One-pot combination of enzyme and Pd nanoparticle catalysis for the synthesis of enantiomerically pure 1 , 2-amino alcohols

Supplementary data .................................................................................... 2 Complete results of ADH screening ......................................................... 2 Activity assay of bacterial ADHs.............................................................. 4 Substrate concentration–activity profiles of selected ADHs...................... 5 Time-course experiments using selected ADHs........................................ 6 Reduction of azido ketones 2a–f by ADHs .............................................. 7 Time course of the azidolysis of chloro ketone 1a................................... 7 Characterisation data of Pd nanoparticles ............................................... 8 Environmental impact assessment of tembamide syntheses................... 10


hemistry
Univ
rsity of Graz
Heinrichstrasse 288010GrazAustria

Wolfgang Kroutil 
Department of Chemistry, Organic & Bioorganic Chemistry
University of Graz
Heinrichstrasse 288010GrazAustria

Nicola D' 
Department of Engineering and Geology (INGEO)
University "G. d'Annunzio" of Chieti-Pescara
Viale Pindaro 4265127PescaraItaly

Frank Hollmann f.hollmann@tudelft.nl 
Department of Biotechnology
Delft University of Technology
Julianalaan 1362628 BLDelftThe Netherlands

Department of Engineering and Geology (INGEO)
University "G. d'Annunzio" of Chieti-Pescara
Viale Pindaro 4265127PescaraItaly
11 September 201341AACEFA165A139FE42E737A277F3D4C10.1039/c3gc41666fReceived 15th August 2013, Accepted 11th September 2013
One-pot combinations of sequential catalytic reactions can offer practical and ecological advantages over classical multi-step synthesis schemes.In this context, the integration of enzymatic and chemo-catalytic transformations holds particular potential for efficient and selective reaction sequences that would not be possible using either method alone.Here, we report the one-pot combination of alcohol dehydrogenase-catalysed asymmetric reduction of 2-azido ketones and Pd nanoparticle-catalysed hydrogenation of the resulting azido alcohols, which gives access to both enantiomers of aromatic 1,2-amino alcohols in high yields and excellent optical purity (ee >99%).Furthermore, we demonstrate the incorporation of an upstream azidolysis and a downstream acylation step into the one-pot system, thus establishing a highly integrated synthesis of the antiviral natural product (S)-tembamide in 73% yield (ee >99%) over 4 steps.Avoiding the purification and isolation of intermediates in this synthetic sequence leads to an unprecedentedly low ecological footprint, as quantified by the E-factor and solvent demand.

Introduction

The integration of several chemical transformations into onepot processes (often referred to as 'domino', 'tandem', or 'cascade' systems) offers advantages with respect to operational simplicity, operating time and costs, safety, and the consumption of energy and materials. 1From an ecological perspective, the one-pot combination of catalytic methods 2 is particularly appealing, and recent years have brought about remarkable developments in the use of heterogeneous, homogeneous, organo-and biocatalysts in cascade systems. 3owever, these approaches typically remain within one individual 'subfield' of catalysis, while the one-pot combination of different types of catalysts is less explored.This is particularly true for combinations of chemical catalysts and enzymes, which are often complicated by divergent reaction conditions and detrimental interactions of the bio-and chemo-catalysts. 4evertheless, several excellent recent studies aimed at bringing chemo-and biocatalysis closer together clearly demonstrate the potential of chemo-enzymatic one-pot systems. 5In this context, true cascadesin which the biocatalytic and chemical reactions proceed concurrentlydefinitely represent the most elegant examples, but they are also most prone to the abovementioned difficulties. 6Sequential chemo-enzymatic one-pot reactions are more easily realised and offer essentially the same environmental advantages (e.g.reduction of solvent use due to elimination of work-up and purification steps).4a,c,7 The 2-amino-1-aryl alcohol moiety is a common structural motif in biologically active compounds, 8 as it forms the basic scaffold of adrenergics (e.g. the hormones adrenaline and noradrenaline, β-adrenergic blockers, and anti-asthma drugs), amphenicol antibiotics, and several bioactive natural products (Fig. 1).In addition, 1,2-amino alcohols have found broad use as chiral ligands and auxiliaries in asymmetric synthesis. 9For both pharmaceutical and synthetic applications, a high optical purity of chiral amino alcohols is desired, causing the need for highly selective asymmetric syntheses of these compounds.Chemical methodssuch as asymmetric hydrogenation 10 or transfer hydrogenation 11 of 2-amino ketones or their syn

etic equivalentshave be
n intensively studied in recent years, yet † Electronic suppl mentary information (ESI) available: Complete results of ADH screening, additional optimisation studies, characterisation data of Pd-NPs, additional details on the environmental assessment of tembamide syntheses, additional experimental procedures, full characterisation data of compounds isolated from the chemo-enzymatic transformations, EATOS and Excel files used in the environmental impact assessment.See DOI: 10.1039/c3gc41666f they often fail to afford enantiomerically pure products.
s a result, classical resolution is still a widely used technique in the preparation of chiral 1,2-amino alcohols, either for upgrading the ee of an optically enriched product or for resolving the racemate.10a Biocatalysis, on the other hand, offers several options for the preparation of optically pure 1,2-amino alcohols in general, such as kinetic resolution or dynamic kinetic resolution of the racemate using lipases, 12 oxidative kinetic resolution of N-protected amino ketones by Baeyer-Villiger monooxygenases, 13 the combination of aldolase-catalysed C-C bond formation and enzymatic decarboxylation, 14 or the combination of lyase-catalysed C-C bond formation with reductive amination catalysed by ω-transaminases. 15However, biocatalytic asymmetric methods for the preparation of 2-amino-1-aryl alcohols in particular are comparably scarce.12e,14,15a,16a Herein, we describe a chemo-enzymatic approach for the asymmetric synthesis of this important substance class: a sequential one-pot combination of stereoselective azido ketone reduction by alcohol dehydrogenases and subsequent azide hydrogenation by palladium nanoparticles (Scheme 1), which provides access to both enantiomers of 2-amino-1-aryl alcohols in high yields and with excellent optical purities.The integration of an upstream azidolysis step into the one-pot process and in situ benzoylation of the crude amino alcohol have also been achieved, enabling the one-pot asymmetric synthesis of the natural product tembamide (Fig. 1) in four steps from the corresponding commercially available bromo ketone 1e.The ecological benefits of this one-pot concept are demonstrated by a basic environmental impact assessment.


Results and discussion


Azido alcohol hydrogenation catalysed by metal nanoparticles

In our one-pot approach, we envisioned combining the asymmetric reduction of 2-azido ketones catalysed by alcohol dehydrogenases (ADHs) with azide hydrogenation catalysed by recently reported 17 lignin-stabilised metal nanoparticles (NPs).Since the latter reaction had not been investigated before, we began our studies by screening six different metal nanoparticle preparations (Pd or Pt, stabilised by three different lignin varieties) in the hydrogenation of 2-azido-1-phenylethanol 3a.Phosphate buffer containing 5% (v/v) of 2-propanol as organic co-solvent was chosen as a reaction medium to ensure compatibility with the conditions for the enzyme-catalysed step.As shown in Table 1, Pd-LC and Pd-LK nanoparticles (for explanation of nanoparticle types, see Table 1 footnotes) gave the best results, leading to full conversion within 4 h (entries 1 and 3).Minor amounts of unidentified side products were formed in all reactions, but subsequent experiments showed  that by carrying out the reduction under basic conditions ( pH 9) the chemoselectivity can be raised further, such that 2-amino-1-phenyl-ethanol 4a was the only product detectable by GC-FID and NMR analysis (Table 1, entry 7).The hydrogenation of 3b-f under identical conditions proceeded with the same high level of selectivity, affording the corresponding 1,2-amino alcohols 4b-f in essentially pure form. 18ymmetric azido ketone reduction catalysed by ADHs

Next, we turned our attention to the ADH-catalysed asymmetric reduction of aromatic 2-azido ketones.Although there is literature precedence for this transformation, 19 a broad survey of suitable enzymes has not yet been reported.Therefore, we carried out an extensive screening of 79 commerci

ADHs of unspecified origin and four bac
erial ADHs (ADH-A from Rhodococcus ruber DSM 44541, TbADH from Thermoanaerobium brockii, LkADH from Lactobacillus kefir, LbADH from Lactobacillus brevis) 20 in the reduction of 2a to 3a.The screening was performed at 50 mM concentration of 2a and in the presence of 5% (v/v) 2-propanol, the latter serving both as co-solvent and co-substrate.Table 2 shows selected results of the ADH screening (for complete results, see ESI †).As a general trend, we identified more anti-Prelog-selective 21,22 ADHs ( 18) with activity on 2a than Prelog-selective ones (9), and on average, the former showed about 5-10 times higher activity than the latter.Only eleven enzymes afforded optically pure 3a, whereby four gave the (R)-enantiomer and seven the (S)-enantiomer. 23ased on the observed activities and stereoselectivities, we considered the commercial ADHs listed in Table 2 (entries 5-8) as well as ADH-A (entry 1), the most promising biocatalysts for the reaction under study.A further selection was made according to the enzymes' tolerance for higher substrate concentrations and their operational stability.We identified ADH-A and KRED-NADH-110 as the most robust of the investigated Prelog-and anti-Prelogselective enzymes, respectively, showing reasonable activity at 100 mM concentration of 2a (Suppl.Fig. 1 and 2, ESI †), and a good performance over 24 h of reaction time (Suppl.Fig. 3 and  4, ESI †).The substrate scope of ADH-A and KRED-NADH-110 was investigated using the representative 2-azido-1-arylethanones 2a-f as substrates.Both biocatalysts converted all six ketones tested with excellent stereoselectivity and with similar relat ve rates; only the reduction of 2f by KRED-NADH-110 was surprisingly slow in comparison to the other 2-azido ketones (Fig. 2).Test transformations with 100 mM substrate concentration and optimised enzyme loadings (ADH-A: 1.5-2.5 mg mL −1 , KRED-NADH-110: 0.2-1.0mg mL −1 ; for details see ESI †) proceeded to complete conversion within 20 h in all cases.


Chemo-enzymatic one-pot transformations

After suitable ADHs had been identified, we proceeded with combining the enzymatic azido ketone reduction and Pd-NPcatalysed azide hydrogenation in a one-pot sequence.The biocatalytic reduction was run to completion (20 h) before adjusting the pH of the reaction mixture to 9 and adding a Pd-NP stock solution.First experiments showed that the activity of the nanoparticles was not impaired by the presence of the enzymes, and complete reduction of the intermediate azido alcohols 3a-f was achieved in 4 h.However, in addition to the expected products 4, the corresponding 2,2-dimethyloxazolidines 5 were also formed in minor amounts (4-10%), apparently by the reaction of 4 with acetone generated as by-product in the ADH-catalysed reduction (Scheme 2). 24When the bioreduction was carried out under reduced pressure, so as to remove the acetone from the solution, the one-pot sequence afforded essentially pure 1,2-amino alcohols 4a-f.In semipreparative-scale experiments (5 mL volume, 0.5 mmol substrate converted), the desired products could be isolated in good yield and excellent enantiomeric excess (Table 3).Only the isolated yields of 4f fall behind compared to the other amino alcohols, which we attri ute to the high aqueous solubility and the resulting difficult extraction of 4f.To further prove the preparative value and the scalability of the one-pot two-step reaction system, we performed the conversion of azidoketone 2b into 1,2-amino alcohol (R)-4b on gram scale (75 mL, 7.5 mmol substrate).ADH-A (1.5 mg mL −1 ) was used as a biocatalyst, and the target compound was isolated in 84% yield (1.08 g) and >99% ee.

Encouraged by the positive results of the ADH/Pd-NP combination, we sought to integrate the in situ formation of azido ketones 2 from the corresponding halo ketones 1 into the onepot system.As a first test, 1a was reacted at 60 °C with 1.2 equiv. of NaN 3 in buffer containing 5% (v/v) 2-propanol and varied amounts of potassium iodide as a nucleophilic substitution catalyst.With 5 and 10 mol% of iodide, the reaction proceeded almost equally fast, and 90% conversion was achieve

within 4 h (Suppl.Fig. 5, ESI †)
Transferring the same conditions (10 mol% KI) to the autoclave setup used for the one-pot sequence, the azidolysis reaction of 1a-e proceeded to completion within 5 h, while the furan derivative 1f only required 2 h for full conversion.In addition, we found that the ADH-catalysed reduction could be performed on the crude reaction mixture of the azidolysis step, as a minor decrease in enzyme activity was easily compensated by slightly raising the enzyme loading.Hence, the sequential combination of all three reactions (azidolysis, ADH reduction, and hydrogenation) proved feasible.All steps of the one-pot process could be run to complete conversion, and no accumulation of any side products was observed.Consequently, the 1,2-amino alcohols 4a-f were obtained with the same level of chemical and enantiomeric purity as in the two-step process (see Table 4).

An exemplary gram-scale conversion (75 mL, 7.5 mmol substrate) of 1c into (R)-4c using ADH-A (2.0 mg mL −1 ) as a biocatalyst provided the 1,2-amino alcohol in 84% isolated yield (0.97 g) and optically pure form (ee >99%).

Finally, we wanted to apply our newly developed chemoenzymatic one-pot reaction system to the asymmetric synthesis of a biologically active molecule.As a target compound, we   chose (S)-tembamide, a naturally occurring benzamide derivative, for which antiviral (HIV) activity has been reported. 25The synthesis required benzoylation of amino alcohol (S)-4e as the final step, which was achieved by simply adding a solution of benzoyl chloride (1.2 eq.) in MTBE to the alkaline, aqueous reaction mixture obtained after the Pd-NP-catalysed hyd ogenation step (Scheme 3).A gram-scale reaction (50 mL, 5.0 mmol substrate, 0.5 mg mL −1 KRED-NADH-110 as biocatalyst) yielded 0.98 g (73% from 1e) of (+)-(S)-tembamide, thus providing access to this natural product in a four-step one-pot operation.


Environmental impact assessment

Because of its highly 'integrative' nature we considered the tembamide synthesis a suitable test case for assessing the ecological benefits of the multi-step one-pot concept.Therefore, we performed a basic environmental impact analysis, in which we compared our four-step one-pot preparation of tembamide to previously reported asymmetric syntheses of this compound.We chose Sheldon's E-factor 26 (mass of aste produced per mass of desired product) as a simple metric.We also decided to exclude solvents from the E-factor analysis and calculate the solvent demand as a second, independent indicator, since solvent waste and non-volatile waste ( particularly salts) require very different processing.Both metrics can only provide a rough estimation of environmental impact, as they do not take into account the chemical co position (and hence the toxicity) of the waste, the energy demand of the involved processes, or the waste generated in the preparation of starting materials and catalysts.On the other hand, such a basic analysis is easily performed, and can thus serve as a quick ecoassessment of several synthetic options.

Table 5 provides an overview of the environmental performance of the six synthetic sequences under investigation.The ch mo-enzymatic four-step one-pot system presented herein achieves the second-highest yield and has clear environmental advantages over the previously published procedures.Only the synthesis developed by Br

yproducts, c
talysts) and the down-stream processing reveals the main advantage of the one-pot concept: the elimination of the isolation and purification steps leads to significant reductions in waste generation, which for all syntheses except the one reported by Yadav et al. (which uses large quantities of carrot root as a catalyst), is mainly linked to work-up and purification rather than loss of material in the reaction itself (Fig. 3).Nevertheless, our synthesis also features the lowest reaction-linked E-factor contribution (3.2) of the six procedures analysed.

Differentiation by the type of waste (see Suppl.Finally, a more detailed analysis of the different types of solvents used (see Suppl.Table 7, ESI †) shows that our chemoenzymatic sequence generally employs more environmentally acceptable solvents 31 (mostly water, ethyl acetate, and ethanol) than the other processes, especially because it avoids the use of chlorinated solvents and does not require eluents for chromatography, which often contain large amounts of hexane.

Scheme 3 Asymmetric synthesis of (S)-tembamide in a chemo-enzymatic four-step one-pot sequence.


Conclusions

In summary, we have developed chemo-enzymatic one-pot reaction sequences that provide access to enantiomerically pure 1,2-amino alcohols either in two steps from the corresponding 2-azido ketones or in three steps from 2-halo ketones.

The biocatalytic reduction of 2-azido ketones using alcohol dehydrogenases (ADHs) is the asymmetric key step in these processes, and by selecting suitable ADHs, both enantiomers of the target compounds can be obtained in excellent enantiomeric excess (ee >99%).The 2-amino-1-arylethanol derivatives 4a-f thus prepared are i

blocks in pharmaceutical rese
rch, for instance in the synthesis of anti-inflammatory, antiviral, or antitumour agents. 32urthermore, the one-pot concept has been applied to the asymmetric synthesis of the antiviral natural product (S)-tembamide, obtained in 73% yield over four steps and >99% ee from commercially available bromo ketone 1e.This synthesis reaches high catalyst turnover (TON = 200 for Pd, 1000 for NAD + , several 10 000 for the ADH), 33 uses only a small excess of reagents (1.2 eq. of NaN 3 and BzCl), and affords a chemically pure product after a final recrystallisation as the sole purification step.Due to these features, our method compares favourably with previous syntheses of tembamide not only in terms of yield, but especially regarding its ecological impact, as quantified by E-factor and solvent demand.Hence, our study highlights the advantages of chemo-enzymatic one-pot rocesses in the multi-step synthesis of chiral compounds, and since it uses catalysts that are either commercially available or easily prepared, we believe that it will also be of practical value to synthetic chemists.


Experimental


General materials and methods

Unless otherwise noted, reagents and organic solvents were obtained from chemical suppliers in reagent grade quality and used without further purification.Petroleum ether (boiling range 40-60 °C) and ethyl acetate used for extraction and column chromatography were purchased in technical grade quality and were distilled prior to use.Pro analysi ( p.a., >99% purity) grade solvents were used for handling the 1,2-amino alcohols 4a-f to avoid any undesired oxazolidine formation due to contaminating acetone.Sulfonated lignin with either calcium or ammonium counterions was obtained from Burgo Group S.p.A., Tolmezzo, Italy.Low-sulfonate Kraft lignin was obtained from Sigma-Aldrich.The halo ketones 1a-e as well as both enantiomers of amino alcohol 4a were obtained commercially; all other substrates and reference compounds were synthesised as described in the ESI.†

The proprietary enzymes used in this study are part of the Codexis Codex KRED screening kit, the Almac selectAZyme CRED screening kit, and the evocatal ADH screening kit.The ADHs from Lactobacillus kefir and Thermoanaerobium brockii were obtained from Sigma-Aldrich.ADH-A from Rhodococcus ruber DSM 44541 and bADH from Lactobacillus brevis were heterologously expressed in E. coli as described in the ESI.† Hydrogenation reactions, as well as analytical-scale (2 mL) and semi-preparative scale (5 mL) chemo-enzymatic one-pot transformations were carried out in magnetically stirred stainless steel autoclaves (16 mL total volume) that are part of a HEL PolyBlock8 parallel reactor system, and reactor temperature as well as stirring speed were controlled using the associated HEL WinISO software (v.2.3.85.1).Gram-scale chemoenzymatic

e-pot transformations were carried out in a a Total number of
chemical transformations.The number of steps carried out individually (with product isolation) is given in parentheses.b Overall yield.c Overall E-factor (excluding solvents).d Overall solvent demand.e Please note that this synthesis starts from 2-azido ketone 2e, which is not commercially available, and therefore also needs to be synthesised.mechanically stirred Parr 4560 series stainless steel autoclave (452HC2 bomb cylinder, 160 mL total volume), and the reactor temperature was controlled using a Parr 4841 heater/controller.Non-enzymatic reactions were generally stirred at 500 rpm, while biotransformations were stirred at 300 rpm (to minimise mechanical stress).Thin layer chromatography was carried out on silica gel 60 F 254 plates (Merck) an compounds were visualized either by UV or by dipping into cerium ammonium molybdate stain
[50 g L −1 (NH 4 ) 6 Mo 7 O 24 •4H 2 O, 2 g L −1 Ce(SO 4 ) 2 •4H 2 O in 10% (v/v)

Azido alcohol hydrogenation catalysed by etal nanoparticles

Screening of metal nanoparticles for activity and selectivity in the hydrogenation of azido alcohol 3a.In a small-scale autoclave reactor (16 mL), 2-azido-1-phenylethanol (3a; 32 mg, 200 µmol; final conc. 100 mM) was dissolved in 2-propanol (100 µL; final conc.5% v/v).Potassium phosphate buffer (1.72 or 1.82 mL, depending on nanoparticle type; 100 mM, pH 7.0, 1 mM MgSO 4 ) and a stock solution of metal nanoparticles in water (Pd: 180 µL of a 5.6 mM stock, Pt: 83 µL of a 12 mM stock; final conc.0.5 mM) were added, a

the mixture was st
rred at 30 °C and 500 rpm under hydrogen atmosphere (10 bar) for 4 h.The reaction mixture was extracted with EtOAc (800 µL), the extract was dried over MgSO 4 and conversion was determined by GC-FID analysis.

Investigation of the tolerance of Pd nanoparticles towards the presence of ADHs and NAD(P) + .Reactions were set up as described above, but contained 0.1-1.5 mg mL −1 of different ADHs and 100 µM NAD + or NADP + .

Hydrogenation of azido alcohols 3a-f catalysed by Pd nanoparticles.Reactions were set up as described above, using 200 µmol (final conc.100 mM) of azido alcohols 3a-f as substrates.After 4 h, the reaction mixture was transferred to microcentrifuge tubes, the product was extracted into EtOAc (2 × 1 mL), and t e extract was dried over MgSO 4 .Evaporation of the solvent under reduced pressure afforded the crude amino alcohols 4a-f.The conversion as well as the purity of the crude products were determined by GC-FID and NMR analysis.


Biotransformations

Screening of ADHs for activity and stereoselectivity in the reduction of azido ketone 2a.In a microcentrifuge tube (2 mL), 2-azidoacetophenone (2a; 4 mg, 25 µmol; final conc.50 mM) was dissolved in 2-propanol (25 µL; final conc.5% v/v, approx.650 mM).Potassium phosphate buffer (425 µL; 100 mM, pH 7.0, 1 mM MgSO 4 ) and a stock solution of ADH (1 mg mL −1 ; final conc.0.1 mg mL −1 ) and NAD(P) + (0.7 mg mL −1 , 1 mM; final conc. 100 µM) in potassium phosphate buffer (50 µL) were added, and the mixture was shaken at 30 °C and 1000 rpm on a thermoshaker for 2 h.The reaction mixture was extracted with EtOAc (800 µL), the extract was dried over MgSO 4 and conversion as well as product ee were determined by GC-FID analysis.

Investigation of the substrate scope of ADH-A and KRED-NADH-110.In a microcentrifuge tube (2 mL), azido ketone 2a-f (25 µmol; final conc.50 mM) was dissolved in 2-propanol (25 µL; final conc.5% v/v, approx.650 mM).Potassium phosphate buffer (425 µL; 100 mM, pH 7.0, 1 mM MgSO 4 ) and a stock solution of ADH (1 mg mL −1 ; final conc.0.1 mg mL −1 ) and NAD(P) + (0.7 mg mL −1 , 1 mM; final conc. 100 µM) in potassium phosphate buffer (50 µL) were added, and the mixture was shaken at 30 °C and 1000 rpm on a thermoshaker for 2 h.The reaction mixture was extracted with EtOAc (800 µL), and the extract was dried over MgSO 4 .Conversion was determined by GC-FID analysis, while product ee was de