• 2018-07
  • 2019-04
  • 2019-05
  • 2019-06
  • 2019-07
  • UBA belongs to the non canonical E enzymes that lack


    UBA5 belongs to the non-canonical E1 enzymes that lack a defined Cys domain but have the active-site Cys within the adenylation domain [24]. Similar to other E1 enzymes, UBA5\'s adenylation domain comprises an eight-stranded beta sheet that is surrounded by helices [21]. Moreover, like the ancestral E1, MoeB, and the autophagy related E1, Atg7, the adenylation domain forms a homodimer with a pseudo-2-fold symmetry [25], [26], [27], [28]. Besides the adenylation domain, both UBA5 and Atg7 possess a sequence outside the adenylation domain that is required for ubiquitin-like protein\'s binding. In UBA5, this sequence comprises 13 amino acids, which are known as the UFM1-interacting sequence (UIS) and are located C-terminal to the adenylation domain [2], [29]. Previously, we and others have shown that the adenylation domain by itself is unable to activate UFM1, but a fragment possessing both the adenylation domain and the UIS successfully activates UFM1 [2], [30]. Interestingly, although UBA5 dimerization is critical for UFM1 activation, UBA5 is not a stable dimer in solution. However, binding of UFM1 to UBA5 possessing the adenylation domain and the UIS stabilizes the dimeric state, which is needed for ATP binding and thereby for UFM1 activation [31]. Ultimately, in order to transfer UFM1 to UFC1, UBA5 contains a short sequence at the C-terminus that is required for UFC1 binding [32]. Human UBA5 has two isoforms, each including the above-mentioned three regions: the adenylation domain, UIS and UFC1-binding sequence [1], [33]. However, only one has an extension of 56 Viomycin N-terminal to the adenylation domain (Fig. 1a). Previously, we and others have shown that the short isoform (without the first 56 amino acids) satisfies activation of UFM1 as well as transfer of UFM1 to UFC1 [21], [34]. This suggests that the 56 amino acids N-terminal to the adenylation domain are not essential for UBA5 function, therefore raising the question of their role in UBA5\'s function. Here we provide structural and biochemical data on the UBA5 N-terminus in order to understand its contribution to UFM1 activation. The crystal structures of the UBA5 long isoform bound to ATP with and without UFM1 show, for the first time, that the N-terminus not only is involved in ATP binding but also affects how the adenylation domain interacts with ATP. This leads to the ATP gamma-phosphate adopting a different position to that in the structure of the short isoform. Moreover, the active-site Cys, which in the short isoform resides on one of the adenylation domain helices, is now moved to the crossover loop. Surprisingly, in the presence of the N-terminus, the 1:2 ratio of ATP to UBA5 is not retained but becomes 1:1. Accordingly, the N-terminus significantly increases the affinity of ATP to UBA5, thereby facilitating UFM1 activation at low ATP concentrations. Finally, the N-terminus, although is not directly involved in UFC1 binding, stimulates transfer of UFM1 from UBA5 to UFC1. Taken together, our results provide the structural mechanism for the role of the UBA5 N-terminus in UFM1 activation.
    Results and Discussion
    Conclusions Here we have provided structural and biochemical insights into the contribution of the UBA5 N-terminal extension to UFM1 activation. Previously, we demonstrated that the short UBA5 isoform, lacking the N-terminal extension, executes UFM1\'s activation in a trans-binding mechanism. Now we have found that while this mechanism is preserved in the long isoform, it is actually improved as the N-terminal extension of one protomer interacts with the adenylation domain of the other protomer as well as with the bound ATP. This not only increases the affinity of UBA5 to ATP but also stabilizes the ATP-bound form of UBA5, thereby stimulating UFM1 activation. Interestingly, even at high ATP concentrations (Fig. 1), the long UBA5 isoform activates UFM1 faster than the short isoform, suggesting that the N-terminal extension contributes not only to the binding of ATP but also to the catalytic process. Indeed, the N-terminal extension stabilizes Arg115 of the adenylation domain, which possesses a catalytic role but makes no contribution to ATP binding (Fig. 6). Taken together, our results suggest that the presence of the N-terminal extension benefits the activation process and enables UFM1 activation under ATP concentrations that do not satisfy activation with the short isoform. Currently, little is known about how the UBA5 isoforms are regulated in the cell, and what cellular benefit accrues from the expression of the short isoform. Indeed, RNA expression analysis of UBA5 isoforms suggests that the long isoform is highly expressed compared to the short isoform [43]. The significant boost the N-terminal extension brings to UFM1 activation raises the question of whether it is regulated in the cell. Previously, it has been shown that Tyr53 of the N-terminus undergoes phosphorylation [44]. In our structure, Tyr53 is 2.5 Å away from Asp141 of the adenylation domain and forms hydrogen bonds with the latter. This therefore suggests that phosphorylation of Tyr53, which provides a negative charge on Tyr53, will prevent interaction with Asp141, thereby interfering with the contribution of the N-terminal extension to UFM1 activation. Overall, our results expand our understanding on UFM1 activation by UBA5 but at the same time call for further investigation of the regulation of UFM1 activation in the cell.