Compounds were synthesized using a facile step convergent sy
Compounds were synthesized using a facile 8-step convergent synthesis ( and ). Briefly, (Boc)cyclen (compound ) was coupled using TBTU to Fmoc- or Cbz-protected amino alkyl acids (–), followed by deprotection using 20% piperidine in DMF, or H on Pd/C, respectively (–) (). Deprotected linkers were coupled using TBTU to an acetal-protected adenosine 5′ Primidone () (), which was formed by TEMPO-mediated oxidation of the 5′ hydroxyl of 2′,3′-acetal protected adenosine () (). Global deprotection of the resulting Boc- and acetal-protected ligands (–) was achieved using a TFA/HO mixture to afford –, followed by treatment with Zn(OTf) to yield the final metal-coordinated complexes (–) (). First, the activity of compounds – was evaluated against UBA5 by measuring the extent of UFM1 transfer to UFC1 in an in vitro transthiolation assay (described in ) ( and ). A modest structure–activity relationship (SAR) was observed on UBA5 inhibition, with respect to the alkyl linker chain length, when evaluating the small compound library in vitro. The extension of the linker from 1 to 5 methylene groups increased inhibitor efficacy as demonstrated by the improved activity of (IC=4.0μM, 95% Confidence Interval (C.I.)=1.74–9.34μM) compared to that of (IC=10.5μM, 95% C.I.=4.23–25.88μM) (). Interestingly, we observed a drop in activity with 3 methylene groups as in (IC=23.3μM, 95% C.I.=7.29–74.36μM), compared with (). There was only a slight improvement in activity by an inhibitor with the 5- () compared to 4- methylene chain (, IC=5.2μM, 95% C.I.=0.88–31.03μM), and due to the insignificant increase in activity, chain lengths beyond 5 methylene groups were not investigated (). Although increasing chain length does not appear to drastically affect inhibitory activity, it is possible that longer chain lengths may contribute to improved potency by optimizing the distance between the adenosine and zinc(II)cyclen functionalities and/or the interactions with their respective binding sites on UBA5. Importantly, removing the Zn metal abolished all inhibitory activity (, and data not shown). From these observations, we identified (A) as the most potent UBA5 inhibitor developed to date. Having derived a lead compound, was modified in order to investigate which structural components were essential for inhibitory activity. First, the importance of the metal for activity was measured by replacing Zn with Cu (), which significantly reduced inhibitor activity (IC >100μM, and ), while Mn, Ni, and Fe complexes demonstrated no activity up to 100μM (). It was concluded that Zn was required for optimal UBA5 inhibition, which indicated that the identity of the metal, and not simply the formal divalent charge, mediated activity. Furthermore, zinc(II)cyclen alone (uncoupled to adenosine) resulted in minimal activity at 50μM, suggesting that the adenosine portion of is crucial for UBA5 inhibition (). Together, these data support the current strategy of targeting the ATP pocket and surrounding acidic residues as a novel approach to UBA5 inhibition. To evaluate the selectivity of for UBA5 over other E1 enzymes, the inhibition of the specific Ub/Ubl transthiolation assays (as described previously, and in ), demonstrated ∼20-fold selectivity for UBA5 over both UAE (IC=78.5μM, 95% C.I.=51.2–120.5μM) and NAE (IC=66.8μM, 95% C.I.=31.4–141.8μM) (). Due to the observed selectivity among these E1 enzymes and given the rationale in targeting the ATP pocket, the inhibition of UBA5 transthiolation was quantified kinetically in the presence of increasing concentrations of ATP. Under normal transthiolation conditions, a substrate inhibition profile was observed for ATP, where at high concentrations of ATP there was a decline in the observed transthiolation (1.90±0.10pmolmin) with an associated of 16.5±2.5μM for ATP and a of 0.218±0.010min (, for Lineweaver–Burk plot see ). Interestingly, in the presence of 5μM , a noncompetitive inhibition profile was observed (), as supported by a decrease in the (0.113±0.016min) and (0.988±0.137pmolmin) with no significant change in the for ATP (24.9±9.6μM) compared with untreated conditions. Unlike the canonical E1 enzymes, UBA5 is an asymmetric homo-dimer, with one monomer mediating UFM1 activation while the other monomer may act as a regulatory, ‘inactive’ subunit. The kinetic data suggests that might not be displacing ATP from the active ATP pocket, and may instead elicit its activity by interacting with the inactive (regulatory) UBA5 monomer. Alternatively, might bind an allosteric site on the UBA5 homodimer. The detailed mechanism by which inhibits UBA5 is currently under investigation.