Characterizing circular peptides in mixtures: sequence fragment assembly of cyclotides from a violet plant by MALDI-TOF/TOF mass spectrometry
- First Online:
- 1.5k Downloads
Cyclotides are a very abundant class of plant peptides that display significant sequence variability around a conserved cystine-knot motif and a head-to-tail cyclized backbone conferring them with remarkable stability. Their intrinsic bioactivities combined with tools of peptide engineering make cyclotides an interesting template for the design of novel agrochemicals and pharmaceuticals. However, laborious isolation and purification prior to de novo sequencing limits their discovery and hence their use as scaffolds for peptide-based drug development. Here we extend the knowledge about their sequence diversity by analysing the cyclotide content of a violet species native to Western Asia and the Caucasus region. Using an experimental approach, which was named sequence fragment assembly by MALDI-TOF/TOF, it was possible to characterize 13 cyclotides from Viola ignobilis, whereof ten (vigno 1–10) display previously unknown sequences. Amino acid sequencing of various enzymatic digests of cyclotides allowed the accurate assembly and alignment of smaller fragments to elucidate their primary structure, even when analysing mixtures containing multiple peptides. As a model to further dissect the combinatorial nature of the cyclotide scaffold, we employed in vitro oxidative refolding of representative vigno cyclotides and confirmed the high dependency of folding yield on the inter-cysteine loop sequences. Overall this work highlights the immense structural diversity and plasticity of the unique cyclotide framework. The presented approach for the sequence analysis of peptide mixtures facilitates and accelerates the discovery of novel plant cyclotides.
KeywordsViola ignobilis Circular Cystine-knot Oxidative folding Vigno Peptidomics
Reversed-phase high performance liquid chromatography
Matrix-assisted laser desorption ionization-time of flight
Cyclotides are a unique class of cysteine-rich macrocyclic mini-proteins of about 30 amino acids in size that are defined by a head-to-tail cyclized backbone and three disulfide bonds in a knotted arrangement referred to as cyclic cystine-knot (CCK) motif (Craik et al. 1999). Their knotted structure makes them exceptionally stable against thermal, chemical and enzymatic degradation (Colgrave and Craik 2004). Cyclotides have been discovered and isolated from plants of the violet (Violaceae), coffee (Rubiaceae), cucurbit (Cucurbitaceae) and legume family (Fabaceae) (Poth et al. 2010). Their distribution within the plant kingdom still remains unclear (Gruber 2010), but they are expected to be far more widespread and the number of different cyclotides may be around 50,000 (Gruber et al. 2008; Simonsen et al. 2005) making them one of the largest peptide classes within plants. In agreement with their anticipated number, recent studies report the presence of more than 70 different cyclotides within one single species (Seydel et al. 2007; Gründemann et al. 2012). The first cyclotide kalata B1 was discovered from “kalata-kalata”, a decoction from leaves of Oldenlandia affinis, which has been used as a remedy during childbirth in African ethnomedicine due to its uterotonic activity (Gran 1970; Gruber and O’Brien 2011). In line with their reported antibacterial (Tam et al. 1999), antifouling (Göransson et al. 2004), anthelmintic (Colgrave et al. 2008) and insecticidal properties (Jennings et al. 2001; Gruber et al. 2007a; Barbeta et al. 2008) their native function seems to be part of the plant defence system.
Usually, amino acid sequencing of cyclotides is performed after enzymatic digestion of peptides that have been laboriously purified by reversed-phase high performance liquid chromatography (RP-HPLC) to produce single linearized peptides that are amenable to tandem mass spectrometry (MS) analysis. However, the complexity of cyclotide plant extracts, which comprise dozens of distinct peptides, limits their analysis and characterization by standard MS analysis. Using endoproteinase GluC (endo-GluC), cyclotides are mostly cleaved to yield a single (‘ring-opened’) peptide fragment due to a conserved glutamic acid in loop 1, whereas the use of trypsin and chymotrypsin usually yields several fragments due to multiple cleavage sites. When applied to the analysis of cyclotide mixtures as they occur in plant extracts, mass spectra may be confusing and hard to evaluate caused by fragment ion overlays. Hence the application of combinations of digests to obtain peptide-specific fragments and the subsequent accurate assembly of sequence fragments may overcome this issue. Particularly for bracelet cyclotides this is of importance since until now the majority (~70 %) of more than 200 published cyclotide sequences accessible on CyBase (Wang et al. 2008b) belong to this subfamily.
Besides the complexity of cyclotide sequence analysis, another issue associated with their great diversity is their chemical and biological synthesis. Previous studies have shown that different enzymes seem to be involved in backbone cyclization and disulfide bond formation (Gruber et al. 2007b; Saska et al. 2007) during biosynthesis of these gene-encoded peptides in planta. However, the community still lacks clarity about this process, in particular with respect to the sequence-folding relationship, i.e., how the inter-cysteine sequences of different cyclotides can influence the formation of the native CCK-motif and hence determine their folding yield. As a consequence, in vitro oxidative folding is still a major challenge in cyclotide engineering. Whereas high-yield chemical synthesis and folding of Möbius cyclotides is possible (Daly et al. 1999), obtaining correctly folded bracelet cyclotides is much more difficult and yields of about 10–40 % or less of native peptide are common (Leta Aboye et al. 2008; Wong et al. 2011). However, chemical synthesis of cyclotides is an important tool to obtain sufficient peptide material for bioactivity studies.
Our aim is to characterize plant cyclotides from Viola ignobilis, a native Iranian species of the violet family that was recently discovered to contain cyclotides (Hashempour et al. 2011). As a rich source for representatives of both cyclotide subfamilies, we analysed cyclotide-containing fractions using matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) MS and tandem MS analytics. Together with the use of different enzymatic digests and detailed analysis of mass spectra, full peptide sequence coverage could be achieved by assembling and aligning various sequence fragments. This approach, which we called sequence fragment assembly turned out to be a powerful tool for cyclotide identification and de novo sequencing even when analysing mixtures. To dissect the influence of cyclotide sequence variability with respect to the formation of their native structure, we performed oxidative refolding experiments using representative vigno (V. ignobilis) cyclotides comprising distinct, but subtle differences in their inter-cysteine loop sequences. Overall the characterization of these novel cyclotides highlights their enormous sequence variability and the proposed sequencing methodology may overcome limitations in the discovery of novel representatives of this unique class of circular plant peptides.
Materials and methods
Plant collection, extraction and RP-HPLC fractionation
Aerial parts of V. ignobilis Rupr. were collected in the mountains at an altitude of 1,500–2,500 m around the village of Negarestan in the region of East Azerbaijan (Iran) in spring 2010. A voucher specimen was identified and deposited at the Institute of Medicinal Plants and Drug Research, Iran (MPH-1917). The dried plant material (~500 g) was ground prior to solvent extraction with a mixture of MeOH:CH2Cl2 (1:1; v/v) overnight under continuous agitation at 20 °C. After adding of 0.5 volume water the aqueous phase was concentrated on a rotary evaporator prior to freeze drying, yielding what is further referred to as crude extract. The crude extract was dissolved in 0.1 M NH4HCO3 buffer (pH ~ 8.1) and immediately used for solid-phase extraction (SPE). C18 SPE cartridges (Macherey-Nagel, Chromabond; 10 g; 50 mL) were activated with 1 bed volume of MeOH and subsequently equilibrated with 1 bed volume of aqueous 1 % FA. After application of the extract, the cartridges were washed with 1 bed volume of 1 % FA. Putative cyclotide containing fractions of 50 and 80 % EtOH were collected and freeze dried. After dissolving in 1 % FA they were fractionated using preparative and semi-preparative RP-C18 HPLC (Knauer, Eurospher I 5 μm; 250 × 16.1 mm; 100 Å) using a Knauer 1200 series unit, with an isocratic flow of 30 % acetonitrile: H2O (v/v) at a flow rate of 8 mL min−1. Fractions were collected manually by UV detection at 210 nm. All samples were extracted by avoiding prolonged exposure to high pH and sample heating to reduce the risk of Asn deamidation.
Reduction, alkylation and enzymatic digest
Prior to MS analysis, cyclotides were enzymatically digested to produce linearized fragments following reduction and alkylation of Cys-residues. Lyophilized samples (~0.5 μg peptide) were dissolved in 0.1 M NH4HCO3 buffer (pH 8.2) and 20 μL aliquots were reduced by adding 2 μL of 10 mM dithiothreitol and were incubated at 20 °C for 30 min. Alkylation was carried out by adding 4 μL of 100 mM iodoacetamide to the reduced samples and incubating for 10 min at 20 °C. After a second incubation step for 10 min with 1 μL of 10 mM dithiothreitol to quench the reaction with iodoacetamide, 2 μL of trypsin, endo-GluC and/or chymotrypsin (all Sigma-Aldrich, Austria) at concentrations of 0.1–0.5 μg μL−1 were added. All digests were incubated at 37 °C between 3 and 16 h, quenched with concentrated acetic acid (final concentration 3 %) and stored at 4 °C−20 °C until further analysis.
MALDI-TOF/TOF analysis and peptide sequencing
Analysis of crude, reduced/alkylated and digested samples were performed on a MALDI-TOF/TOF 4800 Analyser (AB Sciex, Canada) operated in reflector positive ion mode acquiring 2,000–3,600 total shots per spectrum with a laser intensity set between 3,200 and 3,800. MS and MS/MS experiments were carried out using α-cyano-hydroxyl-cinnamic acid matrix at a concentration of 5 mg mL−1 in 50 % (v/v) acetonitrile. 0.5 μL of each sample was mixed with 3 μL of matrix solution and the mixture was spotted onto the target plate. Tandem mass spectra were acquired using laser energy of 1 kV with and without the use of collision-induced dissociation and processed using the Data Explorer Software. Cyclotides were identified by sequence fragment assembly (as explained below) and manual peptide sequencing. Automated database searches using the ERA-tool (Colgrave et al. 2010) and DeNovoExplorer software were used to compare manual annotated sequences. The MS/MS spectra were examined and sequenced based on assignment of the N-terminal b-ion and C-terminal y-ion series. The disulfide connectivity of CI–IV, CII–V and CIII–VI was assigned based on homology with published sequences.
Oxidative refolding of cyclotides
Cyclotides were purified by RP-HPLC on a Dionex Ultimate 3000 HPLC unit (Dionex, Netherlands) using semi-preparative (250 × 10 mm) and analytical (250 × 4.6 mm) Kromasil C18 columns (5 μm; 100 Å) with linear gradients of 0.1–1 % min−1 or isocratic flow of 25–35 % buffer B (90 % acetonitrile in ddH2O, 0.08 % TFA) at flow rates of 3 and 1 ml min−1, respectively. The control peptide kalata B1 was isolated from Oldenlandia affinis extract as described earlier (Gründemann et al. 2012). The same procedure was applied for the purification of cycloviolacin O2 from Viola odorata. Reduction was performed as described above and stopped after 30 min incubation by adding concentrated TFA (Sigma-Aldrich, Austria) and samples were immediately subjected to HPLC purification. Folding of 60 μL aliquots, containing 2.5–10 μM peptide, was performed at final concentration of 2 mM reduced (GSH) and 0.1 mM oxidized (GSSG) glutathione (Sigma-Aldrich, Austria). Freeze-dried aliquots were resolved in three different folding-buffers, i.e., 25 and 75 % isopropanol (Roth, Germany) and 35 % DMSO/5 % dodecyl-β-maltoside (DBM) in 0.1 M NH4HCO3 buffer (pH 8.2). For control experiments with cycloviolacin O2 the folding conditions included final concentrations of 2 mM GSH and 2 mM cystamine in 35 % DMSO/5 % DBM buffer and GSH/cystamine (2/2 mM) in 0.1 M Tris–HCl buffer (pH 8.5) at 4 °C and 20 °C. Aliquots were analysed at several time points (15 min, 1 h and 24 h) after incubation at 20 °C. Folding reaction was quenched by adding 1 μL of concentrated TFA and samples were analysed by RP-HPLC on an Aeris Peptide XB-C18 (150 × 2.1 mm; 3.6 μm; 100 Å) column (Phenomenex, Germany) at a flow rate of 0.3 ml min−1 with a gradient of 2 % min−1 buffer B. Folding yields were determined using the peak integration tool of Chromeleon software 6.8 with a peak detection limit set at 0.07 × signal (mAU at 214 nm) × retention time (min). Folding kinetic graphs and calculations of rate constant and half-time were prepared using the one-phase association fit in GraphPad Prism 5 software.
Cyclotide homology modelling
The structural models of vigno 1, 2 and 10 were modelled using the CycloMod application for cyclotide structure modelling within Cybase (http://www.cybase.org.au/). The models were generated using Modeller 9.10 and analysed by Molprobity (Davis et al. 2007). The percentage of residues in the most favoured Ramachandran region and the Molprobity scores are: vigno 1 (92.59 % and 2.71), vigno 2 (89.29 % and 2.75) and vigno 10 (89.66 % and 3.27).
Results and discussion
The discovery and hence the pharmaceutical value of cyclotides is limited by an efficient and reliable protocol for peptide sequence analysis, in particular in crude plant extracts and fractions that contain mixtures of different cyclotides. Therefore, the main goal of this study was to describe a robust method for cyclotide sequence characterization using MALDI-TOF/TOF analytics.
Identification of novel cyclotides from Viola ignobilis
De novo cyclotide sequencing using ‘sequence fragment assembly’
Cyclotide-containing mixtures were chemically modified to yield S-carbamidomethylated Cys-residues and digested to produce linear peptides amenable to fragmentation by tandem MS. Completely reduced and alkylated samples were digested using single enzymes or combinations of trypsin, endo-GluC and chymotrypsin. Resulting mass spectra were manually analysed by assigning N-terminal b- and C-terminal y-ions. Most novel cyclotide sequences were independently confirmed by automatic database searches using the ERA-tool (Colgrave et al. 2010) and DeNovoExplorer software.
Sequence alignment of cyclotides identified from Viola ignobilis
Sequence variation of novel vigno cyclotides
The 13 identified peptides from Viola ignobilis belong to both subfamilies (Table 1). One Möbius cyclotide, known as varv peptide A, has been previously isolated from Viola arvensis (Claeson et al. 1998), as well as two bracelet cyclotides, known as cycloviolacin O2 and O9, have been originally found in Viola odorata (Craik et al. 1999). This is not surprising since some cyclotides such as varv peptide E (=cycloviolacin O12), occur in many different Viola species such as V. tricolor, V. odorata, V. arvensis, V. bashoanensis, V. yedoensis and V. abyssinica and, therefore, seem to be genus-specific. Besides these rather rare examples of inter-genus identity, each single plant species seems to express an abundant array of specific cyclotides. In V. ignobilis the most abundant Möbius cyclotides are vigno 1 and vigno 2. Sequence analysis of these two peptides revealed the presence of an AGGT motif in loop 2 which was recently also described for V. abyssinica cyclotides (Yeshak et al. 2011).
Besides the conserved six Cys-residues and the glutamic acid (E) in loop 1, all six Möbius cyclotides have the same typical GET motif in loop 1 and a serine in loop 4, which is conserved within all newly identified cyclotides from V. ignobilis. Furthermore, the GES motif in loop 1 and the VWIP motif in loop 2 are conserved within all bracelet cyclotides from V. ignobilis. The differences and novelties are within loop 3, 5 and 6 which are known to show the highest amino-acid variability (Supplementary Fig. S5; Table 1). For vigno 2 a novel sequence motif for loop 6 of cyclotides, VRDGSSPL, has been discovered. Although all amino acids are known to occur in this loop, the presence of two serine residues next to each other has not been reported hitherto. The presence of two serine residues and an aspartic acid makes this loop more hydrophilic and confers the peptide with an overall net charge of −1. To distinguish between Asn and Asp residues in loop 6 of vigno 2 and vigno 10, we have additionally analysed the molecular weight and isotopic distribution of diagnostic fragment ions (Poth et al. 2010) (Supplementary Figure S6). Further sequencing of the co-eluting Möbius cyclotides vigno 3 and vigno 4 (Fig. 2) revealed the presence of an alanine or glycine within loop 3 and corresponds to the mass difference of 14 Da in the crude sample (Fig. 4a). A combined digest using trypsin and endo-GluC yields the fragments of 2,416.9 and 2,430.9 Da that allowed the sequence determination and confirming the difference of a glycine (NTPG) and alanine (NTPA) at the last position of loop 3. This is to our knowledge the first report of an alanine residue at this position and expands the known possibilities at this position which was primarily thought to be a conserved glycine (Craik et al. 1999). Within the sequence of vigno 5 a glycine at the first position of loop 5 was found, which so far has only been shown for bracelet cyclotides such as cycloviolacin Y1-3 (Wang et al. 2008a) and tricyclon A and B (Mulvenna et al. 2005). Vigno 6 shows the very common KSKV sequence for loop 5, which is intersected by a glycine, KGSKV. The identification of ten novel peptides underlines the high flexibility and sequence variability caused by single amino acid changes at various positions. It is obvious that sequence variability accounts for different biological and chemical behaviour due to varying physico-chemical properties. As an example, we decided to characterize the in vitro oxidative refolding properties of representative vigno cyclotides, since it is a valuable model to analyse their sequence-folding relationship, which has broader implications on the synthesis and design of cyclotides as tools in pharmaceutical applications.
Sequence-folding relationship of novel vigno cyclotides
The influence of certain residues on oxidative folding and the correct formation of the native disulfide bonds of cyclotides are still not fully understood. In the current study it has been our particular interest to elucidate the sequence-folding relationships of three vigno cyclotides with respect to their yield and folding kinetics using different folding conditions. Therefore, the oxidative refolding of the most abundant Möbius cyclotides in V. ignobilis, vigno 1 and vigno 2 was studied, in comparison to the prototypic cyclotide kalata B1, and the bracelet cyclotide vigno 10 in comparison to cycloviolacin O2, a well-studied bracelet cyclotide isolated from V. odorata.
Overview of yields and folding kinetics of vigno cyclotides
25 % isopropanol (aqueous)
75 % isopropanol (aqueous)
35 % DMSO/5 % dodecyl-β-maltoside (DBM)
25 % isopropanol (aq.)
75 % isopropanol (aq.)
35 % DMSO/5 % DBM
25 % isopropanol (aq.)
75 % isopropanol (aq.)
35 % DMSO/5 % DBM
25 % isopropanol (aq.)
75 % isopropanol (aq.)
35 % DMSO/5 % DBM
25 % isopropanol (aq.)
75 % isopropanol (aq.)
35 % DMSO/5 % DBM
35 % DMSO/5 % DBMf
35 % DMSO/6 % Brij35 (Tris, 4 °C)f
35 % DMSO/6 % Brij35 (Tris, 20 °C)f
Comparison of the sequences and structures provides insights into the differences observed in the folding efficiency in terms of yield and kinetics. Vigno 1 and vigno 2 have high sequence similarity to kalata B1, and all three peptides have high folding yields in 75 % isopropanol buffer (Table 2). However, there are differences in folding using the 25 % isopropanol and 35 % DMSO/detergent buffers (Table 2). The three Möbius peptides differ in loop 6, i.e., VRNGLPL in vigno 1, VRDGSSPL in vigno 2 and TRNGLPV in kalata B1 (Table 2; Supplementary Fig. S8). The two adjacent serine residues and an aspartic acid in vigno 2 make this loop more hydrophilic and confers the peptide an overall single negative net charge. This may explain higher folding yields of vigno 2 for the 25 % isopropanol buffer and lower yields for the more hydrophobic 75 % isopropanol buffer as compared to vigno 1, a neutral cyclotide. The more hydrophilic nature of loop 6 of vigno 2 (Supplementary Fig. S8) (Aboye et al. 2011) may also contribute to the different folding yields in the DMSO/detergent buffer, which vary between 62.7 % for vigno 1, 30.5 % for vigno 2 and 87 % for the control peptide kalata B1. Furthermore, loop 2 of vigno 1 and 2 are very similar; they both have a slightly more hydrophobic nature compared to kalata B1 (Supplementary Fig. S8), which probably also contributes to the folding differences and the overall later elution on RP-HPLC (Fig. 7f).
This work has broadened the knowledge about the immense sequence diversity of plant cyclotides, a unique class of naturally occurring backbone-cyclized peptides built around a conserved cyclic cystine-knot. By characterizing 13 sequences from an Iranian violet species it has been confirmed that cyclotides are one of the most abundant peptide class within the plant kingdom. The characterization of cyclotides in mixtures using MALDI-TOF/TOF analytics may overcome laborious isolation and challenges in de novo peptide sequencing. The use of different proteases as well as the assembly and alignment of sequence fragments facilitates the discovery of novel cyclotide sequences. In addition, by performing oxidative refolding studies on representative cyclotides the knowledge of their in vitro oxidative folding behaviour was extended and this underlines the high dependency of folding yield to their inter-cysteine loop sequences and careful choice of the folding conditions. These studies have further implications taking into account that cyclotides have numerous bioactivities and hence display a scaffold that is extensively used for peptide-based drug design.
Work on cyclotides in the lab of CWG is financially supported by the Austrian Science Fund (FWF): P22889-B11. Mass spectrometry data of this research have been obtained by access to the MS core facility of the Center for Physiology and Pharmacology (Medical University of Vienna). Work on isolation of cyclotides in the lab of AG was supported by the research deputy of Shahid Beheshti University. NLD is supported by an Australian Research Council Future Fellowship.
This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.