Assessment of Potential Off-target Activity of 4b
In addition to the HDAC profiling performed, we also queried 4b for any apparent potential off-target effects, which could complicate any finding of efficacy or adverse events in vivo due to `polypharmacy’. This was accomplished by screening against a `diversity panel’ of 72 binding and 29 enzyme assays, comprised of roughly equal numbers of selective central and peripheral therapeutically relevant targets (Cerep Diversity Profile). This panel included representative targets from diverse enzyme families, G protein coupled receptors, steroid nuclear receptors,
Figure 1. Biochemical and Cellular HDAC inhibition by 4b. (A) % inhibition of human recombinant Class I enzymes HDAC1 (red), HDAC2 (green), HDAC3 (black) and HDAC8 (blue) by 4b. (B) No inhibition of ClassIIa/b enzymes by 4b; HDAC 4(green), HDAC5 (red), HDAC7 (purple), HDAC9 (orange), HDAC6 (brown). (C) Time-dependence of human recombinant HDAC3 inhibition by varying preincubation time of 4b with enzyme (as shown). (D) Cellular inhibition of endogenous Class I HDACs/HDAC6 using Boc_Lys_Ac (black traces) or Class IIa/HDAC8 HDACs using Boc_Lys_TFA substrate (red traces) by 4b (closed circles) or reference compounds SAHA and Compound 26. (E) Time- dependence of cellular Class I HDAC inhibition by varying preincubation of 4b with cells (as shown). (F) Plot of IC50 values versus compound-cell preincubation time for SAHA (green) and 4b (black). voltage- and ligand-gated ion channels. While not totally inclusive, it is a useful tool for discovery compounds to assess their viability as proof of concept molecules to specifically query a particular target. 4b was initially screened at 10 mM in duplicate. The full set of results is provided as Table S1. 4b showed a `clean’ profile, with only 1 target, the CCK1 receptor showing 72% displacement of ligand binding on addition of 10 mM 4b. A follow up concentration-effect curve gave a Ki of 8.6 mM for 4b against CCK1 binding (Fig. S1). This clean in vitro profile was considered acceptable to proceed to ADME and pharmacokinetic studies.
Solubility and Physicochemical Stability of Free Base and Salt Forms of 4b
Compound 4b prepared as the free base proved to be highly insoluble in water at neutral pH. We next attempted to formulate 4b according to the given literature procedure [30], which had been adapted from a similar procedure previously used in the preparation of SAHA [22]. This involved the complexing of 4b to 2-hydroxypropyl-b-cyclodextrin in a 1:5 molar ratio to aid aqueous solubility. Under the prescribed formulation procedure targeting a 1 mg/mL solution, we found that 4b as a free base did not dissolve under these conditions even on heating, but instead formed a suspension. LC-MS analysis of this suspension on day 1 and after 7 days indicated no degradation of 4b when prepared either with or without heating. Filtering the suspension and quantifying the filtrate showed that only 0.1 mg/ml of 4b had dissolved. Thomas and colleagues [30] reported that a solution of 4b at 1 mg/mL was previously obtained using this procedure. A further examination of the reference cited for 4b preparation by this group showed that the 4b used was purified by preparative HPLC with trifluoracetate (TFA) in the eluent [5] suggesting that the authors had most likely isolated and used the TFA salt in their studies. We therefore prepared both TFA and hydrochloride (HCl) salts of 4b to establish whether they could offer improved solubility. Neither salt form of 4b was particularly soluble in water (,1 mg/ml). When prepared according to the prescribed formulation procedures, the TFA salt of 4b did not dissolve, even after heating. Analysis of the filtrate after filtering the suspension prepared without heating showed that only 0.3 mg/mL had fully dissolved. A study of the stability of the TFA salt also raised another more worrying issue in the use of 2-hydroxypropyl-bcyclodextrin�complexed 4b salts for potential in vivo studies. We observed that both salt forms of 4b cyclised slowly when left in solution (eg. aqueous acetonitrile) to a benzimidazole product at ambient temperature and more rapidly when heated (see Fig. 2 schema). The 4b. TFA cyclodextrin suspension targeted at 1 mg/ ml with heating gave an initial (day 1) measured value of 0.75 mg/ ml total 4b, with only 0.27 mg/mL solubilised as quantified in the sample after filtration. The lower than expected initial quantification of 4b was most likely due to the TFA salt rapidly degrading upon heating. Indeed, we always found the presence of the cyclised benzimidazole conversion product, named throughout as product C1, in the salt formulations even at t = 0 after preparation. Furthermore, the suspension was not stable upon standing in the dark with gentle shaking at room temperature, with a further 28% decrease observed in parent (4b) concentration on day 7 analysis, compared to day 1. Importantly, we anticipate that the conversion of benzamide 4b to the benzimidazole C1 is not reversible under physiological conditions and we have no experimental evidence indicating that C1 could be converted back to 4b. In order to assess the benzimidazole conversion product that we detected, we prepared a pure sample. This purified product C1 was readily soluble in aqueous media.
However, profiling againstthe biochemical and cellular HDAC assays determined that C1 retained no activity against any HDACs, or in the cellular Boc_Lys_Ac assay, at concentrations up to 50 mM, as would be expected due to the cyclisation of the `warhead’ benzamide of the compound, which is needed to chelate to the zinc moiety in the HDAC catalytic site. It did however retain the ability to displace CCK1 agonist radioligand binding in the Cerep assay (IC50 = 6.1 mM, Fig. S1). In addition, a further profiling of C1 at 10 mM against the Diversity Profile (Table S1) also showed significant additional displacement of antagonist radioligand binding for the I2 receptor (73% displacement). In conclusion, the insolubility and instability of 4b. TFA in aqueous solutions, which is exacerbated by heating as occurred during the formulation procedure previously used, casts grave doubt on the final 4b/C1 ratio administered to mice during the chronic oral drinking water study previously performed in the Thomas et al study [30], where the formulation in drinking water bottles were also reported to be changed weekly. Indeed, achieving a 1 mg/mL solution at all with 4b in this formulation would indicate that the resultant dosing solution may have been converted to predominantly the `inactive’ and readily soluble C1 product.