Extended Data Fig. 9: Structural comparison of E3 ligases.

a, There is a strong precedence for dimerization of RING/U-box domain E3 ubiquitin ligases^68,69,70. RING/U-box E3s exist both as homo- and heterodimeric complexes, for example, Rad18–Rad18, CHIP–CHIP, RNF8–RNF8, BRCA1–BARD1, RING1b–BMI1 and Hdm2–Hdmx^{17,18,19,20,21,71}. Structures of homo- and heterodimeric RING/U-box E3 ligases are shown here with the RING/U-box in orange. Surprisingly, these E3s display functional and structural asymmetry: in all the dimers listed above, only one protomer binds to an E2 enzyme. The homodimeric CHIP E3 ligase has a strikingly asymmetric structure that clearly demonstrates why only one U-box binds E2 enzyme¹⁹. The FANCL RING subunit is also an asymmetric dimer within the FA core complex and it is possible that only one of these binds E2. However, unlike the smaller E3 s, the FANCL RING fingers are not near each other. Together, this suggests that asymmetric dimerization may be a general feature of RING E3s. b, Comparison of FA core complex with cullin–RING ubiquitin ligases (CRLs). Many large complexes are predominantly helical suggesting that α-helices are commonly used as building blocks for complexes. In addition, β-propellers often mediate protein–protein interactions. The CRL complexes and FA core complex are long and extended with substrate-recognition (green), scaffold (yellow) and RING (orange) subunits residing in three different regions of the structure. However, the structural details of these complexes differ. Interestingly, the activities of some multisubunit RING-containing E3 ligases including APC/C and CRL complexes are stimulated by dimerization^22,23. Thus, dimerization may underpin physiological ubiquitination activity in many E3 ligases. c, d, Ubiquitin discharge assay, in which free lysine is used instead of the FANCD2–FANCI substrate. In these experiments, the FA core complex is incubated with E1, E2, ubiquitin and free lysine. If FANCL is active without a substrate, ubiquitin will be conjugated to lysine, resulting in a shift in its molecular weight; however, if substrate binding is required to activate the E3 ligase activity (for example, through allosteric changes), this will not occur. Coomassie gels of reaction products were run in non-reducing (c) and reducing (100 mM DTT) conditions (d). Ubiquitin is transferred to free lysine as shown by the increase in molecular weight of ubiquitin as well as a decrease in intensity of the E2–ubiquitin band when compared to the lane containing no free lysine. Thus, substrate binding is not required for activity. Reducing conditions do not eliminate the UBE2T–ubiquitin conjugate, as previously shown⁷². Additionally, DNA is not required for FA core complex E3 ligase activity on free lysines, suggesting that DNA activates the substrate, not the E3. The ubiquitin discharge assays were repeated three times independently with similar results (c, d). e, Distributions of patient mutations are indicated on the FA core complex by heat map colouring of subunits and in percentage. f, Ubiquitination assay using several subcomplexes (Extended Data Fig. 3b) and the full FA core complex, analysed by western blot with HA antibody to detect HA-tagged ubiquitin. The migration positions of monoubiquitinated FANCD2 and FANCI are indicated but FANCI is not substantially modified, as in Extended Data Fig. 1a. All complexes have similar activities but isolated FANCL is less active. This assay was repeated at least two times independently with similar results. For gel source data, see Supplementary Fig. 1.