B.R.S.M. Help! I'm trapped in a molecule factory!


Woodward Wednesday 3: Chlorophyll a

Okay, so it's not Wednesday, but I've just finished this and I'm not going to sit on it for another 4 days. Merry Christmas!

The Total Synthesis of Chlorophyll a

Full Paper: R. B. Woodward et al., Tetrahedron, 1990, 46, 7599-7659 [PDF]

Communication: R. B. Woodward et al., J. Am. Chem. Soc., 1960, 82, 3800 [PDF]

Lecture: R. B. Woodward,  Pure Appl. Chem., 1961, 2, 383 [PDF] FREE!


One thing I enjoy most about reading Woodward’s work is the variety of targets and natural product classes he worked on. He conquered steroids, alkaloids, polyketides and amino acid derivatives with equal aplomb, pushing the limits of complexity in each field, and always looking for the highest peak to climb. Woodward’s synthesis of chlorophyll a has been somewhat overshadowed by his truly epic collaborative synthesis of vitamin B-12 (some years later), but for its time was an outstanding accomplishment, unmatched in the field of porphyrin chemistry. Unfortunately, most books on classic or collected total syntheses only cover B-12, but plenty of interesting chemistry (as well as breathtaking experimental skill) was brought to bear in this earlier, simpler campaign.[1] I’m not going to write anything here about the importance of chlorophyll as it's a bit obvious, so on to the chemistry.[2]

In this case, as with so many of Woodward’s ‘total syntheses’ (quinine being particularly controversial),[3] the titular compound itself was never produced in his lab. Instead, he reached a compound from which conversion to the final target was known i.e. he achieved what would now be more correctly called a formal (total) synthesis. Here, presumably for reasons of simplicity (and the awful yields in the final stages of the route), Woodward decided to target the slightly simpler derivative chlorin e, as its trimethyl ester. This compound is only a Dieckmann cyclisation, transesterification and magnesiation away from chlorophyll a itself, and several different reported procedures exist for each step. One immediately striking feature of the route chose to this compound is that Woodward planned to construct all four of the fairly different pyrrole rings from a single precursor, the well known Knorr’s pyrrole. Thanks to a collaboration with a buddy at Chas. Pfizer & Co., the production of this compound was optimised and scaled up to pilot plant (i.e. multi-kilogram) scale, making it readily available for studies. One immediately impressive feature is the scale on which the initial steps were carried out, in many cases several hundred grams to one kilo.

The route to the A-ring pyrrole began with selective hydrolysis and decarboxylation of the β-ester group. The now free position opened up was then formylated under Vilsmeier-Haack conditions (on up to 2.5 kg at a time!), this sequence providing a neat and surprisingly high yielding solution to adjusting the oxidation level of the group at this position without affecting the α-ester. This aldehyde was then protected by condensation with malononitrile to allow oxidation of the α-methyl group to the corresponding methyl ester. A global hydrolysis of the methyl and ethyl esters, as well as the dicyanovinyl protecting group, with concentrated sodium hydroxide solution, gave the formyldiacid. The unmasked aldehyde was then condensed with nitromethane in a Henry reaction to give the nitroalkene, and this was then reduced to the nitroalkane using sodium borohydride in methanol. Both carboxylic acids were then removed by decarboxylation in sodium acetate – potassium acetate melt and final catalytic reduction of the nitroalkane to the primary amine using hydrogen and a platinum catalyst gave the required A-ring pyrrole. It was anticipated that Hofmann elimination later in the sequence could be used to convert this aminoethyl chain to the vinyl group present in the target. Although the preparation of this compound took 10 steps, the longest of any of the four pyrroles, the yields were generally good, and skilful optimisation allowed the required reactions to be performed on large scale to provide sufficient material.

The synthesis of the B-ring began with the same two steps as for the previous ring to selectively remove the β-ester, and the free position was then filled by an acetyl group, introduced by a Friedel-Crafts reaction with acetyl chloride in the presence of aluminium trichloride. This was then reduced all the way to the ethyl group using a Wolff-Kischner reduction, and the harsh conditions required for this transformation also caused hydrolysis and decarboxylation of the α-ester. Formylation of the α-position under Vilsmeier-Haack conditions gave an aldehyde that was again protected by condensation with malononitrile in essentially quantitative yield. Finally, chlorination of the α-methyl group using sulfuryl chloride in acetic acid gave the B-ring pyrrole in an excellent overall yield of 39% over 7 steps.

The synthesis of the C-ring pyrrole began with the oxidation of the α-methyl group to the corresponding acid that was then removed decarboxylatively by heating neat with copper bronze. The α-ester was then selectively hydrolysed without affecting the β-ester, and removed in the same fashion to give the C-ring pyrrole in just four steps.

The first three steps to the D-ring were shared with those of the A-ring, comprising selective hydrolysis of the β-ester, decarboxylation of the acid obtained, and formylation of the free position. This aldehyde was then condensed with malonic acid in the presence of aniline to give the unsaturated diacid as the Knoevenagel-type product. Hydrogenation with Raney Nickel under a hydrogen atmosphere in aqueous sodium hydroxide solution reduced the double bond, and effected monodecarboxylation to give the β-propionic acid. The α-methyl group was oxidised to the carboxylic acid, employing slightly different conditions to those used in the synthesis of the C-ring. Treatment of this compound with sodium hydroxide solution then simultaneously caused decarboxylation of this acid, as well as hydrolysis and decarboxylation of the ethyl ester. The propionic acid sidechain was then esterified using diazomethane and a semi-regioselective Vilsmeier-Haack formylation then gave a mixture of regioisomeric aldehydes. These could be separated by hydrolysis of their methyl esters to give two regioisomeric acids with very different solubilities in water. Re-esterification with yet more diazomethane gave the D-ring pyrrole in 8 steps and around 16% overall yield.

With all four pyrrole subunits in hand, the time had come to investigate conditions for their union. First the lefthand (AD) component was synthesised by simple condensation of the A and D pyrroles under acidic conditions to give the pyrromethene dibromide shown. It was found that swift isolation of the product from the reaction mixture was crucial to obtain a high yield as, although stable when pure, such salts were generally unstable in solution (and difficult to extract). For this reason, the reaction was carefully performed at -25 °C in methanol containing a small amount of water (100:1), conditions under which the desired compound simple crystallised out and could be obtained by filtration.

The righthand (BC) fragment was prepared by a similar reaction between the α-chloromethyl B-ring pyrrole and the disubstituted C-ring pyrrole, to give the dipyrrylmethane in modest yield. This was then subjected to a Friedel-Crafts type acylation to give the righthand α,α'-dicarbonyl compound originally intended for the macrocyclisation. So far, both halves of the macrocycle were obtained as essentially single regioisomers due to the inherant reactivities of the components, and optimisation of the reaction conditions. However, when the condensation of the lefthand and righthand components was attempted it was found that regardless of the conditions tried complex mixtures of compounds were obtained. This problem was attributed to an insufficient difference in reactivity between the ketone and aldehyde on the righthand half. In order to favour the desired macrocyclic product the group had to find a way to make the aldehyde more reactive. This was done by first forming the ethyl imine and then treating this with hydrogen sulfide in benzene (eek!) to give the corresponding thioaldehyde. When this thioaldehyde was stirred with the dipyrrylmethane (prepared by in situ reduction of the lefthand component) the expected imine formation with the ethylamino group occurred, extruding hydrogen sulfide (mmm eggy). This compound then spontaneously rearranged to form the desired AB linkage, and treatment with methanolic HCl effected the closure of the macrocycle by promoting the DC condensation. Finally, iodine was used to oxidise the intermediate compound (a phlorin; see next scheme) to the porphyrin, and the ethylaminogroup was protected as the acetamide to facilitate handling. Overall, the desired porphyrin was obtained  in 50% overall yield from the dipyrromethene.

The experimental procedure for this step is absolutely diabolical, requiring total fastidiousness in some aspects, yet surprising sloppiness in others, as well as an unusual experimental setup.[4] The main reason for this was the highly unstable nature of the dipyrrylmethane formed by reduction of the lefthand component that required it to be used immediately after preparation. A precise sequence of additions and a change of solvent with minimal exposure to the air were also required; if you have a subscription to Tetrahedron I’d definitely recommend a look.[5] The reaction apparently moves though a scintillating sequence of colours from orange to yellow, red and brown, with the final product obtained as violet needles. One fascinating detail of this reaction was the finding that, despite all other precautions being taken to exclude air and water, measurement of the methanolic HCl in an open measuring cylinder was found to give superior results to dispensing it more carefully. I’ll just copy a Woodward quote from the paper here:

“The necessity of this whimsical prerequisite for the successful execution of the reaction series with the stated yield was repeatedly verified. Without it, final yields of porphyrin lower by 10-15% were obtained. One of us, who swore a colourful oath that he would never, by pouring the acid, be party to such an offence against what he regarded (wrongly) as sound scientific procedure, was forced by cruel experience to suffer the ignominy of having a colleague pour his acid for him in order to achieve results comparable with his fellows. Numerous experiments designed to elucidate the cause of the effect led us to no conclusive rationalisation, nor did enable us to substitute a more respectable – if not easier – alternative procedure”

Hilarious. Apparently over 50 grams of the porphyrin were ultimately prepared by this route, although in batches of only a few grams at a time.

Before I continue discussing this synthesis, I’m just going to explain a little bit about nomenclature as it’ll make things easier in the long run. Most of us are probably familiar with the structure of porphyrins. Counting the double bonds which are cyclically conjugated there are 9, making these 18 electron systems and therefore Huckel aromatic (i.e. 4n +2). This makes them fairly stable, which is why nature uses them for all kinds of things. It also means they prefer to be planar, although they actually distort quite easily if enough adjacent substituents are added around the ring. The four positions between the pyrrole rings are known as the meso- positions. Now, if we add a molecule of hydrogen to a porphyrin we do this different ways. If we remove one of the double bonds that isn’t part of the aromatic annulene system then we get the chlorins, one of which is the target of this campaign.

During the optimisation of the previous step, Woodward discovered a new isomeric ring system, which he named the phlorins, where the aromaticity of the system had been disrupted by removal of one of the annulene double bonds. The first instance of this system characterised was as an intermediate in the above sequence, which although not usually isolated, could be obtained as the dihydrobromide salt by modification of the reaction conditions. Although, obviously a number of tautomers are possible for these systems, this one seemed to exist only in a single form, and was quite stable considering the general ease of gaining aromaticity. The reason for this, which is discussed at length in paper during the initial retrosynthesis, is what Woodward referred to as ‘peripheral overcrowding’. As I’ve tried to highlight in the phlorin below, the large meso propionate substituent clashes with the two ester groups on the nearby pyrroles. In tautomers where the meso carbon is sp3 hybridised then this strain is relieved as the subtituents can avoid each other to an extent. Conversely, this strain is exacerbated when this carbon is sp2 as all three substituents are forced to be coplanar, and this effect disfavours isomers where this the case. It is worth noting that although these compounds are stable enough to isolate and characterise, oxidation can still be effected using fairly mild oxidants such as the quinones DDQ and chloranil, or even molecular oxygen.

While investigating the chemistry of the newly available porphyrin system the group discovered another new phlorin – and a remarkable reaction. It was found that when the porphyrin was heated under nitrogen in acetic acid for just one hour then two hydrogen atoms from the meso-propionic acid sidechain migrated to give the phlorin acrylate ester shown (bottom left). Although this wasn’t planned, and certainly hadn’t featured in the group’s retrosynthesis (if such a term was in use at the outset of this project), it did open up possibilities for the synthetic route. Further investigation found that if the reaction was conducted under air or oxygen then oxidation of the phlorin took place to give the porphyrin acrylate ester in excellent yield. This compound, when isolated and heated for much longer, with acid but under nitrogen, underwent an unusual cyclisation.[6] Finally, the acetamide was hydrolysed under acidic conditions and the rather unstable resulting amine was subjected to Hofmann elimination by treatment with dimethyl sulfate and aqueous sodium hydroxide.

In order to avoid the extensive degradation that normally resulted from such conditions, all of the porphyrins prepared in this project were handled with minimal exposure to air and light. However, it was found that irradiation of the vinylpurpurin with a tungsten lamp under air and carefully controlled conditions effected a most unusual oxidative cleavage of the acrylate double bond in a manner reminiscent of ozonolysis. The reaction was thought to proceed via a [2 + 2] with 1O2 to give a dioxetane that then underwent a retro [2 + 2] to give the compound shown, the pupurin itself acting as both the photosensitizer and the substrate.

Comparison of this oxidation product with chlorin e6 trimethyl ester reveals that only three transformations were now required to react the target: excision of the methoxalyl group, resolution and homologation of the aldehyde to the longer chain ester. The first of these operations, a retro-Claisen condensation, was performed in fairly low yield by treatment with methanolic KOH (followed by regeneration of the other esters with diazomethane), to give the trans disposed D-ring ring isomer. These conditions also caused cyclisation of the C-ring ester onto the nearby aldehyde to give the methoxylactone. This slightly unexpected transformation was explained by the fact that cyclisation at this position had the effect of reducing the afore mentioned peripheral overcrowding at this part of the ring. In any case, the this compound was treated with aqueous sodium hydroxide in dioxane to hydrolyse the one remaining methyl ester (and incidentally convert the methoxylactone to the hydroxylactone), in order to provide a handle for resolution. Unfortunately, although resolution (via the quinine salt) could be successfully carried out (under entirely non-obvious conditions), it proved exceptionally difficult and the yield of optically pure material obtained was only 4% from the acid. Finally, treatment of this compound with yet more diazomethane gave the corresponding purpurin dimethyl ester in a surprisingly poor yield for this step. Crucially for the group, the disappointment of the resolution could now be put behind them because this compound could easily be derived from natural material (methyl pheophorbide a). Thus, a relay point had been reached and they were able to replenish their supply of material before pushing on for the finish. Comparison of various synthetic compounds at this points showed identity with naturally derived material, confirming the success of the route so far.

The diester was then treated with (freshly distilled) hydrogen cyanide and triethylamine to form the corresponding cyanolactone that was then reductively opened by cleavage of the C-O bond using Zinc dust in acetic acid. This represents a most unusual method for increasing the length of the sidechain by one carbon atom, and although the yield is dire the reaction was at least consistent, giving the same yield on 20 to 115 mg. Finally, with a simple hydrolysis of the nitrile group to the third methyl ester, chlorine e6 trimethyl ester was isolated, completing the formal total synthesis of chlorophyll a.

I would guess that the reason Woodward didn’t go all the way with this one was largely due to time and material concerns, presumably due to the low yielding final steps. In typical a Woodwardian fashion he celebrated in style, sending each member of the team a leaf marked with a red tick, his sign during lectures that a discussion of a particular target or subject had been concluded.

References and Addenda

1. It is a great shame that the full experimental details of the B-12 project were never published. I live in hope that one day time will be found, as for chlorophyll a, to go back and collect together the information for publication.

2. The history of chlorophyll’s structural elucidation, as with so many famous natural products of its era, was an enormous effort, spanning over 50 years. In the early years of the 20th century Willstatter managed, over a decade and a half of painstaking classical degradation spanning hundreds of reactions, to successfully assign the correct empirical formula to the molecule. Given the extreme difficulty of even isolating pure chlorophyll a (in fact, Woodward remarks that even 50 years later this is non-trivial), and the complete lack of the physical and spectroscopic methods on which we tend  to rely, this was an achievement well ahead of its time. In 1940 Fischer successfully proposed the correct structure of chlorophyll a, with only the stereochemical information undetermined. The final piece of the puzzle was eventually added by the British chemist Linstead (of dot notation fame), who eventually confirmed Fisher’s proposal and determined the missing relative stereochemistry in 1956. The absolute stereochemistry, although now known, was a mystery at the outset of Woodward’s studies but as his plan was to effect a late stage resolution, and presumably select the resolved compound which displayed the correct optical rotation when compared to a natural sample or the derivative of one, this was not important.

3. Woodward's synthesis of quinine has generated a vast amount of discussion related to its validity. To start with, the paper was titled 'The Total Synthesis of Quinine', although in modern parlance it was only a formal total synthesis via some earlier work by Rabe. The main objection voiced was that Rabe did not publish experimental information for the steps in question, and Woodward never repeated them. This lead some chemists, particularly Gilbert Stork, who claimed (rightfully) the first stereoselective synthesis of quinine, to dispute the fact that Woodward's work constituted a synthesis of the natural product at all. For a brilliant account of the debate, and some interesting historical information then there's a FREE Angewandte paper by Seeman that's well worth reading (Angew. Chemie. Int. Ed., 2007, 46, 1378). Another, non-free Angewandte paper published the following year by Williams titled 'Rabe Rest in Peace' contained a detailed analysis of Rabe's work and ultimately was able to successfully reproduce the steps in question, confirming the validity of Woodward's claim to a (formal) synthesis (Angew. Chem. Int. Ed., 2008, 47, 1736 –1740). The fact that two Angewandte papers have been published on this topic in the past 5 years shows just how hot the debate has been.

4. Although the paper wasn’t written by R.B. himself (he did make a start before his untimely death but the author, professor Raymond Bonnett, began from scratch), the level of experimental detail is still incredible and the full paper is well worth a look for this alone. A few quotes and diagrams of Woodward's were reused.

5. Irritatingly, I had to go and photocopy the entire 60 page paper out of the library as it was too old for our electronic subscription, but there are worse ways to spend one's teabreak.

6. Incidentally, the compound resulting from the formation of the new ring is what's known as a purpurin, that is a chlorin with an electron withdrawing groups in the meso-position, so called due to its purple colour.

Comments (8) Trackbacks (0)
  1. How about changing it from Woodward Wednesday to Woodward Work or World?
    Did anyone ever figure out a reason for that “whimsical prerequisite”?
    The description of the D ring preparation neglects to mention the first methylation.
    Are you counting the formylation, hydrolysis, separation of the isomers, & remethylation as 1 step to get a total of 8 steps to make the D ring? Is there any standard way of defining what constitutes a “step”?
    When rings B & C are combined the succinyl appears 1 step too early.

  2. In the reaction of joining rings B and C, the ethyl ester group in compound C appears as methyl ester in the reaction product. Please make the correction.

  3. Thanks guys! I’ve fixed both the mistakes in that scheme (unfortunately not at the same time). I’m surprised there aren’t more stupid mistakes in this as I wrote it over several weeks/continents/operating systems.

    Gippgig: I’ve not seen an explanation offered anywhere, regarding that strange requirement. As for the step count yes, I have counted that as one step. None of the intermediate compounds were isolated or purified, and it’s just one paragraph in the experimental for that sequence. Some would call it a ‘process chemist’s step’. For what it’s worth, I definitely think that step is distinct from ‘reaction’; for example there are several cases in the sequence above where aqueous base is used to do one thing and then the crude product is treated with CH2N2 to reform esters in the compound that underwent unwanted hydrolysis. I would probably argue that’s one step. What do people think?

    • None of the intermediate compounds were isolated or purified? Isn’t separating the isomers purifying?
      Why is the product of the unusual cyclization shown with a specific configuration of the ring D methyl (but not the propionyl ester)? The product is racemic since it is later resolved.

      • Separating the isomers only amounts to an aqueous workup. Here’s the procedure so you can decide for yourself: First the formylation is carried out with POCl3 in DMF. When this is done most of the solvent is removed and the mixture is dumped into a beaker of aq. NaOH and heated. After washing with ether to remove some oily residue the aqueous phase was then acidified to pH 3.5, causing the 5-formyl-4-methylpyrrole3-propionic acid to crash out. It is then collected by filtration while the 2-formyl remained in solution. I think up to this point this could be probably argued to be a single step. Reesterification of the crude product then gave the final compound.

        Regarding the methyl stereogenic centre – I didn’t mean to imply that this is a single enantiomer, because, as you say, it hasn’t been resolved at this point. Honestly, I think I only drew it this way as that’s how it is in the paper. I’m not sure why; it’s possibly a bit misleading. Might change that later…

        • I agree that “up to this point this could be probably argued to be a single step”. What was done after the isolation of the product (separating it from the other isomer) by filtration should be considered a separate step (the reesterification).

  4. A wonderful post! Thanks for your thoughtful overview of this incredible piece of work; it’s really useful to have it summarized like this.

    I’m struck by the many regioselectivity issues they had to face. A bold endeavor, for sure.

    I would have enjoyed seeing Woodward give a seminar on this work. I can see why his talks were three hours or more.

    • Thanks. Just saw your blog for the first time – looks great! Given the single digit yields reported by Fischer and others for previous porphyrin syntheses I think this was an almost unbelievably audacious undertaking for its time. It’s amazing that Woodward looked at all the unpromising precedent and decided he could do better. My supervisor, who has a photo of Woodward in his office (next to a photo of his favourite football team), has often said that never seeing Woodward lecture was one of the few disappointments of his time as a chemist. He was supposed to be spectacular.

Leave a comment

No trackbacks yet.