Orally active drugs need to dissolve in water or in lipids. When the compounds are crashing out even from straight ethyl acetate one should urgently look for better-behaved analogues before going too far down the SAR potency/selectivity road. One should not hope to fix the solubility later - there is only so much that the solubilizing side-chains can do. And when a poorly-soluble polycyclic molecule is decorated with amines a mirage of improved solubility and potency in cell-based assay may appear but very little of it usually goes towards a sound SAR. (Also the PK of such series is most likely going to be awful).
Compounds generally appear more soluble when prepared in form of a lyophilisate. Scaling-up a poorly-soluble clinical candidate that was never before prepared in a pure crystalline form is bound to produce unpleasant surprises…
Developing poorly soluble compounds is a frequent source of frustration in kinase projects; the kinase ligand-binding sites naturally like polyheterocyclic compounds. Nowadays many medchem programs start with the high-throughput screening of chemical libraries - the lead compounds obtained from combichem/parallel-synthesis are notorious for the high number of NH bonds. Also, over-optimistic crystallography-driven drug design can lead to terrible molecules: X-ray co-crystal picture offers the illusion of a great insight (“we should try to pick a H-bond here and reach a pocket back there”) and it encourages decorating the molecule with too large greasy or excessively-polar substituents - the fast way into potent undruggable series.
When trying to fix a poor compound solubility:
1. Abandon the series and work on something else – the most reliable option. One should consider this seriously when the compounds are so terribly insoluble that they form gels.
2. Get rid of some NH bonds - alkylate the nitrogens, replace them with C or O, etc. Tertiary amides are less troublesome than primary, secondary amides or sulfonamides. Stay away from N,N’-diaryl ureas. Amine NHs are fine.
3. If you must have the NH bonds in your molecule, put some groups right next to the NH, to shield it - in ortho or alpha positions. An ortho methyl substituent can make all the difference.
4. Get rid of some hydrogen bond acceptors – remove sulfonyls, carbonyls etc.
5. Avoid a combination of el deficient with el rich aromatic rings within the same molecule – the charge transfer between electronically unequal partners produces exceptionally strong pi-stacking. (Picric acid forms a stable co-crystal with naphtalene)
6. Mess up the symmetry: Avoid big symmetrical substituents like p-Br-phenyl, try to add some ortho substituents to keep rings out of plane
7. Add solubilizing side-chains containing tertiary (or secondary) amines or try to pro-drug
8. Disconnect rings or replace aromatic rings with non-aromatic ones. Polycyclic aromatic systems with more than 2 rings are Very Bad
9. Reduce the size of the molecule
Purifying insoluble compounds is hard, especially when some structurally-related impurity carries over into the solids. You cannot do a silica column and the reverse-phase prep HPLC purification is no picnic either (you inject your material in DMF solution - then pray it wont crash out in the injection loop, autoinjector needle, connecting tubing or on the top of the very expensive column). If you must make poorly-soluble compounds, it is best to rigorously purify the intermediates right up to the moment when the material becomes very insoluble (such as the deprotection or coupling step) and spend enough time optimising the reaction conditions and workup in the following steps so that the impurities are kept at low levels from then on.