In addition to PDCA we also tested
In addition to 2,4-PDCA, we also tested other broad-spectrum 2OG oxygenase inhibitors as vCPH inhibitors. N-Oxalyl glycine (NOG, IC50 = 21.2 ± 2.1 µM), a close 2OG isostere, and IOX1 (IC50 = 47.2 ± 2.7 µM), which is a relatively broad-spectrum JmjC demethylase inhibitor, were not as potent in inhibiting vCPH as 2,4-PDCA (IC50 = 5.3 ± 0.3 µM), nor the tested ‘selective’ PHD inhibitors.
We then investigated whether selected inhibitors compete with 2OG for binding to the active site, as shown in some cases for PHD2 and C-P4H.1, 3 As implied by Carr-Purcell-Meiboom-Gill (CPMG) NMR analyses, 2OG does not bind to apo-vCPH in the absence of metal (within detection limits; Supplementary Fig. 14). The 2OG NMR signal in solution is depleted on addition of Ni(II), Zn(II) or Mn(II) to vCPH, consistent with the binding of 2OG, as reported for some other 2OG oxygenases.1, 3, 38 Addition of NOG displaces 2OG from vCPH, and restores the 2OG resonance; at higher metal concentrations, addition of NOG did not restore the resonance, likely due to chelation of free Mn(II) by 2OG and/or NOG (Supplementary Fig. 14B). Upon addition of 2,4-PDCA, Roxadustat or Vadadustat, the 2OG resonances were restored, indicating that u73122 2OG is displaced on addition of these inhibitors. These observations indicate that, consistent with the expected binding modes of these compounds,1, 3 NOG, 2,4-PDCA, Roxadustat and Vadadustat compete with 2OG (Supplementary Fig. 14C).
The MALDI-TOF MS assay was then used to determine Ki values for selected inhibitors (Supplementary Fig. 15). The rank order of the Ki values was reflective of the IC50 values; of the compounds tested, NOG (Ki = 48.74 ± 15.96 µM) and Molidustat (Ki = 16.05 ± 7.40 µM) were found to be the weakest inhibitors of vCPH, while the most potent inhibitors were Roxadustat (Ki = 1.07 ± 0.11 µM), FG-2216 (Ki = 0.71 ± 0.11 µM) and Vadadustat (Ki = 0.28 ± 0.06 µM). Daprodustat (Ki = 7.65 ± 2.05 µM), similar to the order observed when ranking IC50 values, has weaker binding that FG-2216, Roxadustat and 2,4-PDCA, but exhibits stronger vCPH binding than NOG and Daprodustat.
Since certain TCA cycle intermediates are proposed to inhibit other related 2OG oxygenases in a pathophysiologically relevant manner, we tested them as vCPH inhibitors. None of these compounds were potent vCPH inhibitors, with the most active being the metal chelators citrate (IC50 = 212.6 ± 7.1 µM) and d-isocitrate (IC50 = 284.0 ± 26.8 µM) (Supplementary Fig. 16).
Conclusions The assay development work led to the identification of conditions suitable for screening potential vCPH inhibitors by MS, by monitoring a single hydroxylation of a relatively short peptide. A notable observation arising from this work was that the ascorbate dependence varies for different hydroxylation sites (Supplementary Fig. 9). This may be of relevance in understanding the effect of ascorbate on 2OG oxygenase catalysis during collagen biosynthesis, and in the development of improved assays for the α2β2 human C-P4Hs. While our preferred assay method employs MS,3, 28 the conditions should be broadly comparable to other assay methods developed for 2OG oxygenases.1, 3, 28 The optimised assay was used to investigate the inhibition of vCPH by known and potential 2OG oxygenase inhibitors. The HIF PHD inhibitors currently in clinical trials all inhibited vCPH, with the greatest potency observed for Vadadustat (Table 5). Although care should be taken in comparing inhibition results using different assay methods, the results imply that the development of highly selective inhibitors for a particular class of prolyl hydroxylase should be possible. For example, whilst the structurally related inhibitors IOX4 and Molidustat both potently inhibit PHD2, IOX4 is a more potent inhibitor of vCPH. Similarly, differences were observed between the structurally related inhibitors Roxadustat and Vadadustat, with the latter being the more potent vCPH inhibitor.