A Unified Approach to Phytosiderophore Natural Products

Abstract This work reports on the concise total synthesis of eight natural products of the mugineic acid and avenic acid families (phytosiderophores). An innovative „east‐to‐west“ assembly of the trimeric products resulted in a high degree of divergence enabling the formation of the final products in just 10 or 11 steps each with a minimum of overall synthetic effort. Chiral pool starting materials (l‐malic acid, threonines) were employed for the outer building blocks while the middle building blocks were accessed by diastereo‐ and enantioselective methods. A highlight of this work consists in the straightforward preparation of epimeric hydroxyazetidine amino acids, useful building blocks on their own, enabling the first synthesis of 3’’‐hydroxymugineic acid and 3’’‐hydroxy‐2’‐deoxymugineic acid.

Micronutrient acquisition is an importantf actor in growth and survival of any living organism.P lants are stationary and need to fill all their needs from the soil they grow in. In calcareous, high pH soils, the solubility and therefore plant availability of some of these crucial metal ions, like iron, zinc and copper, is diminished, to ad egree that it greatly inhibits plant growth and leads to chlorosis of the leaves. [1][2] On top of reduced yield, micronutrient deficienciesi nm ajor crops like wheat, barley,r ice and maize carry over to local consumers, causing micronutrient deficiency in humans ("hidden hunger") with severe negative effects on child growth, development and disease resistance. [3] Au nique strategy used by gramineousp lants for the uptake of Fe andp otentially also Zn and Cu relies on phytosiderophores (PS), which are multidentate chelators of metal ions. PS are exuded by roots of grass speciesi nto the surrounding soil (i.e. rhizosphere) where they can complex Fe III ions from soil particles. [4] The PS-Fe III 1:1c omplex is then taken up as whole complex andt he iron liberated within the cell. This complexation strategy renders grass speciesm ore efficient in Fe acquisition compared to non-grasss pecies, particularly in high pH soil, where soluble Fe concentrations are low.I nvestigations into plants molecularm echanisms, and consequentlye fforts towardst he synthesis of these natural products have been reporteds ince the late 1970s. [5a-h] Variation in naturally occurring PS (see Figure 1) arises from hydroxy groups present on C-2 of the western (left) and middle subunits, respectively,w hile the easternh ydroxyacid fragment is conserved throughout the mugineic, avenic and distichonic acids eries. The general synthetic strategy is closely relatedt ot he biosynthesis [6] and specific synthetic solutions were published fora venic acid A( AVA, I), deoxymugineic acid (DMA, IV), mugineic acid (MA, VIII)a nd 3''-epi-hydroxy-MA (VI).
In the state-of-the-art mugineic acid synthesis by Namba et al., [5c-d] which used a" west-to-east" approach, stereoselectiv- ity issues arosed uring the introduction of the 2'-hydroxygroup by allylic oxidation requiring late stage separation of diastereomers by meanso fp reparative HPLC, thus, limiting access to larger quantities of MA (VIII). In earlier work CS and co-workers synthesized 13 C 4 -labelled PS in as imilarf ashion.
[5e] Thea ctivity of the synthetic community notwithstanding, to date, useful quantities of all PS in their naturala nd isotopically ( 13 C) labelled form (used as internal standards for trace analysis in soil) are currently not available, making it difficult for groups studyingt hese plant mechanismst oe fficiently carry out their research. In this light,n og eneral approacht ot he more complex mugineic acids (II, III, VI and VII)h as so far been reported, most likely due to lack of availability of building blocks 4 and 5.A imingf or more members of this class of compounds and due to the high diversity of "western"f ragments presenti n naturallyo ccurring PS, the opposite direction of assembly was recognized as at entatively superior approachw hich has thus far been widely neglectedb yt he synthetic community. [5a] Within the framework of an interdisciplinary project we set out to establish ag eneral solution for all members of the mugineic and avenic acid family in natural form compatible with an application towards 13 C 2 -labelled versions. These are required as standards for high performance trace quantification in complex matrices as well as for biodegradation experiments.

Results and Discussion
As mentioned above our aim was to synthesize eight different PS from "east-to-west" from one common set of building blocks. Thus, the best transformationsa vailable to date to arrive at the necessary buildingb locks 1-7 efficiently (see Figure 2) werer equired. Three of the required building blocks (3, 6, 7)a re commercially availablei nu nprotected form. While there was ample literature on the synthesis of eastern subunit 1,t he chemistry of hydroxylated azetidinesl ike 4 and 5 was still underdeveloped. Ap alladium catalyzed CÀHc yclization [7] was identified as ap ossible access to these unnatural amino acids. Regardingb uildingb lock 2 we envisioned introducing the two required stereocenters simultaneously in an aldol-type 1,2-addition of ag lycine equivalent, [8] which in the future can be prepared from commerciallya vailable 13 C 2 -labelled glycine. The olefinsi n2 and 3 serve as masked aldehydes to be liberated for reductivea minationd uring the course of the synthesis. Notably,t his strategy also allows fort he preparation of distichonic acids and nicotianamines (Figure 1) by starting from amine protected derivatives of 2 and 3 as eastern fragments, thus unifyingt he access to all naturally occurring PS through these seven building blocks. Throughoutt he synthesis acid labile protecting groups (tBu, Boc) weree mployed to reduce step count and enable ac lean deprotection during the endgame.
For the synthesis of middle subunit 2 ah ighly stereoselective addition of ag lycine equivalent to crotonaldehyde was required. The method of SolladiØ-Cavallo [9a,b] was conceivably qualified to produce the required erythro-aminoalcohol 2 in high optical purity.A ccordingly,t he chiral pinanone auxiliary 13 was attached to tert-butyl protected glycine (14)u nder Lewis acid catalysis (See Scheme 1). The resultingi mine (15) was then added in at itaniumm ediated aldol reaction to crotonaldehyde, giving rise to ah ydroxyimine( not depicted), which due to limited stability( retro-Aldol) was subsequently hydrolysed under acidic conditions delivering subunit 2 in three steps in a7 7% overall yield. Protected non-labelleda llylglycine 3 was prepared in as traightforward fashion in one step from commercial l-allylglycine (12).
The synthesis of hydroxylated l-azetidines 4 and 5 was carried out starting from protected l-o rd-threonine derivatives 18 and ent-18 in at hree-step procedure. Initial attachment of a2 -picolinamide (PA) directing group (DG)w as followed by cyclisation under palladium catalysed oxidative CÀHa ctivation conditions developed by Chen et al. [7] The tert-butyl protecting groups in 19 and ent-19 proved to be ideal for the overall success of this route duet ot heir excellent stability,w hiles ilyl ethers and other protecting groups were not suitable.
Next, the methyl ester and DG cleavage wase valuated, at first under basic conditions. When ah ydroxide base was employed at room temperature, epimerization to the more stable [10] trans-azetidines (21 and ent-21)p roceeded rapidly and with almost full conversion. Subsequently,t he amide was cleaved by further heating the reaction mixture. This turned out to be av ery efficient solutiont oa ccess 5.A ni nfluence in temperature and base was found, with epimerization being slower than ester cleavage at lower temperatures. Nevertheless, only a6 2:38 mixture of the desired (cis-azetidine 4)a nd undesired product (ent-5)c ould be obtained with this method. The assigned stereochemistry was proven by X-ray diffraction of pure ent-5 (see Scheme 2, Deposition Number 2007994 contains the supplementary crystallographic data for this paper. These data are providedf ree of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum KarlsruheA ccess Structures service www.ccdc.cam.ac.uk/structures). While the 56 %i solated yield of 4 in this step are acceptable for such an elusive material, as elective approachw as also developed. Thus,t he methyl ester in 19 wasc leaved first with the mild reagent Me 3 SnOH. [11] No epimerization at C-2 was observed and the picolinamidec ould be subsequently cleaved with base after carefuln eutralization of the carboxylic acid. The amino acids 4 and 5 were separated from any minor isomersp resentb yp reparative HPLC and obtainedi n> 60 % overall yield from 18 and ent-18.
The eastern subunit aldehyde 1 was prepared in ag ood yield over 5steps as reported [5c,e] and assembly of the two central dimeric intermediates could commence. Aminoesters 2 and 3 were alkylated under reductivec onditions (NaCNBH 3 )w ith aldehyde 1.Asmoothr eaction was achievedi nb oth instances giving 83 %a nd 80 %o fd imeric amines 22 and 24 upon chromatographic purification at decent scale (Scheme 3). Attachmento faBoc protecting group was realized in high yield (Boc 2 O, THF) and the resulting carbamates 23 and 25 proved to be suitably stable storage compounds. The dimericc ompounds 23 and 25 now enter the second stage of diversification towards two sets of four phytosiderophores each, with a2 '-hydroxy and 2'-deoxy-structure, respectively.T he olefin was cleaved by ozonolysis yielding aldehydes 26 and 27 which were subsequently treated with one of the four respective western building blocks 4-7 under reductive amination conditions, as before (Scheme 4). Thanks to the universal protecting group strategy throughout the synthesis, global deprotection of the moleculesp rovedf acile and wasi nduced by treatment with 6 m HCl. The resultinga mino acids I-VIII were purifiedo nD owex resin andi solated as the respective ammonium salts. For highly pure material additional triturations teps or preparative HPLC can be carriedo ut. [5e] To the best of our knowledge this represents the first total synthesis of 3''-epi-hydroxy-DMA (II), 3''-hydroxy-DMA (III), hydroxyavenic acid A(V)a nd 3''-hydroxy-MA (VII).
Ac oncisea nd modular synthesis of phytosiderophore natural products was achieved. Startingf rom l-malic acid, targetc ompounds I-VIII could be prepared in 10 or 11 steps longest linear sequence (15-25 %o verall yield). The required building blocks could all be efficiently accessed by CÀHa ctivation and stereoselective aldol reaction. Common key intermediates (23 and 25)a llow very flexible and fast resynthesis and delivery of materials for the planned applicationsw ithin our current project and beyond. As mentioned before, by employing properly protected building blocks 2 and 3 as "eastern" fragments and compound 14 as "western" fragment, accesst o the nicotianaamine and distichonic acid family is well within the scope of the developed methodology.
Further work on fully assembled 13 C 2 -labelled versionso f compounds I-VIII using the presented strategy as well as development towards scale up (gram-scale) of the avenic acid and mugineic acids yntheses and other PS are currently ongoing in our laboratories and will be reported in due time.