For the planned synthesis of porphobilinogen and of structural analogues of prophobilinogen we needed the corresponding silyl enol ether of a protected derivative of 5-aminolevulinic acid (see figure 18).

Figure 18: Retrosynthesis for the planned biomimetic approach to porphobilinogen or to structural analogues thereof.

Trials to submit the 5-azido levulinic acid methyl ester 40, or the 5-tert.-butyldimethylsilyloxy-levulinic acid methyl ester 43 to the conditions worked out by Miller[57] inevitably lead to the unwanted regioisomers of the silyl enol ether die 42 and 45 (see figure 19).

Figure 19: Unsuccessful trials to achieve the transformation of adequately substituted ketones under thermodynamic conditions into the wanted regioisomeric silyl enol ether.

The 5-bromo levulinic acid methyl ester 46 was not stable under Miller's conditions. Even when the conditions developed by House were used no trace of the wanted silyl enol 47 ether could be isolated. However replacing DMF by CH2Cl2 as solvent allowed to achieve the transformation at room temperature in the presence of DMAP. Under these conditions a 60 % yield of the silyl enol ethers 47 and 48 could be isolated. But even under these reaction conditions the ratio in favour of the wanted product 47 was only 2 : 1 and 40 % of the product of elimination was isolated as well. Using TMS-triflate in a slight excess in the presence of triethylamine yielded a cleaner mixture of the two regioisomeric silyl enol ethers 47 and 48. Adding after a short reaction time an additional 20 Mol % excess of 5-bromo methyl levulinate 46 and TMS-triflate and stirring the reaction mixture for another 15 days at room temperature lead to equilibration. Under these conditions 73 % of a mixture could be isolated containing the two wanted diastereomeric silyl enol ethers 47 in 37 %, and the unwanted silyl enol ether 48 in 13 % and the starting material in 23 % yield.

The synthetic problem could be finally solved using the 5-phthalimido-levulinic acid methyl ester 49 submitting it to Miller's conditions but changing the solvent to chloroform (see figure 20).[58] Adding after 17 hours dry hexane allowed to precipitate the salts which had been formed. The wanted silyl enol ether 51 was obtained in 93 % yield as 1 : 1 mixture of the diastereoisomers containing only 4 % of the unwanted silyl enol ether 50 as side product.

The silyl enol ether 51 could be stored for months at - 20 °C in the refrigerator. The regioselectivity of the formation of the silyl enol ether is surprising, because one has to assume the methylene group in the a-position to the phthalimido group should be clearly more acidic than the protons at the C3 methylene group. The regioselectivity can probably be attributed to the steric hindrance which can not be avoided if the silyl enol ether towards the position C3 is formed. The X-ray structure of the starting material the 5-phthalimido levulinic acid methyl ester 49 corroborates this proposal (see figure 21).

The phthalimido group is in an orthogonal position compared to the plane defined by the carbonyl group. In this conformation the oxygen atoms of the phthalimido group avoid the unfavourable steric and electronic interactions with the carbonyl oxygen. At the same time a favourable intramolecular p-p-stacking can be observed.

Having the correct regioisomer 51 in our hands experiments were undertaken to couple the silyl enol ether with the acetal of the 5-azido levulinic acid methyl ester 35 under standard conditions.[55] Unfortunately all our trials to couple these two precursors of 5-amino levulinic acid according to Mukaiyama were totally unsuccessful. Despite our intensive efforts to achieve the formation of this crucial C-C-bond we were unable to accomplish this transformation.

In order to check the reactivity in and the potential of the a-phthalimido silyl enol ether for the use in the Mukaiyama crossed aldol reaction we decided to use simple model compounds and to submit them to our reaction conditions. Phthalimido acetone (52) can be obtained in one step from chloro acetone and potassium phthalimid.[59] Treating 52 with two equivalents of TMSI and HMDS in CHCl3 during 3 h lead in almost quantitative yield to a mixture containing 90 % of the desired silyl enol ether 53 containing 7 % of the regioisomeric silyl enol ether and 3 % of the starting material (see figure 22).

Work-up avoiding the contact with water allowed to obtain the product in sufficient purity for our further studies. Also the silyl enol ether 53 could be stored in the deep-freezer for months. Already our first trials to submit the silyl enol ether 53 to the Mukaiyama aldol coupling met with success. Using the dimethylacetal of levulinic acid methyl ester (54) as coupling partner allowed to obtain the aldol product rac 55 in 31 % isolated yield. Using the same standard conditions the acetals of the azido acetone (56) and of 5-azido levulinic acid methyl ester (35) could be coupled in 53 respectively 68 % yield. The two aldol products 57a and 57b could be reductively transformed in 46 % respectively 41 % to the corresponding substituted pyrrole 58a and 58b (see figure 25). Pd on charcoal was used as catalyst and methanol as solvent. Studying a sample which has been taken after 3 h in the 1H- and 13C-NMR showed the presence of an intermediate, whose spectral datas were in accordance with the D1-pyrrolenine structure. The final product, the pyrrole 58b, could be crystallised the structure could be determined with an X-ray diffraction study (see figure 23).

Having shown that the aldol coupling with the model for our silyl enol ether could be successfully achieved, we tried to react the silyl enol ether 51 under our optimized conditions. In order to check the reactivity of 51 in the crossed aldol reaction we first tried the coupling reaction using the dimethyl acetal of benzaldehyde (59) (see figure 24).

The reaction of 51 and 59 in the presence of 1.1 equivalents of TiCl4 during 3 1/4 h gave 60 in 63 % after crystallisation. Careful recrystallisation allowed to isolate the main diastereoisomer which could be submitted to X-ray analysis (see figure 25). The relative configuration of the main diastereoisomer is unlike.

Despite our considerable efforts to achieve also the Mukaiyama Aldol coupling using a protected form of 5-amino levulinate, we were unable to isolate products which could be traced back to the central C-C-bond formation. Using TiCl4 as a catalyst for the aldol coupling starting from the silyl enol ether 51 at temperatures below -40 °C no reaction could be observed. Increasing the temperature above -40°C rapid destruction of the reaction partner was observed. Using Lewis acids like TMSOTf[60,61] or the "super-Lewis acid" B(OTf)4TMS according to Davis[62] the aldol reaction between 51 and the dimethyl acetal of levulinic acid methyl ester (54) could be achieved. Using Noyori's conditions [60] whereby 0.11 equivalents of TMSOTf are utilised 30% of one pure diastereoisomer 61 could be isolated. Even when these stronger Lewis acids were used we were unable to achieve the crucial C-C-bond forming process starting from an adequate precursor of 5-amino levulinate. The use of the more reactive, but also more aggressive catalyst TMSI[63] at -80 °C lead to the destruction of both starting materials: the silyl enol ether 51 and the acetal.

The only way out seemed to be to increase the inherent reactivity of the carbonyl component. Reacting the silyl enol ether 51 with the succinic acid mono chloride mono methyl ester yielded the b-diketone 62 in 35 % isolated yield (see figure 26).

Treating this diketone with hydrazine hydrate in methanol for one hour yielded the pyrazole 63 in 83 % yield. The protecting groups could be removed by boiling 63 in 1N HCl for 24 h. Using an ion exchange column Amberlite XAD-2 the phthalic acid could be separated from the pyrazole 64 which was obtained in 87 % yield. Crystalisation from acetone/H2O gave an analytically pure sample.

Structurally the pyrazole 64 has a strong resemblance to PBG. However the reactivity of the two compounds is totally different. Pyrazoles in general a clearly less electron rich than pyrroles. Therefore we could boil the pyrazole for 24 h in 1N aqueous HCl. It is well-known that PBG or even precursors of PBG would not resist to such drastic conditions. It has been reported that PBG forms porphyrines, oligomers of pyrroles and polymers like pyrrole black under such forcing conditions .[64,65]

Pyrazole are present in solution in their two tautomeric forms. The X-ray analysis of the protected pyrazole showed the presence of only one of the two possible tautomeric forms in the solid state (see figure 27).

In order to get information about the presence of one or two tautomeric forms in solution we measured an 15N,1H HMQC-spectrum. The signal ati -170.7 ppm, which can be attributed to the NH of the pyrazole is correlated with the H2C(31)-group of the propionic acid side chain. This is a strong indication that in solution the same tautomeric form is present as the one which was observed in the solid state.

In order to obtain the porphobilinogen itself we tried to use the mono methyl succinic acid monocyanide (7) as activated carbonyl component. In this strategy it should be possible to combine two partners which contain all the carbon, oxygen and nitrogen atoms necessary for the construction of prophobilinogen. Deprotection of the aldol product should then induce the ring closing and aromatisation process. In view of this analysis we reacted the silyl enol ether 51 with the mono methyl succinic acid monocyanide (65). The cyano hydrine could be detected in the raw product of the reaction. Extraction against water and purification with column chromatography yielded 35 % of the b-diketone rac-62 as hydrolysis product (see figure 28).

Figure 28: Aldol coupling between mono methyl succinic acid monocyanide (65) and the silyl enol ether 51

Under optimised conditions at 20° C and using TiCl4, which had been freed from HCl by distillation over polyvinyl pyridine, the aldol product rac-66 could be obtained in 60 to 87 % (see figure 29).[58] One diastereoisomer of the aldol product rac-66 could be obtained analytically pure by crystallisation in 47 % yield. Trials to reduce the cyano hydrine directly met with limited success. For the synthesis we protected the unpurified aldol product using acetone enol acetate. The acetylated aldol product rac-67 could be obtained in 56 % yield. Even the reduction of the acetylated cyano hydrine rac-67 proved to be difficult. Finally the cyanohydrine rac-67 could be reduced smoothly at 65° C under 120 atm H2 in the presence of Raney nickel. After column chromatography we obtained the fully protected porphobilinogen 68 in 54 % yield analytically pure. Removal of the protecting groups over two steps has already been described in the literature.[33]

In conclusion we were able to obtain the protected porphobilinogen 68 in a convergent way starting from two easily obtainable starting materials. The central step of the synthesis is the Mukaiyama aldol reaction between the regioselectively formed silyl enol ether 51 as the nucleophile with succinic acid mono methyl ester mono cyanide (65) as electrophile. Reducing the acetylated cyano hydrine rac-67 yields directly the protected porphobilinogen 68. This synthesis follows the proposal for the biosynthesis made by Shemin exactly 30 years ago. The correctly functionalised side chains are introduced on the level of the two starting materials used for the synthesis of the pyrrole ring. Subsequent functionalisation is therefore not necessary. In this synthetic scheme the same bonds are formed as in the biosynthesis catalysed by porphobilinogen synthase. The overall yield starting from 5-phthalimido methyl levulinate is 25 %. The synthesis can be used to obtain selectively labelled porphobilinogen.