Two different gene families have been proposed to act as sorting receptors for vacuolar storage cargo in plants: the vacuolar sorting receptors (VSRs) and the receptor homology-transmembrane-RING H2 domain proteins (RMRs). However, functional data on these genes is scarce and the identity of the sorting receptor for storage proteins remains controversial. Through a genetic screen we have identified the mtv2 mutant, which is defective in vacuolar transport of the storage cargo VAC2 in shoot apices. Map-based cloning revealed that mtv2 is a loss of function allele of the VSR4 gene. We show that VSR1, VSR3 and VSR4, but not the remaining VSRs or RMRs, participate in vacuolar sorting of VAC2 in vegetative tissues, and 12S globulins and 2S albumins in seeds, an activity that is essential for seedling germination vigor. Finally, we demonstrate that the functional diversification in the VSR family results from divergent expression patterns and also from distinct sorting activities of the family members.
The distinguishing feature of spermatophyta is the production of seeds, where nutrients are stored for use by the embryo during germination. In some species, including Arabidopsis, large amounts of storage proteins are synthesized in embryo cells during seed development and are densely packed in protein storage vacuoles (PSVs) that occupy most of the cellular volume by maturity. This massive amount of seed proteins constitutes one of the main agricultural commodities obtained from crops. Even though transport of vacuolar storage proteins is predominant in seeds, it also occurs in vegetative tissues (Sanmartin et al., 2007). Therefore, understanding the trafficking of storage proteins is essential not only to modify protein accumulation in seeds, but also to improve the storage capacity of vegetative tissues.
It is well documented that the default destination for soluble proteins in the secretory pathway of plants is the apoplasm (Denecke et al., 1990; Crofts et al., 1999; Phillipson et al., 2001). Transport to the vacuole requires an active targeting mechanism that selects the vacuolar cargo and diverts it from bulk flow secretion. This selection process relies on the presence of vacuolar sorting determinants (VSDs) in the cargo protein sequence. It is assumed that VSDs are recognized by membrane bound sorting receptors that recruit them into vesicles destined for the vacuole. However, the identity of the sorting receptors for the different vacuolar cargo, including storage proteins in seeds, remains unknown. In yeast, a single vacuolar cargo sorting receptor, Vps10p, has been found (Marcusson et al., 1994). In mammalian cells, two types of mannose-6-phosphate receptors (Ghosh et al., 2003) and sortilin, a Vps10p homologue (Petersen et al., 1997), have been shown to function as sorting receptors for vacuolar proteins. The Arabidopsis genome does not encode homologues of either Vps10p or the mannose-6-phosphate receptors that could act as vacuolar sorting receptors for seed storage proteins. Instead, two distinct plant-specific protein families have been proposed to fulfill that role. The vacuolar sorting receptor (VSR) family, made up of seven members in Arabidopsis, and the receptor homology-transmembrane-RING H2 domain protein (RMR) family, made up of six members in Arabidopsis. Biochemical evidence suggests that VSRs can interact in vitro with sorting signals from several vacuolar storage proteins (Watanabe et al., 2002; Jolliffe et al., 2004; Fuji et al., 2007). VSRs are present at the trans-side of the Golgi apparatus, most likely at the trans-Golgi network (TGN), and at the pre-vacuolar compartment (PVC) in Arabidopsis (Sanderfoot et al., 1998), a distribution consistent with a role in vacuolar cargo sorting. Moreover, VSRs are recycled by retrograde transport to participate in further rounds of cargo sorting (daSilva et al., 2005; Oliviusson et al., 2006; Niemes et al., 2010). Direct evidence for a role in vacuolar sorting was provided by the analysis of the vsr1 mutant, which is defective in trafficking of endogenous seed storage proteins (Shimada et al., 2003). However, the effect of the vsr1 null mutation on storage protein transport is partial and mutations in the other VSR genes have no effect (Shimada et al., 2003), which has been interpreted as evidence that VSRs are just salvage receptors for stray storage protein that has escaped the main sorting mechanism (Hinz et al., 2007; Craddock et al., 2008). In addition, other evidence supports the role of RMRs as sorting receptors for seed storage cargo. RMR1 interacts with VSDs from storage proteins and dominant negative versions of RMR block the exit of phaseolin from the Golgi in Arabidopsis protoplasts (Park et al., 2005, 2007). However, genetic evidence for a role of RMRs in sorting endogenous storage proteins has not been presented yet. Recently, co-localization of RMRs and VSRs with storage cargo in Arabidopsis seeds was analysed (Otegui et al., 2006; Hinz et al., 2007). The results obtained in those studies are compatible with either family acting as storage protein sorting receptors, and therefore have not settled the issue.
We have developed a genetic screen to search for Arabidopsis mtv mutants (for ‘modified transport to the vacuole’) that are affected in the trafficking of the vacuolar storage protein cargo VAC2 (Sanmartin et al., 2007). VAC2 is a translational fusion of the extracellular ligand CLAVATA3 to the VSD from barley lectin driven by the 35S promoter (Rojo et al., 2002). If trafficking is unperturbed, VAC2 is transported to the vacuole, where it is inactive, and plants have normal meristems. If trafficking is blocked and VAC2 is secreted, it induces premature shoot apical meristem (SAM) and flower meristem termination, allowing identification of the mtv mutants. In this screen transport is assayed in vegetative tissues, where storage protein trafficking is less prevalent than in seeds and may be more sensitive to perturbations. Indeed, this distinctive feature has allowed the identification of factors that are limiting for transport in vegetative tissues but not in seeds (Sanmartin et al., 2007; this study; M. Sauer, O. Delgadillo, and E. Rojo, unpublished results). We report here that defective storage protein trafficking in the mtv2 mutant is due to a mutation in VSR4. Moreover, analysis of the whole VSR and RMR gene families shows that VSR1, VSR3 and VSR4 are key sorting receptors for specific storage cargo in seeds and vegetative tissues.
MTV2 encodes the VSR4 protein required for VAC2 sorting to the vacuole
Using the mtv genetic assay, mutants with different degrees of meristem termination, depending on the levels of VAC2 secreted, can be isolated. In mutants with strong phenotypes, such as the mtv2 mutant reported here, most SAMs terminate prior to production of floral meristems, and the few formed do not produce stamens and carpels. However, production of flowers with all the whorls can be restored in these mutants by exogenous treatment with cytokinins, which counteract CLV3-induced meristem termination. Incidentally, this supports the hypothesis that cytokinins participate in the feedback loop that determines SAM stem cell pool size downstream of WUSCHEL (Tucker and Laux, 2007), and epistatically on CLV3 (our observations). By treating the apices of the mtv2 mutant repeatedly with 6-benzylaminopurine, a few flowers eventually developed (Figure 1a), which were used for crossing with the Col-0 ecotype for mapping purposes. The F1 plants had wild type (Wt) meristems, indicating that the mtv2 mutation is recessive. In the F2 mapping population we recovered plants with extreme meristem termination that did not produce flowers with sexual organs, similar to the original mtv2 plant (Figure 1b), and also plants with an intermediate phenotype (similar to that shown in Figure 1g). The strongest phenotype corresponded to mtv2 mutants homozygous for VAC2, while the intermediate phenotype corresponded to mtv2 plants with a single VAC2 copy. The distribution of plants with the strong phenotype (mtv2 homozygous for VAC2) in the mapping population (115 out of 1923 plants) is consistent with the number expected for a recessive mutation unlinked to the insertion site of the VAC2. Indeed, the mutagenized parental line has a single VAC2 transgene inserted in chromosome III while MTV2 maps to chromosome II. Mapping data from those 115 F2 individuals with the strong phenotype restricted the position of MTV2 to a 169-gene interval (At2g14080–At2g15360) close to the centromere, which included 78 genes listed as transposable element genes or pseudogenes. The remaining set of 91 genes included two members of the VSR family (Figure 1c). We sequenced both genes and detected a G/A substitution at the splice acceptor site of the eighth intron of the VSR4 gene. The substitution creates a new splice acceptor site displaced 1 bp from the original one, which leads to a frame shift in the mRNA and the presence of a stop codon four amino acids downstream of the mutation. The mutated mRNA was predicted to code for a truncated protein lacking two EGF repeats, the transmembrane domain and the C-terminal cytosolic tail (Figure 1d). We detected an extra band of higher mobility recognized by anti-VSR antibodies in protein extracts from mtv2 plants (Figure 1e), proving that a truncated VSR4 protein was indeed synthesized in the mutant. The transmembrane and cytosolic domains are necessary for linking cargo binding by the luminal domain to recruitment into vesicles. Thus, the truncated VSR4 would not be functional, which is consistent with the recessive nature of the mtv2 mutation. We have isolated another allele of VSR4 from the SALK T-DNA collection, designated vsr4-2. The T-DNA in vsr4-2 is inserted in the fourth exon and the mutant does not accumulate VSR4 transcript or protein (Figures 2a and 5a), suggesting that the allele is null. We crossed-in the VAC2 construct and obtained homozygous vsr4-2 VAC2 plants that displayed strong meristem termination phenotypes similar to homozygous mtv2 VAC2 plants (Figure 1f). We then performed an allelism test by crossing mtv2 plants heterozygous for VAC2 with vsr4-2 plants without the transgene. Out of 28 plants from the F1 generation, 15 plants had a copy of the VAC2 transgene and showed an intermediate meristem termination phenotype (Figure 1g), while the other 13 plants without the VAC2 transgene had Wt meristem development. These results suggest that mtv2 and vsr4-2 are allelic, and confirm that VSR4 is required for efficient vacuolar trafficking of VAC2 in cells of the shoot apex. Moreover, the mtv2 and vsr4-2 plants without VAC2 are indistinguishable from Wt plants, implying that the trafficking defects are not due to gross abnormalities in the mutants that could indirectly affect endomembrane transport.
VSR1 and VSR3 are involved in VAC2 sorting in shoot apices
To test whether other members of the VSR family also participate in VAC2 sorting in the shoot apex we obtained T-DNA mutants in VSR1, VSR2, VSR3, VSR5, VSR6 and VSR7, and crossed-in the VAC2 transgene. In vsr1, vsr2, vsr3 and vsr5 the T-DNAs are inserted in exons, in vsr6 in an intron, and in vsr7-1 and vsr7-2 in the 3′ UTR. Except in the case of the vsr7 alleles, full-length transcripts or transcripts that bypass the T-DNA and include sequences after the insertion do not accumulate (Figure 2a). Even if partial transcripts 5′ of the insertion accumulate and are translated, the truncated proteins encoded would not be functional because they would lack critical parts of the protein such as the transmembrane and cytosolic domains. In contrast, the vsr7-1 and vsr7-2 alleles accumulate full-length transcript, albeit at reduced levels (Figure 2b). The F1 generation from all the crosses displayed Wt meristems. In the F2 populations we observed plants with terminated meristems only in the crosses with vsr1 and vsr3 plants. We confirmed by genotyping that only vsr1 and vsr3 mutants containing a VAC2 transgene had a meristem termination phenotype, whereas vsr2, vsr5, vsr6, vsr7-1 and vsr7-2 mutants homozygous for VAC2 had Wt meristems (Figure 2c,d). Interestingly, the vsr3 mutant displayed a weaker termination phenotype than vsr4 or vsr1 mutants (Figure 2d). More rosette leaves were produced before termination of the primary SAM, and when the vsr3 VAC2 plants eventually bolted from lateral meristems, they generated flowers with all the whorls that fructified normally (Figure 2d, right panel). The VSR3 and VSR4 genes share the same probes in microarray chips but ESTs can be assigned to either gene. The relative EST abundance suggests that the expression of VSR3 is several-fold lower than that of VSR4 in aerial tissues, consistent with the weaker phenotype of vsr3 VAC2 mutant. These results indicate that VSR4 and VSR1, and to a lower degree VSR3, are rate limiting for VAC2 sorting in shoot apices, while VSR2, VSR5 and VSR6 are not. Although the vsr7 alleles are not null and the participation of VSR7 in VAC2 sorting cannot be excluded, our results below suggest that VSR7 has a different sorting activity than the VSR1, VSR3 and VSR4 group.
VSR1, VSR3 and VSR4 are the main sorting receptors for seed storage proteins
Processing of seed storage proteins occurs in the pre-vacuolar compartment and the vacuole (Otegui et al., 2006). The accumulation of unprocessed precursors of seed storage proteins is a hallmark of mutants that secrete these proteins (Shimada et al., 2003; Fuji et al., 2007; Ebine et al., 2008) or are blocked in earlier steps of the pathway (Li et al., 2006). For instance, the vsr1 mutant accumulates precursors of the main Arabidopsis storage proteins, 12S globulins and 2S albumins, and antibodies against those proteins label abnormal electron-dense aggregates found in the extracellular space of mature seeds (Shimada et al., 2003). In contrast, mutants in the other VSR genes do not secrete seed storage proteins (Shimada et al., 2003). Moreover, since the bulk of 2S albumins and a large fraction of 12S globulins still reach the vacuole in the vsr1 mutant, it has been suggested that VSRs are not the main receptors for storage protein sorting but may instead redirect to the vacuole storage proteins that has escaped earlier aggregation-based sorting in the cis-Golgi (Hinz et al., 2007; Craddock et al., 2008; Pompa et al., 2010). However, an alternative possibility is that VSRs are the sorting receptors for storage proteins in seeds, but redundancy within the gene family reduces the effect of mutations in single VSR genes. The promoters of VSR1 and VSR4 are active in developing embryos (Figure 3a), consistent with the expression of the endogenous genes in those tissues inferred by RT-PCR analysis (Laval et al., 2003). The overlapping expression is compatible with VSR1 and VSR4 displaying genetic redundancy in seeds. Indeed, the vsr1 and vsr4 mutations synergistically block transport of 12S globulins when combined (Figure 3b–e), suggesting that they cooperate in sorting this class of proteins. In contrast, the ratio of secreted to vacuolar forms of 2S albumins in the vsr1vsr4 mutant is very low and unchanged relative to the single vsr1 mutant (Figure 3b,c), while in a vsr5vsr6 mutant sorting of 12S globulins and 2S albumins is not affected (Figure S1). Importantly, in the vsr1vsr4 mutant there is almost no accumulation of the vacuolar forms of 12S globulins (Figure 3b), suggesting that VSR1 and VSR4 are bonafide vacuolar sorting receptors for these proteins and not part of a salvage pathway. The PSVs in vsr1vsr4 embryos are smaller than in vsr1, vsr4 and Wt plants (11.5 ± 6.3 μm2 in vsr1vsr4; 14.2 ± 7.4 μm2 in vsr1; 18.5 ± 6.3 μm2 in vsr4 and 19.8 ± 9.2 μm2 in Wt). As a consequence, PSVs in double mutant embryos do not fill the whole cellular volume and maintain rounded shapes (Figure 3d). In contrast Wt and vsr4 embryos have tightly appressed and amorphous PSVs, while vsr1 embryos have amorphous PSVs that appear less tightly packed than those in Wt seeds. In addition, the vsr1vsr4 mutant accumulates large amounts of extracellular electron-dense aggregates, which are not observed in Wt plants and are less prominent in vsr1 mutants (Figure 3e). Taken together, these results suggest that VSR1 and VSR4 have overlapping expression domains and a redundant function in embryos, where they are responsible for sorting the bulk of 12S globulins to the PSVs. VSR3 is also expressed in seeds (Avila et al., 2008), indicating that the residual amounts of vacuolar forms of 12S globulins still detectable in the vsr1vsr4 mutant may due to the activity of this gene. Although a triple mutant would unequivocally confirm this, VSR3 and VSR4 are located in tandem in chromosome II, so the isolation of such a mutant would be extremely laborious, if not unfeasible because of a complete block in germination (see below). Importantly, the double vsr1vsr3 mutant also has reduced levels of the vacuolar forms of 12S globulins relative to the single vsr1 mutant (Figure 3f), suggesting that indeed VSR3 contributes to transport of this class of proteins and together with VSR1 and VSR4 sorts the whole pool of 12S globulins. Moreover, in the vsr1vsr3 mutant the levels of 2S albumin precursors are drastically increased relative to the vsr1 mutant, indicating that VSR1 and VSR3 sort a significant fraction of this class of storage proteins. Considering that 12S globulins and 2S albumins are transported together from the Golgi to the pre-vacuolar compartment (Otegui et al., 2006), the differential effect on transport of these proteins is consistent with sorting, not general vesicle trafficking, being affected in the vsr mutants.
VSR1, VSR3 and VSR4 determine germination vigor
During germination, seed storage proteins are hydrolysed to provide amino acids for the growing embryo. Interestingly, vsr1 seedlings have been reported to have normal germination and growth (Shimada et al., 2003), even though they have partial defects in vacuolar accumulation of storage proteins. This may be because the alterations are not sufficient to affect germination. Alternatively, since the secreted precursors are also hydrolysed during germination (Shimada et al., 2003), vacuolar localization may not be a requisite for their mobilization and use. To resolve this question we analysed germination in the vsr1vsr3 and vsr1vsr4 mutants, where vacuolar accumulation of storage proteins is largely inhibited. We observed a very strong reduction in vigor in the vsr1vsr3 and vsr1vsr4 mutants (Figure 4a), which compromised severely seedling establishment of the double mutants when sown directly in soil (data not shown). Analysis of the total cellular proteins in the vsr1vsr4 mutant shows that the reduced vigor and seedling establishment is coupled to a delay in synthesizing the proteins necessary for initiating photoautotrophic growth (Figure 4b, compare levels of Rubisco). These results suggest that VSR1-, VSR3-, and VSR4-dependent deposition of storage proteins in PSVs during embryogenesis is essential later on for vigorous germination and seedling establishment.
Redundant function of VSR1 and VSR4 in AtAleurain sorting in leaves
Expression analysis of VSR1 and VSR4 in the Genevestigator database (Zimmermann et al., 2004) suggests that transcript accumulation is maximal in seeds, but is also significant in vegetative tissues. This is consistent with the activity of their promoters (Figure 3a) and with their role in VAC2 trafficking. Indeed, western blot analysis confirms that VSR1 and VSR4 proteins accumulate in leaves (Figure 5a). However, the endogenous cargo for these proteins in vegetative tissues is unknown. We analysed leaf apoplastic fractions for possible secretion of vacuolar cargo in vsr mutants. In single vsr1 and vsr4 mutant we did not detect secretion of any of the vacuolar proteins tested (Figure 5b), and neither in vsr5, vsr6 or vsr5vsr6 plants (Figure 5c). Importantly, we detected secretion of AtAleurain, but not of other vacuolar cargo (AtCPY, VPEγ and TGG2) in vsr1vsr4 plants, indicating that VSR1 and VSR4 have a redundant role in specific sorting of AtAleurain in leaves. Previous experiments suggested that heterologous expression in Arabidopsis leaves of an ER-retained form of PV72, a seed specific VSR from pumpkin, partially blocks exit of AtAleurain from the ER (Watanabe et al., 2004). Our data provide genetic proof that VSRs, and in particular VSR1 and VSR4, contribute to AtAleurain vacuolar targeting. Importantly, secretion of AtAleurain in the vsr1vsr4 plants is marginal, indicating that VSR1 and VSR4 sort only a minor fraction of the total pool of that protein. AtAleurain may have two different VSDs (Hinz et al., 2007), so VSR1 and VSR4 could function as receptors for one of them but not for the other, which would be responsible for vacuolar targeting of the bulk of AtAleurain. In this regard, other mutants that secrete storage proteins also secrete marginally AtAleurain in leaves (Zouhar et al., 2009), suggesting that only a minor fraction of AtAleurain utilizes a storage trafficking pathway to the vacuole.
RMRs are not rate limiting for storage protein sorting in seeds or vegetative tissues
Although several reports suggest a function of RMRs in storage protein transport (Park et al., 2005, 2007), genetic evidence backing that hypothesis has not been presented. Moreover, our results suggest that sorting of certain storage cargo in Arabidopsis is solely dependent on VSR1, VSR3 and VSR4 activity, implying that at least those proteins would not require additional receptors for their vacuolar targeting. To analyse the possible contribution of RMRs to storage protein transport, we obtained T-DNA insertional alleles. The T-DNAs are inserted in exons, and, in rmr2, rmr3, rmr4, rmr5 and rmr6 mutants, no full-length transcripts, or transcripts that include the sequence after the T-DNA, were detected (Figure 6a). If truncated proteins were synthesized, they would lack the transmembrane and cytosolic domains, except in the case of rmr3 allele that would only lack the serine rich C-terminal domain of unknown function. In the rmr1 mutant a hybrid transcript of T-DNA and RMR1 mRNA sequence accumulates (Figure 6a,b asterisk) that lacks the start codon of RMR1. If this transcript were translated in the RMR1 frame from the next methionine, the protein encoded would lack half of the RMR1 sequence, including the signal peptide and the protease-associated (PA) domain (Figure 6c), suggesting that it would be non-functional. These results indicate that rmr1, rmr2, rmr4, rmr5 and rmr6 are null alleles, while rmr3 could potentially retain function. The rmr mutants homozygous for the VAC2 transgene have Wt meristem development, indicating that vacuolar transport of VAC2 is independent on RMR activity. Moreover, in seeds of rmr mutants no precursor forms of 12S globulins and 2S albumins are observed and they accumulate Wt levels of the processed forms (Figure 6d), suggesting that transport of these proteins to the vacuole is not perturbed. Indeed, PSVs in embryos from these mutants are similar in size and shape to those Wt embryos, and fill the whole cellular volume (Figure 6e). Analysis with the ACT data-mining tool (Manfield et al., 2006) shows that RMR1 and RMR2 have a high degree of co-expression (Pearson correlation coefficient r = 0.709262), and they are the members of the RMR family most highly expressed in seeds, so redundancy could be masking their effect on storage protein trafficking. However, an rmr1rmr2 mutant did not show any seed storage protein phenotype and neither did an rmr3rmr4 mutant (Figure 6f), suggesting that the lack of storage protein trafficking defects are not due to redundancy in the gene family. In addition, we have not found any effect of the rmr1 mutation when combined with the vsr4 and vti12 mutations (Figure 6f), which in contrast have a clear seed storage phenotype when combined with the vti11 (Sanmartin et al., 2007) or vsr1 mutations (this work). These data suggest that RMRs, in contrast to VSRs, are not limiting for sorting of storage proteins in seeds, nor in vegetative tissues.
Functional diversification within the VSR gene family
Our results show that VSR1, VSR3 and VSR4 function in vacuolar transport of VAC2 in shoot apices, and demonstrate the genetic redundancy between those genes in sorting storage proteins in seeds and AtAleurain in leaves. In contrast, no evidence for VSR2, VSR5, VSR6 and VSR7 functioning in sorting storage cargo in seeds nor in vegetative tissues was found by examining the corresponding mutants (Figure 2; Shimada et al., 2003). However, these latter genes could have similar sorting activity as VSR1/VSR3/VSR4, but may be expressed at low levels in the tissues analysed and their contribution may have remained unnoticed. For instance, we could only detect VSR2 transcripts in flower tissues (Figure 2a), consistent with previous results (Laval et al., 2003) and with data from the Genevestigator database showing pollen specific expression. Thus, although VSR2 has high sequence identity with VSR1/VSR3/VSR4 in the PA domain (Table S1) that confers ligand-binding specificity, and probably has a similar sorting activity, this would only be revealed if tested in pollen. Some seed storage genes are expressed in pollen (Santos-Mendoza et al., 2008) and VSR2 may be involved in sorting them to the vacuole. In contrast, the PA domain of VSR5, VSR6 and VSR7 are more divergent and may confer different sorting activities to these proteins. Previous results suggest that VSR5, VSR6 and VSR7 are expressed in vegetative tissues but not in mature seeds (Laval et al., 2003). Moreover analysis of microarray and proteomic data at Genevestigator and the AtProteome database (http://fgcz-atproteome.unizh.ch) indicates that they are expressed mainly in roots, where they would be the predominant VSRs. To determine their spatial pattern of expression, we analysed the activity of promoter-GUS fusions. VSR5 and VSR6 have almost identical transcriptional patterns (Pearson correlation coefficient r = 0.8366), and as the absolute expression is higher for VSR6, we studied the VSR6 promoter as representative for these two genes. Strong activity of the VSR6 and VSR7 promoters is limited to roots, coinciding with the microarray and proteomic data. There was no activity of the VSR6 and VSR7 promoters in embryos or in shoot apices (Figure 7a,b), implying that a phenotype in seeds and shoot apices would not be expected for vsr5, vsr6 and vsr7 mutants, even if the corresponding proteins had the same sorting activity as VSR1 and VSR4. Therefore, to unequivocally resolve whether VSRs have or not distinct sorting activities, we compared the ability of the coding sequences from the different VSR cDNAs, expressed under the control of the VSR1 promoter and the nopaline synthase (NOS) terminator sequence, to complement sorting of 12S globulins in the vsr1 background. We analysed T2 seeds from mixtures of independent T1 transgenic lines. By pooling independent T1 lines we normalized for differences due to insertion site of the transgenes and we limited any bias that could occur during selection of homozygous transgenic lines. It should be noted that in the T2 pools, one-fourth of the seeds on average are not transgenic and thus not complemented. Even so, we detected clear restoration of 12S globulin sorting with the VSR1 and VSR4 coding sequences in all the independent pools analysed. In contrast, no complementation was observed in vsr1 plants transformed with the VSR5, VSR6 and VSR7 coding sequences (Figure 7c). Moreover, expression of the VSR coding sequences under the VSR4 promoter and the NOS terminator confirmed that VSR1 and VSR4, but not VSR5, VSR6 or VSR7, complemented the 12S globulin sorting defects of vsr1 mutants (Figure 7d). In addition, a slightly higher rate of complementation was observed for the VSR1 promoter and coding sequences compared to the corresponding VSR4 sequences, which was more evident in a vsr1vsr4 background (Figure 7e). The increased effectiveness of the VSR1 sequences in restoring 12S globulin sorting explains the higher contribution of the endogenous VSR1 to the transport of these proteins. While testing these constructs, we generated an additional set with the VSR coding sequences under the control of the VSR1 promoter and 3′ UTR, including this time the VSR3 coding sequence in the analysis. Analysis of seeds from T2 pools showed that VSR1, VSR3 and VSR4 efficiently sort 12S globulins while the activity of VSR5, VSR6 and VSR7 is negligible (data not shown). In plants homozygous for single insertions of the transgene, full complementation was obtained with the VSR1 coding sequence, and almost full complementation with the VSR3 and VSR4 sequences, while no complementation was observed with the VSR5, VSR6 and VSR7 sequences (Figures 7f and S2). Although the anti-VSR sera that we have available does not recognize VSR5, VSR6 and VSR7 (data not shown), we confirmed accumulation of transgenic transcripts in the lines generated (Figure S2 and data not shown).
All together, the results presented here firmly establish that VSR1, VSR3 and VSR4 function redundantly as sorting receptors for particular storage cargo in seeds and vegetative tissues, whilst they suggest that VSR5, VSR6 and VSR7 have distinct activities. VSRs have all been localized to the pre-vacuolar compartment (Miao et al., 2006), indicating that their distinctive activity is not due to differential subcellular localization but rather to different receptor affinities.
The presence of multiple VSR genes confers robustness and functional diversification
The organization of the endomembrane system has unique features in plants. Pertinent cases are the presence of distinct functional vacuoles, such as the lytic and storage vacuoles, compared for example to the single vacuole present in yeast, and a possible plant-specific retrograde pathway for VSR recycling (Niemes et al., 2010). The higher complexity of the plant vacuolar system may in part explain the increase in the number of vacuolar sorting receptors found (thirteen putative sorting receptors in Arabidopsis versus only one in yeast). Receptors with a range of sorting activities may select different cargo into diverse trafficking pathways. Our results suggest that within the VSR family in Arabidopsis at least two functional groups can be distinguished: VSR1, VSR3 and VSR4 have similar sorting specificity, which is distinct from that of VSR5, VSR6 or VSR7. This functional clustering coincides with the segregation based on sequence identity in the PA domain (Table S1): VSR1, VSR3, and VSR4 show high sequence conservation among themselves (identity 70–100%), and less with VSR5, VSR6 and VSR7 (identity 45.9–51.4%). The PA domain confers ligand-binding specificity in vitro (Cao et al., 2000) and this data is consistent with the PA domain determining substrate specificity in vivo. Supporting this, the PA domains of VSR1, VSR3 and VSR4 are more similar to those of PV72 and BP-80 (identity 64.7–85.7%), seed specific VSRs possibly involved in storage protein trafficking in pumpkin and pea (Kirsch et al., 1996; Watanabe et al., 2002), than to the paralogous VSR5, VSR6 and VSR7 proteins (identity 46.5–53.4%). Thus, it may be possible to predict a function in seed storage trafficking for uncharacterized VSR genes from other plant species based on their paralogous and orthologous identity with the Arabidopsis proteins. Interestingly, although functional domains have tighter constrains for variation, the identity between the VSR1/VSR3/VSR4/BP-80/PV72 group and the VSR5/VSR6/VSR7 group is lower in the PA domain than in the whole protein (Table S1), suggesting selective pressure for modifications in this domain that may have been the basis for functional specification within the VSR family.
In addition to functional diversification, an increased number of sorting receptors may provide robustness to crucial trafficking pathways. For instance, VSR1, VSR3 and VSR4 share sorting specificities and also expression patterns (Pearson correlation coefficient r = 0.5293, Table S2), which may confer stability against fluctuations in the expression and activity of the individual genes. This is manifested by the limited or null effect that single knockouts in these genes have on trafficking. Moreover, transcript levels of VSR1, VSR3 and VSR4 are coordinately induced coinciding with the massive synthesis of storage proteins in embryos (Genevestigator database), suggesting that their transcriptional activation serves to increase transport capacity in those cells. In this regard, their expression is induced by overexpression of LEC1, which is a master regulator of seed development that, among other functions, controls the expression of seed storage proteins (Santos-Mendoza et al., 2008). VSR5, VSR6 and VSR7 are also highly co-expressed (r ≥0.6045 in paired comparisons, Table S2), suggesting that they may also share a common function coordinately modulated at the transcriptional level. Interestingly, VSR6 and VSR7 are among a short list of primary targets for transcriptional induction by the key defense regulator NPR1 (Wang et al., 2005). Several antimicrobial vacuolar PR proteins accumulate in response to salicylic acid treatment or pathogen challenge in an NPR1-dependant manner (Wang et al., 2005) and may be the targets for sorting by VSR5, VSR6 and VSR7. Indeed roots deploy constitutively many of those antimicrobial defenses to fend off the abundant pathogens present in the soil, which could explain the high expression of VSR5, VSR6 and VSR7 in those tissues.
How are 2S albumins sorted?
In the vsr1vsr3 and vsr1vsr4 mutants there are high levels of processed 2S albumins, suggesting that a large fraction of these proteins is still transported to the vacuole. This is in contrast to the accumulation of processed 12S globulins, which is largely reduced in these mutants. Indeed, considering the effects of each mutation, we can infer that VSR1, VSR3 and VSR4 sort the entire pool of 12S globulins, while a similar conclusion cannot be directly reached for 2S albumins. This may mean that an independent mechanism, such as the aggregation-based sorting proposed for vacuolar storage proteins from other species (Castelli and Vitale, 2005; von Lupke et al., 2008; Pompa et al., 2010), participates together with VSR receptors in sorting 2S albumins in Arabidopsis. Alternatively, the transport of vacuolar 2S albumins in the vsr1vsr3 and vsr1vsr4 plants may depend, respectively, on the VSR4 and VSR3 activity remaining. In that case, an affinity of VSR1, VSR3 and VSR4 for 2S albumins that is greater than for 12S globulins may explain the weaker effect that the double mutants have on transport of 2S albumins.
Plant materials and growth conditions
The EMS-mutagenized population, derived from the L1 transgenic line (Sanmartin et al., 2007), was previously described (Sohn et al., 2007). The vsr1-1 allele is in the Ws background (Shimada et al., 2003). The other vsr and rmr mutants are in the Col-0 background and were obtained from the Nottingham Arabidopsis Stock Centre. The trafficking of all the cargo (VAC2, 12S globulins, 2S albumins and apoplastic fluids) was tested in the different pure and mixed Wt backgrounds of the single and double mutants, and in all cases no evidence of altered sorting was obtained. In figures involving mutants with different backgrounds, only one of the corresponding Wt backgrounds is shown for clarity. Plants were grown under standard long-day conditions (16 h of light, 8 h of dark) at 22°C.
Mapping of the mtv2 mutation
The tissue samples were taken from 115 plants with the strong phenotype (Figure 1b). Bulk segregant analysis was used to determine the approximate chromosomal location of the mutation. Subsequently, we restricted the interval containing the mutation by searching for recombination events in the individual plants and mapping with SSLP and CAPs markers.
The VSR1, VSR4, VSR6 and VSR7 promoters were amplified from genomic DNA of Arabidopsis thaliana seedlings ecotype Col-0 and cloned into pENTR/D-TOPO (Invitrogen, http://www.invitrogen.com/) to produce the pPVSR1, pPVSR4, pPVSR6 and pPVSR7 plasmids that were checked by sequencing. For the constructs using the NOS terminator, the VSR1, VSR3, VSR4, VSR5, VSR6 and VSR7 coding sequences were amplified with cDNA synthesized from total mRNA extracted from 10-day-old Arabidopsis thaliana seedlings ecotype Col-0, with XhoI sites in both oligonucleotides. The amplified coding sequences were cloned in pGEM-T (Promega) and verified by sequencing. The inserts were then excised with XhoI, ligated into the XhoI site of the pPVSR1 and pPVSR4 plasmids and the correct orientation verified by restriction mapping. The resulting promoter and promoter:coding-sequence constructs were transferred into the binary pGWB3 and pGWB1 plasmids respectively, using Gateway technology (Invitrogen) and used for Arabidopsis transformation.
For the constructs using the VSR1 3′ UTR, the VSR1 promoter and the 3′ UTR region were amplified from genomic DNA of Arabidopsis seedlings ecotype Col-0. The promoter was cloned into pDONR P4-P1R and the 3′ UTR into pDONR P2R-P3, and verified by sequencing. The VSR1, VSR3, VSR4, VSR5, VSR6 and VSR7 coding sequences were amplified with cDNA synthesized from total mRNA extracted from 10 days old Arabidopsis thaliana seedlings ecotype Col-0 with the primers indicated below, cloned into pDONR 207, and verified by sequencing. The VSR coding sequences in pDONR 207, the VSR1 promoter in pDONR P4-P1R, and the VSR1 3′ UTR in pDONR P2R-P3 were transfered by an LR reaction into the destination vector pGWB MS. The resulting binary plasmids were used for Arabidopsis transformation. The oligonucleotides used for cloning are listed in Table S3.
The isolation and analysis of apoplastic fluids were carried out as described (Sanmartin et al., 2007). To analyse total proteins, 1 mg of dry seeds or 20 seedlings were homogenized in a fixed volume of SDS/PAGE sample buffer and equal volumes of the extract were analysed by SDS/PAGE followed by Coomassie Blue R-250 staining. The antibodies used in this work were previously described: anti-TGG2 3D7 monoclonal antibody (Andreasson et al., 2001), anti-AtAleurain and anti-VSR (Ahmed et al., 2000), anti-AtCPY and anti-VPE (Rojo et al., 2003).
Flowers were marked at the time of anthesis and siliques collected 11 days after. The embryos were dissected from the ovules and fixed overnight at 4°C in 4% formaldehyde/4% glutaraldehyde in 50 mm phosphate buffer. They were post-fixed in 1% osmium tetroxide in 50 mm phosphate buffer for 1 h at 4°C, washed, dehydrated in an ethanol series and included in LR white resin. Ultra-thin sections were stained with uranyl acetate and lead citrate by standard procedures, and analysed in a JEOL 1200-EX II electron microscope operating at 100 kV.
VSR1: At3g52850; VSR2: At2g30290; VSR3: At2g14740; VSR4: At2g14720; VSR5: At2g34940; VSR6: At1g30900; VSR7: At5g20110; RMR1: At5g66160; RMR2: At1g71980; RMR3: At1g22670; RMR4: At4g09560; RMR5: At1g35630; RMR6: At1g35625; AtAleurain: At5g60360; AtCPY: At3g10410; AtVPEγ: At4g32940; TGG2: At5g25980. Gene annotation data is included with the manuscript.
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors is Enrique Rojo (firstname.lastname@example.org).
We are grateful to Prof. I. Hara-Nishimura for providing vsr1-1 seeds. We also thank Pilar Paredes for her technical assistance. This work was supported by the Spanish Ministerio de Educación y Ciencia (BIO2009-110784 to E.R.), the Comunidad de Madrid (P2006/GEN-0191 to E.R.), and by a post-doctoral I3P fellowship to J.Z. and a JAE Doc contract to A. M. from the Consejo Superior de Investigaciones Científicas-CSIC.