Chemical Synthesis and Functional Analysis of VarvA Cyclotide

Cyclotides are circular peptides found in various plant families. A cyclized backbone, together with multiple disulfide bonds, confers the peptides’ exceptional stability against protease digestion and thermal denaturation. In addition, the features of these antimicrobial molecules make them suitable for use in animal farming, such as aquaculture. Fmoc solid phase peptide synthesis on 2-chlorotrityl chlorine (CTC) resin using the “tea-bag” approach was conducted to generate the VarvA cyclotide identified previously from Viola arvensis. MALDI-TOF mass spectrometry determined the correct peptide amino acid sequence and the cyclization sites-critical in this multicyclic compound. The cyclotide showed antimicrobial activity against various Gram-negative bacteria, including recurrent pathogens present in Chilean aquaculture. The highest antimicrobial activity was found to be against Flavobacterium psychrophilum. In addition, membrane blebbing on the bacterial surface after exposure to the cyclotide was visualized by SEM microscopy and the Sytox Green permeabilization assay showed the ability to disrupt the bacterial membrane. We postulate that this compound can be proposed for the control of fish farming infections.


Introduction
Cyclic peptides have attracted great interest in recent years due to their increased stability over linear peptides and wide range of bioactivities [1].
Most peptides show a linear structure with open ends, which makes them targets for proteolytic enzymes, thus decreasing their bioavailability [2,3]. In contrast, cyclic peptides have the unique feature that their N and C termini are joined in an amide bond to form a cyclic backbone, and they show greater stability than their linear counterparts [4][5][6][7][8].
Measuring 28-37 amino acid residues in length, cyclotides are the largest known family of cyclic peptides [9]. To date, more than 150 cyclotides have been characterized from plants of the Rubiaceae, Violaceae, Cucurbitaceae, Solaceae, and Fabaceae families [10][11][12]. The six highly conserved cysteine residues of cyclotides lead to a knotted arrangement of three disulfide bonds. An embedded ring is formed by two disulfide bonds and their connecting backbone segments, and the third disulfide bond

Chemical Synthesis of VarvA Cyclotide
The Fmoc-SPPS chemical synthesis approach on 2-chlorotrityl chlorine (CTC) resin was used for the VarvA cyclotide synthesis. The first Fmoc-Asn(Trt)-OH was introduced onto CTC-resin, resulting in a loading of 0.62 mmol/g resin. Next, the resin was subjected to various coupling-deprotection steps to build the linear peptide as the precursor for the cyclic peptide ( Figure 1). The course of the couplings was monitored using the bromophenol test [34], which allows in situ control of the reaction since the absence of color indicates that the reaction has been completed. The VarvA backbone (cyclo(GLPVCGECFGGTCNTPGCSCDPWPMCSRN)) was generated, starting from 2-chlorotrityl chloride resin loading with Fmoc-Asn(Trt)-OH. The second step was to place the resin in a polypropylene bag for the "tea bag" SPPS approach. Successive deprotection/coupling steps were carried out to build the peptide backbone. The peptide was then cleaved from the resin, followed by the cyclization step. Additionally, the final product yield for this step based on initial resin loading is shown. Finally, peptide oxidation was carried out to ensure correct folding and the product yield is shown.
Before the cleavage of the protected peptide, an aliquot of the peptide resin was treated with a high concentration of TFA to release the free peptide. The molecular mass of the resulting linear peptide (2992.0 Da) was confirmed by mass spectrometry, which showed a clean synthesis ( Figure  2A). Next, a peptide concentration of 0.5 mM was used for the cyclization. In this case, the success of the cyclization was demonstrated by mass spectrometry after the global deprotection of the cyclic peptide ( Figure 2B). Taking into account that ammonium bicarbonate has been successfully used for The VarvA backbone (cyclo(GLPVCGECFGGTCNTPGCSCDPWPMCSRN)) was generated, starting from 2-chlorotrityl chloride resin loading with Fmoc-Asn(Trt)-OH. The second step was to place the resin in a polypropylene bag for the "tea bag" SPPS approach. Successive deprotection/coupling steps were carried out to build the peptide backbone. The peptide was then cleaved from the resin, followed by the cyclization step. Additionally, the final product yield for this step based on initial resin loading is shown. Finally, peptide oxidation was carried out to ensure correct folding and the product yield is shown.
Before the cleavage of the protected peptide, an aliquot of the peptide resin was treated with a high concentration of TFA to release the free peptide. The molecular mass of the resulting linear peptide (2992.0 Da) was confirmed by mass spectrometry, which showed a clean synthesis ( Figure 2A). Next, a peptide concentration of 0.5 mM was used for the cyclization. In this case, the success of the cyclization was demonstrated by mass spectrometry after the global deprotection of the cyclic peptide ( Figure 2B). Taking into account that ammonium bicarbonate has been successfully used for folding synthetic cyclic inhibitor peptides [35,36], we selected this buffer. The mass spectrometry analysis showed the loss of six hydrogen atoms from the three-disulfide bond formation ( Figure 2C). folding synthetic cyclic inhibitor peptides [35,36], we selected this buffer. The mass spectrometry analysis showed the loss of six hydrogen atoms from the three-disulfide bond formation ( Figure 2C). Moreover, the I-Tasser server was used to determine the three-dimensional structure. The hydrophobic surface of the structural model shows the presence of hydrophobic and hydrophilic patches ( Figure 3). Moreover, the I-Tasser server was used to determine the three-dimensional structure. The hydrophobic surface of the structural model shows the presence of hydrophobic and hydrophilic patches ( Figure 3).  Kyte & Doolittle scale in UCSF Chimera [37]. Four different views of the surface are shown to facilitate visualization of the surface.

Antimicrobial Activity of VarvA Synthetic Cyclotide against Fish Bacterial Pathogen
Here, we focused on the antimicrobial study of the VarvA synthetic cyclotide against various Gram-negative bacterial pathogens that affect fish aquaculture. The MIC concentration of synthetic cyclotide against V. anguillarum, V. ordalii, F. psychrophilum, A. hydrophila, and A. salmonicida was determined by a microdilution assay. The resulting MICs are shown in Table 1. According to this assay, F. psychrophilum was the bacteria most susceptible, with an MIC of 12.5 μM. Moreover, the peptide demonstrated good activity against A. hydrophila and A. salmonicida, with an MIC of 22.5 μM. Finally, the highest MIC was obtained against V. anguillarum and V. ordalii (30 μM).

Bacterial Membrane Damage Induced by VarvA Synthetic Cyclotide
The impact of VarvA on membrane integrity was studied by the SYTOX Green permeabilization assay. SYTOX green is selectively taken up into cells with compromised membrane integrity and exhibits greatly enhanced fluorescence upon DNA binding [38]. SYTOX green and VarvA were

Antimicrobial Activity of VarvA Synthetic Cyclotide against Fish Bacterial Pathogen
Here, we focused on the antimicrobial study of the VarvA synthetic cyclotide against various Gram-negative bacterial pathogens that affect fish aquaculture. The MIC concentration of synthetic cyclotide against V. anguillarum, V. ordalii, F. psychrophilum, A. hydrophila, and A. salmonicida was determined by a microdilution assay. The resulting MICs are shown in Table 1. According to this assay, F. psychrophilum was the bacteria most susceptible, with an MIC of 12.5 µM. Moreover, the peptide demonstrated good activity against A. hydrophila and A. salmonicida, with an MIC of 22.5 µM. Finally, the highest MIC was obtained against V. anguillarum and V. ordalii (30 µM). Table 1. Values of minimal inhibitory concentration (MIC) for synthetic cyclotide on different Gram-negative bacterial fish pathogens.

Bacterial Membrane Damage Induced by VarvA Synthetic Cyclotide
The impact of VarvA on membrane integrity was studied by the SYTOX Green permeabilization assay. SYTOX green is selectively taken up into cells with compromised membrane integrity and exhibits greatly enhanced fluorescence upon DNA binding [38]. SYTOX green and VarvA were added simultaneously to log-phase bacteria (E. coli and F. psychrophilum), and the SYTOX Green fluorescence was quantified. Phospholipase-A2-derived synthetic peptide was used as a positive control [39]. The SYTOX green uptake was detected as early as 1 min after control peptide and VarvA treatment ( Figure 4). The SYTOX green intensity increased strongly within 1 to 10 min.
Molecules 2018, 23, x FOR PEER REVIEW 6 of 15 added simultaneously to log-phase bacteria (E. coli and F. psychrophilum), and the SYTOX Green fluorescence was quantified. Phospholipase-A2-derived synthetic peptide was used as a positive control [39]. The SYTOX green uptake was detected as early as 1 min after control peptide and VarvA treatment ( Figure 4). The SYTOX green intensity increased strongly within 1 to 10 min. In addition, SEM microscopy was used to provide direct evidence for the antimicrobial effect of VarvA synthetic cyclotide. The untreated A. salmonicida cells, prepared for SEM micrographs in phosphate buffer, displayed a smooth and intact surface ( Figure 5A). However, after incubation with an MIC of the synthetic cyclotide, multiple blisters of various shapes were observed ( Figure 5B). In addition, the partial detachment of an outer membrane was seen on A. salmonicida cells ( Figure 5B-II). Similarly, under control conditions, the surface of F. psychrophilum cells and V. ordalii were smooth ( Figure 5C,E). However, after exposure to the synthetic cyclotide, numerous blisters and partial detachment of the blisters from the membrane were observed ( Figure 5D,F). In addition, SEM microscopy was used to provide direct evidence for the antimicrobial effect of VarvA synthetic cyclotide. The untreated A. salmonicida cells, prepared for SEM micrographs in phosphate buffer, displayed a smooth and intact surface ( Figure 5A). However, after incubation with an MIC of the synthetic cyclotide, multiple blisters of various shapes were observed ( Figure 5B). In addition, the partial detachment of an outer membrane was seen on A. salmonicida cells ( Figure 5B-II). Similarly, under control conditions, the surface of F. psychrophilum cells and V. ordalii were smooth ( Figure 5C,E). However, after exposure to the synthetic cyclotide, numerous blisters and partial detachment of the blisters from the membrane were observed ( Figure 5D,F).

Discussion
Cyclotides are amenable to significant sequence variation. Specifically, the backbone portions between cysteine residues, referred to as loops, can be modified [40,41]. Thus, the ability to chemically synthesize cyclotides is an important goal, both for the practical purpose of mutagenesis studies to understand their mechanism of action, as well as studies of new pharmaceutical applications. Here, we use the CTC-resin for the elongation of the peptide chain of VarvA cyclotide, an approach first described by Craik's group [35]. The simplicity of the method was exemplified by the use of a simple tea bag as a reactor for peptide elongation. Taking advantage of the lability of the CTC-resin to acids, we used a low concentration of TFA to release the protected peptide from the resin. These conditions assure that amino acid side chains remain protected and facilitate peptide cyclization [42]. Nevertheless, for minimizing the probability of oligomerization and improving the cyclization efficiency, it is necessary to use a high dilution [43,44].

Discussion
Cyclotides are amenable to significant sequence variation. Specifically, the backbone portions between cysteine residues, referred to as loops, can be modified [40,41]. Thus, the ability to chemically synthesize cyclotides is an important goal, both for the practical purpose of mutagenesis studies to understand their mechanism of action, as well as studies of new pharmaceutical applications. Here, we use the CTC-resin for the elongation of the peptide chain of VarvA cyclotide, an approach first described by Craik's group [35]. The simplicity of the method was exemplified by the use of a simple tea bag as a reactor for peptide elongation. Taking advantage of the lability of the CTC-resin to acids, we used a low concentration of TFA to release the protected peptide from the resin. These conditions assure that amino acid side chains remain protected and facilitate peptide cyclization [42]. Nevertheless, for minimizing the probability of oligomerization and improving the cyclization efficiency, it is necessary to use a high dilution [43,44].
Several strategies have been included for peptide disulfide bridge arrangement, including the use of orthogonally protected cysteine residues and/or oxidation reactions promoted by reagents such as iodine, thallium trifluoroacetate, potassium ferricyanide, or dimethylsulphoxide [45][46][47][48]. However, strong oxidant compounds can also affect other amino acid residues, like tryptophan and tyrosine [36]. As an alternative, the air oxidation method performed in a buffered aqueous medium can be used. Here, we used ammonium bicarbonate for VarvA folding, supporting previous works that have used this buffer for the cyclotides disulfide bridge arrangement [49]. An additional advantage of ammonium bicarbonate is that after acidification, it is removed by lyophilization. Therefore, and although cyclotides have a difficult structure, our study demonstrates that they can be folded just by using the air oxidation method.
Cyclotides were initially studied because their main function in plants is in the control of opportunistic pathogens [50]. Given that cyclotides have an amphipathic character similar to that of classical antimicrobial peptides, it was hypothesized that they have antimicrobial activities [51]. Diverse antimicrobial properties have been described for cyclotides, including antiviral, antifungal, insecticidal, antiparasitic, and antibacterial activities [22,25,[52][53][54]. Given their broad antimicrobial spectrum, cyclotides emerge as an interesting target to exploit for the improvement of animal health.
Fish mortality caused by infectious diseases is a significant problem in aquaculture worldwide. Intensive aquaculture conditions induce fish stress, which in turn makes them susceptible to invasion by opportunistic bacterial pathogens [55]. In this regard, intensive aquaculture provides an ideal scenario in which to explore the capacity of antimicrobial peptides to protect fish against bacterial infections. Moreover, there are only a handful of approved antibiotics available and consequently the number of resistant bacteria to existing antibiotics is increasing. Here, we demonstrated the antimicrobial activity of the VarvA synthetic cyclotide against various Gram-negative bacterial pathogens that affect fish aquaculture. In addition, low MIC values were obtained for all bacteria tested. Thus, cyclotides could be of great interest even in aquaculture, as they are attractive candidates for antimicrobial therapeutic approaches.
The antibacterial properties of natural and synthetic cyclotides have been described in different works [51,56]. However, contradictory results have been reported. For example, preliminary studies showed that native kalata B1 is not active against S. aureus, but is active against a Gram-negative strain. Conversely, synthetic kalata B1 was found to be active against the Gram-positive S. aureus, but relatively inactive against Gram-negative bacteria [57]. These discordances could be explained by structural differences or by the experimental conditions used [25]. Here, we used SEM microscopy to provide direct evidence for the antibacterial effect of the synthetic cyclotide. The SEM micrographs showed that untreated bacteria displayed a smooth and intact surface, but after incubation with VarvA, multiple blisters and partial membrane detachment were observed. This membrane disruption was supported by the SYTOX Green permeabilization assay. Taken together, these results suggest that cyclotides like VarvA exert antimicrobial activity by disrupting bacterial membranes, along with previous reports.
Biophysical studies have described that cyclotides target biological membranes [58]. Most cyclotides have conserved molecular surface regions, including hydrophobic and bioactive patches involved in the insertion into the membranes [59]. Several models for the interaction of AMPs with membranes, such as the "barrel stave," "toroidal pore," or "carpet" model, have been postulated [60]. In the carpet model, the peptides form a layer of "carpet" that induces membrane weakness, which ultimately results in membrane collapse by a detergent-like action. Moreover, the presence of membrane blebbing is associated with this model [61][62][63]. Thus, the SEM images support the notion that VarvA exerts a detergent-like mechanism to disrupt bacterial membranes. This idea is in agreement with the putative structure of the peptide, with hydrophobic patches able to interact with the membrane. Particularly, loop 6 composed of the first and the last amino acids form a cavity with the most hydrophobic residues on one side and the most hydrophilic on the other. This kind of hydrophobic distribution could favor the interaction with the bacterial membrane, first by the hydrophilic and charged residues to made the superficial contact, and second by the hydrophobic ones to interact with the lipid moiety of the membrane. Therefore, this study provides new evidence about the mechanism of action of cyclotides against the bacterial membrane.
The loading efficiency for the first amino acid was determined as previously described [64]. A sample of resin (5 mg) was placed in 2 mL microcentrifuge tubes (triplicate), and 1 mL 20% v/v piperidine in DMF was added to each tube. The tubes were vortexed briefly and allowed to agitate on a rotatory shaker (150 rpm) at room temperature for 20 min. Aliquots of 30 µL of each of the samples were diluted to 3 mL with DMF. The absorbance of each UV absorbance of each sample (300 nm) was determined against DMF as the reagent blank. Loading was calculated using the following equation: Loading = 101(A)/7.8(W), where A is the absorbance and W is the mass of the resin [64].
The cyclotide was synthesized on 40 mg resin (Fmoc-Asn(Trt)-O-CTC-resin) placed in polypropylene bags (74 µm mesh). The resin in these bags (quintupled) was swelled in a polypropylene bottle with DMF. The Fmoc group was removed by the addition of 20% v/v piperidine in DMF, for 10 min (twice). Next, the resin bags were washed 3 × 1 min in DMF, 1 × 1 min in isopropyl alcohol (IPA), 1 × 1 min in bromophenol blue (1% in DMF) for free amino group testing, and finally with 2 × 1 min with DMF and 2 × 1 min with DCM. The coupling reaction procedure was as follows: Fmoc-amino acid (10-fold excess) was activated by N-[(1H-benzotriazol-1-yl)-(dimethylamino) methylene]-N-methylmethanaminium hexafluorophosphate N-oxide (HBTU) in the presence of ethyl cyanohydroxyiminoacetate (OxymaPure ® ) and DIEA (5/5/5/7.5 milliequivalent (meq), respectively) in DMF. The coupling reaction was performed under vigorous shaking at room temperature for 3 h. The resin was then washed 2 × 1 min in DMF, and the second coupling reaction using TBTU as the activation reagent was performed immediately. Throughout the synthesis, DMF was used to wash the resin and to dissolve the amino acid, and all coupling and washing steps were performed under vigorous shaking. The cycle, starting with removal of the Fmoc group, was repeated until the last Fmoc-amino acid had been coupled.

Cyclotide Cyclization and Oxidative Folding
Side chain protected linear peptides were first cleaved from the resin with 2% trifluoroacetic acid (TFA) in DCM for 30 min at room temperature (3 mL per bag). Next, 3 mL of water (Milli Q grade) was added, and TFA and DCM were then evaporated. The cyclization reaction was performed under highly diluted conditions in acetonitrile (ACN) with a peptide concentration of 0.5 mM. The cyclization reaction was carried out with 1% DIEA and 1 mM HBTU/Oxima and was stirred at room temperature for 1 h. After solvent evaporation, the peptide was dissolved in DCM, and sodium bicarbonate was added. The solution was centrifuged at 3000× g for 5 min. Sodium sulfate was then added and the mixture was centrifuged one more time. Finally, DCM was rota-evaporated.
For the oxidative folding step, 10 mg of cyclic peptide was dissolved in water with 0.1 M ammonium bicarbonate (NH 4 HCO 3 ) pH 8.0 [35]. The solution was stirred at room temperature for 1 h. Then aqueous solution was applied onto a Sep-pak C18 Vac cartridge (Waters Associates, Milford, MA, USA) equilibrated in acidified water (0.05% TFA) (Milli Q grade). After washing with acidified water (six times), the peptides were eluted at a flow rate of 1 mL/min with 5%, 10%, 20%, 30%, 40%, 60%, and 80% ACN in water. The appropriate fractions were collected, and the ACN was evaporated on a SpeedVac centrifuge. The fractions were then analyzed by reversed-phase (RP)-HPLC on a Water Corp XBridge™ BEH C18 column (100 mm × 4.6 mm, 3.5 µm) using a 0-70% ACN gradient, water containing 0.05% TFA as solvent A, and ACN containing 0.05% TFA as solvent B, at a flow rate of 1 mL/min for 8 min.

Prediction of Peptide Structure
The Basic Local Alignment Search Tool (BLAST) was used to determine the homology of the sequence in the Uniprot database, and a comparison with the sequences found was made by multiple alignment in Jalview [65]. The I-Tasser server [66] was used to determine the three-dimensional structure, and the model obtained was refined using UCSF Chimera (http://www.rbvi.ucsf.edu/ chimera) [37]. Additionally, a hydrophobic surface was generated using a 1.4 angstrom probe and the Kyte & Dolittle hydrophobicity scale (used by default in Chimera).

Antibacterial Assay
Antibacterial activity was determined using the microplate assay, as previously described [67][68][69][70]. A range of peptide concentrations (1-50 µM) were mixed with 100 µL of an exponential phase bacterial culture of Vibrio anguillarum, Vibrio ordalii, Flavobacterium psychrophilum, Aeromonas hydrophila, and Aeromonas salmonicida. In addition, a phospholipase-A2-derived synthetic peptide variant was used as a positive control [39]. The test was performed at a starting OD of 0.001 at 620 nm in the following: tryptic soy broth (TSB) for A. salmonicida, and A. hydrophila; TSB containing 1.5% NaCl for V. anguillarum and V. ordalii; and Anacker and Ordal's (AOAE) liquid medium for F. psychrophilum [70][71][72]. Absorbance was measured after 16 h of incubation. Minimum inhibitory concentrations (MIC) were defined as the lowest concentration of peptide that inhibited the visible growth of bacteria [73]. MICs were measured in quadruplicate.

SYTOX Green Bacteria Permeabilization Assay
The SYTOX Green uptake assay was performed according to a previously described procedure [74,75]. Cultures of exponentially-grown E. coli and F. psychrophilum were diluted in 10 mM sodium phosphate buffer (pH 7.2) to a cell density of 1 × 10 6 CFU/mL. Then, aliquots of 90 µL of this cell culture were deposited in optics real time PCR tubes and 5 µL of the solution of VarvA synthetic cyclotide (MIC concentration) and 5 µL of 100 µM SYTOX Green were added to the wells, and then the tubes were placed in the thermocycler (Agilent Mx3000p qPCR System). The thermocycler program was performed using the SYBR green filter selected, 40 cycles of 30 s at 37 • C (E. coli) or 24 • C (F. psychrophilum) with a reading at the end of each cycle. Control experiments were performed under the same conditions without the addition of peptide.

Scanning Electron Microscopy (SEM)
Aliquots of mid-log phase A. salmonicida, V. ordalii, and F. psychrophilum were harvested by centrifugation at 1000× g for 5 min. Cell pellets were washed twice with 10 mM saline phosphate buffer (PBS) and resuspended in the same buffer. The cell suspension was incubated at 24 • C for 20 min with the MIC concentration of the peptide. After incubation, the cells were centrifuged and washed three times at 1000× g for 5 min with PBS. Bacterial pellets were deposited on a glass coverslip in a Petri dish for 20 min and then fixed in 500 µL of 2.5% v/v glutaraldehyde in PBS. Subsequently, the bacterial samples were dehydrated with a graded ethanol series, critical-point dried, and coated with platinum-palladium to avoid charging in the microscope. Microscopic examination was performed using a Hitachi SU 3500 scanning electron microscope (Hitachi Ltd. Tokyo, Japan).

Conclusions
Fmoc solid phase peptide synthesis on 2-chlorotrityl chlorine (CTC) resin using the "tea-bag" approach was used to generate the VarvA cyclotide identified previously from Viola arvensis. The antimicrobial activity of this synthetic cyclotide was studied against Vibrio anguillarum, Vibrio ordalii, Flavobacterium psychrophilum, Aeromonas hydrophila, and Aeromonas salmonicida, being the highest against Flavobacterium psychrophilum. In addition, membrane blebbing on the bacterial surface was observed after exposure to VarvA, showing the ability of this cyclotide to disrupt the bacterial membrane. It is proposed that this compound exerts a carpet mechanism, a notion that is also consistent with the proposed structural model. Thus, cyclotides represent an interesting alternative to the use of antibiotics in the control of fish farming infections.