Synthesis of nonracemic hydroxyglutamic acids

Glutamic acid is involved in several cellular processes though its role as the neurotransmitter is best recognized. For detailed studies of interactions with receptors a number of structural analogues of glutamic acid are required to map their active sides. This review article summarizes syntheses of nonracemic hydroxyglutamic acid analogues equipped with functional groups capable for the formation of additional hydrogen bonds, both as donors and acceptors. The majority of synthetic strategies starts from natural products and relies on application of chirons having the required configuration at the carbon atom bonded to nitrogen (e.g., serine, glutamic and pyroglutamic acids, proline and 4-hydroxyproline). Since various hydroxyglutamic acids were identified as components of complex natural products, syntheses of orthogonally protected derivatives of hydroxyglutamic acids are also covered.


Introduction
L-Glutamic acid (1, Figure 1) plays an important role in the biosynthesis of purine and pyrimidine nucleobases [1]. It also takes part in metabolic transformation to L-glutamine by L-glutamate synthetase (GS) which is crucial for cell maintenance.
In neoplastic cells synthesis of L-glutamine is interfered as a result of reduced activity of GS [2]. γ-Glutamyl transpeptidase (GGT) which catalyses transfer of the γ-glutamyl group from glutathione is another enzyme relevant in cancer. High activities of GGT are observed during neoplastic transformation [3].
Several derivatives of L-glutamic acid functioning as anticancer agents have been reported [4]. But primarily L-glutamic acid is known as the major excitatory neurotransmitter in central nervous system which acts by binding to glutamate receptors [5][6][7]. However, these interactions are linked to several neurodegenerative diseases (Alzheimer [8], Huntington [9], Parkinson [10]) as well as to stroke [11] and epilepsy [12].
Two main classes of receptors, each of them containing three subclasses which are further divided into subtypes have been established for glutamic acid. To understand the physiological role of each receptor, recognition of their specific ligands is necessary. This, in turn, may pave a way for development of drug candidates for future therapeutic applications. These goals can be achieved by synthesis of glutamic acid analogues modifying the structure of 1 through installation of additional substituents, tuning the conformational flexibility of analogues and introducing groups capable of hydrogen bonding. Crystallographic data obtained for glutamate receptors [13][14][15] showed complex set of atoms interacting electrostatically and through hydrogen bonds and the conclusions from these studies should facilitate the development of new ligands.
In terms of mapping of glutamate receptors hydroxyglutamic acids 2-4 ( Figure 2) should be of great interest since an additional hydroxy group is capable of acting as a hydrogen bond donor as well as a hydrogen bond acceptor. In fact (2S,4S)-3 showed similar potency at mGlu 1a R and mGlu 8a R as L-glutamic acid [16] while its affinity for AMPA and NMDA receptors was low [17]. On the other hand, (2S,4R)-3 demonstrated significant preference for the NMDA receptor [17]. Furthermore, it was found that (2S,3S,4S)-4 acts as a selective agonist of mGluR1 and as a weak antagonist of mGluR4 [18]. Excitatory amino acid transporters (EAAT) are effected by hydroxyglutamic acid in various degrees. For example, (2S,4S)-3 appeared to be a substrate at EAAT1-3, while (2S,4R)-3 did not interact with them [19,20]. A number of studies revealed that several giant neurons of the African giant snail appeared to be sensitive to various extents to all stereoisomers of 2 [21][22][23].
Hydroxyglutamic acids are widely spread in nature, especially in plants but they were also found in other species or as components of more complex molecules of interesting biological activity. Indeed, the interest in 3-hydroxyglutamic acid started many years ago by the discovery of this amino acid in hydrolysates of an antibiotic peptide S-520 [24]. It has been proved recently that it was actually the isomer (2S,3R)-2 and it is a fragment of a cyclohexapeptide [25]. (2R,3S)-2 and (2R,3R)-2 were found as components of antifungal and antimicrobial hexadepsipeptides called kutznerides isolated from the actinomycete Kutzneria sp. 744 [26,27]. And finally, threo-3hydroxyglutamic acid was identified in the cell wall of Mycobacterium lacticum [28].
Natural occurrence as well as possibilities of glutamate-like biological activity modulated by additional hydrogen bonding with hydroxy groups inspired the interest in the synthesis of stereoisomers of hydroxyglutamic acids 2-4 ( Figure 2). Since they contain two or three stereogenic centers their orthogonally protected derivatives could be considered as extremely valuable chirons in syntheses of various natural products. Their 1,2-and 1,3-aminohydroxy fragments can serve as pharmacophores of interest in medicinal chemistry. In this paper we wish to review chemical syntheses of non-racemic 3-hydroxy-(2), 4-hydroxy-(3) and 3,4-dihydroxyglutamic acid (4) to summarize achievements in this area. The protected forms of 3-hydroxyglutamic acid are of significant value as intermediates in the synthesis of complex peptides.
The majority of asymmetric syntheses of 3-hydroxyglutamic acid employ serine or similar three-carbon chirons as starting materials. Configuration at Cα is retained in the final products and it also induces chirality at the Cβ(OH) center. The hydroxymethyl group of serine can serve as a precursor of the carboxyl fragment but when oxidized to aldehyde it may be attacked by nucleophiles to introduce the required two-carbon residue.

From serine-derived precursors
When Garner's aldehyde (R)-5 prepared from D-serine was subjected to ZnCl 2 -catalyzed cyclocondensation with Danishefsky's diene a (>9:1) mixture of diastereoisomeric pyranones 6 was formed with the threo isomer 6a prevailing. Oxidative removal of two carbon atoms was followed by formate hydrolysis, formation of methyl ester and silylation to give 7 after separation from the minor diastereoisomer. After selective hydrolysis of the acetal the hydroxymethyl fragment was oxidized and all protective groups were removed to give (2S,3R)-2 as the hydrochloride (Scheme 1). The observed stereoselectivity of the cyclocondensation step is best explained by the attack on a re-face of the C=O group due to chelation of Zn 2+ to the carbonyl oxygen and amide nitrogen/oxygen atoms [49].
A better approach in terms of carbon atom economy relied on the addition of allylmagnesium chloride to the aldehyde (R)-5 which after O-benzylation provided an inseparable 1:3 mixture of compounds 8a and 8b. A six-carbon chain was shortened by a diol formation-diol cleavage sequence followed by aldehyde oxidation and esterification to give 9a and 9b after chromato-graphic separation. They were transformed into (2S,3R)-2 and (2S,3S)-2 in several steps including hydroxymethyl to carboxyl oxidations (Scheme 2) [50].
N-Fmoc protection of the amino group in L-serine together with transformation of the carboxylic function into an orthoester allow for the racemization-free oxidation to aldehyde 10, which was immediately subjected to Reformatsky reaction to give a 92:8 mixture of (2S,3R)-11 and (2S,3S)-11, respectively. The Treatment of a trilithium salt of the N-protected acid 13 derived from L-serine with allylmagnesium bromide provided ketone 14 which was reduced to diastereoisomeric diols in a 9:1 syn to anti ratio when L-selectride was applied. They were separated as isopropylidene derivatives and the syn isomer 15 was subjected to ozonolysis and oxidation to give acid 16. To complete the synthesis of di-tert-butyl ester of (2R,3S)-2 compound 16 was first transformed into the ester and later deprotected to the diol 17 which was selectively oxidized, again esterified and finally the phenylsulfonyl group was removed electrochemically (Scheme 4) [52].
The aldehyde (S)-18 prepared from O-benzyl-L-serine in three standard steps [53] was elongated by a two-carbon fragment employing a Wittig reaction to give Z-alkene 19. To introduce the next center of chirality of the required configuration a iodocyclocarbamation reaction was applied to give trans-oxazolidin-2-one (4S,5S)-20 after reduction of the carbon-iodine bond formed in the primary products of cyclization (via iodonium ion 22). Hydrogenolytic debenzylation preceded oxidation of the hydroxymethyl group to afford diester (4R,5S)-21 which after hydrolysis gave (2R,3S)-2 as the hydrochloride (Scheme 5) [53]. Starting from O-benzyl-D-serine (2S,3R)-2 was obtained in a similar way.
Configurationally stable D-serinal derivative (R)-23 (prepared from D-serine [54]) which primarily exists as hemiacetal was subjected to cis-olefination with Stille's reagent at −30 °C to produce (S)-24 in good yield. However, when the reaction mixture was warmed to 0 °C before quenching, an intramolecular cyclization occurred under basic conditions to give the oxazolidine (4S,5R)-25 as an almost (>20:1) pure diastereoisomer. The hydroxy group which acted as a nucleophile preferred to attack the re-face of the double bond for steric reasons. Selective removal of the silyl protective group allowed for the hydroxymethyl to carboxyl transformation to (4S,5R)-26, and hydrolysis afforded (2S,3R)-2 as the hydrochloride (Scheme 6) [55].

From homochiral aziridine
An interesting approach to protected (2S,3R)-2 makes use of the aziridine (2R,1′S)-27 as a synthetic equivalent of L-serine (Scheme 7) [56]. Stereoselective reduction of ketone (2R,1′S)- 28 gave hydroxyaziridine 29 as the major (10:1) product which, after the protection of the hydroxy group, was subjected to the regioselective aziridine ring opening, catalytic removal of the chiral auxiliary with simultaneous formation of a N-Boc derivative 30. The hydroxymethyl to carboxylate transformation to form the protected diester (2S,3R)-31 required prior basic deacetylation followed by standard oxidation and esterification. Diastereoisomer (2S,3S)-31 was also prepared employing the same methodology.

From N-Boc-D-phenylglycinal
Since the phenyl group has been applied for many occasions as a precursor of the carboxylic function selection of D-phenylglycine as a starting material in the synthesis of the N-Boc-protected (2S,3R)-2 makes a useful addition to the existing methodologies (Scheme 8) [57]. Thus, N-Boc-D-phenylglycinal (R)-32 was in situ treated with benzylmagnesium chloride to give N-Boc-aminoalcohol (1R,2R)-33 as a major (9:1) product easily separable chromatographically. Before oxidative degradation of both phenyl groups (1R,2R)-33 was protected as an acetonide. Intermediary diacid was first esterified with diazomethane, then the isopropylidene acetal was hydrolyzed, and diester saponification gave N-Boc-protected compound (2S,3R)-35.

Via ketopinic acid functionalized 2(3H)-oxazolones
When oxazolone 36 derived from (R)-(−)-ketopinic acid was reacted with bromine and trimethyl orthoacetate the enantiomerically pure bromomethoxy derivative (4R,5R)-37 was prepared after crystallization of the reaction mixture. The precursor of the carboxymethyl group was first introduced with full retention of configuration employing a stannate chemistry to give (4R,5R)-38 after removal of a chiral auxiliary with lithium dibutylcuprate. Next, titanium tetrachloride-catalyzed cyanation secured another carboxy group and after a few transformations an oxazolidinone (4S,5R)-39 was obtained as a major (7:1) product readily purified chromatographically. To complete the synthesis of (2S,3R)-2 N-Boc protection preceded the cleavage of the oxazolidine ring while silylation of the hydroxy group was necessary before oxidation of the C=C bond (Scheme 9) [58].
Further applications of the ketopinic acid framework as a chiral auxiliary relied on fine tuning of the steric environment around the carbonyl group. Thus, when compound 40 prepared using readily available (S)-(+)-ketopinic acid was reacted with phenylselenyl chloride in methanol the adduct 41 was formed with high diastereoselectivity (de 96%) and was later separated chromatographically. Further transformations into dimethyl ester of (2S,3R)-2 involved attachment of allyl and vinyl groups to form (4R,5R)-42 which was next oxidized to diacid and finally esterified to give dimethyl ester of (2S,3R)-2 as the hydrochloride (Scheme 10) [59].

By formation of the pyrrolidine ring
Important synthetic strategies towards 3-hydroxyglutamic acids take advantage of the intermediary formation of the pyrrolidine ring. Addition of the dianion of 43 to acrolein gave a 69:31 mixture of diastereoisomers with compound 44 predominating which was easily separated on silica gel. When imine 45 was treated with iodine a stereoselective iodolactamization occurred to produce lactam 46 having the same configurations as found in (2S,3R)-2. To complete the synthesis of (2S,3R)-34 first the iodomethyl group was transformed in two steps into the hydroxymethyl moiety, both hydroxy groups were silylated, the chiral auxiliary was removed and the amide nitrogen was protected as N-Boc to furnish (4R,5R)-47. Under basic conditions the pyrrolidin-2-one ring was cleaved to provide a five-carbon chain of the target molecule. The final steps included esterification, desilylation and selective oxidation of the hydroxymethyl group followed by esterification (Scheme 11) [60]. From L-malic acid (S)-Acetoxypyrrolidin-2,5-dione (51), readily available from L-malic acid [62], was carefully reduced and immediately acetylated to (S)-52 which was reacted with furan to produce a 67:33 mixture of readily separable (2S,3S)-53 and (2R,3S)-53, respectively. Steric hindrance of the acetoxy substituent controls the formation of higher amounts of the trans-isomer.
The other strategy which also commences from L-malic acid [64] showed much better carbon atom economy since the acetate (S)-55 was reacted with cyanide while to the acetate  From D-glucose D-Glucose may be used as a chiral template for the synthesis of (2S,3R)-2 since configurations at C3 and C4 in the hexose are retained in the target compound. The disclosed strategy relied on prior transformation of D-glucose into azidofuranoside 57 [66] and next to acid 58. Homologation of acid 58 was accomplished by the Arndt-Eistert reaction to give the methyl ester 59 from which benzyl ester 60 was obtained for easy hydrogenolytic removal in the last step. Hydrolysis of the isopropylidene acetal was followed by periodate cleavage of the C1-C2 bond in the furanose, oxidation of the already formed aldehyde to the acid and basic hydrolysis of the formate to afford the acid (2S,3R)-61. Its allylation provided the ester (2S,3R)-62, a protected precursor of 3-hydroxyglutamate, from which (2S,3R)-2 can be prepared by catalytic hydrogenolysis (Scheme 15) [67].
In connection with the total synthesis of thiopeptide antibiotic nosiheptide an orthogonally protected (4S)-4-hydroxy-Lglutamic acid derivative 66 (Scheme 16) was required and it was obtained as a single diastereoisomer from 65 in the same way [75,76]. when compared with 69 and (2S,4S)-3 [78]. Two other stereoisomers were synthesized in a similar way from (R)-67.

By a nitrone-acrylate cycloaddition
The isoxazolidine ring can be considered as another cyclic precursor to 4-hydroxyglutamic acids due to the easy cleavage of the N-O bond and high trans diastereoselectivities of 1,3dipolar cycloadditions which allow to control stereochemistries at C3 and C5 [79,80]. To illustrate this concept the E/Z mixture of nitrone 70 was reacted with acrylamide 71 prepared from (2S)-bornane-10,2-sultam to afford mainly (20:1) the isoxazolidine (3S,5S)-72 easily separable from minor cycloadducts. The trans stereochemistry of the isoxazolidine ring in 72 was the consequence of the endo and exo additions to the Zand E-nitrones, respectively [80]. Further steps to the orthogonally protected (2S,4S)-73 required selective hydrolysis of the chiral auxiliary, installation of the tert-butyl ester function and finally hydrogenolytic opening of the isoxazolidine ring with simultaneous protection of the amino group (Scheme 18).   [82]. To synthesize (2S,4S)-81 the inversion of configuration at C4 executed by Mitsunobu reaction preceded oxidation at C5 and the ring opening [82]. O-Benzyl ethers of (2S,4R)-3 and (2S,4S)-3 were prepared by the same methodology [50]. sis of antibiotic nosiheptide [83,84] employs the N-Boc derivative of natural (2S,4R)-4-hydroxyproline 82 as a starting material (Scheme 21) [84,85]. The inversion of configuration at C4 was carried out by intramolecular lactonization to form 83 by implementation of the Mitsunobu reaction. After opening of the lactone ring with trichloroethanol and silylation of the hydroxy group oxidation at C5 was performed in the usual way to give a pyroglutamate 84. Benzyl or p-methoxybenzyl esters 85a or 85b were next obtained after cleavage of 84 under basic conditions.

From pyroglutamic acid
In case of low availability of selected stereoisomers of 4-hydroxyprolines asymmetric syntheses of enantiomeric 4-hydroxypyroglutamates have been elaborated employing 1,3dipolar cycloadditions of homochiral nitrones and acrylates [86- 88] or a Diels-Alder reaction using acylnitroso compounds [89]. However, when compared with these multistep approaches hydroxylation of pyroglutamic acid derivatives seems to be the simplest option. Treatment of the lithium enolate of benzyl N-Boc-pyroglutamate (S)-86 with Davis oxaziridine produced (2S,4R)-87 (Scheme 22) [90][91][92]. HPLC investigation of the reaction mixture showed that (2S,4S)-87 was not formed [90]. Stereospecific hydroxylation occurred on the opposite side to the benzyloxycarbonyl group, i.e., only re-face of the enolate was attacked for steric reasons. It is worth mentioning that hydroxylation of lithium enolates of pyroglutamate and glutamate results in the opposite stereochemical outcome at C4 (R vs S) and formation of a single diastereoisomer for the cyclic system and a 9:1 mixture for the linear one.
Several methodologies toward enantiomeric 3,4-dihydroxy-Lglutamic acid have been developed. In terms of carbon atom economy syntheses using 5-carbon synthons, e.g., pyroglutamic acid derivatives or pentoses, are the most valuable.

From pyroglutamic acid
Cleavage of the 5-membered ring in the protected epoxide 88 obtained from (S)-pyroglutamic acid [93][94][95] gave the methyl ester 89 which, when adsorbed on silica gel, smoothly underwent stereospecific epoxide ring opening to give the oxazolidinone 90 (Scheme 23) [96]. Before installation of the second carboxylic group the secondary hydroxy group in compound 90 was transformed to the silyl ether while the hydroxymethyl fragment was subjected first to hydrolysis of the acetal, then to oxidation and esterification of the acid to provide 91. After acidic hydrolysis (2S,3S,4R)-3,4-dihydroxyglutamic acid [(2S,3S,4R)-4] was obtained as the hydrochloride.
To avoid racemization at Cα in sensitive amino acids the carboxy group was frequently masked as an orthoester. To illustrate this strategy dihydroxylation of the orthoester 92 (derived from L-pyroglutamic acid [97]) was performed to afford a single diastereoisomer 93 since the bulky orthoester residue allows the osmylation to occur from the opposite side (less hindered face). After purification of the diacetate 94 the recovery of acid (2S,3R,4R)-4 was performed (Scheme 24) [98]. However, the hydrolysis was carried out under mild conditions to prevent decomposition of this stereoisomer including racemization at Cα.

From pentose via 2,3-aziridino-γ-lactone
In the so called "2,3-aziridino-γ-lactone methodology" [18,99,100] ribose (or lyxose) is used as a starting material  ment of the known stereochemistry. To demonstrate this strategy cyclic imides 106a (R = TBDMS) and 106b (R = Ac) readily prepared from L-tartaric acid [104,105] were reduced and the respective hydroxylactams were acetylated to produce acetoxylactams 107a and 107b, necessary intermediates in the next step (Scheme 26) [106,107]. The introduction of the cyano group was accomplished by boron trifluoride-catalyzed reaction with trimethylsilyl or tributyltin cyanides and the stereochemical outcome of these reactions strongly depends on the protecting group. Diastereoisomeric excesses of 60-80% were observed in the cyanation of tert-butyldimethylsilyl ether 107a and 108a was the major product, while for the acetate 107b the selectivity was lower (de 54-64%) with 109b predominating.

From D-serine
An interesting strategy to (2S,3S,4S)-4 (Scheme 28) [110] employs a protected serinal (R)-23 [54]. Wittig olefination ex-tended the alkyl chain by two carbon atoms and simultaneously installed the C=C bond which was subjected to the intramolecular epoxidation to give a >20:1 mixture of aminoepoxides with the isomer (2S,3R,4R)-117 dominating. Without isolation this compound underwent another intramolecular cyclization in the 5-exo mode to form the oxazolidinone 118. To complete the synthesis of (2S,3S,4S)-4 the secondary hydroxy group was protected as a pivalate, the hydroxymethyl fragment was oxidized after hydrolysis of the silyl ether and finally all protecting groups were removed by concentrated acid.

Synthetic applications of enantiomeric hydroxy-L-glutamic acids
Besides numerous applications of hydroxyglutamic acids in studies on glutamate receptors they have also been used as starting materials in syntheses of other compounds including complex natural products ( Figure 6).

Conclusion
The synthesis of nonracemic hydroxyglutamic acids is an active area of research and it was stimulated by studies on glutamate receptors to modulate the biological activity of L-glutamic acid from one side and applications as starting materials in total syntheses of complex natural products from the other. In general, the syntheses started from other amino acids and were designed to preserve the stereochemical integrity at Cα while inducing chirality at Cβ-or Cγ-OH centers. Thus, both enantiomers of serine or their synthetic equivalents, glutamic acid, 4-hydroxyproline and pyroglutamic acid were most frequently employed. Alternatively, α-hydroxy acids (malic, tartaric) offered the opportunity to induce chirality at Cα-N while the stereochemistry at C-OH was retained. Monosaccharides (glucose, ribose) also appeared attractive providing two or three predefined stereogenic centers. In more sophisticated approaches application of chiral auxiliaries allowed to generate vicinal or 1,3-aminoalcohol units of the required stereochemistries.
Currently available synthetic methodologies towards hydroxyglutamic acids significantly differ in terms of carbon atom economy and preparative simplicity although carbon-wasteless approaches do exist. For future use as starting materials in total syntheses of complex natural products synthetic methodology to orthogonally protected hydroxyglutamic acids were also discussed which allow, for example, to differentiate between α and ω carboxy groups.
Although syntheses of particular enantiomers of hydroxyglutamic acids look to be optimal, e.g., (2S,4S)-3 or (2S,4R)-3 via hydroxylation of the protected glutamic or pyroglutamic acids, synthetic methodologies to the other enantiomers may require improvements or even designing new ways especially when larger quantities are needed and we hope this review will stimulate further research in this area.