Discovery and characterization of pseudocyclic cystine-knot a-amylase inhibitors with high resistance to heat and proteolytic degradation Phuong Q. T. Nguyen, Shujing Wang, Akshita Kumar, Li J. Yap, Thuy T. Luu, Julien Lescar and James P. Tam
School of Biological Sciences, Nanyang Technological University, Singapore
Keywords Obesity and type 2 diabetes are chronic metabolic diseases, and those cis-proline; cystine knot; pseudocyclics; affected could benefit from the use of a-amylase inhibitors to manage a wrightide; -amylase inhibitors starch intake. The pseudocyclics, wrightides Wr-AI1 to Wr-AI3, isolated from an Apocynaceae plant show promise for further development as Correspondence a J. P. Tam, School of Biological Sciences, orally active -amylase inhibitors. These linear peptides retain the stability Nanyang Technological University, known for cystine-knot peptides in the presence of harsh treatment. They Singapore are resistant to heat treatment and endopeptidase and exopeptidase degra- Fax: +65 6515 1632 dation, which is characteristic of cyclic cystine-knot peptides. Our NMR Tel: +65 6316 2833 and crystallography analysis also showed that wrightides, which are cur- E-mail: [email protected] rently the smallest proteinaceous a-amylase inhibitors reported, contain the backbone-twisting cis-proline, which is preceded by a nonaromatic residue (Received 22 February 2014, revised 19 June 2014, accepted 15 July 2014) rather than a conventional aromatic residue. The modeled structure and a molecular dynamics study of Wr-AI1 in complex with yellow mealworm doi:10.1111/febs.12939 a-amylase suggested that, despite having a similar structure and cys- tine-knot fold, the knottin-type a-amylase inhibitors may bind to insect a-amylase via a different set of interactions. Finally, we showed that the precursors of pseudocyclic cystine-knot a-amylase inhibitors and their bio- synthesis in plants follow a secretory protein synthesis pathway. Together, our findings provide insights for the use of the pseudocyclic a-amylase inhibitors as useful leads for the development of orally active peptidyl bio- actives, as well as an alternative scaffold for cyclic peptides for engineering metabolically stable human a-amylase inhibitors.
Database The nucleotide sequences for Wr-AI1 to Wr-AI3 have been deposited in the GenBank database under GenBank accession numbers KF679826, KF679827, and KF679828, respectively. The Wr-AI1 solution structure solved for 10 ensembles with the lowest target function is available in the Protein Data Bank under accession code 2MAU. The coordinates of Wr-AI1 crystal structure are available in the Protein Data Bank database under accession code 4BFH.
Introduction Small disulfide-rich, proteinaceous bioactives are (CK) motif, with a three-disulfide knotted structure prominent in toxins, hormones, growth factors, and formed by two disulfide bonds, together with the protease inhibitors [1]. Many contain a cystine-knot connecting backbones, forming an embedded ring
Abbreviations AAI, amaranth a-amylase inhibitor; CCK, cyclic cystine-knot; CK, cystine-knot; ER, endoplasmic reticulum; HSA, human salivary a-amylase; PCK, pseudocylic cystine-knot; PDB, Protein Data Bank; TMA, Tenebrio molitor a-amylase; UPLC, ultra-performance liquid chromatography.
FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS 4351 Pseudocyclic cystine-knot a-amylase inhibitors P. Q. T. Nguyen et al. through which the third bond penetrates [2]. Of par- proteomic and genomic methods, we identified three ticular interest in drug development is the knottin AAI-like a-amylase inhibitors, wrightide-amylase- family of CK peptides containing 25–45 residues, and inhibitors Wr-AI1 to Wr-AI3, from the medicinal often possessing protease inhibitory functions, from plant Wrightia religiosa (Apocynaceae family). We which the name was derived [3]. Knottins form com- showed that they are resistant not only to heat treat- pact and defined structures with extensive internal ment and endopeptidase degradation, but also to exo- hydrogen bonding, endowing them with resistance to peptidase. The structure of Wr-AI1 was analyzed in proteolytic degradation by endopeptidases and dena- both solution and crystal form by NMR and X-ray turation by heat or chemicals, as shown by numerous crystallography (to 1.25-A resolution), respectively. studies, including those using sequencing experiments Modeling the Wr-AI1–TMA complex with docking to determine their primary structures. Certain CK and molecular dynamics suggests that a-amylase inhi- peptides of the knottin family have further evolved bition by knottins occurs via an overall shape-fitting as macrocycles such as cyclotides, harboring cyclic mechanism rather than through a particular set of CKs (CCKs) with no termini, a feature that has polar or ionic interactions in the TMA active site made them resistant to degradation by exopeptidases pocket. We also showed that the precursors of knot- [4]. Cyclotides, generally consisting of 28–37 residues, tin-type a-amylase inhibitors contain a three-domain are known be ultrastable to proteolytic and heat deg- structure common to CK peptides. Taken together, radation, and possess robust qualities comparable to our findings provide new insights into the sequence, those of small-molecule drug candidates. All of these structure and biosynthesis of CK a-amylase inhibitors, features bode well for the development of orally which could be used as stable scaffolds in engineering active peptidyl bioactives. human a-amylase inhibitors. In a program to identify potentially orally active peptidyl bioactives for the treatment of metabolic dis- Results eases such as diabetes, we have initiated MS profiling to identify cysteine-rich peptidyl a-amylase inhibitors Isolation of a-amylase inhibitors from in traditional medicines. Plants and microorganisms W. religiosa produce a diverse group of proteinaceous a-amylase inhibitors that function in defense pathways. These Our preliminary MS profiling of crude extracts of inhibitors vary greatly in structure and size, ranging W. religiosa leaves and flowers revealed strong posi- from small peptides (3 kDa), such as amaranth a-amy- tive signals in the mass range of 3–5 kDa, indicative lase inhibitor (AAI) [5], to large proteins, such as of cysteine-rich peptides (Fig. 1). We therefore per- a-AI1, a 23-kDa a-amylase inhibitor from kidney bean formed extraction of the putative cysteine-rich pep- (Phaseolus vulgaris) [6]. They are structurally classified tides from fresh W. religiosa leaves from Vietnam into seven groups: knottin-type, c-thionin-like, CM- and Singapore in 50% ethanol, and purified them proteins, Kunitz-type, thaumatin-like, legume-lectin- through several rounds of RP-HPLC and strong cat- like, and microbial [7]. These classes of a-amylase ion-exchange HPLC. The most abundant peptides inhibitor have attracted attention as tools in agricul- from Vietnam and Singapore leaves were named ture and for antidiabetes management. wrightide-amylase-inhibitors Wr-AI1 and Wr-AI2, The smallest proteinaceous a-amylase inhibitor respectively. Each purified wrightide was fully known to date, the 3-kDa AAI, is currently the only reduced by dithiothreitol, and then digested with member of the knottin-type group to be reported. AAI trypsin and chymotrypsin. The resulting fragments comprises 32 residues harboring a CK core. This were sequenced by tandem MS, and their sequences inhibitor specifically inhibits the yellow mealworm were deduced by analyzing b-ions and y-ions (Fig. 2). Tenebrio molitor a-amylase (TMA), but is inactive By genetic analysis, we also obtained the sequence of against human and bovine a-amylases [5]. Although wrightide Wr-AI3, which could not be detected in the detailed structural study of the inhibition mechanism MS profile. of AAI on TMA has been reported, little is known Wrightides Wr-AI1 to Wr-AI3, all 30 residues in about its knottin-type homologs or their genetic pre- length, contain six cysteines, three glycines, and two cursors. prolines. Together, these three residues account for Here, we report on the discovery and characteriza- > 35% of the sequences. Wrightides share high tion of a group of linear knottins with characteristics sequence homology with each other (93–96%), differ- and a potential for use in drug development compara- ing by one or two residues (Fig. 3), and high sequence ble to those of CCK peptides. Using a combination of homology with AAI (48%).
4352 FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS P. Q. T. Nguyen et al. Pseudocyclic cystine-knot a-amylase inhibitors
Fig. 1. Tissue-specific and region-specific expression profiles of wrightides from flower and leaf of W. religiosa. A 50% ethanol extract of 1 g of each plant sample was purified with a C18 solid- phase extraction column. The eluate with 80% acetonitrile was profiled with MALDI- TOF MS to determine the occurrence of putative CK peptides in different W. religiosa plant parts from Singapore and Vietnam.
Solution structure of Wr-AI1 determined by donors on the basis of the structures. Wr-AI1 contains 1H-NMR two prolines, whose conformations were identified as cis-Pro17 and trans-Pro23. The proline cis and trans With the distance, dihedral angle and hydrogen bond 1 conformations were confirmed by the observation of restraints derived from H-NMR experiments N a N NOE crosspeaks H Asp16–H Pro17 and H Gln22– (Table 1), the solution structures of Wr-AI1 showed d a H , respectively, as the H strips of these four resi- that it adopts a similar CK scaffold as AAI, with the Pro23 dues were not identified from the noise region of H O same three disulfide linkages: CysI–IV, CysII–V, and 2 around 4.7 p.p.m. CysIII–VI, where CysIII–VI is the penetrating disulfide bond (Fig. 3A,B). The structure contains three short b-strands: Tyr7–Cys8, His19–Cys20, and Gly27–Ala30; Structure of Wr-AI1 determined by X-ray His19–Ala30 forms a b-hairpin (Fig. 3C). The crystallography b-strands are connected by four b-turns, two pointing Wr-AI1 showed a high propensity to form fiber-like towards the N-terminal and C-terminal ends on one precipitates at neutral pH. Crystals of Wr-AI1 suitable side of the molecule, and the other two towards the for X-ray crystallography were successfully obtained opposite side. This compact fold is also stabilized by after 1 day of incubation, and diffracted to 1.25-A res- an abundance of intramolecular hydrogen bonds olution at a synchrotron beamline. The complete pep- (Fig. 3C), as reported in other cysteine-rich peptides tide chain comprising 30 residues was unambiguously such as plant defensin PhD1 and cyclotide kalata B5 traced (Table 2 and Movie S1), clearly confirming the [8,9]. Moreover, the structure is devoid of N-terminal disulfide connectivity of CysI–IV, CysII–V and Cys- or C-terminal tails that would extend away from the III–VI as determined by NMR. In addition, the crystal CK core stabilized by three disulfide bonds. Approxi- structure of Wr-AI1 agreed with its solution structure mately 30% of the amide proton signals remained in ensemble (Fig. 3D), with an average rmsd of 0.93 A the 1D spectra after 18 h of H/D exchange in D2O. for backbone atoms. Minor differences were observed, These amide protons are identified as hydrogen bond mainly in the loop region and side chain orientations,
FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS 4353 Pseudocyclic cystine-knot a-amylase inhibitors P. Q. T. Nguyen et al.
Fig. 2. Tandem MALDI-TOF/TOF MS/MS profiles of two tryptic fragments (m/z 2072 and and m/z 1198) provide the full wrightide Wr-AI1 sequence. Ile/Leu assignment was determined from the genetic sequence and X-ray and NMR structures. including the disulfide bonds. This could be attribut- formation. The structure of Wr-AI1 also resembles able to the flexibility in solution of the loop region those of several spider toxins (Fig. 3F): Hainan tox- and side chains. This observation is consistent with the ins III and IV (PDB codes: 2JTB and 1RYV, respec- b overlapping of the chemical shifts of H Cys, which cre- tively), which are neuronal sodium channel inhibitors ated ambiguities in clearly defining the orientation of comprising 33 and 35 residues [10], and the GXTX-1E disulfide bonds by NMR. high-affinity tarantula toxin (PDB code: 2WH9), A systematic search for homologous structures which is a potassium channel inhibitor [11]. deposited in the Protein Data Bank (PDB) by use of Further analysis of the crystal structure unambigu- the DALI server (http://www.ebi.ac.uk/) returned four ously established the two prolins as cis-Pro17 and homologous structures with a Z-factor > 3.0 (Fig. 3A). trans-Pro23 (Fig. 4). In the structure of Wr-AI1, the The structure of Wr-AI1 is most similar to that of cis peptide bond between Asp16 and Pro17 causes a AAI (PDB code: 1QFD in its free form, and 1CLV as local backbone twist (Mobius-like structure similar to a complex with the a-amylase from the yellow meal- Mobius cyclotides). This energetically unfavorable worm): a superposition of 29 a-carbon atoms returns twist is partly stabilized by a strong hydrogen bond an rmsd of 1.10 A, with strict conservation observed between main chain atoms of Cys15 and Tyr18 for the inhibitor core and disulfide bridges. Variations (Table 3 shows the list of intramolecular hydrogen between the two structures are confined to the two bonds). Previous studies showed that the cis conforma- turns connecting the individual inhibitor strands that tion occurs at a higher frequency in X–Pro peptides, come into contact with the a-amylase upon complex where X is an aromatic residue [12]. This high
4354 FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS P. Q. T. Nguyen et al. Pseudocyclic cystine-knot a-amylase inhibitors
A phenol ring is 4.9 A. Thus, the interaction between the aromatic side chain and the pyrrolidone ring could contribute to the stabilization of cis-Pro17 form.
Modeled complex between Wr-AI1 and TMA The complex between TMA and AAI was previously BC characterized by Pereira et al., using X-ray crystallog- raphy [13]. In this structure, AAI inserts into a V-shaped crevice located at the interface of TMA domains A and B that forms the active site accom- modating the carbohydrate residues. A total of 18 residues of AAI are in contact (distance of < 4.0 A) with 24 residues of TMA. Among them, several resi- D E dues of the AAI inhibitor occupy or block the entrance to the six carbohydrate-binding subsites at the TMA active site cleft. These residues include Lys4–Arg7, Met12, Tyr27, and Tyr28, all of which are located at the hydrophilic accessible surface of the inhibitor molecule. In particular, Arg7 forms a salt bridge with the catalytic residue Asp287 from F TMA. This residue is also involved in a water-medi- ated hydrogen-bonding network with two other cata- lytic residues, Glu222 and Asp185, from TMA. To better understand how Wr-AI1 can inhibit the amy- lase activity of TMA, we built an atomic model for their interaction by using the AAI–TMA complex as a template (Fig. 5) and assuming an overall conserva- tion of the molecular orientation of the inhibitors in Fig. 3. The 3D structure of Wr-AI1. (A) Secondary structure the TMA active site pocket. With the exception of illustration for the sequence of Wr-AI1. The sequence of Wr-AI1 is Gly27 (Wr-AI1 numbering), no residue located at the aligned with sequences of: AAI, Hainan toxin III (HT-III), Hainan interface with the enzyme is strictly conserved toxin IV (HT-IV), and GxTX-1E Guangxiensis toxin 1E (GxTX-1E) (PDB codes: 1QFD, 2JTB, 1RYV, and 2WH9, respectively). b between Wr-AI1 and AAI. Upon complex formation, 2 stands for b-strand; yellow bridges indicate disulfide bonds; and the buried surface area is 1831 A , which is compara- 2 the red turn depicts the b-hairpin. (B) Solution structure of Wr-AI1 ble to that of the AAI–TMA complex (2085 A ) [13]. (PDB code: 2MAU). (C) Illustration of intramolecular hydrogen The network of interactions that stabilizes the Wr- bonds. (D) Backbone trace alignment of the crystal structure of Wr- AI1 complex is detailed in Table 3 and Fig. 6 AI1 (blue) (PDB code: 4BFH), with the 10 solution structure (Movie S2). Lys4, Glu6, Tyr7 and Thr21 from Wr- ensemble (tan). (E) Superposition of the solution structures of Wr- AI1 form hydrogen bonds with several negatively AI1 and AAI. (F) Structure alignment of Wr-AI1 with its structural homologs. The figure was prepared with PYMOL. charged residues from TMA (Fig. 6). This set of hydrogen bonds are preserved in > 95% of all frequency of occurrence is explained by the interaction configurations sampled along the last 250 ns of the between the aromatic side chain and the proline, which molecular dynamics simulation. The total DG is gives rise to ring-current-induced shifts for the cis con- 40.56 1.07 kcal mol 1 for this complex. An formers but not for the trans conformers in NMR analysis of the various components contributing to experiments. Here, we observed parallel stacking molecular complex stabilization gives the following between the phenol ring of Tyr18 and the pyrrolidine values: van der Waals, –88.88 0.93 kcal mol 1; ring of Pro17. Significant shifts from the average electrostatics, 138.44 4.47 kcal mol 1; and nonpolar d chemical shifts of H (2.39 with a reference solvation, –77.49 3.79 kcal mol 1. This analysis c average value of 3.63) and H (0.71 with a reference suggests that the enthalpic contribution of the associ- average value of 2.02) were also observed on the basis ation between TMA and Wr-AI1 is mainly driven by of the NMR assignments. The distance between the van der Waals interactions, with a smaller contribu- centers of the Pro17 pyrrolidine ring and the Tyr18 tion of electrostatic interactions.
FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS 4355 Pseudocyclic cystine-knot a-amylase inhibitors P. Q. T. Nguyen et al.
A C
Fig. 4. Ribbon diagram and structural features of Wr-AI1. (A and B) Stereoview
of the electron density (2Fobs Fcalc) and D schematic presentation of the backbone B twist caused by cis Pro17. The electron density was contoured at 1.0 r. (C and D) Pro17 and Pro23 adopt the cis and trans conformations, respectively. The three beta strands are shown in blue and disulfide bonds highlighted in yellow. The figure was prepared with the program PYMOL.
Heat and proteolytic stability Wr-AI2 had inhibitory activities against TMA in a dose-dependent manner, with IC values of 1.9 and To determine whether wrightides would resist degrada- 50 2.3 lM, respectively (Fig. 7D). Like AAI, Wr-AI1 and tion by boiling or protease treatment, which are Wr-AI2 did not inhibit a-amylases from fungus or important for administering decoctions in traditional mammals at concentrations up to 100 lM. medicines, Wr-AI1 was heated at 100 °C for 1 h or incubated with chymotrypsin or carboxypeptidase A for 4 h. More than 95% of Wr-AI1 remaining intact Biological activity of amylase inhibitors was observed at the same retention time in the ultra- The cytotoxic, hemolytic and antibacterial activities of performance liquid chromatography (UPLC) profiles Wr-AI1 and Wr-AI2 were tested. In our experiments, after heat treatment (Fig. 7A). The MS profiles of cor- wrightides did not show appreciable toxic, hemolytic responding peaks showed that both peaks contained or antibacterial activity at concentrations up to native Wr-AI1 (m/z 3246) with a small amount of 100 lM. degraded products. To confirm that the CK structure is important for its proteolytic stability, Wr-AI1 was fully reduced by Cloning of wrightide-encoding genes dithiothreitol, and served as the control in a chymo- Using 30-RACE and 50-RACE PCRs, we obtained the trypsin stability assay. A nine-residue linear peptide Wr-AI2 full-length gene from an RNA extract from a was used as the control in a carboxypeptidase A assay. Singapore plant. Subsequently, we used two primers The control peptides were almost completely hydro- derived from the 50-UTR and 30-UTR of a Wr-AI2 lyzed after 4 h of incubation with chymotrypsin or 1 h clone, and successfully amplified DNA sequences of of treatment with carboxypeptidase A at 37 °C. Under Wr-AI1, Wr-AI2, and a novel wrightide, Wr-AI3, similar conditions, the native peptide Wr-AI1 was which was not found at the protein level. resistant to protease degradation, with > 95% of pep- Figure 8 shows the deduced 87-residue precursors of tides remaining intact (Fig. 7B,C). Our results provide Wr-AI1 to Wr-AI3 and their alignment with previ- strong evidence for the stability of wrightides against ously characterized CK trypsin inhibitor and x-cono- thermal, endopeptidase and exopeptidase treatments. toxin precursors. In general, wrightide precursors contain a 21-residue endoplasmic reticulum (ER) sig- nal sequence followed by a 36-residue prodomain and a-Amylase inhibitory activity a 30-residue wrightide domain at the C-terminus. We performed inhibition assays with TMA and a-amy- Comparison between RACE and DNA PCRs showed lases from human saliva, porcine pancreas, and fungus that wrightide genes contain a phase 1 intron in the (Aspergillus oryzae), by using the Bernfeld method middle of the ER signal. The signal sequence and [14]. Preliminary results showed that both Wr-AI1 and prodomain of wrightide precursors are almost
4356 FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS P. Q. T. Nguyen et al. Pseudocyclic cystine-knot a-amylase inhibitors
ABC
DEF
Fig. 5. Cartoon and surface view of superimposition of AAI and Wr-AI1 in complex with TMA. (A) Model of the complex of AAI (magenta) and Wr-AI1 (cyan) with TMA (gray) derived from the AAI–TMA complex (PDB code: 1CLV). (B, C) Zoom-view of the binding region between AAI (B)/Wr-AI1 (C) and TMA. Residues of AAI/Wr-AI1 with atoms within 6 A of TMA are colored in magenta/cyan, and residues of TMA with atoms within 6 A of AAI/Wr-AI1 are colored in blue/red. Residues that are > 6 A away from TMA in both AAI and Wr-AI1 are colored orange. (D) Superimposition of AAI and Wr-AI1 at the active site of TMA. The three CysI–IV, CysII–V and CysIII–VI disulfide bonds arranged in a CK motif are highlighted in yellow. (E, F) Close-ups of AAI (E) and Wr-AI1 (F) residues at the TMA active site. The figure was prepared with PYMOL. identical, except for a three-residue difference in the as cyclotides. In wrightides, the N-terminus and C-ter- prodomain and several silent mutations at the gene minus are protected by disulfide bonds at the ultimate level, as highlighted in Fig. 8A. or penultimate residues. Our structural analysis showed that this arrangement enables the termini to Discussion loop back to the peptide chain via disulfide bonds like ‘pseudocyclics’, particularly at the N-terminus of In this study, we used proteomic, genomic and struc- wrightides. These pseudocylic CK (PCK) peptides, tural methods to characterize the 30-residue knottin- with or without one extra residue flanking the disul- type a-amylase inhibitors Wr-AI1 to Wr-AI3 from fide-looping terminus, would probably escape degrada- W. religiosa of the Apocynaceae family. Since the dis- tion by exopeptidases. covery of AAI in 1994, the 32-residue AAI has remained the only representative of the knottin group The backbone-twisting cis-proline in PCK that shows a-amylase inhibitory activity [5]. The dis- inhibitors covery of wrightides thus extends the list of the family of knottin a-amylase inhibitors. With two fewer resi- The presence of cis-proline in naturally occurring cys- dues than AAI, the wrightide family represents the teine-rich peptides generally causes a twist in the pep- smallest proteinaceous a-amylase inhibitors reported. tide backbone. This was used as a benchmark to Interestingly, these wrightides are resistant to both classify cyclotides into Mobius€ (with cis-proline) and heat denaturation and proteolytic degradation, includ- the bracelet (without cis-proline) subfamilies [4]. In this ing exopeptidase treatment. Thus, wrightides have the study, we found that the PCK wrightide Wr-AI1 also favorable stability features of cyclic CK peptides such contains one backbone-twisting cis-proline. Also, in
FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS 4357 Pseudocyclic cystine-knot a-amylase inhibitors P. Q. T. Nguyen et al.
study of local interactions that stabilize the cis-prolines in Wr-AI1 and other PCK a-amylase inhibitors may reveal diverse mechanisms of cis-proline formation in H189 D185 cis-proline-rich peptides. V151 D225 E6 Y11 Shape-fitting inhibition mechanism between PCK a-amylase inhibitors and TMA Y7 A model of the interaction between Wr-AI1 and TMA K4 was constructed on the assumption of overall conser- vation of molecular orientation in the TMA active site pocket as compared with AAI. Interestingly, despite D287 T21 the lack of sequence conservation between both pep- tide inhibitors, several side chains that project from the surfaces of the two inhibitors are placed in similar N331 positions in the active site crevice of TMA, and small movements would allow them to make equivalent con- tacts with the enzyme. Molecular dynamics study of the modeled complex suggests that Wr-AI1 binds to the TMA active site depression via an interaction net- Fig. 6. Stable hydrogen bond interactions between Wr-AI1 work composed largely of nonpolar interactions and (residues shown as sticks in cyan) and TMA (residues shown as completely lacking the critical salt bridge observed for green sticks) following molecular dynamics simulation (see text). AAI (Table 3). A crystal structure for the TMA–Wr- AI1 complex is needed to confirm this hypothesis. our unreported work, we found PCK peptides in three other Apocynaceae plants, each of which contained Wrightides follow the biosynthesis pathway for three or four prolines, and in four determined by secretory proteins NMR spectroscopy to have one or two cis-prolines. Together, these results suggested that the occurrence Our genetic analysis showed that wrightide precursors of cis-proline bond could be underestimated in pro- consist of an ER signal domain with a phase 1 intron, line-rich cysteine-rich peptides. a prodomain, and a single wrightide domain at the Surveys of protein databases revealed approximately C-terminus. The gene organization, starting with a 35% and 6–8% cis-proline in small polypeptides and signal peptide, provides hints on the biosynthesis native proteins, respectively [15]. The percentage of pathway of wrightides, which are gene-encoded and cis-proline amide bonds increases to as high as 12– ER-targeted following the conventional pathway for 16% when proline is preceded by an aromatic residue secretory proteins (Fig. 8B), as suggested for many in protein primary sequences. The steric repulsion cysteine-rich peptides [16]. The signal peptide is gener- between the pyrrolidine rings of a proline and the two ally removed by SPase I from the precursor to release neighboring Ca atoms generally renders the cis config- the propeptide. A single cleavage between the 36-resi- uration energetically less favorable than the trans con- due prodomain and the 30-residue functional domain figuration. In peptides with cis-proline preceded by an subsequently produces the native wrightide. These aromatic residue, clustering of the aromatic side chain characteristics distinguish wrightides as ribosomally and the pyrrolidine ring provides stability to the steri- synthesized peptides from smaller peptides of 5–12 cally constrained cis-proline, which is manifested in residues that are synthesized by nonribosomal multien- part by the selective ring-current-induced shifts of pro- zyme complexes [17,18]. line Ha and Hb in NMR spectroscopy [12]. Our analy- The three-domain precursor structure is commonly sis of the Wr-AI1 structure demonstrated the found in many CK peptides, both cyclic and linear. occurrence of one cis-proline in X–Pro amide bonds, Examples of such cyclic plant CK peptide precursors where X is a nonaromatic residue. The backbone twist include cyclotides from the Rubiaceae, Violaceae and caused by this cis-proline is probably stabilized by the Solanaceae families [17,19], and squash trypsin inhibi- hydrogen bond between the neighboring residues tors from Momordica cochinchinensis [20]; selected Cys15 and Tyr18 rather than by direct stacking of the examples of linear plant CK peptide precursors are preceding aromatic side chain and proline. Thus, the acyclic cyclotides from the Violaceae, Rubiaceae and
4358 FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS P. Q. T. Nguyen et al. Pseudocyclic cystine-knot a-amylase inhibitors
A
B C
D
Fig. 7. Stabilities and a-amylase inhibitory activity of wrightides. (A) Thermal stability of Wr-AI1 and Wr-AI2. The minor peak after heat treatment contained mainly Wr-AI1/Wr-AI2, with a small amount of degraded products, as determined from the MS profile. (B) Chymotrypsin stability of wrightide Wr-AI1. (C) Carboxypeptidase A stability of Wr-AI1. (D) Inhibition of T. molitor a-amylase by wrightides. Peptides were preincubated with TMA for 20 min at 37 °C. Hydrolysis was started by addition of 1% starch. The reaction was allowed to proceed for 5 min, and stopped by addition of a color reagent containing 3,5-dinitrosalicylic acid. The IC50 values are 1.9 and 2.3 lM for Wr- AI1 and Wr-AI2, respectively. The error bars show standard deviations.
Poaceae families [21–23], and towel gourd trypsin blockers: x-conotoxins [25] and d-conotoxins [26]. It inhibitors [24]. This precursor organization is also used should be noted that, within the plant kingdom, by animals such as cone snails to produce ion channel diverse structures are used to organize CK peptide
FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS 4359 Pseudocyclic cystine-knot a-amylase inhibitors P. Q. T. Nguyen et al.
Fig. 8. Precursor structure of Wr-AI wrightides. (A) Alignment of wrightide precursors with precursors of other CK peptides, including x-conotoxin SO-3 and towel gourd trypsin inhibitor TGTI-II. The ER signal peptide was assigned to x-conotoxin SO-3 by SIGNALP 3.0, whereas this domain was not recognized by SIGNALP 3.0 for TGTI-II. (B) Secretory protein synthesis pathway for wrightides. precursors. One example is provided by precursors A potential application of considerable interest in with multiple repeats of cyclic or linear CK peptides, drug development is the engineering of wrightides to such as those reported for cyclotides from the Rubia- be orally active mammalian a-amylase inhibitors for ceae and Violaceae families [17,19], and TIPTOP the treatment of obesity and type 2 diabetes mellitus. squash trypsin inhibitors [20]. Other examples are chi- The literature shows that extended hydrophobic inter- meric precursors of both CK peptides and other types actions could be important for AAI inhibition of of protein, such as cliotide precursors of both cyclo- mammalian a-amylases [13]. Human salivary a-amy- tides and legume albumin PA1a in Clitoria ternatea lase (HSA) (PDB code: 3DHP) and TMA share high [27,28]. Given such a diversity of precursor organiza- sequence homology (65%) and structural homology tion even within a CK peptide family, understanding (Z-score of 57.1 with 468 equivalent residues at an the genetic sequences of each CK peptide family, here rmsd of 1 A, by the DALI server). Superimposing the PCK a-amylase inhibitors, is thus beneficial for their HSA–Wr-AI1 complex on the TMA–Wr-AI1 complex applications in crop protection, and also provides reveals four additional loops present in HSA at the insights into their biosynthesis. interface of the active site, including loops Asn53– Phe55, Asn137–Gly146, Gly304–Ala310, and Trp344– Val358. The conformational flexibility of these loops Knottin-type a-amylase inhibitors with might be responsible for the low-affinity binding of applications in engineering peptidyl bioactives Wr-AI1 to HSA. Our docking experiments suggested The CK structure has been employed in nature as a that it is possible that careful incorporation of aro- scaffold for a variety of unrelated protein families matic and positively charged residues into wrightide found in microorganisms, animals, and plants. In par- templates could improve their contact with the nega- ticular, a-amylase inhibitors adopting CK folds are tively charged enzyme active sites, to make wrightides small, extraordinarily stable against heat and endopep- active against mammalian a-amylases. In this regard, tidase and exopeptidase degradation, and highly toler- our work showing the interaction of Wr-AI1 and ant to sequence variation [29]. Thus, small CK TMA provides new insights for a structure-guided peptides such as wrightides with molecular masses of approach to designing potentially useful orally active 3–5 kDa possess appealing features as potential pep- a-amylase inhibitors for managing type 2 diabetes tide therapeutics [30,31]. First, the small size makes mellitus and obesity. wrightides more amenable to chemical synthesis [32,33]. Second, the CK peptides in general are highly Experimental procedures tolerant to sequence variations and the spacing of the half-cystines, allowing a-amylase inhibitors to poten- Isolation of a-amylase inhibitors tially serve as scaffolds for protein engineering to attain new functions, such as in the successful grafting W. religiosa leaves (800 g) were homogenized and extracted of the bradykinin antagonist peptides DALK or DAK twice in 50% (v/v) ethanol. After centrifugation onto the cyclotide kalata B1 scaffold [34]. (8500 g, 10 min), the supernatant was partitioned with
4360 FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS P. Q. T. Nguyen et al. Pseudocyclic cystine-knot a-amylase inhibitors dichloromethane. The aqueous upper layer was concentrated, During the structure calculation, hydrogen bond restraints filtered, and loaded onto a C18 flash column (Grace Vydac, of 1.8–2.2 A for the NH–O distance and 2.2–3.2 A for the Hesperia, CA, USA). Elution was performed with increasing N–O distance were applied on nine identified hydrogen concentrations of ethanol. The presence of cysteine-rich pep- bonds according to the slowly exchanging amide protons. tides in all fractions was monitored by MALDI-TOF MS. To Φ angles were constrained to the range of –150° to –90° for 3 purify individual peptides, several dimensions of strong cation JHN-Ha > 8 Hz. Structures were displayed and analyzed exchange and RP-HPLC were employed. with PYMOL and PROCHECK-NMR, respectively [40]. The experimental and structural statistics are summarized in Table 1. De novo sequencing with MALDI-TOF MS/MS l Approximately 40 g of each purified peptide was dissolved Crystal structure determination in 50 mM ammonium bicarbonate buffer (pH 7.8) contain- ing 50 mM dithiothreitol, and incubated at 37 °C for 2 h. With the sitting drop vapor diffusion method, the native Digestion with endoproteinase Glu-C, trypsin or chymo- crystals were obtained from the mixture of 1 lL of Wr-AI1 trypsin was carried out at room temperature for 5 min, and solution (4.8 mg mL 1) and 1 lL of precipitant solution this was followed by MALDI-TOF MS/MS sequencing, as (3.6 M sodium formate, 10% glycerol) after 1 day of incu- previously described [27]. Isobaric residues were assigned bation at 16 °C. The crystals were stabilized in the precipi- on the basis of gene sequences for Wr-AI1 and Wr-AI2, tant solution supplemented with 40% (v/v) glycerol, and and confirmed on the basis of the X-ray or NMR structure flash-frozen in liquid nitrogen. Diffraction intensities to for Wr-AI1. Wr-AI3 was sequenced only at the gene level. 1.25-A resolution were collected at 100 K at the Swiss Light Source Beamline PXIII with a Pilatus 6M detector Solution structure determination with NMR (Dectris, Baden, Switzerland). Integration, scaling and merging of intensities were carried out with XDS [41] and spectroscopy SCALA [42] from the CCP4 suite [43]. Data collection statistics The NMR sample was prepared by dissolving lyophilized are summarized in Table 2.
Wr-AI1 in 95% H2O/5% D2O or 99.9% D2O directly The structure was determined by molecular replacement (~ 1mM protein and pH/pD 3.3). All NMR experiments with PHASER [44]. The search probe was the structure of were carried out on a Bruker 600-MHz NMR spectrometer AAI (PDB code: 1CLV [13]). ARP-WARP [45] was used for equipped with a cryogenic probe. Two-dimensional (2D) chain tracing and map improvement, and the resulting TOCSY and NOESY experiments were performed with mix- Table 1. NMR experimental and structural statistics of Wr-AI1. ing times of 80 and 200 ms, respectively [35]. The 2D data were acquired at 298 K. Water suppression was achieved NOE constraints 681 with modified WATERGATE pulse sequences [36]. The Intraresidue (|i j| = 0) 332 Sequential (|i j| = 1) 193 NMR spectra were processed with NMRPIPE [37]. The amides Medium-range (1 < |i j| < 5) 36 involved in hydrogen bonding were identified by hydrogen– Long-range (|i j| ≥ 5) 120 deuterium exchange (1D 1H) experiments [38]. Dihedral angle restraints 10 Sequence-specific assignments were achieved with 2D Hydrogen bonds 7 TOCSY and NOESY, and NOEs were assigned from 2D PROCHECK-NMR Ramachandran plot (%) NOESY results with NMRSPY (http://yangdw.sci Most favored region 70.9 ence.nus.edu.sg/Software&Scripts/NMRspy/index.htm). The Additionally allowed region 28.7 chemical shifts are deposited in BioMagResBank (accession Generously allowed region 0.4 number: 18983). Distance restraints were derived from the Disallowed region 0 φ Average maximum violations per structure peak intensities of the assigned NOEs. Dihedral angles ( ) 3 Distance (A) 0.02 0.002 were obtained from JHN-Ha coupling constants measured Van der waals ( A) 3.7 0.4 from the 1D 1H-spectrum. Hydrogen bond restraints were Torsion angles (°) 0.25 0.11 incorporated on the basis of the observation of amide pro- 2 1 CYANA target function value (A ) 1.35 0.15 tons in the 1D H-spectra recorded after resuspension of the Average rmsd to mean structure ( A) ° lyophilized Wr-AI1 in D2O for up to 18 h at 25 C. All backbone atoms (1–30) 0.38 0.08 Structure was calculated with a simulated annealing All heavy atoms (1–30) 0.85 0.12 approach with CYANA 2.0 [39]. Distance restraints are divided into three classes: 1.8 < d < 3.4 A (strong NOEs), 1.8 < d < 4.2 A (medium NOEs), and 1.8 < d < 5.5 A model was corrected manually (Table 2). The Ramachan- (weak NOEs). Disulfide bond restraints of 2.0 < d dran plot calculated with PROCHECK [46] revealed that 87% c c b c (S i,Sj) < 2.1 A, 3.0 < d(C i,Sj) < 3.1 A and 3.0 < d of the residues were in the most favored region and 13% c b (S i,Cj) < 3.1 A were employed for structure calculation. were in the additional allowed region.
FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS 4361 Pseudocyclic cystine-knot a-amylase inhibitors P. Q. T. Nguyen et al.
Table 2. Crystallography data collection and refinement statistics Before the dynamic simulations, the solvated system was of Wr-AI1. relieved of any unfavorable interactions by subjecting it Data collection to 100 steps of energy minimization. Harmonic restraints a Space group during the equilibration were placed on C atoms with a 1 2 Cell parameters P212121 1 kcal mol A to the energy-minimized coordinates. The a, b, c (A) a = 16.19, b = 29.13, c = 47.70 system was heated to 300 K in steps of 100 K, and this a, b, c (°) a = b = c = 90 was followed by gradual removal of the positional – – Resolution range (A) 29.1 1.25 (1.28 1.25) restraints and a 10-ns unrestrained equilibration at Observed reflections 21 115 (2864) 300 K. Analysis of the resulting trajectories revealed that Unique reflections 6362 (856) the simulated complex reached stability after 50 ns, with Completeness (%) 95.4 (90.3) an RMSD of < 1.8 A. The first 10 ns of simulation were Multiplicity 3.3 (3.3) b performed in NPT, and the production run of 500 ns was Rmerge 0.029 (0.071) Mean I/r(I) 23.2 (11.9) performed in NVT. The simulation temperature of 300 K Refinement was set with Langevin dynamics, with a collision fre- Resolution range ( A) 24.88–1.25 (1.28–1.25) quency of 0.1 ps 1. The pressure was maintained at Number of reflections 6017 (410) 1 atm by the use of weak coupling with a pressure relax- used for refinement ation time of 1 ps. During the simulation, all long-range Number of reflections used 317 (21) electrostatic interactions were treated with particle mesh for Rfree calculation (5%) Ewald methods [49], with a real space cut-off distance of R (%)c, R (%)d 15.8, 15.9 factor free 9 A. Bonds involving hydrogen atoms were constrained Number of nonhydrogen atoms 223 with the M-SHAKE algorithm [50]. A time step of 4 fs with Number of water molecules 41 Mean B-factor ( A2) hydrogen mass repartitioning was used, and coordinates Protein whole chain 5.4 were saved every 100 ps. Hydrogen bond analysis was Water 32.9 performed with the PTRAJ module in AMBER for the last rmsd from ideality 250 ns of the stabilized trajectory, with a cut-off distance Bond lengths (A) 0.020 of 3.5 A. Binding energy analysis based on molecular Bond angles (°) 1.85 mechanics/generalized Born surface area [51] was per- Ramachandran plot (%) formed on the simulated trajectory to calculate the free Most favored regions 87 energy of binding of Wr-AI1 to TMA. For binding Additional allowed regions 13 energy calculation, a total of 100 structures were G-factore 0.07 extracted at regular intervals from the last 250 ns of the a The numbers in parentheses refer to the last (highest) resolution trajectory. A salt concentration of 150 mM and a Born b shell. Rmerge = ΣhΣi|Ihi
Molecular docking and molecular dynamics study Purified Wr-AI1 was heated in boiling water for 1 h and of the Wr-AI1–TMA complex then subjected to UPLC. Wr-AI1 without heat treatment To investigate the stability of the TMA–Wr-AI1 complex was used as a control. Peaks collected from UPLC were obtained from docking (which was initially obtained by monitored by MALDI-TOF MS. simply superimposing Wr-AI1 onto AAI in the TMA– AAI crystal structure), we performed three molecular Proteolytic stability test dynamics simulation of 500 ns each, using ACEMD [47] and all-atom ff12SB forcefield parameters. Hydrogen Purified Wr-AI1 was incubated with or without chymotryp- atoms were added to this initial complex with the XLEAP sin (at a final peptide/enzyme ratio of 10 : 1 mol/mol) in module of AMBER [48]. The system was solvated with 20 mM ammonium bicarbonate (pH 7.8) at 37 °C for 4 h. TIP3P water molecules to form a box with at least 10 A Purified Wr-AI1 that had been completely reduced with separating the solute atoms and the edge of the box. A 50 mM dithiothreitol (2 h, 37 °C) was treated in the same total of 92 sodium ions and 70 chloride ions, correspond- way, and used as a control. Treated samples or controls ing to a salt concentration of 150 mM, were added to the were subjected to UPLC, and the collected peaks were system by replacing water molecules random positions. monitored by MALDI-TOF MS.
4362 FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS P. Q. T. Nguyen et al. Pseudocyclic cystine-knot a-amylase inhibitors
Table 3. Potential intramolecular hydrogen bonds in Wr-AI1 and intermolecular interactions between PCK inhibitors and TMA (distance of < 3.4 A). Intramolecular distances were determined by X-ray crystallography. Intermolecular distances between AAI and TMA were derived from the crystal complex of AAI and TMA (PDB code: 1CLV), and the distances between Wr-AI1 and TMA were calculated from the molecular dynamics simuation of the TMA–Wr-AI1 complex for the last 250 ns of the trajectory.
Wr-AI1 Wr-AI1
Residue/atom Residue/atomDistance ( A) Residue/atom Residue/atom Distance ( A)
A2 N Q13 O 2.91 S9 OG L12 N 3.06 A2 O C15 N 2.84 V10 N P23 O 2.97 Q3 N E6 OE1 2.87 C15 O Y18 N 2.80 Q3 O G5 N 3.30 C15 O P17 N 3.31 Q3 O E6 N 3.00 P17 O H19 NE2 3.21 K4 NZ A30 OXT 2.88 H19 N A30 O 2.91 G5 N C29 O 2.76 H19 O A30 N 3.01 E6 O C29 N 2.93 C20 O Q22 N 3.35 C8 N G27 O 2.99 T21 N I28 O 2.92 C8 O G27 N 3.01 T21 OG1 I28 N 3.01 S9 N L12 O 3.02 Q22 O G26 N 2.79 S9 O Y11 N 3.15 Q22 O G27 N 3.00 S9 O L12 N 3.32 Q22 OE1 V24 N 2.94 S9 O L12 O 3.03 Q22 OE1 I25 N 3.01 S9 OG Y11 N 3.06
AAI TMA Wr-AI1 Residue/atomDistance ( A) Residue/atomDistance ( A) Residue/atom Occupancy (%)
C1 N 2.80 N137 O K4 NZ 2.73 Q295 OE1 K4 NZ 3.27 Q295 NE1 N6 OD1 2.81 K188 NZ D13 O 2.93 E135 OE1 T24 O 3.18 N331 ND2 D332 OD2 S25 OG 2.76 D332 OD1 N30 ND2 3.19 D287 O N30 ND2 3.06 R290 O S32 OG 3.09 G292 N V151 O 2.75 Y11 OH 49.89 D185 OD2 2.80 Y7 OH 99.09 H189 NE2 2.75 E6 OE1 97.04 D225 O 2.81 K4 NZ 99.07 D225 OD1 2.81 K4 NZ 48.70 D287 OD2 2.64 T21 OG1 98.44
In the carboxypeptidase A stability assay, Wr-AI1 was (in 20 mM sodium phosphate buffer, pH 6.7) for 5 min. incubated with or without enzyme (at a final peptide/enzyme Color reagent (3,5-dinitrosalicylic acid and sodium ratio of 40 : 1 mol/mol) in 50 mM NaCl/Tris and 1 M NaCl potassium tartrate; Sigma, St. Louis, MO, USA) [56] was dis- (pH 7.5) at room temperature for up to 24 h. A linear nine- pensed into each well, and color was allowed to develop for residue peptide was used as a control. Degradation products 20 min at 100 °C. Absorbance at 540 nm was read to deter- were monitored by UPLC and MALDI-TOF MS. mine the a-amylase activity. Similar inhibition experiments were performed for human salivary, porcine pancreatic and A. oryzae a-amylases (Sigma). Assay for a-amylase activity a -Amylase was isolated from lavvae of the yellow mealworm, Hemolysis assay T. molitor, with the procedure described previously [55]. Assays for a-amylase were carried out in 96-well plates with Fresh type AB blood was donated by a healthy volunteer. the Bernfeld method [14]. TMA with or without treatment The hemolysis assay was performed as described elsewhere with peptides (20 min, 37 °C) was incubated with 1% starch [27].
FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS 4363 Pseudocyclic cystine-knot a-amylase inhibitors P. Q. T. Nguyen et al.
Cytotoxicity assay Author contributions The cytotoxicity of the purified wrightides was tested with P. Q. T. Nguyen performed or was involved in all of PrestoBlue Cell Viability Reagent (Invitrogen, Carlsbad, the experiments except for NMR experiments. CA, USA). African green monkey kidney (Vero) cells S. Wang performed NMR spectroscopy experiments seeded onto 96-well plates were incubated with Wr-AI1 and and sequence calculation. A. Kumar analyzed molecu- Wr-AI2 at 1–100 lM for 24 h at 37 °C. After incubation, lar dynamics results and enzyme alignment. L. J. Yap the PrestoBlue reagent (Invitrogen, Carlsbad, CA, USA) performed X-ray crystallization experiments. T. T. ° was dispensed into the wells and left at 37 C for 2 h. The Luu contributed to peptide and enzyme extraction. J. fluorescence was subsequently read as instructed by the Lescar analyzed X-ray crystallography data, built the manufacturer. Triton X-100 solution (1%) was used as a structure, and modeled the complex. J. P. Tam ana- positive control. lyzed the data. P. Q. T. Nguyen, J. Lescar and J. P. Tam contributed mainly to manuscript preparation. Antibacterial assay All of the authors discussed the results and contrib- uted to the writing of the manuscript. The antibacterial activity of wrightides was assessed with a radial diffusion assay, as described previously [57], on Gram-negative Escherichia coli (FDA strain Seattle 1946) References and Gram-positive Staphylococcus aureus. D4R, an in- 1 Cheek S, Krishna SS & Grishin NV (2006) Structural house peptide dendrimer with potent antibacterial activity, classification of small, disulfide-rich protein domains. was used as a positive control. The experiments were per- J Mol Biol 359, 215–237. formed in duplicate. 2 Pallaghy PK, Nielsen KJ, Craik DJ & Norton RS (1994) A common structural motif incorporating a Cloning of a-amylase inhibitor genes cystine knot and a triple-stranded beta-sheet in toxic and inhibitory polypeptides. Protein Sci 3, 1833–1839. Total RNA extraction was performed with the PureLink 3 Le Nguyen D, Heitz A, Chiche L, Castro B, Boigegrain Mini RNA purification kit (Invitrogen, Carlsbad, CA, USA), RA, Favel A & Coletti-Previero MA (1990) Molecular with addition of 3% 2-mercaptoethanol and 4% polyvinyl- recognition between serine proteases and new bioactive pyrrolidone to the lysis buffer. A total RNA extract of Singa- microproteins with a knotted structure. Biochimie 72, pore W. religiosa leaves was subsequently converted to 431–435. 30-RACE and 50-RACE cDNA libraries with the 30-RACE 4 Craik DJ, Daly NL, Bond T & Waine C (1999) Plant System for Rapid Amplification of cDNA Ends (Invitrogen) cyclotides: a unique family of cyclic and knotted and the SMARTer RACE cDNA Amplification Kit (Clon- proteins that defines the cyclic cystine knot structural tech, Takara Biotechnology, Dalian, China), respectively. motif. J Mol Biol 294, 1327–1336. 30-RACE PCR products obtained with the degenerate primer 5 Chagolla-Lopez A, Blanco-Labran A & Patthy A (1994) targeting the sequence CAQKGE (50-TGTGCTCAr- A novel a-amylase inhibitor from amaranth (Amaranthus AArGGnGA-30) were gel-purified, cloned into pGEM-T hypocondriacus) seeds. J Biol Chem 269, 23675–23680. Easy Vector (Promega Madison, WI, USA), and sequenced. 6 Le Berre-Anton V, Bompard-Gilles C, Payan F & A reverse primer based on the newly obtained partial Rouge P (1997) Characterization and functional sequence was designed to reveal the remaining encoding gene properties of the alpha-amylase inhibitor (alpha-AI) in 50-RACE PCR. To determine the DNA sequences of from kidney bean (Phaseolus vulgaris) seeds. Biochim wrightide genes, we performed PCR on the W. religiosa Biophys Acta 14,31–40. DNA extract with two primers: Wr2speF (50-TAG- 7 Svensson B, Fukuda K, Nielsen PK & Bønsager BC GCGCAAACAACATGGCTAAGC-30) and Wr2speR (50- (2004) Proteinaceous alpha-amylase inhibitors. Biochim CCACATAGCTCG-TAGAACAAGCTTACAG-30). The Biophys Acta 1696, 145–156. ER signal peptides were predicted with SIGNALP 3.0 (http:// 8 Plan MR, Rosengren KJ, Sando L, Daly NL & Craik www.cbs.dtu.dk/services/-SignalP-3.0/). DJ (2010) Structural and biochemical characteristics of the cyclotide kalata B5 from Oldenlandia affinis. Pept Acknowledgements Sci 94, 647–658. 9 Janssen BJC, Schirra HJ, Lay FT, Anderson MA & We thank P. Q. T. Nguyen, C. H. Teo and Y. S. Lam Craik DJ (2003) Structure of Petunia hybrida for technical assistance with this project. This research defensin 1, a novel plant defensin with five disulfide was supported in part by the Competitive Research bonds. Biochemistry 42, 8214–8222. Grant from the National Research Foundation in Sin- 10 Li D, Xiao Y, Xu X, Xiong X, Lu S, Liu Z, Zhu Q, gapore (NRF-CRP8-2011-05). Wang M, Gu X & Liang S (2004) Structure–activity
4364 FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS P. Q. T. Nguyen et al. Pseudocyclic cystine-knot a-amylase inhibitors
relationships of hainantoxin-iv and structure variation and the implications for structure, function determination of active and inactive sodium channel and stability. Biochem J 400,1–12. blockers. J Biol Chem 279, 37734–37740. 23 Nguyen GK, Lian Y, Pang EW, Nguyen PQ, Tran TD 11 Lee S, Milescu M, Jung HH, Lee JY, Bae CH, Lee & Tam JP (2013) Discovery of linear cyclotides in CW, Kim HH, Swartz KJ & Kim JI (2010) Solution monocot plant Panicum laxum of Poaceae family structure of GxTX-1E, a high-affinity tarantula toxin provides new insights into evolution and distribution of interacting with voltage sensors in Kv2.1 potassium cyclotides in plants. J Biol Chem 288, 3370–3380. channels. Biochemistry 49, 5134–5142. 24 Ling MH, Qi HY & Chi CW (1993) Protein, cDNA, 12 Wu WJ & Raleigh DP (1998) Local control of peptide and genomic DNA sequences of the towel gourd conformation: stabilization of cis proline peptide bonds by trypsin inhibitor. A squash family inhibitor. J Biol aromatic proline interactions. Biopolymers 45,381–394. Chem 268, 810–814. 13 Pereira PJ, Lozanov V, Patthy A, Huber R, Bode W, 25 Colledge CJ, Hunsperger JP, Imperial JS & Hillyard Pongor S & Strobl S (1999) Specific inhibition of insect DR (1992) Precursor structure of omega-conotoxin alpha-amylases: yellow meal worm alpha-amylase in GVIA determined from a cDNA clone. Toxicon 30, complex with the amaranth alpha-amylase inhibitor at 1111–1116. 2.0 A resolution. Structure 7, 1079–1088. 26 Woodward SR, Cruz LJ, Olivera BM & Hillyard DR 14 Bernfeld P (1955) Amylases a and b. Methods Enzymol (1990) Constant and hypervariable regions in conotoxin 1, 149–158. propeptides. EMBO J 9, 1015–1020. 15 Milner-White EJ, Bell LH & Maccallum PH (1992) 27 Nguyen GKT, Zhang S, Nguyen NTK, Nguyen PQT, Pyrrolidine ring puckering in cis and trans-proline Chiu MS, Hardjojo A & Tam JP (2011) Discovery and residues in proteins and polypeptides. Different puckers characterization of novel cyclotides originated from are favoured in certain situations. J Mol Biol 228, 725– chimeric precursors consisting of albumin-1 chain a and 734. cyclotide domains in the Fabaceae family. J Biol Chem 16 Mergaert P, Nikovics K, Kelemen Z, Maunoury N, 286, 24275–24287. Vaubert D, Kondorosi A & Kondorosi E (2003) A 28 Poth AG, Colgrave ML, Lyons RE, Daly NL & Craik novel family in Medicago truncatula consisting of more DJ (2011) Discovery of an unusual biosynthetic origin than 300 nodule-specific genes coding for small, for circular proteins in legumes. Proc Natl Acad Sci secreted polypeptides with conserved cysteine motifs. USA 108, 10127–10132. Plant Physiol 132, 161–173. 29 Norton RS & Pallaghy PK (1998) The cystine knot 17 Jennings C, West JL, Waine C, Craik DJ & Anderson structure of ion channel toxins and related MZ (2001) Biosynthesis and insecticidal properties of polypeptides. Toxicon 36, 1573–1583. plant cyclotides: the cyclic knotted proteins from 30 Tam JP & Lu Y-A (1997) Synthesis of large cyclic Oldenlandia affinis. Proc Natl Acad Sci USA 98, cystine-knot peptide by orthogonal coupling strategy 10614–10619. using unprotected peptide precursor. Tetrahedron Lett 18 Marahiel MA (2009) Working outside the protein- 38, 5599–5602. synthesis rules: insights into non-ribosomal peptide 31 Taichi M, Hemu X, Qiu Y & Tam JP (2013) A synthesis. J Pept Sci 15, 799–807. thioethylalkylamido (TEA) thioester surrogate in the 19 Dutton JL, Renda RF, Waine C, Clark RJ, Daly NL, synthesis of a cyclic peptide via a tandem acyl shift. Jennings CV, Anderson MA & Craik DJ (2004) Org Lett 15, 2620–2623. Conserved structural and sequence elements implicated 32 Wong CTT, Taichi M, Nishio H, Nishiuchi Y & Tam in the processing of gene-encoded circular proteins. JP (2011) Optimal oxidative folding of the novel J Biol Chem 279, 46858–46867. antimicrobial cyclotide from Hedyotis biflora requires 20 Mylne JS, Chan LY, Chanson AH, Daly NL, Schaefer high alcohol concentrations. Biochemitry 50, 7275–7283. H, Bailey TL, Nguyencong P, Cascales L & Craik DJ 33 Tam JP, Lu Y-A & Yu Q (1999) Thia zip reaction for (2012) Cyclic peptides arising by evolutionary synthesis of large cyclic peptides: mechanisms and parallelism via asparaginyl-endopeptidase-mediated applications. J Am Chem Soc 121, 4316–4324. biosynthesis. Plant Cell 24, 2765–2778. 34 Wong CTT, Rowlands DK, Wong C-H, Lo TWC, 21 Nguyen GK, Zhang S, Wang W, Wong CT, Nguyen Nguyen GKT, Li H-Y & Tam JP (2012) Orally active NT & Tam JP (2011) Discovery of a linear cyclotide peptidic bradykinin b1 receptor antagonists engineered from the bracelet subfamily and its disulfide mapping from a cyclotide scaffold for inflammatory pain by top-down mass spectrometry. J Biol Chem 286, treatment. Angew Chem Int Ed 51, 5620–5624. 44833–44844. 35 Kumar A, Ernst RR & Wuthrich K (1980) A two- 22 Ireland DC, Colgrave ML & Craik DJ (2006) A novel dimensional nuclear Overhauser enhancement suite of cyclotides from Viola odorata: sequence (2D NOE) experiment for the elucidation of complete
FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS 4365 Pseudocyclic cystine-knot a-amylase inhibitors P. Q. T. Nguyen et al.
proton–proton cross-relaxation networks in biological potential functions for simulating liquid water. J Chem macromolecules. Biochem Biophys Res Commun 95,1–6. Phys 79, 926–935. 36 Piotto M, Saudek V & Sklenar V (1992) Gradient- 49 Darden T, York D & Pedersen L (1993) Particle mesh tailored excitation for single-quantum NMR spectros- Ewald – an N.Log(N) method for Ewald sums in large copy of aqueous solutions. J Biomol NMR 2, 661–665. systems. J Chem Phys 98, 10089–10092. 37 Delaglio F, Grzesiek S, Vuister GW, Zhu G, Pfeifer J 50 Krautler V, Van Gunsteren WF & Hunenberger PH & Bax A (1995) NMRPipe: a multidimensional spectral (2001) A fast SHAKE: algorithm to solve distance processing system based on UNIX pipes. J Biomol constraint equations for small molecules in molecular NMR 6, 277–293. dynamics simulations. J Comput Chem 22, 501–508. 38 Saether O, Craik DJ, Campbell ID, Sletten K, Juul J & 51 Bashford D & Case DA (2000) Generalized Born Norman DG (1995) Elucidation of the primary and models of macromolecular solvation effects. Annu Rev three-dimensional structure of the uterotonic Phys Chem 51, 129–152. polypeptide kalata B1. Biochemistry 34, 4147–4158. 52 Onufriev A, Bashford D & Case DA (2000) 39 Guntert P, Mumenthaler C & Wuthrich K (1997) Modification of the generalized Born model suitable for Torsion angle dynamics for NMR structure calculation macromolecules. J Phys Chem B 104, 3712–3720. with the new program DYANA. J Mol Biol 273, 53 Hubbard SJ & Thornton JM (1993) NACCESS. 283–298. Department of Biochemistry and Molecular Biology, 40 Laskowski RA, Rullmannn JA, MacArthur MW, University College, London. Kaptein R & Thornton JM (1996) AQUA and 54 Humphrey W, Dalke A & Schulten K (1996) VMD: PROCHECK-NMR: programs for checking the quality visual molecular dynamics. J Mol Graph Model 14,33–38. of protein structures solved by NMR. J Biomol NMR 55 Strobl S, Gomis-Ruth€ F-X, Maskos K, Frank G, 8, 477–486. Huber R & Glockshuber R (1997) The a-amylase from 41 Kabsch W (2001) Integration, scaling, space-group the yellow meal worm: complete primary structure, assignment and post refinement. In International Tables crystallization and preliminary X-ray analysis. FEBS for Crystallography, Volume F: Crystallography of Lett 409, 109–114. Biological Macromolecules (Rossmann MG & Arnold 56 Miller GL (1959) Use of dinitrosalicylic acid reagent E, eds), pp. 218–225. Springer, Dordrecht, Netherlands. for determination of reducing sugar. Anal Chem 31, 42 Evans P (2006) Scaling and assessment of data quality. 426–429. Acta Crystallogr D 62,72–82. 57 Lehrer RI, Rosenman M, Harwig SSSL, Jackson R & 43 Winn MD, Ballard CC, Cowtan KD, Dodson EJ, Eisenhauer P (1991) Ultrasensitive assays for Emsley P, Evans PR, Keegan RM, Krissinel EB, Leslie endogenous antimicrobial polypeptides. J Immunol AG, McCoy A et al. (2011) Overview of the CCP4 Methods 137, 167–173. suite and current developments. Acta Crystallogr D 67, 235–242. 44 McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn Supporting information MD, Storoni LC & Read RJ (2007) Phaser crystallo- Additional supporting information may be found in graphic software. J Appl Crystallogr 40, 658–674. the online version of this article at the publisher’s web 45 Langer G, Cohen SX, Lamzin VS & Perrakis A (2008) site: Automated macromolecular model building for X-ray Movie S1.Stick representation of Wr-AI1 with 2Fo Fc crystallography using ARP/wARP version 7. Nat map overlaid. The complete chain was unambiguously 3 – Protoc , 1171 1179. traced, and contains three disulfide bonds in a 46 Laskowski RA, MacArthur MW, Moss DS & cystine-knot motif. Thornton JM (1993) PROCHECK: a program to check Movie S2. Movie of molecular dynamics simulation the stereochemical quality of protein structures. J Appl trajectory of TMA bound to Wr-AI1. The TMA pro- Crystallogr 26, 283–291. tein (white) and Wr-AI1 (blue) are shown as ‘car- 47 Harvey MJ, Giupponi G & De Fabritiis G (2009) toons’. Interacting residues from TMA (Asp225, ACEMD: accelerating biomolecular dynamics in the microsecond time scale. J Chem Theory Comput 5, His189, Asp185, Val151, and Asp287) and Wr-AI1 1632–1639. (Lys4, Glu6, Tyr7, Tyr11, and Thr21) are shown as 48 Jorgensen WL, Chandrasekhar J, Madura JD, Impey sticks. See also Table 3. For clarity, hydrogen atoms RW & Klein ML (1983) Comparison of simple are not shown.
4366 FEBS Journal 281 (2014) 4351–4366 ª 2014 FEBS