2. Synthesis of derivatives of levulinic acid as potential analogues of postulated intermediates of the biosynthesis of porphobilinogen


In our first attempts to study the inhibition capacity of a series of substrate analogues we had reasoned that one of the enantiotopic protons at the methylene group a to the keto function of 5-aminolevulinic acid has to be deprotonated. So we replaced one of the protons by a hydroxy group. This hydroxy substituent should be able to form a hydrogen bond to the base on the active site of the enzyme. Implicitly we assumed that enough space for an additional substituent should be available in the active site. This assumption was based on the fact, that during the enzyme catalysed process, both protons have to be removed and that at one point during the enzyme catalysed reaction a C-C-bond has to be formed to the keto function of the P-site substrate. The 3-hydroxy levulinic acid could in principle be modified on the hydroxy function to create a series of different analogues of the substrate or of potential intermediates. Unfortunately the 3-hydroxy levulinate is chemically too unstable to allow easy modification. The major problem seems to be the retro-Michael reaction (Scheme 10).[86]

Scheme 10: Chemical transformations of the 3-hydroxy levulinate

Obviously the use of the 3-chloro or the 3-bromolevulinate is hampered by the same problem. Preliminary studies clearly indicated that these two compounds are rapidly degraded in aqueous neutral buffer solution as it is used for the inhibition studies. During the time needed for an inhibition experiment (30 minutes) a considerable percentage of the starting material is transformed. This observation ruled out the quantitative determination of the inhibition constants for the 3-chloro and of the 3-bromolevulinate. In the literature the use of these two derivatives as inhibitors of PBGS has been reported.[75] The inhibition constant for 3-chlorolevulinate was reported to be 11 µM.[75] However to obtain irreversible binding much higher concentrations of 3-chlorolevulinate had to be applied. This somewhat surprising result might be the consequence of the hydrolysis of the 3-chlorolevulinate. The formed 3-hydroxy levulinate is a good inhibitor in its own right. To obtain the alkylation of the enzyme much higher concentration of 3-chlorolevulinate are necessary. These higher concentration might be necessary, to assure that sufficient concentrations of the unhydrolysed 3-chlorolevulinate are present at the active site.

In order to avoid the problems with hydrolysis and elimination of the substituent at the 3-position of levulinic acid we decided to concentrate our efforts on the synthesis of derivatives of levulinic acid which contain a hydroxymethyl group or a synthetic equivalent of this group at the 3 position.

2.1. Synthesis using alkylation of b-ketoesters

The easiest way to obtain the desired skeleton seemed to be the use of an alkylation reaction of ethyl acetoacetate with an adequate ester of bromoacetate. The alkylation in homogenous solution using NaH as a base worked well (Scheme 11).[87] The major problem was the formation of appreciable amounts (11 %) of the dialkylated product 3. The best solution to achieve selectively this transformation was to proceed to the reaction in solid state using the conditions developed by Ranu (Scheme 11).[88]

Scheme 11: Alkylation of ethyl acetoacetate and trials to obtain the compound 6.

Adsorbing the base, potassium tert.-butoxide, onto aluminium oxide activated at 180°C followed by the addition of the ethyl acetoacetate gave after 30 minutes of stirring a powder to which the benzyl bromoacetate was added. This solid mixture containing all the reagents was stirred for 24 h until the TLC showed the completion of the transformation. Extraction of the product from the aluminium oxide with dichloromethane, evaporation of the solvent and finally purification of the product by bulb-to-bulb destillation gave a 79 % yield of the product 2.

Reduction of the benzyl ester using Pd on charcoal as catalyst, ethanol as solvent at 70 atm H2-pressure gave in excellent yield (87%) the corresponding diethyl ester 7. This undesired side reaction could be suppressed using ether as solvent instead. Flash chromatography gave an 84 % yield of the analytically pure product.

Trials to reduce the ester without touching the acid were unsuccessful.[89] Forming the ate complex between DIBAH and n-BuLi at -78°C and adding either NaBH3CN or even NaBH4 in ethanol at this temperature and warming the reaction mixture up to room temperature did not give the desired product. We were only able to reduce the keto function to the alcohol, without touching one of the two carboxylic acid derivatives. Forming the same ate complex at -78°C and then warming the solution up to room temperature before adding NaBH4 gave an untractable mixture, which still contained the ester function as a major component.

In this context the diketo acid 9 was of interest to us and has been synthesised in a direct two step procedure. Alkylation of acetylacetone with methyl bromoacetate using NaH in THF as a base[87] gave a 81 % yield of the monoalkylated product 8 (Scheme 12).[90] Hydrolysis of the ester function using commercial porcine liver esterase gave a good 85 % yield of the desired diketo acid 9.[91]

Scheme 12: Synthesis of the diketo acid 9.

2.2. Synthesis using aldol reactions and Mukaiyama aldol reactions

In view of the difficulties encountered for the reduction of the compounds of the type 4, we decided to try to modify the protecting group for one of the carboxylic groups in our starting material. Furans have been oxidatively transformed into carboxylic acid functions.[92,93] So we decided to synthesise derivatives of the structure 10 which potentially would be precursors for compounds of the type 4 (Scheme 13).

Scheme 13: Synthesis of furan protected precursors 10 via aldol reaction.

The aldol process was executed according to the procedure developed by Lehnert.[94-96] Adding fufuraldehyde and ethyl acetoacetate to the yellow precipitate obtained by adding a CCl4 solution of two equivalents of TiCl4 to THF and then adding a large excess of pyridine at -5°C allowed to obtain a 77 % yield of mixture of the (E)- and (Z)-diastereoisomers (E : Z ratio = 3 : 4). The reduction of the double bound proved to be difficult. Of the many methods used some did not touch the substrate and untouched starting material could be recovered. This was the case for reduction conditions using the following reagents: Mg in methanol,[97] 1 atm of H2 pressure in the presence of Raney nickel.[98] Under most conditions however the starting material was decomposed. Treating the starting material 10 with NaBH4,[99] LiAlH4,[100] 1 atm of H2 pressure in the presence of Pd on charcoal,[101] and also using cob(I)alamin in the presence of acetic acid [102] lead only to decomposition of the starting material. In our hands only two methods could be applied successfully (Scheme 14). Using the mixture obtained by treating tellurium powder with NaBH4 in ethanol[103] could be used in small scale experiments for the selective reduction of the a,b-unsaturated ketone.[104] The reduction product 11, where the double bond and the ketofunction had been reduced at the same time was obtained in 81 % yield. For further transformation the so obtained alcohol 11 was protected as the TBDMS ether 12 in 83 % yield.[105] Unfortunately we were not able to reproduce this transformation on a large scale. The preparation of the reducing agent proved to be very capricious.

Scheme 14: Successful reduction of the condensation product 10.

The procedure developed by Ojima where an a,b-unsaturated ketone is reduced by a formal 1,4-addition triethylsilane could be successfully applied to the condensation product 10.[106-108] Treating the condensation product 10 with one equivalent of triethylsilane in the presence of 1 mol % of tris(triphenylphosphine)rhodium chloride in dry benzene at 55 °C allowed to isolate the mixture of the 1,4-addition product 13 and the product 14 obtained by hydrolysis.[107] A 92 % yield of this crude mixture could be isolated. Direct hydrolysis of this mixture gave an 87 % overall yield of relatively pure product (93 % pure by NMR). Trials to purify by flash chromatography were successful, but heavy losses had to be accepted. Only 29 % of the material put onto the column could be recovered as pure keton 14. It is known that alkyl substituted furans are sensitive against acids [109-112] and the acidity of the silica gel of the chromatography column was enough to diminish the yield of the isolated product considerably.

In view of the sensitivity of the intermediates and the irreproducibility of one of the reduction steps we decided to modify our approach.

Using the chemistry developed in our laboratory for the synthesis of alkyl substituted pyrroles,[54] we were able to introduce a methoxymethyl group regioselectively onto the levulinic acid (Scheme 15).

Scheme 15: Application of Mukaiyama reaction

Starting from the silyl enol ether 15[113] obtained from the corresponding bromo ester [114] the additional carbon could be introduced under standard Mukaiyama conditions.[115] An 81 % yield of the pure compound 16 could be isolated after flash chromatography. This compound could be stored, but slowly lost methanol if kept in the refrigerator. Treating 16 with commercial pig liver esterase lead to the relatively slow hydrolysis of the ester group. The corresponding acid 17 could be isolated in 72 % yield after flash chromatography. The elimination of methanol was fast and spontaneous and the a,b-unsaturated compound 18 (7 %) could be identified already after the isolation procedure.

2.3. Synthesis using the Baylis-Hillman approach

The Baylis-Hillman reaction has gained recently in importance.[116-121] The Baylis-Hillman reaction is interesting because of the mechanism, which is based on the "Umpolung" [122] of the a,b-unsaturated ketone and because highly functionalized building blocks can be obtained in one simple transformation.[123]

Applying the Baylis-Hillman conditions [124] to furfuraldehyde or glyoxalate gave the product 19 - 21 in yields of purified product between 74 % and 89 % (Scheme 16).

Scheme 16: Application of the Baylis-Hillman reaction

We hoped to be able to obtain the desired product, in a two-step procedure: First reducing the alcohol function and then or in some cases concomitantly with the reduction of the alcohol function we planed to execute a Michael addition to the a,b-unsaturated ketone. Unfortunately the benzylic alcohol 19 proved to be in our hands too sensitive. Trials to remove the alcohol function reductively using triethylsilane in the presence of BF3 etherate at 0 °C [125] lead to complete destruction of the substrate (Scheme 17). Lowering the temperature to -30 °C stopped the whole process and only starting material could be isolated. Using scandium triflate as Lewis acid [126] at room temperature lead to decomposition. Using the mixture of triethylsilane and a Broenstedt acid lead either to complete decomposition in the presence of trifluoroacetic acid [127] or if only acetic acid was used no reaction could be observed.[128] Trials obtain the rearranged product as a consequence of a formal SN2' process met with no success in our first trials. Treating the starting material 19 with N-bromosuccinimide in the presence of dimethylsulfide [129,130] only the complete destruction of the starting material could be observed. Trials to use the possibility to activate the alcohol via silylation and then to undergo the SN2' process using iodide or mechanistically to obtain the 1,4-addition product to the a,b-unsaturated ketone induced by silylation and then followed by elimination of the alcohol due to the double activation of the furan ring and the silyl enol ether met no success.[131]

Scheme 17: Unsuccessful trials to transform Baylis-Hillman product 19

Transformation of the Baylis-Hillman products 19,20 into the acetate 22,23 could be achieved quite efficiently in 83 and 80 % yield.[132] We hoped to transform this product by thermal treatment into the thermodynamically more stable rearranged product (Scheme 18). We were unable to achieve this transformation.

Scheme 18: Successful SN2' reactions starting from the Baylis-Hillman products 19,20

Using the Baylis-Hillman products 19 and 20 the SN2' reaction could be achieved under Mitsunobu conditions[120] using benzoates as nucleophiles (Scheme 18). Starting with the Baylis-Hillman product 19 obtained from furfuraldehyde only the (E)-diastereoisomers 25 and 26 were obtained in moderate to good yields. When the Balis-Hillman product 20 obtained using glyoxalate was used a mixture of the (E)- and (Z)-diastereoisomer 27 was isolated in a disappointingly low yield. Trials to reduce the a,b-unsaturated ketone by a rhodium catalysed 1,4-addition of triethylsilane were unfortunately unsuccessful.[107].


Go to top
Continue to Conclusion
Go to References
Back to main menu