AE following a reported procedure. To verify the activity of PCP tagged NAE, we labeled the PCP domain of the fusion protein with biotin by Sfp phosphopantetheinyl transferase that conjugated the biotin-phosphopantetheinyl group derived from biotin-CoA to the Ser residue of PCP. The biotin-labeled PCP-NAE fusion was immobilized in 96-well plates coated with streptavidin. Nedd8 with an Nterminal HA tag was added to the plate with ATP and the formation of Nedd8,NAE conjugate was detected with a mouse anti-HA antibody and an anti-mouse IgG antibody linked to horse radish peroxidase . We observed significant immobilization of HA-Nedd8 in NAE coated wells in the presence of ATP. In contrast, only low levels of HA-Nedd8 were bound to the plate when either NAE or ATP was excluded from the reaction. These results suggest that PCP tagged NAE immobilized on the streptavidin plate retained catalytic activity with Nedd8 and could be used for the selection of the UB library based on the catalytic formation of UB,NAE conjugates. We next carried out phage selection of a UB library that was constructed in a previous work with randomized residues replacing 71 LRLRG75 at the UB C-terminus. Selection of the phage library was based on the formation of thioester conjugates between phage displayed UB variants and NAE immobilized on the streptavidin plate. Control reactions were set up in parallel in which either NAE was not immobilized on the plate or ATP was excluded from the reaction mixture. Phage bound to the plate were eluted by DTT and amplified for further rounds of selection. The amounts of phage input, NAE immobilization and reaction time were reduced through iterative rounds of selection to specifically enrich phage displaying UB that were reactive with NAE. There was a steady increase in phage enrichment from the selection over the controls as the selection conditions became more stringent. The amount of phage eluted from the fifth round of selection 17636045 was 1,500-fold and 120-fold higher than the controls missing either NAE or ATP, respectively, suggesting that a substantial fraction of the SNDX-275 selected phage pool displayed UB variants that were catalytically active with NAE. Sequencing of the UB clones after the fifth round of selection showed a convergence of the UB C-terminal sequences recognized by NAE. The gating residue Arg72 of UB is most commonly replaced by hydrophobic residues with aliphatic side chains including Ala, Leu, Ile, Val, or infrequently with polar residues such as Ser or Gln, or large aromatic side chains such as Phe and Tyr. No positively charged residue such as Arg, Lys or His was selected suggesting that NAE rejects positively charged residues at position 72 to avoid activation of the wt UB. Out of 58 clones sequenced, only one clone has a Ser replacing Gly75 suggesting that the Gly-Gly motif at 22634634 the C-terminal end of UB is important for reactivity with NAE. Compared to residues 
72 and 75, residues selected at positions 71, 73 and 74 can be more diverse with a significant preference for aromatic side chains replacing the native residues Leu71, Leu73 and Arg74 at the C-termini of wt Nedd8 and wt UB. UB clone N26 with the single Arg72Ala mutation appeared twice among the 58 clones sequenced. This clone has the same C-terminal sequence as Nedd8. This suggests that other UB clones in the selected pool with substantial sequence variations can compete with the native C-terminal sequence of Nedd8 for NAE activation. Interestingly clone N