General remarks and comparison of the Shemin mechanism with the Knorr pyrrole synthesis
Concentrating only on the step connecting the two substrate molecules via a covalent bond for the first time, there have been three mechanisms postulated so far for the biosynthesis of porphobilinogen. In two of these mechanisms the central decisive step for the building of the pyrrole ring is the formation of the carbon-carbon bond between C3 of one 5-aminolevulinic acid (12) reacting as a nucleophile and the keto function of the other 5-aminolevulinic acid. It was assumed that the nucleophilic 5-aminolevulinic acid partner is bound to the enzyme as an enamine. One would expect that the formation of the central carbon-carbon bond should be slowest and therefore the rate determining step of the biosynthetic sequence. This sequence is in remarkable contrast to the mechanism of the Knorr pyrrole synthesis (see figure 9).[36,49-51]
Figure 9: Mechanism for the Knorr pyrrole synthesis.
For the Knorr pyrrole synthesis the steps leading to the pyrrole nucleus are the same, but the sequence is clearly different from the Shemin mechanism for the biosynthesis. In the Knorr synthesis the carbon-nitrogen bond is formed first and the aldol-like carbon-carbon bond forming reaction is therefore an intramolecular process.
Following the mechanistic reasoning first proposed by Shemin, one can ask the question if pyrroles can be synthesised using the same sequence of transformations and if it will finally be possible to synthesise even porphobilinogen (11) in a biomimetic way using such a methodology?
The Mukaiyama crossed aldol reaction seemed to be ideally suited for our trials to obtain the crucial carbon-carbon bond.[52]
Figure 10: Novel pyrrole synthesis using acetamino acetone (22).
Applying the Mukaiyama conditions to the carbon-carbon bond formation between the silyl enol ethers of levulinic acid methyl ester 20 and 21 and the acetal of acetamino acetone (22),[53] we were able to isolate small quantities of pyrrole 24 already in our first trials (see figure 10). Optimising the reaction conditions we could isolate up to 48% of the pyrrole 24. A large excess of titanium tetrachloride was necessary to obtain good yields of the product.
Trials to use this new pyrrole synthesis for the formation of other pyrroles met with mixed success (see figure 11).[53]
Figure 11: Novel pyrrole synthesis.
Treating the silyl enol ether of cyclopentanone (27) with the acetal of acetamino acetone (22) gave the mixture of the diastereoisomeric aldol products 28 in 83% yield. These aldol products 28 could be quantitatively transformed into the annelated pyrrole 29 using benzene as solvent and p-toluene sulphonic acid as catalyst. Using deactivated silyl enol ethers like the silyl enol ether from 1,3-cyclohexadione (30) the reaction conditions had to be much harsher and still the yield of the pyrrole 31 was disappointingly low. Using the silyl enol ether deactivated by the cyano group 32 we were unable to isolate any pyrrolic product.
We decided to study two ways to improve the crossed aldol reaction:
1) synthesising the pure regioisomers of the silyl enol ethers;
2) replacing the amide function by another protecting group, compatible with the crossed-aldol reaction.
For the synthesis of the isomerically pure silyl enol ether we developed a modification of the procedure of Rubottom[54] for the reductive silylation of the corresponding bromoketone. The reductive silylation had to be carried out in the absence of a base. To avoid problems during work-up the zinc salts had to be precipitated using TMEDA to complex the zinc salts and then adding pentane. As long as most of the zinc salts could be removed by filtration the silyl enol ether 20 could be distilled.
Figure 12: Regioselective synthesis of the silyl enol ethers 20 and 21.
To synthesise the regioisomeric silyl enol ether 21 the 5-bromo levulinic acid methyl ester (34) was treated according to the same procedure and a 77% yield of the silyl enol ether 21 could be isolated (see figure 12).
Instead of the amide protecting group, we decided to use the azido group. Already in our first preliminary trials with the mixture of the silyl enol ethers we were able to isolate the aldol products (see figure 13).[55]
Figure 13: The crossed-aldol reaction followed by the Staudinger reaction applied to the synthesis of the pyrroles 38 and 39.
The aldol products were treated with triphenylphosphine in benzene to induce a Staudinger reaction.[56] We were able to isolate the pyrroles formed 38 and 39. The separation of the triphenylphosphineoxide from the alkylpyrroles 38 and 39 was delicate and therefore the yields were not satisfactory. The use of the regioisomerically pure silyl enol ethers 20 and 21 considerably improved the yield of the aldol process (see figure 14).[55]
Figure 14: The crossed-aldol reaction using the pure silyl enol ethers 20 and 21.
Replacing the triphenylphosphine by triethylphosphine, the water soluble triethylphosphine oxide is formed which can be easily removed by extraction (see figure 15).
Figure 15: Modified Staudinger reaction.
Catalytic reduction is another mild method to transform the azido group into the corresponding amine. Using palladium on charcoal as catalyst and methanol as solvent the aldol products could be reduced (see figure 16). The amino ketone formed spontaneously the corresponding pyrrole. The work-up using these conditions was very convenient.
Figure 16: Catalytic reduction
The new two-step pyrrol synthesis allows to synthesise mono-, di-, tri- and tetraalkylpyrroles in good yield (see figure 17). The yield of pyrrole obtained by reduction was extremely low for the annelated products.[55]
Figure 17: Pyrroles synthesized using the Mukaiyama crossed aldol condensation
The synthesis is complementary to the classical Knorr pyrrole synthesis. It allows to introduce the side chains at the correct positions and with the needed functionalities already in the pyrrole forming step. The reaction conditions for the pyrrole formation are sufficiently mild to allow also the isolation of highly sensitive pyrroles. At this stage of the project we hoped to be able to apply our reaction conditions to a synthesis of porphobilinogen avoiding many of the pitfalls of the former synthesis.