Thirty six of the analogs synthesized
Thirty-six of the analogs synthesized showed ATX inhibition greater than 50% at 10μM with FS-3, twelve of which had improved potency over the lead (Fig. 5). Of these, 22 was sub-micromolar (IC50 670nM). The 3,5-dichlorophenoxy ring of 22 is reminiscent of the 3,5-dichlorophenylthiourea ring in Hoeglund 5 (IC50 1.6μM, Fig. 6), which was previously shown to be a competitive inhibitor. Like 22, PF-8380 (IC50 2.8nM, Fig. 6) also has a 3,5-dichloro substituted aromatic ring, but is attached through a methyl 1-piperazinecarboxylate. Although 22 is similar to both Hoeglund 5 and PF-8380, neither of those scaffolds have a substituted phenoxy ring, which may lend to the activity of 22. A structure–activity relationship study of PF-8380 also showed the importance of a 3,5-dichloroaromatic ring, but this study also did not use a substituted phenoxy ring. Only one of the analogs with the nitro group ortho to the piperazine ring (11, IC50 3μM) and two ortho carboxy analogs (42, IC50 1.8μM and 44, IC50 2.2μM) showed improved potency over the lead, indicating ortho substitution on the phenyl piperazine ring may not lead to further improvement. A similar trend was seen with boronic Q-VD(OMe)-OPh substitution on an aromatic ring by Albers et al. (Fig. 6), where Albers 74 (IC50>5μM) was far less potent than Albers 72 or 73 (IC50 28nM and 5.7nM, respectively) when monitoring choline release from the natural substrate, LPC.
In general, using a shorter linker than the one in GRI392104 may help limit off-target effects by reducing flexibility and does not seem to have detrimental effects on ATX inhibition (32, IC50 3.7μM is quite similar to GRI392104, IC50 4μM). It is also of note that only two of the fluoro-containing compounds (42 and 50, IC50 2.2μM) improved potency over the lead. Of the fluoro-containing analogs, both 42 and 50 contain trifluoromethyl-substituted phenoxy rings in the para and meta positions, respectively. ortho substitution on the phenoxy ring decreased potency (Table 3, Table 4), following a similar pattern noticed by both Albers et al. and Hoeglund et al. during optimization of disparate lead compounds (Fig. 7).51, 53 All but five of the twelve compounds with improved activity over GRI 392104 had dichloro substitution on the phenoxy ring. This may suggest that, like PF8380 and Hoeglund 5, having a dichloro substitution may improve potency, as also seen during structure–activity relationship studies of PF8380. Across the 66 compounds tested, minor changes in structure caused vast differences in biological activity. These drastic changes are activity cliffs, as described by both Dimova et al. and Hu et al.58, 59Figure 7 shows a generic structure of these compounds, highlighting similarities in PF8380 and Hoeglund 5. Based on results presented here and by extrapolation to related work of others we proposes that optimization of the polar head group of this scaffold can serve as a targeting handle whereas changes to the non-polar aromatic moiety can be implemented to improve pharmacokinetics and biological stability—although the 3,5-dichloro moiety is preferred.
Although ATX activity is reduced by these inhibitors, they are unlikely to be used as the lone treatment against ATX-associated diseases without further study in more complex systems. These compounds may be useful by causing inhibitor-mediated decreases in ATX activity, which would sensitize tumor cells to radiotherapy and apoptosis caused by treatments such as Taxol. It is important to explore several different scaffolds when developing targeted inhibitors because of the high attrition rate moving hit compounds through the development pipeline to approved drugs.62, 63 By diversifying the number of ATX inhibitors, there is greater hope of successfully treating ATX-related diseases.
Conclusion Here we describe a novel ATX inhibitor scaffold that was synthetically modified to improve potency into the sub-micromolar range. This study also demonstrates that shortening the carbon linker on this scaffold has no apparent detrimental effect and, in fact, may be of more use for further development and modification to produce an orally bioavailable drug. While flexible compounds may be useful for binding to ATX, they may also adopt additional conformations leading to off-target effects. Out of the sixty-six new analogs synthesized, twelve showed improved potency with one of those improving potency into the sub-micromolar range (22, IC50 670nM vs GRI 392104, IC50 4μM). Future endeavors are ongoing to further optimize this lead and to characterize both pharmacokinetic and pharmacodynamic effects.