Structural basis for the glucan phosphataseactivity of Starch Excess4Craig W. Vander Kooia,1, Adam O. Taylora, Rachel M. Pacea, David A. Meekinsa, Hou-Fu Guoa,Youngjun Kimb, and Matthew S. Gentrya,1

aMolecular and Cellular Biochemistry and Center for Structural Biology, University of Kentucky, Biomedical Biological Sciences Research Building, 741South Limestone, Lexington, KY 40536-0509; and bDepartment of Applied Biochemistry, Konkuk University, 322 Danwol-dong, Chungju-city, Chungbuk,380-701, Korea

Communicated by Jack E. Dixon, Howard Hughes Medical Institute, Chevy Chase, MD, June 30, 2010 (received for review May 25, 2010)

Living organisms utilize carbohydrates as essential energy storagemolecules. Starch is the predominant carbohydrate storage mole-cule in plants while glycogen is utilized in animals. Starch is awater-insoluble polymer that requires the concerted activity ofkinases and phosphatases to solubilize the outer surface of the glu-can and mediate starch catabolism. All known plant genomesencode the glucan phosphatase Starch Excess4 (SEX4). SEX4 candephosphorylate both the starch granule surface and soluble phos-phoglucans and is necessary for processive starch metabolism. Thephysical basis for the function of SEX4 as a glucan phosphatase iscurrently unclear. Herein, we report the crystal structure of SEX4,containing phosphatase, carbohydrate-binding, and C-terminal do-mains. The three domains of SEX4 fold into a compact structurewith extensive interdomain interactions. The C-terminal domainof SEX4 integrally folds into the core of the phosphatase domainand is essential for its stability. The phosphatase and carbohydrate-binding domains directly interact and position the phosphataseactive site toward the carbohydrate-binding site in a single conti-nuous pocket. Mutagenesis of the phosphatase domain residueF167, which forms the base of this pocket and bridges the two do-mains, selectively affects the ability of SEX4 to function as a glucanphosphatase. Together, these results reveal the unique tertiaryarchitecture of SEX4 that provides the physical basis for its functionas a glucan phosphatase.

carbohydrate ! Lafora disease ! laforin ! phosphorylation

Plants and animals store carbohydrates as starch and glycogen,respectively. Starch is produced in diurnal cycles and is com-

posed of <10% w"w amylose and >80% w"w amylopectin inArabidopsis thaliana leaves (1). Amylose is a linear molecule com-posed of glucose moieties linked by !-1,4-glycosidic linkages withvery few branches. Amylopectin, which is similar to glycogen, iscomposed of !-1,4-glycosidic linkages with !-1,6-glycosidicbranches, but amylopectin branches are arranged in clusters atregular intervals and the branches form double helices that packtogether to form crystalline lamellae (2, 3). The decreasedbranching and crystalline lamellae of amylopectin are key contri-butors to the insolubility of starch, while glycogen has morebranches and is water-soluble.

Starch is a water-insoluble polymer whose surface is inacces-sible to most enzymes. Recent work convincingly demonstratesthat reversible starch phosphorylation and dephosphorylation isessential for processive starch metabolism (reviewed in refs. 4–7).An essential signal triggering starch catabolism is phosphoryla-tion on the C6 position of glucose moieties on the surface ofstarch by glucan water dikinase (GWD/R1) (8, 9). C6 phosphor-ylation triggers C3 phosphorylation by phosphoglucan water di-kinase (PWD) (8, 10, 11). Recent data suggest that C6 phos-phorylation fits within the unphosphorylated structure of theamylopectin helix, but C3 phosphorylation imposes significantsteric effects and is predicted to induce a conformational change(12–15). This suggests that GWD-directed C6 phosphorylationpromotes local hydration of crystalline lamellae and that

PWD-directed C3 phosphorylation induces helix strain. Thishelix strain allows substrate access to "-amylases that release mal-tose from the surface of the starch molecule (4). However, in or-der to efficiently digest the starch molecule, glucan phosphataseactivity is necessary to prevent accumulation of phosphorylatedstarch and phosphorylated starch breakdown intermediates (16).

All known Archaeplastida/Kingdom Plantae genomes encodefor the glucan phosphatase Starch Excess4 (SEX4), but genomesoutside of Kingdom Plantae lack a SEX4 ortholog (17). SEX4contains an amino-terminal chloroplast Targeting Peptide (cTP),followed by a dual specificity phosphatase (DSP) domain, andcarbohydrate-binding module (CBM) (17, 18). Mutations inthe SEX4 gene result in substantial accumulations of starch inA. thaliana leaves due to decreased rates of degradation and theaccumulation of soluble phospho-oligosaccharides (16, 19–21). Inaddition to dephosphorylating amylopectin (18), recombinantSEX4 also dephosphorylates crystalline maltodextrins (22),starch granules isolated from A. thaliana leaves (16), and phos-pho-oligosaccharides (16). While C6 and C3 phosphorylation iscarried out by two dikinases, SEX4 is able to dephosphorylateboth C6 and C3 positions with similar kinetics (22). Thus, itappears that SEX4 possesses a broader substrate specificity thanthe kinases, in terms of phosphate position, length of glucan, andsolubility of the glucan.

SEX4 is a member of the larger glucan phosphatase family thatincludes the human protein laforin (17–19). Mutations in thegene encoding laforin result in Lafora disease (LD, OMIM254780), an autosomal recessive progressive myoclonus epilepsy(23). A hallmark of LD is the accumulation of intracellular water-insoluble glucans, i.e., glucose polymers linked by glycosidicbonds, termed Lafora bodies (LBs) (23–27). Laforin containsa CBM and DSP, like SEX4, but in the opposite orientation.Similar to SEX4, laforin dephosphorylates phospho-glucans invitro (28), and LBs from LD patients and a LD mouse modelhave increased phosphate compared to glycogen (29–32). Strik-ingly, mutations in the genes encoding both laforin and SEX4 re-sult in the accumulation of insoluble glucans (19–21). In addition,laforin partially complements mutations in SEX4, highlighting afunctional overlap between divergent glucan phosphatases (18).

The structure, mechanism, and basis for specificity of glucanphosphatases are all unknown. Herein, we report a previously un-described structure of a glucan phosphatase. The structure re-veals that the domains of SEX4 form an integral structural unit,

Author contributions: C.W.V.K. and M.S.G. designed research; C.W.V.K., A.O.T., R.M.P.,D.A.M., H.-F.G., and M.S.G. performed research; D.A.M., H.-F.G., Y.K., and M.S.G.contributed new reagents/analytic tools; C.W.V.K., A.O.T., R.M.P., D.A.M., H.-F.G., andM.S.G. analyzed data; and C.W.V.K. and M.S.G. wrote the paper.

The authors declare no conflict of interest.

Data deposition: The atomic coordinates and structure factors have been deposited in theProtein Data Bank, www.pdb.org (PDB ID code 3NME).1To whom correspondence may be addressed. E-mail: craig.vanderkooi@uky.edu ormatthew.gentry@uky.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1009386107/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1009386107 PNAS Early Edition ! 1 of 6

BIOCH

EMISTR

Y

with extensive interdomain interactions. The unique multidomainstructure of SEX4 serves to align the phosphatase active site andcarbohydrate-binding face. This extended contiguous surfaceproduces an active site incorporating both catalytic and glucansubstrate binding functionalities, providing a structural model forthe physical basis of SEX4 function as a glucan phosphatase.

ResultsStructure of SEX4. A. thaliana SEX4 (SEX4) is a 379 amino acidprotein containing three known domains: an amino-terminal cTP,a DSP domain, and a CBM family member 48 (Fig. 1A). SEX4(residues 90–379, C198S) was crystallized, selenomethioninesingle-wavelength anomalous dispersion (SAD) data collected,and the structure determined at 2.4 Å (Table 1). The structureof the SEX4 glucan phosphatase reveals a unique set of extensiveinterdomain interactions producing a complex tertiary architec-ture (Fig. 1B). The SEX4 DSP domain consists of a central five-stranded "-sheet flanked by eight !-helices (Fig. 1C and Fig. S1).The SEX4 CBM domain possesses six "-sheets that fold into a

characteristic compact " sandwich composed of antiparallelsheets (Fig. 1C). C-terminal to the CBM is an extended regionthat possesses two !-helices (Fig. 1C). The structure reveals anintegrally folded unit composed of an N-terminal DSP domain,CBM, and a previously unrecognized domain at the C terminus.

The SEX4 DSP Domain. The DSP domain (residues 90–252) has acharacteristic !"! protein tyrosine phosphatase (PTP) fold. DSPdomains contain a number of conserved elements (reviewed inref. 33) (Fig. S1). The SEX4 active site sequence HCTAGMGRA(residues 197–205), also known as the PTP loop (33, 34), lies be-tween "5 and !6 at the base of the active site cleft. The cysteine inthis motif functions as a nucleophile during catalysis, attackingthe phosphorous atom of the substrate and forming a phosphoen-zyme intermediate (34). The D-loop of SEX4 (residues 162–172)is arranged similarly to other DSP domains, with D166 in positionto act as the general acid catalyst to enhance hydrolysis of thephosphoenzyme intermediate (33, 34) (Fig. S1). In addition,the variable insert and recognition domain of DSPs, includingSEX4, contribute to the depth of the active site (33, 35).

The SEX4 structure is the catalytically inactive C198S mutantand has phosphate tightly bound (Fig. S2A). The SEX4 active siteis formed by the variable insert, recognition domain, D-loop, andR-motif (Fig. S2B). A structural search of the Protein Data Bankusing DALI identifies relatively modest structural homology withprotein phosphatases (36). The closest structural homologue tothe SEX4 DSP domain is a PTP from the Archaea thermophileSulfolobus solfataricus, SsoPTP (PDB 2I6O) with an rmsd of 2.5 Åand 18% identity by sequence (37). The limited structural simi-larities between the glucan phosphatase SEX4 and PTP familymembers are due to two features within the DSP domain andthe unique C-terminal domain of SEX4.

The most striking physical differences between the DSP do-main of SEX4 and other DSP domains are found within the vari-able insert and AYLM motif. The SEX4 variable insert (residues131–157) is longer than many proteinaceous DSPs, includinghuman Vaccinia virus H1-related phosphatase (VHR) (Fig. S3).Additionally, while the variable insert of DSP domains typicallycontain little to no secondary structure, the variable insert ofSEX4 has two !-helices (!3 and !4) with !3 forming the apex ofone side of the active site cleft (Fig. 1C and Fig. S2B). The AYLM

A

B

C

Fig. 1. Structure of SEX4. (A) SEX4 domain structure. The active site of SEX4is denoted with a black line and labeled Cx5R. (B) Structure of SEX4 showingthe integrated architecture of DSP (pink), CBM (green), and C-terminal (blue)domains. (C) Ribbon diagram of SEX4 (residues 90–379) with the active sitelabeled and S198 in gold sticks. Elements of secondary structure are num-bered consecutively from N to C termini.

Table 1. Data collection and refinement statistics

Data collection

Beamline APS 22-IDSpace group P212121Wavelength 0.9792Unit-cell parameters 59.408, 73.934, 162.382Unique reflections 28453Completeness (%) 99.2 (94.7)Resolution (Å) 2.4 (2.49–2.40)Rmerge (%) 10.0 (44.7)Redundancy 3.2 (2.5)I"!!I" 11.4 (2.04)RefinementResolution Limits (Å) 20.0–2.40# reflections/# to compute Rfree 24660"2000R (Rfree) 19.8 (24.9)# protein residues 579# solvent molecules 197# phosphate molecules 2RamachandranMost favored 93.3Add. allowed 6.3Generously 0.4Disallowed 0.0RMS deviationBond, Å 0.006Angle, ° 0.997

2 of 6 ! www.pnas.org/cgi/doi/10.1073/pnas.1009386107 Vander Kooi et al.

motif, located in !6 (residues 211–214, TYMF in SEX4), is one ofthe few DSP motifs outside the PTP loop that is highly conservedamong the DSPs (33). This motif forms part of the extendedsignature sequence of the DSP family, but it is not conservedin SEX4 (Fig. S1 and S4). In the SEX4 structure, the final residueof this motif, F214, directly interacts with the C-terminal helix(Fig. 2A).

The C Terminus of SEX4 Integrally Folds into the DSP. The C-terminaldomain (residues 338–379) is composed of an extended loop fol-lowed by two !-helices (Fig. 1B, 1C and Fig. S1). The C-terminalregion of SEX4 makes the most extensive interdomain contactswith the DSP domain, with 1010 Å2 interface accessible surfacearea (Fig. 2A). The two helical regions in the C-terminal domainintimately associate with the DSP, cradling the final helix of theDSP domain. The region following the CBM (residues 338–358),including !9, interacts with !8 of the DSP domain. The final helix,!10 (residues 361–373), associates with several regions of theDSP, interacting with !5, !6, and !8 (Fig. 2A).

Many dual specificity phosphatases contain little more than aDSP domain and can be readily expressed in Escherichia coli(38–41). The DSP of SEX4 cannot be independently expressed,and we hypothesized that the intimate association of the C-ter-minal domain with the DSP underlies this phenomenon. We

identified and aligned 19 SEX4 orthologs from Kingdom Plantaegenomes, including a green alga, a moss, and land plants/trees(Fig. S4). The orthologs are 34–64% identical to At-SEX4 atthe amino acid level, with the DSP domain sharing the highestdegree of conservation (Table S1). An alignment of SEX4 ortho-logs demonstrates striking conservation in the DSP-contactingresidues in the C terminus (Fig. 2B). Within the C-terminal do-main, !9 contains a conserved -RXRL- motif. In addition, L359,located between the two helices, is invariant. Finally, !10 containsa highly conserved -ERXXLXXXL- motif.

To determine the functional significance of this region, we gen-erated a C-terminal deletion construct, SEX4 (90–340). The crys-tallized construct, containing the C terminus (residues 90–379), isproduced upon induction, produces predominantly soluble pro-tein, and is readily purified from the soluble fraction (Fig. 2C).SEX4 lacking the C-terminal domain (residues 90–340) is alsoproduced upon induction but is entirely insoluble with no proteinrecovered from the soluble fraction (Fig. 2C). Thus, the C-term-inal domain of SEX4 is necessary for soluble expression of SEX4.

The DSP and CBM Domains Interact. Based on CAZy classification(42), the SEX4 CBM domain (residues 253–338) belongs to theCBM48 family, the same family as the AMPK" CBM. Consistentwith this classification, structural analyses of the Protein DataBank using DALI identifies AMPK"1 (PDB 1Z0M) as the clo-sest structural homologue (rmsd # 1.3 Å, 30% identity by se-quence) (43).

The DSP and CBM domains of SEX4 directly interact with457 Å2 of interfacial surface area. A broad surface formed bymultiple residues creates the extensive interface between theCBM and DSP domain (Fig. 3A). Specifically, the D-loop/!5 re-gion of the DSP possesses a highly conserved FDXFDLR motif(residues 167–173) that packs against the "7/8 loop of the CBMthat possesses a corresponding LDIGWG motif (residues 274–279) (Fig. 3A). An alignment of SEX4 orthologs demonstratesstriking conservation in these DSP/CBM interacting residues(Fig. S4). Additionally, !8 of the DSP packs against "10 of theCBM. These interactions position the two domains such that theyform a continuous surface that runs the length of the enzyme.

Intriguingly, the two SEX4 molecules in the asymmetric unitdiffer slightly in the relative orientation of the DSP and CBMdomain via a 10° rotation (Fig. 3B). This rotation is coupledto alteration in the packing of !9 from the C-terminal domainand results in a slight flexing of the hinge region between theCBM and DSP domain.

A

B

C

Fig. 2. Interaction of the C-terminal domain with the DSP domain. (A) The C-terminal domain wraps around the final DSP !8 helix and additionally inter-acts with both !5 and !6. F214 in !6, the final residue of the AYLM motif, ishighlighted in purple. (B) Multiple sequence alignment of SEX4 orthologs re-veals that the residues contributing to the DSP interface (highlighted withred asterisks) are highly conserved. (C) The C-terminal helix is essential forsoluble expression of SEX4. #89-SEX4 lacks the N-terminal 89 residues buthas the full-length C terminus, to residue 379. 90–340 has both N- andC-terminal truncations. U # Uninduced cells, I # Cells induced with IPTG,P # Pellet of insoluble protein, S # Soluble protein, E # eluted from IMACcolumn.

BA

90o

10º

!8

D/!5

"7/8

"10

Fig. 3. Interaction of the CBM and DSP domains. (A) The DSP domain andCBM of SEX4 directly interact via several structural elements. The D-loop/!5region of the DSP (D/!5) possesses packs against the "7/8 loop of the CBM.Additionally, !8 of the DSP packs against "10 of the CBM. Residues at theinterdomain interface are highlighted in blue sticks. (B) Comparison of thetwo molecules in the asymmetric unit reveals that the DSP/CBM interfaceis maintained but has rotational flexibility. Overlay generated by superimpos-ing the DSP domains of the two molecules.

Vander Kooi et al. PNAS Early Edition ! 3 of 6

BIOCH

EMISTR

Y

SEX4 Possesses a Unique Ligand Binding Active Site. The DSP activesite is located between "5 and !6 while the carbohydrate-bindingsite of SEX4 is located in the deep cleft formed by the extendedloop between "6 and "7 and the five-sheet face of the domain(18, 43, 44). The unique multidomain structure of SEX4 servesto align the DSP active site and carbohydrate-binding face of theCBM into a single surface (Fig. 4A and Fig. S5). This extendedsurface produces a contiguous active site incorporating bothcatalytic and glucan binding functionalities.

F167 is an invariant residue located in a highly conserved re-gion within the D-loop (Fig. S5C). F167 is physically oriented be-tween D166 of the DSP domain and W278 of the CBM (Fig. 4Aand Fig. S5). D166 acts as the general acid-base catalyst to en-hance leaving group expulsion in most DSPs, and W278 is a cri-tical glucan binding residue located at the apex of the interactingregions (Fig. 4A and Fig. S5) (33, 34). F167 forms the base of thebridge that connects the CBM binding site to the DSP active site.

Most DSP family members possess a short chain hydrophilic re-sidue, S/T/N/H, at the position corresponding to F167 (Fig. S5C).Only one other DSP, VHZ, contains a phenylalanine residue inthe same position.

Based on the structure, we predicted that F167 functions incoupling the substrate binding and catalysis functions of SEX4by positioning the phosphoglucan directly into the extendedSEX4 catalytic cleft. To test this prediction, we mutated F167 toa serine, tyrosine, and methionine. Proteins were expressed, puri-fied, and assayed for phosphatase activity. We assayed bothglucan phosphatase activity, utilizing amylopectin (Fig. S5D),and generic phosphatase activity, utilizing the exogenous sub-strate para-nitrophenylphosphate (p-NPP) (Fig. S5E). We thencompared the ratio of these two activities (i.e., glucan activity/generic activity) to determine specific glucan phosphatase activity(Fig. 4B). F167S, a short chain hydrophilic residue most com-monly found across the DSP family, showed decreased specificactivity as a glucan phosphatase, supporting the hypothesis thatF167 specifically functions in SEX4 glucan phosphatase activity.F167Y, which is expected to minimally physically perturb the sys-tem, but whose hydroxyl would partially occlude the glucan bind-ing site of the CBM, substantially decreases the glucan phos-phatase activity of SEX4 without affecting p-NPP activity. Ofnote, F167M, the corresponding residue in the glucan phospha-tase laforin, did not impact the specific glucan phosphataseactivity of SEX4. Cumulatively, these data demonstrate a rolefor F167 in the glucan phosphatase activity of SEX4.

DiscussionOur data demonstrate the essential tertiary architecture neces-sary for the glucan phosphatase activity of SEX4. The SEX4structure reveals multiple features unique to SEX4 comparedto other DSPs, including a variable insert that contains two !-he-lices, adaptation of the AYLMmotif for interdomain interaction,a novel C-terminal domain that is structurally integrated into theDSP domain, extensive DSP–CBM interactions that align theDSP active site and CBM binding site into a common pocket,and a channel formed by the CBM-DSP interface. These featuresserve to structurally differentiate SEX4 from phosphatases thatdephosphorylate proteinaceous substrates.

The majority of phosphatases are protein phosphatases (45–48). The PTP superfamily is divided into the classical PTPs thatdephosphorylate only phosphotyrosine; the DSPs that dephos-phorylate phosphotyrosine, phosphoserine, and phosphothreo-nine; and a subset of DSPs that dephosphorylate nonprotein-aceous substrates (35, 47, 49, 50). Within the PTP superfamily,the depth of the active site contributes a significant portion tosubstrate specificity (33, 47). The active site depth of proteinac-eous DSPs is #6 Å, allowing access of both pS/Tand pY (33, 51).Alternatively, tyrosine-specific PTPs possess an #9 Å cleft,allowing the longer pY access but limiting the shorter pS/T(33, 52). The nonproteinaceous phosphoinositol phosphatasesphosphate and tensin homolog (PTEN) and myotubularin-related protein 2 (MTMR2) have an 8 Å and 13 Å deep pocket,respectively, with an elliptical opening nearly twice as wide as thetyrosine-specific PTP1B (53, 54).

The unique active site cleft of SEX4 forms a 21 Å pocket fromphosphatase active site to CBM glucan binding region. If SEX4requires simultaneous engagement of both glucan binding andphosphatase domains by a single substrate, then it would mini-mally require a glucan composed of approximately five to sixglucose moieties (i.e., phospho-maltopentaose or -hexaose) tospan this region. Therefore, it would require a linker region ofthree to four glucans to span the two sites, with an additional glu-cose bound at the CBM and a phospho-glucose moiety at theactive site. Our data demonstrate that F167 is a critical residuefor SEX4 glucan phosphatase activity. F167 bridges the two do-mains and physically links the CBM binding site to the DSP active

A

B

S198

F167

W278

WT C198S F167S F167Y F167M0

50

100

150

200

250

300

spec

ific

pho

spha

tase

activ

ity (g

luca

n/pN

PP)

Fig. 4. SEX4 possesses a contiguous ligand-binding active site. (A) Surfaceview of the DSP active site to CBM binding site pocket. F167 (yellow),S198 (red), and W278 (green) are all highlighted. (B) SEX4 specific phospha-tase activity expressed as the ratio of glucan phosphatase activity/generic p-NPP activity. Glucan phosphatase activity was quantified aspmol min$1 mg$1 of phosphate release. The activity against p-NPP was quan-tified as "mol min$1 mg$1. #89-SEX4 (WT), C198S, and three F167 mutantswere analyzed. Error bars indicate means$ the standard deviation.

4 of 6 ! www.pnas.org/cgi/doi/10.1073/pnas.1009386107 Vander Kooi et al.

site. It is tempting to speculate that F167 functions by positioningthe linker region so that the phosphoglucan is properly presentedto the SEX4 active site. In addition to the 21 Å wide pocket, theDSP–CBM interface also forms a channel on the face of theenzyme that contains both the glucan binding site and DSPactive site. This deep interfacial channel terminates at the openbinding pocket positioned between DSP active site and the CBMbinding site. This open CBM terminus suggests a physical basis bywhich SEX4 could accommodate both linear phospho-oligosac-charides and also longer nonlinear phosphoglucan substrates,such as those found in the starch granule surface.

In addition to these features, the SEX4 structure reveals thatthe C-terminal domain of SEX4 integrally binds to the DSP do-main, and our results demonstrate that this domain is essentialfor protein stability. As previously noted, the AYLM motif,located in !6 (residues 211–214), is one of the few DSP motifsoutside the PTP loop that is highly conserved among DSPs(33). However, this motif is not conserved in any of the knownglucan phosphatases, including SEX4. In SEX4, the final residueof this motif, F214, directly interacts with the C-terminal helix.The interaction between F214 and the C-terminal helix suggeststhat the AYLMmotif is adapted in SEX4 to support DSP domaininteraction with the C-terminal domain and may prove a generalfeature of glucan phosphatases.

While the C-terminal domain is essential for maintaining thefold and stability of SEX4, the CBM is essential for substratebinding. The interdomain orientation of the DSP domain andCBM is essential for glucan phosphatase activity of SEX4. Inter-estingly, the overall positioning of the SEX4 DSP and CBMdomains are quite similar to that of the phosphatase and C2 do-mains of PTEN, with the equivalent elements in the phosphatasedomain, D loop/!5 and !8, employed in the interdomaininteraction (54).

Previous data from H/D exchange mass spectrometry of SEX4indicated that no major interdomain conformational change oc-curred in SEX4 upon substrate binding (55). Our data are con-sistent with this conclusion but suggest that some rotation of theDSP/CBM interface is possible. Intriguingly, the two SEX4 mo-lecules in the asymmetric unit differ by a 10° rotation that altersthe packing of !9 from the C-terminal domain. This rotation pro-vides a possible manner to regulate substrate entry, which we arecurrently investigating. SEX4 dephosphorylates both the starchgranule surface as well as soluble phosphoglucans (16). In addi-tion, SEX4 can dephosphorylate both the C3 and C6 position ofglucose (22). Therefore, SEX4 possesses a broad substratespecificity and must accommodate both different glucans (i.e., so-luble and crystalline) as well as differently positioned phosphates(i.e., C3 versus C6). We are currently investigating if this con-formational lability accommodates binding to inherently chemi-cal heterogeneous glucans and/or allows processive substratedephosphorylation. The SEX4 structure begins to address howthe enzyme manages these sterically different substrates.

Materials and MethodsProtein Expression and Purification. Cloning of Arabidopsis thaliana SEX4 waspreviously described (18). Based on data from secondary structure predic-tions, disorder predictions, and deuterium exchange mass spectrometryexperiments, we generated a construct of A. thaliana SEX4 lacking the first89 amino acids (#89–SEX4) (55). #89–SEX4 lacks the cTP (predicted to beresidues 1–54) alongwith residues up to the DSP recognition domain (Fig. S1).#89–SEX4, point mutants, and #89–SEX4(90–340) were subcloned into pET28

(Novagen) using NdeI and XhoI to encode HIS6, a thrombin cleavage site, andSEX4. BL21-CodonPlus Escherichia coli cells (Stratagene) or T7 Express Crystal(NEB) were transformed with expression vectors for production of native andselenomethionine labeled protein, respectively. Cells were grown at 37 °C in2 ! YT or M9 supplemented with 100 mg"L selenomethionine to an OD600 #0.6, placed on ice for 20 min, induced with 1 mM isopropyl "-D-thiogalacto-side (IPTG), grown at 20 °C for 16 h, and harvested by centrifugation. Cellswere lysed in 20 mM Tris-HCl (pH 7.5), 100 mM NaCl, 15 mM imidazole,and 2 mM dithiothreitol (DTT), centrifuged, and proteins were purified usinga Profinia IMAC column (Bio-Rad) with a Profinia protein purification system(Bio-Rad). Protein was dialyzed in 20 mM Tris-HCl (pH 7.5), 100 mM NaCl, and2 mM dithiothreitol (DTT) overnight at 4 °C in the presence of thrombin.Protein was then reverse purified over the Profinia IMAC column. Proteinwas purified to homogeneity using a HiLoad 16"60 Superdex 200 size exclu-sion column (GE Healthcare).

Crystal Structure Determination. Single high quality crystals, with two proteinmolecules in the asymmetric unit, were obtained via hanging drop vapor dif-fusion using aMosquito crystallization robot (TTPLabtech) with a 200 nL dropusing a 100%100 mixture of selenomethionine labeled SEX4 (12 mg"mL):0.4 M lithium citrate and 24% PEG3350. Crystals were briefly soaked in21% PEG3350 and 0.5 M ammonium hydrogen phosphate and then flashfrozen. SAD data was collected at the selenomethionine peak wavelength,based on a fluorescence energy scan, on the 22-ID beamline of SER–CAT atthe Advanced Photon Source, Argonne National Laboratory (Table 1). Datawas processed using HKL2000 (56). PHENIX (57), employing HYSS (58),PHASER (59), and RESOLVE (60), was used to locate all twelve expectedselenomethionine sites, obtain phase information, perform density modifica-tion, and generate an initial structural model. The structure was then fullybuilt and refined via iterative model building and refinement using Coot(61) and Refmac5 (62), respectively (Table 1).

Structural and Sequence Analysis. Interdomain interfaces were analyzed usingPROTORP (63). Molecular graphics were prepared using the programs Pymoland MOLMOL (64). The sequences of A. thaliana SEX4 orthologs wereobtained by performing tBLASTn searches using the GenBank “dbEST” data-base or BLASTp and PSI-BLAST searches using GenBank “eukaryote genome”and “nonredundant” (nr) databases, PlantGDB, Department of Energy JointGenome Institute Resource, and Phytozome. Amino-acid sequences of SEX4orthologs were aligned by ClustalW and refined manually using MacVector.

Phosphatase Assays. Phosphatase assays utilizing p-NPP have been previouslydescribed and were performed with the following modifications (18, 28): Hy-drolysis of p-NPP was performed in 50 $l reactions containing 1X phos-phatase buffer (0.1 M sodium acetate, 0.05 M bis-Tris, 0.05 M Tris-HCl pH6.0, 2 mM dithiothreitol), 50 mM p-NPP, and 100–500 ng of enzyme at 37°C for 15–20 min. The reaction was terminated by the addition of 200 $lof 0.25 M NaOH and absorbance was measured at 410 nm. Malachite greenassays were performed as previously described with the following modifica-tions (65): 1X phosphatase buffer (0.1 M sodium acetate, 0.05 M bis-Tris,0.05 M Tris-HCl pH 8.0, 2 mM dithiothreitol), 100–500 ng of SEX4, and45 $g of amylopectin (Sigma) in a final volume of 20 $l. The reaction wasstopped by the addition of 20 $l of 0.1 M N-ethylmaleimide and 80 $l ofmalachite green reagent. Absorbance was measured after 40 minutes at620 nm. Specific glucan phosphatase activity is expressed as phosphaterelease from amylopectin divided by the specific activity against p-NPP.

ACKNOWLEDGMENTS. We thank Dr. Qingjun Wang for technical assistance,Dr. Carol Beach in the Molecular Basis of Human Disease COBRE ProteomicsCore, the Molecular Basis of Human Disease COBRE administrative and pro-tein analytical cores, and Drs. Carolyn Worby, Oliver Kötting, Doug Andres,Mike Begley, and Choel W. Kim for technical assistance and discussions. Thiswork was supported by National Institutes of Health (NIH) grantsR00NS061803, P20RR020171, and R01NS070899; University of KentuckyCollege of Medicine startup funds to M.S.G.; NIH grant P20RR0202171 toC.W.V.K.; and the Korea Research Foundation grant KRF-2008-331-E00031to Y.J.K.

1. Zeeman SC, Smith SM, Smith AM (2007) The diurnal metabolism of leaf starch.Biochemical J 401:13–28.

2. Buleon A, Colonna P, Planchot V, Ball S (1998) Starch granules: Structure and biosynth-esis. Int J Biol Macromol 23(2):85–112.

3. Gallant DJ, Bouchet B, Baldwin PM (1997) Microscopy of starch: Evidence of a newlevel of granule organization. Carbohyd Polym 32(3–4):177–191.

4. Blennow A, Engelsen SB (2010) Helix-breaking news: Fighting crystalline starch energydeposits in the cell. Trends Plant Sci 15(4):236–240.

5. Fettke J, et al. (2009) Eukaryotic starch degradation: Integration of plastidial andcytosolic pathways. J Exp Bot 60(10):2907–2922.

6. Kotting O, Kossmann J, Zeeman SC, Lloyd JR (2010) Regulation of starch metabolism:The age of enlightenment? Curr Opin Plant Biol 13(3):321–329.

7. Zeeman SC, Kossmann J, Smith AM (2010) Starch: Its metabolism, evolution, andbiotechnological modification in plants. Annu Rev Plant Biol 61:209–234.

8. Ritte G, et al. (2006) Phosphorylation of C6- and C3-positions of glucosyl residues instarch is catalysed by distinct dikinases. FEBS Lett 580(20):4872–4876.

Vander Kooi et al. PNAS Early Edition ! 5 of 6

BIOCH

EMISTR

Y

9. Ritte G, et al. (2002) The starch-related R1 protein is an alpha -glucan, water dikinase.Proc Natl Acad Sci USA 99(10):7166–7171.

10. Baunsgaard L, et al. (2005) A novel isoform of glucan, water dikinase phosphorylatespre-phosphorylated alpha-glucans and is involved in starch degradation inArabidopsis. Plant J 41(4):595–605.

11. Kotting O, et al. (2005) Identification of a novel enzyme required for starch metabo-lism in Arabidopsis leaves. The phosphoglucan, water dikinase. Plant Physiol 137(1):242–252.

12. Hansen PI, et al. (2008) Structure and hydration of the amylopectin trisaccharide build-ing blocks—Synthesis NMR and molecular dynamics. Biopolymers 89(12):1179–1193.

13. Hansen PI, et al. (2009) Starch phosphorylation–maltosidic restrains upon 3'- and6'-phosphorylation investigated by chemical synthesis, molecular dynamics andNMR spectroscopy. Biopolymers 91(3):179–193.

14. Kozlov SS, Blennow A, Krivandin AV, Yuryev VP (2007) Structural and thermodynamicproperties of starches extracted fromGBSS and GWD suppressed potato lines. Int J BiolMacromol 40(5):449–460.

15. Sanderson JS, Daniels RD, Donald AM, Blennow A, Engelsen SrB (2006) ExploratorySAXS and HPAEC-PAD studies of starches from diverse plant genotypes. CarbohydPolym 64(3):433–443.

16. Kotting O, et al. (2009) STARCH-EXCESS4 is a Laforin-like phosphoglucan phosphataserequired for starch degradation in Arabidopsis thaliana. Plant Cell 21(1):334–346.

17. Gentry MS, Pace RM (2009) Conservation of the glucan phosphatase laforin is linked torates of molecular evolution and the glycogen metabolism of the organism. BMC EvolBiol 9(1):138.

18. Gentry MS, et al. (2007) The phosphatase laforin crosses evolutionary boundaries andlinks carbohydrate metabolism to neuronal disease. J Cell Biol 178(3):477–488.

19. Niittyla T, et al. (2006) Similar protein phosphatases control starch metabolism inplants and glycogen metabolism in mammals. J Biol Chem 281(17):11815–11818.

20. Sokolov LN, Dominguez-Solis JR, Allary AL, Buchanan BB, Luan S (2006) A redox-regulated chloroplast protein phosphatase binds to starch diurnally and functionsin its accumulation. Proc Natl Acad Sci USA 103(25):9732–9737.

21. Zeeman SC, Northrop F, Smith AM, Rees T (1998) A starch-accumulating mutant ofArabidopsis thaliana deficient in a chloroplastic starch-hydrolysing enzyme. Plant J15(3):357–365.

22. Hejazi M, Fettke J, Kotting O, Zeeman SC, Steup M (2010) The Laforin-like dual-spe-cificity phosphatase SEX4 from Arabidopsis hydrolyzes both C6- and C3-phosphateesters introduced by starch-related dikinases and thereby affects phase transitionof alpha-glucans. Plant Physiol 152(2):711–722.

23. Lafora GR (1911) Uber des Vorkommen amyloider KJrperchen im innern der Ganglien-zellen. Virchows Arch A 205:295–303.

24. Harriman DG,Millar JH, Stevenson AC (1955) Progressive familial myoclonic epilepsy inthree families: Its clinical features and pathological basis. Brain 78(3):325–349.

25. Schwarz GA, Yanoff M (1965) Lafora’s Disease. Distinct clinico-pathologic form ofUnverricht’s Syndrome. Archives of neurology 12:172–188.

26. Yokoi S, Austin J, Witmer F (1967) Isolation and characterization of Lafora bodies intwo cases of myoclonus epilepsy. J Neuropath Exp Neur 26(1):125–127.

27. Yokoi S, Austin J, Witmer F, Sakai M (1968) Studies in myoclonus epilepsy (Lafora bodyform). I. Isolation and preliminary characterization of Lafora bodies in two cases. ArchNeurol 19(1):15–33.

28. Worby CA, Gentry MS, Dixon JE (2006) Laforin: A dual specificity phosphatase thatdephosphorylates complex carbohydrates. J Biol Chem 281(41):30412–30418.

29. Sakai M, Austin J, Witmer F, Trueb L (1970) Studies in myoclonus epilepsy (Lafora bodyform). II. Polyglucosans in the systemic deposits of myoclonus epilepsy and in corporaamylacea. Neurology 20(2):160–176.

30. Schnabel R, Seitelberger F (1968) Histophysical and histochemical investigations ofmyoclonus bodies. Pathol Eur 3(2):218–226.

31. Tagliabracci VS, et al. (2008) Abnormal metabolism of glycogen phosphate as a causefor lafora disease. J Biol Chem.

32. Tagliabracci VS, et al. (2007) Laforin is a glycogen phosphatase, deficiency of whichleads to elevated phosphorylation of glycogen in vivo. Proc Natl Acad Sci USA 104(49):19262–19266.

33. Alonso A, Rojas A, Godzik A, Mustelin T (2003) The dual-specific protein tyrosinephosphatase family (Springer, Berlin), pp 333–358.

34. Denu JM, Stuckey JA, Saper MA, Dixon JE (1996) Form and function in proteindephosphorylation. Cell 87(3):365–368.

35. GentryMS, Dixon JE,Worby CA (2009) Lafora disease: Insights into neurodegenerationfrom plant metabolism. Trends Biochem Sci 34(12):628–639.

36. Holm L, Sander C (1995) Dali: A network tool for protein structure comparison. TrendsBiochem Sci 20(11):478–480.

37. Chu HM, Wang AH (2007) Enzyme-substrate interactions revealed by the crystalstructures of the archaeal Sulfolobus PTP-fold phosphatase and its phosphopeptidecomplexes. Proteins 66(4):996–1003.

38. Agarwal R, Burley SK, Swaminathan S (2008) Structure of human dual specificityprotein phosphatase 23, VHZ, enzyme-substrate/product complex. J Biol Chem 283(14):8946–8953.

39. Koksal AC, Nardozzi JD, Cingolani G (2009) Dimeric quaternary structure of theprototypical dual specificity phosphatase VH1. J Biol Chem 284(15):10129–10137.

40. Sun JP, et al. (2005) Structure and biochemical properties of PRL-1, a phosphataseimplicated in cell growth, differentiation, and tumor invasion. Biochemistry 44(36):12009–12021.

41. Yuvaniyama J, Denu JM, Dixon JE, Saper MA (1996) Crystal structure of the dualspecificity protein phosphatase VHR. Science 272(5266):1328–1331.

42. Boraston AB, Bolam DN, Gilbert HJ, Daview GJ (2004) Carbohydrate-binding modules:Fine-tuning polysaccharide recognition. Biochem J 382:769–781.

43. Polekhina G, et al. (2005) Structural basis for glycogen recognition by AMP-activatedprotein kinase. Structure 13(10):1453–1462.

44. Polekhina G, et al. (2003) AMPK beta subunit targets metabolic stress sensing toglycogen. Curr Biol 13(10):867–871.

45. Janssens V, Longin S, Goris J (2008) PP2A holoenzyme assembly: In cauda venenum (thesting is in the tail). Trends Biochem Sci 33(3):113–121.

46. Shi Y (2009) Serine/threonine phosphatases: Mechanism through structure. Cell 139(3):468–484.

47. Tonks NK (2006) Protein tyrosine phosphatases: From genes, to function, to disease.Nat Rev Mol Cell Biol 7(11):833–846.

48. Rigden DJ (2008) The histidine phosphatase superfamily: Structure and function.Biochem J 409(2):333–348.

49. Alonso A, et al. (2004) Protein tyrosine phosphatases in the human genome. Cell 117(6):699–711.

50. Robinson FL, Dixon JE (2006) Myotubularin phosphatases: Policing 3-phosphoinosi-tides. Trends Cell Biol 16(8):403–412.

51. Guan KL, Broyles SS, Dixon JE (1991) A Tyr/Ser protein phosphatase encoded byvaccinia virus. Nature 350(6316):359–362.

52. Barford D, Flint AJ, Tonks NK (1994) Crystal structure of human protein tyrosinephosphatase 1B. Science 263(5152):1397–1404.

53. Begley MJ, et al. (2003) Crystal structure of a phosphoinositide phosphatase, MTMR2:Insights into myotubular myopathy and Charcot-Marie-Tooth syndrome. Mol Cell 12(6):1391–1402.

54. Lee J-O, et al. (1999) Crystal structure of the PTEN tumor suppressor: Implications for itsphosphoinositide phosphatase activity and membrane association. Cell 99(3):323–334.

55. Hsu S, et al. (2009) Structural insights of glucan phosphatase dynamics using amidehydrogen/deuterium exchange mass spectrometry. Biochemistry 48(41):9891–9902.

56. Otwinowski Z, Minor W (1997) Processing of X-ray diffraction data collected in oscilla-tion mode. Method Enzymol 276:307–326.

57. Adams PD, et al. (2010) PHENIX: A comprehensive Python-based system for macro-molecular structure solution. Acta Crystallogr D 66(Pt 2):213–221.

58. Grosse-Kunstleve RW, Adams PD (2003) Substructure search procedures for macro-molecular structures. Acta Crystallogr D 59(Pt 11):1966–1973.

59. McCoy AJ, et al. (2007) Phaser crystallographic software. J Appl Crystallogr40(Pt 4):658–674.

60. Terwilliger TC, et al. (2008) Iterative model building, structure refinement and densitymodification with the PHENIX AutoBuild wizard. Acta Crystallogr D 64(Pt 1):61–69.

61. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot.Acta Crystallogr D 66(Pt 4):486–501.

62. Murshudov GN (1997) Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D 53(Pt 3):240–255.

63. Reynolds C, Damerell D, Jones S (2009) ProtorP: A protein-protein interaction analysisserver. Bioinformatics 25(3):413–414.

64. Koradi R, Billeter M, Wuthrich K (1996) MOLMOL: A program for display and analysisof macromolecular structures. J Mol Graphics 14:51–55.

65. Harder KW, et al. (1994) Characterization and kinetic analysis of the intracellulardomain of human protein tyrosine phosphatase beta (HPTP beta) using syntheticphosphopeptides. Biochemical J 298:395–401.

6 of 6 ! www.pnas.org/cgi/doi/10.1073/pnas.1009386107 Vander Kooi et al.