Confirmation of the existence of an E1•E2 disulfide complex

a. Two time-dependent chase assays in which the order of addition of an E2 enzyme (UBE2L3 or UBE2L6) or a Ubl (Ub or ISG15) protein is added to the mixture containing an E1 enzyme (UBA1 or UBA7). The expected products are E1∼Ubl and E2∼Ubl if no E1•E2 complex is formed. b. The two chase reactions i and ii produced similar yields of UBA1∼Ub* and UE2L3∼Ub*, where * represents a fluorescein probe. In contrast, the UBA7 plus UBE2L6 premixure in reaction i resulted in less UBE2L6∼ISG15* product than detected for the UBA7 plus ISG15* premixure in reaction ii. The Coomassie-stained gel shows a high molecular weight (MW) band at ∼125 kDa, which is not ISG15*-related but may be an UBA7•UBE2L6 covalent complex. c. The active cysteines in bovine UBA7 (bUBA7) and UBE2L6, respectively, are responsible for UBA7•UBE2L6 complex formation since C597A (UBA7) or C86A (UBE2L6) mutation resulted in no reaction. UBA7-UBE2L6 pairs from human, mouse and bovine all form a disulfide complex (indicated by red arrows), but not human UBA1 mixed with UBE2L3 or UBE2L6. d. SEC-MALS chromatographs establish high-precision MWs of UBA7 alone (110.13 kDa theoretical mass) and for the UBA7•UBE2L6 complex. The blue and red dashed lines represent the predicted MWs of UBA7 and UBA7•UBE2L6 complex, respectively. e. Cartoon representation of the UBA7•UBE2L6 complex.

Cryo-EM data collection, refinement, and validation statistics.

Structural analysis of bUBA7•UBE2L6∼ISG15 complex.

a. Cryo-EM map of the MBP-UBA7•UBE2L6 complex. UBE2L6 is depicted in sky blue, while the SCCH domain of UBA7 is highlighted in salmon, and the FCCH domain is emphasized in yellow. b. Cryo-EM map of the bUBA7•UBE2L6∼ISG15 complex. The domains of UBA7 are displayed in the same color codes, while ISG15 is represented in green. Zoomed-in views show the cryo-EM density of adenylated ISG15 bound to UBA7, as well as the disulfide bond formed between the catalytic cysteines of UBA7 and UBE2L6. c. Superimposed structures of three key interfaces within the UBA7/UBE2L6/ISG15 complex from both bovine and human (PDB ID: 8SEB). The interfaces include: (I) UBA7/ISG15, (II) UBA7(UFD)/UBE2L6, and (III) UBA7(SCCH)/UBE2L6. d. Comparative analysis of the AAD and ISG15 interaction in the UBA7/UBE2L6/ISG15 complex between bovine and human. Structural alignment highlights the interaction of the AAD of UBA7 with ISG15, revealing both conserved and distinct interaction patterns across species. Key contact residues are shown, emphasized by the ball and stick style. e. Interface analysis of the UFD and E2 structures in the UBA7/UBE2L6 complex. The UFD of bovine UBA7 is displayed as a colored surface representing electrostatic potential, while the hA helix of UBE2L6 is shown as a cartoon. Based on electrostatics, the left and right areas of the hA are identified as interaction Site-1 and Site-2, respectively. The superimposition of the bovine and human’s UFD of UBA7 (bovine’s colored in light salmon and human is white) and UBE2L6 (bovine: sky blue, human: pink) structure, illustrates structural differences. Sequence alignment of the different species’ UBE2L6. The blue asterisk marked the residues that differ between the UBE2L6s’ hA of bovine and human. Magnified views of Site-1 and Site-2. The structures are represented as transparent cartoons and the residues contained in the contact are shown as ball and stick.

The role of the SCCH domain in the UBA7 and UBE2L6 interaction

a, b. Magnified views of the interface between the SCCH domain of UBA7 and UBE2L6 in both bovine and human structures. Contacting residues at the interface are depicted in ball-and-stick representation. c. Structural alignments of the SCCH domain of E1 in its apo form and in complex with E2. The catalytic cysteine capping loop (CCL) is defined by the loop structure flanked by two helices: an entry helix at the N-terminal and an exit helix at the C-terminal ends. In the yeast E1-E2 complex model, the CCL region is partially missing, but it becomes visible in the electron density map (EMD_44223) when the contour level is lowered. d. Superimposed structural models of the SCCH domain from UBA7 and UBA1 across different species, obtained from the AlphaFold database (UniProt IDs displayed below each species’ name). The entry and exit helices, as well as the catalytic cysteine capping loop (CCL), are highlighted in color, while the rest of the structure remains transparent. The number of residues in the CCL for each species is counted and annotated along a number line.

HDX-MS reveals dynamic structural characteristics of E1 enzymes

a-b. Bovine UBA7 and human UBA1 display structural similarities, yet the hydrogen-deuterium exchange rates in their respective SCCH domains are substantially different. The UBA7 SCCH domain exhibits a mean ΔHDX >15%, whereas ΔHDX values for most of the residues in the UBA1 SCCH are <5%. c-d. Upon UBA7 binding to UBE2L6, ΔΔHDX values for the UBA7 SCCH are notably reduced, implying that UBE2L6 stabilizes SCCH. e. When human UBA1 binds UBE2L3, ΔΔHDX values for the UBA1 SCCH region are notably increased. In this scenario, the CCL is more exposed to the solvent (resulting in increased ΔΔHDX), implying the CCL is in an open state (as illustrated in f). g. The dynamic characteristics of the CCL of UBA1 are consistent with the crystal structure of the yeast UBA1•CDC34 disulfide complex (PDB ID 6NYA), for which the structure of the CCL spanning N773-I798 is too dynamic to be resolved.

The enzymatic kinetics of UBA1, UBA7 and their variants

a. Schematics of the domains in UBA1 and UBA7, as well as their variants. The deleted and extended sequences in UBA1(dCCL) and UBA7(eCCL) are indicated. Chimeras in which the UFD was swapped between UBA1 and UBA7 are also shown. In the chimeras, the CCL remains unchanged. b-c. Production yields of E2∼Ub* or E2∼ISG15* (E2 = human UBE2L3 or bovine UBE2L6) using a wildtype (wt) E1 enzyme (UBA1 or UBA7) or a mutant variant. The reaction comprised a mixture of E2 enzyme, Ubl (Ub or ISG15), MgCl2 and ATP, and it was initialized by adding an E1 enzyme into the solution. The 30-min reaction was evaluated by SDS-PAGE. For hUBA1, both UBE2L3 and UBE2L6 can transfer Ub* from wt UBA1. UBA1(dCCL) and UFD-swapped UBA1(U7) generated elevated levels of UBE2L6∼Ub*, but they had no significant impact on UBE2L3∼Ub* production. Note, UBA7 exclusively binds to UBE2L6 as its cognate E2 enzyme for ISG15* transfer. When UBA7(eCCL) or UFD-swapped UBA7(U1) were used as the E1 enzyme, UBE2L6∼ISG15* production was dramatically increased compared to when wt UBA7 was used. UBA7(U1) could also conduct a small amount of UBE2L3 transthioesterification to generate UBE2L3∼ISG15*. d-g. The Michalis-Menten kinetics of wt UBA7, UBA7(U1), and UBA7(eCCL), revealing diminished kcat values for the mutant variants. The dissociation constants (KD, Bio-Layer Interferometry, BLI) and Michalis-Menten constants (KM) calculated for wt UBA7, UBA7(eCCL), and UBA7(U1) are highly consistent. For example, the UBA7(U1) chimera exhibits reduced KM and KD values (50-90 µM) relative to wt UBA7 (g). The E1 or E2 enzymes used in the BLI assay are catalytic dead to prevent the spontaneous disulfide complex formation.

The disulfide complexes of UBA7 variants and the pKa scores of selected E1-E2 pairs

a-c. The UBA7•UBE2L6 complex was monitored using SDS-PAGE by titrating UBA7 with varied concentrations of UBE2L6. Production of the complex was assessed and the data was used to fit the kinetics profile. Production of the complex was significantly reduced upon using the UBA7(eCCL) variant, as its extended CCL covers the catalytic cysteine. The UFD-swapped UBA7(U7) variant presented both reduced catalytic efficiency and binding affinity. The black and blue arrows in the SDS gels indicate bands for the disulfide complex and the E1 enzyme alone, respectively. The red arrow in c denotes UBA7(U1) variant contamination. d. Evaluation of disulfide complex formation for hUBA1 and the UBA1(dCCL) variant with UBE2L3. Due to low binding affinity and catalytic efficiency, only a single reaction at 30-min was collected. The ratio of disulfide complex production over the total amount of E1 enzyme was calculated. The hUBA1(dCCL) variant significantly enhances disulfide complex formation. e. Schematic of the alkylation reaction, using iodoacetamide to react with the thiolate group of cysteine. f. The alkylation ratios of UBA1, UBA7, and their variants are pH-dependent. g. The alkylation ratios of various E2 enzymes (single cysteine variants) were directly measured according to changes in heat release in different pH buffers. In f and g, datapoints for the E1 and E2 enzymes were fitted to determine pKa values.

Summary of E1-E2 interactions in the Ub/Ubl cascade

a. In the ubiquitin reaction cascade, Ub is first charged by UBA1, followed by transfer from UBA1 to an E2 enzyme and sequential ubiquitylation reactions. When Ub is ultimately released, UBA1 is free to reinitiate the cascade. b. The ISG15 system operates via two distinct pathways. At the immune resting state, UBA7 and UBE2L6 favorably form a disulfide complex to suppress intracellular ISGylation (pathway 1). However, when the immune response is elevated, UBA7 charges ISG15 to initialize ISGylation of target proteins (pathway 2). c. Co-expressing GFP-UBA7 and UBE2L6-V5 in HEK293T cells. Non-reducing immunoblot of UBE2L6-V5 reveals a high molecular weight complex (>130 kDa) which is reduced by β-ME. The control groups (GFP and UBE2L6-V5, GFP and GFP-UBA7) demonstrate that the complex only forms when UBE2L6-V5 and GFP-UBA7 are co-transfected. Blue circles indicate the groups expressing UBE2L6 and UBA7 separately, then mixed after sample buffer addition. The fluorescent gel confirms GFP-UBA7 expression and anti-GAPDH is used as a loading control.

The UBA7•UBE2L6 complex is sensitive to a reducing agent.

UBA7 was used to charge ISG15 and to transfer it from UBA7 to UBE2L6. After reaction at 37 °C for 30 mins, the reaction was quenched using 4X non-reducing Laemmli sample buffer. β-mercaptoethanol (β-ME, 10 mM) was also added in the experimental (reduced) samples. The SDS gel shows that the thioester-linked products, including UBA7∼ISG15* and UBE2L6∼ISG15*, are disrupted by β-ME treatment, as observed in both Coomassie-stained and fluorescence images. A band corresponding to the UBA7•UBE2L6 complex is only detected in the Coomassie-stained gel, and it is also sensitive to β-ME treatment, implying a disulfide-bonded complex.

Cryo-EM analysis of the MBP-UBA7•UBE2L6 complex (UBA7•UBE2L6)

a. Schematic representation of the formation of the MBP-UBA7•UBE2L6 complex. The first 11 residues of UBA7 have been deleted and replaced by a linker sequence (-NAAA) attached to the C-terminus of maltose-binding protein (MBP). b. The representative micrograph out of all images used for processing. c. The cryo-EM data processing flow chart. The final map was reconstructed to 3.83 Å.

Cryo-EM data processing of the UBA7•UBE2L6 adenylated ISG15 complex (UBA7•UBE2L6-ISG15(a))

a. Schematic depicting the formation of the UBA7•UBE2L6-ISG15(a) complex. In the complex, the active site cysteines of UBA7 and UBE2L6 are cross-linked by a disulfide bond. b. The representative micrograph out of all images used for processing. c. The cryo-EM data processing flow chart. The final map was reconstructed to 2.99 Å whereas local resolution was estimated (FSC0.143) suggesting FCCH and SCCH domains exhibit lower resolution (i.e. 4-7 Å).

Models of chimeric proteins for functional analyses of E1 enzymes

a-b. Extension or deletion of the cysteine capping loop (CCL) region in UBA7 and UBA1, respectively, results in the catalytic cysteine either being exposed or covered. c. In the case of the chimeric E1 variants UBA7(U1) and UBA1(U7), UFD domains were swapped while their linker regions remained unchanged. Substitutions of the UFDs were initiated at amino acid positions Q896 and R944 for UBA7 and UBA1, respectively. These positions correspond to residues located between the first β-sheet of the UFD and the linker.

a. Disulfide bond formation assay of UBA7 and UBE2L6 among three species: mouse (M), bovine (B), and human (H). UBA7 and UBE2L6 were mixed and incubated for 30 minutes at 37 °C, resulting in cross-reactivities. A contamination band for hUBA7 (R&D Biosystems, USA) is indicated by *. b. UFD-deleted UBA7 was used to validate ISG15 activation, transthioesterification, and UBA7(dUFD)•UBE2L6 complex formation. Use of UBA7(dUFD) does not impair ISG15 activation, but transthioesterification and disulfide complex formation are not observable in the Coomassie-stained SDS gel. H2O2 treatment of an UBA7(dUFD) and UBE2L6 mixture results in no disulfide complex formation. c. Curve fitting and sensorgrams of BLI-determined binding affinities for E1-E2 pairs. The UBA7(C597A) or UBE2L6(C86A) mutant was used for BLI experiments on wtUBA7•UBE2L6 or UBA7(eCCL) •UBE2L6, respectively, to avoid irreversible disulfide complex formation during the BLI experiments.

AlphaFold2-generated models of complexes comprising E1 variants and their cognate E2 enzymes.

AlphaFold2-predicted E1•E2 complex structures, including wt UBA7•UBE2L6 (a), UBA7(eCCL)•UBE2L6 (b), wt UBA1•UBE2L3 (c), and UBEA1(dCCL)•UBE2L3 (d). E1, E2 and the CCL are colored in gray, green and red, respectively. Zoomed-in views of the contacts near the two active site cysteines show the distances between the two sulfur atoms of the cysteines. Both wt and eCCL variants of UBA7 can form disulfide complex with UBE2L6, but only the UBA1(dCCL) variant can do so.

pH-dependent inactivation of the active site cysteine of E1 enzymes

a. The catalytic cysteine of E1 enzymes is influenced by the local acidic patch, as revealed by the electrostatic surface potential drawn in PyMOL 2.2. Cartoon and surface views show the locations of the CCL and catalytic cysteines. b. E1 enzymes were alkylated at desired pH, then diluted to pH 7.5, before undergoing an immediate E1∼Ubl charging reaction. c. At higher pH, greater amounts of alkylated E1 result in lower quantities of E1∼Ubl* product. The ratios of alkylated E1 versus pH were used to obtain the pKa values for each E1 enzyme. Control lanes are E1 enzymes without alkylation, eliciting maximal amounts of E1∼Ubl* product.

UBA1 forms disulfide complex with UBE2L3 in ROS solution

UBA1•UBE2L3 disulfide complex formation was tested in the presence of H2O2. In the absence of H2O2, wt UBA1 cannot form significant amounts of UBA1•UBE2L3 (SS) complex, even when a 40-fold concentration of UBE2L3 (60 µM) to UBEA1 (1.5 µM) was used. In contrast, upon addition of 0.5 mM H2O2, strong signal of disulfide complex was detected. When the CCL-shortened UBA1(dCCL) variant was used, disulfide complex spontaneously formed in the absence of H2O2, and H2O2 treatment did not further facilitate the ability of the UBA1(dCCL) variant to generate disulfide complex.