Insight into the redox regulation of the phosphoglucan phosphatase SEX4 involved in starch degradation

Authors


G. B. G. Moorhead, BI144 - Biological Sciences Building, University of Calgary, Calgary, Alberta T2N 1N4, Canada
Fax: +1 403 289 9311
Tel: +1 403 220 6238
E-mail: moorhead@ucalgary.ca
Website: http://www.bio.ucalgary.ca/contact/faculty/moorhead.html

Abstract

Starch is the major carbohydrate reserve in plants, and is degraded for growth at night. Starch breakdown requires reversible glucan phosphorylation at the granule surface by novel dikinases and phosphatases. The dual-specificity phosphatase starch excess 4 (SEX4) is required for glucan desphosphorylation; however, regulation of the enzymatic activity of SEX4 is not well understood. We show that SEX4 switches between reduced (active) and oxidized (inactive) states, suggesting that SEX4 is redox-regulated. Although only partial reactivation of SEX4 was achieved using artificial reductants (e.g. dithiothreitol), use of numerous chloroplastic thioredoxins recovered activity completely, suggesting that thioredoxins could reduce SEX4 in vivo. Analysis of peptides from oxidized and reduced SEX4 identified a disulfide linkage between the catalytic cysteine at position 198 (Cys198) and the cysteine at position 130 (Cys130) within the phosphatase domain. The position of these cysteines was structurally analogous to that for known redox-regulated dual-specificity phosphatases, suggesting a common mechanism of reversible oxidation amongst these phosphatases. Mutation of Cys130 renders SEX4 more sensitive to oxidative inactivation and less responsive to reductive reactivation. Together, these results provide the first biochemical evidence for a redox-dependent structural switch that regulates SEX4 activity, which represents the first plant phosphatase known to undergo reversible oxidation via disulfide bond formation like its mammalian counterparts.

Abbreviations
BAM1/BAM3

β-amylases 1 and 3

DSP

dual-specificity phosphatase

DTT

dithiothreitol

LSF1/LSF2

like-SEX4 1 and 2

pNPP

p-nitrophenyl phosphate

PTP

protein tyrosine phosphatase

SEX4

starch excess 4

TCEP

Tris(2-carboxyethyl)phosphine

TRX

thioredoxin

Introduction

Reversible glucan phosphorylation at the starch granule surface is the primary mechanism that governs leaf starch breakdown in the chloroplasts of plants [1–3]. This mechanism is initiated by two kinases: α-glucan, water dikinase [4,5] and phosphoglucan, water dikinase [6]. The α-glucan, water dikinase phosphorylates the C6 position of starch glycosyl residues, providing pre-phosphorylation priming for the phosphoglucan, water dikinase to phosphorylate the C3 position [7,8]. These phosphates disrupt the helical, semi-crystalline structure of amylopectin, triggering unpacking of starch for hydrolysis and release of sugars (glucose and maltose) for growth at night [9,10]. The hydrolytic enzymes isoamylase 3 and β-amylases 3 and 1 (BAM3 and BAM1) act on the starch granule to release maltose and linear oligosaccharides [11–13]. Simultaneously, the phosphoglucan phosphatase starch excess 4 (SEX4) dephosphorylates both the C6 and C3 positions of glycosyl residues to remove phosphates that impede movement of the BAMs along the glucan polymer [1,2]. Together, these enzymes operate in a coordinated manner at the granule surface to degrade starch throughout the night.

Dephosphorylation of starch by SEX4 (At3g52180; EC3.1.3.48) is essential for proper starch breakdown. Loss-of-function plants contain three times more starch than wild-type plants by the end of night [14,15]. Moreover, sex4 mutants accumulate phospho-oligosaccharides [1], suggesting that these plants have a decreased pool of hydrolyzable starch that results in stunted growth and delayed flowering [14,16]. Outside of the kingdom Plantae, an analogous phosphatase named laforin controls the phosphorylation state of glycogen [17]. Mutations in the laforin gene (EPM2A) cause the fatal neurodegenerative disorder Lafora disease, in which patients accumulate hyper-phosphorylated glycogen inclusion bodies [17,18]. If laforin is chloroplast-localized in sex4 plants then normal starch levels are restored, indicating that these phosphatases are functional equivalents [19]. The presence of SEX4 and laforin among species with differing evolutionary lineages argues that reversible phosphorylation of glucans spans multiple kingdoms to regulate carbohydrate metabolism [20].

SEX4 belongs to the dual-specificity phosphatase (DSP) family of the protein tyrosine phosphatase (PTP) super-family [21]. Together with two other DSPs involved in starch degradation – like SEX4 1 (LSF1) [22] and C3 position-specific like SEX4 2 (LSF2) [23] – they constitute the phosphoglucan phosphatase clade of DSPs in Arabidopsis. Furthermore, SEX4 and LSF1 are the only plant phosphatases that contain a carbohydrate-binding module [24]. This domain has numerous interactions with the phosphatase domain in SEX4, forming a wide 21 Å active site to accommodate the long phosphoglucan polymer [25]. This structural architecture may also regulate SEX4 activity by occluding the active site in the absence of glucan binding [26].

PTPs possess the signature motif HCX5R at the active site, with the invariant catalytic cysteine forming a phosphorylcysteine intermediate [27,28]. This cysteine exists as a thiolate anion even at neutral pH (pKa 4.7–5.7) [29], making PTPs susceptible to oxidation [30]. Initial oxidation rapidly converts the thiolate (-S) to sulphenic acid (-SOH), rendering the protein inactive [31]. PTPs protect this cysteine through reversible oxidation by forming a disulfide bond with a nearby cysteine or a sulphenyl amide bond with the protein backbone, which is then reduced back to the thiolate ion [32,33]. These structural rearrangements prevent the cysteine from becoming irreversibly oxidized to sulphinic acid (-SO2H) and sulphonic acid (-SO3H) [34]. Reversible oxidation of PTPs also serves as a regulatory mechanism for activity. For instance, certain mammalian PTPs are inactivated by oxidation during hormonal induction to perpetuate phosphorylation-dependent signal transduction by protein kinase receptors [32]. The activity of phosphatase and tensin homolog (PTEN) is also modulated under redox conditions to adjust intracellular levels of the 3′-phosphorylated phosphoinositides to regulate cell growth and survival [35]. To date, the AtPTP1 from Arabidopsis is the only plant PTP that has been shown to undergo reversible oxidation [36,37].

The evidence for redox regulation of PTPs/DSPs suggests that SEX4 could also participate in reversible oxidation to control its phosphatase activity. Sokolov et al. [38] showed that SEX4 is inactivated by oxidation; however, the mechanism of reversible oxidation were not explored. We show that SEX4 phosphatase activity is modulated by switching between reduced (active) and oxidized (inactive) states, much like other PTPs. Oxidation of SEX4 promotes formation of a disulfide linkage between the catalytic cysteine at position 198 (Cys198) and the cysteine at position 130 (Cys130) within the phosphatase domain, and mutation of the latter residue renders SEX4 redox-impaired. Not only does this disulfide bridge protect Cys198 from irreversible oxidation, these results provide the first biochemical evidence for a redox-dependent structural switch that regulates SEX4 activity.

Results

Reversible oxidation of SEX4

As SEX4 activity is inhibited by oxidation [38], we investigated whether SEX4 could undergo reversible oxidation from an inactive, oxidized state (SEX4ox) to an active, reduced state (SEX4rd). We first confirmed that SEX4 can be inactivated by oxidation by exposing SEX4rd to increasing concentrations of hydrogen peroxide (H2O2) and measuring phosphatase activity via p-nitrophenyl phosphate (pNPP) hydrolysis. Activity of SEX4 was abolished at 50 μm H2O2 (Fig. 1A). SEX4ox was then incubated with the chemical reductants dithiothreitol (DTT) and Tris(2-carboxyethyl)phosphine (TCEP) to determine whether phosphatase activity could be restored. Both reactivated SEX4ox to almost 30% of the activity of SEX4rd, with TCEP recovering the activity of SEX4 more readily at lower concentrations than DTT (Fig. 1B). Similar results were obtained for SEX4-mediated release of inorganic phosphate from solubilized potato amylopectin using the malachite green assay. SEX4ox had no detectable activity and DTT recovered SEX4ox activity to roughly 30% of that of SEX4rd (Fig. 1C). These results indicate the SEX4 can switch between active and inactive forms by reversible oxidation in vitro.

Figure 1.

 Modulation of SEX4 phosphatase activity. (A) Inactivation. SEX4 was exposed to H2O2 for 10 min prior to pNPP phosphatase assay. Activity is expressed relative to SEX4 with no oxidant. (B) Reactivation. SEX4 was oxidized with 50 μm H2O2 for 10 min, buffer-exchanged, and then exposed to increasing concentrations of either DTT or TCEP for 10 min. Reactivation is expressed relative to non-oxidized SEX4. (C) Amylopectin dephosphorylation. Reduced or oxidized SEX4 was incubated with solubilized potato amylopectin for 10 min, the reaction was stopped with N-ethylmaleimide, and malachite green reagent was added. Data are means ± SE (= 3).

To explore possible in vivo reductants for SEX4ox, we tested thioredoxins (TRXs) that reside in the chloroplast, where starch degradation occurs. TRXs function in thiol–disulfide exchange via the vicinal cysteines in the conserved WCGPC motif [39]. We tested six of the nine chloroplastic TRXs (f1, m1, m3, m4, x and y1) encoded in the Arabidopsis thaliana genome [40]. TRX f2, m2 and y2 were omitted as their catalytic domains have very high percentage identity to their closest isoforms f1, m1 and y1 (90%, 81% and 82%, repectively), and equivalent activities towards previously studied protein targets [41,42]. The efficiency of TRXs in restoration of SEX4 activity was calculated from rates of pNPP hydrolysis over 1 h using mixtures of purified SEX4ox and TRX (Fig. S1). The TRX f1 reactivated SEX4ox the most effectively (293 ± 33 pmol p-nitrophenol·min−1), and the TRX m3 reactivated SEX4ox the least effectively (63.9 ± 6.9 pmol p-nitrophenol·min−1), an effect equivalent to that of 10 mm DTT (71.7 ± 11.1 pmol p-nitrophenol·min−1) (Table 1). The order of TRX efficiency for SEX4 reactivation was f1 > m4, y1, m1 > >>> m3. TRX-containing samples contained 0.2 mm DTT to recycle reduced TRXs, but this concentration of DTT alone did not reactivate SEX4ox (Fig. S1B), and no activity was detected for oxidized SEX4. Additional experiments measured the recovery of phosphatase activity of SEX4ox compared to SEX4rd after SEX4ox was incubated with TRXs for 20 min. Once again, each TRX showed specific efficiency for reactivating SEX4ox, and only TRX f1 completely restored SEX4 activity (Fig. 2). Moreover, all TRXs restored SEX4 activity better than the chemical reductant DTT, suggesting that TRXs may reduce SEX4 in vivo.

Table 1. Reactivation of SEX4 by Arabidopsis chloroplastic TRXs and DTT. Activity was derived from the slope of the linear region of each reactivation curve (see Fig. S1). Data are means ± SE (= 3).
ReductantActivity of reactivated SEX4 (pmol p-nitrophenol·min−1)
TRX f1293 ± 33
TRX m1217 ± 40
TRX m363.9 ± 6.9
TRX m4240 ± 38
TRX x175 ± 23
TRX y1228 ± 42
DTT71.7 ± 11.1
Figure 2.

Recovery of SEX4 activity using chloroplastic TRXs. SEX4 was oxidized with 50 μm H2O2 for 10 min, buffer-exchanged, and then exposed to either 10 μm TRX (as indicated) or 10 mm DTT for 20 min. Recovery is expressed relative to reduced SEX4. Data are means ± SE (= 4).

Structural basis for SEX4ox inactivation

To determine whether SEX4ox forms a disulfide bond during reversible oxidation, we compared profiles of pepsin-digested SEX4ox and SEX4rd proteins using LC-MS/MS (Fig. 3A). We identified 67 peptides that covered all six cysteines, and measured the mass spectral intensities of all cysteine-containing peptides for SEX4ox and SEX4rd. Several peptides with reduced cysteines showed depleted SEX4ox [see spectra for CLQQDPDLRY(130–139) and YKAVKRNGGVTYVHC(184–198) in Fig. 3B]. Of the six cysteines within SEX4, only peptides containing Cys130 or the catalytic residue Cys198 showed a greater than twofold change in intensity between reduced and oxidized states (Table 2). From this analysis, we also identified a peak corresponding to the disulfide-linked peptides CLQQDPDL(130–137) and VHCTAGMGRAPAVAL(196–210) from SEX4ox that showed higher intensity than for the SEX4rd sample (Fig. 3B). This confirms the existence of an oxidation-dependent disulfide bond between Cys130 and Cys198 in the phosphatase domain of SEX4. Both of these residues are conserved in all SEX4 orthologs, including several algal species (Fig. S2). Cys130 is located on the β3–β4 loop on the edge of the active site pocket within SEX4 [25], and is structurally analogous to the corresponding cysteine position of other redox-regulated DSPs [kinase-associated phosphatase 1 (KAP1) and the PTEN] (Fig. 4). The distance between the cysteine pair of SEX4 is longer (8.1 Å; based on Cβ–Cβ distance) than in KAP1 (5.8 Å) and PTEN (5.7 Å), but within the range for all known redox-regulated DSPs (5.7–10 Å) [43]. The greater distance between cysteines in SEX4rd may be attributed to the wider cleft of the active site needed to accommodate the phosphoglucan substrate.

Figure 3.

 Mass spectrometric detection of disulfide-linked cysteines in oxidized SEX4. (A) Schematic for LC-MS/MS workflow. SEX4 was incubated with either 50 μg·mL−1 spinach TRX m, 10 mm DTT or 0.5 mm H2O2 for 10 min, digested with pepsin, and analyzed by LC-MS/MS. Peptides that form disulfide linkages (shown in red) appear as a single peak upon oxidation. (B) Identification of Cys130–Cys198 disulfide linkage. Representative spectra of reduced peptides (using TRX/DTT in black/blue) and oxidized peptides (H2O2 in red) from SEX4 containing Cys130 and Cys198. Peaks that differ between redox states are indicated by asterisks (see Table 2 for complete list of identified peptides). m/z, mass/charge ratio.

Table 2. Summary of cysteine-containing peptides identified by LC-MS/MS. Ratios were calculated by dividing the normalized peak intensity for TRX-reduced samples (Rd) by that for H2O2-oxidized (Ox) samples for each peptide. Data for spectra in Fig. 3B are highlighted in bold.
Mass (Da)ChargeSequence of SEX4Cysteine residueRatio (Rd/Ox)
  1. a Involved in disulfide linkage.

797.43SCLQTPEDVDKLRKIGVKTIF(109–129)1101.4
931.41 CLQQDPDL(130–137)1302.2a
1224.0 1 CLQQDPDLEY(130–139) 130 4.3a
812.92YAKKYSDIQHIRC(150–162)1621.2
847.9 2 YKAVKRNGGVTYVHC(184–198) 198 4.0a
752.42YVHCTAGMGRAPAVA(195–209)1985.0a
808.92YVHCTAGMGRAPAVAL(195–210)1983.0a
727.42VHCTAGMGRAPAVAL(196–210)1985.7a
727.41SKRSCF(230–235)2341.4
681.31LETCSE(371–376)3741.6
Figure 4.

 Comparison of disulfide cysteines in SEX4 with other redox-regulated DSPs. Catalytic pocket of reduced proteins for human kinase-associated phosphatase 1 (KAP1; Protein DataBank ID 1FPZ) [59], human phosphatase and tensin homolog (PTEN; Protein DataBank ID 1D5R) [60] and Arabidopsis SEX4 (Protein DataBank ID 3NME) [25]. Structures were aligned together using the HCX5R motif sequence and exported using MacPyMOL. Proteins in which the catalytic cysteine (Cys198 in SEX4) was substituted for serine are indicated in curly brackets.

We next mutated Cys130 to serine (C130S) to assess the importance of Cys130 during oxidation of SEX4. Theoretically, this mutation should disrupt disulfide bond formation and promote the irreversible oxidation of Cys198 over time, making SEX4 permanently inactive. Even though SEX4 and C130S proteins had equivalent activity at non-oxidized conditions ([H2O2] = 0 μm), C130S was nearly three times more sensitive to H2O2 inhibition than SEX4, with IC50 values of 13.4 and 37.4 μm, respectively (Fig. 5A). Interestingly, oxidized C130S protein was reactivated by the addition of DTT; however, the degree of recovery was less than observed with SEX4 at DTT concentrations < 10 mm (Fig. 5B). These results suggest that the C130S mutation alters the reversible oxidation mechanism by making SEX4 more susceptible to oxidation and less amenable to reactivation.

Figure 5.

 Assessment of the C130S mutation in SEX4. (A) Inactivation. Proteins were reduced with 10 mm DTT, buffer-exchanged, and then exposed to H2O2 for 30 min prior to pNPP phosphatase assay. Activity is expressed relative to SEX4 with no oxidant. (B) Reactivation. Proteins were oxidized with 50 μm H2O2, buffer-exchanged, and then exposed to DTT for 10 min. Activity is expressed relative to reduced SEX4. Data are means ± SE (= 3).

Discussion

Redox regulation of SEX4

The function of SEX4 as a phosphoglucan phosphatase is an essential enzyme in the degradation of starch in plants [1]. How this activity is regulated in the chloroplast stroma is not well understood. Transcriptional expression of SEX4 fluctuates diurnally, with maximal expression at the end of the day, and correlates with that of other enzymes involved in starch breakdown [24]. Protein levels of SEX4 remain constant throughout the day and night (O. Kötting, Department of Biology, Eidgenössische Technische Hochschule (ETH) Zurich, Zurich, Switzerland). There is no evidence for phosphorylation of SEX4 in the Arabidopsis thaliana phosphorylation site database PhosPhAt (http://phosphat.mpimp-golm.mpg.de/) [44,45], nor any documented protein interactors, allosteric inhibitors or other post-translational modifications. Therefore, an alternative method may regulate SEX4 activity. We describe a plausible mechanism of redox regulation to modulate SEX4 between active and inactive forms. Oxidation of SEX4 results in formation of a disulfide bond between Cys130 and Cys198 (Fig. 3B), rendering the protein inactive as the catalytic Cys198 is unable to form a thiolate anion. Upon reduction, this structural rearrangement is reversed, making SEX4 active again. This mechanism acts as an intramolecular redox switch within the phosphatase domain, and represents the first evidence shown to control the phosphatase activity of SEX4. Moreover, this represents the first plant DSP whose activity is controlled by reversible oxidation like mammalian DSPs, and the result strongly suggests that redox switch regulation is intrinsic to these phosphatases in all eukaryotes.

Whether a similar mechanism exists for the other phosphoglucan phosphatases LSF1, LSF2 and laforin is not known. Conservation of Cys130 was not maintained when these proteins were aligned, even though several cysteines do exist within the DSP domain for each protein (data not shown). Deducing the correct cysteine partner for a disulfide linkage with the catalytic cysteine using protein sequence alignments may be inadequate since the protein fold largely dictates the orientation and distance of a suitable cysteine [43]. This is exemplified by comparing the cysteines involved in disulfide bond formation for SEX4, PTEN and KAP1, which show almost identical orientation of these cysteines despite differences of 20–30 residues in sequence length between them (Fig. 4). Therefore, more structural data are needed before a generalized mechanism can be proposed for all phosphoglucan phosphatases.

Our aim of creating an irreversibly oxidized variant of SEX4 by mutating Cys130 to serine (C130S) was ineffective. Although the C130S protein was more sensitive to oxidation compared to SEX4, C130S was still reactivated when reduced by DTT (Fig. 5). As PTPs/DSPs harbor the highly reactive catalytic cysteine within the active site pocket [28], the C130S mutation probably forced the protein to form a less favourable disulfide bond between the catalytic Cys198 and an alternative cysteine. Of the remaining four cysteines in SEX4, Cys110 and Cys234 are the next closest to Cys198 and could serve as alternative disulfide bond partners (Fig. S2) [25]. Therefore, the structural fold of SEX4 may harbor additional cysteines to protect Cys198 from irreversible oxidation if Cys130 was mutated, functioning as a fail-safe system to regulate phosphatase activity. Biochemical studies of mitogen-activated protein kinase phosphatase 3 support the suggestion that more than two cysteines can be involved in reactivation of an oxidized phosphatase [46]. It was postulated that a cysteine located distal to the active site transfers the initial disulfide link (Cys130–Cys198 equivalent) to the exterior surface of the protein to facilitate reduction by a TRX. Therefore, SEX4 may use cysteines additional to Cys130 to reduce SEX4ox.

TRX reactivation of enzymes involved in starch metabolism

Several other proteins involved in starch metabolism in addition to SEX4 are also redox-regulated [47]. For instance, the tetramer ADP-glucose pyrophosphorylase catalyzes the first step in starch synthesis, and its substrate affinity is attenuated by formation of disulfide bonds between subunits when oxidized [48,49]. The activity of α-glucan,water dikinase is also inhibited during oxidation, and can be reactivated by chloroplastic Spinacia oleracea spinach TRXs f and m [50]. The endoamylase BAM1 requires a reduced cysteine for catalysis, and forms a disulfide bond with a neighboring cysteine [51]. A subsequent study showed that BAM1 is reactivated by numerous chloroplastic TRXs from Arabidopsis [52]. Interestingly, the order of specificity of TRXs for BAM1 is similar to our data for SEX4 reactivation (Table 2, and see Results). This suggests that the relative abundance of each TRX isoform could control the activation of specific targets, such as those involved in starch degradation. Alternatively, the spatial distribution of these isoforms within the chloroplast may differ between the starch granule surface and the stroma. Therefore, insight into the protein abundance and sub-organelle localization of each TRX isoform within the chloroplast is required. At present, no published data exist for protein levels of TRXs throughout the diurnal cycle. Development of antibodies for each TRX isoform would enable more specific detection of each protein in immunoblotting (protein abundance) or immunofluorescence (protein localization) assays. Our laboratory successfully employed isoform-specific antibodies to detect the three mammalian protein phosphatase 1 isoforms and determined which isoform forms a complex with taperin in the nucleus and cytoplasm [53].

Reactivation of SEX4 by numerous TRXs indicates these proteins do associate in vitro. Reaction mixtures containing SEX4ox alone were inactive in all experiments (Fig. S1B). However, despite our best efforts, we could not demonstrate a direct interaction by co-immunoprecipitation from leaf tissue (data not shown). Perhaps the SEX4–TRX interaction is too transient in vivo or unstable during purification to permit isolation of this complex. Other studies have used modified TRX variants in which one of the cysteines within the WCGPC active site motif is mutated to a serine to isolate mixed disulfide-linked complexes [54]. None of these proteomic screens identified SEX4 as a potential TRX-binding partner, possibly because the amount of SEX4 that interacted with TRX was below detection limits. Use of the C130S mutant to identify specific TRX isoforms that associate with SEX4 could possibly circumvent this limitation by enriching for SEX4-specific interactions. Moreover, determining the impact of the C130S mutation of SEX4 in planta will provide further understanding of the role of redox regulation on starch degradation.

Experimental procedures

Protein cloning and expression

SEX4 (At3g52180) lacking the N-terminal chloroplast transit peptide (residues 54–379) [16,24] was cloned into pET101/D/TOPO using directional TOPO® cloning (Invitrogen, Burlington, Ontario, Canada). The SEX4 sequence was terminated with a stop codon, and proteins were purified as described below. The point mutation C130S of SEX4 was generated using a QuikChange® II site-directed mutagenesis kit (Stratagene, Mississauga, Ontario, Canada) according to the manufacturer’s instructions. Recombinant proteins were expressed in Escherichia coli strain BL21 Star™ (DE3) (Invitrogen) and grown to an attenuance at 600 nm of approximately 0.4 at 37 °C before induction for 4 h at 22 °C using 0.5 mm isopropyl thio-β-d-galactoside (BioShop, Burlington, Ontario, Canada). Cells were harvested by centrifugation at 4000 g for 10 min at 4 °C, resuspended in lysis buffer (50 mm Tris/HCl pH 7.4, 1 mm EDTA, 1 mm EGTA, 5% glycerol, 0.5 mm phenylmethanesulfonyl fluoride, 0.5 mm benzamidine), disrupted using a cell press (1000 psi; three passes) and clarified by centrifugation at 100 000 g for 35 min at 4 °C. Before loading onto a Q-Sepharose™ anion-exchange column (GE Healthcare, Mississauga, Ontario, Canada), 2 mm DTT were added to the supernatant. Proteins were eluted using 0–200 mm NaCl gradient over 15 column volumes at 3 mL·min−1 using an ÄKTA FPLC system (GE Healthcare). Fractions (5 mL) were resolved by SDS/PAGE, and those containing SEX4 were pooled, concentrated (200 μL) and separated using a Superdex™ 75 size-exclusion pre-packed column (GE Healthcare) at 1 mL·min−1 with 25 mm Tris/HCl pH 7.4, 200 mm NaCl, 1 mm EDTA, 1 mm EGTA, 1 mm DTT and 5% glycerol. Fractions containing pure SEX4 (1 mL) were pooled, concentrated, flash-frozen and stored at −80 °C.

Phosphatase activity assays

The phosphatase assays involved measurement of in vitro enzyme activity for hydrolysis of pNPP and dephosphorylation of solubilized potato amylopectin. The kinetic parameters of SEX4 and C130S for pNPP are shown in Fig. S3. For inactivation assays, purified SEX4 or C130S was incubated with 10 mm DTT for 10 min, and buffer-exchanged into pNPP buffer (100 mm Hepes, pH 7.0, 150 mm NaCl, 1 mm EDTA) using an Amicon Ultra centrifugal filter (Millipore, Billerica, MA). Reduced SEX4 (100 ng) and C130S (200 ng) were exposed to various concentrations of H2O2 for 10 min. Samples were then incubated with 4 mmpNPP (BioShop) for 30 min at 30 °C, the reaction was stopped using 1 volume of 2 m NaOH, and absorbance was measured at 405 nm. Data were fitted to calculate the half-maximal inhibitory concentration (IC50) using Sigmaplot version 11 (Systat Software Inc., San Jose, CA, USA). Reactivation assays used oxidized SEX4 or C130S prepared by 10 min exposure to 50 μm H2O2. Following buffer exchange into pNPP buffer, oxidized SEX4 or C130S was incubated with various concentrations of DTT or TCEP (Gold Biotechnology, St Louis, MO, USA) for 10 min. Reactions were then mixed with 5 mmpNPP for 20 min at 30 °C, neutralized and measured for absorbance. Recovery of activity was calculated relative to non-oxidized SEX4. Malachite green assays were performed using a modification of the protocol described by Worby et al. [55] to reduce precipitation of the phosphomolybdate–malachite green complex. The 20 μL samples contained 50 ng SEX4rd/SEX4ox and approximately 45 μg solubilized potato amylopectin (Fluka, Oakville, Ontario, Canada) in buffer (50 mm Hepes, pH 7.0, 100 mm NaCl) with or without 10 mm DTT. Reactions were incubated for 10 min at 30 °C, stopped with 1 volume of 0.1 mN-ethylmaleimide, and visualized using 4 volumes of malachite green reagent [1.25% w/v (NH4)6Mo7O24 (Sigma-Aldrich, Oakville, Ontario, Canada) and 0.15% w/v malachite green (Sigma-Aldrich) in 1 m HCl] for 5 min. Absorbance was measured at 620 nm, and the amount of inorganic phosphate released was determined from a standard curve of K2HPO4.

The specificity of TRXs for SEX4 was determined by measuring the activity of SEX4ox reactivated by various TRXs. Recombinant isoforms of the Arabidopsis chloroplast TRXs f1, m1, m3, m4, x and y1 were expressed and purified as described previously [41,42]. Purified SEX4 was oxidized using 50 μm H2O2 for 10 min, and buffer-exchanged into pNPP buffer. The 200 μL reactions contained 100 ng SEX4ox, 5 mmpNPP, 10 mm DTT or 10 μm TRX with 0.2 mm DTT (for recycling the pool of reduced TRX). Controls lacking TRX or TRX and DTT were used to show that SEX4ox remained inactive. Sample absorbance was measured at 410 nm every 45 s for 1 h at 30 °C using a GENios Microplate Reader (TECAN, San Jose, CA, USA). Rates were derived from the slope of the linear region of each reactivation curve as shown in Fig. S1. In addition, recovery of SEX4 activity was quantified by incubating SEX4ox (100 ng) with each TRX or with DTT for 20 min at 30 °C. Measurements were relative to the SEX4rd control.

Analysis of disulfide linkage by mass spectrometry

Purified SEX4 was dialyzed into 1× NaCl/Pi and 0.1 mm EDTA, and pre-treated with 10 mm DTT or 50 μg·mL−1 spinach TRX m for SEX4rd and 0.5 mm H2O2 for 10 min for SEX4ox. Recombinant spinach TRX m was expressed and purified as described previously [56]. SEX4 was digested for 5 min using an immobilized pepsin slurry (Pierce, Rockford, IL) in 200 mm glycine hydrochloride (pH 2.3). SEX4 peptides were separated from the pepsin beads, and 15 μL of this supernatant was injected onto a microbore C18 reversed-phase chromatography column (assembled in-house, 7 cm height × 200 μm internal diameter) for analysis by LC-MS. Samples were loaded onto the column at a rate of 4 μL·min−1, in 3% acetonitrile, 0.02% formic acid and 0.03% trifluoroacetic acid. Peptides were then separated using a linear gradient of 5–90% acetonitrile over 15 min. All peptides were sequenced using a recursive independent data acquisition approach with a QStar Pulsar i quadrupole time-of-flight MS (MDS Sciex, Toronto, Ontario, Canada). Data were collected using Analyst qs version 1.1 (Applied Biosystems, Streetsville, Ontario, Canada). The resulting MS/MS spectra were searched against the SEX4 sequence using mascot version 2.1 [57] with conventional identification criteria (i.e. mass tolerance ± 0.5 units for MS and mass tolerance 0.8 units for MS/MS). This identified 67 peptides with 91% sequence coverage of SEX4, including all six cysteines. All matches were manually verified. To assess the presence of disulfides in SEX4, the intensity of each cysteine-containing peptide was measured, and normalized to that of its nearest neighboring non-cysteine-containing peptide. The ratio of normalized intensity was then calculated by dividing the results for the reduced sample by those for the oxidized sample.

Protein structure

All protein structure data files were acquired from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (http://www.pdb.org/pdb/) [58]. Structures were viewed, aligned using the HCX5R catalytic motif sequence, analyzed (distance measurement), and exported using Macpymol version 1.3r1 (http://www.pymol.org/).

Acknowledgements

We wish to thank members of the Moorhead laboratory for their critical review of the manuscript and fruitful discussions. We would like to acknowledge some preliminary experiments by Nick Bandy. We greatly appreciate access to equipment for microplate experiments by Kenneth Ng (Department of Biological Sciences, Univeristy of Calgary). M.A.G. thanks Samuel C. Zeeman and Oliver Kötting (Department of Biology, ETH Zurich) for advice and support while working at ETH Zürich. The research was supported by the National Sciences and Engineering Research Council of Canada (D.M.S. and G.B.G.M.), Alberta Innovates: Technology Futures (D.M.S.), and the Canadian Foundation for Innovation (D.C.S.).

Ancillary