Total Synthesis of GE81112A: An Orthoester-Based Approach

The GE81112 series, consisting of three naturally occurring tetrapeptides and synthetic derivatives, is evaluated as a potential lead structure for the development of a new antibacterial drug. Although the first total synthesis of GE81112A reported by our group provided sufficient amounts of material for an initial in depth biological profiling of the compound, improvements of the routes toward the key building blocks were needed for further upscaling and structure–activity relationship studies. The major challenges identified were poor stereoselectivity in the synthesis of the C-terminal β-hydroxy histidine intermediate and a concise access to all four isomers of the 3-hydroxy pipecolic acid. Herein, we report a second-generation synthesis of GE81112A, which is also applicable to access further representatives of this series. Based on Lajoie’s ortho-ester-protected serine aldehydes as key building blocks, the described route provides both a satisfactory improvement in stereoselectivity of the β-hydroxy histidine intermediate synthesis and a stereoselective approach toward both orthogonally protected cis and trans-3-hydroxy pipecolic acid.


■ INTRODUCTION
The worldwide problem of increasing multidrug resistance is compounded by the decline in new classes of antimicrobials entering into clinical practice. This fact led to renewed interest in novel scaffolds as starting points for new antimicrobial agents. GE81112A (1), 1,2 a nonribosomally synthesized tetrapeptide, and its congeners GE81112B (2) and GE81112B1 (3) (Figure 1) represent such a novel chemical structure. Since the anti-Gram-negative activity is based on ribosome targeting, 2,3 the GE81112 series gained high interest as a potential lead.
The first total synthesis of GE81112A (1) accomplished by our group led to the revision of the originally published structure 4 and provided first batches of 1 on a 100 mg scale for in depth biological profiling. 5 Structure−activity relationships were expanded by the Renata group which developed a concise total synthesis of GE81112B1 (3) and 10 non-natural analogues thereof. 6,7 After careful analysis of our (firstgeneration) route toward 1 and Renata's synthesis of 3, we identified the syntheses of both β-hydroxy amino acid moieties as major areas to be improved.
While a nonstereoselective approach toward all four stereoisomers of the 3-hydroxy pipecolic acid moiety was considered acceptable in the first total synthesis of GE81112A with regard to ambiguity concerning the absolute configuration and planned SAR studies (Scheme 1, entry G), an adapted route featuring Teoc-protected (2S,3R)-hydroxy pipecolic acid 5 in a stereocontrolled manner was required for scale-up and synthesis of derivatives. In the same line, poor stereoselectivity experienced in the synthesis of the C-terminal β-hydroxy histidine building block 22 using Garner's aldehyde (dr 2:1) was identified as an additional issue (Scheme 2). In the total synthesis of GE81112B1 (3) by the Renata group, 6,7 (2S,3R)hydroxy pipecolic acid derivative 6 was prepared from (2S)pipecolic acid (14) by chemoenzymatic oxidation 8 in two steps (Scheme 1, entry H).
Our strategy for a concise stereoselective synthesis of GE81112A (1) is based on a detailed analysis of its revised structure revealing an intriguing feature: both the β-hydroxy pipecolic acid and the β-hydroxy histidine display the same threo configuration. Therefore, we reasoned that it should be feasible to deliver intermediates for both moieties from one common building block defining the required threo configuration in a highly selective manner. We first focused on the βhydroxy pipecolic acid targeting the Teoc-protected cis-(2S,3R)-configured product 5 as key intermediate. Due to the total synthesis of GE81112B1 (3) by the Renata group, Boc-protected compound 6 became a target as well. Besides high yields, technical feasibility, and low number of synthetic steps, the high degree of stereoselectivity and enantiomeric excess (ee) represent important criteria for such an alternative route. While numerous synthetic protocols for 3-hydroxy pipecolic acid derivatives had been reported, our search for cisselective methods resulted in surprisingly few convincing examples (Scheme 1): In a synthesis from Chavan et al. starting from readily available chiral pool material L-ascorbic acid (Scheme 1, entry A 9 ), the advantage of cheap starting materials is completely outweighed by in total 18 synthetic steps and one racemization event. Syntheses from D-sugars such as D-glucose (not shown) 10 (12) affording cis-selectively rac-4 (Scheme 1, entry F 15 ), but separation of the enantiomers remains an issue. 16 In comparison, synthetic strategies based on nucleophilic addition to serine derivatives such as Garner's aldehyde, 17,18 other serinals 19 or a serine Weinreb amide 20 seemed to be more attractive since serine is a cheap chiral starting material readily available in both D-and L-configurations. Since we 4 and others 18,21 experienced only low to moderate diastereoselectivities in addition reactions of various nucleophiles to Garner's aldehyde, we were looking for an alternative approach. In this context, Lajoie's orthoester-protected serine aldehydes 21a-c 22 gained our attention (Scheme 2). Transformation of the carboxylic acid into its corresponding trioxabicyclo [2.2.2] ortho (OBO) ester installs a very bulky group enabling the attack of the nucleophile to the aldehyde moiety in the preferred orientation and at the same time keeping the oxidation state constant. The formed product possesses the syn (or threo) selectivity, which is fully in line with the nonchelation-controlled Felkin−Anh model. Consequently, Lajoie's aldehyde has been successfully applied to the synthesis of several β-hydroxy, 23,24 β-methoxy, 25 and other unusual amino acids. 26 In 2014, Karjalainen and Koskinen reported a scalable route based on Garner's aldehyde 18 yielding trans-(2R,3R)-3hydroxy pipecolic acid 15 in 11 steps with an overall yield of 31% and without the need for chromatographical purification of intermediates (A, Scheme 2). 27 Our initial idea was to adapt the concept of a stereoselective addition of a metal alkyne species to Lajoie's aldehyde 21c combined with a stringent strategy of functionalization and ring closure to minimize the number of necessary steps. Since reductive amination is an established method of piperidine ring formation, 28 we envisaged an addition reaction of an acetal-protected propiolaldehyde 20 to orthoester 21c as the key step to deliver intermediate 19, the precursor for the ring closure to 4. 29 After ring formation under reductive hydrogenation conditions, the free carboxylic acid 4 would be obtained directly after a hydrolysis-saponification sequence (B, Scheme 2) in contrast to the Garner's aldehyde based route A which requires an additional oxidation step. Furthermore, we planned to investigate serinals 21a and 21b as a potential substitute for Garner's aldehyde 18 employed in the first total synthesis of GE81112A for the preparation of βhydroxy histidine intermediate 22. For this purpose, we chose the Fmoc-or Boc-protecting groups since hydrogenolytic Cbz removal would be incompatible with the chlorine present in the histidine moiety (C, Scheme 2).

■ RESULTS AND DISCUSSION
Starting from alkyne 20, 30 metalation with ethyl magnesium bromide followed by reaction with aldehyde 21c at low temperature delivered products 19 and 25 with yields up to 39% and a diastereomeric ratio (dr) of merely 1.8:1 for the desired syn (threo) isomer 19 (Scheme 3). 31 The low yield could be linked to an incomplete deprotonation of the terminal alkyne, which required extensive periods of time. In order to curb the long deprotonation periods, we substituted the Grignard species by a lithium organyl resulting in an increase of the yield to 81% and of the dr of >20:1 for the syn-product 19. The reaction was upscaled to 2g resulting in a yield of 71% and a constant ee compared to utilizing aldehyde 21c (ee = 97%). After recrystallization target, compound 19 was obtained in 60% yield and with an ee > 99%.
The conversion of intermediate 19 into literature-known cis-3-hydroxypipecolic acid methyl ester 27 was achieved by hydrogenation under high pressure (85 bar) in a Thales Nano H-Cube at 70°C using Pearlman's catalyst followed by OBO ester opening and transesterification under acidic conditions (Scheme 3). Reaction control (LC/MS) indicated that hydrogenation of the 1,5-dihydro-3H-2,4-benzodioxepin protecting group is the rate-limiting step which required a hydrogen pressure of 85 bar to obtain full conversion. Additionally, compound 27 was converted to target molecule 5 by Teoc-protection and saponification. Notably, the hydrogenation of 19 was possible under milder conditions (room temperature, atmospheric pressure) when diluted acetic acid was added. We hypothesized that protonation of the basic amine may prevent partial catalyst poisoning and facilitate hydrogenetic cleavage of the 1,5-dihydro-3H-2,4-benzodioxepin protecting group. The crude mixture was treated with aqueous hydrochloric acid to drive the orthoester hydrolysis to completion. The synthesis of the desired derivatives 5 and 6 was completed by carbamate protection of intermediate 29 and saponification (Scheme 3). Analogously, isomer 25 was successfully converted to the trans-3-hydroxypipecolic acid methyl ester 32. 32 In comparison with the successfully established entry to the pipecolic acids 5 and 6, an effective synthesis of β-hydroxy histidine intermediate 22 depended on the yield and stereoselectivity of the addition step of 2-chloro imidazole derivative 24 (prepared by lithiation of commercially available imidazole 33 and trapping of the lithium species with C 2 Cl 6 ) to serinal 21a (Scheme 4). Applying conditions similar to those of the previously described syntheses of alkyne addition products 19/25 gave access to desired stereoisomer 23a and minor diastereomer 34a as an inseparable mixture with a preparative yield of 67% and a synthetically useful selectivity (dr > 5:1). Since the separation of the diastereomers could not be achieved and test reactions demonstrated that the Fmocprotecting group was not compatible to the basic transesterification conditions required for the synthesis of target compound 22 (Scheme 5), we switched to Boc as the protecting group. We were able to increase the yields while reducing the required equivalents of iodo-imidazole 24 which is irreversibly consumed by the iodine-magnesium exchange and cannot be recycled (in contrast to alkyne building block 20 that is metalated by deprotonation cf. Scheme 3). However, the selectivity dropped significantly (dr 1.8:1). Switching to the corresponding lithiated imidazole as a nucleophile retained the selectivity at a dr of 5:1. In contrast to the Fmoc-series, the isolation of pure diastereomer 23b was possible by careful chromatography. 33 To obtain 23b in a preparative useful yield of 62% (after two chromatography runs), careful control of time, temperature, and equivalents of n-BuLi employed was required since the reaction turned out to be prone to decomposition when reaction conditions were modified. The remaining mixture of 23a/34a was converted to 23b and 34b by a one-pot Fmoc-deprotection Boc-protection sequence, and the desired isomer 23b was isolated as a pure compound after chromatography.
Intermediate 23b was converted to TIPS-protected compound 35 followed by OBO ester opening and simultaneous trityl deprotection. The transesterification with K 2 HPO 4 in methanol resulted in methyl ester 37 in good overall yield. 24 Final Boc deprotection completed the synthesis of target compound 22 (Scheme 5). The synthesis of key intermediate 22 was thereby shortened, and the yields were increased from 15% over six steps to 48% over five steps. The sensitivity of OBO-protected intermediates 23b and 35 toward ortho-ester hydrolysis was identified as the only disadvantage hampering chromatography, analytics, and storage of these intermediates.
Additionally, a second-generation synthesis based on the highly stereoselective addition (dr 10:1) of trityl-protected imidazole 24 to Garner's aldehyde (18) was developed by us which gave a significant improvement in overall yield (cf. SI). The desired stereoisomer was obtained in high purity by crystallization on a 20 g scale. Unfortunately, even though the trityl group gave a higher selectivity, a change to the SO 2 NMe 2 protection group was nonetheless required since it was more acid-tolerant. Thus, it was able to withstand the following acetonide cleavage and alcohol oxidation to the carboxylic acid 22 after initial TIPS protection. With these two improved synthetic routes toward intermediate 22 in hand, we not only achieved the formal total syntheses of GE81112A (1), but also secured a robust and scalable material supply of 22, which is crucial for further SAR investigations. Compared to the remarkably short and fully diastereoselective synthetic access to β-hydroxy histidine intermediate 42 by Renata et al. 7 (Scheme 6), our improved routes reached a similar level of convenience while relying solely on chemical means.
The most critical step within the endgame of our firstgeneration synthesis of GE81112A (1) was the saponification of methyl ester 49 (Scheme 7). Intermediates 49 and 50 were quite sensitive to basic conditions (resulting in partial decomposition). Only under well monitored conditions, a moderate yield of 42% for the saponification was achieved, and the reproducibility of that reaction�especially on scale� turned out to be difficult to control. 4 These findings motivated us to modify the protection group strategy and to utilize TMSE-protected building block 44 (for details regarding the synthetic procedure cf. SI) for the endgame (Scheme 7). After coupling 44 with dipeptide 43 (derived in 14 steps from D- xylose and L-His(Trt)-OMe) to 45, the reduced amine 46 could be coupled with building block 5, derived now from the OBO route. With the TMSE ester in place on the C-terminus, the separate saponification step could be omitted, and all fluoride labile protecting groups (TMSE, TIPS, TBS, and Teoc) of protected tetrapeptide 47 were removed in one step by TAS-F under mild and neutral conditions. Thus, compound 48 could be isolated in an acceptable yield. 34 After final trityl deprotection, GE81112A (1) was obtained in 67% yield. Overall, this modification shortened the endgame by one synthetic step, facilitated chromatographic purification due to less side product formation, and improved the yield significantly from 9 to 29% over five steps.

■ CONCLUSIONS
We have demonstrated that Lajoie's OBO-protected serinals 21a-c are readily available and synthetically useful building blocks for the synthesis of β-hydroxy amino acids in threo configuration. Compared to the widely used Garner's aldehyde 18, the high threo selectivity caused by the steric bulk of the OBO-orthoester (in line with the nonchelation-controlled Felkin−Anh model) and no need for a final oxidation of the addition product to the carboxylic acid are prime advantages of the corresponding routes over the established ones. In the synthesis of cis-3-hydroxy pipecolic acids 5 and 6, a concise and selective access was developed, which was in our hands more convenient than other literature-known strategies. Furthermore, Boc-protected serinal 21b enabled us to improve our first-generation synthesis of the β-hydroxy histidine building block 22 significantly. As main drawbacks, the tendency of aldehydes 21a-c to epimerize and the sensitivity of the OBO-orthoester protecting group toward hydrolysis were identified, which require increased care during experimentation and precautionary measures and may be a reason why the use of 21a-c remained a niche in the synthesis of βhydroxy amino acids compared to Garner's aldehyde (18).
Nevertheless, our synthetic studies enabled us to improve the sophisticated total synthesis of GE81112A (1) allowing us to provide the building blocks for synthetic analogues of the GEseries. Since key questions concerning the biological activity and the pharmacological and safety profile of GE81112 remain to be adressed, 5 modification of the structural backbone and therefore synthetical access is of utmost importance to answer these. The reported results helped us to tackle a number of these issues by generating a number of new derivatives which will be reported in due course. ■ EXPERIMENTAL SECTION General Information. All chemicals and solvents/anhydrous solvents were commercially supplied and used without further purification. For heating of reaction mixtures, aluminum flask carriers in different sizes from IKA were used. Reactions were monitored using thin layer chromatography (TLC) or using one of the following LCMS systems: 1100 HPLC (Agilent) with DAD equipped with an Esquire 3000plus MS detector (Bruker) or 1200 HPLC (Agilent) with DAD equipped with MSD (Agilent) ESI quadrupole MS. TLC was performed on precoated silica gel glass plates (Merck TLC Silica gel 60 F254), and compounds were detected under UV light (254 nm) and/or by staining with an aqueous solution of KMnO 4 with K 2 CO 3 and NaOH, or an aqueous solution of phosphomolybdic acid, cerium(IV) sulfate, and H 2 SO 4 , or a solution of ninhydrin in n-BuOH/AcOH (100:3) followed by heating with a heat gun. Products were purified by flash column chromatography using silica gel 60 M (Macherey-Nagel) or silica gel from Merck (particle size: 40−63 μm, 60 Å average diameter) or by using automated flash column chromatography systems (puriFlash XS 520Plus from Interchim or Reverlis PREP device from Buchi) equipped with ISOLUTE Flash SI II columns of different sizes from Biotage or PF-15SIHC flash columns of different sizes from Interchim (eluants are given in parentheses). For all OBO-orthoester-containing compounds, conditioning with 2% NEt 3 in n-heptane prevented OBO-orthoester hydrolysis by the slightly acidic silica. Reverse-phase flash column chromatography was performed on a FlashPure EcoFlex C18 (Bruker) equipped with FlashPure EcoFlex C18 end capped columns (eluants are given in parentheses). Preparative HPLC was performed
Condition 2 (Deprotonation with n-BuLi, Scheme 3, Condition b). Alkyne 20 (0.79 g, 4.5 mmol, 4.0 equiv) was dissolved in THF (15 mL) and then cooled to −78°C. Then n-BuLi (2.5 M in n-hexane, 1.8 mL, 4.5 mmol, 4.0 equiv) was added dropwise, and the reaction mixture was stirred for 10 min. The reaction mixture was then stirred without a cooling bath for 10 min and afterward cooled to −78°C again. Aldehyde 21c (0.36 g, 1.1 mmol, 1.0 equiv) was dissolved in THF (10 mL) in a separate flask, cooled to −78°C, and then added dropwise to the alkyne-solution. The reaction mixture was stirred at −78°C for 0.5 h and was then diluted with saturated aqueous NH 4 Cl (25 mL) and ethyl acetate (80 mL). The layers were separated, and the organic layer was washed with saturated aqueous NaHCO 3 (20 mL) and saturated aqueous NaCl (30 mL) and was dried over MgSO 4 and filtered, and the solvent was removed under reduced pressure. The crude product was purified via column chromatography (silica, conditioning with 2% NEt 3 in n-heptane, 0−100% ethyl acetate in nheptane) to obtain the diastereomeric mixture of 19 and 25 (0.45 g, 0.91 mmol, 81%, dr > 20:1; dr determined by 1 H-NMR) as a colorless solid.

Conditions 2 (Scheme 4, Conditions d).
Iodo-imidazole 24 (836 mg, 1.78 mmol, 3.00 equiv) was dissolved in THF (9 mL) and cooled to −78°C. Then n-BuLi (1.54 M in n-hexane, 1.15 mL, 1.18 mmol, 3.00 equiv) was added dropwise, and the reaction mixture was stirred for 30 min. Aldehyde 21b (170 mg, 0.592 mmol, 1.00 equiv) was dissolved in THF (6 mL) in a separate flask, cooled to −78°C, and added to the lithium reaction solution at −78°C. After 5 h, saturated aqueous NH 4 Cl (10 mL) was added. The solution was warmed to room temperature and diluted with CH 2 Cl 2 (10 mL). The aqueous layer was extracted with CH 2 Cl 2 (3 × 10 mL). The combined organic layers were washed with saturated aqueous NH 4 Cl (10 mL) and saturated aqueous NaCl (10 mL), dried over Na 2 SO 4 , and filtered, and the solvent was removed under reduced pressure. The resulting crude product was purified via column chromatography (silica, conditioning with 2% NEt 3 in petroleum ether, petroleum ether/ethyl acetate 4:1 to 100% ethyl acetate) to obtain 23b and 34b as a colorless solid and a mixture of diastereomers with a dr of 5:1(determined by 1 H-NMR). The diastereomers were separated by a second purification via flash chromatography (25 g PF-15SIHC flash column, conditioning with 2% NEt 3 in petroleum ether, 0−100% ethyl acetate in n-heptane in 60 min) to obtain the desired main diastereomer 23b (231 mg, 0.364 mmol, 62%) as a colorless solid (all fractions containing traces of 23b and the undesired minor diastereomer 34b were discarded).

■ ASSOCIATED CONTENT Data Availability Statement
The data underlying this study are available in the published article and its online Supporting Information.
Detailed experimental procedures including spectroscopic data and determination of enantiomeric excess (PDF) ■ AUTHOR INFORMATION