[Reprinted from the Journal of Orsanic Chemistrv. 10, 2516(1975i.1 Copyright 1975by the American Chemical Society and reprinted b!' permissionrf the copyright owner.

Large-Scale Synthesis of Diammonium Acetyl Phosphatel

George M. Whitesides,* Merrell Siegel, and Patricia Garrett

Department of ChemistrTt,Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

ReceiuedApriL 11,1975

A detailed procedure for the large-scale synthesis of diammonium acetyl phosphate (1) is presented. Ketene is used to acylate 1007ophosphoric acid in ethyl acetate at -10o, and the resulting mixture of mono- and polyacetyl phosphoric acids converted to I by treatment with anhydrous ammonia in ethyl acetate-methanol at -10o. The product is obtained as an easily filtered, crystalline solid in ca. 90% yield and ca. 90% purity.

One limitation to the use of enzymatic catalysis in large- ADP is produced from ATP and AMP by phosphoryl scale organic synthesis has been the expense of many of the transfer catalyzed by adenylate kinase. ADP is converted common cofactors. As part of an effort to devise techniques to ATP by reaction with acetyl phosphate (AcP) catalyzed that would make enzymatically catalyzed reactions requir- by acetate kinase. Acetyl phosphate, the ultimate phospho- ing adenosine triphosphate (ATP) useful in practical syn- rylating agent in this sequence, had been synthesized pre- thesis, we have developed the reaction sequenceoutlined in viously from phosphoric acid by acylation with acetyl chlo- eq 1 and 2 as a method for regenerating ATP from AMP ride,3 ketene,a isopropenyl acetate,5 and acetic anhy- and/or ADP.2 dride,6'? and isolated as the lithium or silver salts.8 All of these procedures contain difficult work-up and isolation se- + Arp AMp #- 2ADp (1) quences. None are suitable for the preparation of acetyl phosphate in large quantity. Here we report a synthesis of 2ADP + 2AcP 2ATP + 2Ac (2) :-kinase diammonium acetyl phosphate from phosphoric acid, ke- Synthesis of Diammonium Acetyl Phosphate I Ors Chem., Vol. 40, No. 17, 1975 2517 tene, and ammonia, which yields product in easily isolated this studv are summarized in Figure 1 for acylations of form. This synthesis provides the most practical method both 85 and 100o/ophosphoric acid. These data establish available for synthesizing large quantities of acetyl phos- that the maximum conversion (90-95ozn)of 100o/"phosphor- phate. ic acid to 1 occurs for molar ratios of ketene to phosphoric Results acid of approximately 1.7. The decreasein yield observed The reaction of ketene with phosphoric acid in an inert solvent yields mono-, di-, and (presumably) triacetylphos- phoric acids.The relative amount of monoacetylphosphoric acid produced depends on the ratio of reactants used and on the extent of hydration of the phosphoric acid: water that is present is converted to acetic acid and acetic anhy- dride. Dilution of the reaction mixture obtained from ke- tene and phosphoric acid with methanol, and treatment of the resulting solution with anhydrous ammonia at -10o, yields diammonium acetyl phosphate (l) as an easily fil- tered, crystalline solid.

CH.:C--C + H,pO: l+ H.Ol -i ) oo il CH COPO.H. + {CH CO).PO H + o oo llri' (cHrco) ,Po [+ cH ]co"H + cH coccH j

oo o o cHTc=o/ H3Pc.4 lllillrl CH.COP(O-NHr.).+ CH,CNH. [* CH CO-NH,.l (3) Figure l. \'ields of diammoniumacetyl phosphate obtained fol- lowingreact.ion ofketene with l00o/ophosphoric acid (o) and 859. 1 phosphoricacid (t). Reactionswere carried out in ethyl acetate Isolation of acetyl phosphate as its diammonium salt has solutionat -1{)o,and the reactionmixtures were allowed to equili- a number of advantages over other isolation procedures. brate for 2 hr at -l0o beforediluting with methanoland adding 'fhe (CHz:C:O/H:IPO+) First, I is sparingly soluble in methanolic solutions, and ammonia. quantity is the number of added.divided bv the total numberof moleso{' precipitates as a crystalline, easily filterable solid. Ammo- molesol'ketene phosphoricacid originallvpresent. In 85% phosphoricacid, 1 mol soluble in methanol and nium acetate and acetamide are of wateris presentfor eachmole o|phosphonc acid. Yields are can be separated on the basis of solubilities. Previous pro- basedon phosphoricacid. The numbers associatedwith each cedures have involved neutralization of the acetylphospho- pointrepresent the ptrritvof the I isolatedat that ptiint;themajor ric acid in aqueous solutions, and have required either the partof the impuritt isammonium phosphate in mostinstances. use of silver(l) salts to effect precipitation or the filtration of the phosphate "slimes" generated by neutralization with for ratios greater tlu . 2 reflects the fact that addition of lithium acetate, carbonate, or hydroxides, followed by pre- ammonia t.othese reactions yields thick, difficultly filtered cipitation with ethanol. In addition, removal of water from suspensions.The use of 85olnphosphoric acid gave lower the dilithium acetyl phosphate required a time-consuming yields, apparently for the same reason. The effects of sol- and not always successfullyophilization or related proce- vent. phosphoric acid water content, and temperature on dure. Second, ammonia is expected to attack the acetyl yields of I were examined briefly: 100o/ophosphoric acid, moiety of diacetyl phosphate more rapidly than that of mo- prepared bv dehydration of 85o/"phosphoric acid,rl gave noacetyl phosphate.eThis appears to underlie the unexpec- higher vields than dioxane diphosphate or 85o/ophosphoric tedly high yields ()90o/o)of I obtained by this procedure. acid; ethyl acetate was superior as a solvent to DMF, DME, Although we have not studied the reactions that occur dur- di-n-butyl ether, and n-butyl acetate;raising the tempera- ing introduction of ammonia into the initial reaction mix- ture above -10o during addition of ketene and equilibra- tures in any detail, it seems that the excessof ammonia tion resulted in lower yields. present at the conclusion of this stage must convert di- and The detailed course of the acylation of phosphoric acid triacetyl phosphates to l. Third, 1 is very soluble in water, with ketene has not been established.l2In particular, it is and ammonium ion is innocuous to most (although not all) not clear how rapidly intermolecular acyl or acetyl group enzymes.l0Thus I can be used directly in the regeneration transfer occurs. A qualitative observation made during this of ATP. Ammonia is inexpensive compared with lithium work does, however, suggestthat intermolecular equilibra- and silver salts. Finally I has adequate storageand solution tion between diacetyl phosphoric acid and phosphoric acid stability (vide infra). occurs under the reaction conditions. A sample of 100o/o The number of products that can be formed by reaction phosphoric acid was allowed to absorb 2 molar equiv of ke- of ketene with phosphoric acid containing some water is tene. The resulting sarnple was treated with 1 additional large. This complexity, combined with uncertainties con- equiv of phosphoric acid, allowed to equilibrate at -10o for cerning the details of this reaction and of the subsequent 2 hr, and then worked up by the usual procedure,yielding I reaction of the product mixture with ammonia, make it dif- in72v" yield based on the fofol phosphoric acid involved in ficult to define a priori the number of equivalents of ketene the mixture. Since the initial reaction mixture would have required to maximize conversion of phosphoric acid to di- yielded ca. 45o/nof | (90o/ol2)on this basis had equilibration ammonium acetyl phosphate. In this work, the ratio of not occurred, the higher yield-which is also that obtained added ketene to phosphoric acid originally present has sim- by direct reaction of 2 equiv of ketene with 2 equiv of phos- ply been varied, and the yield of I determined. Results of phoric acid-suggests intermolecular equilibration. 2518 J Org Chem., Vol. 40, No. 17, 1975 Whitesides, Siegel, and Garrett

o o NMR assays.A Radiometer Model PHM 62 pH meter was used to determine pH values. Microanalysis were obtained by Midwest (CHTCO)"PO'H+ HrPOr r- zCHTCOPOTH" (4) Microlab, Ltd., Indianapolis, Ind. Diammonium Acetyl Phosphate (l). A 2-1.three-necked flask Ammonium ion, although normally innocuous as a com- was fitted with a thermometer, a gas inlet tube, and an overhead ponent of an enzymatic reaction mixture, does occasionally stirrer. The stirrer shaft entered the flask through a fitting reduce enzymatic activity,l0 and might interfere with other equipped with a side arm which served as a gas outlet. Ethyl ace- aspects of a synthetic sequence catalyzed by enzymes. It is tate (750 ml) and 100% phosphoric acid (100 g, 1.02 mol) were possible to convert I to disodium acetyl phosphate by transferred into the flask. and the resulting solution was cooled to -10o using an ethylene glycol-acetone-Dry Ice bath. Ketene was treatment with an ion exchange resin in water, although bubbled through the stirred solution for 10 hr (1.98 mol), after the yield is only moderate by the procedure we employed. which 750 ml of methanol, precooled to -10o, was added. Anhy- It is also possible to use other amines (e.g.,aniline) to neu- drous ammonia, directly from the tank, was passed through alumi- tralize the initial reaction mixture. The salts resulting from num coils immersed in the cooling bath, then over the surface of these reactions are less crystalline and more soluble in the rapidly stirred solution, and finally out through a bubbler methanol, and this type of work-up offers no obvious ad- linked to the flask through the gas outlet on the stirrer. This addi- tion was continued for 1.5 hr at a rate such that bubbles passed vantages. through the outlet bubbler at a rate of approximately one per sec- Compound I does contain a potential nucleophile (am- ond. During this time, the internal temperature of the solution monia) in the presenceof a reactive carbonyl group, and it graduall-"-rose to -7o and then fell to -10o, signalling the end of was important to examine its stability. Solid I could be the reaction.A total of 65 g of ammonia (3.82mol) was used (as de- stored for extended periods at 4o without decomposition so termined by weighing the tank before and after reaction), although long as it was protected from atmospheric moisture: no de- not all was consumed by the reaction mixture. The fine solid which filled the flask was collected by suction filtration on a Buchner creasein the purity of I in a desiccatorwas observedover 2 funnel. It was washed with 200 ml of methanol and 200 ml of anhy- months at 40. Storage in a desiccatorfor 1 month at 25" re- drous ether and transferred to a 1000-ml erlenmeyer flask. Metha- sulted in a 30olodecrease in acetyl phosphate content. The nol (350 ml) was added. and the resulting suspensionwas magneti- solution stability of acetyl phosphate has been extensively cally stirred for 10 min at room temperature. The solid was filtered studied.e'I3'14In the region between pH 5.5 and 9.5, hydrol- as before and washed in succession with 150 ml of methanol and ysis of dilithium acetyl phosphate takes place by P-O bond 500 ml of anhydrous ether. It was dried by covering the funnel with a piece of neoprene rubber, through which protruded a drying cleavage,apparently by a processinvolving metaphosphate tube containing l)rierite, and drawing air through it. Final drying anion.15Direct reaction with free amines does occur. The to constant weight under vacuum gave 180.2g of solid. Enzymatic oo oo assar'(videinfra) shriwedthat the solid contained89% I by weight, ll ll--,) llrl corresponding to a 9lq' r'ield based on phosphoric acid. A NMR ('H,('()P: + + (] t assal-tvide inlra) indicated a composition ratio of 91% 1,4.4%' -o- CH,CO- P:O 'lhe I acetamide,and 4.,1%ammonium acetate. sample was stored at o- 4o in a desiccatrlr. Anal. Calcd for CzHrrNzOsP:C, 111.80;H.6.37; N, 16.09.Found: rate of addition of ammonia in equilibrium with ammo- C, 12.37;H,6.56; N, 16.21. nium ion would not, however, be expected to be competi- Solvent was evaporated from the filtrate and the residue was tive with the rate of reaction 5 at pH 6-8.e To check this dissolved in acetone. Filtration o1'this solution gave 266 mg of prediction, the stability of 1 in buffered solutions at 39o solid. Concentration of the filtrate and addition of ether gave 34.2 g (0.58 mol) of acetamide,mp 80-81.5 (lit. mp 81). Removal of sol- was determined by observing its disappearance with time vent from the filtrate left 11.0g of yellow oil of undetermined com- by means of the enzymatic assay. The hydrolysis of l, fol- plife was phate. ammonirrnracetate. and acetamidecollectivelv account for found to be 3 hr. The rate constants obtained in this work 80'v.of' t he keteneirnd I l'l' of the ammonia used. are in excellent agreement with those reported by Kosh- Enzl'matic Assal' for l. The enzy'maticassa,v used to determine the and puritv ot'l is basedon three coupledenzymaticsteps: land for hydrolysis of dilithium acetyl phosphate.r:r -vield reaction of adenosine diphosphate (ADP) and acetyl phosphate Experimental Section yielding adenosine triphosphate (ATP) catalvzed bv acetate ki- nase; conversion of glucose to glucose 6-phosphate using this ATP General. All chemicals were reagent grade and were not further catalyzed by hexokinase;and reduction of nicotinamide adenine purified. Enzymes used in the assay of l-acetate kinase (EC dinucleotide phosphate (NADP+) to NADPH by this glucose 6- 2.7.2.1) and a commercial mixture of glucose6-phosphate dehydro- phosphate catalyzedbv glucose6-phosphate dehvdrogenase. genase(EC 1.1.1.49)/hexokinase(EC 2.7.1.|)-were obtained from Sigma Chemical Co. Authentic acetyl phosphate (Li, K salt), ADP ,l! ct lt. l.iil.ts(' 1 + AI)P -'---} ATP cH co - (6) (Na salt), and NADP* (Na salt) were also obtained from Sigma. Anhydrous ammonia was obtained from Matheson, and was used ir. r.rL.Lntsc + directly from the tank without purification. Phosphoric acid ATP D-ft-lucose (100%) was made by the slow addition of 191.5 g of phosphorus ADP + o-glucose 6-lthoslthute (7) pentoxide to 500 g ofstirred 85% phosphoric acid at -10o (ice-ace- tone bath).Ir The final solution spontaneously crystallized after o-Glucose 6-phosphltte + NADP. standing at room temperature for 48 hr. Ketene was produced by thermal cracking of acetone in a conventional apparatus similar to that described by Williams and Hurd.16 The rate of generation, de- 6-phospho-o-gluconic acid + NADPH (B) terminedlT by bubbling the output of the ketene generator through 50 ml of a cooled (0o), stirred solution of ethanolamine (3.184N) The conditions used in the assay are such that the equilibrium in 2-propanol for 30 min and then titrating excess amine with constants for each reaction lie far to the right. Under these condi- standardized 1.0 N hydrochloric acid (Methyl Red as indicator), tions, the number of equivalents of NADPH produced (measured was found to be 0.198 + 0.014 mol/hr under conditions used repro- spectrophotometrically at 340 nm) is equal to the number of equiv- ducibly throughout this work. Water used in enzymatic assayswas alents of I added originally. This general assay scheme has been distilled twice, the second time using a Corning Model AG-1b dis- used previously.ls The following standard solutions were prepared. tillation apparatus. Small-volume aliquots for these assays were Solution 1. To 300 mg of D-glucose and a mixture of 500 units of obtained using a Clay-Adams suction apparatus (obtained from hexokinase and 250 units of D-glucose 6-phosphate dehydrogenase Bectin Dickerson) and calibrated disposable micropipettes. Ultra- was added enough triethanolamine buffer (0.2 M, pH 7.6) contain- violet absorbance was measured using a Gilford Model 220 spec- ing magnesium chloride (0.03 M) to give 200 ml of solution. Solu- trophotometer. A Varian Model T-60 spectrometer was used for tion 2. Water was added to 250 mg of ADP (Na salt) to give 1.0 ml Synthesis of Diammonium Acetyl Phosphate J. Org. Chem., Vol. 40, No. 17, 1975 2519 of solution. Solution 3. Water was added to 30 mg of NADP+ (Na The decrease in concentration of I at 39o was followed until less salt) to yield 1.0 ml of solution. Just prior to the assay, approxi- than 25o/"remained. At least seven points were taken for each solu- mately 70 mg of I was brought to 10.0 ml with water (solution 4). tion. The disappearance of I followed first-order kinetics, and led To 5.0 ml of solution 1 was added 0.05 ml of solution 2, a 0.01-ml to these rate constants 4o6"6 (103 sec-t) (pH):4.33 (5.83); 3.78 aliquot (ca. 8.5 units) of a suspension of acetate kinase in 3.2 M (6.90);3.78 (8.00); 4.35 (9.30). ammonium sulfate solution (supplied at 850 units/ml), and 0.05 ml Disodium Acetyl Phosphate. A 2 x 30 cm chromatography col- of solution 3. The solution was allowed to incubate at 25o until its umn (Pharmacia) was filled with 40 ml of the washed Bio-Rad AG absorbance (measured at 340 nm using a l-cm cell) had reached a MP-50 ion exchange resin (H+ form, 100-200 mesh, i.86 mequiv/ plateau (ca. 3 min), and this absorbance (Ar) recorded. Another ml) resin, and 100 ml of sodium hydroxide (1.0 N) was slowly (1 solution was prepared as above, but to this solution was added 0.01 ml/min) passed through it. This neutralization was followed by ml of solution 4. The absorbance of this second solution (Az) was washing with 250 ml of water. The column was equilibrated for 2 then determined at 340 nm after a similar incubation. The absorb- hr at 4o. Doubly distilled water was added to 1.0 g of I to make 4.0 ance A I corrects for the small amount of ATP present as a contam- ml of solution, and this solution was placed on the ion exchange inant in the ADP, any NADPH contaminant in NADP+, as well as column. Doubly distilled water was passed through the column at 1 other species which may have absorbance at 340 nm. For absorb- ml/min, and 2.5-ml fractions were collected. These fractions were ances obtained using 1-cm cells, the following equation relates the tested for the presence of acetyl phosphate by the hydroxylamine- difference, (Az - ,A1), to the numbers of moles of diammonium ferric chloride test.le Fractions 1-6, pH 4.5. were devoid of acetyl acetyl phosphate present in solution 4. phosphate; fractions 7-14, pH 5.5. contained acetyl phosphate. These fractions, when combined, were found to contain 85% of the total volume of assay solution moles of 1 : initial acetyl phosphate by enzymatic assay. (Nenpn 103(.1"- A,) : B.2ox 1o-1(.1.-,lr) Registry No.-1, 55660-58-7;ethyl acetate, 141-76-8;phosphor- ic acid, 7664-38-2:ketene, 463-51-4; ammonia, 7664-41-7;dianil- NMR Assay for l. Approximately 130 mg of the reaction prod- inium acetyl phosphate, 55660-59-8; aniline, 62-53-3; disodium uct was dissolved in 0.4 ml of DzO, and to this solution was added acetyl phosphate,55660-60- 1. 0.01 ml of dioxane. The solution was stirred for 10 sec on a vortex- mixer, then transferred to an NMR tube, and the spectrum was re- References and Notes corded. The acetyl protons of I fall 1.63 ppm upfield from the di- (1) Supportedby the NationalScience Foundation (RANN), Grant Gl 34284 oxane protons, and are split into a doublet by coupling to phospho- (2) C. R. Gardner,C. K. Colton,R. S. Langer,B. K. Hamilton,M. C. Archer, - present rus (J 1.2 Hz). Acetyl protons from acetamide in the and G. M Whitesides,"Enzyme Engineering", Vol 2, E. K. Pyeand L. B. sample are found 1.75ppm upfield from dioxane, while those from Wingard,Jr., Ed.,Plenum Press, New York, N.Y., 1974,p 209; G. M. ammonium acetate are found 1.83ppm upfield from dioxane. Inte- Whitesides,A. Chmurny,P. Garrett,A. Lamotte,and C. K. Colton,ibid., gration of the dioxane peak and of the peaks due to I acetamide p 217. (3) F. Lynen,Chem.9er.,73,367 (1940);(b) F. Lipmannand C. Tuttle,J. and ammonium acetate allows calculations of the percentages of Biol.Chem..153. 571 (1944). these compounds found in the reaction mixtures. (4) R. Bentley,J. Am. Chem.Soc.,7O,2183 (1948). Dianilinium Acetyl Phosphate. To 150 ml of ethyl acetate was (5) F. Lipmannand E. R. Stadtman,J. Biol.Chem., 185, 549 (1950);D. E. added 20 g (0.204 mol) of 100o/ophosphoric acid, the solution was Koshland.Jr . J Am Chem.5oc.,73,4103 (1951). (1955); co<-rledto -10o, and ketene was bubbled through the stirred solu- (6) A. W D Avison,J. Chem 9oc.,732 R. W. Porter,M. O. Mo- debe,and G. R Stark.J. Biol Chem.,244,1846 (1969). for 2 hr. A 75-ml portion of this solution was withdrawn, and tion (7) E. Heyde, A Nagabhushanianand J. F Morrison,Biochemistry, 12, to the remainder was added 125 mi of precooled methanol, fol- 4718(1973). krwed by the dropwise addition of 30 g of aniline (2.S-fold excess) (8) E. Cherbuliezin "OrganicPhosphorus Compounds", Vol. 6, G. M. Kos- over a period of 10 min. During this time, the internal temperature alopoffand L Maier,Ed., Wiley-lnterscience, New York, N.Y., 1973, p o1'the solution remained at -10o. The resulting crvstallinesolid 211t| (9) DiSabatoand W P Jencks,J. Am.Chem Soc.,83,4393(1961). by' suction filtration on a sintered giass funnel, G. was collected (l0) Ammoniumron does act as an inhibitor(both competitive and noncom- washed with 75 ml of acetone and 150 ml of ether, and dried by petitive)toward certain enzymes. For examples,see J. P. Hoareand K. covering the funnel with a piece o{'neoprenerubber through rvhich J. Laidler,J. Am. Chem.9oc.,72,2487 (1950); F W. Sayreand E. Rob- protruded a drying tube containing Drierite, and drawing air erts,J. Biol.Chem.,233,1128 (1958); P. Maebaand B. D. Sanwal,Erb- (1966).Also R. P. Williams, Rev., Chem. through the product. Enzymatic assay showed it to be 89.9o/"diani- chemistry,5, 525 see J. Q. (1970);A S. Mildmanin "The Enzymes",Vol. ll, P. D. phosphate. An NMR spectrum (DMSO-d6) with di- 9oc.,24,331 linium acetyl AcademicPress. New York,N.Y., 1970, Chapter 9. (aromatic Boyer,Ed.. oxane as internal standard showed a broad multiplet (11) R N Beli,Anat Chem. 19, 97 (1947|,L F Adriethand O. F. Hill,J. prot