# Synthesis Golf I: Sodium Fluvastatin

A full FAQ exists on meta.chem.SE explaining the premise of synthesis golf, why we're doing this, and the ground rules. Please read this before posting an answer.

The target for this first round of synthesis golf is sodium fluvastatin, an artificial statin drug used to treat hypercholesterolemia and in the prevention of cardiovascular disease. You must provide a synthetic route to the target molecule.

In addition, to narrow the scope of this question:

• the synthesis must include a way of making the indole (i.e. no buying of the indole core and carrying out multiple functionalisation reactions).
• all of the stereo centres must be set. Relative stereochemistry only is fine (the 1,3-syn relationship between the hydroxyls), but if you can think of a way to set them with absolute configuration, then great.
• This is a neat idea - hopefully there are enough experts in organic synthesis on the site to make for a robust set of answers. – hBy2Py May 23 '17 at 21:47
• (may/may not be helpful to anyone but this doesn't cost me much) SMILES for target molecule O=C(C[C@H](C[C@H](/C=C/C(N1C(C)C)=C(C2=CC=CC=C21)C3=CC=C(F)C=C3)O)O)O and InChI InChI=1S/C24H26FNO4/c1-15(2)26-21-6-4-3-5-20(21)24(16-7-9-17(25)10-8-16)22(26)12-11-18(27)13-19(28)14-23(29)30/h3-12,15,18-19,27-28H,13-14H2,1-2H3,(H,29,30)/b12-11+/t18-,19-/m0/s1 – orthocresol May 23 '17 at 23:26
• This is so interesting! Can't we fully adopt this idea and create a new stack exchange like the code golf? – Kartikeya Badola May 24 '17 at 4:40
• @KartikeyaBadola you would need to have sufficient interest, if these become popular then it might be a consideration. See: area51.stackexchange.com/faq – orthocresol May 24 '17 at 12:07
• Maybe put a limit on the adjusted cost per kilo or cost per mole to eliminate some advanced intermediates that are readily available? I.e. To make it more commercially realistic? – Beerhunter May 27 '17 at 19:28

Edit: step count is 15, longest linear sequence of 9 steps

Just to (hopefully) get the ball rolling, here's something I scribbled down. I am sure people here will have better answers. And I have not done a lot of literature research to see whether the steps are plausible, so there might be fatal flaws... but hopefully not; as far as I know/can tell, everything here is relatively sensible.

Retrosynthesis first. I chose a Horner–Wadsworth–Emmons reaction to make the (E)-double bond, and something similar to the Reissert synthesis for the indole. It's not exactly the same because I didn't acylate the benzylic carbon, but the initial disconnection is similar. This leads to the key building blocks:

I chose the C(sp2)-C(sp3) bonds in 2 to disconnect. One can be made by SNAr with the amazing SNAr substrate 1-fluoro-2-nitrobenzene (5), and the other by Pd cross-coupling (I'm guessing cheaper ways probably exist). I've never actually seen SNAr reactions with enolates as nucleophiles, but I guess that if it doesn't work, SRN1-type conditions should still make that step possible:

As for 3, I figured it could be made from a symmetrical intermediate:

## Forward synthesis

Synthesis of 3, starting from cyclopentadiene:

(i) $\ce{BH3.THF}$, then $\ce{H2O2, NaOH}$; (ii) TBSCl, imidazole; (iii) $\ce{O3}$, then $\ce{H2O2}$; (iv) $\ce{TMSCHN2}$; (v) enzymatic desymmetrisation (I did not look up specifically which, but I think it should be possible); (vi) 7, NaH

Next, the synthesis of 2, starting from methylglyoxal:

(vii) ethylene glycol, TsOH; (viii) Pd cat., ligand, $\ce{NaO^{t}Bu}$, 6; (ix) NaH, 5

Forming the indole:

(x) Zn, AcOH; (xi) $\ce{NaH, ^{i}PrI}$; (xii) aq. HCl

and finally the HWE, the TBS deprotection, and the last redox step, a diastereoselective boron reduction. I wasn't entirely sure if it would work on an α,β-unsaturated ketone, but Comp Org Synth II, section 8.01 has several examples of it being used on an α,β-unsaturated ketones.

(xiii) LiHMDS, 3; (xiv) $\ce{TBAF, H2O}$; (xv) $\ce{Et2B(OMe), NaBH4}$

• Would the conversion from cyclopentadiene to the other compound work? I think cyclopentadiene is quite reactive, and undergoes a Diels-Alder reaction with itself. – Pritt says Reinstate Monica May 28 '17 at 14:00
• You buy the Diels-Alder adduct (sigmaaldrich.com/catalog/substance/…), then heat it to get the monomer (e.g. orgsyn.org/demo.aspx?prep=CV4P0238); you can use the monomer in further reactions. Wikipedia article has more info. – orthocresol May 28 '17 at 14:18
• I'm a bit concerned about your synthesis of compound 4. Methylglyoxal, being rather unstable in its pure form, is typically sold as a 40% solution in water, which would make acetal formation (and concomitant removal of water) rather difficult. Additionally, the formation of the acetal with TosOH and ethylene glycol would probably not proceed smoothly, likely forming the diacetal and some polymeric material. Methylglyoxal 1,1-dimethyl acetal is commercially available from Sigma-Aldrich, and avoids this problem (and saves a step). – JSK May 28 '17 at 21:42
• @JSK thanks! That's much more tractable. I've not done a lot of stuff in the lab yet - so my practical knowledge will be rather limited too. On paper I suppose it would be possible to selectively protect the aldehyde over the ketone, but I realise that in practice it's going to be an entirely different matter. – orthocresol May 28 '17 at 23:22
• @orthocresol no problem! I did like your use of hydroboration on cyclopentadiene; I wasn't sure of the regioselectivity on cyclic dienes, but it appears to be quite regioselective in modest yields. It seems like most syntheses of cyclopent-3-en-1-ol use either epoxidation of cyclopentadiene, followed by lithium aluminium hydride reduction, or use a metathesis reaction on hepta-1,6-dien-4-ol, but your method works nicely as a single step (in my opinion, but at least single pot) reaction from a simple starting material. – JSK May 29 '17 at 20:16

## Strategy

The indole core can be made with one of numerous methods; having had classes taught by Dr. Richard Larock, my first inclination was to use the Larock indole synthesis. The Larock indole synthesis uses an N-alkylated/acylated/tosylated 2-iodoanilines and a di-substituted alkyne to produce 1,2,3-trisubstituted indoles in one step. However, care must be taken to ensure appropriate regioselectivity of the product. The use of a TMS-substituted phenylacetylene produces a 2-trimethylsilyl-3-phenylindole (Larock, JOC, 1998, 63 (22), 7652).

The main complexity of the molecule is found in the side-chain, containing a syn-1,3-diol with alkene and carboxylate functionality; installation of a chiral alcohol $\beta$ to a ketone would allow diastereoselective formation of the diol.

## Synthesis

The side chain may be made in a few steps from the cheap and commercially available 2,2,6-trimethyl-4H-1,3-dioxin-4-one (1), which is first deprotonated and reacted with TMSCl to produce compound 2. A Mukaiyama aldol-type reaction, mediated by a titanium (IV) compound, e.g. titanium tetrachloride or Ti(OiPr)$_3$Cl, between 2 and acrolein produces the racemic alcohol 3.

Refluxing 3 in methanol and toluene with catalytic pTSA removes the acetal-like protecting group to form the methyl $\beta$-ketoester 4. The Narasaka–Prasad reduction on 4 uses the boron reagent diethylmethoxyborane (available from Sigma-Aldrich) as a chelating reagent to permit the diastereoselective reduction with sodium borohydride to the alkenyldiol 5 (relative stereochemistry is shown).

It should be noted that the aldol reaction to produce 3 may also be performed enantioselectively, as in Singer and Carreira, JACS 1995, 117, 12360, to give the desired enantiomeric alcohol; compound 4 retains this stereochemistry, and the diol 5 would then be the single diastereomer shown. Alternatively, chiral resolution could be performed at this point, or the racemic compound could be used to produce racemic Fluvastatin.

As the Larock indole synthesis allows for the use of a 2-iodoaniline and a trimethylsilyl phenylacetylene derivative, the main core is synthesised using this reaction. A palladium- and copper-catalyzed Sonogashira coupling between 1-bromo-4-fluorobenzene (6) and ethynyltrimethylsilane gives the TMS-substituted 4-fluorophenylacetylene 7. Meanwhile, N-isopropyl-2-iodoaniline (9) is synthesized from 2-iodoaniline (8, from Sigma-Aldrich) and acetone via formation of the imine in acetic acid, followed by the in situ reduction with sodium triacetoxyborohydride.

The palladium-catalyzed Larock indole synthesis is then used to produce indole 10 from alkyne 7 and aniline 9. The 2-TMS-substituted indole is capable of undergoing a Heck reaction (see Larock, JOC, 1998, 63 (22), 7652) with alkenyldiol 5 to form 11, the methyl ester of Fluvastatin. Ester hydrolysis with sodium hydroxide in water gives the sodium carboxylate, Fluvastatin, 12, in nine steps.

I thought given the challenge, it was only fair that I 'wasted' some time thinking about this myself. The following route is hopefully plausible, even if not 100% thought through. Comments welcomed, and hopefully provides some competition to Ortho's great attempt.

### Strategy and retrosynthesis

The synthesis proposed here disconnects fluvastatin into two main components: an indole core, and a side chain containing the stereocentres.

Fragment union is proposed to take place via selective cross metathesis (mainly because @Orthocresol made use of a HWE and I thought some variety would be nice [though seriously, cross metathesis is great for not having to use pre-functionalised precursors]).

The indole core itself is proposed to be brought together using a modern variant of a Bischler Indole synthesis by cyclisation of a linear precursor under L.A. conditions, avoiding the potential regioselectivity issues of many indole syntheses involving carbonyl condensations.

For the side-chain, a vinylogous Mukaiyama aldol reaction should provide the required stereochemistry, using a copper-bisoxazoline catalyst to control the absolute facial selectivity.

### Synthesis of the Indole core

Starting from commercially available fluorobenzene 1 (Aldrich: 500G/16.30GBP) a Friedel Crafts acylation using the ethylene glycol (Aldrich: 1000mL/52.60GBP) derived acyl chloride in the presence of a Lewis acid catalyst could be carried out to give the acylated ring 2. Regioselectivity of the Friedel-Crafts is well precedented based on directing effects of the fluorine, with the 4-position both activated electronically and sterically most accessible. Due to the availability of the starting materials, poor regioselectivity at this stage not crucial, and the isomers could be separated.

Subsequent cleavage of the PMB under oxidative conditions furnishes the free alcohol 3 which may undergo an Appel-type iodination (could also do a Finkelstein via formation of the tosylate), displacement, and reductive amination with acetone to provide the indole precursor 6.

Upon treatment with a Lewis acid catalyst and heat, 6 should undergo (what I think I can call) a Bischler indole synthesis to furnish the indole core of fluvastatin 7, via this method no regioselectivity issues arise, since the two reacting partners are tethered together, with only the 5 membered ring being favoured.

Failing the milder Lewis acidic conditions, polyphosphoric acid (PPA) is common, and given the lack of delicate functionality in the molecule, this should pose no issue.

### Completion of the synthesis

With the indole core 7 in hand, functionality must be installed at the 2-position. Standard Vilsmeyer-Haack conditions afford aldehyde 8 which may undergo Tebbe olefination (or similar) to provide a terminal olefin 9 capable of undergoing (E) selective cross metathesis with side chain 18 (synthesis described below). Final ester saponification using NaOH provides the sodium salt of Fluvastatin 11.

### Stereoselective synthesis of the side-chain

The crucial step of this synthesis is the setting of the 1,3-syn diol with complete absolute and relative stereochemistry.

Chan's diene 12 has been shown to undergo enantio- and diastereo-selective Mukaiyama aldol reaction to furnish the 1,3-syn diol 13 using a copper-bisoxazoline catalyst. Protection of the diol as the acetonide, reductive debenzylation, DMP oxidation and Tebbe olefination then affords the required terminal olefin 18 for the planned cross metathesis.

### Conclusion

Based on the proposed synthesis, Fluvastatin Sodium 11 would be synthesised in 19 steps overall, with 13 steps in the longest linear sequence.

• Why not do the acylation of fluorobenzene with chloroacetyl chloride? – Waylander May 25 '17 at 10:30
• I wasn't confident which chloribe would get ripped off first by the Lewis acid - I'm sure there's precedence, I just would take my chances – NotEvans. May 25 '17 at 10:49
• Org. Syn. Coll. Vol 3, 191 4-F-phenacyl chloride is commercially available – Waylander May 25 '17 at 10:53
• I admittedly forgot about the existence of Vilsmeier formylation. Regarding the Mukaiyama, does that produce the diol in one step? All I can find is Evans' report on it, where he used $\ce{Me4NBH(OAc)3}$ to reduce the beta-hydroxyketone to the anti diol: JACS 1996, 118, 5814 P/S Did you use ChemDraw for this? – orthocresol May 26 '17 at 15:16
• Frutiger (Frutiger Neue to be precise IIRC). – NotEvans. May 26 '17 at 17:54

Reference through step 2: Synthesis of carbon–14 labeled fluvastatin, Journal of Labeled Compounds and Radiopharmaceuticals volume 41, pages 1-7.

Step 0 (product is commercially available):

Fluorobenzene + Bromoacetylchloride $\ce{->}$ 2-Bromo-4′-fluoroacetophenone

(Aluminum chloride, 75 °C)

Step 1:

(intermediate created under nitrogen atmosphere, ethanol solvent, 78 °C, 1 hour. Intermediate then reacted with zinc chloride, ethanol solvent).

(nitrogen atmosphere, phosphorous oxychloride, acetonitrile solvent)

Then see Facile and Highly Enantioselective Synthesis of (+)- and (−)-Fluvastatin and Their Analogues Journal of Organic Chemistry, volume 75, pages 7514–7518.

Step 3:

3-[3-(4-Fluorophenyl)-1-isopropyl-1H-indol-2-yl]-propenal + diketene $\ce{->}$ 3-[3-(4-Fluorophenyl)-1-(1-methylethyl)-1H-indol-2-yl]-5-hydroxy-3-oxo-(6E)-heptenoic acid-1-isopropyl ester

(in the presence of Ti(O-i-Pr)4 and chiral Schiff base, -40 °C, methylene chloride)

Step 4: reduction of ketone with $\ce{NaBH4}$ to yield fluvastatin isopropyl ester

Step 5: saponification to yield fluvastatin product.

(Step 6: preparation of the chiral Schiff base from 3-tert-Butyl-2-hydroxybenzaldehyde and valinol)

For alternatives to steps 3-6, see An Improved Manufacturing Process for Fluvastatin Organic Process Research & Development volume 11, pages 13-18.

And for alternatives generally see:

Section 9.1 of Fluvastatin in Pubchem

US patents 4739073 and 5354772

Cardiovascular Drugs in Ullmann's Encyclopedia of Industrial Chemistry.

I'm stuck with the lateral chain, but here are my two cents for the formation of the indole with just a few steps/reagents (I admit I'm quite a newbie, yet an enthusiast in organic chemistry).

Maybe help is not admitted, but I hope my idea could be integrated by others:

Sigma-Aldrich has readily available the p-fluorophenylacetic acid.

A typical Fischer synthesis for the indole might be performed with said compound and (the diazonium salt of) aniline.

The product is brominated with $\ce{PBr3}$. This results in 2-bromo-3-(4-fluorophenyl)-1H-indole

The amine might be alkylated by reaction with a 2-halopropane.

A Grignard reagent could then be formed, and by reaction with a haloalkene, a Kumada reaction could attach the side chain. Yet, I admit that I didn't come up with a decent way of preserving the stereochemistry of the side chain.

"Ai posteri l'ardua sentenza"

A synthesis which involves an Evans auxiliary to induce asymmetry, and a Narasaka-Prasad reduction to control diastereoselectivity.