Cryo-EM structures of UBA6 reveal mechanisms of E1–E2 specificity and dual FAT10/ubiquitin thioester transfer

Abstract

UBA1 and UBA6 define parallel ubiquitin (Ub) activation systems that perform non-overlapping roles in Ub and ubiquitin-like protein (Ubl) signaling. Whereas UBA1 supports the canonical Ub pathway, UBA6 also activates the Ubl FAT10, linking Ub signaling to immune-regulated proteostasis. In addition to selective Ub/Ubl activation, UBA1 and UBA6 engage distinct sets of E2s, yet how these enzymes achieve selective E2 engagement has remained unclear. Using chemical trapping and high-resolution cryo-EM, we determine four structures of UBA6–E2 complexes representing the thioester-transfer step with either FAT10 or Ub, revealing how this E1 distinguishes its cognate partners. UBA6 achieves E2 specificity through coordinated contributions of the UFD and SCCH domains, a dual-domain mechanism that contrasts with the UFD-dominated selectivity of UBA1. The structures further show that an existing inositol hexakisphosphate (InsP₆)–binding site, unique to UBA6, stabilizes an expanded SCCH cleft that pre-organizes the enzyme for selective engagement of UBA6-specific E2s. These findings define principles for E1–E2 recognition and identify InsP₆ as a cofactor shaping specificity within the Ub-like conjugation network.

Specificity in ubiquitin (Ub) and Ub-like protein signaling is essential. Here, the authors use cryo-EM to show how UBA6 selectively engages its cognate E2 for dual Ub and FAT10 transfer, revealing a role for an InsP₆-binding site and illuminating molecular rules governing pathway specificity.

Introduction

Posttranslational modification of proteins by ubiquitin (Ub) and Ub-like proteins (Ubl) is a key regulatory mechanism governing nearly all aspects of eukaryotic cell biology^1–3. Ub/Ubl conjugation alters target protein properties such as stability, subcellular localization, intermolecular interactions, and activity, thereby regulating essential cellular processes, including cell cycle control, DNA repair, signal transduction, and immunity^4–8. Ub/Ubl signaling is mediated by three enzymes, E1, E2, and E3, which work sequentially to conjugate Ub/Ubl to target proteins^9–11. E1 enzymes initiate the process by coupling ATP hydrolysis to the formation of a high-energy E1-Ub/Ubl thioester linkage^9–11. This enables the recruitment of E2-conjugating enzymes and transfer of activated Ub/Ubl to a catalytic cysteine on E2, a process known as E1-E2 thioester transfer^9–11. E2-Ub/Ubl intermediates then interact with an array of E3 ligases to complete the conjugation process, attaching Ub/Ubl to target proteins either as single molecules or polymeric chains, thereby dictating the biological outcomes of Ub/Ubl signaling^12–15.

The canonical E1 enzymes for Ub^16–24, SUMO^25–28, NEDD8^29–31, FAT10^32,33, and ISG15^34,35 act as gatekeepers of their respective conjugation pathways, each built around a conserved multidomain architecture in which individual domains play distinct functional roles. In all canonical E1s, the adenylation domain, consisting of active and inactive adenylation domains (AAD and IAD, respectively), catalyzes ATP-dependent activation of the Ub/Ubl C-terminus to form an adenylate intermediate, while the catalytic cysteine (Cys) domain, comprising first and second catalytic cysteine half-domains (FCCH and SCCH), harbors the active-site cysteine that forms the transient E1~Ubl thioester intermediate. A C-terminal Ub-fold domain (UFD) recruits and positions the cognate E2 enzyme to facilitate E1-E2 thioester transfer. By coordinating domain dynamics and large-scale conformational rearrangements, canonical E1s orchestrate the sequential steps of adenylation, thioester formation, and E2 loading that drive Ubl conjugation³⁶.

Despite their conserved structural and mechanistic framework, Ub/Ubl conjugation pathways regulate distinct repertoires of cellular processes. Fidelity across these systems is essential to cell biology and relies on each E1 acting as a molecular gatekeeper that selectively engages its cognate Ubl and E2 while excluding noncognate partners in order to preserve the integrity of pathway-specific signaling. While most Ubl pathways are served by a single E1 enzyme, the human Ub system is unusual in having two E1 enzymes, UBA1 and UBA6^37–41. UBA1 is dedicated exclusively to Ub activation, whereas UBA6 is distinctive in that it can also activate FAT10^37,40, a Ubl involved in mitotic progression⁴², immunity^43–46, and implicated in cancer^47–52. Although UBA1 executes the majority of Ub activation in the cell³⁸, UBA6 defines a distinct, functionally indispensable branch. UBA6 is not only involved in selective Ub-conjugation events such as degradation of RGS (Regulator of G Protein Signaling) proteins⁵³ and restriction of autophagy via LC3B⁵⁴, but as the only E1 for FAT10, UBA6 promotes proteasomal targeting of substrates upon TNF-α and IFN-γ stimulation, coupling its activity to cytokine-controlled proteostasis³⁹. In addition to specificity at the level of E1–Ubl interaction and activation, UBA1 and UBA6 interact with and charge distinct subsets of more than thirty Ub E2 enzymes^38,40,41. While certain E2s, such as members of the UBCH5 family, can be charged by both UBA1 and UBA6, others display strict selectivity with UBE2R1 (CDC34) functioning exclusively with UBA1³⁸, and UBE2Z³⁸, BIRC6^54–56, and UBE2O⁵⁷ functioning solely with UBA6. Interestingly, unlike most E2s, UBE2Z, BIRC6 and UBE2O possess both N- and C-terminal extensions flanking the UBC core domain and are classified as class IV E2’s⁵⁸. This division of labor across E1–E2 pairs implies that precise molecular rules govern E2 selection, ensuring that each branch of the Ub system engages the appropriate downstream machinery. While previous studies have elucidated the structural elements within the E1 adenylation domains and Ubls that govern selective activation of Ub by UBA1 and dual activation of Ub and FAT10 by UBA6^32,33,59, the molecular basis for E2 recognition and selective thioester transfer by these enzymes remains poorly understood. While the UFD of canonical E1s is recognized as a key determinant of E2 specificity, the structural principles underlying UBA1 and UBA6 selectivity, and whether additional domains such as the Cys domain also contribute, remain unknown due to the absence of UBA6–E2 complex structures.

In this work, we seek to elucidate the molecular basis of E1–E2 thioester transfer by UBA6 and the principles governing E2 specificity between UBA1 and UBA6. Using chemical trapping and cryo-EM, we determine four structures of UBA6–UBE2Z complexes bound to either FAT10 or Ub thioester-like states. Biochemical analysis reveals that UBA1 selectivity for its cognate E2s resides primarily within the UFD, whereas UBA6 employs a dual-domain mechanism, in which both the UFD and SCCH domains cooperatively determine specificity for UBA6-type E2s. In contrast, UBCH5 family members function efficiently with any UFD–SCCH combination from UBA1 or UBA6, consistent with their broad compatibility. The cryo-EM structures visualize the molecular interfaces and conformational rearrangements that enable UBA6 to recruit UBE2Z and mediate thioester transfer of both Ub and FAT10, and further reveal that an existing inositol hexakisphosphate (InsP₆)–binding site, unique to UBA6, plays a previously unrecognized role in stabilizing the SCCH domain and pre-organizing UBA6 for selective class IV E2 engagement. Together, these data reveal how coordinated contributions of the UFD, SCCH, and InsP₆-binding architecture encode E2 specificity, enable dual-Ubl reactivity, and enforce pathway fidelity across the Ub-like conjugation network.

Results

Overall Structure of the UBA6-UBE2Z/FAT10(a) Complex

Given the distinct E2 specificities of UBA1 and UBA6, we next sought to define how UBA6 engages its cognate E2, UBE2Z, during FAT10 activation and transfer. UBE2Z is a class IV E2 enzyme comprising a conserved catalytic UBC core domain (residues ~99–258) flanked by extended N- and C-terminal regions (Fig. 1a, b). The N-terminal ~100 residues are predicted to be intrinsically disordered and enriched in glycine, alanine, and serine. For structural and functional analyses, we used a truncated construct (residues 94–354; ΔNter-UBE2Z), hereafter referred to as UBE2Z. Both full-length and truncated proteins displayed comparable Ub and FAT10 thioester transfer activity with UBA6 (Fig. 1c).

Fig. 1. Overall architecture of a UBA6-UBE2Z/FAT10(a) complex.

a Domain organization of UBA6 and UBE2Z; regions of disorder are shown as hatched boxes, with amino acid boundaries for selected domains and structural motifs indicated. b Schematic of the UBA6/UBE2Z/FAT10 thioester transfer reaction (left) and the disulfide-trapped UBA6-UBE2Z/FAT10(a) mimetic characterized in this study (right). c E1–E2 thioester transfer assays comparing full-length and N-terminally truncated UBE2Z variants. A representative gel is shown from experiments repeated independently with similar results. d E1–E2 thioester transfer assays testing UBA6 and UBA1 with the indicated E2 enzymes and Ubls (Ub or FAT10). A representative gel is shown from experiments repeated independently with similar results. Source data are provided as a Source Data file. e Cryo-EM reconstruction of the UBA6-UBE2Z/FAT10(a) complex (left) and corresponding cartoon representation (right). The IAD and AAD domains of UBA6 are shown in slate blue and light pink, respectively; the SCCH domain in magenta, FCCH in hot pink, and UFD in orange. UBE2Z is colored cyan, and FAT10(a) in lime green.

We next assessed E1 specificity across representative E2s using both Ub and FAT10 as substrates. UBE2Z was efficiently charged by UBA6 with either modifier, whereas BIRC6 was charged by UBA6 with Ub only (Fig. 1d). Neither UBE2Z nor BIRC6 supported thioester transfer by UBA1. In contrast, CDC34 and UBE2B were charged by UBA1 but not UBA6. As expected, the promiscuous E2 UBCH5B formed a Ub thioester with either E1 (Fig. 1d). Collectively, these assays establish distinct, non-overlapping E2 preferences for UBA6 versus UBA1 in vitro.

To define the molecular basis of this specificity, we employed a disulfide-trapping crosslink that covalently links the catalytic cysteines of UBA6 (C625) and UBE2Z (C188) (Fig. 1b) in the presence of free FAT10 and ATP/Mg²⁺. Cryo-EM analysis of the trapped complex yielded a 3.24 Å map of the UBA6-UBE2Z/FAT10 adenylate (UBA6-UBE2Z/FAT10(a)) (masked FSC 0.143) (Fig. 1d; Supplementary Fig. 1 and 2; Supplementary Table 1). The overall architecture of the complex is conserved: the IAD and AAD of UBA6 interact with FAT10(a), while UBE2Z bridges the UFD and SCCH domains (Fig. 1e). UBE2Z is recognized through coordinated contacts from both domains, burying ~2,050 Ų of surface area at the interface. The FAT10 C-terminal Gly165 occupies the active site with continuous density consistent with an acyl-adenylate intermediate, and the engineered disulfide between UBA6 C625 and UBE2Z C188 is clearly resolved (Supplementary Fig. 1d). As observed in a previous UBA6/FAT10(a) structure, the FAT10 C-terminal domain binds at the same AAD/IAD interface that accommodates Ub in UBA1, whereas the FAT10 N-terminal domain contacts the three-helix bundle within the IAD of UBA6 (Fig. 1e). Together, these features reveal a modular binding mode that simultaneously accommodates both FAT10 domains and positions the C-terminal tail for adenylate formation and subsequent thioester transfer. Notably, density for an InsP₆ molecule is visible within the basic pocket of the SCCH domain, consistent with the previously described UBA6-specific binding site (Fig. 1e; Supplementary Fig. 1d)³³. Collectively, these findings establish the overall architecture of the UBA6-UBE2Z/FAT10(a) complex and set the stage for a detailed analysis of the molecular features that define E2 recognition and specificity.

Conformational Changes Accompanying UBA6-UBE2Z/FAT10(a) Thioester Transfer

To elucidate how UBA6 engages UBE2Z and enables thioester transfer, we compared the UBA6-UBE2Z/FAT10(a) complex with previously determined apo and Ub-bound structures of UBE2Z and UBA6 (Fig. 2a–d). In the apo structure (PDB 5A4P), UBE2Z adopts the canonical UBC core fold but contains four distinctive surface loops (LA–LD) that define its class IV identity. The LA loop (residues 169–173) is a short insertion unique to UBE2Z and BIRC6; deletion of this segment renders UBE2Z insoluble, indicating a role in folding or stability. The LB loop (residues 194–197) arches over the catalytic cysteine (C188) and features Trp195, which shields the active site and reduces solvent accessibility. The LC loop (residues 218–258) replaces the third α-helix found in canonical E2s (hC) and extends from the UBC core to contact the C-terminal extension, which includes two α-helices (hE and hF) enclosing the LD loop (residues ~290–326). Together, these elements form a compact structural module that stabilizes UBE2Z while providing the flexibility required for E1 engagement (Fig. 2a).

Fig. 2. Conformational changes accompanying thioester transfer in the UBA6-UBE2Z/FAT10 complex.

a Middle, Superposition of apo UBE2Z (PDB: 5A4P, gray) and UBE2Z from the UBA6-UBE2Z/FAT10(a) complex (cyan). UBA6 domains are colored as follows: SCCH (magenta), FCCH (hot pink), UFD (orange), and FAT10(a) (lime green). InsP₆ bound to the SCCH domain is shown as spheres. Left, magnified view of the helix A–UFD interface with an accompanying schematic representation of the hA helix rotation ( ~ 12°) and translation ( ~ 0.6 Å) between apo and E1-bound UBE2Z. Right, close-up of the LC loop highlighting conformational differences between apo and E1-bound states of UBE2Z. b Superposition of apo UBA6 (PDB: 7SOL, gray) and UBA6 from the UBA6-UBE2Z/FAT10(a) complex (colored as in a). Major conformational changes are observed in the UFD (left) and SCCH (right), with the α23–α24 loop of the SCCH repositioned to accommodate UBE2Z. c Comparison of the UFD conformations in the thioester-transfer–inactive (PDB: 7SOL, top) and thioester-transfer–active (this study, bottom) states. Interacting residues are shown as sticks; hydrogen bonds are indicated by dashed lines. d Comparison of the SCCH domains from the E2-free (PDB: 7SOL, center) and UBE2Z-bound (this study, left) structures. The cysteine cap is ordered in the apo state but becomes disordered upon E2 binding. Right, superposition of both SCCH domains, highlighting conformational rearrangements in the cysteine cap and α23–α24 loop.

Superposition of the apo and E1-bound UBE2Z structures revealed a series of coordinated conformational adjustments that promote stable E1–E2 interaction (Fig. 2a; Supplementary Fig. 1f). Helix A rotates by ~12° and translates by ~0.6 Å, enhancing complementarity with the UFD of UBA6. Notably, the LC loop, one of the defining structural features of class IV E2s, undergoes a localized rearrangement upon UBA6 binding, shifting to place UBE2Z within the UBA6 cleft while avoiding steric interference with the SCCH domain (Fig. 2a). This repositioning reorients the catalytic cysteine (C188) toward the active-site channel of UBA6, aligning it for productive thioester transfer.

UBA6 itself undergoes substantial remodeling upon E2 engagement. Comparison of the UBE2Z-bound and Ub-bound structures (PDB 7SOL) shows that the UFD rotates by ~24° and translates by ~7.6 Å, a concerted motion that aligns the catalytic cysteines of UBA6 (C625) and UBE2Z (C188) for thioester transfer (Fig. 2b). This rearrangement is stabilized by new intramolecular contacts between D555 in the AAD and Y1021/R1004 in the UFD, along with a ~90° rotameric shift of R957 that preserves a hydrogen bond with the AAD backbone carbonyl of A583 (Fig. 2c). Additional adjustments occur in the α23–α24 loop of the SCCH, which moves toward the LC loop of UBE2Z to complete the interface (Fig. 2b). These subtle yet coordinated shifts bring the E1 and E2 active sites into near-ideal proximity for thioester exchange.

Within the SCCH domain, engagement of UBE2Z triggers pronounced remodeling of the cysteine-cap region (Fig. 2d). In the E2-free state, residues S817–A831 form a short helix that covers C625, likely protecting it from oxidation. Upon E2 binding, this helix melts and becomes largely disordered, with only weak density visible for L828–A831. The SCCH domain simultaneously rotates by ~6° and translates by ~1.5 Å, movements that retract the cysteine cap and expose the catalytic cysteine for thioester formation. Cap destabilization and SCCH reorientation thus appear prerequisite for establishing a catalytically competent E1–E2 configuration. By bringing the active sites into precise alignment and restructuring the catalytic environment, these conformational changes define the platform for selective molecular recognition between UBA6 and UBE2Z described below.

Molecular Recognition of UBE2Z by UBA6

The UBA6-UBE2Z/FAT10(a) structure reveals that UBE2Z engages UBA6 through three principal domains, the UFD, SCCH, and AAD, via an extensive network of complementary electrostatic and hydrophobic interactions (Fig. 3a–e). Together, these contacts bury approximately 2,050 Ų of surface area and create a composite interface that orients the catalytic cysteines for thioester transfer. The UFD contributes a broad recognition surface that buries roughly 700 Ų and provides both charge complementarity and hydrophobic anchoring (Fig. 3a, c). An acidic patch on the UFD (Asp1035–Asp1037) pairs with basic residues on helix A of UBE2Z where extensive electrostatic interactions occur (Fig. 3a). Arg106 of UBE2Z forms salt bridges with UBA6 Asp1035 and Glu1036, while UBE2Z Lys105 interacts with UBA6 Asp1037. Mutational analysis underscores the importance of this network of interactions as reversing the UFD acidic patch (D1035K/E1036K/D1037K) or introducing charged substitutions within either hydrophobic pocket abolished thioester formation (Fig. 3c; Supplementary Fig. 3a). Two adjacent hydrophobic clusters reinforce this electrostatic core. In one, UBE2Z Leu102 and Met128 pack against UFD Met985 and Val987; in the other, UBE2Z Met109 and Tyr112 contact UFD Met992, Val997, Met998, and Pro999 (Fig. 3a). Importantly, reciprocal E2 mutations (L102D/M128D or M109A/Y112A) reduced activity to less than 10% and approximately 50% of wild-type levels, respectively (Fig. 3c; Supplementary Fig. 3b). Together, these structural and mutational analyses establish the UFD–hA interface as the dominant docking surface that anchors UBE2Z on UBA6 and enables subsequent SCCH-mediated catalytic alignment.

Fig. 3. Molecular recognition of UBE2Z by UBA6 reveals determinants of E2 recruitment.

a Left, network of contacts between the UBA6 UFD and UBE2Z, with interacting residues shown as sticks (oxygen, red; nitrogen, blue; sulfur, yellow). Hydrogen bonds are indicated by dashed lines. Right, electrostatic surface representation of the UBA6 UFD with UBE2Z overlaid, colored as in the left panel. b Interaction interfaces between UBE2Z and the UBA6 SCCH and AAD domains. Insets show detailed views of four contact sites (I–IV) within the composite UBE2Z–UBA6 interface. c–e E1–E2 thioester transfer assays assessing the contribution of individual interfaces to complex formation and activity. c UFD–UBE2Z mutants. d SCCH–UBE2Z mutants. e AAD–UBE2Z mutants. Data are presented as bar graphs showing mean ± SD from three technical replicates, normalized to wild-type activity. Individual replicate values are shown as black dots. Source data are available in the accompanying Source Data file.

The SCCH domain provides three additional points of contact that consolidate E2 positioning around the catalytic cleft (Fig. 3b, d). Site I, centered on the UBA6 (Cys625) and UBE2Z (Cys188) catalytic cysteine residues which are flanked by hydrophobic residues Phe624 and Pro623 that cradle the E2 active site and likely play a role in stabilizing the active sites in proximity to each other. The importance of these residues to E1-E2 thioester transfer is evidenced by the significant reduction in activity of a P623A/F624K double mutant of UBA6 (Fig. 3d; Supplementary Fig. 3c). This junction is reinforced by a side chain mediated hydrogen bond between UBE2Z His161 and SCCH Glu619 (Fig. 3b). Site II lies within the partially melted cysteine cap, where SCCH Glu829 and Asn799 hydrogen bond with the UBE2Z backbone at Arg305 and Gly306, while UBE2Z Asn172 and Lys260 engage SCCH Gly676–Gly681. Alanine substitution of Asn172 produced only a moderate (~30%) loss of activity (Fig. 3d; Supplementary Fig. 3d). Site III involves the α23–α24 loop of UBA6 packing against loop LC of UBE2Z. Hydrophobic residues from UBA6 (Ile726, Phe734, Phe722, Cys721) contact UBE2Z Phe228, Phe301, and Arg232, the latter also forming hydrogen bonds with the UBA6 H720/C721 backbone. Disrupting this patch through a quadruple mutation in UBA6 (H720A/C721D/I726K/F734D) nearly eliminated thioester activity, whereas reciprocal mutations in UBE2Z (E225A/F228A/R232D) caused only a modest loss (Fig. 3d; Supplementary Fig. 3c, d), indicating that this interface is predominantly defined by E1-side contacts and is essential for catalytic alignment. A smaller auxiliary contact occurs at Site IV, where AAD residues Arg575 and Arg615 interact with UBE2Z Gln99 and Asp157 (Fig. 3b, e). Although these interactions contribute modestly to binding, the combined AAD double mutant (R575D/R615D) resulted in a stronger loss of activity than either single substitution (Fig. 3e; Supplementary Fig. 3e, f). Of note, the reduced activity of the AAD double mutant (R575D/R615D) likely reflects defects at multiple steps in the reaction, including less efficient formation of the E1 ~ FAT10 thioester intermediate relative to wild type (Supplementary Fig. 3e).

Collectively, these structural and mutational analyses define a multi-site recognition mechanism in which the UFD establishes the primary docking surface, the SCCH forms catalytic and structural contacts that guide Cys-to-Cys alignment, and the AAD provides fine-tuning interactions that stabilize the productive E1–E2 configuration. Together, these findings reveal how UBA6 achieves stable recruitment and catalytic alignment of its cognate E2, providing a structural framework for understanding how UBA6 and UBA1 achieve distinct E2 specificities.

Distinct Domain Contributions to the E2 Specificity of UBA6 and UBA1

Previous work has established that the UFD of E1 enzymes mediates E2 recruitment and plays a central role in E2 charging^25,31,38,60. However, prior observations hinted that the UFD alone might not fully dictate E2 specificity^38,61. For instance, while UBA6 efficiently charges UBE2Z, a UBA1 chimera bearing the UBA6 UFD (UBA1^U6UFD) cannot³⁸. In contrast, both UBA1 and UBA6 chimeras carrying the UBA1 UFD (UBA6^U1UFD) readily charge CDC34 with Ub, suggesting that additional domains within E1s contribute to the recognition and alignment of cognate E2s³⁸. Together, these observations implied that E2 selectivity arises from combinatorial contributions of multiple E1 domains rather than from the UFD alone. To test this hypothesis directly, we generated a panel of UBA1 and UBA6 chimeras in which the UFD, SCCH, or both were swapped between the two E1s (Fig. 4a).

Fig. 4. Differential Contributions of the SCCH and UFD Domains to UBA6 and UBA1 E2 Specificity.

a Cartoon representations of chimeric E1 constructs generated by swapping the UFD, SCCH, or both domains between UBA1 and UBA6, colored accordingly. b E1–E2 thioester transfer assays using the eight chimeric E1s described in a and four E2s (UBCH5B, UBE2Z, BIRC6, and CDC34). Biochemical assays were repeated at least three times. c Quantification of data from panel b presented as bar graphs showing mean ± s.d. of three technical replicates. Data were normalized to thioester formation of each E2 with wild-type UBA6. Individual replicates shown as black dots. Source data are provided as a Source Data file. d Same as c, except data were normalized to thioester formation of each E2 with wild-type UBA1. e Summary table showing the relative dependence of each E2 on the UFD or SCCH domain of the E1 enzyme.

All eight variants, including wild-type enzymes, efficiently activated Ub as assessed by E1~Ub thioester formation, confirming that the chimeras were properly folded and catalytically active (Supplementary Fig. 4a). Each variant supported thioester transfer of Ub to the promiscuous E2 UBCH5B, consistent with UBCH5B’s broad compatibility with both UBA1 and UBA6 (Fig. 4b–d; Supplementary Fig. 4b). In contrast, full-length UBE2Z was efficiently charged only by UBA6^WT, whereas chimeras bearing swapped UFD domains (UBA6^U1UFD and UBA1^U6UFD) were inactive (Fig. 4b–d; Supplementary Fig. 4b). Similarly, replacing the SCCH domain of UBA6 with that of UBA1 (UBA6^U1SCCH) abolished activity, demonstrating that both the UFD and SCCH domains of UBA6 are required for efficient UBE2Z charging. Conversely, transferring both UBA6 domains onto UBA1 (UBA1^{U6UFD & U6SCCH}) conferred modest activity toward UBE2Z, whereas single-domain swaps produced only weak or negligible transfer (Fig. 4b, d; Supplementary Fig. 4b). Testing a second UBA6-specific E2, BIRC6, further supported the conclusion that both domains are critical for activity with UBA6, though their relative contributions differ (Fig. 4b–d; Supplementary Fig. 4b). These results indicate that while the UFD provides the docking platform, the SCCH contributes essential determinants that together specify UBA6’s E2 selectivity for its cognate E2s. By contrast, specificity of the UBA1-specific E2 CDC34 relied primarily on the UFD, with minimal contribution from the SCCH domain (Fig. 4b–d; Supplementary Fig. 4b).

Consistent with these observations, although UBCH5B charging was preserved across all chimeras, we observed modest quantitative differences between UBA1- and UBA6-based chimeric variants. Under our assay conditions, WT UBA1 charged UBCH5B with slightly lower efficiency than WT UBA6 (~85% of UBA6 activity; Fig. 4c, d), and introduction of domain swaps into the UBA1 scaffold further reduced activity. In contrast, analogous swaps in the UBA6 background had little effect on UBCH5B charging. This asymmetry is consistent with UBCH5B’s relaxed interface requirements and suggests that, while UBCH5B does not depend on specific SCCH-mediated contacts, the UBA1 scaffold is more sensitive to domain perturbation than UBA6. Thus, differences in overall E1 domain compatibility, rather than E2-specific recognition, likely account for the reduced UBCH5B activity observed in UBA1-based chimeras.

Altogether, these findings reveal that UBA6 and UBA1 employ distinct domain architectures to encode E2 specificity. In UBA6, the UFD and SCCH domains act cooperatively to recognize and align specialized E2s such as UBE2Z and BIRC6, with both domains contributing to productive E2 engagement. In contrast, UBA1 relies primarily on its UFD for E2 recruitment, while its SCCH domain contributes little to specificity. The promiscuous E2 UBCH5B remains unaffected by either domain swap, underscoring its relaxed interface requirements (Fig. 4b–e; Supplementary Fig. 4b). Together, these results establish a hierarchical domain code for E1–E2 pairing, in which the UFD dictates partner identity and the SCCH fine-tunes catalytic alignment. This dual-domain framework provides a mechanistic foundation for understanding how UBA6 evolved specialized E2 recognition and sets the stage for structural analysis of the SCCH–E2 interface underlying this selectivity.

SCCH–E2 Interface Features Defining E1–E2 Specificity

The structural and functional analyses above demonstrated that the SCCH domain of UBA6 plays a decisive role in establishing UBE2Z specificity. We therefore examined the structural basis for this selectivity by comparing the UBE2Z interface with corresponding regions in other E2 enzymes and by analyzing how distinct SCCH architectures across E1s accommodate these partners (Fig. 5a–i).

Fig. 5. Structural features of the SCCH–E2 interface defining E1–E2 specificity.

a Superposition of Ub E2s onto UBE2Z highlighting its three unique elements (LA, LB, and LC loops). b Cartoon representation of the UBA6–UBE2Z interface showing secondary structure elements and the LA loop (red). Inset: hydrogen-bonding network between the UBE2Z LA loop and the UBA6 α21–α22 region. Bottom: thioester transfer assays of LA loop mutants. Data are presented as bar graphs with the mean ± s.d. of three technical repeats, displayed as a percentage of the WT value. Individual replicates shown as black dots. Source data are provided as a Source Data file. c, d Structure-based sequence alignments of Ubl E2s centered on the LA c and LC d loops of UBE2Z, with secondary structure assignments above the sequences. Shaded boxes indicate residues contacting E1 in corresponding complexes. e Alignment of E1 E2 (LA loop)-binding regions showing residues that contact E2s in known Ubl E1–E2 complexes. f Docking of SCCH domains from representative E1s onto UBA6 (this study) showing steric clashes between the UBE2Z LA or LC loops and corresponding α21–α22 regions of non-UBA6 E1s. g Open-book electrostatic surface of the UBA6–UBE2Z interface showing the UBE2Z surface oriented toward InsP₆. h, i Docking of CDC34 h and UBCH5B i onto UBA6, showing that both can be accommodated without steric conflict.

Superposition of UBE2Z with other Ubl E2s shows that the UBE2Z core adopts the canonical ellipsoidal UBC fold but contains three insertion elements, LA, LB, and LC, defined earlier (Figs. 1a and 5a). LA is a short loop insertion following the β4 strand that is unique to class IV enzymes such as UBE2Z and BIRC6 and is partially conserved in UBE2O (Fig. 5a, c). LB is an insertion in proximity to the catalytic cysteine of UBE2Z, and LC replaces the conventional helix C of canonical E2s which is absent from UBE2Z, forming an extended loop that contacts the C-terminal extension of UBE2Z (Fig. 5a, d). Among these structural elements, loop LA emerged as a defining determinant for UBA6 recognition. In the UBA6–UBE2Z complex, LA projects toward the SCCH domain, where residues Asn172 and Thr174 form hydrogen bonds with residues Leu679 and Gly681 within UBA6 helices α21 and α22 (Fig. 5b, c). Substituting this segment with a poly-alanine stretch (GNNTV → AAAAA) nearly abolished thioester transfer activity, underscoring the functional importance of LA in stabilizing the E1–E2 interface (Fig. 5b, inset; Supplementary Fig. 4c). The spatial relationship between LA and α21–α22 suggests that UBA6 positions these SCCH helices in an orientation pre-optimized for class IV E2 engagement.

To test whether this arrangement is unique to UBA6, we compared the SCCH domain from UBA6 with those of other E1 enzymes, including UBA7, S. cerevisiae UBA1 (scUBA1), and S. pombe UBA1 (spUBA1) (Fig. 5f). In each case, E1 helices corresponding to α21–α22 of UBA6 extend inward toward the E2, resulting in steric clashes with the LA loop of UBE2Z (Fig. 5f). By contrast, both the UBE2Z-bound and apo UBA6 structures retain the same outward displacement of α21–α22, demonstrating that this configuration is intrinsic to UBA6 rather than induced by E2 binding (Fig. 5f). In addition to the LA–α21–α22 incompatibility, the LC loop of UBE2Z collides with the α21–α22 region of UBA7 and UBA1, creating a second steric barrier to class IV E2 binding (Fig. 5f). This dual clash would prevent proper alignment of the UBE2Z catalytic center and its LC insertion, precluding a productive E1–E2 configuration in non-UBA6 enzymes. Together, these comparisons indicate that UBA6 is structurally preconfigured to accommodate class IV E2s by maintaining an open SCCH geometry that prevents clashes and aligns the active site for catalysis. Differences in both sequence composition and secondary structure length within this region likely underlie these conformational distinctions (Fig. 5e, f). Furthermore, the InsP₆-binding pocket lies immediately adjacent to α21–α22 and may contribute to stabilizing its outward orientation. Thus, InsP₆ binding within this UBA6-specific basic pocket could help maintain the open SCCH cleft required for UBE2Z engagement (Fig. 5b, f, g).

To assess whether this orientation also permits engagement of other E2s, we docked CDC34 and UBCH5B, representative UBA1- and dual-specific E2s, onto the UBA6 structure (Fig. 5h, i). Both models fit readily within the same SCCH pocket, with the outwardly rotated α21–α22 helices providing sufficient clearance (Fig. 5h, i). These results are consistent with biochemical data showing that UBCH5B is readily charged by both UBA1 and UBA6, and that CDC34 retains partial activity when paired with UBA6 chimeras containing the UBA1 UFD (Fig. 4b).

Altogether, these analyses reveal that UBA6 achieves selective recognition of UBE2Z through a structurally pre-arranged SCCH domain that accommodates the distinctive LA and LC loops of class IV E2s without steric conflict, while maintaining favorable electrostatic complementarity (Supplementary Fig. 4e,f). This open and pre-stabilized SCCH geometry, supported by a nearby InsP₆ binding site, provides the mechanistic foundation for UBA6’s ability to engage UBE2Z and related E2s, distinguishing it from UBA1 and other E1s whose SCCH conformations would sterically or electrostatically exclude class IV E2s.

UFD–E2 Interface Features Defining E1–E2 Specificity

Having established that the SCCH domain of UBA6 provides a structural framework for accommodating class IV E2s, we next examined how the UFD contributes to E2 selectivity. Across characterized E1–E2 systems, the UFD serves as the principal docking platform for E2 engagement, providing electrostatic complementarity and orientational control of the catalytic cysteine. In UBA6, this role is executed by a compact acidic patch centered on the β30–β31 loop, which pairs with helix A of UBE2Z (Fig. 6a, b). The basic side chains of Lys105 and Arg106 on UBE2Z form salt bridges with Asp124 and Asp126 on UBA6, stabilizing the interface. This electrostatic complementarity positions UBE2Z precisely for thioester transfer and constitutes a key determinant of E2 recognition by UBA6.

Fig. 6. Structural basis of UFD–E2 interactions defining E1–E2 specificity.

a Electrostatic surface representation of the UBA6 UFD highlighting UBE2Z-interacting residues. Comparison models were generated by docking UBE2Z onto UBA7, UBA1, or the SUMO E1 based on their respective E2-bound structures. The UBA6–UBE2Z complex (left) serves as the reference. b Cartoon representations of UFD–E2 interfaces from UBA6–UBE2Z, UBA7–UBE2L6, UBA1–CDC34, and SUMO E1–UBC9 complexes, showing helix A and the corresponding β-loop regions (colored green). c Structure-based sequence alignment of Ubl E1 UFDs, centered on the acidic patch that mediates E2 binding. Secondary structure elements are indicated above the sequences; acidic residues are boxed in red. d Docking of UBE2L6, CDC34, and UBC9 onto the UBA6 UFD reveals steric clashes between the UBA6 β30–β31 loop and noncognate E2s. The UFD is shown as a cartoon, and E2 surfaces are colored by electrostatic potential. e Structure-based alignment of Ubl E2 regions that contact E1 UFDs, with basic residues in helix A highlighted in blue. Shaded boxes denote contacting residues across complexes. f Left, docking of BIRC6 onto the UBA6 UFD showing its N-terminal extension (green). Right, open-book electrostatic view of the interface, highlighting complementary charge features and labeled secondary structure elements.

The β30–β31 loop in UBA6 displays a unique composition and geometry compared with other E1 enzymes (Fig. 6b). All acidic contributors to the UBA6 patch reside within the loop itself, whereas in UBA7, UBA1, and UBA2, the corresponding acidic residues are distributed across the loop and neighboring secondary elements (Fig. 6b, c). Moreover, the UBA6 β30–β31 loop packs closely against helix A of UBE2Z, whereas the analogous β31–β32 loop in UBA7 is longer and adopts an orientation that leaves a gap between the E1 and E2 surfaces. In UBA1 and the SUMO E1, the β30–β31 and β12–β13 loops, respectively, extend away from their cognate E2 helices, generating less favorable contact geometry (Fig. 6a, b). These architectural distinctions translate directly into specificity. Docking UBE2L6 or CDC34 onto UBA6 reveals steric clashes at the UFD interface, while the distinct β1–β2 loop of UBC9 collides with the acidic surface of UBA6 (Fig. 6d). Thus, the UBA6 β30–β31 loop provides a structurally and electrostatically optimized interface that selectively accommodates class IV E2s such as UBE2Z.

Consistent with this specialization, helix A residue Lys105 of UBE2Z, a key participant in the salt bridge network, is not conserved among E2 enzymes with the exception of S. pombe UBC15 (Fig. 6e). Among surveyed orthologs, only S. pombe UBC15 retains a positively charged residue (Arg11) at the equivalent position (Fig. 6e), suggesting a conserved but restricted pairing mechanism. Docking analyses reinforce this view as superimposing UBE2Z onto UBA7 reveals steric clashes between the UBE2Z Lys105/Arg106 pair and the UBA7 acidic surface, while docking onto UBA1 produces a single clash at Lys105, with Arg106 rotated away from the patch (Fig. 6a). Similarly, modeling UBE2Z onto the SUMO E1 predicts electrostatic repulsion between the UBE2Z D124/T125 region and the negatively charged E1 UFD surface (Fig. 6a). Together, these comparisons highlight the matched electrostatic and steric complementarity between the UBA6 β30–β31 loop and the basic pair on UBE2Z helix A as a co-evolved recognition module that ensures selective UBA6–UBE2Z pairing.

To explore whether this recognition logic extends to other UBA6-specific E2s, we examined BIRC6, which, unlike UBE2Z, contains an extended N-terminal domain preceding the UBC core. Docking of BIRC6 onto UBA6 revealed an acidic region contributed by the BIRC6 N-terminal extension and β1–β2 loop that complements a basic patch on UBA6 encompassing helices α32–α33 (Fig. 6f). This additional contact surface provides a structural rationale for the dominant role of the UFD in defining BIRC6 specificity, consistent with the biochemical data showing that UBA6–BIRC6 pairing is primarily UFD-dependent (Fig. 4b–e).

Altogether, these analyses establish that the UFD of UBA6 encodes a dual-mode recognition surface tailored for class IV E2s: one sub-interface centered on the β30–β31 loop engages helix A of UBE2Z through a unique acidic-basic pairing, while a second basic patch near α32–α33 complements the extended acidic features of BIRC6. The combination of these electrostatic and steric modules allows UBA6 to recognize chemically diverse E2 partners while preserving specificity distinct from UBA1, UBA7, and other E1s. This modular strategy, together with the SCCH architecture described above, defines a cohesive two-domain framework for E2 selectivity within the UBA6 branch of the Ub activation system.

Architecture of a Double-Loaded UBA6-UBE2Z-FAT10(t)/FAT10(a) complex

To understand how the UBA6–UBE2Z complex accommodates the thioester transfer-ready FAT10, we examined the architecture of its catalytically engaged state. During the E1 reaction cycle, the enzyme transiently forms a ‘double-loaded’ intermediate in which one Ubl is noncovalently bound at the adenylation site (Ubl(a)) and a second is thioester-linked to the active-site cysteine (Ubl(t)). This configuration aligns the E1 and E2 catalytic centers and primes the complex for thioester transfer. To capture this intermediate, we reconstituted a disulfide-linked UBA6–UBE2Z–FAT10(t)/FAT10(a) mimetic and determined its cryo-EM structure at 2.73 Å resolution (masked FSC 0.143) (Fig. 7a, b; Supplementary Fig. 5 & 6; Supplementary Table 1). Both the NTD and CTD of FAT10 molecules adopt canonical β-grasp folds, although their NTDs were weakly resolved in the cryo-EM maps, consistent with conformational variability (Supplementary Fig. 5). Despite incorporation of the thioester-linked FAT10(t) mimetic, the overall architecture closely resembles the single-loaded UBA6-UBE2Z/FAT10(a) complex, indicating that dual modifier occupancy preserves the global E1–E2 arrangement described above.

Fig. 7. Architecture of the double-loaded UBA6-UBE2Z-FAT10(t)/FAT10(a) complex.

a Cryo-EM map (left) and model (right) of the double-loaded UBA6-UBE2Z-FAT10(t)/FAT10(a) complex. Domains are color-coded as indicated. Catalytic cysteines are shown as yellow spheres; AMP and InsP₆ are shown as cyan and orange spheres, respectively. b Cartoon representation of UBE2Z–FAT10(t) with the FAT10(t) NTD modeled. Right, open-book electrostatic surface maps showing complementary charge distribution at the FAT10(t)–UBE2Z interface. c Detailed view of the FAT10(t)–UBE2Z CTD interface with interacting residues shown as sticks; hydrogen bonds are indicated by dashed lines. Bottom, thioester transfer assays for UBE2Z and FAT10 mutants. Bars show mean ± s.d. of three technical replicates (individual values in black), normalized to WT. Source data are provided. d Comparison of LB-loop interaction surfaces. Electrostatic surfaces of UBE2Z–FAT10(t), UBCH5B–Ub, and CDC34–Ub complexes are shown, with FAT10 CYCIGG or Ub LRLRGG motifs overlaid. LB-loop boundaries are outlined in black (solid or dashed). e Conformational change of the UBE2Z LB loop upon UBA6 binding. The loop from the UBE2Z–FAT10(t) complex is superimposed on apo UBE2Z (PDB: 5A4P). Trp195 (sticks) rotates ~34° and shifts ~3.3 Å, exposing Cys188 (spheres). f Thioester transfer assays for the UBE2Z LB-loop mutant (WTGP → AAAA) showing reduced activity with both FAT10 and Ub. Bars show mean ± s.d. of three technical replicates (individual values in black), normalized to WT. Source data are provided. g Structure-based sequence alignment of CYCIGG- or LRLRGG-interacting regions across E2s, with UBE2Z secondary structure elements indicated above. Shaded boxes mark residues contacting FAT10 or Ub.

In the double-loaded complex, the FAT10(t) CTD engages UBE2Z on its front face in a closed conformation analogous to catalytically active E2–Ub(t) states (Fig. 7c, top; Supplementary 7a-c). The FAT10(t)CTD–UBE2Z interface is dominated by polar interactions centered on helix B of UBE2Z, contrasting the hydrophobic network of interactions observed in canonical E2-Ub complex structures. FAT10 residues S95, T133, and A159 contact UBE2Z S208, I211, Q214, and P117, while UBE2Z E219 forms salt bridges with FAT10 K143 and K137 (Fig. 7c,top). Additional hydrogen bonds reinforce the interface, including FAT10 R138 with UBE2Z S215 and FAT10 K137 with the UBE2Z D140 backbone carbonyl. Mutational analysis corroborated these observations as UBE2Z Q214K/I211K reduced thioester formation by ~90%, and E219K/D140K/F139A nearly abolished activity (Fig. 7c, bottom; Supplementary Fig. 3g, h). Reciprocal FAT10 variants K137E, F157A, and R138D caused severe defects, whereas K143E/E140A produced a partial but synergistic loss (Fig. 7c, bottom; Supplementary Fig. 3h, i). These results indicate that a polar β-grasp–to–hB interface defines the closed, catalytically competent FAT10(t)–UBE2Z complex and that electrostatic complementarity, rather than the hydrophobic patch typical of Ub(t) transfer, governs UBA6-dependent FAT10 conjugation.

Closer inspection of the complex revealed a dynamic role for the UBE2Z LB loop (residues 195–198; WTGP) in coordinating thioester formation. The FAT10(t) C-terminal tail (CYCIGG motif, replaced by SYSIGG in the analog) rests on a hydrophobic patch formed by the LB loop, which is conserved among UBA6-reactive E2s such as BIRC6 and UBE2O but absent from canonical UBA1 E2s like UBCH5B or UBC13 (Fig. 7d, g). Substituting this loop (WTGP → AAAA) sharply reduced thioester transfer for both FAT10 and Ub, demonstrating that the LB region facilitates catalysis independent of modifier identity (Fig. 7f; Supplementary Fig. 4d). Structurally, Trp195 undergoes a pronounced rearrangement upon FAT10(t) binding: relative to E1-free UBE2Z (PDB 5A4P), W195 rotates by ~34° and translates ~3.3 Å, shifting from a buried position shielding the catalytic cysteine (C188) to an exposed configuration that increases its solvent accessibility from 0.9 Ų to 25.3 Ų (Fig. 7e). This movement alleviates steric conflict with FAT10(t) and opens a channel for thioester formation, suggesting that LB-loop motion functions as a gating mechanism coupling E1 engagement to activation of the E2 catalytic center (Fig. 7e).

Although the cryo-EM density for the FAT10(t) NTD was too weak to justify its inclusion in the final atomic model, an AlphaFold-predicted NTD fit the residual density plausibly (Fig. 7b). Electrostatic surface analysis of this model revealed strong charge complementarity: a negatively charged patch on the C-terminal extension of UBE2Z aligns with a basic region on the FAT10(t) NTD (Fig. 7b). This acidic feature is absent from the corresponding surfaces of BIRC6 and UBE2O, and may contribute to the higher efficiency of FAT10 loading onto UBE2Z relative to BIRC6 (Fig. 1b; Supplementary Fig. 7d). Interestingly, the same basic patch of the FAT10 NTD also forms extensive electrostatic contacts with the adenylation domains of UBA6 in the FAT10(a)-bound complex, where it contributes to the specificity of the UBA6–FAT10 interaction³². Thus, our cryo-EM data and structural analyses suggest that the FAT10 NTD likely fulfills a dual mechanistic role in UBA6-dependent FAT10 conjugation by stabilizing E2 engagement through complementary electrostatics and reinforcing selective pairing with UBA6 through shared surface chemistry.

Because UBA6 activates both Ub and FAT10, we next sought to capture single- and double-loaded UBA6–UBE2Z–Ub complexes using the same strategy. Cryo-EM reconstructions were obtained at 3.14 Å resolution for the UBA6–UBE2Z/Ub(a) complex and 3.86 Å for the double-loaded UBA6–UBE2Z–Ub(t)/Ub(a) assembly (Supplementary Fig. 8a, b; Supplementary Fig. 9, 10, 11, 12). Comparison of the two models revealed no major architectural differences (Supplementary Fig. 8c). In the double-loaded state, both Ub(a) and Ub(t) were clearly resolved, although their local resolution was moderate (Supplementary Fig. 11). Ub(t) remained positioned adjacent to UBE2Z in a closed, catalytically poised conformation similar to that observed for FAT10(t). In contrast to the polar surface of FAT10(t) which engages hB of UBE2Z, the surface of Ub(t) that engages hB is largely hydrophobic. Among available double-loaded E1–E2–Ubl complexes, the UBA6–UBE2Z–Ub(t)/Ub(a) structure most closely resembles the UBA1–CDC34–Ub(t) complex (PDB: 7K5J), in which Ub(t) likewise adopts a closed conformation (Supplementary Fig. 8f). Relative to FAT10(t), the Ub(t) moiety in the UBE2Z–Ub(t) complex is rotated by ~14° and translated by ~3 Å, reflecting subtle adjustments that accommodate the smaller and more hydrophobic Ub surface (Supplementary Fig. 8d, e). These findings reveal that UBE2Z uses a structurally conserved but chemically adaptable binding surface to engage two distinct Ubls.

Together, these findings reveal that UBA6 and UBE2Z catalyze modifier transfer through a precisely choreographed double-loaded intermediate that aligns both Ubls for efficient thioester exchange. The polar β-grasp-to-hB interface imparts FAT10 specificity, while inducible movement of the UBE2Z LB loop transiently exposes the catalytic cysteine to enable thioester formation. Through subtle, context-dependent remodeling of its hB region and LB loop, UBE2Z is able to engage the more hydrophobic interaction surface of Ub. Thus, UBE2Z employs a conserved structural scaffold that flexibly accommodates both FAT10 and Ub, an uncommon property among E2 enzymes that reflects its evolutionary specialization within the UBA6 system. This dual reliance on electrostatic precision at the FAT10(t) NTD/UBE2Z and FAT10(a)/UBA6 interfaces and conformational flexibility at the β-grasp-to-hB interfaces provides a unified mechanism for UBA6-dependent activation of both Ub and FAT10, highlighting how shared architecture and substrate adaptability together define the catalytic versatility of this unique E1–E2 system.

Discussion

UBA1 and UBA6 define parallel branches of the Ub system that activate distinct sets of E2 enzymes. The cryo-EM structures and biochemical analyses presented here and in another study published while ours was under review⁶², reveal how these two E1s achieve selective E2 engagement through fundamentally different domain architectures. In the panel of E2s characterized in our study, UBA1 relies primarily on its UFD for E2 recognition, enabling interaction with broadly reactive E2s such as CDC34 and UBCH5. In contrast, UBA6 integrates its UFD and SCCH domains into a cooperative recognition module that specifies class IV E2s such as UBE2Z and BIRC6. This dual-domain mechanism provides a structural explanation for the non-overlapping E2 repertoires of the two E1s and illustrates how modular evolution of E1 architecture diversifies Ub conjugation pathways.

In UBA6, the UFD forms an acidic β30–β31 loop that engages the basic helix A of UBE2Z, establishing initial docking, while the SCCH domain supplies a pre-organized catalytic scaffold that accommodates the unique LA and LC loops of class IV E2s. Domain-swap experiments show that neither domain alone is sufficient for UBA6-type activity, emphasizing that specificity arises from the combined contributions of docking and catalytic alignment. Structural comparison further reveals that UBA6 maintains an expanded SCCH cleft (Supplementary Fig. 4e, f), stabilized by a nearby InsP₆–binding pocket, that prevents steric clashes with UBE2Z insertions, an intrinsic feature absent in UBA1. The presence of this cofactor highlights how small-molecule interactions can modulate E1 domain geometry, offering a broader principle for cofactor-assisted regulation in Ub-like systems. This pre-configured geometry ensures that UBA6 remains “E2-ready” for class IV enzymes without compromising the integrity of other Ub-like pathways.

The single- and double-loaded complexes capture UBA6 and UBE2Z across the thioester-transfer reaction and illuminate the local rearrangements that drive catalysis. E2 binding induces concerted shifts in the UFD and SCCH that align the active sites, while movement of the UBE2Z LB loop exposes its catalytic cysteine through a gating mechanism coupled to E1 engagement. Minimal global changes between adenylate and thioester states suggest that UBA6 achieves catalysis through precisely tuned local adjustments rather than large-scale domain motions. Functionally, these structural adaptations endow UBA6 with the ability to activate both Ub and FAT10. UBE2Z uses a common catalytic scaffold but remodels its hB region and LB loop to match the surface chemistry of each modifier, polar complementarity for FAT10 and hydrophobic packing for Ub. This context-dependent flexibility explains how UBA6 maintains dual-Ubl reactivity while preserving specificity across its dedicated E2 network. Lastly, it is interesting to note that the same basic NTD of FAT10(a) engages an acidic surface unique to UBA6 during FAT10 activation while that of the FAT10(t) is positioned in proximity to an acidic surface that is unique to UBE2Z. In contrast, Ub lacks a corresponding NTD. Thus, the positively charged FAT10 NTD plays an important role in mediating E1 recognition at the FAT10 activation step and also E2 recognition at the E1-E2-FAT10 thioester transfer step.

Together, these findings establish a comparative framework for understanding E1–E2 selectivity in the Ub system. UBA1 and UBA6 achieve fidelity through distinct strategies for the panel of E2s characterized in this study: UBA1 primarily by electrostatic recognition within its UFD, and UBA6 by a dual-domain code combining UFD docking with SCCH-mediated catalytic alignment. This architectural divergence underlies the emergence of a functionally specialized UBA6 branch capable of activating both Ub and FAT10, thereby linking canonical Ub signaling to immune-regulated proteostasis. More broadly, these results provide a model for how architectural specialization and cofactor integration within E1 enzymes expand the regulatory capacity of Ub/Ubl signaling networks.

Altogether, our results suggest that E2 specificity across ubiquitin and ubiquitin-like conjugation systems is not governed by a single universal mechanism. Instead, different E1 enzymes appear to encode partner selectivity through distinct combinations of docking geometry, catalytic alignment, and conformational pre-organization. In the case of UBA6, E2 recognition arises from cooperative contributions of the UFD and SCCH domains, supported by an intrinsically expanded SCCH architecture stabilized by an adjacent InsP₆-binding pocket. This contrasts with the predominantly UFD-driven specificity of UBA1 and illustrates how diversification of E1 domain organization can generate parallel signaling branches within the same modifier system. The structural and chimera-based framework established here provides a generalizable approach for dissecting E1–E2 specificity across the Ub/Ubl landscape and offers a conceptual foundation for understanding how E1 enzymes evolve specialized E2 networks while preserving pathway fidelity.

Methods

Cloning

Human UBA6 (residues 37–1052), UBA1 (residues 49-1058) and any variants were synthesized (Gene Universal) and cloned into modified vector pSMT3.4 with a N-terminal ULP1-cleavable SMT3 tag. Human E2s like BIRC6 (residues 4498-4820), UBE2Z (residues 93-354 or full-length), CDC34, UBE2B, UBCH5B and their variants, if any, were cloned into pSMT3.4 and pET28a. The DNA fragment encoding human FAT10 (C7T/C9T/C134L/C160S/C162S) was inserted into pGEX-6P1 vector with a N-terminal PreScission cleavable GST-tag. Ub was inserted into NcoI/XhoI sites of vector pET29NTEV with an N-terminal TEV-cleavable 6 × His tag. For structural studies, UBA6 (37–1052) was cloned into pFastBac-HTB (NcoI/NotI) for insect cell expression, with an N-terminal TEV cleavable 6X His tag. All mutations were generated using site-directed mutagenesis with specifically designed primers (Supplementary Table 2) followed by PCR amplification.

Domain-swapping experiments between UBA1 and UBA6 were designed guided by sequence alignments to identify conserved regions. A UBA6 UFD chimera was generated in which the UFD domain of UBA6 was replaced with that of UBA1 (aa 958–1058), and reciprocally, a UBA1 UFD chimera was constructed by replacing its UFD with that of UBA6 (aa 958–1052). Similarly, chimeras were generated in which the SCCH domain of UBA6 (aa 623–899) was replaced by the corresponding domain of UBA1 (aa 629–899), and vice versa. In addition, two double-domain chimeras were constructed in which both the UFD and SCCH domains were swapped between UBA1 and UBA6. All constructions were cloned into the pEG-BacMAM vector between the RsrII and NotI restriction sites, incorporating an N-terminal 6×His tag for purification.

Protein expression and purification

For all proteins produced in bacterial system, we use E. coli BL21(DE3) Codon Plus cells (Agilent; Cat.No.230280) to express the target proteins as described previously^33,63,64. Large scale cultures were grown at 37 °C in Luria Broth medium to nearly A600 OD 2.0, and then cooling down 4 °C placing the flasks on the ice with the addition of 1.5% ethanol(v/v). Protein was induced by the addition of isopropyl-β-D-thiogalactoside (IPTG) to a final concentration of 0.2 mM followed by shaking at 18 °C for 20 h. Human UBA6 protein purification from insect cell was carried out as previously described following minor modification³³. Briefly, high titer recombinant baculovirus produced from Sf9 were used to infect High Five cells (ThermoFisher Scientific; Cat.No. 11496015 and B85502, respectively) at a cell density of 1.5 × 10⁶ cells/ml. Cells were harvest by centrifuging at 2000 × g after 48 h of infection.

Bacterial- or insect cell-expressed UBA6 crude cell lysates in the presence of DNase and Lysozyme were prepared by sonication in lysis buffer (20 mM Tris Hcl pH 8.0, 350 mM NaCl, 10 mM Imidazole, 0.5 mM TCEP), followed by centrifugation at 100,000 × g for 45 min. The supernatant was applied to 5 ml Ni-NTA resin (Qiagen, 1018142), followed by sequential washes of the resin with 50 ml wash buffer same as lysis buffer except with 20 mM Imidazole, final protein evolution was by adding 25 ml Lysis buffer containing 250 mM Imidazole. The SMT3 or 6XHis tag was cleaved by adding ULP1 or TEV protease at a ratio of 1:1000 (w/w) and incubating overnight at 4 °C. Further purification involved subjecting the protein to Superdex 200 26/600 pg columns (GE Healthcare) using a buffer consisting of 20 mM Tris HCl pH 8.0, 350 mM NaCl, and 0.5 mM TCEP. The target proteins, if needed, were then pooled and subjected to a MonoQ anion exchange column (GE Healthcare) using buffer A (20 mM Tris HCl pH 8.0, 50 mM NaCl, 0.1 mM TCEP) and buffer B (20 mM Tris HCl pH 8.0, 1000 mM NaCl, 0.1 mM TCEP) for further purification. Human UBA1 was purified similarly to UBA6. All E2s like BIRC6, UBE2Z, CDC34, UBE2B, UBCH5B and their variants were purified using the same method as UBA6, except they were subjected to gel filtration chromatography using Superdex 75 26/600 pg columns (GE Healthcare). FAT10 was purified using GST-affinity chromatography followed by PrePresiccion cleavage and subjected to Superdex 75 26/600 pg gel filtration (GE Healthcare). Human Ub was purified using Ni-NTA affinity and Superdex 75 26/600 pg columns gel filtration (GE Healthcare). After purification, all the proteins were concentrated to 3-10 mg/ml, aliquoted, and snap frozen in liquid nitrogen.

UBA1/UBA6 domain-swapped constructs were expressed in Expi293F^TMcells (Thermofisher, Cat. No. A14635). Cells were transfected with 1 µg of plasmid DNA (per mL of culture volume) mixed with ExpiFectamine^TM 293 reagent at a 1:3 ratio in total of 3 × 10⁶/mL cells with ≥95% viability (37 °C, 8% CO₂). After 72 h post transfection cells were harvested, resuspended in ice cold lysis buffer (50 mM Tris-HCl pH 8.0, 300 mM NaCl, 10 mM imidazole, 0.5 mM TCEP, 0.1% NP-40, 5% glycerol, 100 µg/mL DNase I, 4 mM MgCl₂, and protease inhibitor cocktail). Cell lysate was subjected to followed by clarification through centrifugation at 18,000 rpm, 30 min at 4 °C. The supernatant was applied to Ni-NTA resin (Qiagen) and washed with buffer containing 30 mM imidazole. Bound protein was eluted with buffer containing 300 mM imidazole and subjected to HiLoad^TM 26/600 Superdex^TM 200 gel filtration column (cytiva) in buffer 20 mM Tris-HCl pH 8.0, 300 mM NaCl and 0.5 mM TCEP. The purified protein was concentrated (2–5 mg/ml), aliquoted and flash-frozen in liquid nitrogen.

E1-E2 thioester transfer assays

All E1-E2 thioester transfer assays were performed at 25 °C. Briefly, 0.1 μM E1, 0.4 μM E2, 2.5 μM FAT10 or Ub, 5 mM MgCl₂, 500 μM ATP, 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄ pH 7.4, 5% glycerol, and 0.1 mM TCEP. Reactions were initiated by addition of ATP and incubated for 30 s or 12 mins, before terminating it with the addition of nonreducing Urea SDS-PAGE buffer. The reactions were subjected to SDS-PAGE and run for 45 min at a constant voltage of 180 V on Any kD Mini-PROTEAN TGX precast gels (Bio-Rad). Gels were stained with Sypro Ruby (BioRad) and imaged using a BioRad GelDoc Go Gel Imaging system (12009077EDU). The intensity of the E2 ~ FAT10 or E2~Ub band was quantified by densitometry in ImageJ (Version 1.53t) and further processed in Prism 7a (GraphPad). Densitometry values were normalized to the percentage of the wild-type control from the same gel. All E1 and E2 mutants were analyzed under identical experimental conditions. Results are shown as the mean of two or three technical replicates, with error bars representing standard deviation. Uncropped images of representative gels used in biochemical assays are included in the Source Data file.

Generation of UBE2Z-FAT10 or UBE2Z-Ub thioester mimetic and purification

Briefly, 10 μM UBE2Z^{Y182K,C100S,C261S,C263S, K186R} and 20 μM FAT10 or Ub were added in a reaction mixture containing 50 mM Tris-Cl pH 9.3, 50 mM NaCl, 10 mM MgCl₂, 2 mM ATP, 0.3 μM UBA6 and incubated for 4 h at RT. The UBE2Z-FAT10(t) or UBE2Z-Ub(t) mimetic were separately purified via cation-exchange (MonoS 10/100; GE Healthcare) or anion-exchange chromatography (MonoQ 10/100; GE Healthcare), respectively. The purity of the peak fractions (containing UBE2Z-FAT10(t) or UBE2Z-Ub(t)) were assessed by SDS-PAGE and then concentrated to 4–7 mg/ml and snap frozen in liquid nitrogen in a final buffer composition of 20 mM HEPES (pH 7.5), 150 mM NaCl, 2 mM DTT.

Trapping of the single/double loaded UBA6 with FAT10 or Ub

To generate the single loaded complex, UBA6-UBE2Z/FAT10(a), UBA6 active Cys625 and UBE2Z’s active Cys188 were cross-linked using 2,2’-Dipyridyldisulfide, as previously described (add ref). Briefly, purified UBE2Z^{C100S, C261S, C263S, Y182A} (150-200 µM) was desalted into cross-linking buffer (20 mM HEPES pH 7.5, 75 mM NaCl) using a HiPrep 26/10 Desalting column. Purified UBA6 (50-100 µM) was also desalted into the cross-linking buffer. To the desalted UBE2Z^{C100S, C261S, C263S, Y182A} protein, equal volume of ‘activating buffer’ (20 mM HEPES pH 7.5, 75 mM NaCl, 2.5 mM 2,2’-Dipyridyldisulfide, 2.5% DMSO) was added and incubated at 25 °C for 10–15 min. This ‘activated’ UBE2Z^{C100S, C261S, C263S, Y182A} sample was subjected to desalting column equilibrated with cross-linking buffer to remove excess 2,2’-Dipyridyldisulfide and DMSO. The ‘activated’ UBE2Z^{C100S, C261S, C263S, Y182A} protein was mixed with the desalted UBA6 and incubated at 25 °C for 30 min in a molar ratio of 10:1. The cross-linking mixture was subjected to MonoQ anion exchange column (GE healthcare), to isolate purified cross-linked UBA6- UBE2Z^{C100S, C261S, C263S, Y182A} complex. In order to generate single loaded complex, UBA6-UBE2Z^{C100S, C261S, C263S, Y182A} complex was incubated with 1.2 molar excess of free FAT10, ATP and Mg²⁺ for 10 min at 4 °C. This mixture was further concentrated to 3 mg/ml for EM grid preparation. For double loaded complex, UBA6-UBE2Z-FAT10(t)/FAT10(a), a similar strategy was used except UBE2Z^{C100S, K130R, K166R, Y182K, K186R, C261S, C263S} -FAT10(t) mimetic was used instead of UBE2Z^{C100S, C261S, C263S, Y182A}.

For single loaded complex with Ub, UBA6-UBE2Z/Ub(a), a similar strategy was used as described above except UBA6-UBE2Z^{C100S, C261S, C263S, Y182A} complex was incubated with 1.2 molar excess of free Ub. However, for double loaded Ub complex, UBA6-UBE2Z-Ub(t)/Ub(a), UBE2Z^{C100S, Y182K, K186R, C261S, C263S} was used and subjected to MonoQ anion exchange column (GE healthcare), to isolate purified cross-linked UBA6-UBE2Z^{C100S, Y182K, K186R, C261S, C263S} -Ub(t).

Cryo-EM sample preparation and data acquisition

Freshly purified UBA6-UBE2Z/FAT10(a), UBA6-UBE2Z-FAT10(t)/FAT10(a), UBA6-UBE2Z/Ub(a) and UBA6-UBE2Z-Ub(t)/Ub(a) complexes were concentrated to around 20 μM and 3 μl of each was immediately applied onto cryo-EM Au grids (UltrAuFoil 1.2/1.3 300 mesh, Electron Microscopy Sciences) that had been glow discharged at 20 mA for 30 s in a Quorum EMS glow discharge. Grids were vitrified using a Vitrobot Mark IV (Thermo Fisher Scientific) maintained at 4 °C and 100% humidity. Grids were subsequently blotted for 3 s, with a blotting force of 10 and plunged into the liquid ethane. Grids were transferred into liquid nitrogen and stored at cryogenic temperatures before screening and data collection.

Data collection of UBA6-UBE2Z-Ub(t)/Ub(a) complex was completed at 200 kV on a Glacios cryo-TEM microscope equipped with a Falcon IV camera and Selectris energy filter (slit width: 10 eV) at UT Health San Antonio Cryo-EM center; Imaging was acquired using Thermo Fisher Scientific’s EPU software with AFIS model. A total of 1761 image frames with EER file format were recorded with an exposure time 7 s, a total dose of 53 e¯/Ų with a 5.65 e¯/pix/s dose rate. A total of 15,755 micrographs were acquired from UBA6-UBE2Z-Ub(t)/Ub(a), with a pixel size of 0.87 Å, a nominal magnification 130,000x, and a defocus range from −0.8 to −2.0 μm. Data collection for UBA6-UBE2Z-FAT10(t)/FAT10(a) and UBA6-UBE2Z/Ub(a) complexes were collected using a Titan Krios G3 microscope operating at 300 kV at UT Austin, Texas with K3 camera and Gatan Biocontinuum Imaging Filter (slit width: 20 eV). Imaging was acquired using SerialEM software with image-shift model. 50 TIF file format frames were recorded with an exposure time around 1.9 s, a total dose of 50 e¯/Ų with around 25 e¯/pix/s dose rate. A total of 27,654 and 14,879 micrographs were acquired from one Cryo grid of UBA6-UBE2Z-FAT10(t)/FAT10(a) and UBA6-UBE2Z/Ub(a), respectively, with a pixel size of 0.8332 Å, a nominal magnification 105,000x, and a defocus range from −1.0 to −2.5 μm. UBA6-UBE2Z/FAT10(a) data was collected using a Titan Krios G3 microscope operating at 300 kV at NYSBC with K3 camera and Gatan Biocontinuum Imaging Filter (slit width: 20 eV). Imaging was acquired using Leginon software with image-shift model. 50 TIF file format frames were recorded with an exposure time around 2 s, a total dose of 53 e¯/Ų with around 30 e¯/pix/s dose rate. A total of 14,879 micrographs were acquired from one cryo grid, with a pixel size of 1.067 Å, a nominal magnification 81,000x, and a defocus range from −0.8 to −2.5 μm.

Cryo-EM data processing

Data images were gain corrected, dose weighted and patch motion corrected using CryoSPARC 3.3.1⁶⁵. The exact defocus value of each micrograph was searched using CryoSPARC’s patch CTF estimation. For micrograph curation, a CTF cutoff value of 5 was applied to the data. Blob picker was used for picking particles that were subsequently extracted with a box size of 256 pixels. Three rounds of 2D classification were performed to remove junk particles and noise followed with 3D initial model reconstruction and 3D classification. Moreover, ab initio models generated were used as references for heterogeneous refinement. For each cycle of heterogeneous refinement, the particles from the best class were kept for the next round of heterogeneous refinement. For UBA6-UBE2Z-FAT10(t)/FAT10(a) complex, 4 rounds of heterogeneous refinement were performed with the best class containing 299,646 particles. This class was refined to 2.8 Å with homogeneous refinement followed by non-uniform refinement that yielded a map of 2.77 Å. Further local refinement was done that improved the resolution to 2.73 Å with 299,646 particles. The final resolution was evaluated based on FSC-0.143 criterion. A full processing tree is shown in Supplementary Fig. 6.

For all other datasets, similar strategy was used as above except another round of 3D classification was used after the homogenous refinement. A full processing tree for UBA6-UBE2Z/FAT10(a), UBA6-UBE2Z/Ub(a) and UBA6-UBE2Z-Ub(t)/Ub(a) are shown in Supplementary Figs. 2, 10, and 12, respectively. The sharpened map of each dataset using a negative B-factor estimated in CryoSPARC was used for model building.

Model building and refinement

For UBA6-UBE2Z/FAT10 and UBA6-UBE2Z/Ub structures, PDB 7SOL (UBA6), PDB 7PYV (FAT10), PDB 6DC6 (Ub), and PDB 5A4P (UBE2Z), Alphafold⁶⁶ models were rigid-body docked into the single and double loaded UBA6-UBE2Z cryo-EM maps using UCSF ChimeraX 1.5⁶⁷. This was followed by iterative rounds of real-space refinement as implemented in PHENIX V1.20.1-4487⁶⁸ followed by manual model building in COOT version 0.9.8.7⁶⁹. The structural model was validated as implemented by PHENIX. The cryo-EM maps and models were prepared with ChimeraX 1.5⁶⁷ and PyMOL version 2.0⁷⁰, respectively.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Protein purification was conducted by D.N., P.S.B., A.S., A.N., C.M.S., P.E., J.T.V., L.Y., F.G., and K.E.C. Structural experiments and analysis were conducted by L.J., E.A.R., D.N., A.A.T., E.V.W., and S.K.O. D.N. conducted biochemical assays. H.C., C.D., P.S., and M.U.G. assisted with experimental design and data interpretation. The figures and manuscript were prepared by D.N., E.A.R., and S.K.O. with input from all authors.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

Atomic coordinates for the UBA6-UBE2Z double-loaded FAT10, UBA6-UBE2Z single-loaded FAT10, UBA6-UBE2Z double-loaded Ub, UBA6-UBE2Z single-loaded Ub complex structures reported in this study are deposited in the RCSB with accession codes 9YKV, 9YLB, 9YKW, and 9YLF, and the corresponding cryo-EM maps have been deposited into the Electron Microscopy Data Bank with accession numbers EMD-73059, EMD-73079, EMD-73060 and EMD-73081. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-69882-3.