Chinese Chemical Letters  2019, Vol. 30 Issue (8): 1468-1480   PDF    
Establishing the structure-activity relationship of teixobactin
Eilidh Mathesonc, Kang Jina, Xuechen Lia,b,*     
a Department of Chemistry, State Key Laboratory of Synthetic Chemistry, The University of Hong Kong, Hong Kong, China;
b Laboratory of Marine Drugs and Bioprodcuts, Qingdao National Laboratory for Marine Science and Technology, Qingdao 266237, China;
c School of Chemistry, University of Edinburgh, Edinburgh, UK
Abstract: In 2015, a new antimicrobial peptide agent was discovered, termed teixobactin. Over the past few years, the structure-activity relationship of teixobactin has been extensively studied. Here, the updated studies have been summarized to provide structure-activity relationship established to date. It can be seen that position 1, 2, 5 and 6 of teixobactin are not tolerant of diversion from the native amino acids. In positions 7 and 11, native amino acids give the highest activity but there is tolerance for other amino acids. Positions 3, 4, 9 and 10 are very tolerant of substitution while maintaining good potency and a broad activity spectrum. Activity does not depend on absolute stereochemistry, but on the relative stereochemistry and positions 1, 4, 5, and 8 must contain D-amino acids. The ring and tail structure are necessary for activity, macrolactone and lactam rings are both acceptable. Some teixobactin analogues show greater activity than native teixobactin. All conducted animal studies show positive results with no animal deaths.
Keywords: Teixobactin     Antibacterial     Peptide synthesis     Structure-activity relationship     Cyclic peptides    
1. Introduction

The production and accessibility of antibiotics is, arguably, one of the greatest feats of modern medicine. However, the ease of availability and use of antibiotics could cause their own downfall. Overuse and incorrect use have led to the acceleration of antibiotic-resistant genes emergence and spreading among bacteria. The World Health Organisation warns that we could be entering a "post-antibiotic era, in which common infections and minor injuries can once again kill" [1]. Within EU countries alone, an estimated 25, 000 people die annually from antimicrobial resistant bacterial infections (ARBI) [2]. By 2050, 10 million people are expected to die from ARBI every single year. The resultant medical care will cost the world economy $10 trillion [3]. The need for new antibiotic discovery and development has never been more pressing. Apart from continuous efforts on modifying the existing antibacterial drugs to restore their efficacy against resistant strains, search for new structural motifs with new antibacterial mechanisms is also actively being explored. Cyclic peptides are very important resources for the development of novel and potent antibiotics. Many natural cyclic peptides exhibit powerful antibacterial activity and low toxicity [4-6], some of which have already been approved to be used in clinic. In this review, the structure-activity relationship studies of teixobactin, a promising natural cyclic peptide, have been summarized, which will be helpful for the new drug discovery.

1.1. Teixobactin discovery

In 2015, teixbaction was discovered by Lewis and coworkers using a new approach, iChip, screening antibacterial agents from nature. This method of cultivating bacterial species has enabled bacteria that would previously not survive in lab conditions to be studied [7]. IChip has allowed for a mass screening of bacteria, approximately 10, 000 different isolated species, looking for activity against S. aureus. A β-proteobacteria showed encouraging results. DNA sequencing identified this strain as a new gram-negative species called Eleftheria terrae, belonging to an entire new genus of bacteria. Mass spectrometry studies found that an isolated compound with antibacterial activity from E. terrae did not match any chemical databases. NMR and Marfey's analysis revealed the structure of this new compound. It was named teixobactin [8]. Teixobactin showed potent activity against gram-positive bacteria, with the minimum inhibitory concentration (MIC) of 0.25 μg/mL for methicillin resistant S. aureus (MRSA) [8]. MRSA MIC values of 0.5–1.1 μg/mL have been quoted for the synthetic teixobactin [9, 10].

1.2. Teixobactin structure

Teixobactin is composed of 11 amino acids (Fig. 1). In its biological synthesis, thioesterase catalyses esterification between terminal isoleucine and threonine, forming a four-amino acid ring [8]. Positions 4, 5 and 8 are D-amino acids: D-glutamine, D-allo-isoleucine and D-threonine respectively. All other amino acids are L-enantiomers. Interestingly, four of the 11 amino acids are isoleucine at positions 2, 5, 6 and 11. Position 10 is occupied by L-allo-enduracididine. Enduracididine is rarely found in nature. It has only ever been isolated in cyclic peptides, all with antimicrobial properties [11].

Download:
Fig. 1. The structure of teixobactin.

1.3. Teixobactin resistance and mode of action

Under the laboratory conditions, bacterial resistance to teixobactin has not been observed, which benefits from its mechanism of action. It is common for antimicrobials to target bacterial proteins. Genes encoding proteins are prone to mutations, allowing the bacteria to overcome any antimicrobial ability of the antibiotic. However, teixobactin does not inhibit enzymes or protein acceptors, but binds to the precursors involved in the production of peptidoglycan and teichoic acid [8]. As a result of inhibiting the production of two cell components, bacteria must achieve multiple mutations in the biosynthesis pathway to change the structure of the building component, thus development of bacterial resistance is slow, as in the case of vancomycin. As shown in Fig. 2, it was found that teixobactin binds to lipid Ⅱ and lipid Ⅲ. Non-competitive binding to these precursors results in their inability to be used in biosynthesis of peptidoglycan (lipid Ⅱ) and teichoic acid (lipid Ⅲ). Peptidoglycan is necessary for cell wall formation, the absence of which leads to an incomplete and damaged cell wall [8]. Teichoic acid and autolysins both bind to peptidoglycan, giving them a reciprocal relationship where the presence of teichoic acid results in increased delocalisation of autolysins. Teixobactin results in the absence of both peptidoglycan and teichoic acid, disrupting the normal relationship and artificially increasing autolysin delocalisation. The free autolysins prompt cell lysis [12]. The accumulation of effects caused by teixobactin's lipid Ⅱ and lipid Ⅲ binding, ultimately leads to bacterial cell death.

Download:
Fig. 2. The proposed binding sites of teixobactin.

The Wen group hypothesised that membrane insertion is the primary mechanistic step, followed by lipid binding [13]. This conclusion came from low abundance of the nature lipid Ⅱ, approximately making up only -1% of bacterial cellular lipids [14]. Molecular modelling and molecular dynamic simulations have revealed more about teixobactin's possible mode of action. The D-allo-Ile5 and Ile6 residues act through hydrophobic interactions as membrane anchors. Two conformations were discovered upon teixobactin binding to lipid Ⅱ. Both of these contained crucial binding interactions between the phosphate moiety on lipid Ⅱ and amide groups on the C-terminal cyclodepsipeptide ring. The phosphate group also acted through hydrogen bonding of the hydroxyl group on Ser7 [13].

This molecular modelling can provide an insight into teixobactin's bactericidal mechanism. Corroboration between this and structure activity relationship studies could provide definitive confirmation on teixobactin's novel mode of action. Understanding this can aid the progression of teixobactin into clinical studies and could provide a pathway to discover new antibiotics acting through the same mechanism. Given the absence of bacterial resistance to teixobactin, this could prove extremely significant in the fight against ARBI [8].

1.4. Structure activity relationship (SAR)

Understanding the relationship between the structure of teixobactin analogues and their activities against different strains of bacteria is imperative for the development of teixobactin-based therapeutics. Establishing the required amino acids and their positions reveals the binding motif, the section(s) of the peptide that interacts with the target molecules, lipids Ⅱ and Ⅲ, and are essential for bacterial activity. It can also reveal which positions are more tolerant, where either a group of amino acids or any amino acid can be substituted without any significant effect on activity. Knowing the SAR can help to streamline and increase the efficiency of analogue screening processes. The end goal is to find the analogues with the highest and broadest spectrum of activity, to take forward, further into the antibiotic development process.

This article reviews published literatures outlining the existing analogues of teixobactin. It will also look at the activity against gram-positive and gram-negative bacteria of the different analogues. The important relationship between the structure of teixobactin and its activity will be discussed. This leads on to the future implications for the development and clinical trials of teixobactin.

2. Teixobactin synthesis 2.1. Challenges with chemical synthesis

Teixobactin is a cyclic tetra-depsipeptide anchored with a septipeptide, a structure with modest complexity for chemical synthesis. The three main problems presented in total synthesis include synthesis of L-allo-enduracidine, esterification between DThr and Ile, and cyclisation to form the tetra-depsipeptide.

2.1.1. L-allo-Enduracidine synthesis

Toward total synthesis of teixobactin, one of the main challenges is the preparation of commercially unavailable L-allo-enduracididine (L-allo-END). Until now, many groups have successfully completed the synthesis of this unnatural amino acid. In 1975, the first synthesis of enduracididine (END) was reported by Shiba et al. [15] As shown in Fig. 3a, L-END was obtained as a mixture of two diastereomers by 4 steps. In 2014, Ling and co-workers also prepared the different diastereomers of END building block and confirmed the correct configuration [16].The synthetic route is shown in Fig. 3b. L-allo-END could be obtained by 4 steps in a high yield. However, L-END was produced at the same time with a 1:6 ratio to the desired product.

Download:
Fig. 3. (a) Synthesis of L-allo-END by Shiba's group [15]; (b) synthesis of L-allo-END by Ling's group [16].

In order to prepare the END building blocks suitable for the total synthesis of teixobactin, the first highly stereoselective and scalable preparation of protected L-allo-END was developed by Yuan et al. in 2015 [17]. Their synthesis was started from commercially available Boc-trans-Hyp-OH (Fig. 4). With the pyrrolidine oxidation and reductive ring-opening as the key steps, L-allo-END was obtained by 10 steps in 31% overall yield with excellent stereoselectivity (dr = >50:1). In 2019, Xu et al. also used a similar strategy to prepare the L-allo-END building block and continued to complete the total synthesis of teixobactin [18].

Download:
Fig. 4. Synthesis of L-allo-END by Yuan's group [17].

In 2016, Payne et al. reported another synthetic route of L-allo-END building block by using Boc-protected L-Asp ester as the starting material [10]. In this strategy, the critical C4 stereochemistry was achieved through a stereoselective reduction of ketone by using L-selectride. The minor diastereoisomeric by-product could be separated through flash column chromatography. After the subsequent guanination and intramolecular cyclization, the Bocprotected L-allo-END was obtained with the correct configuration. Then the following acidic deprotection and Fmoc-installation afford the target building block in an overall 21% yield (Fig. 5). Reddy et al. also prepared the L-allo-END building block in gramscale by using a similar strategy [19]. In their method, the guanidine moiety was fully protected by three Cbz groups. Unfortunately, these Cbz protecting groups were unable to be completely removed by hydrogenation. As a result, the total synthesis of teixobactin could not be carried to the end.

Download:
Fig. 5. Synthesis of L-allo-END by Payne's group [10].

Although L-allo-END synthesis is a laborious and low-yielding process, the synthesis of this unusual amino acid opened the possibility to total synthesis of native teixobactin.

2.1.2. Esterification

Another challenge in total synthesis of teixobactin is the esterbond formation between D-Thr and Ile during the solid-phase peptide synthesis (SPPS). In 2015, Albericio et al. reported the first synthesis of a teixobactin analogue with END10 being replaced with Arg (Fig. 6) [20]. In this study, it was found that the key onresin esterification was very difficult to be achieved through a general SPPS procedure. And the esterification of D-Thr and Ile only took place smoothly by using DIC and DMAP as coupling reagents. Several couplings were also required to give an acceptable yield of this step. In order to prevent the O to N-acyl migration during the SPPS, the Ile was also protected as Nα-Alloc rather than Nα-Fmoc. In following studies, other groups have adopted this method and completed their synthesis successfully [21].

Download:
Fig. 6. Synthesis of a teixobactin analogue via on-resin esterification by Albericio's group [20].

In 2016, Li et al. reported another strategy to accomplish the esterification between D-Thr and L-Ile (Fig. 7) [9]. In their total synthesis of teixobactin, depsipeptide Alloc-D-Thr-O(Fmoc-Ile)-OH was first prepared through solution-phase coupling and then loaded onto 2-Cl-Trt resin. Alloc and Fmoc were used as two orthogonal protecting groups. And this method enabled the extension of peptide from both C terminus and N terminus. In this case, the possible de-esterification of the resin-bound peptide was also not observed in the subsequent SPPS steps.

Download:
Fig. 7. Solution phase esterification between D-Thr and L-Ile by Li's group [9].

2.1.3. Cyclisation

Due to the constrained geometry of cyclo-tetra-depsipeptide, the macrocyclization is a very difficult step in the synthesis of teixobactin and its analogues. There are two requirements in macrocyclization: (a) high cyclization efficiency; (b) minimal racemization. In previous references, different cyclization points (point Ⅰ, Ⅱ and Ⅲ) have been selected taking into account different requirements (Fig. 8) [9, 20]. Due to the lower efficiency of esterification than lactamization, the point Ⅳ between D-Thr and Ile has not been chosen for macrocyclization.

Download:
Fig. 8. Different cyclization points in teixobactin.

In most cases, the macrocyclization was carried in solution phase and many different coupling methods have been used, for example, HATU/DIEA in DMF/DCM [21, 22], PyAOP/OxymaPure/ DIEA in DMF/DCM [20], HATU/HOAT/OxymaPure/DIEA in DCM [9, 23, 24], DMTM·BF4/DIEA in DMF [7], HBTU/HOBT/DIEA in CAN/ THF/DCM [25], and so on. In order to avoid the dimerization, the cyclization was always carried in a very dilute solution [9, 20, 21].

Some on-resin cyclization methods have also been reported. In 2017, de la Torre et al. synthesized a series of END10Lys-teixobactin analogues by using on-resin cyclization (Fig. 9a) [26]. It was started by incorporating Fmoc-Lys-OAllyl into the 2-Cl-Trt resin. After the esterification between Alloc-Ile-OH and D-Thr, the Allyl and Alloc protecting groups were removed with Pd(0). Then the macrocyclization was carried out by using PyAOP/OxymaPure/DIEA in DMF for 2 h and gave an excellent yield (>95%). Su et al. developed another effective cyclization strategy on solid phase (Fig. 9b) [27]. Their works were started by attaching Fmoc-Ala-OH to Fmoc-Hydrazinobenzoyl AM resin. After SPPS, the hydrazide groups were oxidized by Cu(OAc)2 to form transient acyl diazene species. Then an intra-molecular nucleophilic attack occurred simultaneously to offer the cyclized product.

Download:
Fig. 9. Synthesis of teixobactin analogues via on-resin cyclization. (a) de la Torre's work [22]; (b) Su's work [23].

2.2. Total chemical synthesis

The total synthesis of teixobactin was first reported by both Li et al. and Payne et al. at the same time in 2016. As shown in Fig. 10, Payne's group used SPPS combined with L-allo-enduracididine precursor synthesis to give the linear teixobactin precursor. Macrolactamisation of this precursor gave the final teixobactin product in 3.3% yield [10]. Since this synthesis, SPPS-based teixobactin synthesis has been repeated by other groups. Chen and co-workers utilised solution-phase for esterification between Ile11 and Thr8 followed by N + 1 SPPS for all remaining residues [28].

Download:
Fig. 10. Total synthesis of teixobactin by Payne's group [10].

In parallel, the Li group published their strategy for total synthesis (Fig. 11). This involved convergent Ser-Thr ligation between the linear hexapeptide containing residues 1–6 and a C-terminal salicylaldehyde ester, with the cyclic depsi-pentapeptide, containing residues 7-11. For the cyclic depsi-pentapeptide, D-Thr8 and Ile11 were coupled in the liquid phase, the resulting depsipeptide was deprotected and loaded onto the chlorotrityl resin.The otheramino acidson this segment were coupled through Fmoc-SPPS. The linear hexapeptide was fully constructed by BocSPPS. These two fragments were coupled with pyridine:AcOH (6:1 molar ratio) reagents to give a teixobactin in a yield of 37% [9]. The synthetic route for Fmoc-End(Cbz)2-OH outlined by Yuan et al. was utilised for the L-allo-END10 residue [19]. Other research groups have utilised convergent strategies, Zong and co-workers performed SPPS synthesis of the linear hexapeptide and cyclic depsi-pentapeptide. The two fragments were cleaved from resin before solution phase coupling using DIEA (3 equiv.), HATU (3), HOAT (3 equiv.), DCM:DMF (9:1) for 24 h [29].

Download:
Fig. 11. Total synthesis of teixobactin by Li's group [9].

In 2019, the first total solution-phase synthesis of teixobactin was completed by Xu et al. (Fig. 12). Retrosynthetic analysis disconnected teixobactin into 4 segments, each of which were synthesised before convergence. The resulting 20-step synthesis gave a resulting yield of 5.6% [18].

Download:
Fig. 12. Total solution-phase synthesis of teixobactin by Xu's group [18].

2.3. Chemoenzymatic synthesis

Chemoenzymatic approaches to teixobactin synthesis have also been explored by Chen et al. The linear peptide precursor was constructed through SPPS before thioesterase catalysed the etherification between D-Thr8 and L-Ile11-COOMe. This resulted in the formation of the cyclic tetrapeptide, and hence, teixobactin. Experimental work determined that the conversion of residues 1, 4 and 5 to their L-enantiomeric counterparts resulted in successful cyclisation. However, there was no observed cyclisation of the linear peptide containing the L-Thr8 mutation. This reveals that analogue synthesis is some-what limited by substrate selectivity for residue 8. A truncated linear peptide, without N-terminal residues 1–3 was successfully cyclised by thioesterase. It is theorised that two forms of thioesterase, TE1 and TE2, are required for biosynthesis [30]. The proposed mechanism is shown in Fig. 13. Since its discovery, a number of groups have successfully synthesised teixobactin through many different routes. There are still advancements that need to be made, specifically with yield, to push teixobactin and teixobactin analogues into clinical trials. Yet, the current progress has been made in only a short period of time, highlighting the potential for teixobactin.

Download:
Fig. 13. Proposed mechanism of chemoenzymatic approaches to teixobactin synthesis [26].

3. L-allo-Enduracididine: Rare, but important?

As mentioned, teixobactin contains L-allo-enduracididine in position 10; this rare amino acid became a significant point of interest for research groups. Several groups have created analogues alternating position 10; L-allo-enduracididine. This rare amino acid and its importance to activity has become a source of fascination. END was hypothesized to bind to the phosphate group of the lipid Ⅱ or lipid Ⅲ, thus the basicity of END was very likely important for the teixobactin activity. To develop simpler teixobactin analogues, basic amino acid residues have been used as potential END isosteres.

The Albericio group was the first to complete total synthesis of a teixobactin analogue where L-allo-End10 was replaced with isosteric L-Arg10. The resulting peptide still showed antimicrobial activity. However, compared to natural teixobactin, the MIC was considerably higher against all the bacteria tested. For S. aureus, the MIC increased from 0.2 nmol/L to 1.6 nmol/L. The L-Arg10 analogue is eight times less active than native teixobactin. For B. subtilis, the MIC increased from 0.016 to 0.40, a 25-fold decrease in activity [20]. Although L-Arg10 analogues are less active than natural teixobactin, it is still significant that activity was observed. Enduracididine is not commercially available and the synthesis is a ten-step process with a documented final yield of only 31% [17]. When mass producing and screening analogues, this is neither desirable nor sustainable. The activity of L-Arg10 consequently means that analogues containing this amino acid can be efficiently produced and screened. This allows the structure activity relationship of other amino acid positions to be explored and determined.

During the total synthesis study of teixobactin, the Li group also reported the teixobactin analogue with End10Orn, which exhibited four times less active against MRSA than teixobactin.

The Wu group went further into probing the role of position 10. They hypothesised that the guanidinium functional group on both arginine and enduracididine is necessary for activity. Lys10 showed the same MIC as L-Arg10 for S. aureus (2 μg/mol) and no activity against E. coli. However, the MIC increased from 0.25 μg/mL to 0.5 μg/mL for B. subtilis [27]. Conversely, Chen et al. determined the activity of Lys10 as higher than Arg10 [25]. The Orn10 -teixobactin analogue reported by Li et al. also exhibits activity against S. aureus comparable to the Arg10 analogue [23] Singh and co-workers tested the MIC against MRSA, S. aureus and B. subtilis for 8 analogues containing isosteres of L-allo-enduracididine. All showed activity, most within a range close to that of teixobactin [31]. This shows the high tolerance of position 10 for cationic, amine side-chain amino acids. In contrast, for the His10-teixobactin analogue, synthesised by Wu et al., a more significant decrease in activity was observed. The S. aureus MIC for His10 (8 μg/mL) is four times as high as the equivalent for L-Arg10 (2 μg/mL), the observed MIC against B. subtilis also increased, 2 μg/mL versus 0.25 μg/mL. Again, there was no observed activity against gram-negative E. coli [27]. The Wu group noted that they had observed that activity of teixobactin analogues does appear to depend on the basicity of the amino acid in position 10. They concluded that position 10 is tolerable to isosteres of enduracididine, with either a guanidinium or anime side-chain [26]. This is necessary to maintain the basicity of this position and the resultant cation formed.

However, an alanine screening by Chen et al. led to an unexpected outcome. They found that Ala10-teixobactin showed good activity [32]. Alanine is non-polar and uncharged, lacking both amine and guanidinium. It was previously believed that position 10 had to be a basic amino acid, resulting in the cationic interaction. It is now understood that position 10 is tolerable to basic and non-polar amino acids, while maintaining good activity. This was agreed by Singh et al. who synthesised a number of analogues with polar, non-polar and cationic L-amino acids in position 10, including Arg, Phe, Gly and Leu. All showed inhibition of MRSA in the range of 0.25–2 μg/mL [2] (Table 1).

Table 1
MIC results of AA10-teixobactin.

Li and co-workers have synthesised a total of 32 analogues with non-isostere substitution of L-allo-enduracididine10 (Fig. 14), allowing a thorough and conclusive study into the relationship between position 10 and activity (Table 2) [24]. Several conclusions were drawn from this study. Substitution in position 10 with the anionic amino acids Asp and Glu resulted in a complete loss of activity. Analogues containing the hydrophilic amino acids Ser, Thr, Gln and Asn sawa significant activity reduction. Position 10 is not tolerantof anionic or hydrophilic amino acids. In position 10, hydrophobic amino acids with aromatic or very short side-chains resulted in severely reduced or even no activity. Hydrophobic amino acids with larger side-chains, however, resulted in little or no reduction [24]. The activity reported for Ile10 and Leu10 agrees with that reported by Singh et al. [17, 30]. This provides confirmation that hydrophobic amino acids are acceptable in position 10. Five analogues with metasubstitution in the aromatic ring in Phe10 were synthesised and screened for bacterial activity. Substitution with Cl, F and NO2 gave analogues with equivalent activity to the reference Phe10-teixobactin. Substitution with CH3 and Br did not affect the activity against S. aureus but eliminated activity against MRSA. Substitution of OH with methoxy in the para position in the aromaticring in Tyr10 saw better activity against S. aureus than the reference Tyr10-teixobactin, however, there was a complete lack of activity against MRSA. Meta substitution with NO3 produces an analogue with double the activity of Tyr10-teixobactin against both MRSA and S. aureus. The group synthesised Gly10-teixobactin, Ala10-teixobactin and a further three analogues with an increasing length of linear aliphatic side chain. Up to three carbons, there was a positive correlation between carbon chain length and activity. Nle10-teixobactin containing a four carbon long chain, however, show half the activity of the Nva10-teixobactin analogue, with the three carbon side chain. Nva10-teixobactin did also express higher potency than native teixobactin; 1–2 μg/mL and 0.25 μg/mL for MRSA MIC and S. aureus MIC respectfully. This data shows that the chain length for optimal potency and activity is three carbons. Three analogues containing cyclopropane and cyclohexane side chains showed equivalent potency levels to native teixobactin. Chg10-teixobactin had MIC values of 1 μg/mL against MRSA and 0.25 μg/mL against S. aureus. This analogue containing a methylcyclohexane side chain displays double the activity of teixobactin. One further analogue exhibited higher potency and activity than teixobactin. Substitution of L-allo-enduracididine10 with the methylthiolethyl containing amino acid, Met, doubled the activity against both MRSA and S. aureus [24].

Download:
Fig. 14. The structures of non-isostere substitution of L-allo-enduracididine in position 10 [24].

Table 2
MIC results of teixobactin analogues with non-isostere substitution of L-allo-END [24].

These studies prove that L-allo-enduracididine10 is not required for high potency, a broad range of cationic and hydrophobic amino acids are acceptable with no or little reduction in activity. Results from the Li group reveal that aliphatic amino acids are preferred over those containing aromatic groups [24]. Studies on the sidechain length suggest that when position 10 is occupied by a hydrophobic residue, a very precise hydrophobicity level is vital for optimised potency [24]. Mechanistic studies are required to determine if cationic10-teixobactin and hydrophobic10-teixobactin have the same mode of action. The tolerance of position 10 opens the door for more synthetically and economically desirable teixobactin analogues. Position 10 screening has facilitated the discovery of three analogues which exert higher activity against MRSA and S. aureus isolates than native teixobactin [24].

4. The four isoleucines

Albericio et al. began a lysine screening process to help determine whether introducing a cation in positions 2–7, 9 and 11 could improve activity. This involved the Arg10-teixobactin analogue and replacing each of the respective amino acids in the mentioned positions with L-lysine. There has allowed the necessity of the isoleucine in positions 2, 5, 6 and 11 to be investigated. In all four positions, replacing isoleucine with lysine eliminated any activity (Table 3) [33]. This study proves that replacing the hydrophobic isoleucine with basic lysine is not tolerable in any isoleucine position.

Table 3
MIC results from lysine substitution in positions 2–9 and 11 in Arg10-teixobactin [33].

The question still stood to whether positions 2, 5, 6 and 11 could only contain isoleucine or if other hydrophobic amino acids are tolerated. Chen et al. conducted alanine screening of Lys10-teixobactin, in positions 1–7, 10 and 11. Alanine is non-polar and uncharged. This determined whether positions 2, 5, 6 and 11 could be hydrophobic amino acids, other than isoleucine. Table 4 shows the results. In position 2, lysine substitution resulted in a decrease in activity for all bacterial strains tested. Substitution in position 5, resulted in no inhibition of S. aureus and an MIC increase of a factor of 16 against B. subtilis. Again, for position 6, there was a major activity decrease. The reported MIC is >32 μg/mL for all bacterial strains tested. Position 11 on the other hand, allows for some diversion from Ile11, a two to four-fold MIC increase is seen for all bacteria [32]. The loss of activity through isoleucine replacement shows the importance of isoleucine in the pharmacophore. Although classed as non-polar and uncharged, a shorter aliphatic side-chain means alanine is less hydrophobic than isoleucine. This suggests that positions 2, 5, 6 and 11 need a highly non-polar amino acid for good activity. Chen et al. suggested a link between the overall solubility of analogues and activity. They concluded there is a negative correlation between the solubility of analogues in phosphate-buffered saline and their activity. The reasoning of this is the enhanced binding to the hydrophobic bilipid cell membrane layer [32]. Current literature suggests that L-Ile2, D-allo-Ile5, L-Ile6 and L-Ile11 give the highest activity against a wide range of gram-positive bacteria. However, L-Ala11 still displays activity, implying amino acid 11, one quarter of the cyclic tetrapeptide, is tolerant of hydrophobic amino acids closely related to isoleucine.

Table 4
MIC results of alanine scan teixobactin analogues [32].

5. Positions 3, 4, 7 and 9

Table 3 shows MIC results for the basic lysine substitution in positions 3, 4, 7 and 9. All four positions natively contain noncharged amino acids. Polar L-serine in position 3, D-glutamate in 4, L-serine in 7 and the non-polar L-alanine in position 9. Lys3-Arg10-teixobactin saw a decrease in activity against S. aureus compared to the Arg10-texiobactin reference. However, the MIC of B. subtilis decreased from 0.5 μg/mL to 0.25 μg/mL, and the MIC of E. coli decreased from 64 μg/mL to 32 μg/mL. For both bacterial strains, Lys3-Arg10-teixoabctin has double the activity of the reference. Although Lys4-Arg10-teioxbactin sees a decrease in the activity against both S. aureus and B. subtilis, it also shows a doubling in the activity against E. coli (MIC of 32 versus 64 μg/mL). Lys substitution in position 9 also sees a decrease in activity against the grampositive bacterial strains. There is a slight decrease in MIC for E. coli, 50 μg/mL for Lys9-Arg10-teixoabctin compared to 64 μg/mL for Arg10-teixobactin. Substitution in position 7 did prompt a more notable activity loss for S. aureus, B. subtilis and E. coli than the other three positions discussed [33]. This is evidences that this position is more sensitive to substitution. Although there is a higher required concentration for bacterial growth inhibition, this inhibition is still observed, demonstrating that position 7 does have a degree of tolerance for basic amino acids.

All four lysine substitutions exhibit activity against P. aeruginosa that did not exist in the reference Arg10-teixobactin [33]. This broad spectrum of activity is highly desirable in clinical application. The ability of these substitutions to increase the spectrum of activity, such that analogues can be effective against both grampositive and negative is a considerable breakthrough and highlights the importance of analogue screening in drug development.

Alanine screening conducted by Chen et al. further reveals the structure activity relationship of positions 3, 4 and 7. As position 9 is natively occupied by alanine, testing this position would not reveal anything further. The resultant MIC against 5 g-positive bacterial strains and the gram-negative E. coli is detailed in Table 2. Ala3-Lys10-teixobactin shows the same activity against the bacterial strains as the reference Lys10-teixobactin. The only variation being the increased MIC against E. durans from 2 μg/mL to 4 μg/mL [32]. This confirms what was discovered by Albericio et al. [33]. Position 3 is accepting of substitution. Polar, non-polar, uncharged and basic amino acids can occupy this position with little effect on activity. Similar observations were made for the alanine substitution of position 4. D-Ala4-Lys10-teixobactin showed a slight reduction in activity for all bacterial strains. The activity against E. coli is quoted as >32 μg/mL for both the reference and D-Ala4-Lys10-teixobactin analogue. However, MIC testing at higher concentrations of the peptide is required to conclusively say that D-Ala4 substitution had no effect on activity against E. coli. The small observed decrease in activity against gram-positive bacteria caused by alanine and lysine substitution and the increase in activity against gram-negative bacteria caused by lysine substitution highlights the tolerance of position 4 [32]. The site can be occupied by non-polar, polar and cationic amino acids while displaying impressive activity. Mirroring the results of the lysine scan, substitution of alanine in position 7 decreases activity. For example, the MIC against S. aureus increases more than 64-fold (0.5 μg/mL for the reference Lys10-teixobactin, >32 μg/mL for Ala7-Lys10-teixobactin) [32]. Position 7 can be considered moderately tolerant. Substitution of a cationic or non-polar amino acid reduces the activity against gram-positive bacteria but does not completely eliminate it. The Lys7 substituted analogue did actually show improved activity against gram-negative bacteria. This agrees with the Wen group who proposed that the hydroxyl group in the Ser7 binds through hydrogen bonding to the phosphate group on lipid Ⅱ [13]. The amine group in Lys7 would allow also allow for hydrogen bonding to lipid Ⅱ.

Aware of the tolerance of positions 3, 4 and 9, the Singh group created analogues with the cationic arginine substituted in these positions. Ile10-texibactin and Leu10-teixobactinwere used as the base teixobactin template. This would maintain the overall number of hydrophobic and cationic residues as in the native teixobactin. Singh et al. hoped that this would lead to analogues with equivalent or improved potency. Arg3-Leu10-teixoabctin, D-Arg4-Leu10-teixoactin and Arg9-Leu10-teixobactin all exhibited a MIC of 0.125 μg/mL for MRSA, these analogues exhibit higher activity than both the Leu10-teixobactin reference and native teixobactin. D-Arg4-Ile10-teixobactin saw the same activity improvement, whereas, Arg3-Ile10-teixobactin and Arg9-Ile10-teixobactin had the same MIC values against MRSA as the reference and native teixobactin. Multiple Arg substitution mutations has a negative effect on potency. Arg3-Arg9-Leu10-teixobactin, D-Arg4-Arg9-Leu10-teixobactin and Arg3-D-Arg4-Arg9-Leu10-teixobactin all displayed MIC of 1 μg/mL for MRSA. This is a four-fold decrease in activity [34]. This demonstrates the significance of the number of hydrophobic and cationic residues in optimising activity. Positions 3, 4 and 9 are highly tolerant of both cationic and hydrophobic residues. As demonstrated by Singh et al. this opens up the opportunity to design and produce analogues with a higher potency than natural teixobactin [34]. This stresses the importance in structure activity relationship studies to ascertaining the most potent analogues to take into clinical studies.

6. D-NMe-Phe1

The N-terminal of the peptide has captured researchers' interest; particularly the presence of the methyl group on the terminal nitrogen. The Wu group used the Arg10-teixobactin analogue to probe the necessity of this methyl group. D-NPhe1-Arg10-teixobactin analogue was synthesised, with an amine N-terminal. This showed similar bacterial activity to the reference Arg10-teixobactin: both have MIC of 2 μg/mL against S. aureus and >128 μg/mL against E. coli, there was only a slight increase in MIC against B. subtilis from 0.25 μg/mL for the reference to 0.5 μg/mL for D-NPhe1-Arg10-teixobactin. They also synthesised and tested the dimethylated N-terminal analogue; D-Me2NPhe1-Arg10-teixobactin. This methylation caused a significant decrease in activity. The MIC against S. aureus is given as 32 μg/mL. Wu et al. suggested that at least one NH proton is required for successful binding to the lipid targets. They also tested the effect of the substitution structurally similar tyrosine. Tyrosine differs from phenylalanine by a para-hydroxyl group on the aromatic ring. Tyr10-Arg10-teixobactin showed low activity: 128 μg/mL against S. aureus. This activity loss suggests the NH terminal is involved in hydrogenbonding with the molecular targets. The OH disrupts the hydrogen bonding pattern and overall polarity of the peptide, disfavouring bonding to lipids Ⅱ and Ⅲ [27]. Zong et al. proposed that D-NMe-Phe1 is also involved in membrane anchoring. The phenyl ring bonds hydrophobically to the lipid chain and the terminal NH interacts through hydrogen bonds to the phosphate membrane sections [29]. The Albericio group synthesised D-NAc-Phe1-Arg10; with the N group acetylated. This analogue showed no activity against the tested bacterial strains, apart from some low activity against B. subtilis (256 μg/mL) [35]. This further confirms that the polarity of position 1 needs to be maintained for peptide bactericidal activity.

Several groups have also investigated the Phe amino acid in position 1 and if it has a role in the pharmacophore. Albericio et al. tested the analogue D-NMe-Lys1-Arg10, containing the cationic Lys amino acid in place of the non-polar Phe. This analogue displayed no activity against any of the bacterial strains tested [36]. The alanine scan conducted by the Chen group shows that D-NMe-Ala1-Lys10 did not inhibit the growth of the bacterial strains tested [32]. These results confirm that the N-terminal must contain Phe for an active teixobactin analogue.

The Rao group further investigated the effect of para-substitution of the aromatic ring in non-methylated D-Phe10. In comparison to the reference D-NMe-Phe1-Arg10-teixobactin (2 μg/mL against S.aureus), substitution with basic NH2 resulted in decreased activity (>32 μg/mL), the para-methylated analogue showed equivalent activity (2 μg/mL) and para-substitution with another phenyl group increased the activity (0.5 μg/mL) [29]. This mirrors the importance of hydrophobicity in position 1, observed by Wu et al. [27]. The presence of anionic or cationic groups on the phenyl group in D-NMe-Phe1, decreases the hydrophobicity of position 1 and hence, the activity of the antibiotic. Substitution of hydrophobic aliphatic and aromatic groups in the same position conservesor increasesactivity. D-NMe-Phe1 plays avital rolein bonding tolipids Ⅱ and Ⅲ, and its high hydrophobicity must be maintained for the successful bactericidal activity of teixobactin. The Zong group then carried out para-substitution of D-para-Ph-Phe1-Lys10-teixobactin. Substitution with hydrophobic phenyl, methyl, n-propyl and hydrophilic CF3 showed very little deviation from the reference potency. This showed that any alternation made on the outer phenyl had little to no effect on the overall activity of the peptide. Zong et al. discovered a positive correlation between the hydrophobicity of the D-NMe-Phe1 residue and bacterial activity of the peptide. This supported their theory that this residue is involved in hydrophobic interactions to lipid Ⅱ, high hydrophobicity is required for activity [29].

7. Stereochemistry: what does it matter?

The importance of the 4 D-amino acids tothe pharmacophore of teixobactin has become intriguing to research groups. Singh et al. synthesised analogues where D-NMe-Phe1, D-Gln4 and D-allo-Ile5 where replaced bytheir L-enantiomeric counterparts. L-NMe-Phe1-Arg10 gave a resultant MIC against S. aureus of 64 μg/mL, a 32-fold decrease in activity compared to the Arg10-teixobactin reference. Both L-Gln4-Arg10 and L-allo-Ile5 showed no inhibition of S. aureus growth [37]. The importance of the D-Thr8 stereochemistry was tested by the Yang group. L-Thr8-Arg10-teixobactin showed no activity against tested bacterial strains [25]. These results confirm the importance of the D-amino acids to the pharmacophore and to an active antimicrobial peptide. Peptides with higher numbers of D-amino acids are less prone to enzyme-mediated proteolysis [38]. This mayapply to teixobactin, contributing to the higher activity of teixobactin analogues containing the D-amino acids. However, the D-amino acids most likely contribute to the 3D conformation of teixobactin and hence, the binding to lipid Ⅱ and Ⅲ. NMR studies conducted by Singh et al. on the L-amino acid analogues they synthesised showed that the D-amino acids have a role in maintaining an amorphous teixobactin. This is what allows it to have maximum interactionwith the target molecules. D-NMe-Phe1 has a less significant effect on the structural flexibility than D-Gln4 and D-allo-Ile5. This matches the inhibition activity observed for substitution of each amino acid [37].

Yang et al. synthesised ent-Arg10-teixobacin. This follows the amino acid sequence of Arg10-teixobactin. However, the stereochemistry of each amino acid is inverted. Each L-amino acid is replaced with the corresponding D-amino acid and vice versa. Ent-Arg showed good activity. The MIC values matched those of the Arg10 reference, with only the MIC against S. epidermis increasing from 1 μg/mL to 2 μg/mL [25]. This determines that is not the absolute stereochemistry of the teixobactin that is of importance, but rather the relative stereochemistry. The available literature outlines the importance of stereochemistry of teixobactin to its activity. Not just the stereochemistry of individual amino acids, but the overall, relative stereochemistry of the peptide.

8. Ring and tail

Teixobactin contains a 4-mer macrolactone ring, attached tothe 7-mer tail. How important are both the ring and the tail to target binding and bacterial inhibition? The Yang group tested seco-Arg10-teixobactin. The analogue has no ester bond between Ile11 and D-Thr8, hence, the ring was uncyclised. The analogue demonstrated no bacterial activity. This proves the macrolactone ring is required for activity and matches the group's proposal that teixobactin binds to lipid Ⅱ through hydrogen bonding to the macrolactone ring. They also synthesised a truncated form of the peptide; missing amino acids 1-5. This also showed no activity. Addition of n-dodecanoyl to the truncated teixobactin created lipobactin 1. Interestingly, this did show activity. Against the reference, the MIC values for all tested bacteria were four times higher. Lipobactin 1 has a quarter the potency of Arg10-teixobactin [25]. However, the presence of activity reveals the role of the tail. The hydrophobicity of the tail is of clear significance tothe mode of action; allowing teixobactin to enter the cell plasma and attract its hydrophobic lipid targets. Lipobactin 1 paves the way for a new set of teixobactin analogues, replacing 5 amino acids with dodecanoyl is synthetically more desirable and is a new route to be explored.

As previously discussed, Zong et al.had successfully synthesised D-para-Ph-Phe1-Lys10-teixobactin, which showed good activity against MRSA with a given MIC of 0.5 μg/mL. A derivative of this analogue was synthesised, replacing the hydroxy group with amine in D-Tyr8, this gives a resultant lactam ring. This oxygen to nitrogen substitution caused an increase in peptide potency. The MIC halved to 0.25 μg/mL. The group hypothesised this was due to a more favoured interaction between the lactam ring and the lipid pyrophosphate component [29].

9. In-vivo studies

The balance of low cytotoxicity, high potency and no detectable resistance certainly makes teixobactin a likely candidate for clinical trialling, with the end goal of clinical application and circulation [8]. Furthermore, a number of teixobactin analogues have been synthesised that show improved potency [24, 29, 34]. These exciting discoveries show the real potential of teixobactin and its relatives in the war against bacterial resistance. Then again, animal physiology and metabolism mean that there is no guarantee that a potent anti-microbial agent will have any ability to treat disease in-vivo.

The first animal studies on teixobactin were published alongside its originally discovery. Mice were injected with a MRSA dose that untreated, would cause 90% mortality within 48 h. One hour post-infection, the mice were administered teixobactin within a dose range from 1 μg/kg to 20 μg/kg. All animals survived. Further animal tests determined the PD50 (protected dose for 50% survival) to be 0.2 μg/kg. This showed favoured results to vancomycin treated mice, a PD50 of 2.75 μg/kg was determined for MRSA treatment in mice [8]. Vancomycin is a circulating drug used primarily for MRSA treatment. This is an example of the capacity for teixobactin to replace current antibiotics that are soon to be rendered useless by antibiotic resistance.

Subsequently, research groups have ventured into in-vivo studies of teixobactin analogues [29, 34]. Upon discovery of a highly potent, non-cytotoxic, D-Arg4-Leu10-teixobactin analogue, shown below, this was selected as a suitable candidate for animal studies, conducted by Singh et al. Topical application of D-Arg4-Leu10-teixobactin reduced the S. aureus population by >99.0% in mouse models of bacterial keratitis, preventing permanent eye damage and vision loss. Corneal thickening and edema formation is an indicator of disease progression and vision loss. The mean starting thickness was reported as 93.8 ± 2.9 μm. In the infected mouse treated with only phosphate-buffered saline (PBS) this increased to 151.7 ±12.7 μm 24 h and 186.2 ±17.5 μm 48 h posttreatment. The mean corneal thickening was less significant in mice treated with D-Arg4-Leu10-teixobactin, 92.3 ± 12.5 and 121.7 ± 3.2 μm for 24 h and 48 h post-treatment, respectively. This corneal thickening was also tested for moxifloxacin-treated mice; 124.2 ± 9.4 and 140.3 ± 10.3 μm for 24 h and 48 h post treatment [34]. These results show that D-Arg4-Leu10-teixobactin is an effective agent in the treatment eye-borne S. aureus infections, it even showed improved corneal edema results compared to moxifloxacin, a circulating antimicrobial therapeutic [34, 39].

Zong et al. also took forward an analogue to mouse model testing. D-para-Ph-Phe1-Lys10-teixobactin showed equivalent activity to teixobactin making it a viable candidate for in-vivo testing. Female mice were injected with 0.2 mL of Streptococcus pneumoniae D39, an adequate dose to inflict 90% mortality within 48 h. An hour after the bacterial infection was administered, some mice in the group were administered the D-para-Ph-Phe1-Lys10-teixobactin, while the control mice were administered only the antibiotic vehicle. After 48 h of infection, all the vehicle-only control mice had died while the survival rate for the teixobactin analogue administered group was 100% [19].

10. Conclusion

Understanding the teixobactin SAR is important to comprehend and fully explain its mode of action. It is a crucial step forward towards the clinical application of a teixobactin antibiotic. This review has discussed existing teixobactin analogues and provided an analysis of activity results. This has built the picture of the SAR (Table 5).

Table 5
Summary of SAR studies of teixobactin.

Fig. 15 shows teixobactin's structure with the necessary pharmacophore in red. The interestingly element of the SAR is not just the individual amino acids but the larger structure of teixobactin. The macrolactone ring is required to enable hydrogen bonding to lipid Ⅱ, although a macrolactam ring. The tail must be predominantly hydrophobic to allow anchorage to the cell membrane. This explains the required hydrophobicity of D-NMe-Phe1, D-allo-Ile5 and Ile6. The relative stereochemistry is essential for the amorphous 3D teixobactin structure and hence, for binding and activity. The findings of this structure activity relationship review match the proposed bactericidal mechanism, found through molecular modelling by the Wen group [13]. Since its discovery four years ago, the volume of literature on teixobactin shows the hope and excitement that it has brought to the scientific community. Animal studies have proved highly successful [8, 24, 29, 34]. This testifies that teixobactin should be taken into clinical trialling, with the intent of becoming a fully circulation antibiotic therapeutic. Teixobactin holds real potential to be a significant effective counter-attack in the on-going war against antibiotic resistance.

Download:
Fig. 15. Tolerance of amino acids in teixobactin toward substitution.

Acknowledgements

This work was supported by the Research Grants Council of Hong Kong (No. C7038-15G), and the Area of Excellence Scheme of the University of Grants Committee of Hong Kong (No. AoE/P-705/16).

References
[1]
World Health Organisation: Antibiotic Resistance, http://www.who.int/newsroom/fact-sheets/detail/antibiotic-resistance (accessed Nov 2018).
[2]
N. Ragnar, M. Powell, B. Aronsson, et al., ECDC/EMEA Jiont Technical Report, 2009, https://doi.org/10.2900/2518.
[3]
J. O'Neill, Review on Antimicrobial Resistance: Tackling a Crisis for the Health and Wealth of Nations, Review on Antimicrobial Resistance, London, in: https://amr-review.org/sites/default/files/AMR%20Review%20Paper%20-%20Tackling%20a%20crisis%20for%20the%20health%20and%20wealth%20of%20nations_1.pdf.
[4]
Y. Li, K. Clark, Z. Tan, Chin. Chem. Lett. 29 (2018) 1074-1078. DOI:10.1016/j.cclet.2018.05.027
[5]
J. Tang, J. Lu, Q. Luo, H. Wang, Chin. Chem. Lett. 29 (2018) 1022-1028. DOI:10.1016/j.cclet.2018.05.004
[6]
H. Luo, H. Yin, C. Tang, P. Wang, F. Liang, Chin. Chem. Lett. 29 (2018) 1143-1146. DOI:10.1016/j.cclet.2018.05.033
[7]
T. Kaeberlein, K. Lewis, S. Epstein, Science 296 (2002) 1127-1129. DOI:10.1126/science.1070633
[8]
L. Ling, T. Schneider, A. Peoples, et al., Nature 517 (2015) 455-459. DOI:10.1038/nature14098
[9]
K. Jin, I.H. Sam, K.L. Po, et al., Nat. Comm. 7 (2016) 12394. DOI:10.1038/ncomms12394
[10]
A. Giltrap, L. Dowman, G. Nagalingam, et al., Org. Lett. 18 (2016) 2788-2791. DOI:10.1021/acs.orglett.6b01324
[11]
D. Atkinson, B. Naysmith, D. Furkert, M. Brimble, Beilstein J. Org. Chem. 12 (2016) 2325-2342. DOI:10.3762/bjoc.12.226
[12]
T. Homma, A. Nuxoll, A. Brown-Gandt, et al., Mbio 60 (2016) 6510-6517.
[13]
P. Wen, J. Vanegas, S. Rempe, E. Tajkhorshid, Chem. Sci. 9 (2018) 6997-7008. DOI:10.1039/C8SC02616E
[14]
M.D. Hartley, B. Imperiali, Arch. Biochem. Biophys. 517 (2012) 83-97. DOI:10.1016/j.abb.2011.10.018
[15]
T. Shinichi, K. Shiochi, S. Tetsuo, Chem. Lett. 4 (1975) 1281-1284. DOI:10.1246/cl.1975.1281
[16]
A. Peoples, D. Hughes, L. Ling, et al., WO Patent: WO2014/089053A1.
[17]
W. Craig, J. Chen, D. Richardson, R. Thorpe, Y. Yuan, Org. Lett. 17 (2015) 4620-4623. DOI:10.1021/acs.orglett.5b02362
[18]
B. Gao, S. Chen, Y. Hou, et al., Org. Biomol. Chem. 17 (2019) 1141-1153. DOI:10.1039/C8OB02803F
[19]
S. Dhara, V. Gunjal, K. Handore, D. Srinivasa Reddy, Eur. J. Org. Chem. 25 (2016) 4289-4293.
[20]
Y. Jad, G. Acosta, T. Naicker, et al., Org. Lett. 17 (2015) 6182-6185. DOI:10.1021/acs.orglett.5b03176
[21]
A. Parmar, A. Iyer, C. Vincent, et al., Chem. Comm. 52 (2016) 6060-6063. DOI:10.1039/C5CC10249A
[22]
A. Parmar, A. Iyer, S. Prior, et al., Chem. Sci. 8 (2017) 8183-8192. DOI:10.1039/C7SC03241B
[23]
K. Jin, K. Po, S. Wang, et al., Bioorg. Med. Chem. 25 (2017) 4990-4995. DOI:10.1016/j.bmc.2017.04.039
[24]
K. Jin, K. Hiu Laam, W. Kong, et al., Bioorg. Med. Chem. 26 (2018) 1062-1068. DOI:10.1016/j.bmc.2018.01.016
[25]
H. Yang, K. Chen, J. Nowick, ACS Chem. Biol. 11 (2016) 1823-1826. DOI:10.1021/acschembio.6b00295
[26]
S. Abdel Monaim, E. Ramchuran, A. El-Faham, et al., J. Med. Chem. 60 (2017) 7476-7482. DOI:10.1021/acs.jmedchem.7b00834
[27]
C. Wu, Z. Pan, G. Yao, et al., RSC Adv. 7 (2017) 1923-1926. DOI:10.1039/C6RA26567G
[28]
L. Liu, S. Wu, Q. Wang, et al., Org. Chem. Front. 5 (2018) 1431-1435. DOI:10.1039/C8QO00145F
[29]
Y. Zong, X. Sun, H. Gao, et al., J. Med. Chem. 61 (2018) 3409-3421. DOI:10.1021/acs.jmedchem.7b01241
[30]
D. Mandalapu, X. Ji, J. Chen, et al., J. Org. Chem. 83 (2018) 7271-7275. DOI:10.1021/acs.joc.7b02462
[31]
A. Parmar, A. Iyer, D. Lloyd, et al., Chem. Commun. 53 (2017) 7788-7791. DOI:10.1039/C7CC04021K
[32]
K. Chen, S. Le, X. Han, J. Frias, J. Nowick, Chem. Commun. 53 (2017) 11357-11359. DOI:10.1039/C7CC03415F
[33]
S. Abdel Monaim, Y. Jad, E. Ramchuran, et al., ACS Omega 1 (2016) 1262-1265. DOI:10.1021/acsomega.6b00354
[34]
A. Parmar, R. Lakshminarayanan, A. Iyer, et al., J. Med. Chem. 61 (2018) 2009-2017. DOI:10.1021/acs.jmedchem.7b01634
[35]
S. Abdel Monaim, Y. Jad, G. Acosta, et al., RSC Adv. 6 (2016) 73827-73829. DOI:10.1039/C6RA17720D
[36]
S. Abdel Monaim, E. Ramchuran, A. El-Faham, F. Albericio, B. de la Torre, J. Med. Chem. 60 (2017) 7476-7482. DOI:10.1021/acs.jmedchem.7b00834
[37]
A. Parmar, S. Prior, A. Iyer, et al., Chem. Commun. 53 (2017) 2016-2019. DOI:10.1039/C6CC09490B
[38]
R. Tugyi, K. Uray, D. Iva'n, et al., Proc. Natl. Acad. Sci. U. S. A. 102 (2005) 413-418. DOI:10.1073/pnas.0407677102
[39]
G. Keating, L. Scott, Drugs 64 (2004) 2347-2377. DOI:10.2165/00003495-200464200-00006