Reprinted from Journal of Organic Chemistry, 1986, 51,25.Copyright O 1986 by the American Chemical Society and reprinted by permission of the copyright owner.

Preparation of Optically Active l,2-Diols and a-Hydroxy Ketones UsingGlycerol Dehydrogenase as Catalyst: Limits to Enzyme-Catalyzed,Synthesis due to Noncompetitive and Mixed Inhibition by Product

Linda G. Lee and George M. Whitesides*

Departments of Chemistry, Haruard Uniuersity, Cambridge, Massachusetts 02138, and Massachusetts Instituteof Technology, Cambridge, Massachusetts 02139

Receiued August 7, 1985

Glycerol dehydrogenase (GDH, EC 1.1.1.6, from Enterobacter aerogenes or Cellulomonas sp.) catalyzes theinterconversion of analogues of glycerol and dihydroxyacetone. Its substrate specificity is quite different fromthat of horse liver alcohol dehydrogenase (HLADH), yeast alcohol dehydrogenase, lactate dehydrogenase, andother alcohol dehydrogenases used in enzyme-catalyzed organic synthesis and is thus a useful new enzyrnic catalystfor the synthesis of enantiomerically enriched and isotopically labeled organic molecules. This paper illustratessyrthetic applications of GDH as a reduction catalyst by the enantioselective reduction of 1-hydroxy-2-propanoneand 1-hydroxy-2-butanone to the corresponding fi 1,2-diols (ee = 95-98%). (R)-1,2-Butanediol-2-d1 was preparedby using formate-d1 as the ultimate reducing agent. Comparison of (R)-1,2-butanediol prepared by reductionof 1-hydroxy-2-butanone enzymatically and with actively' fermenting bakers' yeast indicated that yield andenantiomeric purity were similar by the two procedures. Reactions prcrceeding in the direction of substrate oxidationusually suffer from slow rates and incomplete conversions due to product inhibition. The kinetic consequencesof product inhibition (competitive, noncompetitive, and mixed) for practicai synthetic applications of GDH, HLADH,and other oxidoreductases are analyzed. In general, product inhibition seems the most serious limitation to theuse of these enzymes as oxidation catalysts in organic synthesis.

The NAD(P)(H)-dependent oxidoreductases are usefulin the synthesis of chiral synthons. Horse liver alcoholdehydrogenase (HLADH)1'2 has been most thoroughlyexplored; lactate dehydrogenase and r,-leucine de-hydrogenase have also been used.3'a The practicality ofthese systems for organic s5'nthesis has increased with therecent deveiopment of effective procedures for enzymeimmobilization and nicotinamide cofactor regeneration.tsTThis manuscript describes the use of another commerciallyavailable oxidoreductase, glycerol dehydrogenase (GDH,EC 1.1.1.6), to reduce cr-hydroxy ketones to chiral 1,2-diols.Previous studies have shown that chiral 1,2-diols having

(1) Jakor.ac, I . J . ; Goocibrand, H. B. ; Lok, K. P. ; Jones. J.B. J. Am.{'hem. Soc. 1982, 104, 4659-4665.

(2) For a review of early work using dehydrogenases, see: Jones, J. B.;Beck. J. F. t rn *Techniques of Chemistrv" ; Weissberger. A. , Ed. ; Wi ley:New \-Lrk, 1976; Vol X, pp 107-401. For recent work using HLADH. see:Ns. C S. \ ' . : \ 'uan, L.-C. : Jakovac. I . J . ; .Jones, J.B. Tetrahedron 1984.l ( j . i 2 : J ; "12 .13 . . l ones , , I . B . ; Takemura , T . Con . J . Chem. 1982 ,60 ,2950-2956. . Iones, .1. B. : F inch, M. A. W.; Jakovac , L j . Can. J. Chem.1982. n 'a) . :007 2011.

(3) F{ i rschbein. B. I - . ; Whi tesides, G. M. . ; . Am. Chem. Soc.1982. 104.4458-4,160.

1.1) Wandrev. L- . ; Buckmann, A. F. ; Kula, M. R. Rtotech. Rioeng. lg8l ,23 .2789-2802 .

(5) Whitesides. G. M.; \4 'ong, C.-H.; Pol lak, A. In *Asymmetr ic Reac-tit, 'ns and Processes in Chemistry'; i l l iel, E. L., Ot-suka, S.. Eds.; Americant lhemical Society ' : Washington, DC, 1982; pp 205-218.

(6) Whitesides. G. M.; Wong, C.-H. Aldr ichimica Acfo 1983. 16.27-34.(7) Findeis, M. A. ; Whi tesides, G. M. Ann. Rep. Med. Chem.l984. 19,

26|1-212.

Table I. Relative Rates of Reduction of a-Hydroxy Ketonesbv GDH/NADH"

reiativerateb K-, mM

substrate C..s. E.a. C. .s. E.a.

dihydroxyacetone(r?) -2,3-d ihvdroxypropanal(S) -2,3-d ihydroxvpropanal1 -hydroxy-2-propanone'

1 -hydroxy- 2- butanone'2-oxopropanal2- hydroxycrvclobutanone2- hydroxycl'ciopent ano ne2- hydroxycyclohexanonei3-hydroxy- 2- butanone

o Abbreviat ions: C.s. , CelLulornonas sp. ; 8.c. . Enterobacter Ge-rogenes. Reactions were conduc'teci at irH 7.6, 25 "C. Details otreact ion condi t ions are given in the Exper imental Sect ion. Theproducts of these react ions were not isolated. nor were their s t ruc-tures expl ic i t lv establ ished. Compounds which did not react underlhese condi t ions include 1,2-cyclohexanedione, 2, iJ- i ru lanedione,2-propanone, 1-chloro-2-propanone, acetaldehyde, 2-hydroxy-4,4-d imeth_v- lcyciohexanone. and 4-hydroxy- l i -hexanone. b The concen-i rat ion ot 'subst . rate in each exper iment was 100 mM unless indi-cated othenvise. 'The c<lncentrat ion <l t ' the subst , rate was 0.2 mMin these determinations c'f relative rate. Lclrver rates were observedat h igher concentrat ions of substrate (100 mNI) .

the same absolute configuration obtained here are obtainedby reduction of a-hydroxy ketones using fermenting bak-ers' yeast.r 'e We compare these enzymatic and fermen-

1 ( ) Q 4 A - ^ - : ^ ^ - r \ L ^ * : ^ ^ l Q . - ^ i ^ + , .

i002i013030

i3290i.101 445

100

2020

0.50.3

0.04

10

n n t t - ' l ' ) A ' l I Q A . / ' l O ( 1 / ) n t K Q n l < n / A -

26 J. Org. Chem., Vol. 51, No. 1, 1g86

tative methods and conclude that they generate productsin similar yields and enantiomeric purities. GDH can alsobe used to prepare chiral a-hydroxy ketones by oxidizingcis-1,2-diols to chiral a-hydroxy ketones or by resolvinga racemic mixture of a-hydroxy ketones kinetically byenantioselective reduction of one enantiomer. Chemicalsyntheses of chiral a-hydroxy esters,l0 chiral dihydroxy-cycloalkanones,rl and chiral a-hydroxy aldehydesl2 areavailable, but there are no g.r.."i methods to synthesizechiral a-hydroxy ketones.

Glycerol dehydrogenase is a nicotinamide cofactor-de-pendent oxidoreductase which catalyzes the interconver-sion ofglycerol and dihydroxyacetone. Several sources ofoxidoreductases having this activity are known, includingEnterobacter aerogenes,Ls-r6 Cellulomonas sp. (the en-zymes from these two sources are commercially availableand are used in this study), Erwinia aroideoe,ro andKlebsiella aerogenes.lT GDH from both Cetlulomonasand Enterobacter is relatively inexpensive ($+071000 Uand $600/1000 U, respectively; 1 U = 1 pmol min-1; 200U of activity converts ca. 1 mol of substrate per day toproduct). GDH is stable and easily immobilized. Theenzyme is used in several assay procedures.ls BecauseGDH contains autooxidizable thiols, reactions using it areconducted under an inert atmosphere. The antioiidantsdithiothreitol and mercaptoethanol activate the enzymeat low concentrations and deactivate the enzyme at higherconcentrations;le these antioxidants should be avoided orused sparingly in preparative-scale reactions.

ResultsSubstrate Specificity of GDH. Table I gives relative

rates of reductions of several substrates catalyzed by GDH;Table II gives relative rates of oxidation. Initial concen-trations of substrates in both sets of experiments were 100mM. These relative rates were obtained by observing thereduction of NAD or oxidation of NADH; products wereidentified only in cases where preparative quantities (>1g) were synthesized. Identifications of compounds thatreact very slowly as substrates should be considered ten-tative: experimental errors due to small quantities ofrapidly reacting impurities become important in these

(8) Guette, J.-P.; Spassky, N. Bull. Soc. Chim. Fr. t972.42t7-4224.(9) For recent wol\ using yeast reductions. see: Takaishi, y.; yang.

Y.-L.; DiTullio, D.; Sih, C. J. Tetrahedron Lett. Ig8Z.23. S4gg-S4gr.Z_hou, B. ;Gopalan, A. S. ; Van Middlesworth, F. ; Shieh, W.-R.; Sih, C. J.J. Am. Chem. Soc. 1983, 105, b92S-5926. Chen. C.-S. : Zhou. g. : Ctr-daukas, G.; Shieh, W.-R.; Van Middlesworth. F. ; Gopalan, A. V. : Sin. C.J. Bioorg. Chem. 1984, 12,98-117.

(10) Brown, H. C. ; Pai , G. G.;Jadhov, P. K. J. Am. Chem. Soc. 19g4.106, 1531-1533.

(11) Johnson, C. R.; Barbachyn, M. R. J. Am. Chem. Soc. 19g4. 106.2459-246t.

(12) Asami, M.; Mukaiyama, T. Chem. Lett.l98l, 98-96.(13) Burton, R. M. In *Methods in Enzymology"; Academic press:

New York, 1955; Collect. Vol. 1, pp 39?-400.(14) Lin, E. C. C.; Magasanik, B. J. Biol. Chem.1960, 2J5. 1820-1823.(15) Strickland, J. E.;Miller, O. N. Blochjm. Biophys. Actq 196g,I59,

22r-226.

^, (16) Sugiura,_M.; Oikawa, T.; Hirano, K.; Shimizu, H.; Hirata, F.Chem. Pharm. Bull. 1978. 26.718-72I.

. q7) _Rgch, F. E., Jr.; Lin, E. C. C.; Kowit, J. D.; Tang, C.-T.; Goldberg,A. L. J. Bacteriol. 1980, 141,1022-1085. The enzvme in Klebsiellafppears tn csftalyze th9 fi_rst_ step in anaerobic assimilation of glycerol.See-: _R_u_ch, F. E.; Lin, E. C. C. J. Bacteriol. L}ZS, I24, g4g-gb2. i.ieilssel,O. M.l Hueting, S.; Crabbendam, K. J.; Tempest, D. W. Arch. Microbiot.t975,104,83-87.

(18) Assays utilizing glycerol dehydrogenase include serum aldolasedetermination (Sugiura, M.; Hirano, K.; Voshimura, H.; Sugiyama, M.;Shimizu, H. Chem. Pharyt. BulI. 1976, 24,2647-26b0), glycer"oi in serumand beverages (Hinsch, W.; Sundaram, p. V. Clin. Chii. Acta lgg0. 104.ry-!4)l and tfiglyceride determination (Abdulrahman, S. A.; Christian,G. D. Anal. Chim. Acta 1979, 106,228-ZBt\.

(19) ryIcc1ego1W. G.;Phill ips, J.; Suelters, C. H. J. Biot. Chem.1974,

249,3132-3139. The authors suggest irreversible complexation ofthe zincat the active site as the mechanism of inactivation.

l ,ee an d \ \ 'h i t es ides

Table II . Relat ive Rates of Oxidation by GDH/NAD,,

relativerateb K-. mN,l

substrate C..s. E.a.

glycerol 100 1001,2-propanediol ' 60 40(R)-1,2-propanediol ' 70 4b(S)-1,2-propanediol" 0 01,2-butanediol ' 60 4i(B)-1,2-butanediol ' 60 40l ,2-hexanediol 45l ,2-octanediol (45 mM) 121,2-ethanediol 13l ,3-propanediol 39l ,4-butanediol 22,3-butanediol 2t 3t)3-mercapto-1,2-propanediol 333-chloro-l,2-propanediol 223-methoxy-1,2-propanediol 103-(methyl th io)-1,2-propanediol 30L-a-glycerophosphate 0Dl-a-glycerophosphate 11(R)-1-amino-2-propanol 4(S)-1-amino-2-propanol 0.5cls-1,2-cyclopentanediol 33 80cls-1,2-cyclohexanediol 6trans-L,2-cyclohexanediol 02-hydroxycyclohexanone 14crs- 1,2-cycloheptanediol 3cis- 1 ,2-cyclooctanediol 0D-galactose 2

"Abbrev ia t i ons : C .s . : Ce l l u lomonas sp . E .a . : En tc r , ,h r : , t r r 6 , . -rogenes. React ions were conducted at pH 9.0. Detai ls ,1 reat . r i .ncondi t ions are given in the Exper imental Sect ion. ( ' .m1.r , , r rndsw i th a re la t i ve ra te o f 0 (us ing GDH f rom ( r , l ! y l , , n1 , ,n , r , r i r r t . l r rde2 - a m i n o - 1 - p r o p a n o l , e t h a n o l , 2 - p r o p a n o l . 3 . 3 - d i m e t h r . l - 1 , 2 -p r o p a n e d i o l , 2 - p h e n y l - 1 , 2 - e t h a n e d i o l . ( , r ! . i r o n , - l . ; l - r ' r . c l o -pentanediol, cis,trans-1,3-cyclohexanediol, cr-c,l.hera.ed imethanol,myo- inosi to l , r ibose, g lucose, mannose, 6-deoxvgala( , tose. ervthr i -t o l , mann i to l . and so rb i to l . ^The c , rn . .en t ra t i r ' , r r , , 1 the .ub r t ra tewas 100 mM, un less ind ica ted o the r l ' i se . 'The t , r r cen t ra t i ' n . fthe substrate was 1 mM for the c leterrr i inat i , r . r . i the re lat i 'e rate.Lower rates were observed at h igher concentrat ions of substrate{ i00 mM) .

types of experiments. GDH from both Cei / ulomonus sp.and Enterobacter oerogenes were examined; similar sub-strate specificities and values of K- were observed for bothenzymes. The two enzymes have different specific activ-it ies and costs: Cellulomonas, g0 tJ l^g, $0.041tJ, En-terobacter, 5 U f mg, $0.6/U.20 For most synthet ic, ap-plications, the Cellulomonas species is superi<lr becauseof its higher specific activity and lower cost.

The data in these tables are not sufficiently svstematicto define all of the structural features required f or activityof an a-hydroxy ketone or l,2-diol as a substrate, but theirmost useful implication is that for ketones (i.e., R' I Hin eq 1) the group R is restricted to H, CH3, or a o'clic aikyl

C..s. E.a

9 90.10 0.090.07 0.09

0 . 1 5 0 . 1 50.09 0.09

20:t: l

1 . 1 l rt ) L

H9 .oa t ; (

/ \, R R "

y'\ !"- - ' ' / \\ )

HO-, gHx>--*.,n n)/ \

R R '\. ,,H, CH3, cyc l i c o tky t CH2X

connected to R'. The group R' can be a number of simplealkyl groups (CH2OH, CH2SH, CH2OCH,r, (CH2)"). Cer-tain of the substances listed in the tables that appear notto react may deactivate the enzyme (chloroacetone, cy-clohexane-1,2-dione) and should probably be discounted

(20) Several of the large-scale reductions were performed with immo-bilized GDH from Enterobacter; GDH from cellulomonas onlv recentlvbecame available.

Optically Active 1,2-Diols and a-Hydroxy Ketones

on this basis. The significant reactivity of 1,3-propanedioland 1,4-butanediol suggest that the two hydroxyl groupsneed not necessarily be adjacent, but the low reactivity of1,3-cyclohexanediol and 1,3-cyclopentanediol indicate thatthere are additional, presently undefined constraints onacceptable geometry for these groups.

Applications of GDH in Organic Synthesis. Ste-reoselective Reduction of Achiral a-Hydroxy Ketonesto Chiral 1,2-Diols.21 Reductions of a-hydroxy ketonesare straightforward. Cofactor regeneration was accom-plished with formate and formate dehydrogenase (FDH)(eq 2; DH is the ultimate hydride donor used for in situregeneration of NADH).4'22 Conversion of 1-hydroxy-2-

J . O re . Chem. , Vo l . 51 , . ^Vo . 1 , 1986 27

6 0

t 6Bose( m L ) e

7" Rocemic

3o Substrote

?o Consumed

t o

a)

o 20 40 60 80 loo 120

T i m e ( h )Figure 1. Quanti ty of 2.5 N KOH added to two reactors main-tained at pH 7.6. The symbol (o) represents a reactor whichcontained 5, glucose-6-phosphate, GDH, G-6-PDH. and NAD.The symbol O represents a reactor which combined startingquantities and concentrations of 5, glucose, and NAD equal t<rthose of the first reactor and approximately the same numberof units of 'both GDH and GlcDH.

acid or gluconic acid generated by oxidation of the hydridedonor; this procedure was more convenient than GLC orenzymatic assa! 's. Figure 1 shows representative results.The ee of the reisolated rr-hydroxy ketones was determinedfrom the tH NMR spectra of their (-)-MTPA esters.2eThe absolute configuration of enantiomerical ly enriched5 was assigned as ,R on the basis of the sign of its rotation(-) at 589 nm.:r0 Treatment of a-hydroxycyclobutanone8 using condit ions similar to those in eq 4 resulted inreduction, but the recovered 8 was stiii racemic (ee = 07").

Equations 3 and 4 give both observed values of ee andthe values of the enantiomeric ratio, E, obtained using theformulas of Sih.n1 The value of E calculated from ex-perimental data is verv sensitive to the extent of conversionof the subst ra te . These extents o f convers ion were no lmeasured indepenclentl l ' here but are estimates based onthe quant i t l 'o f ' the reduc ing agent (DH, eq2) added. Asa resu l t , r 'a iues o f l - in these equat ions. and e lsewhere inth i s pape r . A re app rox ima l e .

Enant io top ic Stereoselect ive Oxidat ion o f Meso-1,2-diols to a-Hydroxy Ketones. Enantioselective oxi-dation of cis-1,2-cyclohexanedioi (9 mmol) was accom-pl ished using diaphorase, methylene blue, and dioxygento regenerate NAD (eq 5).t ' Reaction stopped sponta-neously when 30% of the start ing material had beenconsumed, apparently due to product inhibit ion. Theabsolute configuration at the chiral carbon of 2-hydroxy-cyclohexanone was not determined, but is assumed to beS. Oxidative resolut ion of 1,2-butanediol was carr ied outby using 2-oxoglutarate and glutamate dehydrogenase(GIDH) to regenerate NAD (eq 6). Reaction progressappeared to stop when -307, of'the racemic suhstrate hadbeen consr-rrneC.

Reduct ions Us ing Act ive ly Ferment ing Yeas i .Bakers' yeast grclwing on glucose in an anaerobic env:-

(28) The reductiori of 7 was carried out at two vaiues of pH: pH 5.5anci 6.5. The reaction at pH 5.5 produced (r?t-2-hydroxvc-vclopenianoneol iATc ee; the reaction at pIJ 8.5 produced (R)-2-hvdroxvcyclopentanoner,1 86% ee.

(29) (R)-5 couir i he inverted at the chira l center wi th the N{ i tsunobureactior.r' i" itriphenylphosphine, dieth-vl azodicarboxl' late, and (-)-,r-N{TPA acid)wi th some loss of enant iomeric pur i t5. ' ( f rom 8E% tai0%).' fhe enant iomeric pur i lv of the C-11 posi t ion of the pro<luct (35i- l l -O( ) -MTPA-2-bt t t .anone was measured by tH NMR and fo lnd to he 7A7"ee. Simi lar t reatment of racemic 5 provided 3-0-NITPA-:)- i rutanone <; , f( t Ia ee at the i ' - ;1 pos. i t ionl no diastereoselect ion in tht react lon wasobserved (see Figure 6) .

(30 ) B lom. R . F I . ; . \ ; n . Chem. Soc . 1945 , 67 ,494 .(31) Chen. i . . - - { . : Fui in ioto. \ ' . ; Girdaukas. G.; Sih, C. . i . ,1. .4rn. Chem.

soc. . lgg2, 10i . T: ] ! r .1--?299.( i l 2 ) Lee . L . {1 . ; \ 1 -h i tes ides . L i . t u i . j . . 4m. Chem. Soc . I985 . /07 .

69!)9-?0t)S.

2 a

A N

4 0

( 2 )(R ) -? (98% ee )( R ) - 4 ( 9 8 % e e )

butanone to (R)-1,2-butanediol was carried out withoutdifficulty on a scale generating 70 mmol of isolated prod-uct. Conversion of 1-hydroxy-2-propanone to (R)-1,2-propanediol was also straightforward. The ee for both diolswas determined to be >9870 by examination of the lH

NMR spectra of the 1-O-tosyl-2-O-(-)-u-methoxy-a-(tri-fluoromethyl)phenylacetyl (MTPA) derivatives.23 Sub-stitution of formic acid-d, for formate2a generated (R)-1,2-butanediol-2-d, (98% dr). These reductions arestraightforward at this scaie and with the quantities ofenzyrne used here but are siow (complete reaction requiresapproximately 19 days).

Kinetic Resolutions: Reduction of Racemic a-[Iy-droxy Ketones. Reduction of racemic a-hydroxy ketonesusing 0.5 equiv of the ultimate hydride donor resulted inenantiomeric enrichment of the remaining unreduceda-hydroxy ketone (eqs 3,4). Giucose-6-phosphate/glu-

O . 5 e q u i vH o o G t c - 6 - P o r G l c H o o H o o H

>-\ ----------------

",,\ - n7--\Fi r.j;

H , C C H , 6 O H H , C C H , H , C C H ,

t R ) - 2(88"A ee)

E = l O

H O on",l---(1 |

''..-r/

tR) -z(70- 86 % ee)

E = l O

cose-6-phosphate dehydrogenase iG-6-PDF{}j: ' or glu-cose/giucose dehydrogenase (GlcDFi)262; were used toregenerate NADH in situ. The couise oi these reactionswas followed by monitoring the quan:;tv of KOH solutionadded at constant pF{tt to neutraiize the 6-phosphogluconic

(21) The stereochemical nomenclature used is ciescribed by N1arch:March, J. In "Advanced Organic Chemistrv: Reactions, Mechanisms. andStructure', 2nd ed.; McGraw-Hill: New York, 197?: po L2l\-124.

(22) Shaked, Z. ; Whi tesides, G. lv i . J . Ant . Cher, t . Soc. 1980. 102.7104-7105.

(23) Dale, J. A. ; Dui l . D. [ , . : Mosher. H. S. i . Org. ( 'hent . 196{ i . ,?J.2543 2549.

(24) Wong, C.-H.; Whi tesides, G. N{. J. Am. Chem. Soc. 1983, 1t l , i .5012-5014.

(25; Wong, C.-H.; Whi tesides. G. M. J. Am. Chem. Soc. l$81, /0.1 j .4890-4899.

(26) Wong, C.-H., pr ivate communicat ion.(27 )Levy , H .R . ; Logq 'ss . F . A . l Vennes land .8 . , i . , : t n i . { ' hem. Soc .

1957, 79. 2949*2953.

O . 5 e q u i v

G l c - 6 - P H O O HH*- - - -H 14)/ \

H O O5 / /x

ronment reduces ketones to alcohols.3s The enzymesresponsible for these reductions are probably in major partnicotinamide cofactor-dependent oxidoreductases, al-though the specific enzymes have not been identified. Theadvantage offered by reduction with yeast is simplicity:no enz)rnes need be isolated or immobilized and cofactorregeneration is accomplished by enzyne systems alreadypresent in the yeast. The disadvantage of yeast reductionsrelative to those using isoiated enz5rnes is that they areintrinsically messy. The reactant and products are presentin dilute solution in a medium containing yeast .e1ls, tr.r-trients, and other metabolites. Further, since the reactantsor products in a ketone reduction using yeast may bemetabolized, yields may not be high. Finally, several en-zymes may be present in the yeast that are capable ofreduction of the substrate. If these enzJrnes have differentenantioselectivity, their combined action mav lower en-antioselectivity.

We were interested in comparing the practical efficacyof ketone reduction using cell-free enz5rnes and fermentingyeast. Scheme I summarizes experimental results.

The major conclusions from this limited comparison arethat these enzymatic and fermentation reductions giveproducts with the same absolute configuration and similar(high) enantiomeric excesses and that the yields are higherfor the enzymatic reductions, the times required to ac-complish the reduction longer, and the isolations simpler.The yeast reductions are operationally simpler to carry out.

Preparat ions of Epoxides. The chiral d io ls (R)-1,2-propanediol and (R)-1,2-butanediol-2-d, were convertedinto the corresponding epoxides in two steps:3a init ialtreatment with hydrogen bromide in acetic acid providedthe (R)-1-bromo-2-acetoxyalkanes; subsequent treatmentwith potassium pentoxide yielded the (R)-1,2-epoxy-alkanes. The ee of the epoxides were determined by re-gioselective ring-opening with thiophenol, followed bypreparation of the 2-O-MTPA esters and examination bytH NMR spectroscopy. The ee was >98% for both (rBi-1,2-epoxypropane and (R)-1,2-epoxybttane-2-d 1.

Bffect of Product Inhibition on Reaction Rates:Optimization of Reaction Conditions. Preparative-scaleoxidations of vicinal diols to a-hydroxy ketones using GDHwith in situ cofactor regeneration are much more difficultto accomplish than are the reductive reactions: productinhibition seriously slows the rates of oxidative reactionsand limits the final concentrations of products which canbe obtained. This section analyzes the influence of productinhibition on the kinetics of reactions carried out underthe constraints imposed by preparative conditions (espe-

(33) For reviews of microbial transformations, see: Sih. C. J.: Chen.C.-5. Angeu:. Chem., Int. Ed. Engl. 1984, 2J, S7O-b78. Fischli, A. In'Modern Synthetic Methods"; Scheffold, R., Ed.; Salle and SauerlAnder:Frankfurt, 1980; Vol. 2, pp 269-350. Sih, C. J.; Rosazza, J. p. In'Techniques of Chemistry'; Weissberger, A., Ed.; Wiley: New york, 1976;Vol. X, pp 69-106.

( i l4) Golding, B.T. ; Hal l , D. R. ; Sakr ikar, S. J. Chem. Soc. . perk inTrans. 1 1973, 1214-1220.

Lee and Whitesides

Scheme I. Comparison of Enzymatic andFermentation Transformati ons

,oi.*,-* ,o5l",

"oi.,",-----' "".-X:"r. zo (s8) 4r (ee)

cially the desirability of high concentrations, completeconversions, and short reaction times) and suggests stra-tegies which minimize the undesirable consequences ofproduct inhibition. This analysis indicates clearly, how-ever, that product inhibition is a practical problem indehydrogenase-catalyzed oxidations of alcohols to ketonesthat may, in fact, be considerably more serious that theproblems posed by cofactor regeneration or enzyme costand operating lifetime.

The GDH-catalyzed oxidation of diols proceeds throughan ordered bi-bi mechanism;le the order of addition isrepresented by eq 7.35 The rate of reaction for an ordered

HO_ -OH

"7-T"( )

I

6 D H , 9 N 9 .

3 0 %

1986

Ho'. ,9

"-fl

( s ) - !g( 9 1 % c e )

( 5 )

(6)

, -Y ie ld, o /o (o/" ee)-1

Enzymolic Fcrmrnlolivr

50 (98) 38 (95)

oH GDH, 9H 9H o i / -

\-z \-,-

26 "h4

o

"o-.-''\.- +

1x

H O H

"o..--\-.-

( . r ) - t( 36 "/" ec)

E = 2 0

A B(NAD) (d io l )

i lP g

(o -hydroxy ke tone) (NADH)

f rE E A ( E A B - r P O ) E Q

( 7 )

bi-bi mechanism in the absence of Q(NADH)36 is givenby eq 835 (V-", = h"urfenzyme]). In this equation, the

U_ =T/Y max

rBr/[K,""(1+ ff)('. fr#)./ K,"e rPl \tB t I I + ;+ + = | | (8 )\ tA l K ,o l l

Michaelis constant for, e.g., Q, is given by K-q and thecorresponding inhibition constant by K,o. The product p(ketone) acts as a "mixed-type inhibitorr' with respect tothe substrate B (diol): the inhibition affects both K6 andV-",.tt Equation 8 may be simplified to eq 11 if the con-centration of A is saturating (eq 9; for dehydrogenases, [A]= [NAD] i 1 mM) and if the quantity (KioK-p/K-q) isexpressed as a function of K1o (eq 10). We define the

[A] >> K^e,, K;^

KtoK^,

KtoK-e

KrrK^,

K-e_ *K io

IB ]

(e)

(10a)

(1ob)

( 1 1 )V^u* +rBr('.fr)

extent of reaction as .R (eq 12). We further assume thatsubstrate is converted only to product and that the productexists only in the form of a single inhibitor I (eq 1B).

(35) For a discussion of the ordered bi-bi mechanism and an analysisof the kinetics, see: Segel, I. H. In *Enzyme Kinetics'; Wiley: New york,1975; pp 561-590.

(36) In practice, Q(NADH) is maintained at a very low concentrationby an efficient NAD regeneration system, e.g., 2-oxoglutanate/glutamatedehldrogenase-32 (See also: Wong, C.-H.; McCurry, S. D.; W*hitesides,G. M. J. Am. Chem. Soc. 1980, 102,7938-7Sgg.\

K-"('.fr)

Optically Active 1,2-Diols and a-Hydroxy Ketones

R = 1 (r2)

Equation 11 can then be rewritten as eq 14 (in this andsubsequent equations describing mixed or noncompetitiveinhibition, K,n is the Michaelis constant of the diol sub-strate and K; is the inhibitor constant of the ketoneproduct).

l r l = tso l - [ s ](1 -n )

V-"*

The value of r in eq 11 determines whether productinhibition is noncompetitive or mixed. When r is unitythe inhibition is noncompetitive (only V-", is affected).In practive, r is rarely unity and product inhibition isusually mixed (both K-s and V-,* are affected in eq 11).The value for .r can be determined experimentally; foryeast and liver alcohol dehydrogenase it has been foundto be 0.19 and 0.15, respectively.3T

For comparison and for reference, the simpler equationsanalogous to eq 11 and 14 describing a reaction in whichthe product is a competitive inhibitor are eq 15

"tt6 16.ss'as

Km

O.OrP

r 0

1986 29

(-t')

,.#.*[?,'-R) +ft']

.(ft) 'l"(?),'11+-(?)

1), mixed (eq 18, x =(eq 19); the values for

N O N C O M P E T I T I V E

I K n / K t = O . O l

o Km/K, = 0.3

A Kn/Kt =5 O

M I X E D

r = O . l 5

(13 )

(14)

( 1 5 )

(16)

( 1 7 )

(18)

v

' m o x

Arbitrarily setting the initial substrate concentration [Ss]as 10 K,"e (eq 17) allows eq 14 and 16 to be rewritten aseq 18 (for mixed inhibition) and 19 (for competitive in-hibition). Figure 2 plots values of u lV-", vs. .B for non-

[So] = 10K,"

( 1 - R )

7-"t

11 + n [ _n)

( 1 - R )

C O M P E T I T I V E

Figure 2. Rate of reaction, u, relative to the maximum rate ofreaction V-"" as a function of extent of reaction R for reactionsin which product is a noncompetit ive (eq 18, x = I), mixed (eq18, r = 0.15), and competit ive inhibitor (eq 19). The calculatedcurves are labeled with the ratio K^f Ki. Experimental valuesare given by the symbols l, (t, a. Experimental values for K-for each reaction were determined from Eadie-Hofstee plots;values for K1 were determined according to the method of Dixon.62Kinetic parameters used were as follows: for reduction of 3-hydroxy-2-butanone to 2,3-butanediol, K,,, = 2 mM, K1 = 200 mM,K^lK, = 0.01; for oxidation of cis-1,2-cvclohexanedimethanol to(1f i ,6S)-8-oxabicyclo[4.3.0]nonan-7-one, K^ = 26 mM, Ki = 80mM, K^f K; = 0.3; for oxidation of glycerol to dihydroxyacetone,K^ = 15 mM, Ki = 0.3 mM, K^f Ki = 50.

K^lK, ranges from 0.01 to 10. Experimental points forthree representative reactions involving NAD(H)-de-pendent dehydrogenases are plotted with the curves cor-responding to noncompetitive inhibition. This particularjuxtaposition of experimental and theoretical curves is anarbitrary one, since the value for r (eq 10) for these re-actions is not known. In practice, however, i f one is in-terested in qualitative estimations of rates and reactiontimes one can best approximate product inhibit ion ofdehydrogenases by assuming simple noncompetit ive in-hibit ion i f no specif ic value for r is avai lable.

Figure 2 demonstrates that reactions with K^f Ki < |proceed rapidly to completion. For K^l Ki ) 1, however,it is difficult to achieve high rates and high conversionssimultaneously. The consequences of incomplete reactionfor GDH-catalyzed reactions are the fol lowing: For oxi-

(mX'.U

l/7 max

U- =( t -

H).'R)

v*"' ffi.1+ R(? ')

V-*

competit ive (eq 18, x =competit ive inhibit ion

(1e)

0.15) andthe ratio

(37) Wrat ten, C. C. ; Cle land, W. W. Biochemistry 1963. 2,935-941.(38) In this example, A(NAD) and Q(NADH) are competitive inhib-

itors and the competitive product inhibition represented in Figure 2would be expected if [P] (ketone) = g and if the concentration of B(alcohol) was saturating. Other examples of the control of the ratio K^lK;in competitive systems are kinases; for derivations and discussion, see:Atkinson, D. E. In "The Enzymes"; Academic Press: New York, 1970;Vol. 1, pp 461-489.

(39) Using the number of units of enzyme available, U, and the desiredextent of reaction, R, one can calculate the optimal volume required tocomplete the reaction from eq 27. For example, consider the HLADH-catalyzed oxidation of 100 mmol of cis-1,2-cyclohexanedimethanol to 95mmol of the lactone (ft = 0.95) using 100 U (pmol product produced/min)of HLADH. The experimentally determined value of the ratio K^lKlis0.3. The optimal value of the ratio [So]/Ki is then found to be 1.4 fromeq 27. The experimental value of K1 is 0.08 M. We take, therefore, thevalue of [So] to be 0.11 M and the optimal volume of the reaction to be0.96 L. The experimental value for the optimal volume is not known.The time t cannot be accuratelv determined unless the parameter r (eq10) is known.

O . O Ex len t o f Reoc f i on (F )

30 J. Org. Chem., Vol. 51, No. I, lgSG

dative kinetic resolutions of diols, the diols are recoveredin low enantiomeric excesses; for oxidative, enantioselectivesyntheses of a-hydroxy ketones, low yields are obtained,although the enantiomeric purity of the a-hydroxy ketonerray be high.

The curves in Figure 2 represent the velocity of thereaction for different values of K^f Ki at a fixed value of[So]/K- = 10. In fact, this value of [Ss] is not necessarilythe one which maximizes the rate of reaction or extent ofcrnversion. we desired a method that would allow us toestimate the best value of [Ss] for known values of Ki andK* and for a chosen extent of reaction R. We define *6est"in this context to mean that value of [s6] which minimizesthe time required to achieve a given value of R.

Defining velocity as a function of the concentration ofinhibitor [I] (eq 20) and with the fact that K-l tsol is thequotient of the two ratios K^lKi and [Ss] lKi Gq 2t), theequation for mixed inhibition (eq 14) can be rewritten aseq 22. Equation 22 was integrated and evaluated from

dtrl d([so]n)l l = -- d r d t

K-l[So] = (K^/K;)([So] /K)-,

Lee and Whitesides

O r B

O o 6

O . 4

O o 2

O r O

l r O

O o 8 '

O Oo4 O.8 | c2 1 .6 ? .O

rrme t[s.] z x.o, [r t t

Figure 3. Time (in units of [56]/V-",) as a function of extentof reaction R (eq 12) for reactions in which product is a non-competitive (eq 23, a = 1), mixed (eq 23,x = 0.15), and competitiveinhibitor (eq 24). The value of the rat io K^f Ki is 0.1.

ference reflects, of course, the fact that for a given valueof [56]/K; and for a particuiar type of inhibition, the valueof [56] f K^in Figure 3 is greater than that in Fisure 4 bv102 .

The curves in Figure 3 (K^lKi = 0.1) represent an "easy"extreme from the vantage of practical catalysis: Thesubstrate binds more t ightly to the active site than theinhibitory product, and high values of I i are achievedrelatively rapidly. Note, however, that even for a constantratio of starting concentrations of substrate to enzyme, theconversion achieved in a given time decredses as thesubstrate concentraticln increases beyond a certain point(from [So]iKi = 1 to [Sn]/Ki = 10) for noncompetitive andmixed inhibit ion. This fal loff in rate with conversion re-ilects the inhibitory effect of product. iinzyme is inhibitedat a concentrat ion of product determined by Ki, and anincrease in [S,y] does not result in an increase in rate, onlyin an increase in the concentrat ion of unreacted sn (thatis . in a decrease in f r ) when inh ib i t ion occurs .

The curves in F igure 4 (K^ lK i = 10) represent a"difficult" extreme for preparative catalysis. The substratebinds /ess t ightly than the inhibitory product, and rates(fbr a given quantity of enzyme) are much slower than forthe corresponding case in Figure i3 in any event. Further,

O .

O .

c

;C)

oE

o

ox

t!

t l

t z

ar = ffrr -

" ,

t [

(20)

(21)

+-[So]E'_ +

Ki

K^

K,

ft-')". r l dR(22)

(25)

(?.(?) '

f = 0 (R = 0) to t = t to obtain eq 28. The expression forcompetitive inhibition (eq 16) was integrated by a similarprocedure to obtain eq24. (The validity of both integra-tions can be checked, if desired, by differentiation.)

,=ff[H''.('-ft)" (ft.

,=ff[(' ?).-tt.(?) '(X)

' ]h(1 -n)l e+y

ff i)h(1-n)] (23)

Figures 3 and 4 plot the extent of reaction r? vs. time fornoncompetitive (eq 23, x = l), mixed (eq 28, r = 0.1b), andcompetitive inhibition (eq 24) for different values of theratio ISu]/K1. In interpreting these curves. it is heipful toremember that the time variables are [56] (since K1 andK,', are fixed by the enzy'rne being considered) and R. Theabcissa in these figures is given in units (with dimensiont ime) of the rat io [So]/V."* (eq 25). Thus, numerical

lSul [So] so= _

v*o* ft"",IEo] A.u,Eo

values along the abscissa depend on the ratio of' thestarting concentration of substrate to enzyme. The curvesior dif ' ferent values of [Se]/K; can thus only be compareddirectly if the value .f [s6]/ [Es] is considered as a constant:that is , as the rat io ISo]/Ki increases, [Eo] increases pro-portionately. Note, also, that the abscissa fbr Fisure 3\K,,lKi = 0.1) differs from that for Figure 4 by alactorof'100. That is, the reactions in F-igure 3 are proceedingroughly 100 times I 'aster than those in Figure 4; this dif-

N O N C O M P E T I T I V E

Km- _ - n I

K / I

O . l -

ro

I

M I X E D , x = O o 1 5

Kp_ - = O . l I

c o M P E T I T I V E I o

Optically Active 1,2-Diols and a-Hydroxy Ketones J. Org. Chem., Vol. 51, No. 1, 1986 3l

I O.2 Oo4 0.6 O.8 loOExtent of Reoct ion @)

Figure 5. Values of the ratio [Ss]/K1, which minimize the timerequired to achieve an extent of conversion R (eq 27). Note thatfor each value of K^f Ki, the optimal value of [36]/K; remainsessentially constant between R = 0.2 and 0.9.

This analysis indicates that for reactions subject tononcompetitive and mixed inhibition by product, there willbe an optimal vaiue of the starting concentration of sub-strate [56]. At low values of [56], the efficiency of thereaction will be low because the system will use dilutesolutions and large volumes, and because rates may be lowif [S0] lK^> I0. At high values of [Ss], rates and con-versions are limited by product inhibition. We pose thefollowing question: for a given ratio of starting concen-trations of substrate and enzyme, what starting concen-tration of substrate minimizes the time required to achievea selected extent of conversion? An alternative phrasingof this question which may be more relevant to practicalsynthesis is as follows: For given quantities of substrateand enzyme (and thus of the ratio So/Eo), what uolumeof solution minimizes the time required to synthesize agiven quantity of product? To solve this (these) prob-lem(s), we note that the tSol/ V-", term in eq 23 is inde-pendent of volume (eq 25). Differentiating eq 23 withrespect to [Ss] (eq 26) (or with respect to volume \/) andsetting the derivative equal to zero yields eq 27 . Figure

dr so [n ' K^ - I- = U = - l = + - - = l n ( 1 - B ) l t Z A tdlSol

- K",,En | 2K, [S^12

--- ' - - ' ' J

a=lAIn(1 -o'1"' (27\3s

5 plots eq 2i for representative values of K^l Ki. Theseplots are most simpiy interpreted physically by consideringKito be the same for all three lines. With this constraint,for higher values of K^ it is necessary to use higher [So](and, proportionately, [E6]) to achieve a chosen conversionin the minimum time. For a given value of K^f Ki, theoptimal value of [56]/K; remains roughly constant betweenR = 0.2 and 0.9. Note that R is not directly correlated withtime in these plots (see Figures 3 and 4); estimates of timecan be derived from eq 23, if desired.

Discussion

Among the many commercially available oxido-reductases, only horse l iver alcohol dehydrogenase(HLADH) has been thoroughly investigated for syntheticutility. The most useful applications of this enzyme havebeen to the enantioselective oxidation of 1,4-diols to lac-tones. This type of reaction seems only moderately in-fluenced b5' product inhibit ion. The data in this papersuggest that glycerol dehydrogenase (GDH) should aiso beuseful in chiral synthesis and that the most useful appli-

l . O

--ro{

t'4 /Ki

I

c

=oooE.

xLrJ

l l

t

t2

I

ts"lKi

I

6

4

2

O .

O .

O r

O .

O . O

l . O

o . 8

M I X E D , x = O o 1 5

Km

K i

EiKi

-\

O . ?

0 . O

l o O

O . 8

0 . 6

O . 4

O . 2

O . O

12 16 20

Time ( [s. ] z x.o1[e"] )

Figure 4. Same as Figure 3, except the value of the ratio K^f Kiis 10.

for any values if [S0] lKi for mixed inhibition it is im-practical to achieve high conversions: Product inhibitionis severe even for moderate extents of conversion [R - 0.3).

Of the three sets of experimental data plotted in Figure2, one-the reduction of 3-hydroxy-2-butanone-is verystraightforward: K^lKt = 0.01. Of the two oxidations, theHlADH-catalyzed oxidation of cis-L,2-cyclohexanedi-methanol to the corresponding lactone falls close to the*easy" kinetic extreme represented by Figure 3: The ratefalls off approximately iinearly with conversion. This typeof reaction has been exploited extensively and successfullyby ,Iones and co-workers for the preparation of 1-10-9quantities of products. Note, however, that a 90To con-version, the rate of catalysis is only l0-20%o that at thestart of the reaction. This falloff suggests difficulties inefforts to scale such reactions to high concentrations andlarge quantities, unless some other strategy is used torninimize product inhibition (such as continuous extractionof the product as it is formed.) The third experimentalcase in Figure 2-the oxidation of glycerol to dihydroxy-acetone-is a *diff icult" case of the type represented formixed inhibition in Figure 4. Product inhibition is severeeven at very low conversions, and it is not possible toachieve high conversions at any practical concentration ofenzvme. We bel ieve this type of product inhibi t ion un-derlies the slow rates and low c<lnversions characterizingthe reactions summarized in eq 5 and 6.

O .

O .

COMPETITIVE

32 J. Org. Chem., Vol . 51, No. 1, 1986

cations will lie in the enantioselective reduction of pro-chiral and racemic a-hydroxy ketones. Product inhibitionin oxidations with GDH is likely to be severe, and unlessthe product can be removed continuously as it is formed(by extraction, distillation, or further reaction) or unlessK, is small, reactions proceeding in the direction of oxi-dation will be slower and complete conversion more dif-ficult than those run in the direction of reduction. Glyceroldehydrogenases from two sources are commerciallyavailable: Cellulomonas sp. and Enterobacter aerogenes.Both enzymes appear to be R specific: Both utilize (r?)-l,2-propanediol but not (S)-1,2-propanediol. The enzymeshave similar substrate specificities. GDH from Cettuto-monas has a higher specific activity and is less expensivethan that obtained from Enterobacter and is thus pre-ferred for most applications. GDH accepts a variety ofdiols and hydroxy ketones as substrates. We found fewcompounds that were substrates for both glycerol de-hydrogenase and horse liver alcohol dehydrogenase.a0Ethanol, cls-cyclohexanedimethanol, and cyclohexanolwere substrates for HLADH but not for GDH; cis-1,2-cyclohexanediol, glycerol, and glyceraldehyde were sub-strates for GDH but not for HLADH. The presence ofvicinal diols in a substrate promotes activity with GDHand hinders activity with HLADH. The presence of morethan three hydroxyl moieties apparently inhibits activitywith GDH: Sugars were generally poor substrates. Bulkysubstituents a to the ketone moiety were not tolerated,although L,2-n- alkanediols were generally good substrates.The group R'in eq 1 is probably l imited in size to CH,.unless the substrate is cyclic: 3-Hydroxy-2-butanone is agood substrate whereas 4-hydroxy-3-hexanone is not.

Reductions catalyzed by GDH were successfully per-formed by using three different NADH regenerating sys-tems: formate/formate dehydrogenase (FDH), glucose-6-phosphate/glucose-6-phosphate dehydrogenase (G-e-PDH), and glucose/glucose dehydrogenase (GlcDH).Formate/FDH is the least efficient because the enzymehas low specific activity and high cost (0.4 lJ l^g, $0.8/U).41Formate and CO2 cause no problems in workup. Glu-cose-6-phosphate/G-6-PDH is very efficient (600 U/mg,$0.02/U), but commercially available glucose-6-phosphateis prohibitively expensive ($1500/mol) for large-scale re-actions. This compound can, however, be easily syn-thesized on multikilogram scale.2s Separation of productfrom 6-phosphogluconate may in some instances be in-convenient. Glucose/GlcDH is probably the most con-venient system for laboratory-scale reactions: The enzymeis moderately expensive but has good specific activity (200U l^9, $0.2/U). Glucose is readily available. Gluconic acidis normally easily separated from product. In each method,one balances cost against efficiency; glucose/GlcDH is agood compromise. GDH successfully resolved racemic3-hydroxy-2-butanone in 8-g quantities with 88% ee;a2 both

Lee and Whi tes ides

glucose/GlcDH and glucose-6-phosphate/G-6-PDH wereused uneventful ly for regeneration.

The most important result to emerge from this work forthe use of enzymes as catalysts in practical organic syn-thesis is the analysis of the importance of noncompetitiveand mixed product inh ib i t ion. In pr inc ip le , the properstrategy for addressing competit iue product inhibit ion isto increase the concentration of reactant. Since increasingthe concent ra t ion o f 'product in the react ion is genera l lyhelpful in svnthesis. the desirabi l i ty of increasing thereactant concentrat ion is not necessari ly a disadvantage(although limits in solubility of reactant, and denaturationof 'enz1'me in s<l lut ions containing high concentrat ions oforganic substances may l imi t the pract ica l i ty o f th isstrateg-v). The corresponding strategy for overcomingnoncompetit i t , t , and mixed inhibit ion must be to l imit theconcentrat ion of product: Simply increasing the concen-trat ion is not. in general. helpful and may, in fact, beclearly undesirable tb5' increasing the concentrat ion ofunconverted reactant remaining in the reaction mixtureat the point at which product inhibit ion causes the levelof enzymatic activitv to fall to an impractically low value).In the la t ter ins tances, s tar t ing reactant concent ra t i< lnmust be l imi ted to va lues suf f ic ' ient lv krw ' that h igh con-version to product is possible before the product concen-trat ion becomes suff icientlv high to cause kinetical lv un-acceptable product inhibit ion. For svst€ms in * 'hich K.,, ,Ki5 1 , t he l im i t a t i on o f r eac tan t concen t ra t i r ) n n ta \ . bep rac t i ca l . howbe i t a t t he i nconven ience o f n ran ipu la t i ngd i l u te so lu t i ons o f p roduc t s and reac tan t s . Fo r svs temsin wh i ch Kn , , 'K , > f . i t i s d i f f i cu l t t o ach ie r . e h i gh con -ve rs i on o f r eac tan t s t o I t r oduc t s a t an ac ' c ,e l ) t ab le ra te ,un less product can i te r€ l l to \ '€d t or . r t inLrous l r - .

An impor tant quest ion concern ing th i : r r ' r , rk . a . * 'e l l aswork d i rected toward svnthet ic appl icat ior . rs , , l HLAI)H,is that o f the c i rcumstances in r r 'h ich \AI ) r ,o f31 '1 , ) r - re-qu i r ing enzymat ic ox idat ion- reduct ion rea( ' t i ( )ns arepreferable to enz5rme-catalyzed, enantioseIective hr.dr,,lr-sisas a method of preparing chiral alcohols and cierir at ir .es.Answering this general question-the question ot rherelat ive ut i l i ty of cofactor-requir ing and cofactor- inde-pendent reactions-is one important focus of current w'orkin appl ied enzymology. No esterase has been tested forenantioselectivity in hydrolysis of esters of a-hydrox1.ketones, and reduction reactions have clear ut i l i ty incertain reactions involving the introduction of isotopes. Inenantioselective production of monoalcohols, 1,2-diols, andrelated substances, however, current research suggests thatlipases or esterases provide the basis for more practicalpreparative methods than do GDH and HLADH.6,7'43-45

Experimental Section

General. UV spectra were measured using a Perkin-ElmerModel552 UV-vis spectrophotometer. GLC analyses were carriedout by using a 5-ft 3To Carbowax column. Separations of en-z1'rne-containing gels from suspension were accomplished by usinga table-top clinical centrifuge. The pH in reaction solutions wasmaintained either by manual addit ion of 2.5 N KOH or by au-tomatic addition with a LKB Model 12000 \Iarioperpex peristalticpump, connected to a Chemtrix pH control ler. Formate de-hydrogenase was obtained from Boehringer Mannheim. F'leish-

M. J. Am. Chem. Soc' . 1984. 1( /6.

(40) Glycerol is actually a substrate for HLADH but is a very Dcxlr one(l% the rate of oxidation of ethanol). Racemic 1,2-propanediol is asubstrate for both GDH and HLADH. There is evidence that HLADHpreferentially oxidizes the S enantiomer2o in contrast to GDH, which isselective for the R enantiomer.

(41) Higher specific activity (a8 U/mg) has been obtained experi-mentally. See: Shutte, H.; Flossdorf, J.; Sahm, H.; Kula, M.-R. Eur. J.Biochem. 1976, 62, 151-160.

(42) Other methods to obtain chiral S include asymmetric hydrogen-ation of 2,3-butanedione (3% ee) (see: Ohgo, Y.; Takeuchi, S-; Na[ori,Y.; Yoshimura, J. Bull. Chem. Soc. Jpn. lg8l, 54, 2124-275b), OolCuoxidation of (28,3R)-butanediol ((5% ee) (see: ref 30), and a ciude,nematode extract catalyzed condensation of acetaldehyde (see: Berl, S.;Bueding, E. J. Biol. Chem. 1951, 191,401-418). The last reaction pro-vided 3 only in small quantities and in dilute, aqueous solution. It is notknown if racemization of 2-hydroxycyclobutanone occurs under the re-action conditions.

B. ; Kl ibanov, A. M. J. Am. Chem. Soc. 198{. 1 l i f i

Optically Active 1,2-Diols and a-Hydroxy Ketones

mann's yeast (cake type) was obtained from the local supermarket.All other enzymes and biochemicals were obtained from Sigma.1-Hydroxy-2-butanone was purchased from BASF and was pu-rified by spinning-band distillation [58-59 'C (30 torr)]. Formicacid-drwas purchased from Stohler Isotope Chemicals. Organicchemicals were purchased from Aldrich. Water was doublydistilled, the second time through a Corning Model AG-1b glassstill. Welding grade argon was used as the inert atmosphere.

Enzymatic Assays. Assays were performed by the literatureprocedures.a6 Units of activity are pmol min-l. Units of GDHrefer to activity with glycerol as substrate, unless otherwise in-dicated. Glucose dehydrogenase (GlcDH) was assayed by aprocedure similar to glucose-6-phosphate dehydrogenase:s glu-cose-6-phosphate was replaced by glucose. Diaphorase (Clos-tridium hluyueri) was assayed in cuvettes containing 0.96 mL ofpH 9.0 glycine buffer, 0.01 mL of NADH solution (10 mg/ml,),and 0.01 mL of methylene blue (0.01 M). An initial rate ofdecrease in absorbance at 340 nm was measured. A solution ofdiaphorase (0.02 mL) was added and the final rate of decreasein absorbance was measured. The difference in initial and finalrates was used to determine the activity of enzyme.

Several assays for glycerol dehydrogenase were used. The assaysfor the relative rates in Table I were performed in 1.5-mL cuvettescontaining 0.86 mL of pH 7.6 Hepes buffer, 0.02 mL of NADHsolution (5 mg/ml-), and 0.10 mL of substrate solution (1 M). Thereaction was initiated by the addition of 0.02 mL of GDH solution(1 mg/ml,), and the absorption at 340 nm was measured. Therelative rates found in Table II were obtained in 1.5-mL cuvettescontaining 0.83 mL of pH 9.0 glycine buffer, 0.05 mL of NADsolution (20 mglmL), and 0.10 mL of substrate solution (1 M).The reaction was initiated by the addition of 0.02 mL of GDH(1 mg/ml,) and the absorbance measured at 340 nm.

Quantitative Assays with GDH. The success of enantiomericresolutions catalyzed by GDH is dependent on the difference inthe rate of reduction (or oxidation) of the two enantiomers. Theassay used here depends on stoichiometric reaction of the substratewith NAD(H). The reaction, initiated by addition of GDH, wasperformed in a cuvette, and the change in absorbance at 340 nmwas measured. After several minutes (10-30) the absorbance wasconstant; the difference between the initial and final absorbancewas measured, and the quantity of substrate consumed wascalculated. This method could be used both for oxidations withNAD and reductions with NADH. The oxidative assays weremeasured in 3-mL cuvettes containing 2.9 mL of pH 9.0 glycinebuffer, 0.05 mL of a solution containing NAD (30 mg/ml,), and0.05 mL of substrate (0.01 M). The reaction was initiated by theaddition of GDH (1 mg); after several minutes (5-20) the finalabsorbance was measured. The difference in the initial and finalabsorbances was used to determine the quantity of substrateconsumed. The quantitative, reductive assays were similar, exceptpH 7.6 Hepes buffer and NADH (0.2 mM) were used. GDHoxidized half of a racemic solution of l,2-propanediol and of1,2-butanediol; solutions of the B enantiomers were completelyoxidized. GDH reduced half of a racemic solution of 3-hydroxy-2-butanone and 2-hydroxycyclopentanone. Greater than85To of a racemic solution of 2-hydroxycyclobutanone was reduced.

Enzyme Immobilization. Enzymes were immobilized withPAN-900 following a general procedure;4? yields were in the rangeof 30-50%. Assays of immobilized enzymes were similar to thosefor soluble enzymes: To a 1-mL cuvette containing the assaysolution was added 20 pL of a buffer containing the gel in sus-pension. The cuvette was capped and inverted several times tomix the contents and the absorbance read at 340 nm. The solutionwas mixed every 30 s over a period of 5 min and the rate of changein absorbance monitored on a chart recorder.

(.R)- and (S)-1,2-Propanediol. The R enantiomer was pre-pared from the zinc salt of o-lactate by LAH reduction of themethyl ester. o-Lactic acid was obtained by the reduction ofpyruvate catalyzed by >lactic dehydrogenase.z The S enantiomer

(46) Bergmeyer, H. U. In *Methods of Enzymatic Analysis'; VerlagChemie and Academic Press: Weinheim and New York, 1974. Forspecific assays, see pp 459 (G-6-PDH), 653 (GID), and 1555 (FDH) of thisreference.

(47) Pol lak, A. ; Blumenfeld, H. ;Wax, M.; Baughn, R. L. ;Whitesides,G. M. J. Am. Chem. Soc. 1980. 102.6324-6336.

J. Org. Chem., Vol. 51, No. 1, 1986 33

was prepared by LAH reduction of ethyl r,-lactate.a8cis-1,2-Diols. The cis-1,2-diols were prepared by oxidation

of the corresponding olefins with N-morpholine N-oxide andcatalytic quantities of osmium tetraoxide according to a generalprocedure.ae cis-lr,2-Cyclopentanediol,s cis-1,2-cyclohexanediol,sand cis-1,2-cyclooctanediol5l were prepared in 1:2 water-acetone.cis-L,2-Cycloheptanediol52 and cis-exo-norbornanediols3 wereprepared in 10:3: 1 terl-butyl alcohol-tetrahydrofuran-water. Allthe diols, except cis -1,2- cy clopentanediol, were crystalline solidsat room temperature and melted within I-2 "C of the reportedtemperature.

a-Hydroxy ketones were prepared by acyloin condensationof the corresponding (di)esters, using chlorotrimethylsilane as atrapping agent, according to a general procedure.sa The inter-mediate 1,2-bis[(trimethylsilyl)oxy]alkenes were hydrolyzed byone of two methods: adding to dry methanolM or refluxing in amixture of 1 N HCI and ether.5s Compounds prepared by theformer method were 2-hydroxycyclobutanone (8) and 4-hydroxy-3-hexanone; compounds prepared by the latter methodwere 2-hydroxycyclopentanone (7) and 4,4-dimethyl-2-hydroxy-cyclopentanone.

Enzymatic Preparation of (R)-l,2-Propanediol [(,R)-2] bvReduction of l-Hydroxy-2-propanone (1). A 1-L, four-necked,round-bottomed flask was equipped with a pH electrode, argoninlet, outlet to a bubbler, and a magnetic stirring bar. Water (100mL), ammonium formate (6.3 g, 100 mmol), l -hydroxy-2-propanone (3.7 g,50 mmol), and 2-mercaptoethanol (0.02 g,0.25mmol) were added. The pH was adjusted to 7.5 with 2 N KOH.GDH (MU, Enterobacter aerogenes), FDH (9 U), and NAD (0.07mmol) were added. The solution was maintained under an argonatmosphere in a pH range of 7.5-8.3. The progress of the reactionwas monitored by GLC; after 9 days less than 27o of the startingmaterial remained. The solution was decanted and centrifugedto separate the immobilized enzymes. The gel was washed withtwo 50-mL portions of pH 7.5 Hepes buffer, and the decantedbuffer washes were combined. The recovered activities of theenzymes were 61Vo for GDH and 62% for FDH. The reactionsolution and the washings were combined and extracted contin-uously with ether for 6 days. The ethereal solution was con-centrated and dist i l led [bp 85-86 'C (10 torr)] through ashort-path dist i l lat ion head to yield a clear oi l [(E)-2; 1.9 g,25mmol, 50%1. The IH NMR spectrum of this material was in-distinguishable from that of authentic material.

Enzymatic Preparation of (R)-1,2-Butanediol [(,R)-a] bVReduction of 1-Hydroxy-2-butanone (3).s A 1-L, four-necked,round-bottomed flask was equipped with a pH electrode, argoninlet, outlet to a bubbler, and a magnetic stirring bar. Water (150mL), ammonium formate (12.6 g,200 mmol), 1-hydroxy-2-buta-none (8.8 g, 100 mmol), and 2-mercaptoethanol (0.04 g,0.5 mmol)were added, and argon was bubbled through the solution for 1h to remove dioxygen. The pH was adjustedtnT.4 with 2 N KOH.Immobilized GDH (88 U, Enterobacter aerogenes) and FDH (11U) and NAD (0.13 mmol) were added. The reaction mixture wasstirred under an argon atmosphere and maintained in a pH rangeof 7.0-7.8 by periodic addition of 2.5 N KOH. After 7 days,additional NAD (0.13 mmol) was added. The progress of thereaction was monitored by GLC; after 19 days no starting materialremained. The mixture was decanted, and the immobilized en-zyrnes were separated by centrifugation. The gel was washed withtwo 30-mL portions of pH 7.6 Hepes buffer. The recoveredactivities of the enzymes were 50% for GDH and 88% for FDH.

(48) Schmidt , U. ;Gombos, J. ; Hasl inger, E. ; Zak,H. Chem. Ber. lg76,109,2628-2644.

(49) Van Rheenen, V.; Cha, D. Y.; Hartley, W. M. Org. Synfh. 1978,58,43-52.

(50) Owen, L. N.; Smith, P. N. J. Chem. Soc. 1952,4026-4035.(51) Cope, A. C.;Fenton, S. W.;Spencer, C. F. J. Am. Chem. Soc. 1952,

74, 5884-5888.(52) Owen, L. N.; Saharia, G. S. J. Chem. Soc. 1953, 2582-2588.(53) Sauers, R. R.; Odorisio, P. A. J. Org. Chem.1979, 44,2980-2983.(54) Bloomfie ld, J. J . ; Nelke, J. M. Org. Synth.1977,57, l -7.(55) Ruhlmann, K. S1'nfhe.sls 1971,236-253.(56) Another method to produce chiral 1,2-butanediol is based on

lipase-catalyzed transesterification of racemic 1,2-butanediol; both en-antiomers are obtained in high enantiomeric excess by using this proce-dure.{s

34 J. Org. Chem., Vol. 51, No. 1, 1986

The aqueous solution was continuously extracted with ether for5 days. The ethereal solution was concentrated and distilledthrough a short-path distillation head [bp 80-81 'C (5 torr)] toyield a colorless oi l [(r?)-a;6.3 g, 70 mmol, 70%1. The lH NMRof this material was indistinguishable from that of an authenticsample.

Preparation of (R)-3-Hydroxy-2-butanone [(n)-5] bVReduction of Racemic 3-Hydroxy-2-butanone (5). A 500-mL,round-bottomed flask was charged with 3-hydroxy-2-butanone(8.0 g, 91 mmol), glucose-6-phosphate (12.8 g,45 mmol),25 andwater (80 mL). The pH was adjusted to 7.5 with KOH. NAD(0.1 9,0.13 mmol) and immobil ized GDH (90 U, Cellulomonassp.) and G-6-PDH (85 U, Leuconostoc mesenteroides) were added.A pH electrode connected to a pH controller was added; the pHwas maintained at 7 .4-7.5 by automatic addition of a 2.5 N KOHsolution, which was contained in a graduated cylinder. After 3days of stirring under an argon atmosphere, no further changein pH was observed; 19 mL of KOH solution (0.52 equiv, 47 mmol)had been added. The immobilized enzymes were removed bycentrifugation: The recovered yields were 130% for GDH andrc}% for G-6-PDH. The aqueous solution was saturated withsodium chloride and continuously extracted with ethyl acetatefor 2 days. The organic solution was concentrated (8.7 g) anddistilled aL 25 torc through a 30-cm Helmholtz column. Twofractions were collected: at 54-55 and 95-100 oC. The firstfraction contained (n)-5 (1.6 g, 18 mmol,20Va;fclp-40.76o (neat),

[ ,r ] 'no -40.8o; d = 1.0 SlmL). The second fract ion contained2.3-butanedio l (6 ; 2 .7 9 ,30 mmol , 34%).

Enzymatic Preparation of (8)-3-Hydroxy-2-butanonet(n)-5] Using Glucose Dehydrogenase to Regenerate NADH.Racemic 3-hydroxy-2-butanone (5) (8.0 g, 91 mmol) was alsoresolved by using GDH; glucose (9.0 g, 50 mmol) and glucosedehydrogenase (GlcDH) were used to regenerate NADH. Im-mobilized GDH (60 U, Cellulomonos sp.) and GIcDH (63 U) wererecovered in 133% and 51% yield, respectively. The aqueoussolution was saturated with sodium chloride and extracted con-tinuously with ethyl acetate for 7 days. The organic solution wasconcentrated (8.0 g) and distilled through a 30-cm Helmholtzcolumn at 26 ton. Two fractions were collected: at 55-56 and96-98 oC. The first fraction contained (rB)-5 (1.7 g, 2lTa; [ttly.-50.74' (neat), [o]mo -50.8";d = 1.0 glmL).57 The second fractionconta ined 6 (3 .2 g ,39%).

Enzymatic Preparation of (n )-2-Hydroxycyclopentanonet(,R)-7] by Reduction of Racemic 2-Hydroxycyclopentanone(7). A 500-mL, round-bottomed f lask was charged with 2-hydroxycyclopentanone (7; 6.0 g, 60 mmol), glucose-6-phosphate(9.6 g, 33 mmol), and water (150 mL). The pH was adjusted to6.558 with KOH. NAD (0.1 g, 0.13 mmol) and immobilized GDFI(60 U, CelluLomonos sp.) and G-6-PDH (ZOO IJ, Leuconostocmesenterolde.s) were added. The pH was maintained at 6.3-6.5bv automatic addition of a 2.5 N KOH solution. After 26 h ol'stirring under an argon atmosphere, no further change in pH wasobserved; 6.0 mL of KOH solution (0.25 equiv, 15 mmol) had beenadded. The color of the mixture was brown; the color of theenzyrne-containing gel was black. The immobilized enzyrne-q wereremoved by centrifugation: The recovered yields were 90% forGDH and 38To for G-6-PDH. The aqueous solution was saturatedwith sodium chloride and continuously extracted with ethyl acefateior 36 h. The organic solution was concentrated (4.9 g) anddistilled at 2 torr through a iO-cm Helmholtz column. Twofractions were collected: at 50-53 and 7O-?2 "C. The first fractioncontained (f i)-7 (1.0 g, 10 mmo|76%). The second fract ion

(57) The physical properties of chiral 5 were quite dlfferent from thoseof racemic 5; racemic 5 solidifies as the dimer at room temperature,whereas chiral 5 remains a mobile, colorless oil even at 0 oC. The absoluterotation of (.R)-3-hydroxy-2-butanone [(R]-51 is [a]20p -58o. This valueis extrapolated from the rotation of material found to be of 88% ee basedon 1H NMR data of the (- ) -MTPA ester : [ .1]20D (-50o)/ [ (0.88)(1.0 g imL)(1 dm)] = -58". Previous workersa2 found a higher absolute rotationof (,R)-5; however, the authors themselves questioned the validitv of therotation since the concentration was determined by colorimetric methodsli.e., the sample was not weighed.

(58) The a-hydroxy ketone was unusually unstable in aqueous solu-tion, particularly at basic pH; 7 dissolved in buffered soiutions at pH 7-9graduallS'darkened to a black color. Decomposition was minimized bvmaintaining the pH aL 6-l.

Lee and Whi tes ides

conta ined 1,2-cvc lopentanedio l (1 .1 g , l l mmol , 18%1.The reaction was repeated at pH 5.5 with the same quanti t ies

of enzymes and start ing materials. After i3 days, 6. l l ml, of ' 2.5N KOH had been added. The products of the reaction wereisolated bv dist i l lat ion as befbre: (R)-2-Hydroxvcyclopentanone(1.3 g , 13 mmol , 22%,) and 1,2-cyc lopentanedio l (1 . .1 g . 1-1 mmol ,23%\ were obta ined.

Enzymat ic Reduct ion o f Racemic 2-Hydroxycyc lo-butanone (8). A 500-mL, round-bottomed flask was charged with2-hydroxycyc lobutanone (8 ; 1 .5 g , 1? mmol) , g lucose (1 .7 g . 10mmol) , Hepes buf fer (pH 7.6 ,50 mL) , NAD (0.1 9 ,0 .111 mmol) .and immobil ized GDH (40 L.r, Cel lulomonos.sp.) and GIcDH (36U, Baci l lus cereus). The pH was maintained at 7 .2-i .4 by au-tomatic addit ion of 2.5 N KOH solut ion. After 16 h of st irr ingunder an argon atmmphere, no further change in pH wa-s observed;4.0 mL of KOH soiut ion (0.59 equiv, 10 mmol) had been added.The immobilized enzymes were removed by centrifugation. Theaqueous solut ion was saturated with sodium chloride and con-tinuously extracted with ether for 24 h. The organic solution wasconcentrated to an oi l (0.6 g); GLC showed the oi l contained a1:1 mixture of a-hydroxy ketone 8 to 1,2-cyclobutanedioi.

Preparat ion o f ( .R)-1 ,2-Butanedio l t (R)-41 and (E)-1 ,2-Propanediol [(.R )-2] bV Reduction of l-Hydroxy-2-butanone(3) and 1-Hydroxy-2-propanone (1) Us ing Ferment ingBakers'Yeast.se A 1-L, round-bottomed flask was charged withsucrose (60 g, 0.3 mol) and water (0.5 L). Yeast (60 g; was sus-pended in water (60 mL), and the slurry was added to the stirredsugar solution. After 2 h, 1-hydroxy-2-butanone (13; 2.i g,37 mmol)was added to the bubbling mixture. After 24 h, no furtherbubbling was observed. The mixture was stirred for iJ days andfiltered through Celite. The filtrate was concentrated and 20 mLol a 1:I solution of ether-ethanol was added. The resulting whiteprecipitate was removed bv centr i fugation and discarded. Thesupernatant rvas c'oncentrated (2.6 g) and dist i l led through ashort-path dist i l iat ion head fbp 64-65 "C (2.5 torr)] as a colorlesso i l [ (R ) -a ; 1 .1 g , 1 .3 mmo l . . 11%1.

A similar prcrcedure was rrsed tri reciuce 1-hydrox5r-2-propanone(1; 6 .4 g , 86 mmol) . The f ie ld o f ' (R)- i .2 -propanedio l [ (R)-2 ] was38% (2.5 g, 33 mmol).

Preparation of (S)-2-Hydroxycyclohexanone by Oxidationof 1,2-Cyclohexanediol. A 500-mi,. round-bottomed f lask wascharged wi th l ,2-cyc lohexanedio l (1 g , 8 .6 mmol) . NAD (0.1 g .0.13 mmol), methylene blue (0.035 g, 0.09 mmol), pH 9.0 glycinebuffer (80 mL), and immobil ized GDH (CelLulomono.s.sp., 1:15Li) and diaphorase (140 U). The reaction mixture was stirred whileopen to the atmosphere; the color was dark blue. The reactionprogress was followed by GLC analysis. After 3 days, the reactionappeared to be 30% complete and no further progress was ob-served. The immobil ized enz-vmes were separated by centr i fu-gation. The recovered activi t ies of '1he enzrrnes were 83% f 'orGDH and 58% for diaphorase. The aqueous solut ion was ex-tracted with two 200-rnl, port ions r i f methylene chloride. Thecombined organic extracls were concentrated, redissolved in ether(20 mL). f i i tered through c' i iaicoal and Celi te. and concenlratedto a colorless oi l (0.2 gi. which crystal l ized upon standing: [ , t ln- 0 . 0 7 ( r ' 0 . 5 3 . C H C l , t . I n l r " r ) - 1 3 . 3 o

Enzymat ic Preparat ion o f (S) -1 ,2-Butanedio l [ (S) -a ] b l 'Ox idat ion o f Racemic 1 ,2-Butanedio l [ (+) -4 ] . To a deoxr ' -genated solut ion ofsodium 2-oxoglutarai-e (5.0 g. 30 mmol), 1.2-butanediol (4,4.5 g, 50 mmol), and 2-mercaptoethanoi (0.00!t g,l) . i mmol) in water (1 L) were added irnmobil ized GDH (65 U,Enterc tbacter ) , GIDH (34 LI ) and NAD (0.1 g ,0 .13 mmol) . ThepH was adjusted with NH4OH to 8.5 and maintaineci in the pHrange 8.;J 8.5 b_"- the addit ion of 2.5 N KOH. After 5 days, ad-ditional NAD (0.1 g. 0.i3 mmol) was added. 'Ihe reac'rir)n progressrvas fol]<;wed b1 GLtl; after 11 days no 1i:rt i rer reacl ion wasobsen'eci. The gel was allowed to setlie and tire solution decantedthrough a stainie-os steel cannula under a positive pressure of argon.The recor,ered enzyme activi t ies were i lOc/o lor GDH anci 18%lbr (llDH. The aqueous sohrtion was exlract,ed continuousl.'u'ithether 1'or ll da.r's. The ethereal solu.tion was concentratecl (i1.8 g)and d is t , i l led through a shor t -path d is t i l la t ion head at 4 . ' ) io r r .Two fract ions were col lected: at 25 and 67-70 'C. T'he f i rst

Optically Active 1,2-Diols and a-Hydroxy Ketones

fraction contained 1-hydroxy-2-butanone (3; 0.71 g, 8 mmol, t6%\;the second fract ion contained (S)-1,2-butanediol [(S)-a; 1.95 g,22 mmoll.

Preparation of (B)-1,2-Butanediol-2-d, [(fi \-4-2 -d ]. Thepreparation of (fr ) - 1, 2-butanediol- 2- il I@) - 4- 2-d tl was similar tothe enzymatic preparation of (R)-1,2-butanediol I(R)-41 exceptthat ammonium formate was replaced with formic acid-d2 (neu-tralized with NH.OH). The substrate, 1-hydroxy-2-butanone (3;4.4 g,50 mmol) was reduced to (R)-1,2-butanediol-2-dr I(R)-4-2-d;bp 62-63 "C (2 .4 mmHg); L9 g,2 l mmol , 42Tol by us ing GDH(20 U, Enterobacter, l00To recovery at the conclusion of thereaction), FDH (5 lJ, 83Ya recovery), and NAD (0.16 mmol) overa period of 22 days. The deuterium incorporation was determinedby GC-MS and found to be 9880, based on mle 60(CHgCH2CH(D) = OH+) tH NMR (CDCI3) 6 3.68 (d, 1 H, J =1 0 . 8 H 2 ) , 3 . 4 6 ( d , 1 H , J = 1 0 . 8 H z ) , 2 . 0 5 ( b r s , 1 H ) , 1 . 9 0 ( b r s ,1 H ) , 1 . 4 9 ( m , 1 H , J = 7 . 4 H z ) , 0 . 9 8 ( t , 3 H , J = 7 . 4 H 2 \ .

Preparation of 1-O-Tosyl-1,2-propanediol and l-O-To-syl-1,2-butanediol. The tosylation of 1,2-propanediol and 1,2-butanediol was performed according to the literature procedure.sThe reaction product was contaminated with small amounts ofditosylate and with the secondary tosylate. The desired l-O-tosylderivatives were isolated by preparative TLC or medium-pressurecolumn chromatography; for both, silica gel was used as thestationary phase and methylene chloride or 3:2 hexane-ethylacetate as the mobile phase. IH NMR spectral data of thesecompounds were in agreement with the literature values.8

Preparat ion o f (R)-2-Acetoxy- l -bromopropane and(,R)-2-Acetoxy-l-bromobutane.34 A 50-mL, round-bottomedflask equipped with a magnetic st irr ing bar was charged with(R)-1,2-butanediol [(R)-4: 1.1 g. 14 mmol: prepared enzl'rnatically]and cooled in ice. HBr in acetic acid (31 % , 43 mmol. 11 mL) wasadded over a period of 5 min. The solution was stirred at roomtemperature for 30 min. \\-ater (40 mL) was added and thesolution neutralized with NaHCOB (12 g) and extracted with three50-mL portions of ether. The combined organic extracts weredried (MgSOa) and concentrated (2.0 i l . The crude (R)-1-bromo-2-acetoxypropane was dist i l led through a short-pathdist i l lat ion head [bp 75-80'C (40 torr)] as a colorless oi l (1.6 g,8.9 mmol, 63%). lH NMR spectral data were in agreement withthe literature values.3a

(R)-1,2-Butanediol-2- dr [@)-a-Z-dr;1.7 g, 19 mmol] was con-verted to (R)-1-bromo-2-acetoxvbutane -2-d1 by a similar proce-dure. The product (2.5 g,13 rfmol) was obtained in 69% yieldlbp 75-80 'C (20 torr) l : lH NMR (CDCI3) 6 3.49 (d, 1 H, J =5 .3 Hz ) ,3 .46 (d , 1 H , J = 5 .3 Hz ) , 2 .10 ( s , 3 H ) , 1 .76 (m , 2 H ) ,0 .93 ( t , 3 H , J = 7 .6H2) .

(R)-1,2-Epoxypropane and (R)-1,2-Epoxybutane.s A b0-mL, round-bottomed flask equipped with a magnetic stirring barwas charged with (R)-1-bromo-2-acetoxypropane (1.6 g,8.9 mmol)and 1-pentanol (2 mL). A solut ion of potassium pentaoxide in1-pentanol (prepared by reaction of 0.35 g of potassium metalwith 10 mL of pentanol) was added to the stirred solution viastainless steel cannula under a positive pressure of argon over aperiod of 5 min. A white precipitate (KBr) formed immediately.The mixture, containing (ft)-1,2-epoxypropane, was distilledthrough a 15-cm Vigreux column (32-35 oC) as a colorless oil (0.2bg,4.3 mmol, 48%). lH NMR spectral data were in agreementwith the literature values.3a

(B)-1-Bromo-2-acetoxybutane-2-di was converted to (R)-1,2-epoxybutane-2-d1by a similar procedure. The product [bp 65-70oC; 0.41 g, 5.6 mmol) was obtained in 44% yield: 1H NMR (CDCI3)6 2 . 7 4 ( d , 1 H , J = 4 . 6 H 2 ) , 2 . 4 8 ( d , 1 H , J = 5 . 4 H z ) , 1 . 5 8 ( q , 2 ,J = 7 .6 Hz ) , 1 .01 ( t , 3 H , J = 7 .6H2) .

Preparat ion o f 1- (Pheny l th io) -2-propanol and (R)- t -(Phenylthio)-2-butanol-2 -d ,. A 5-mL test tube was chargedwith (R)-1,2-epoxypropane (0.08 g,1.4 mmol), thiophenol (0.29g, 2.6 mmol), and methanol (0.5 mL). After 12 h, the productwas isolated by chromatography on silica Sel (5 g): The columnwas eluted initially with hexane to remove thiophenol, followedby methylene chloride. 1-(Phenvlthio)-2-propanol was obtainedas a colorless oi l (0.05 g, 0.3 mmol, 2l%): tH NMR (CDCI3) d7.4-7.2 (m, 5 H) , 3 .82 (m, 1 H) , 3 .12 (dd, 1 H, J = 13.8, 3 .8 Hz) ,2 .84 (dd, L H, J = 14.0, 8 .6 Hz) , 2 .4L (s , L H) , 1 .27 (d , B H, J =5.9 Hz). A similar procedure was used to obtain (R)-1-(phenyl-th io) -2-butanol -2-dr : lH NMR (CDCI3) 67.4-7.2 (m,5 H) ,3 .16

2

( m M )

2 0- - l a '

v-, ' ' ) ,.2

'o-.1 .a'(mM-'min)

q

xo-lloH GoH,

ror;ojot

Ki = Q.3 mM

t , 19g6 35

o.2 M

t!

( b )o

)\..IOH

60H._+OHI

I

OH

o.2 o.4 O.6 qrg

LrJ (M)

911 HLAOHo H -

' A ' \t t o\-ry

K, o 80 mMO

-80 -zlo o 40 80 120 t60

[r l (mM)Figure 6 . f ) ixon p lo ts used in determinat ions o f K; . (a) GDH-ca ta l vzed ox ida t i on o f g l vce ro l (Ce l l u l omn nos . sp . ) . ( b ) GDH-catal l 'zed reduction of l l -hvdroxr'-2-butanone (C el luIomrtnas sp.).(c) HLAD- cat alyzed oxidations of c'is- 1,2-cyclohexanedimethanol.

( d , 1 H , J = 1 3 . 5 H z ) , 2 . 8 4 ( d , 1 H , J = 1 1 1 . 5 H z ) , 2 . 3 9 ( s . 1 H ) .1 . 5 7 , ( m , 2 H ) , 0 . 9 7 ( t , 3 H , J = 7 . 4 H 2 ) .

Mitsunobu Reaction Using (.R )-3-Hvdroxy-2-butanonel (R)-51 and (+) -1-Hydroxy-2-butanone (5) .60 A 2b-ml , ,round-bottomed flask was charged with (R)-3-hydroxy-2-butanone

[(r?)-5; 0.05 g,0.57 mmol], tr iphenylphosphine (0.15 g,0.57 mmol),( - ) -MTPA ac id (0 .13 g,0 .57 mmol) , and THF (5 mL) . D ie thy lazodicarboxylate (DEAD, 0.57 mmol) in THF (5 mL) was addedvia sy'ringe. The yellow color of the DEAD solution disappearedimmediately upon addition. After 4 h, the solution was concen-trated. Ether (20 mL) was added and the white precipitateremoved by filtration. The ethereal solution was concentratedto an oi l (0.4t g1. Purif icat ion by preparative TLC on si l ica gel(to remove residual tr iphenyl phosphine oxide) using 1:1 meth-ylene chloridrpetroleum ether provided 3-O-MTPA-2-butanoneas an oi l (90 mg, 0.27 mmol,547a). Racemic 3 was converted tothe (-)-O-MTPA ester by a similar procedure.

Racemization of (R)-3-Hydroxy-2-butanone (5). (n)-g-Hydroxy-2-butanone (5) was found to be stable to 1 N HCI butto rapidly racemize in 1 N NaOH. The rate of racemization wasmeasured by monitoring the decrease in the polarimeter readingof a solution of 5 dissolved in I N NaOH. Chiral 3-hydroxy-2-butanone [(R)-5;0.11 g, 1.3 mmol, 80% ee] was dissolved in 1 mLof water. Aliquots of the solution were diluted in 1 mL of distilledwater, 1 mL of 1 N HCl, and 1 mL of 1 N NaOH. The opticalrotation of the diluted solutions were measured over time. Theneutral and acidic solutions showed constant rotations of -0.65"and -O.70o. After 30 h, the acidic solution had a rotation of -O.69o.The rotation of the basic solution decreased rapidly; a mea-surement was taken every 30 s. A first-order plot of the loss of

(60) Mi tsunobu, O.; Eguchi , M. Bul l . Chem. Soc. Jpn. tg7t . 44.3427-3430.

( m M j m i n )

t o

o.o6

oooS

op4V

( m M m i n - ' ) o . o .

o.o2

crs -cyclohexonediol

2 3 -bu toned iol

36 J. Org. Chem., Vol. 51, No. 1, 1986

GDH (Cellulomonos spJ

Lee and Whitesides

representative Eadie-Hofstee plot is shown in Figure 6; the re-mainder of the plots are given as supplementary material.

Determination of Values of Ki. Product inhibition constants(K1) for three reactions were determined: GDH-catalyzed oxi-dation of glycerol, GDH-catalyzed reduction of 3-hydroxy-2-bu-tanone, and horse liver alcohol dehydrogenase (HLADH) catalfzedoxidation of cis-L,2-cvclohexanedimethanol. The inhibition foreach reaction was assumed to be noncompetitive for these sub-strates, and the method of Dixon6l'62 was used to determine eachvalue of Ki. For the oxidation of glycerol at pH 9 the glycerolconcentration was maintained at 0.1 M and the dihydroxyacetoneconcentration varied between 0.2 and 2 mM. For the reductionof 3-hydroxy-2-butanone at pH 7.6, the substrate was maintainedat 10 mM and the 2.il-butanediol concentration varied between0.3 and 0.8 M. For HLADH catalyzed oxidation of cis-1,2-cyclohexanedimethanol at pH 9.0, the substrate concentrationwas maintained at 40 mM and the lactone63 concentration variedbetween 20 and 60 mM. Plots of I lV vs. [I] are given in Figure7 .

Acknowledgment. We thank Dr. R. DiCosimo forperforming the GC-MS analysis of 1,2-butanediol-2-dr,Professor C.-H. Wong for providing a sample of o-lactate,and Steve Wasserman for aid in plotting the curves inF'igures 2-5. This research was supported by the NationalIns t i tu tes o f Heal th (Grant GM 30367) .

Registry No. l, 1 16-09-6; (R\-2, 4254-14-2: 3, 5077 -67 -8; (,?)-4,J0318 -66 -1 ; (S ) -4 , i 3522 -77 -5 ; (+ ) -4 , 26171 -83 -5 (R \ -4 -2 -d t ,e91-{ 1 -00-6; (* )-5. 52277 -02-4; (ft )-5, 53584-56-8; (+)-7, 99440-98-9;(R )-7, 99.193-88-6r (*)-8, 99440-99-0; GDH, 9028-14-2; HLADH,90;11 -72-5; (S)-2-hydroxycyclohexanone, 53439-93-3; I ,2-cyclo-heraned iol. glJ I - I 7 -9 : ( R ) - 1 - bromo- 2-acetoxypropan e, 99457 - 42-8;t r? ) - 1 - l r r trrno- ) - a t 'et oxvbutane-2- d b 9944I-01-7; (E)- 1,2-epoxy-propane. I ;-1-{b- I ; - : : t r? ) - 1,2-epoxyb utane- 2- d 1, 99441-02-8; 1-r phenl ' l thio )- 2- prropanol. 9; l7-56-4; (f t ) - t - (phenylthio)-2-buta-no l -2-d1, 994.11-03-9; ; l r l - I ITPA-2-butanone, 99441-04-0; d i -hydroxyacetone, 96- 26--1 : t r? t - 2.; l-d ihvdroxl 'propanal, 453- 17-8;(S)-2 ,3-d ihydroxypropanal . '197-09-6: 1 -hvdroxy-2-butanone,5077 -67 -8; 2-oxopropanai, 76-98-8: 2-hvdroxvcyclobutanone,L7082-63-2; 2-hydroxycyckrpentanone, -1711-84-?; 2-hydroxy-cyclohexanone, 533-6G8; glycerol, 56-81 -5; 1,2-propanediol, 57-55-6;(R)- 1,2-propanediol, 4254-14-2; (S)-1,2-propanediol, 4254-15-3;1, 2- hexaned iol, 6920 -22- 5; 1,2-octanediol, 1 1 1 7- 86- 8; 1, 2-ethanediol,I07-21-l; 1,3-propanediol, 504-63-2; 1,4-butanediol, 110-63-4;2,3-butanediol, 513-85-9; 3-mercapto-1,2-propanediol, 96-27 -5;l l -chloro- 1,2-propanediol, 96-24-2; 3-methoxy- 1,2-propanediol,62il -'.19 - 2; 3- ( methylthio) - 1, 2- propan ed iol, 2255I-26- 4; (t-) - rr-gly'cerophosphate, 57 46-57 -6; (ot )-cr-glycerophosphate, 1509-81-5;(R)-1-amino-2-propanol , 2799-16-8: (S) -1-amino-2-propanol ,2799-17-9; cis-1,2-cyclopentanediol, 5057-98-7; cis-\,2-cyclo-hexanediol, 1 792-81 -0 ; t rans - 1,2-cy'clohexanediol, 1460-57 -7 ; c is -l, 2-cycloheptanediol, 42174-7 2-7 ; c ls- i. 2- c.r'ckxrctaned iol. 27 607 -3 l l -6 ; o-ga lactose. 59-23-4; c is -1 ,2-cvc loheranedimethanol ,15?53-50-1.

Supplementary Material Avai lable: 'H \! IR data t 'or\{'fPA esters and Eadie-Hofstee plots useci in the determinationcl K,, vaiues (8 pages). Ordering int ' t-rrmation is given on anylurrent masthead page.

(61) Product inhibition for the natural substrates of both GDH andHLADH are noncompetitive for the alcohol vs. the ketone (or aldehyde).rri

i 52 ) D ixon . M. B iochem. J . 1953 , 55 . 170-171 .,63) The product of the oxidat ion, (+)-( lR,65)-c l .s-S-oxabic-v*c lo-

1-1. lJ .0] nonan- i -one. was obtained f rom a large-scale HLADH-cataivzedoxidat ion ot c is- j . l -cvclohexanedimethanol .

c

^ \

4 6

V

tal

S t o t z

( m i n x l O 4 )

l 1 2

l .o

v(mM m in - l ) o .8

O o 6

o.4

or2

o.o

c/s- cyclopentonediol

o oeo4 o.og o..l 2o .o2 0.06 o oto

vt s ]

( m i n - t '

Figure 7. EadirHofstee plots used in the determination of valuesof K^ for GDH catalyzed oxidations (Cellulomono.s sp.).

rotation of (fi)-5 indicated that the half-life of the racemizationwas 6.3 min.

Preparation of (-)-MTPA Esters. The (-)-MTPA esterswere prepared according to a general procedure23 using 2 equivof (-)-MTPA-CI per equiv of alcohol. iH NMR spectral data aregiven as supplementary material.

FtI- Values for Substrates for GDH and HLADH. Thevalues of K- of substrates in GDH-catalyzed oxidations weredetermined in glycine buffer (pH 9.0) with (NHa)zSOr (10 mM)and NAD (3 mM). The concentration of cis-i,2-cyclohexanediolwas varied from 25 to 500 mM, of 2,3-butanediol and cls-c1'clo-pentanedioi from 3 to 40 mM, and of l ,2-propanediol and 1,2-butanediol from 0.3 to 0.4 mM. The K- values for GDH-catal-vzedreductions were determined in Hepes buffer (pH 7.6) with (N-I{4)2SO4 (10 mM) and NADH (0.2 mN'I). The concentrat ion of2-hydroxycyclopentanone was varieci f ' rom 2.5 fo ir mM, of 3-hydroxy-2-butanone from 0.4 to 20 mM, of dihydroxyacetone from0.05 to 0.8 mM, and of 1-hydroxv-2-propanone from 0.03 to 0.20mM. The value of K- of cis-cyclohexanedimethanol with HLADHwas determined in glvcine buff'er (pH 9) with NAD (il mN{). Theconcentrat ion of ' the substrate was varied from 6 -10 mM. A

' l

t \

V. "r..V \\ 1 . ,xv

Y.v