Synthesis of Cyclotetrapeptides Analogues to Natural Products as Herbicides

The synthesis of cyclotetrapeptides analogues of the natural products tentoxin and versicotide D was achieved in good yield by solid phase peptide synthesis (SPPS) of their linear precursors and solution phase cyclization. All the cyclopeptides and several open precursors were evaluated as herbicides. Five cyclopeptides and five lineal peptides showed a significant inhibition (>70%) of Ryegrass seed’s radicle growth at 67 μg/mL. The evaluation at lower concentrations (4–11 μM) indicates two cyclopeptides analogs of tentoxin, which present one (N-Methyl-d-Phe), and two N-MeAA (N-Methyl-Ala and N-Methyl-Phe), respectively, as the most active of them, showing remarkable phytotoxic activity. In two cases, the open precursors are as active as their corresponding cyclopeptide. However, many linear peptides are inactive and their cyclization derivatives showed herbicidal activity. In addition, two cyclopeptide analogues of versicotide D showed more improved activity than the natural product. The results indicate that the peptide sequence, the amino acid stereochemistry and the presence of N-methyl group have important influence on the phytotoxic activity. Moreover, several compounds could be considered as lead candidates in the development of bioherbicides.


Introduction
Agrochemicals play a key role in agriculture as their use has dramatically increased productivity. Weeds represent one of the most important pests that need to be controlled. In fact, about 50% of the commercial pesticides used worldwide are herbicides [1]. However, many factors are causing an urgent need for the development of novel herbicides. In the first place, the toxicity and long-term impact on human health and the environment of extensively used agrochemicals, such as glyphosate, have been deeply studied during the last decades [2][3][4]. As a result, many agrochemicals have been banned by governmental agencies in many countries [5]. Furthermore, agrochemicals have caused the eutrophication of water bodies, with environmental consequences, such as the increase in cyanobacteria blooms [6], which produce harmful toxins [7], affecting animal and human health. On the other hand, many herbicides have become ineffective by the development of weed resistance [8][9][10], and the discovery of new ones during the recent decades have been scarce [11][12][13].
Natural products have long been a source of inspiration in the discovery of bioactive molecules. Furthermore, natural products could provide an environmentally friendlier approach to weed management [14][15][16]. In the search for herbicides with novel modes of action and safer for both human health and the environment, plants, fungi extracts and metabolites have been investigated as bioherbicides [17].
Naturally occurring cyclic peptides and synthetic cyclic peptides inspired in natural products have found application in a large variety of fields, such as drug discovery [18], great binding affinity, low toxicity, and the capability of targeting traditionally "undruggable" protein surfaces [21]. In addition, cyclopeptides exhibit increased metabolic stability in comparison with their linear counterparts.
In particular, a relatively large number of cyclic tetrapeptides reported in the literature have shown interesting bioactivities. However, due to their size, their synthesis can be challenging. As the ring size decreases, peptide cyclization becomes more difficult due to energy constraints [22]. The presence of turn-inducing motifs, such as D-amino acids or N-methyl amino acids (N-MeAA), are relevant in promoting a preorganized conformation for cyclization. In addition, cyclopeptides containing N-MeAA could produce an impact on bioactivity as cell permeability and chemical stability are increased [23].
Our group has recently reported the phytotoxic evaluation of natural products versicotides, among which versicotide D (3, Figure 1) shows low cytotoxicity on HepG2 cells [31], high phytotoxicity (74% radicle growth inhibition at 67 μg/mL) and the ability to inhibit cyanobacteria population with a substantial depletion of the concentration of microcystins in the media [32]. Here, we report the synthesis of a library of cyclic tetrapeptides inspired on tentoxin and versicotide D and the evaluation of their phytotoxicity against Ryegrass (Lolium multiflorum) seeds. The design of the compounds was focused on small modifications of the peptide backbone, exploring the influence on the bioactivity of L-or D-amino acids and N-MeAA.

Synthesis
The synthesis of the linear tetrapeptides, 4-15, as shown in Table 1, and the corresponding cyclopeptides 16-27, as shown in Figure 2, were carried out following the procedure shown in Scheme 1. The methodology was based on the Fmoc-strategy of SPPS (solid phase peptide synthesis), employing the 2-chlorotrityl chloride resin (2-CTC resin). To avoid racemization, and thereby the formation of diastereomers during ring closure, we started most peptide sequences with Glycine at the C-terminus, which also minimizes steric hindrance during the macrocyclization process. The yields and purities of the obtained tetrapeptides are shown in Table 1. HBTU was used for coupling to primary amine Our group has recently reported the phytotoxic evaluation of natural products versicotides, among which versicotide D (3, Figure 1) shows low cytotoxicity on HepG2 cells [31], high phytotoxicity (74% radicle growth inhibition at 67 µg/mL) and the ability to inhibit cyanobacteria population with a substantial depletion of the concentration of microcystins in the media [32]. Here, we report the synthesis of a library of cyclic tetrapeptides inspired on tentoxin and versicotide D and the evaluation of their phytotoxicity against Ryegrass (Lolium multiflorum) seeds. The design of the compounds was focused on small modifications of the peptide backbone, exploring the influence on the bioactivity of L-or D-amino acids and N-MeAA.

Synthesis
The synthesis of the linear tetrapeptides, 4-15, as shown in Table 1, and the corresponding cyclopeptides 16-27, as shown in Figure 2, were carried out following the procedure shown in Scheme 1. The methodology was based on the Fmoc-strategy of SPPS (solid phase peptide synthesis), employing the 2-chlorotrityl chloride resin (2-CTC resin). To avoid racemization, and thereby the formation of diastereomers during ring closure, we started most peptide sequences with Glycine at the C-terminus, which also minimizes steric hindrance during the macrocyclization process. The yields and purities of the obtained tetrapeptides are shown in Table 1. HBTU was used for coupling to primary amine and HATUwhen coupling to an N-MeAA.       In order to maximize the expected outcome of the cyclization reaction, the linear sequences of the peptide precursors were carefully chosen. Gly (compounds 4-8, 11-13) or In order to maximize the expected outcome of the cyclization reaction, the linear sequences of the peptide precursors were carefully chosen. Gly (compounds 4-8, 11-13) or N-MeGly (compounds 9-11 and 14-15) were selected as C-terminals to avoid epimerization of this residue and minimize steric hindrance during the cyclization process.
The anchorage of Fmoc-Gly-OH or Fmoc-N-MeGly-OH to the resin was achieved with excess of DIPEA in CH 2 Cl 2 . For the elongation of the peptide chain, we carried out successive steps of deprotection employing piperidine and amide bond formation, using the corresponding Fmoc-AA-OH and HBTU (N,N,N ,N -Tetramethyl-O-(1H-benzotriazol-1yl)uronium hexafluorophosphate) for coupling to primary amines and HATU for secondary The elongation of the peptide chain was monitored by HPLC; in some cases, additional steps of deprotection or coupling were needed. Once the desired linear tetrapeptides were reached, cleavage with a solution of TFA 1% in CH 2 Cl 2 rendered them as trifluoroacetate salts in very good yields and excellent purities.
Diketopiperazine formation is an important side reaction that can hinder peptide synthesis on solid phase. This usually occurs during the coupling between the second and third amino acid in the sequence. It is well known that the use of 2-CTC resin decreases the formation of diketopiperazine due to the presence of the bulky trityl group [33]. Nevertheless, when the second amino acid is an N-MeAA, the formation of diketopiperazine can be promoted. In order to avoid this, short times during Fmoc deprotection of N-MeAA were used. This procedure allowed us to obtain compounds 7, 12 and 13 in very good yields. However, low yields were obtained for compounds 6, 9 and 11, possibly due to their particular peptide sequences.
Once the linear precursors were obtained as trifluoroacetate salts, we proceeded with the head-to-tail lactamization in solution phase, obtaining the corresponding cyclotetrapetides. It is well known that macrocyclization reaction is a low yielding process. The E-geometry of the amide bond prevents the peptides from adopting a ring-like conformation conducive to cyclization [34,35]. In addition, the cyclization yield is very dependent on the peptide sequence and on the ring size. In general, the cyclization of peptides with less than seven amino acids is a difficult process, even though small cyclic peptides containing a β-turn such as a D-amino acid, an N-MeAA or proline, a thiazole or oxazole ring, etc., can be prepared in higher yields.
To obtain the desired cyclotetrapeptides, high dilution conditions were used, and cyclization reactions were performed for 1-5 days (Scheme 1). Figure 2 and Table 2 show the twelve cyclotetrapeptides analogues to tentoxin (cyclopeptides 16-24) and versicotide D (25-27) that were prepared using HBTU/DIPEA in the cases where the N-terminal residue is a primary amine and HATU/DIPEA or Oxyma/DIC when the N-terminal residue was a secondary amine. To obtain compound 17, we also tried the combination of Oxyma/DIC, instead of HATU or HBTU, but the best results were obtained with HATU. For the cyclization of compound 15, Oxyma/DIC was used and 27 was obtained in 54% yield. Table 2. Peptide precursors, cyclotetrapeptides and macrocyclization yield.

Peptide Precursor Cyclopeptide Macrocyclization Yield (%)
Ala-Leu-D-Phe-Gly (4) 16 5 Ala-Leu-Phe-N-MeGly (6) 18 40 Ala-Leu-NMe-D-Phe-Gly (7) 19 66 It is important to highlight that in the case of 17 vs. 20, and 18 vs. 21, the change of Phe by D-Phe, respectively, leads to a higher cyclization yield. However, 16 containing a D-Phe, was obtained in the lowest yield (5%). Nevertheless, this yield was not optimized and could be improved by the use of others coupling reagents. Taking into account the number of N-MeAA in the peptide precursor, most of the higher yields were for compounds with one N-MeAA (17)(18)(19)(20)(21). An interesting result is the case of the compound 25, which contains two N-MeAA and was obtained in very good yield. In fact, this was the highest yield for the cyclopeptides synthesized in this work. In contrast, when Phe was substituted by Leu (25 vs. 24, respectively), the yield decreased.

Herbicidal Activity
The herbicidal activity of cyclotetrapeptides 16-25, 27 and the linear precursors 4-8, 10, 12-15 were evaluated for their influence on germination, leaf development and radicle length growth of ryegrass seeds at 67 µg/mL using germination in agar methodology [32]. DMSO was used as negative control and the herbicide S-metolachlor as positive control. In this trial, the results showed that root length was the only variable for which significant differences were found after a statistical analysis. The results were expressed as % of radicle growth inhibition, calculated as the percentage between the treatment with each compound and radicle length growth for the negative control (DMSO), as shown in Figure 3. Cyclopeptide 26 was tested at 45 µg/mL and showed 67% inhibition.
The herbicidal activity of cyclotetrapeptides 16-25, 27 and the linear precursors 4-8, 10, 12-15 were evaluated for their influence on germination, leaf development and radicle length growth of ryegrass seeds at 67 μg/mL using germination in agar methodology [32]. DMSO was used as negative control and the herbicide S-metolachlor as positive control. In this trial, the results showed that root length was the only variable for which significant differences were found after a statistical analysis. The results were expressed as % of radicle growth inhibition, calculated as the percentage between the treatment with each compound and radicle length growth for the negative control (DMSO), as shown in Figure 3. Cyclopeptide 26 was tested at 45 μg/mL and showed 67% inhibition. The most active compounds are the cyclotetrapeptide 25, and the tetrapeptides 7 and 6, for which an inhibition of 96%, 94% and 92% at 67 μg/mL was observed, respectively. At this concentration, the cyclization of 13 to obtain 25 leads to an increase in activity. On the contrary, the cyclization of 6 and 10 to render 18 and 22, respectively, leads to a decrease in herbicidal activity.
By comparison of the herbicidal activity of 16 and 19, we concluded that the presence of N-methyl group is relevant for the bioactivity. Moreover, cyclopeptides 20 and 19, which differ in the position of N-methyl group, present root growth inhibition of 74% and 87%, respectively. On the other hand, 17, which presents a Phe instead of a D-Phe, seems to be as active as 20. However, the substitution of Phe in 6 (NH2-Ala-Leu-Phe-N-MeGly-OH) by D-Phe (9) leads to a decrease on the herbicidal activity.
The activity results for cyclotetrapeptides analogues of tentoxin, which contain two N-MeAA (22)(23)(24), allowed us to conclude that the position of this type of amino acid is relevant for the bioactivity, whereas 23 is inactive (inhibition 16%), 22 and 24 present 66% and 81% of inhibition, respectively. It is important to note that unlike 24, 22 and 23 present two contiguous N-MeAA.
In the case of cyclotetrapeptide 24, the change of Leu for Phe (compound 25) leads to an increase on the activity, and for their corresponding linear precursors (12 and 13), this fact becomes more noticeable.
The most active compounds, which exhibited inhibition greater than 85% at 67 μg/mL (compounds 6, 7, 10, 13, 19, 25, Figure 3) and cyclopeptides 17, 20, 24 and 26, were selected to assay them at a lower concentration in order to obtain more experimental results for a deep discussion about chemical structure requirements, as shown in Table 3. The most active compounds are the cyclotetrapeptide 25, and the tetrapeptides 7 and 6, for which an inhibition of 96%, 94% and 92% at 67 µg/mL was observed, respectively. At this concentration, the cyclization of 13 to obtain 25 leads to an increase in activity. On the contrary, the cyclization of 6 and 10 to render 18 and 22, respectively, leads to a decrease in herbicidal activity.
By comparison of the herbicidal activity of 16 and 19, we concluded that the presence of N-methyl group is relevant for the bioactivity. Moreover, cyclopeptides 20 and 19, which differ in the position of N-methyl group, present root growth inhibition of 74% and 87%, respectively. On the other hand, 17, which presents a Phe instead of a D-Phe, seems to be as active as 20. However, the substitution of Phe in 6 (NH 2 -Ala-Leu-Phe-N-MeGly-OH) by D-Phe (9) leads to a decrease on the herbicidal activity.
The activity results for cyclotetrapeptides analogues of tentoxin, which contain two N-MeAA (22)(23)(24), allowed us to conclude that the position of this type of amino acid is relevant for the bioactivity, whereas 23 is inactive (inhibition 16%), 22 and 24 present 66% and 81% of inhibition, respectively. It is important to note that unlike 24, 22 and 23 present two contiguous N-MeAA.
In the case of cyclotetrapeptide 24, the change of Leu for Phe (compound 25) leads to an increase on the activity, and for their corresponding linear precursors (12 and 13), this fact becomes more noticeable.
The most active compounds, which exhibited inhibition greater than 85% at 67 µg/mL (compounds 6, 7, 10, 13, 19, 25, Figure 3) and cyclopeptides 17, 20, 24 and 26, were selected to assay them at a lower concentration in order to obtain more experimental results for a deep discussion about chemical structure requirements, as shown in Table 3. The linear peptide 6, did not show inhibition at 23 µM and 10 presented only 15% of root grow inhibition at 11 µM. All the other compounds (7, 13, 17, 19, 20, 24, 25 and 26) showed considerable inhibition at concentrations between 4 and 11 µM. The open precursors of the cyclopeptides 19 and 25, the peptides 7 and 13, presented similar activity, suggesting that cyclization does not have influence on the bioactivity. However, 26 is a very active compound and its open precursor (14) showed lower activity at 67 µM (53%). At lower concentration (6 µM), cyclopeptides 20 and 17, which differ on the stereochemistry of Phe, present different bioactivity, 32% and 10% inhibition, respectively.
Compound 24 containing two NMe-AA in the (1, 3) relative position in the cycle, presents substantial activity, 62% inhibition at 10 µM. In addition, the open peptide precursor, 12, is one of the less active of this series (Figure 3).

Synthesis
All reactions were carried out under nitrogen atmosphere with dry, freshly distilled solvents, under anhydrous conditions, unless otherwise stated. All solvents were purified following procedures described in the literature [36]. The cyclopeptides were analyzed using HPLC-DAD-ESI-MS/MS Shimadzu LCMS 8040 equipped with LC-20 AD pump, a DGU solvent degasser solvent, a SPD-M20AD detector, a CTO.20A oven, a Sil-20A injector. The mass spectrometer is connecting by a split 4:1 of flow. The data were processed by the Labsolutions LCMS software. Chromatographic analysis were developed using a Kinetex EVO C18 (100 mm × 4.6 mm, 5 µm solid core particle), using a flow of 1.25 mL/min and 40 • C. Analysis were developed by a gradient solvent system using 0.1% formic acid (mobile phase A) and acetonitrile (mobile phase B). The gradient program was: t = 0'8% B, t = 18'90% B.
Analysis of LC-DAD-MS was recorded by UV absorbance in a 220-360 nm range, and by full scan ESI + ions with a range of 200-1000 uma. 1 H and 13 C-NMR spectra were recorded at 25 • C on a Bruker Neo 400 using a BBO z-gradient probe, operating at 400.13 and 100.62 MHz for 1 H and 13 C, respectively.
Solid Phase Peptide Synthesis of linear peptides (4-15): (I) Resin loading: 500 mg of 2-Chlorotrityl chloride resin (2-CTC) were added to a plastic syringe. The resin was swelled in CH 2 Cl 2 (3 × 5 min). A solution of first protected amino acid Fmoc-AA-OH (1 eq. for 0.8 mmol/g loading) and DIPEA (3 eq.) in CH 2 Cl 2 was added and shaken for 10 min. Then, an extra 7.0 eq. of DIPEA were added, and shaking was continued for 50 min. MeOH (0.8 mL/g of resin) was added to the previous mixture and shaken for 10 min. After filtering, the resin was washed with CH 2 Cl 2 (×3), MeOH (×3), CH 2 Cl 2 (×3), DMF (×3). (II) Removal of NHFmoc group: The resin was shaken with piperidine-DMF solution (1:4) (1 × 1 min and 2 × 5 min). In exceptional cases, deprotection step was accomplished by a single treatment with piperidine-DMF solution for 5 min in order to prevent side reactions. (III) Coupling of subsequent N-Fmoc protected amino acids to primary or secondary amines: After removal of NHFmoc-protecting group, as previously described, the resin was washed with DMF (×3), CH 2 Cl 2 (×3) and DMF (×3). Then, a solution of Fmoc-AA-OH (3 eq.) and DIPEA (6 eq.) in DMF was added to the resin, followed by a solution of HBTU, for coupling to primary amines, or HATU (2.9 eq.) in DMF, in case of coupling to an N-methylamino acid. The mixture was stirred for 60 min. After the coupling was completed, the resin was washed with DMF (×3) and CH 2 Cl 2 (×3). Successive deprotection and coupling cycles were made with the appropriate amino acids to obtain the desired compound. The coupling was monitored by colorimetric assays; Kaiser test in case of primary amines and Chloranil test for secondary amines.
Coupling procedure was repeated in case of positive results. (IV) Cleavage: The peptide was cleaved from the resin by treatment with 1% TFA in CH 2 Cl 2 for 2-3 min at room temperature, followed by filtration and collection of the filtrate in MeOH. The treatment was repeated three times and then the resin was washed with CH 2 Cl 2 (×5) and MeOH (×3). Solvents were removed under vacuum to obtain the crude peptide. LC-MS was used to identify the desired product (V) General procedure for macrocyclization in solution phase to obtain (16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27): Method I: Macrocyclization reaction was performed in diluted conditions (1-5 mM) using HBTU or HATU (1.5 eq.), DIPEA (3 eq.), 4-DMAP (catalytic) in dried CH 2 Cl 2 at room temperature during 1-5 days. The reaction mixture was washed with HCl 5%, later with saturated aqueous NaHCO 3 , dried over MgSO 4 , filtered and concentrated in vacuo. The crude was purified by flash chromatography to obtain the pure macrocycle. Method II: The trifluoroacetate salt of the corresponding linear peptide was dissolved in dried CH 2 Cl 2 and diluted to a concentration of 1-5 mM. DIPEA (1 eq.) was added to enable dissolution. EDCI (1.2 eq) and oxyma (1.2 eq.) were added at 0 • C and the reaction mixture was stirred for 10 min. Then, the reaction mixture is allowed to reach room temperature and stirred for 48 h. The reaction mixture was washed with HCl 5%, later with saturated aqueous NaHCO 3 , dried over MgSO 4 , filtered and concentrated in vacuo. The crude was purified by flash chromatography to obtain the pure macrocycle.    13  Cyclo diameter) in three replicates per treatment. Ten ryegrass seeds were germinated in a growth chamber (20 • C, day/night temperature). The seeds were previously sterilized by immersing them in 70% alcohol for 10 s. When distributed in the Petri dish on the agar, the seeds were placed in such a way as to ensure that they remained submerged in the solution.
Agar-water solution was prepared at 0.3%, 3 g of agar was placed in 1 liter of deionized water, and the solution was autoclaved at 100 • C for 45 min. Once the agar medium had cooled to approximately 60 • C, the solutions were prepared.
The negative control-DMSO control was prepared by adding 100 µL of DMSO per plate in 15 mL of agar and then the seeds were distributed, as mentioned above. A control without DMSO was also carried out to check that the product was not altering the correct development of the ryegrass seeds. For this test, 15 ml of agar were placed in each Petri dish and, before it solidified, the seeds of the species evaluated were placed on top. The herbicide treatment, positive control-Control S-Metolachlor (960 g/L), was carried out for a conversion of 1/8 of the dose of 1 L/ha of commercial product. For this purpose, a stock solution of S-Metolachlor was prepared by placing 0.28 mL of the herbicide in a volumetric flask and topping up to 1000 mL. A total of 25 mL of this stock solution were taken, placed in a volumetric flask and brought to 200 mL, thus, generating the 1/8x solution of S-Metolachlor. A volume of 3 mL of 1/8x herbicide solution was mixed with 45 mL of the agar solution to bring 16 mL into each Petri dish. This ensured that there was 1 mL of 1/8x S-Metolachlor solution per plate: 0.28 µL of herbicide solution. Seeds were arranged in the same way. Using the same method, the agar media corresponding to each plate (15 mL) was mixed with the cyclopeptides dissolved in 100 µL of DMSO.
Germination, root length and leaf development were evaluated 12 days after preparation. The variables of germinated plants and plants with developed leaves of the total number of plants placed to germinate were analyzed by fitting a generalized linear model, since they presented a binomial distribution.
Glinmix procedure of the SAS statistical package. Based on the model and for the comparison of the treatments with the different controls, the contrasts of interest were carried out. The effect of the treatments on the root length variable was studied by comparing means using a Tukey test (p-value < 0.05) in INFOSTAT.

Conclusions
In summary, cyclopeptides and their open precursors, analogues of tentoxin and versicotide D were successfully synthesized by SPPS of their linear precursor and solution-phase macrolactamization. After evaluating the herbicidal activity of cyclopeptides analogues of tentoxin, we can conclude that 19 and 24 are the most active of them showing remarkable phytotoxic activity. Cyclotetrapeptides 19 and 24 present one (N-Methyl-D-Phe) and two N-MeAA (N-Methyl-Ala and N-Methyl-Phe), respectively. The cyclopeptide without N-MeAA (16) is inactive. In two cases, the open precursors (7 and 13) are as active as their corresponding cyclopeptide (19 and 25). However, many linear peptides are inactive and their cyclization derivatives showed herbicidal activity. An example of these observed results is 24 and its open precursor 12, which are the most active and one of the less active compounds, respectively. Moreover, the cyclopeptide analogues of versicotide D, 25 and 26, showed improved activity over the natural product.
All these findings allowed us to conclude that the conformation adopted by the peptides and cyclopeptides would have great influence on the herbicidal activity. As a consequence, the peptide sequence, the amino acid stereochemistry and the presence of N-methyl group would play an important role on the phytotoxic activity of this type of compound.