Type a reaction class (ex: alkylation) or name (ex: Lossen rearrangement)

Alternate Synthetic route and process proposal of AZD 3264 an IKK2 Inhibitor


Original synthesis procedure and route (used for large scale synthesis)

Exploiting the Differential Reactivities of Halogen Atoms: Development of a Scalable Route to IKK2 Inhibitor AZD3264, Pharmaceutical Development, AstraZeneca India Pvt. Ltd, Hebbal, OffBellary Road, Bangalore 560024, India, Org. Process Res. Dev.2014, 18, 646−651

The synthesis is already optimized (chosen route: scheme 4), but there are some drawbacks:
  • The use of boronic derivatives, which are now classified as mutagenic, if avoided, should be better.
  • The isoxazole derivatives is expensive (1260 $/kg – molbase price)
  • Cryogenic conditions to prepare the unstable boronic derivative 3 with n-Hexyl lithium.
  • The process described in the patent for the compound 6 (an in house product) use a toxic reagent to prepare the 2-[(aminocarbonyl)amino] : chlorosulfonylisocyanate.

Alternate route proposal
Cheaper starting materials (1336 $/kg less) but with 2 more steps (thiophene moiety excluded since this is an in-house product). This route avoids a cryogenic stage with n-Hexyl lithium (health and safety and plant capabilities considerations), and optionally avoids the use of boronic derivatives which are now classified as mutagen.

Sum-up of the modifications:
Modification of the starting material with three possibilities, essentially to introduce the isoxazole moiety: According to some lectures about the VNS of H (reference mentioned later), it maybe possible to use 1-halo-2-nitrobenzene which is cheaper (27$/kg (Cl) 101$/kg (F) – molbase) and 3-chloropentane-2,4-dione (358$/kg), to take advantage of the nitro EWG behavior. It will be reduced later, followed by a diazotation and CuBr/KBr or KI dependently of the method, which will avoid a cryogenic step. Also, to go back to the original route, the diazonium salt could be reacted with B2(OH­)4 which afford the boronic derivative (see reference later)

If doesn’t work, 1-chloro-5-fluoro-2-nitrobenzene probably do, which is unfortunately more expensive than the trihalobenzene (675$/kg – molbase), but the exceeding price of 480$/kg should be absorbed by other starting materials, pentane-2,4-dione (110$/kg - molbase) and hydroxylamine sulfate (25$/kg - molbase).

6 is an in-house product, but to avoid the use of the toxic reagent, i will use in this proposal, the 2-amino-3-cyano-thiophene (450 $/kg – mol base), which is commercially available and treat it with CDI/formamide to obtain the 2-[(aminocarbonyl)amino].

Also if the original compound 6 is used, an exchange could be made with i-PrMgClBis[2-(N,N-dimethylamino)ethyl] Ether Complexes, followed by a treatment with trimethylborate which lead to the boronic acid derivatives (see reference), instead of using the unstable aryl boronic derivative.

Aromatic Nucleophilic substitution (7) with 1-chloro-2-nitrobenzene
Making the orientation into para position need a thermodynamic control with a weak base (K2CO3), a polar solvent like NMP or DMSO, if necessary a PTC, and an appropriate reaction temperature. The cited reference indicates a very good selectivity into para position for a thermodynamic control.

Isoxazole preparation (8) (one-pot possible)
Most probably, by adding the hydroxylamine sulfate on the precedent mixture, the cyclization occurs. If the end of reaction is slightly “asthmatic”, the reaction could be driven to completion by water removal (anhydrous Na2SO4), which is also needed if it is chained with the next step.

At the end of the reaction, by removing inorganic solids (KHCO3/K2CO3/Na2SO4), the mixture is ready for the next step.

2nd nucleophilic aromatic substitution (9) (one-pot possible)
Most probably, by using the conditions mentioned in the original route, it should work. Also if a one-pot is considered, the ArSNX in cited references below is made in a polar solvent (MeTHF or tBuOH). Probably by using NMP with tBuOK as base, it works.

Reduction of NO2 (10)
The reduction could probably be done in a mix water / non-miscible organic solvent to prepare the next step by washing the organic layer by water to remove remaining dithionite and bisulfite, followed by an acid extraction with AcOH to avoid the N-Boc deprotection at a suitable molarity to prepare the next step.

Diazotation – Iodination (11) (“semi one-pot” possible if the previous workup is made)
To the aqueous mixture add NaNO2, adjust AcOH quantity for the diazotation. If the diazonium salt is unstable, an exchange could be done with TsO- or BF4- via a resin exchange (but add cost) and add KI to prepare the iodo derivatives. Since there is a N-Boc protective group, HCl or p-TsOH are not suitable.

As mentioned earlier, by making a Sandmeyer borylation with B2(OH)4, the route go back to the original designed one.

References for a draft procedure:
(This one should be taken with precautions...)

Sandmeyer with CuBr cat/KBr:

Sandmeyer borylation:

2-[(aminocarbonyl)amino] derivative (12)
Instead of using a toxic reagent, i am pretty sure that a treatment of 2-amino-3-cyano-thiophene with carbonyl diimidazole followed by addition of formamide afford the “protected” 2-[(aminocarbonyl)amino].

5-chloro-thiophene derivative preparation (13)
An interesting publication use NCS with FeCl­3 in acetonitrile for the chlorination. Since Fe(acac)3 or FeCl3 could be used in a catalyzed cross-coupling reaction in the next step, a feasibility test and solvent screening (with and without catalyst since there is potentially a dimerization) should be made to afford the use of the solution with a pre-treatment to remove the succinimide.

In the publication, thiophene was used with FeCl3 and only polymerization was observed. Since the 2-position is substitued, there is maybe a dimerization. Due to the substituent effects on the ring, it should be tested to see if there is chlorination or not in these conditions.

If doesn’t work, instead of using ammonium nitrate (which is slightly “petatoire” in french), use the procedure in the patent WO 2001058890 A1 in DMF.

Lastly, the standard procedure is NCS/AcOH.

If the choice is made to prepare the boronic derivative, NBS must be used.

For the boronic derivative:

Coupling reaction (14)

As mentioned earlier, there are several avenues:
1 – Coupling reaction with the aryl boronic derivative obtained from the diazonium salt
2 – Coupling reaction with the thiophenyl boronic derivative

The method for case 1 and 2 above is already developed in the publication, the third case is through a Grignard complex, followed by a cross coupling with FeCl3 or Fe(acac)3. This third case is purely exploratory and need to be tested.

The Grignard must be made on the aryl derivative. The substituent in ortho-position increase the electronic density on the ring which make the exchange more difficult (that’s why i have chosen I as halide), but with the EWG group on the thiophene derivative, these reagents configuration increase the probability of a coupling with an iron catalyst. Also, since the Grignard reagent is deactivated, maybe the homo-coupling is avoided like with R-Cu(CN)MgX.

An attention should be taken about the N-formyl group and the quantity of Grignard needed.

Also, the haloalkane from the exchange must be removed brfore the cross-coupling reaction by a distillation at a reduced pressure (if the stabilized Grignard is unstable above room temp), and at a constant volume with a co-solvent or the reaction solvent.

Hydrolysis and final deprotection (5)

Acid hydrolysis of nitrile to amide is a bit rough (concentrated H2SO4), but could afford the targeted molecule by N-Formyl removal, N-Boc removal and nitrile hydrolysis. The last step of N-Boc deprotection use IPA/aqueous HCl. This condition could also remove the N-formyl, CN hydrolysis must be tested over a longer reaction time.

Else, a basic hydrolysis could be used with Amberlyst A26 in IPA/water, followed by a filtration to remove the resin, and adding aqueous HCl and make the N-Formyl and N-Boc deprotection.

This is only a comparison of major starting material and reagent cost (solvent excluded) between the two routes (molbase prices):

Original synthesis
Trihalo benzene: 195 $/kg
Isoxazole: 1260 $/kg
n-hexyl lithium: 311 $/kg
Total: 1766 $/kg

Alternate route
1-Chloro-2-nitrobenzene: 27$/kg
1-Chloropentane-2,4-dione: 358 $/kg
Hydroxylamine: 25$/kg
Sodium Dithionite: 10$/kg
Sodium nitrite: < 10$/kg
Total: 430 $/kg

Maximum additional fixed charges for alternate route to 2 supplementary steps: 1336 $/kg and a minimum global yield of 57%

Lastly, the cryogenic step involving the use of n-hexyl lithium is avoided which is a health and safety concern, and there is a possibility to avoid the use of boronic derivatives which are now recognized as mutagen.

This is some personal works on paper only, i have no responsibility in any way if somebody would try this route and has all sort of troubles, including but not limited to: injuries and money loss. This is for experienced chemists only, and tests must be conducted in a suitable lab only.

But if my work is used to synthesize the targeted molecule described here, please, send a word, even if it fails, chemistry is always an experimental science. This will make me pleased, thank you.

© David Le Borgne, 2015, specialist in chemical process development and optimization.


  1. Update about (14) : add "Also, the haloalkane from the exchange must be removed before the cross-coupling reaction by a distillation at a reduced pressure (if the stabilized Grignard is unstable above room temp), and at a constant volume with a co-solvent or the reaction solvent."

  2. If some have this reference: Iron Catalysis in Organic Synthesis, Ingmar Bauer and Hans-Joachim Knölker*, Chem. Rev., 2015, 115 (9), pp 3170–3387, (DOI http://dx.doi.org/10.1021/cr500425u), please contact me through comment on this page or LinkedIn, i have no access to this publication, and the content seems to be VERY interesting, thank you per advance.