Notes on Terrible Reactions

There is a very informative discussion thread In the Pipeline today. It is about reactions that never work or are underwhelming in terms of yield and ease of workup/purification. So, to steal the subject, I would like to add few random notes:

PPA: A less viscous reagent alternative is called Eaton reagent, it is a P2O5 solution in methanesulfonic acid and you can buy it ready-made from Lancaster.

Azide reduction with phosphine: the iminophosphorane adduct is slow to hydrolyze in case of triphenyl phosphine, always use trimethylphosphine solution instead (stinky and expensive, but much faster, the phosphine is volatile and the phosphine oxide extracts into water).

Skraup: We called this kind of drastic conditions “Chemical Inquisition” in Prague. I guess it depends on the substrate. There are some high-yielding Sraups published in Org Process R&D journal, the process people typically use 3-nitrobenzenesulfonic acid as co-oxidant because the crap produced from the spent oxidant is soluble.

MnO2 oxidations: Like with silica or alumina, MnO2 is desactivated by moisture. You can re-activate it in glassware-drying oven, at 110-130C. (Works much faster after oven activation – but can be less selective. You can also add 4A powdered sieves). Non-polar solvents improve the reaction rate, sometimes doing the oxidation in cyclohexane instead of DCM is what is needed. Always use >20 equivalents to get to completion. BaMnO4 is a pretty good alternative to MnO2, and it is very easy to make (permanganate,  Ba salt and KI as a reducing agent)

Grignard: Initiating the reaction is the tricky part, people have used 1,2-dibromomethane, iodine, TMSCl – but for me the best working initiation technique is to place few equivalents of Mg turnings into an oven-dried flask with a large egg-shaped stirbar, flush it thoroughly with dry argon, add few drops of Br2 and dry-stir the Mg turnings in the Br2 vapors overnight. Then add freshly distilled ether solvent via canula (the bromine color disappears) and then carefully your substrate. I got some Grignards like BrMg(CH2)3MgBr by this technique that are hard to make by other methods (unless you want to mess with Rieke Mg).  The Mg turnings are fairly fragile and crushing them in oxygen-free and nitrogen-free environment uncovers a highly-reactive newly-formed surface which is further protected by MgBr2 formation. MgBr2 is soluble in ether. Please note that one has to use Ar because N2 reacts with fresh Mg surfaces, to produce dark Mg nitride.

Also, the Knochel transmetallation with iPrMgCl or iPrMgBr always worked great in my hands, with aryl iodides, at -20C.

Swern-like oxidation with SO3.pyridine in DMSO: the commercial SO3. pyridine often contains pyridine hydrogen sulfate impurity that can upset the reaction, always add few extra drops of pyridine and let the complex with DMSO form, before adding the alcohol.

Mitsunobu: The complex formation is pretty exothermic and overheated complex of DEAD with PPh3 decomposes quickly. Always start the reaction on ice bath. You can add phenol to phosphine before DEAD addition but secondary alcohols should not be premixed with the DEAD/PPh3 complex – a significant amount of eliminated product could form in absence of nucleophile. Use DIAD as a cheap and lazy DEAD alternative (the reactions take >2 times longer). You can use a slight excess of PPhe3 and kill the unreacted phosphine during workup, by adding few drops of 30% H2O2 (the oxidation is instantaneous). PPhe3O crystallizes quite nicely from a benzene-cyclohexane mixture so if your product is reasonably soluble you can dissolve the mix in benzene, dilute with cyclohexane and let it precipitate. The filtrates can be often applied onto a column directly, without evaporation.

DCC couplings: Acetonitrile or carbon tetrachloride are great solvents for DCC coupling, as dicyclohexyl urea is much less soluble there than in DCM or chloroform.  You can dissolve the crude product in a small volume of ethyl acetate and put it in the fridge, to precipitate the last bits of the urea.

Lazy silyl protections: TfOSiR3 + lutidine usually works marvels, but ClSiR3 with DBU as a base plus catalytic DMAP in acetonitrile is in my opinion much more general than imidazole in DMF.

Hydrogenations: Pt on charcoal (5%) is much faster than Pd-C for ArNO2 reduction and unlike Pd-C, with some care one can reduce chloro-nitro compounds without the ring dechlorination (0.05-0.1 wt ratio of 5%Pt-C to substrate, in ethyl acetate or ethanol, H2 baloon, RT, 1 hour). Also, don’t use MeOH as a solvent in hydrogenations – unless you like to play with fire.

Vilsmeyer formylation: Oxalyl chloride in DMF usualy works faster and cleaner than the more traditional POCl3 in DMF. Oxalyl chloride is added first, with cooling on ice (gas evolution!), followed by substrate.

Thioamide preparation: Lawesson reagent is not fun to work up, for many substrates P2S5 is a nicer alternative. 

Strecker aminonitrile hydrolysis to aminoacids: Aminonitriles tend to crap up during acidic hydrolysis to aminoacid, partial retro-Strecker happens. But if you formylate first with excess of formic-acetic mixed anhydride at RT without any base (96% formic acid is first mixed with an equal volume of acetic anhydride and allowed to sit under Ar for 4 hours, then the substrate is added) the acid hydrolysis of the formyl aminonitrile is very clean and N-formyl later falls off during HCl hydrolysis. This trick is was used by Vachal and Jacobsen in their tert-leucine paper. (I told them to try it.)

Initiating  periodnane oxidations: Dess-Martin periodnane sometimes needs initiation, typicialy when a nice, freshly-made reagent is used. People have been adding a drop of tert-butanol or even a trace of water, to get the reaction started. But the addition of a small amount of pyridine works even better and the added advantage is that pyridine protects highly acid-labile groups in the molecule, like TES-O or 1-ethoxyethyl, from being cleaved by the reagent.

SeO2 allylic oxidations: Working up stoechiometric SeO2 reaction is awful, there is a nice Sharpless procedure for using catalytic SeO2 oxidation (5mol%) with salicylic acid as a co-catalyst and anhydrous tBuOOH (2-3 eqivs) as co-oxidant. It works well at 40C in DCM (in 1 day), the only complication is that salicylic acid gets slowly eaten and has to be replenished, I found that tetrazole is a better co-catalyst that does not have this stability problem. (Anhydrous tBuOOH solutions are somewhat expensive on large scale, diluting the 70% aqueous solution of tBuOOH with dichloromethane, saturating the aqueous layer with MgSO4 and extracting with dichloromethane and drying the extract with MgSO4 is easy way to make your own anhydrous solution).
______________________________________________________________________________________ 

(4S)-(4,5-dihydro-4-tert-butyl-2-oxazolyl) ferrocene and (2S)-N-(1-hydroxy-3,3-dimethylbutyl) ferrocenamide

(2S)-N-(1-hydroxy-3,3-dimethylbutyl) ferrocenamide

Ferrocene monocarboxylic acid (16.1 g, 70 mmol, Strem) was suspended in dichloromethane (340 mL) in a 3-neck flask fitted with a nitrogen inlet, septum, and an outgas tube bubbling into a 1L flask Erlenmeyer flask filled with 500 mL of 1 M NaOH. A small amount of DMF (~0.05 mL) was added to the suspension and oxalyl chloride (8.1 mL, 92 mmol) was added to the suspension dropwise over 5 min. The reaction became homogeneous as it was stirred for 1 h. The resulting solution was concentrated and the residue dissolved in dichloromethane (150 mL). The red-orange solution was transferred by cannula to a solution of L-tert-leucinol (8.50 g, 72.5 mmol) and triethylamine (31.7 mL, 220 mmol) in dichloromethane (370 mL) maintained at 0 °C. The reaction mixture was allowed to warm to rt over 3 h and ether (300 mL) was added. The reaction mixture was washed with a solution of NaOH (500 mL, 1M aqueous), dried over MgSO4, filtered through a pad of Celite and concentrated. The orange-brown residue was dissolved in boiling dichloromethane (300 mL) and the solution cooled to –20 °C and maintained for 12 h. The formed orange crystals were collected by vacuum filtration (17.85 g, 54 mmol, 77%). The filtrate was concentrated. The residue was dissolved in boiling dichloromethane (30 mL) and the solution was cooled to –20 °C and maintained for 12 h. A second crop of orange crystals was collected by vacuum filtration (4.94 g, 15 mmol, 21%). The combined crops of crystals (22.55 g, 68.6 mmol, 98%) were identical in all respects to the material prepared by alternative methods. (Ref 1)

Note: Clean your rotovap from any residual HCl after the first concentration. Recommend aspirator vs. diaphragm pump for the first concentration

(4S)-(4,5-dihydro-4-tert-butyl-2-oxazolyl) ferrocene

Methanesulfonyl chloride (MsCl, 8.3 mL, 100 mmol) was added to a solution of (2S)-N-(1-hydroxy-3,3-dimethylbutyl) ferrocenamide (22.55 g, 68.5 mmol) and triethylamine 30 mL (200 mmol) in dichloromethane 0.8L maintained at 0C. The reaction was allowed to warm to RT over 12 h and a solution of saturated aqueous NaHCO3 (0.5L) was added to the reaction mixture. The phases were separated and the aqueous portion was extracted with dichloromethane (2×300 mL) and the combined organic portions were dried over MgSO4 and filtered through a pad of Celite. The filtrate was concentrated and the resulting residue was taken up in boiling hexanes (200 mL) and the solution was decanted, leaving behind a small amount of insoluble brown residue. The resulting homogeneous solution was cooled to –20C and maintained at that temperature for 12 h. The orange needles that formed were collected by vacuum filtration and washed with hexanes (10 mL) to provide the title compound (18.49 g, 59.5 mmol, 87%). The filtrated was concentrated and the resulting residue dissolved in boiling hexanes (20 mL). The solution was cooled to –20 C for 12 h and a second crop of orange needles was collected as above (1.90 g, 6.1 mmol, 9%). The combined crops of needles (20.39 g, 65.6 mmol, 96%) were identical in all respects to the material prepared by alternative methods. (Ref 1)

Ref 1: Sammakia, T.; Latham, H. A.; Schaad, D. R. J. Org. Chem. 1995, 60, 10–11.

__________________________________________________________

Org Prep Daily has a new contributor

I am delighted to announce that Chris Douglas submitted his improved procedure for making a ferrocene oxazolidine ligand to Org Prep Daily.

Thank you Chris – and best luck with your new group in Minnesota!

Milkshake Manifesto

or The Twelve Theses For All Erudite Medicinal Men To Follow, To Purge The Sins Of Abominable Design And To Attain The Heavenly Joys of a Clinical Success

1. Do one change at a time. Keep your molecule small and simple. Cyclize. Fluorine is your friend.

If you make too many intuitive jumps in your structure without knowing the contribution of each change, you will not understand what did what. Your SAR table will become a jungle and you will miss important things.  If your compounds become hard to make early in the project, your progress will be slow. It is easy to gain potency by adding a big greasy piece like biphenyl or bromo-substituted benzyl but this back-fires in the longer run. Cyclic compounds have often better properties (cell permeation, PK) , the potency and selectivity can also dramatically improve by restricting the conformations. Replacing hydrogens with fluorines helps to fix the metabolism problem. One can gain additional potency with correctly placed fluorine with only a small penalty on the molecule lipophilicity and size. (Fluorine is also a wonderful NMR label that can be used to follow the chemistry and quickly determine the isomeric purity of compounds).

2. God created sharp SARs. 

Don’t despair if the permissible substitution pattern of your active compounds is very narrow – it is a good thing. Lousy and un-druggable molecules are usually from the muddy category (no matter what you change the molecules have always about the same potency). 

3. Lipinski rules and Polar surface area scores are only a shorthand. 

Lipinski rules and PSA scores are fairly crude but they point in the right direction – Don’t put too many hydrogen bond donors in your molecule if you want a cell permeable orally-active compound. More than 3 Amide/urea NH is bad news. Too many amide and heterocycle NHs together with good hydrogen bond acceptors (carbonyls, hetero-N) in the same flat polyaromatic molecule will likely result in lousy solubility – this is the typical problem with kinase inhibitors. (A small substituent like Cl, Me group next to NH can remedy the solubility/permeability problem). Be careful about sulfonamides and sulfones – these groups are quite polar and can cause problems with cell/brain penetration. Please note that the requirements are more stringent for brain-penetrable compounds. 

4. Acidic compounds have high plasma protein binding

If your molecule has acidic groups like carboxyl, tetrazole, isothiazolindione etc, your intrinsic activity has to be very high (single digit nM or better in cells) to ofset the extremely-high plasma protein binding of acidic compound. Greasy molecules are  problematic also – but acidic compounds are almost guaranteed to have 99+ % plasma protein binding and hence reduced relative potency in the whole blood assay. 

5. Basic molecules have typically improved solubility and cell potency – at a price

More basic amines (primary, secondary) have more pronounced accumulation within cell than less-basic amines (such as N-sunstituted morpholines). Adding amine side-chain to a molecule can also cause lowered selectivity and unexpected organ toxicity. Quartenary ammoniums tend to produce particularly nasty side-effects and should be avoided. Tertiary amines are metabolised by oxidative dealkylation quickly. (Putting fluorine or oxygen beta to amine can suppress the oxidative metabolism and at the same time lower the amine basicity.) Strongly basic molecules have typically very high volume of distribution – which means that due to partition of the compound deep into tissue the dosing has to be increased to achieve decent concentration in the blood – but at the same time the high volume distribution prolongs the half-life of the compound, even if the compound has rather high clearance due to rapid metabolism.

6. Amidines and guanidines are frustrating

It is easy to get an active compound with amidine but it is hard to get it orally active afterwards. Doing chemistry with amidine in the molecule is not fun either.  Other not-so-druggable functional groups are hydroxamic acids, aldehydes, thiols, reactive halogens. Nitro groups are a liability but they are not as bad as many people think – though you may want to try to replace a nitro group in the end. 

7. Test CYPs 

Basic heterocycles and polyaromatic compounds have problem with inhibition or induction of cytochomes. This is undesirable because of the drug-drug interaction liabilities. It is important to test for CYPs inhibition relatively early in the project, especially if the molecule has troublesome pieces like 4-monosubstituted pyridine, 5-azaindole, 1,2,4-triazole, N-monosubstituted imidazole, etc.

8. Do not trust docking, do crystallography early.

Docking can provide some idea in absence of a real information (co-crystal X-ray structure) but most of time the docking amounts to a wishful thinking in front of the monitor. A co-crystal (or at least soaked-crystal) X-ray crystallographic data can be very illuminating. Use the crystallography as a source of ideas but don’t let it rule your design; X-ray data does not address electronic and entropic effects well, and it over-emphasizes hydrogen bonds and steric factors; you may easily get sidetracked by focusing on marginal interactions while overlooking some other important effects. Over-reliance on X-ray in finding additional interactions usually produces ugly compounds that have lots of floppy polar appendages and are hard to make. 

9. Get cell-based assay as early as possible to guide your project.

Biochemical data tels you about hidden potential of your compound but cell-based activity is the real thing. You don’t have a real lead compound until you can make it work in cell. Usualy everything that works in cell works also in biochemical assay but this is not true in the opposite direction – things will or won’t work in cell for mysterious reasons and only persistence and patience sometimes  helps to find the cell-active version of your molecule. Once you got a decent activity (say, 100nM in biochemical assay) it is important to have cell-based assay as the primary screen; it makes no sense to overoptimise on biochemical assay alone because without cell-based assay you will likely stray in wrong direction.

10. Ignore Caco-2 and do rodent PK tests instead, use human plasma and whole blood

Caco-2 permeability model is useless. Oral absorbtion/brain penetration tests in rodent should be done early in the project. Metabolic studies with liver microsomes can help to identify some metabolically weak spots (use LCMS to identify the metabolites) but microsome studies are not terribly helpful in prioritising the molecules – it is foolish to say “I have to have microsome half-time above 30 minutes.” (If all your molecules have a decent microsomal half-life and suddenly one has a very short one, maybe this one molecule can have a liability towards oxidative metabolism – but this suspicion cannot be taken as a substitute for doing a real PK in rat). Do plasma protein-binding studies with human plasma only – the calf-derived plasma is a poor substitute, often producing misleading results. Do a whole blood assay with human blood. If your advanced compounds are highly crystalline, measure the solubilities and prioritize based on solubility studies

11. Be nice but sceptical 

Be nice to your biologists but watch about the reliability of their data – if they got a weird curve or noisy assay they should be forthcoming about it. Don’t give the biologists the entire stock of your compound – you may not get it back.
Don’t force people to work on your ideas – let your lab colleagues choose what part of project they work on, which particular molecules to make. Make the project more enjoyable for them by scaling up the building blocks for the entire lab and by  ordering the labor-intensive advanced intermediates from a custom-synthesis company.

12. Patents are not scary 

Don’t worry over too much about patents – unless the exact compounds is in the patent experimental examples (with a good preparative procedure and NMR spectra) and is claimed there for the same protein. Very broad patent claims are easy to cast but they are also easy to challenge or skirt; it is almost always possible to get your own narrow patent on things that are within the scope of someone’s patent by describing “new and unexpected” properties of a class of compounds (that were claimed but not documented in the broad patent). 
Getting out of patent is much easier than many chemists think – look what Sepracor did, and how many closely-related me-too compounds are there on the market, from the competing companies.  Also, the patent-busting changes to the molecule can be quite minimal and can be made very late in the project.  It is good to get a broad patent coverage for your project – but if you cannot, all what one needs in the end to have patent-protected is just one compound, the one which eventually becomes a drug.

 Credit: Adolf Born

Note: Totally Medicinal has a related discussion on Tricks of trade in medicinal chemistry.

Update: Derek Lowe has written many times on these subjects In the Pipeline and his take is always insightful. Although there is one idea on which we disagree – Derek claims that “nothing good ever came easy or cheap”. I would like to point out that my mother-in-law did pass away; and I did not have to hire anyone

4-(2′,4′-difluorophenylamino)-piperidine dihydrochloride

1-Cbz-4-piperidone 4.012g (17.2 mmol) with 4A powdered activated molecular sieves 9.1g and 2,4-difluoroaniline 2.582g (20 mmol) was suspended in anhydrous 1,2-dichloroethane (150mL) and acetic acid 6mL was added. The mixture was stirred under Ar overnight (11h30 min). Solid sodium triacetoxyborohydride 12.72g (60mmol) was added and the reaction was continued at RT over an extended weekend (3 days). The reaction mixture was then filtered, the solids were washed with ethyl acetate 350mL. The combined filtrates were washed with saturated NaHCO3 250mL and then with half-saturated NaHCO3 (250mL), the aqueous phases were re-extracted with ethyl acetate (250mL). The combined organic extracts were dried (MgSO4) and evaporated. The residue was dried on highvac to remove some of the difluoroaniline,  then purified on a column of silica (190g) in a 3:1 mixture hexane-EtOAc.

The obtained Cbz-protected piperidine (pink syrup, 3.8g, 64%Y) was hydrogenated under baloon of H2 in ethanol 200mL in presence of Pd-C(10%),  2.5g and 4M HCl in dioxane 5mL for 19 hours (overnight). The catalyst was removed by filtration through a pad of Celite, the filtrated were evaporated and the residue dried on highvac. Y=3.075g (62.5% overall) of a white hygroscopic solid.  LC/MS(+ESI): 213 (M+1)

1H(d6-DMSO, 400MHz): 9.5-7.5(very br s, 2H), 9.272(br m, 1H), 8.975(br m, 1H), 7.158(m, 1H), 6.940(m, 2H), 3.547(m, 1H), 3.268(br d, 13.0Hz, 2H), 2.930(br q, 10.4Hz, 2H), 2.008(br d, 11.5Hz, 2H), 1.703(m, 2H)