Identification of SUMO Targets by a Novel Proteomic Approach in Plants

Authors


Tel: +34 91 336 4577; Fax: +34 91 7157721; E-mail: jc.delpozo@upm.es

Abstract

Post-translational modifications (PTMs) chemically and physically alter the properties of proteins, including their folding, subcellular localization, stability, activity, and consequently their function. In spite of their relevance, studies on PTMs in plants are still limited. Small Ubiquitin-like Modifier (SUMO) modification regulates several biological processes by affecting protein-protein interactions, or changing the subcellular localizations of the target proteins. Here, we describe a novel proteomic approach to identify SUMO targets that combines 2-D liquid chromatography, immunodetection, and mass spectrometry (MS) analyses. We have applied this approach to identify nuclear SUMO targets in response to heat shock. Using a bacterial SUMOylation system, we validated that some of the targets identified here are, in fact, labeled with SUMO1. Interestingly, we found that GIGANTEA (GI), a photoperiodic-pathway protein, is modified with SUMO in response to heat shock both in vitro and in vivo.

Introduction

Recent advances in sequencing technologies have made available an increasing number of completed plant genomes. Attention is now driven towards addressing how information contained within DNA sequences can be used to elucidate the structure, function, and control of biological systems. In recent years, following the availability of genomic sequences, many gene expression studies have been performed. However, it is becoming evident that functional proteome analyses are also needed to understand biological processes. In many cases, the relationship between protein amount and transcript level is not linear; thus, quantitative proteome analysis has been proposed as an alternative or, at least, a complementary method to study gene expression at steady state or after environmental changes. Additionally, it should be taken into account that proteomes are highly dynamic, because they are continuously modified post-translationally. These post-translational modifications (PTMs) alter protein properties, function, stability, subcellular localization, and/or the ability to interact with other proteins or nucleic acids (Krueger and Srivastava 2006).

The ubiquitin system is one of the major protein modification systems required for the highly selective degradation of specific proteins in eukaryotic cells. In plants, the number of processes regulated by the ubiquitin pathway has increased very rapidly, including germination, flowering, cell division, and response to environmental stresses (Seo et al. 2004; Smalle and Vierstra 2004; Lechner et al. 2006; Dreher and Callis 2007; Jurado et al. 2008). In an effort to identify ubiquitin targets, recent proteomic analyses have reported that several hundred proteins are likely targets of ubiquitin (Maor et al. 2007; Manzano et al. 2008). However, considering the very large number of E3 ubiquitin ligases identified in plants (Lechner et al. 2006; Jain et al. 2007), it is reasonable to speculate that the number of target proteins would be much higher than the number reported. In addition, we also have to include those targets that will be modified in response to environmental changes. Other proteins, called ubiquitin-like proteins, also modulate protein function in a reversible post-translational modification, but have been shown to be involved in several cellular regulatory pathways and responses to stress aside from protein degradation. One of these ubiquitin-like proteins is the Small Ubiquitin-Related Modifier (SUMO). Protein modification with SUMO (SUMOylation) is involved in a variety of cellular processes, including signal transduction, cell cycle, transcription regulation, and DNA repair (Bossis and Melchior 2006). Proteomic approaches have been used to identify SUMOylated substrates by purification of SUMO conjugates from Saccharomyces cerevisiae (Denison et al. 2005) or mammalian cell lines (Rosas-Acosta et al. 2005). In plants, the SUMO pathway plays an important role in controlling development and response to external stimuli (Downes and Vierstra 2005; Hay 2005; Miura et al. 2007; Catalá et al. 2007; Miura et al. 2009). Recent studies using proteomic analysis (Miller et al. 2010) and two-hybrid screening (Elrouby and Coupland 2010) have reported the identification of several Arabidopsis SUMO targets, some of which were validated in a bacterial in vivo system.

Post-translational modifications are dynamic, reversible, have a fast turnover, and the stoichiometry of the modified proteins is low relative to the unmodified ones. In addition, the modifications can be removed during the protein purification process, making the identification of proteins containing PTMs difficult (Seo et al. 2004). To overcome some of these limitations and to uncover new plant proteins that have been post-translationally modified, we have developed a new proteomic approach that combines: (i) protein extraction in denaturing conditions; (ii) complete proteome fractioning by 2-D liquid chromatography using a high capacity system; and (iii) a sensitive immunodetection of PTMs (Figure S1), referred to here as 2-D liquid chromatography immunoblotting (2-D-LC-IB). In this study, we describe a novel proteomic strategy to identify proteins containing SUMO conjugates. This method also allows comparative analyses of SUMOylation under different conditions, and may also be suitable to identify other PTMs, such as phosphorylation. These immunodetected proteins were identified by mass spectrometry (MS) analyses, and SUMOylation of some targets was analyzed in a bacterial SUMO modification system. Interestingly, we showed that one of these targets is GIGANTEA (GI), a photoperiodic-pathway protein. Immunoprecipitation analyses using gi-2/35S:GI-HA plants revealed that GI is modified with SUMO in vivo in response to heat shock.

Results

Identification of SUMO targets

We have primarily used the 2-D-LC-IB methodology (Figure S1 and Materials and Methods) to identify proteins modified with SUMO, but the method is also suitable to identify other protein modifications such as phosphorylation (Figure S2). SUMO modification plays an important role in plant development as well as in plant responses to different types of stress (Downes and Vierstra 2005; Hay 2005; Miura et al. 2007; Catalá et al. 2007; Miura et al. 2009; Miller et al. 2010; Elrouby and Coupland 2010). In this study, we generated Arabidopsis transgenic plants that overexpress SUMO1 fused to the HA tag (HA3-SUMO1ox), which is specifically detected by immunoblotting using anti-HA (Figure 1). Total protein extracts from HA3-SUMO1ox plants were fractionated by 2-D liquid chromatography (see Materials and Methods). The 2-D fractions were transferred to polyvinylidene difluoride (PVDF) membranes and analyzed by immunoblotting using the anti-HA antibody (Figure 2A). These HA-immunoreactive spots were identified in the 2-D chromatography. To determine the identity of the proteins contained in the immunoreactive fractions, a subset of the immunoreactive spots were analyzed by MS, or, when indicated, by liquid chromatography MS/MS. These mass fingerprints allowed us to identify putative SUMO targets (Table 1). Next, the identified proteins were analyzed using the SUMOplot program (http://www.abgent.com/doc/sumoplot) to identify the SUMOylation consensus sequence ψ-K-x-D/E (lysine is the modified amino acid, ψ  a hydrophobic residue, and x any amino acid) in their primary sequences. As shown in Table 1, the majority of the identified proteins showed a high score SUMOylation consensus site.

Figure 1.

Immunodetection of Small Ubiquitin-like Modifier (SUMO) modified proteins.
Total proteins (20 μg) from 6-d-old wild-type (lane 1) or HA3-SUMO1ox (lane 2) seedlings or 10 μg of chloroplast protein purified from 21-d-old leaves (lane 3) were separated by sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) and analyzed by immunoblotting using immunoglobulin Gs against the HA epitope to detect Small Ubiquitin-like Modifier (SUMO) conjugates. Arrow indicates free 3xHA-SUMO1 protein, the bracket indicates HA-SUMO conjugates and the arrowhead points to a 34 kDa protein modified with SUMO that is enriched in the chloroplast fraction.

Figure 2.

Identification of Small Ubiquitin-like Modifier (SUMO) targets.
(A) Protein extracts (2.5 mg) from HA3-SUMO1ox plants were fractionated using the PF2D ProteomeLab system. All samples collected from the second dimension were transferred to polyvinylidene difluoride (PVDF) membranes by dot blot and incubated with anti-HA antibody, and the immunoreactive fractions were detected by chemiluminescence. The picture corresponds to immunoblotting analyses of fractions 27 (from spot 1 to spot 24), 28 (from spot 25 to spot 48), and 29 (from spot 49 to spot 72). (B) Representative example of the 2-D absorbance profile at 214 nm. This profile corresponds to the isoelectric-point fraction number 28 that has been separated in a reverse phase C18 column. Peaks marked with A, B, and C correspond to the immunoreactive fractions shown in A (fraction 28). Proteins contained in these fractions and other immunoreactive spots were analyzed by mass spectrometry (MS) to determine their identity (see Table 1).

Table 1.  Identification of Small Ubiquitin-like Modifier (SUMO) targets
Protein annotationIons score*TAIR numberSUMOylation consensus sequence
  1. Proteins identified as a Small Ubiquitin-like Modifier (SUMO) targets. (*) Ions score is −10*Log(P), where P is the probability that the observed match is a random event (MASCOT 2.3.5 program). SUMOylation consensus sequences were calculated with the SUMOplot program (http://www.abgent.com/doc/sumoplot). The motifs and lysines (bold) with high and low probability are indicated. (1)Identification by peptide mass fingerprint. (2)Identification by mass spectrometry (MS)/MS. (a)Protein identified in a total protein extract. (b)Protein identified in nuclei extract from heat shock-treated plants.

PBP1 (PYK10-binding protein 1) (1) (b)423At3g16420LKEE (K269)
GIGANTEA(1) (b)171At1g22770LKPP (K226) AKDD (K975)
PSBO-2/photosystem II subunit O-2 oxygen evolving)(1) (a)245At3g50820VKAD (K145)
PSBO-1 (oxygen-evolving enhancer 33)(1) (a)227At5g66570VKAD (K146)
DREPP plasma membrane family protein (2) (a) 54At4g20260VKAE (K145)
MYB4R1 (MYB domain protein 4R1)(2) (a)  3At3g18100LKQE (K364) LKKE (K495)
Pentatricopeptide repeat-containing protein (2) (a)  3At4g14820IKPD (K341)
PBP1 (pYK10-binding protein 1)(1) (a)121At3g16420LKEE (K269)
RuBisCO small subunit 2B (RBCS-2B)(1) (a) 81At5g38420Low affinity
Ania-6a type cyclin (2) (a) 29At2g01008LKVE K131
Argininine-rich cyclin 1 (2) (a) 29At2g26430LKVE K76
Lipase/Acylhydrolase with GDSL-motif family (2) 41At1g54010LKND (K189)
Putative AAA-type ATPase (2) (a)  9At2g29080IKLD (K509)
Protein phosphatase 2C-like (2) (a)  6At5g53140MKHP (K45)
Phosphatidylinositol-phosphatidylcholine transfer protein (PPT)(2) (a)  4At2g21520VKEE (K424)
MADS-box family protein (2) (a)  4At3g04100LKNP (K82)
TTL1(tetratricopetide-repeat thioredoxin-like 1)(2) (a)  1At1g53300FKLG (K511)

Unexpectedly, one of the identified candidates of SUMOylation with a high score corresponded to the chloroplast-localized protein photosystem II subunit O-2 oxygen evolving (PSBO-2; or PSBO-1, because we cannot discern which one was identified by MS due to their high homology). Because the SUMO pathway machinery has not been described in chloroplasts, we decided to analyze whether SUMO-modified proteins were localized in these organelles. We purified chloroplasts from HA3-SUMO1ox plants, and their proteins were analyzed for the presence of HA-SUMO by immunoblotting with the anti-HA antibody. As shown in Figure 1, chloroplasts contained proteins modified with SUMO, with a predominant band around 40 kDa. We detected a stronger signal of a 40 kDa HA-immunoreactive protein in the chloroplast-enriched fraction than in the total extract, suggesting that this protein is chloroplast-localized or chloroplast-membrane-attached (Figure 1, lane 3 compared to lane 2).

Identification of SUMO targets in response to heat stress

It has been reported that SUMO conjugates rapidly accumulate in response to heat stress in the nucleus (Saracco et al. 2007). For this reason, we decided to carry out comparative analyses of SUMO targets in heat shock-treated (42 °C for 30 min) versus control plants. In this case, we performed nuclei isolation in order to enrich nuclear proteins. By using an anti-HA antibody, we detected a higher amount of HA-SUMO-modified proteins in heat shock-treated plants compared to non-treated plants (Figure 3A). Both heat shock-stressed and non-stressed nuclear-enriched proteomes were fractionated and analyzed by the 2-D-LC-IB technique. First, we analyzed the first-dimension fractions of both samples by comparative immunoblotting using an anti-HA. We were able to identify several differential spots that were specifically recognized by the anti-HA (Figure 3B). We selected the fraction number 30 from the first dimension for further separation using a High Performance Reverse Phase (HPRP) 2-D column. The different fractions collected in this second chromatography were transferred to PVDF membranes and analyzed by immunoblotting using the anti-HA antibody (Figure 3C). We identified three immunoreactive spots, whose fractions were analyzed by LC high-performance liquid chromatography MS/MS. A fingerprint search allowed us to identify the PYK10-binding protein 1 (PBP1) protein, which was present in spots A and B, and the GI protein, which was present in spot C, as SUMO targets in response to heat shock. PBP1 is the PYK10-BINDING PROTEIN 1 chaperone localized predominantly in the cytosol that facilitates the correct polymerization of PYK10 when tissues are damaged and subcellular structures are destroyed by pests (Nagano et al. 2005). GI is a nuclear-localized protein that, together with CONSTANTS (CO) and FLOWERING LOCUS T (FT), promotes flowering under long days in a circadian clock-controlled flowering pathway (Imaizumi and Kay 2006). To analyze whether these two proteins were modified with SUMO, we used a bacterial SUMOylation system (Mencía and de Lorenzo 2004). When PBP1 was co-expressed with SUMO and its conjugating machinery, we found that SUMO was attached to Arabidopsis PBP1 (Figure 4). In the case of GI, in silico analyses with SUMOplot software revealed two possible SUMO attachment sites (K226 and K975). To study whether both sites were susceptible to being SUMOylated or not, we cloned GI in two halves, each containing one site (GI-Nt and GI-Ct). Using the bacterial SUMOylation system, we found that only GI-Ct is modified by SUMO, indicating that the site located in the C-terminal part of the protein is likely to be the functional one (Figure 4). Next, we wanted to study whether GI is modified with SUMO in vivo, and we thus used gi-2/35S:GI-HA plants to immunoprecipitate GI (David et al. 2006). First, we analyzed the effect of heat shock on GI by analyzing the protein level in gi-2/35S:GI-HA. Immunoblotting analyses showed that GI-HA was not degraded by heat shock, although a higher molecular band was detected (Figure 5A). GI-HA was immunoprecipitated from a total protein extract of 7-d-old seedlings (control) and from 7-d-old seedlings treated at 42 °C for 30 min. The immunoprecipitated proteins were analyzed by immunoblotting using an anti-AtSUMO1 serum. As shown in Figure 5B, GI-HA was modified by SUMO in response to heat shock stress. In addition, we found that root growth of gi-2 mutants plants was reduced compared to wild-type or gi-2/35::GI-HA after heat shock (Figure 5C). Unexpectedly, gi-2/35::GI-TAP also showed this root growth reduction, likely because the GI-TAP construction did not completely rescue the gi-2 mutation, because these plants were still exhibiting late flowering in long-day conditions (data not shown).

Figure 3.

Identification of nuclear Small Ubiquitin-like Modifier (SUMO) targets in response to heat stress.
(A) Nuclear protein extracts (5 μg) from HA3-SUMO1ox non-treated (line 1) and treated for 30 min at 42 °C plants (line 2) were separated by sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) and analyzed by immunoblotting using immunoglobulin Gs against the HA epitope to detect Small Ubiquitin-like Modifier (SUMO) conjugates. M, molecular weight. (B) Nuclear protein extracts (0.5 mg) from HA3-SUMO1ox non-treated (No HS; upper panel) and treated plants (42 °C, 30 m; bottom panel) were analyzed by 2-D liquid chromatography immunoblotting (2-D-LC-IB). All fractions from the first dimension were transferred to polyvinylidene difluoride (PVDF) membranes by dot blot and the SUMO-modified proteins were detected by immunoblotting using the anti-HA antibody. Dashed square indicates fraction number 30 in the heat shock treatment that was analyzed by the second dimension. (C) Fraction number 30 was fractionated in the second dimension using a reverse phase column, and then the different fractions were collected and analyzed by immunoblotting with anti-HA antibody. A, B, C, and D indicate different peaks that were analyzed by mass spectrometry (MS). (D) Second dimension absorbance profile at 214 nm. This profile corresponds to the isoelectric-point fraction number 30 that was separated in a reverse phase C18 column. Peaks marked with A, B, C, and D correspond to the immunoreactive fractions shown in (C). Proteins contained in these fractions were analyzed by MS to determine their identity.

Figure 4.

PBP1 and GIGANTEA (GI) are SUMOylated in bacteria.
Rosetta cells containing plasmids pET28b (1), pET28b-At1g27770-Ct (GI-Ct) (2), pET28b-At1g27770-Nt (GI-Nt) (3), and pET28b-At3g16420 (PBP1) (4), were incubated with 0.5 mmol/L Isopropyl-β-d1-thiogalactopyranoside (IPTG) at 37 °C for 2 h, and at 30 °C for 15 h. Soluble proteins were purified by using Ni+ binding columns, separated in 11% acrylamide SDS-PAGE, and analyzed by immunoblotting using anti-SUMO. Asterisks indicate GI-Ct and PBP1 modified with Small Ubiquitin-like Modifier (SUMO). Arrow and arrowheads point to higher molecular weight products of GI-Ct or PBP1 detected by the anti-SUMO, respectively, which likely correspond to poly-SUMO modified proteins. Bracket indicates higher molecular weight proteins recognized by the anti-SUMO antibody in only GI-Ct and PBP1 lanes.

Figure 5.

GIGANTEA is (GI) modified with Small Ubiquitin-like Modifier (SUMO) in vivo.
(A) Columbia-0 (Col) and overexpressing gi-2/GI-HA (35:GI-HA) were grown for 7 d at 22 °C and then one plate was transferred to a dark oven at 42 °C and another plate to a dark oven at 22 °C for 30 min. Afterwards, plates were transferred again to the growth chamber for 15 min more and then frozen in liquid nitrogen. Total protein were extracted from these samples in a TAP buffer and analyzed by immunoblotting using an anti-HA immunoglobulin Gs. The blot was stained with Ponceau to check the protein loading. Arrow indicates a higher molecular weight band that appears in the heat shock-treated sample. 1, Col at 22 °C; 2, gi-2/35S:GI-HA at 22 °C; 3, gi-2/35S:GI-HA at 42 °C. (B) Total protein from wild-type and 35S:GI-HA plants grown at 22 °C or heat shock (30 min at 42 °C) were extracted and used for immunoprecipitation analyzes using the anti-HA antibody (IPαHA). The immunoprecipitated proteins were analyzed by immunoblotting using anti-AtSUMO1 serum. The blot was stained with Ponceau to check the protein loading. Arrow indicates a GI modified with SUMO1. 1, Col at 22 °C; 2, Col at 42 °C; 3, gi-2/35S:GI-HA at 22 °C; 4, gi-2/35S:GI-HA at 42 °C; 5, (IPαHA) Col 22 °C; 6, (IPαHA) Col 42 °C; 6, (IPαHA) gi-2/35S:GI-HA 22 °C; 7, (IPαHA) gi-2/35S:GI-HA 42 °C. (C)gi-2 is more susceptible to heat shock. Col-0, gi2/35S:GI-HA, gi-2/35S:GI-TAP and gi-2 mutants were grown for 4 d in Murashige and Skoog (MS) medium and then transferred to a dark oven at 45 °C or 22 °C for 45 min. Afterwards, plates were moved back to the growth chamber and plants were grown for another 4 d. The root growth was quantified using Imagen J software and represented as a percentage of root growth of heat shock treated seedlings respect to untreated seedlings. *Student's t-test P < 0.005, n≥ 15.

Discussion

In this study, we have developed a method that combines LC and immunoblotting to detect post-translational modifications. Here, we only focused on detecting proteins modified by SUMO, but this approach can also be used to detect other modifications, such as phosphorylation, among others (Figure S2). This methodology has several advantages, because a large amount of total protein extract (up to 7 mg) can be loaded into the 2-D LC system and the results are highly reproducible between different experiments. This reproducibility allows the comparison of protein profiles from different experiments with a high degree of confidence. In addition, PTMs are detected by immunoblotting, a highly sensitive technique.

Protein modification with SUMO plays an important role in many plant responses and in development (Novatchkova et al. 2004). Genetic and biochemical analyses have shown that SIZ1 is one of the most important E3 ligases of SUMO. SIZ1 is involved in phosphate starvation (Miura et al. 2005), cold acclimation (Miura et al. 2007), heat shock tolerance (Yoo et al. 2006), drought response (Catalá et al. 2007), and abscisic acid signaling (Miura et al. 2009). Recently, the identification of several putative SUMO targets in Arabidopsis has been reported using affinity enrichment of SUMOylated proteins combined with tandem MS analyses (Miller et al. 2010). In addition to this work, using two-hybrid screening with ESD4, a SUMO protease, or SCE, a SUMO E2 conjugating enzyme, over 200 possible SUMO targets have been identified (Elrouby and Coupland 2010). The proteomic approach presented in this study is complementary to the aforementioned research, and may help uncover new SUMO targets, as in the case of GI. Thus, our approach allows for fine-tuned comparative analyses of SUMO modification in different treatments, stimuli, or mutants that can be combined simultaneously with other PTMs analyses, such as phosphorylation (Figure S2).

Proteomic analyses in yeast and mammalian cells have uncovered a large number of transcription factors and DNA-associated proteins that may be targets of the SUMO pathway, suggesting that SUMOylation plays important roles in regulating DNA transcription (Denison et al. 2005; Rosas-Acosta et al. 2005; Miller et al. 2010). In Arabidopsis, PHR1 and ICE1, MYB, and a MYB-like transcription factor involved in phosphate starvation and cold-acclimation responses, respectively, are SUMOylated by SIZ1 (Miura et al. 2007). Interestingly, we have identified the transcription factor MYB4R1 and a MADS-box protein, AGL53, as SUMO targets. It is remarkable that the majority of the proteins identified in this work to be SUMO targets seem to be regulated by cold and/or heat stress in Arabidopsis seedlings (Figure 6). Among them, the MADS-box transcription factors AGL57 and MYB4R1 are highly and moderately upregulated, respectively, by heat stress in Arabidopsis seedlings. Although it is speculative, it is possible that SUMO attachment to these proteins has a functional role during the response to heat shock.

Figure 6.

Gene expression levels of identified SUMOylated proteins during cold (blue) or heat (red) stresses.
Values are represented as average of replicate treatments relative to average of appropriate control. The picture shows a Graphical View Representation of Log Transformed Clustered Data (source: http://bar.utoronto.ca).

We also found that PSBO-2 (and/or PSBO-1), which is localized in chloroplasts, is modified with SUMO. In other proteomic studies, several chloroplast proteins were identified as SUMO targets, and some of them were shown to be modified in a bacterial system (Elrouby and Coupland 2010), supporting the idea that this organelle contains SUMOylated proteins. Nevertheless, neither PSBO-1 nor PSBO-2 proteins were identified in this study, indicating that the 2-D-LC-IB system can discover new SUMO targets. However, we cannot exclude the possibility that PSBO-1/PSBO-2 also localized to other compartments or was membrane-attached when modified with SUMO. In addition, a large number of mitochondrial SUMO targets have also been identified (Braschi et al. 2009), indicating that SUMOylated proteins are localized into different compartments in the cell. Because the SUMO conjugation machinery has not been located in chloroplasts or in mitochondria, it is possible that these targets were previously modified with SUMO in the cytoplasm before being imported into these organelles. However, future work needs to be done to clarify this point.

As mentioned above, SUMO modification plays an important role during plant stress adaptation. Our comparative analyses using 2-D-LC-IB showed that Arabidopsis PBP1 is modified with SUMO in response to heat. Human PBP1 has been identified as a SUMOylated protein (Vertegaal et al. 2006), and a proteomic study in Arabidopsis identified PBP1 as a SUMO target in response to H2O2 and heat stress (Miller et al. 2010). PBP1 is a carbohydrate-binding protein (lectin) which may act as a chaperone that facilitates the correct polymerization of the PYK10 protein – a β-glucosidase located in the endoplasmic reticulum (ER) bodies – when tissues are damaged. Thus, it is possible that the modification of PBP1 with SUMO protects it from denaturation or promotes its interaction with more proteins during heat shock to protect them.

In plants, SUMOylation/de-SUMOylation has been reported to be involved in flowering-time regulation. Several examples indicate that SUMO homeostasis is important for flowering-time regulation. For example, mutations in EARLY SHORT DAY FLOWERING 4 (ESD4), a SUMO-specific protease, or SIZ1, resulted in early short-day flowering, suggesting a negative role of these genes in flowering (Murtas et al. 2003; Jin et al. 2008). SIZ1 promotes FLOWERING LOCUS C (FLC) expression by repressing FLOWERING LOCUS D (FLD) activity through SUMOylation (Jin et al. 2008). GI promotes the transition from the vegetative stage to flowering in response to a long-day photoperiod (Mizoguchi et al. 2005). In Arabidopsis, a link between flowering time and the cold stress response through GI has been suggested (Cao et al. 2005). However, to date, there has been no evidence that GI is regulated by heat shock. In this study, we showed that GI expression is indeed induced by heat shock, and the GI protein is modified both in vivo and in vitro with SUMO in response to heat shock stress (Figure 6), suggesting that GI may play a role in the response to heat. In addition, it has been reported that heat stress accelerates flowering in Arabidopsis (Balasubramanian et al. 2006). Although speculative, it is tempting to suggest that modification of GI with SUMO may play a role in the induction of flowering in response to high temperatures. Because GI establishes several protein-protein interactions with different proteins (Demarsy and Fankhauser 2009) to regulate their activities, it is possible that modification with SUMO affects the interaction of GI with some of its partners. Interestingly, GI seems to be degraded through the ubiquitin-proteasome system in the dark (David et al. 2006). In this context, it is possible that heat stress induces SUMOylation of GI, preventing its degradation and favoring specific protein-protein interaction to accelerate flowering. Because SUMOylation may act antagonistically to ubiquitination by blocking the ubiquitin attachment sites, SUMO attachment in GI could protect it from degradation. This competition is implicated in regulating gene transcription, chromatin structure, changes in protein–protein interaction, and subcellular localization (Desterro et al. 1998; Hietakangas et al. 2003; Yang et al. 2003; Gill 2004; Lin et al. 2004). One of the best known examples of this competition is the regulation of the Iκ-Bα protein, which undergoes polyubiquitination for 26S-dependent degradation or SUMOylation to prevent its degradation (Ulrich 2005). In plants, ICE1 is also modified by SUMO and ubiquitin, indicating that this protein may be regulated in a similar fashion (Chinnusamy et al. 2007). Another potential example is the lipase/acylhydrolase-GDSL protein. This protein has been identified as being ubiquitinated in Arabidopsis (Maor et al. 2007; Manzano et al. 2008). In this study, we have found that the lipase/acylhydrolase is modified with SUMO (Table 1), suggesting that it may also be regulated by both peptide tags.

The proteomic approach described here has allowed for the discovery of new SUMO targets, as well as the others that were previously identified. The use of this method will open the possibility to analyze proteome changes and to identify targets of PTMs during programmed development or in response to external stimuli. In addition, the use of this method is not restricted to plants, as it could also be used to analyze different types of PTMs in any organism.

Materials and Methods

Chemicals

Urea, thiourea, sodium chloride, iminodiacetic acid, Tris, glycerol, SB3-10 (3-(decyldimethyl-amonio) propanesulfonate inner salt), methanol G, 2-propanol, ammonium hydroxide, phenylmethylsulfonyl fluoride (PMSF), protease inhibitor cocktail, iodoacetamide, dithiothreitol (DTT), hexylene glycol, PIPES, Percoll, Miracloth, sucrose, ammonium bicarbonate, and trypsin were purchased from Sigma (St Louis, MO, USA). Acetonitrile (ACN) and trifluoroacetic acid (TFA) were obtained from J. T. Baker (Deventer, Holland). A MicroBCA Protein Assay Kit was purchased from Pierce (Rockford, IL, USA). N-Octylglucoside was obtained from Melford (Suffolk, UK). 1,1,2-Trichlorodifluoroethane was purchased from Panreac (Barcelona, Spain). PD-10 desalting columns were purchased from GE Healthcare Bio-Science (Uppsala, Sweden). Chromatofocusing (CF) start buffer (SB) and CF eluent buffer (EB) were commercialized by Beckman-Coulter (Fullerton, CA, USA). The anti-phosphothreonine (clone H2) and anti-ubiquitin (clone P4D1) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA), the anti-HA (clone 3F10) was purchased from Roche (Indianapolis, IN, USA), the anti-histidine Clontech, the anti-human-SUMO1 were purchased from Santa Cruz Biotechnology and the anti-Arabidopis-SUMO1 from Abcam (Cambridge, UK). The chemiluminescence substrate and the ZipTipC18 columns were purchased from Millipore (Billerica, MA, USA).

Plant material

We used Arabidopsis thaliana plants, Columbia-0 ecotype and gi-2 (NASC), and gi-2/35S:GI-HA and gi-2/35S:GI-TAP (David et al. 2006). Plants were grown in a controlled chamber at 22 °C with a photoperiod of 16 h light and 8 h dark, on solid MG medium (Murashige and Skoog Basal Medium salts plus 1% sucrose, 10 mmol/L 2-(N-Morpholino)ethanesulfonic acid, pH = 5.7 and 1% of plant Agar (Duchefa, Haarlem, the Netherlands)). To generate transgenic plants overexpressing HA-SUMO (HA3-SUMO1ox), the SUMO1 (At4g26840) cDNA clone U17495 was obtained from The Arabidopsis Biological Resource Center (ABRC) and cloned into the binary vector pBA002, which contains a 35S promoter and three copies of the HA epitope. The SUMO coding region was fused in-frame with the 3xHA epitope at the N-terminus. Transformation of Arabidopsis was performed following the floral dip method (Clough and Bent 1998).

Nuclei and chloroplast isolation and western blot analyses

Arabidopsis nuclei were isolated from 10-d-old control and heat-shocked (30 min at 42 °C) HA3-SUMO1ox plants by the Percoll gradient method (Folta and Kaufman 2000).

To analyze the presence of SUMOylated or ubiquitinated proteins in chloroplasts, we purified these organelles. We used 20-d-old leaves from wild-type or HA3-SUMO1ox plants, because the yield of isolated chloroplasts was higher and of better quality than those from 6-d-old seedlings. The leaves were chopped with a blade in Xpl buffer (3% sorbitol, 50 mmol/L HEPES pH = 7.5, 1 mmol/L MgCl2, 2 mM ethylenediamine tetraacetic acid, 2.5 μg bovine serum albumin (BSA)/mL, 5 mmol/L sodium ascorbate) and incubated for 5 min on ice. Next, the homogenate was filtered through two layers of Miracloth. Intact chloroplasts were isolated as described by Weigel and Glazebrook (2002), and total protein was extracted by adding 200 μL sodium dodecylsulfate (SDS) loading buffer containing 5% of β-ME and boiling for 10 min.

Proteins were separated by SDS polyacrylamide gel electrophoresis (PAGE), and were then transferred to PVDF membranes, which were analyzed by immunoblotting using antibodies against HA (Roche) to detect SUMO.

Protein sample preparations and 2-D LC analysis

Total protein was extracted from Arabidopsis seedlings or from nuclei preparations. Extraction was carried out in general lysis-denaturing buffer (6 mol urea, 2 mol thiourea, 10% glycerol (v/v), 50 mmol/L Tris-HCl pH = 7.8), 2% (w/v) n-octylglucoside), containing 1× plant inhibitor cocktail and 1 mmol/L PMSF. To extract proteins, 4 mL of general lysis-denaturing buffer was added to approximately 3 g of ground plant materials. The solution was vortexed and incubated on ice for 30 min. Afterwards, it was centrifuged at 16 500 g at 4 °C for 15 min, and the supernatant was transferred to a fresh tube. To further extract non-soluble proteins, the precipitated material was resuspended in 2 mL of membrane lysis-denaturing buffer (general lysis-denaturing buffer plus 2.5% (w/v) SB3-10, 1× plant inhibitors cocktail and 1 mmol/L PMSF). The mixture was vortexed, incubated on ice for 15 min, and centrifuged again at 16 500 g at 4 °C for 15 min. The supernatant was combined with the first one and passed through a 0.2 μM sterile filter. The extracts were desalted and equilibrated with CF start buffer in a PD-10 desalting column following the manufacturer's instructions. Protein quantification was performed using the MicroBCA Protein Assay Kit with different concentrations of BSA protein diluted in CF start buffer as standards.

Protein extracts (2.5 mg of protein) from Arabidopsis wild-type or HA3-SUMO1ox plants were subjected to 2-D LC analysis using a ProteomeLab PF2D instrument (Beckman-Coulter) and the procedure recommended by the manufacturer. The first-dimension separation was carried out by CF on a High Performance ChromatoFocusing (HPCF) 1-D column (250 mm × 2.1 mm internal diameter, 300 armstrong pore size). The column was equilibrated at pH = 8.5 with CF start buffer for 250 min at 0.2 mL/min. The pH gradient began after 20 min of sample injection when the CF eluent buffer at pH = 4.0 moved through the column, gradually decreasing the pH from 8.5 to 4.0. Proteins were eluted according to their isoelectric points (pI) and, in the final step, the most acidic ones were eluted with 1 mol NaCl, 0.2% n-octylglucoside. All fractions were collected in 96 well plates using an automated collector.

All the different pH fractions collected from the first dimension were resolved on a reverse phase C18 column (HPRP column: 4.66 mm × 3.3 mm, 1.5 μm particle size). Of each fraction, 200 μL was run through the column in solvent A (0.1% v/v TFA in water), and the proteins were then eluted with a linear gradient (0–100%) of solvent B (0.08% v/v TFA in ACN) for 35 min. Separation was performed at 0.75 mL/min, and the temperature column was maintained at 50 °C. Eluted proteins were monitored by ultraviolet light at 214 nm of absorbance. The different fractions of the first dimension were collected in 12 plates (each 96 well) using an automated collector.

All CF profiles were elaborated and compared using 32 Karat V1.01 software (Beckman-Coulter). Quantitative analysis of the protein peak areas and heights were performed using the Mapping tools software V1.0 (Beckman-Coulter).

Immunodetection and immunoprecipitation assays

The complete proteome of Arabidopsis was collected in 96 microwell plates as described above, and then transferred onto PVDF membranes (Millipore) by dot blot. Membranes were blocked with 5% BSA in phosphate-buffered saline (PBS) buffer for 1 h at room temperature, and then incubated for 2 h with monoclonal antibodies against phosphothreonine 1:500, against HA at 1:5000 to detect HA-SUMO, or against ubiquitin at 1:1.000 dilution in PBS 0.1% Tween-20. Afterwards, these membranes were washed four times in PBS 0.1% Tween-20, and were then incubated with antimouse immunoglobulin (Ig)G1 peroxidase-conjugated antibody or antirat (1:50.000 dilution in PBS 0.1% Tween-20). Detection of IgG-binding components was carried out by enhanced chemiluminescence (Millipore).

Arabidopsis gi-2/35S:GI-HA and col-0 were grown at 22 °C for 7 d, and then transferred to a dark oven at 42 °C or 22 °C for 30 min. Afterwards, total protein was extracted from ground plant material in a TAP buffer (50 mmol/L Tris-HCl, 150 mmol/L NaCl, 0.5% NP-40, 5% glycerol, 1 mmol/L PMSF, 1× plant protease inhibitors (Sigma) and 100 μmol/L MG132). Extracts were sonicated and incubated in ice for 30 min. Afterwards, they were centrifuged at 16,500 g at 4 °C for 15 min, and the supernatant was transferred to fresh tubes. For immunoprecipitation assays, approximately 1 mg of total protein in 2 mL was mixed with 0.5 μg of high-affinity anti-HA (Roche) and 150 μL of protein G plus (Santa Cruz Biotechnology). The mix was incubated overnight at 4 °C and then washed with TAP buffer four times at 10 min each. The precipitated proteins were released from the beads by heating them at 100 °C in Laemmli loading buffer for 5 min, and were resolved in a 9% SDS-PAGE and then transferred to PVDF membranes. They were analyzed by immunoblotting using anti-AtSUMO1 (Abcam, Cambridge, UK) and anti-HA to check the immunoprecipitation efficiency.

MS analysis

To identify Arabidopsis proteins modified by SUMOylation, ubiquitination, or phosphorylation, immunoreactive fractions were analyzed by MS. Eluted fractions were evaporated to a final volume of 10 μL. Protein digestions were carried out by incubating the samples in 50 mmol/L NH4HCO3 and 10 mmol/L DTT at 60 °C for 1 h. The alkylation of the reduced sulfhydryl groups was performed by adding 55 mmol/L iodoacetamide at 25 °C for 30 min in the dark. Proteins were digested by adding 1.5 μL of trypsin (125 μg/mL) and incubating at 37 °C overnight. The reaction was stopped with 1% of formic acid. Tryptic peptides were desalted and concentrated with ZipTipC18 columns according to the manufacturer's recommendation. Peptides were eluted in 0.1% TFA, 50% ACN for matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF)-MS analysis, and with 1% formic acid, 50% methanol for electrospray MS analysis. To increase salt removal, samples were washed with 3–5 cycles of 0.1% TFA as wash solution. The solution was spotted directly onto a MALDI target and analyzed by MALDI-TOF/TOF off-line coupled LC/MALDI-MS/MS. MS analyses were performed automatically with a 4700 Analyzer MALDI-TOF/TOF instrument (Applied Biosystems, Carlsbad, CA, USA). First, MS spectra of all spotted fractions were acquired in the positive reflector mode for peak selection (S/N>20, excluded precursor with 200 resolution), and further MS/MS spectra acquisition was done using the Collision Induced Dissociation of selected peaks. The search of filtered peptides was performed in batch mode using GPS Explorer V 3.5.0 software with a licensed version of MASCOT, in the Swiss-Prot Database. The MASCOT search parameters were: (i) species, Arabidopsis thaliana; (ii) allowed number of missed cleavages (only for trypsin digestion), 1; (iii) considered modifications, Cys as carboamidomethyl derivate and Met as oxidized methionine; (iv) peptide tolerance, ±150 p.p.m.; (v) MS/MS tolerance, ±0.4 Da; and (vi) peptide charge, +1 (Bairoch and Apweiler, 2000).

In vivo SUMOylation in an Escherichia coli system

To validate the SUMOylated proteins that have been identified by MS analyses, we used the Escherichia coli system that contains all components of the pathway (Mencía and de Lorenzo 2004). The cDNA coding for PBP1 (At3g16420) was cloned into a pET28b plasmid which fused the proteins to a H6x histidine tag. The cDNAs coding for GI (At1g22770) were cloned in a pET28b plasmid in two halves: the N-terminal region, including from the ATG to the nucleotide 1203 (44 kDa), and the C-terminal region, including from nucleotide 2,250 to the STOP codon in position 3,522 (45.5 kDa). In all cases, the cloning was done by using restrictions enzymes and verified by sequencing. We transformed these constructions and the control pET28 along with the SUMO pathway enzymes (pBADE12 and pHRSUMO plasmids) into the expressing BL21(DE3) Rosetta cells. Protein expression was induced by adding Isopropyl-β-d1-thiogalactopyranoside (IPTG) at 0.5 mmol/L to 50 mL of the cultures for 2 h at 37 °C and 15 h at 30 °C. To analyze whether the target proteins were modified with SUMO, first we extracted the total protein from the bacteria and carried out histidine purification. Because SUMO E1 and E2 are also fused to 6x histidine tag, we purified recombinant proteins using nickel beads. The purified proteins were analyzed by immunoblotting with anti-SUMO antibodies (Santa Cruz Biotechnology) at 1:2000 dilution in PBS 0.1% Tween-20. Afterwards, these membranes were washed four times in PBS 0.1% Tween-20, and were then incubated with antirabbit IgG peroxidase-conjugated antibody (1:25000 dilution in PBS 0.1% Tween-20). Detection of IgG-binding components was carried out by enhanced chemiluminescence (Millipore).

(Co-Editor: Qi Xie)

Acknowledgements

The authors are indebted to V. Fernández and S. Navarro for technical assistance. We would like to thank Dr J. Putterill and NASC for providing the gi-2 and gio-2/35S.GI-HA and 35S:GI-TAP seeds. We thank the INIA for providing financial support to acquire the Proteome Lab PF2D equipment. We also thank the proteomic facility at the UCM-PCM, a member of ProteoRed network, for the proteomics analyses and for the help provided in interpretation of the results. This work was supported by the grants S-GEN-0191-2006 (CAM) and BIO2007-62517 (MEC), CSD-2007-00057, and BIO2011-28184-C02-01 to J. C. P., and S-GEN-0191-2006 (CAM), BIO2007-65284 (MEC) and GEN2006-27787-E (MEC) to J. S. G. L. T. was supported by a postdoctoral contract (Comunidad de Madrid).

Ancillary