Concise and Straightforward Asymmetric Synthesis of a Cyclic Natural Hydroxy-Amino Acid

An enantioselective total synthesis of the natural amino acid (2S,4R,5R)-4,5-di-hydroxy-pipecolic acid starting from D-glucoheptono-1, 4-lactone is presented. The best sequence employed as a key step the intramolecular nucleophilic displacement by an amino function of a 6-O-p-toluene-sulphonyl derivative of a methyl d-arabino-hexonate and involved only 12 steps with an overall yield of 19%. The structures of the compounds synthesized were elucidated on the basis of comprehensive spectroscopic (NMR and MS) and computational analysis.

Mario Simirgiotis has worked previously on the synthesis of small molecules starting from sugars as chiral templates [26] and embarked on the synthesis of this compound several years ago (under the mentorship of Dr. Oscar Varela, at the University of Buenos Aires, Argentina), but the results were unpublished.

Retrosynthetic Analysis for the Synthesis of Compound 1
We reasoned (Scheme 1) that 1 can be synthesized by imine hydrogenation after direct carbonyl addition by the amino group at C-2 (synthon I), a strategy that was previously employed for the synthesis of (2S,4R) 4-hydroxypipecolic acid [22]. Synthon I can readily be obtained through O-protection of α amino lactone II. Key intermediate II can be in turn obtained from commercially available 2 as a chiral template via a series of elimination of the HO-3 subsequent oxidation, and conversion of the HO-2 into NH2.

Total Synthesis of Compound 1
We started with commercially available and inexpensive D-glycero-D-gulo-heptono-1,4-lactone (2) using a previously published methodology [27], which, on peracetylation and β-elimination of the acetate group on C-3, followed by hydrogenation, O-deacetylation, and isopropylidenation of the resulting aldonolactone with 2,2-dimethoxypropane, yielded 3-deoxy methyl ester 3, which has the desired (4R,5R) configurations (Scheme 2). The free hydroxyl group at C-2 of 3 was sulfonylated with tosyl chloride in pyridine to give the tosylate derivative 4 in 12 h with 96% yield. Spectroscopic data for compound 4 resembled the distorted planar zigzag conformational preference in solution as previously reported for compound 3 [27] with an anti disposition for H-5 and H-6 (J5, 6  Treatment of tosylate 4 with sodium azide in DMF produced the desired inversion of C-2 configuration affording azide 5 in 87% yield. Selective hydrolysis of the isopropylidene terminal group of 5 yielded diol 6, which on degradative oxidation of the 1, 2 diol system using sodium periodate produced aldehyde 7. Aminolysis of 7 in a "one pot" procedure (Scheme 3, Figure 1) yielded key compound 8 through a sequence of catalytic hydrogenation of the azide group, nucleophilic attack of the resulting amine on the aldehyde (7a), dehydration of the resulting alcohol 7b to the imine 7c, and reduction of the imine 7c to give the expected piperidine 8, which crystallized from hexane-EtOAc 10:1 (mp 119 °C, [α]D −47.62). In order to demonstrate this reaction and try to understand the low yield (27%) obtained, an Internal Reaction Coordinate (IRC) calculation was performed. An IRC calculation allows one to map out a reaction pathway by integrating the intrinsic reaction coordinate. This method examines the reaction pathway proposed, leading to a transition state in a map of the potential energy surface. The results of this calculation are showed in Figure 1. The transition state (TS) was found with an imaginary frequency, demonstrating that this is, as expected, a first-order and six-membered cyclic TS. Furthermore, an intense cross-peak observed between H-6' and H-2 confirmed the S configuration for C-2. The correlations were corroborated by HMQC and HMBC spectra. Density Functional Theory (DFT) calculations (Gaussian 9.0) [28] indicated that compound 8 has two minimum-energy conformations (boat and chair) but the boat conformation ( 4 BN) is 2.8 KJ/mol more stable (Figure 3), which is in concordance with the spectroscopic data found for this compound. The isopropylidene group of amino-acid 8 was removed by stirring with trifluoracetic acid-water (1:1) for 8 h at room temperature, while removal of the remaining ester function at C-1 was achieved by subsequent alkaline treatment with NaOH 1 M to yield the target cyclic amino acid 1. Scheme 3. One pot synthesis of piperidine 8 from azide 7.
The preparation of key compound 8 from 7 (Scheme 3) resulted in a disappointing 27% yield. We assayed different concentrations of catalyst, solvent, hydrogen pressure, and reaction times (Table 1), but in all cases after consumption of all starting material (monitored by TLC) we obtained compound 8 (yields below 27%) and a mixture of polar compounds, according to 1 NMR a mixture of adducts produced by intermolecular addition of the amine group at C-2 of one molecule of 7 to the aldehyde function at C-6 of another molecule of 7. We assumed that the low yield obtained for this reaction could be attributed in part to the partial conformational impediment (Figure 1) produced by the isopropylidene group for the nucleophilic addition of the C-2 amine group to the aldehyde.

Synthesis of Compound 1 Starting from Azide 5
The low yield obtained for key compound 8 from 7 prompted us to use the alternative synthetic sequence depicted in Scheme 4. Catalytic hydrogenation of 5 in EtOAc generated the amino group that was in turn protected with benzyl chloroformate to obtain carbamate 9 in 88% yield. Furthermore, the benzyloxycarbonyl was a convenient protective group for the amine group of this small class of sugar compounds, since a very stable carbamate derivative can be obtained and the protecting group can be easily removed by hydrogenolysis in high yields and short reaction times [22]. We considered that 10 was a convenient precursor of 1 that could be degraded to the aldehyde 11 and this compound could be properly reduced to the primary alcohol, which also has the right configuration and hydroxyl group for the functionalization with a convenient leaving group at C-6 to be displaced by the amino group at C-2, after removal of the benzyloxycarbonyl group. Thus, selective hydrolysis of the terminal isopropylidene group with a mixture of acetic acid-water at 50 °C over 12 h produced 10 in 82% yield, which could be converted to 11 using the procedure applied previously in the synthesis of 7, followed by reductive amination with NaBH3CN at pH = 4 in 4 h with 91% yield. The tosylate 12 was then prepared as explained above to generate the necessary leaving group for the intramolecular nucleophilic displacement by the amino group.  Deprotection of the amine group of 12 was performed with 10% palladium on charcoal under hydrogen atmosphere (45 psi) in EtOAc for 12 h, and after filtration and evaporation of the solvent, the isopropylidene and ester groups of the latter compound were removed by hydrolysis with trifluoracetic acid: water (1:1). The resulting crude product, after evaporation of the solvent, was dissolved in water and treated with K2CO3 for 2 h to produce the nucleophilic amine group necessary for the ring closure (Scheme 4 and Figure 4). As shown in Figure 4, the activation energy for the closure of intermediate 12a (reactant) into 1 (product) is almost half that for the intramolecular nucleophilic attack of amine on the aldehyde of intermediate 7a to obtain 8 ( Figure 1). This can explain in part the different yield obtained for the mentioned reactions.
After 5 h of reaction only one product reactive to ninhydrin was obtained, and the solution was filtrated, acidified to pH = 6, and loaded onto an ion exchange column (Dowex 50W-H + resin), which was rinsed with water and eluted with pyridine 0.1 N to obtain the target syrupy pipecolic amino-acid 1 with an [α]D −5.4, and HRESIMS: [M+H] + = 162.0763, without epimerization in 83% yield. 13 NMR spectra for 1 showed the expected six carbons, including one carboxylic acid (δ 173.35), three oxygenated methynes (C-2, C-4, and C-5 at δ 53.5, 65.4, and 64.9, respectively), and two methylene carbons (C-3 and C-6, at δ 28.4 and 44.2). All assignments were performed by careful analysis of the COSY, NOESY, HMQC, and HMBC spectra. The 1 H-NMR spectral data for this compound was published early in 1976 [25], but recorded using a 100 MHz equipment. As in the 1 H-NMR spectra of pipecolic derivative 8, coupling constants for H-2 (J2,3ax = 11.4, J2,3eq = 4.2 Hz) for H-3ax (J3ax, 3eq = 14.9, J3ax,4 = 2.8 Hz) and H-3eq (J2,3eq = J3eq,4 ≈ 4.2) confirmed the S, R, and R configuration for C-2, C-4, and C-5, respectively, while NOE cross peaks between H-5-H-6', H-2 with H6', and H-3ax with H-5 and H-4 suggested the chair conformation 4C N . The optimized structure of compound 1 ( Figure 5) was built into the Gaussian 9.0 platform and the energy minimized by the DFT method [28], which coincides with the NMR data obtained and configuration proposed. High resolution NMR (see Experimental section) and ESI-MS spectra reported in this work added more accurate information about this natural product structure, for which the NMR data were published a long time ago [25].

General Experimental Procedures
Melting points were determined using a Stuart Scientific SMP3 melting point apparatus (Bibby Scientific Ltd, Staffordshire, UK). Analytical thin-layer chromatography (TLC) was performed on silica gel 60 F254 (Merck, Darmstadt, Germany) aluminum-supported plates (layer thickness 0.2 mm). Medium pressure column chromatography was performed with Silica gel (Kieselgel 60 H Merck, Darmstadt, Germany), 55 mm particle size, FMI QG 150 medium pressure lab pumps (Syosset, NY, USA) and Ace Glass Inc. medium pressure columns (Vineland, NJ, USA), using mixtures of n-hexane:ethyl acetate of different polarities as solvent system (flow rate: 5 mL/min.). Optical rotations were measured on an Autopol III automatic polarimeter (Rudolph Research Co., Hackettstown, NJ, USA). The NMR experiments ( 1 H: 400.12 and 600.13 MHz; 13 C: 100.25 and 150.09 MHz) were performed using either a Bruker Avance 400 or Bruker Avance II (Biospin, Rheinstetten, Germany) 600 UltraShield spectrometer with CD3OD or deuterated MeOD as solvent and TMS as internal standard. Optical rotations were measured on an Autopol III automatic polarimeter (Rudolph Research Co., Hackettstown, NJ, USA). IR spectra were measured using a Thermo Nicolet Nexus 470 FT-IR spectrometer (Thermo Nicolet, Madison, WI, USA) with KBr disks. The molecular weight was determined by low and high resolution with a mass spectrometer (Finnigan Mat 900 XLT, Thermo Fisher, Bremen, GmbH, Germany). Another mass spectrometer equipped with electrospray ion source and qToF analyzer MicrOTOF Q II (Bruker Daltonics Inc., Billerica, MA, USA) was used for HR-ESI-MS analysis. All reagents used for reactions were of analytical grade and purchased from Merck (Santiago, Chile) or Sigma-Aldrich (Santiago, Chile).

Computational Details
Calculations were carried out with the B3LYP hybrid functionals and 6-31+G(d) basis set. Full geometry optimizations and transition structure (TS) searches were carried out with the Gaussian 09 package 18 [28]. The possibility of different conformational isomers was taken into account for all structures. Frequency analyses were carried out at the same level used in the geometry optimizations, and the nature of the stationary points was determined in each case according to the appropriate number of negative eigenvalues of the Hessian matrix.

Conclusions
The amino acid (2S,4R,5R) 4,5 dihydroxy-pipecolic acid (1) was synthesized by two enantiospecific sequences, using D-gluco-heptono-1,4-lactone as a chiral template. The key step in the first approach involved the preparation of compound 8 through nucleophilic attack after catalytic hydrogenation of the azide group from aldehyde 7, obtained by degradative oxidation of diol 6. The second strategy enabled the construction of the piperidine ring by intramolecular substitution of the tosylate at C-6 by the amine group at C-2, produced by hydrogenation of carbamate 12, after removal of the isopropylidene group with F3CCO2H. The latter and more efficient sequence involved 12 steps and gave a 19% overall yield, employing a commercially available and inexpensive sugar lactone.