lines expressing the VSRNTs. One possible explanation is that the intense competition between the VSRNT and the endogenous VSR proteins for Ataleurain interaction may result in little secretion of the cargo for detection. Indeed, endogenous Ataleurain was found to be secreted only in vsr1vsr4 double mutants, but not in vsr1 or vsr4 single mutants (Zouhar et al., 2010; Lee et al., 2013), indicating that both AtVSR1 and AtVSR4 play a part in vacuole transport of Ataleurain. We would thus expect to see the secretion and detection of Ataleurain in transgenic cell lines co-expressing both AtVSR1NT and AtVSR4NT.

Vacuolar sorting determinants

This study identified 11 proteins with potential ssVSDs. The ssVSD usually located immediately after the predicted signal peptide cleavage site, but some functional ssVSDs are also found in the middle or at the C-terminus of the protein (Frigerio et al., 2001). Thus, the putative ssVSDs identified in this study may represent another target for studying the ssVSD pathway in Arabidopsis.

The ctVSD pathway has been proven to be functional for targeting ctVSD proteins from other plant species to the vacuole in Arabidopsis (Rojo et al., 2002; Hunter et al., 2007; Lee et al., 2013). It is thus likely that Arabidopsis proteins carry functional ctVSDs that are yet to be identified. In this study, we have identified several Arabidopsis proteins with a highly hydrophobic C-terminal extension, including the Arabidopsis chitinase (At3 g12500), two glycosyl hydrolase family 17 proteins (At4 g34480 and At5 g58090) and subtilase (At5 g67360). Chitinase comes in two different types that function in different organelles in tobacco. The basic chitinase is located in the vacuole while an acidic chitinase is secreted outside (Neuhaus et al., 1991, 1994); the ctVSD is shown to be present in the tobacco basic chitinase, but not in the tobacco acidic chitinase. The Arabidopsis chitinase (At3 g12500) identified in this study lacks the NPIR-type VSD motif but contains the hydrophobic C-terminal sequence (GLLEAAI), which is similar to the VSD motif (GLLVDTM) found in tobacco basic chitinase. Interestingly, the putative chitinase (At2 g43570) in our VSR cargo list contains the NPIR-type motif (Table 2). Therefore, Arabidopsis probably contains two distinct vacuolar sorting pathways (i.e. ssVSD and ctVSD) for chitinases.

One identified VSR cargo protein, GRX, does not contain a classical NPIR-like motif nor a hydrophobic C-terminal extension, but alanine substitution mutation for the hydrophobic sequence after the signal peptide (VKKTI) affected its PVC-vacuole transport. This result indicates that alternative mechanisms for the ssVSD and ctVSD pathways are likely to exist for vacuolar transport in Arabidopsis. The ER bodies may represent another method of protein storage (Matsushima et al., 2003; Wang et al., 2013). However, GRX does not contain any known ER retention signal and GRX reached the vacuole via PVC.

In conclusion, we have successfully shown that expression of truncated VSR proteins in Arabidopsis PSB-D culture cells results in the co-secretion of VSR-interacting cargo proteins, which is a useful platform for identification of cargo proteins from the culture medium of transgenic culture cells through mass spectrometry, and thus allow us to elucidate the potential function of individual VSR proteins in their trafficking of cargo proteins in the plant secretory pathway. Through this platform we can proceed from the pea BP80 and the two Arabidopsis VSRs to study the other Arabidopsis VSR proteins, and furthermore to study the functions of VSR proteins from the model plant, rice, where currently little is known on the function of the rice VSR proteins.

Experimental Procedures

Plasmid construction

Construction of truncated BP80 (BP80NT-T7) and Arabidopsis AtVSR4 (AtVSR4NT-T7) has been described previously (Suen et al., 2010). The truncated Arabidopsis AtVSR1 (AtVSR1NT-T7) were generated similarly in PBI221 and PBI121. All fluorescent fusion constructs used for transient expression in protoplasts were PCR amplified and cloned into the pre-made GFP backbone in pBI221. The primers used to generate these constructs are listed in Table S2. All constructs were confirmed by restriction mapping and DNA sequencing.

Transient expression and transformation of Arabidopsis suspension cells

Transient expression using Arabidopsis protoplasts has been described previously (Miao and Jiang, 2007). Transformation and generation of Arabidopsis PSB-D cells have also been described (Wang et al., 2010; Cai et al., 2012).

Confocal microscopy

Confocal images were collected at specific time points after transformation using an Olympus FV1000 system ( For each experiment, more than 50 individual cells were observed for imaging that represented >80% of the cells showing similar expression levels and patterns. Images collected from 10 individual cells were used for data processing and analysis. Images were processed using adobe photoshop software ( as previously described (Jiang and Rogers, 1998).

Secretion assay and immunoblot analysis

Secretion assay was performed as described previously (Suen et al., 2010). Intracellular and medium proteins were prepared from 7-day-old WT or transgenic Arabidopsis cells expressing VSRNT, followed by protein separation via SDS-PAGE and Western blot analysis using either VSR or other control antibodies as indicated. The immunoblots were quantified by measuring the intensity of the protein bands with imagej software.

Protein preparation from transient expressed protoplasts

Protein extractions from protoplasts and culture medium were performed as described previously with some modification (Kim et al., 2005; Shen et al., 2013). Proteins in protoplasts were extracted with the extraction buffer [250 mm 2-amino-2-(hydroxymethyl)-1,3-propanediol (TRIS)-HCl, pH 7.4, 750 mm NaCl, 5 mm EDTA, 0.5 mm phenylmethanesulfonyl fluoride (PMSF), and 25 μg ml−1 leupeptin) by sonication for 5 sec. The supernatant was recovered for analysis after centrifugation at 25 000 g at 4°C for 10 min. To prepare proteins from the culture medium, cold trichloroacetic acid (TCA) (100 μl) was added to the medium (1 ml) in the presence of BSA as a carrier, and protein aggregates were precipitated by centrifugation at 25 000 g at 4°C for 30 min. Protein aggregates were washed twice with acetone, air-dried and dissolved in an SDS-PAGE sample loading buffer.

Pull-down assay and co-immunoprecipitation

Synthetic peptides containing wild type (WT) VSDs or their mutated form of vacuolar proteins were synthesized by Genescript. Construction of peptide sepharose and pull-down assay from cell culture medium was done as before (Suen et al., 2010). To perform pull-down experiments from transient expressed protoplasts, the membrane fraction was collected as previously described (Gao et al., 2012) and resuspended in CHAPS buffer (Paris et al., 1997) containing 1% 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS). The supernatant was mixed with sepharose conjugated with peptides at 4°C overnight. After washing with CHAPS buffer, the proteins in sepharose were subject to SDS-PAGE and Western blot by specific antibodies.

To perform co-immunoprecipitation experiments from transient expressed protoplasts, the soluble fractions were prepared in co-immunoprecipitation buffer, containing 25 mm HEPES, pH 7.1, 150 mm NaCl, 1 mm MgCl2, 1000 μm CaCl2 and 0.5% (v/v) Triton X-100, and were then incubated with GFP-Trap agarose beads (ChromoTek, overnight at 4°C in a top-to-end rotator. After incubation, the beads were washed with co-immunoprecipitation buffer and then eluted by boiling in SDS sample buffer.

Protein preparation for mass spectrometry analysis

Secreted proteins present in the culture medium were collected from day-4 culture media of WT and transgenic PSB-D, and were then precipitated by methanol/chloroform (Wessel and Flugge, 1984). The protein pellet was then cold-dried, rehydrated in 25 mm ammonium bicarbonate containing sequencing grade modified trypsin (Promega, After tryptic digestion overnight at 28°C, peptides were finally redissolved in 2% acetonitrile/0.05% trifluoroacetic acid for NANO-LC-MS/MS or 0.1% formaic acid for FT-MS.

Nano liquid chromatography tandem mass spectrometry

The digested peptide mixture was loaded to PepMap100 C18 nano column of UltiMate 3000 Intelligent LC system (Dionex, in order to perform NANO-LC. Collected samples were further analyzed using a matrix-assisted laser desorption/ionization (MALDI) time-of-flight/time-of-flight (TOF/TOF) analyzer (ABI4700 Proteomics Analyzer; Applied Biosystems, The search software mascot (Matrix Science, was used to manually search against the National Center for Biotechnology Information database. Only peptides achieving a significant mascot ion score above 30 and containing a sequence tag of at least three consecutive amino acids were accepted. At least three independent repeats were performed, and for each of the repeats, the digested peptides are injected twice.

Fourier-transform ion cyclotron resonance mass spectrometry (FT-MS)

Analyses were performed using an electrospray ionization source (Apex ultra Hybrid Qq-FTMS; Bruker Daltonics, coupled to a liquid chromatography (Ultimate 3000; Dionex). The system was controlled with the software package hystar (v.3.4 build 8; Bruker Daltonics). Sample extract was loaded onto PepMap100 C18 nano column and was eluted when the spectra were acquired in a Fourier transformed full scan model. The MS/MS raw spectral data were converted to mgf files using dataanalysis 4.0 SP1 (Bruker Daltonics) and analyzed by means of biotools 3.1 (Bruker Daltonics). mascot was used to manually search against the National Center for Biotechnology Information database. Only peptides achieving a significant mascot ion score above 30 and containing a sequence tag of at least three consecutive amino acids were accepted. At least three independent repeats were performed, and for each of the repeats the digested peptides are injected twice.

Analysis of the secreted proteins

Secreted proteins identified in the Arabidopsis WT PSB-D were pooled together. Only those proteins identified in the transgenic PSB-D that are found in at least two of the independent repeated mass spectrometry experiments are accepted as secreted proteins of the transgenic PSB-D cells. By comparing transgenic cell protein with the WT, unique proteins that found only in the transgenic Arabidopsis PSB-D cells were identified. Only those unique proteins that have a mascot ion score above 50, with at least two peptides, and which contain a signal peptide (required for default entry into the ER) but no TMD were accepted as potential cargo proteins. Signal peptide was predicted by signalp v3.0 (; Nielsen et al., 2007). The transmembrane domain was predicted by tmhmm v2.0 ( Protein masses and isoelectric points were determined with the TAIR bulk protein search tool (

To identify the putative sequence-specific vacuolar sorting determinants (ssVSDs), the program meme (Bailey and Elkan, 1994) was performed using the pattern discovery approach based upon other known ssVSDs from sporamin (Matsuoka and Nakamura, 1991), aleurain (Holwerda et al., 1992), PT20 (Koide et al., 1999), ricin (Frigerio et al., 2001) and 2S albumin (Matsuoka and Neuhaus, 1999). The hydrophobicity profile was determined with the emboss package (

Accession numbers

Sequence data in this article are in the GenBank under the following accession numbers: U79958 (BP80), O22925 (AtVSR2), AF126550 (ManI), NM_102617 (SYP61), U31094 (aleurain), AY063095 (AtRabF2b).


We thank Professor Takashi Ueda (University of Tokyo, Japan) for the cDNA encoding ARA7(Q69L) constructs. This work was supported by grants from the Research Grants Council of Hong Kong (CUHK466309, CUHK466610, CUHK466011, CUHK465112, and CUHK2/CRF/11G), NSFC/RGC (N_CUHK406/12), NSFC (31270226), Shenzhen Peacock Plan (KQTD201101) and CUHK Schemes to LJ.

Conflict of Interest

The authors declare that they have no conflict of interest.