In recent years, the ubiquitin system of intracellular protein degradation has been implicated in the control of many fundamental cellular processes — from cell cycle to apoptosis and oxygen homeostasis to inflammation. In addition, defects in this system seem to be directly linked to the development of human diseases, including cancer and neurodegenerative diseases. It is therefore not surprising that the mechanisms underlying regulated ubiquitin-mediated protein degradation are continuing to attract the attention of a broad research community.

During ubiquitination of substrate proteins, ubiquitin is passed from an E1 ubiquitin-activating enzyme to an E2 ubiquitin-conjugating enzyme. The final step of this process is the ligation of ubiquitin to the substrate, catalysed by an E3 ubiquitin protein ligase. The class of E3 enzymes are of particular interest, as they mediate the crucial step of substrate recognition1. As the stability of a large number of proteins is controlled by the ubiquitin system, it is of crucial importance to determine how the the cell achieves sufficient diversity among E3s so that each one selectively recognizes only one or a few substrates in the sea of cellular proteins present at any time. One answer to this seems to be a superfamily of multicomponent, cullin-based, E3 ubiquitin protein ligases, which recruit protein substrates to a core ubiquitination machine through variable substrate-specific adaptor proteins. Work published in this issue of Nature Cell Biology2 (and in the recent issues of Nature3,4 and Molecular Cell5), report on the identification of members of the BTB-domain protein family as substrate-specific adaptors of Cul3-based E3 ubiquitin ligase complexes, and thus fill an important gap in our understanding of the molecular mechanisms that underlie substrate targeting of cullin-based E3 ligase complexes.

The archetype of cullin-based E3s are the Skp1–cullin–F-box (SCF) complexes, which consist of a core complex composed of Cul1 (or CDC53 in budding yeast), the RING protein Rbx1/Roc1/Hrt1 and Skp1, and any one of a large number of substrate-specific adaptor subunits known as F-box proteins6,7,8. F-box proteins recognize different substrates through specific protein–protein interaction domains and recruit them to the core catalytic complex through the F-box motif, which is a binding site for Skp1. Within this core complex, Skp1 binds directly to Cul1, which recruits the RING protein Rbx1/Roc1/Hrt1 and an E2 enzyme (Fig. 1). This combinatorial theme is recapitulated in the closely related Cul2-based E3s9. Here, the core ubiquitination complex is composed of Cul2 (or Cul5), which binds the RING protein Rbx1/Roc1/Hrt1 and the Skp1-related protein elongin C. In turn, elongin C interacts with one of a diverse group of BC-box-containing proteins, such as the von Hippel–Lindau tumor suppressor, pVHL (ref. 10), or the suppressor of cytokine signalling (SOCS) family of proteins11, which provide substrate specificity analogous to F-box proteins (Fig. 1). Although the molecular composition and protein substrates of Cul1- and Cul2-based E3s are beginning to be elucidated, to date little is known about the substrate-targeting mechanisms and substrates of the other members of cullin-based E3s.

Figure 1: Specific classes of substrate-specific adaptors serve distinct cullin-based E3s.
figure 1

The archetypal SCF complex contains Cul1/Cdc53, Skp1 and Rbx1/Roc1/Hrt1 and an F-box protein (a). Analogous proteins assemble with parallel architecture in the Cul2, elongin C/B, Rbx1/Roc1/Hrt1 and BC-box protein complex (b), as well as in the newly identified Cul3, Rbx1/Roc1/Hrt1 and BTB-domain protein complex (c). Note that unlike F-box and BC-box proteins, which interact with cullins through the linker protein Skp1 and elongin C, BTB proteins seem to associate directly with Cul3 through their BTB domain. The use of different substrate adaptors in complexes with distinct cullin-based E3 complexes allows for binding to selected protein substrates, and may underlie the diversity of regulated ubiquitination. SCF and SCF-like substrates are generally recognized by their cognate adaptor in the manner involving prior post-translational modification of the substrate, for example, phosphorylation a or prolyl-hydroxylation b. Whether binding of the BTB protein MEL-26 to MEI-1 involves such modifications or additioinal ones is not known at present.

In a new study on page 1001 of this issue, Furukawa et al.2 employed a biochemical approach to identify novel Cul3-interacting proteins. They immunopurified human Cul3-containing complexes from whole cell extracts and identified co-immunoprecipitating proteins by mass spectrometric analysis. Among the proteins Furukawa and colleagues identified were two closely related, but previously uncharacterized proteins that contain a BTB domain and six Kelch repeats. The BTB domain has been initially identified in the Drosophila melanogaster transcriptional repressors broad complex, tramtrack and bric-a-brac, and can be found in approximately 200 human proteins, whereas the Kelch domain is thought to mediate protein–protein interactions. Deletion analysis of Cul3 identified a BTB protein-binding site on the amino terminus of Cul3. Interestingly, the analogous set of sequence is used in Cul1 and Cul2 to interact with Skp1 and elongin C, respectively, and Cul3 bound to BTB proteins in a BTB domain-dependent manner. However, this interaction seems to occur in the absence of any linker protein, suggesting that the BTB domain may contact Cul3 directly. These observations, in addition to the observation that many distinct BTB domain proteins interact with Cul3 (and not with other members of the cullin family) increased the speculation that BTB proteins may function — analogous to F-box and BC-box proteins — as specificity determinants of Cul3-based E3 ligase complexes. However, although F-box and BC-box proteins interact with their cullin partners through Skp1 and elongin C respectively, BTB domain proteins do interact with Cul3 directly, arguing that BTB proteins combine the properties of Skp1/F-box proteins and elongin C/BC-box proteins into a single polypeptide (Fig. 1). Consistent with this view, Skp1, elongin C and the BTB domain display similar three-dimensional structural features12,13.

In a separate study, Pintard and colleagues4 applied a genetic approach to search for novel components of the Caenorhabditis elegans Cul3-based ubiquitin ligase. Cul3-based E3 activity has been implicated in the degradation of MEI-1, a subunit of the katanin-like microtubule-severing heterodimer MEI-1–MEI-2 (ref. 14). As MEI-1 is required for meiotic spindle function, but is inhibitory to mitotic spindle function, its degradation is critical for the assembly of a functional mitotic spindle at the meiosis-to-mitosis transition. A mutant was identified that displayed similar phenotypes to those seen after reducing CUL-3 function in C. elegans through RNAi. This implies that this mutation may affect the function of a gene whose product participates in the MEI-1 degradation pathway. This was confirmed in a series of experiments, as MEL-26 — a protein previously demonstrated to be required for MEI-1 elimination — was identified and its interaction with CUL-3 verified in vivo14. In addition, MEL-26 contains a BTB-domain and a meprin and TRAF homology (MATH) protein–protein interaction domain. Structure–function analysis demonstrated that the aforementioned domains of MEL-26 are critically important in the binding of CUL-3 and MEI-1. Hence, MEL-26 functions analogous to an F-box and BC-box protein as a substrate-specific adaptor of a CUL-3-based E3 ligase, as it recuits the substrate MEI-1 through its MATH domain and links up to CUL-3 through its BTB domain (Fig. 1).

In parallel, Xu and co-workers3 were pursuing interacting proteins of C. elegans CUL-3 by screening C. elegans two-hybrid cDNA libraries and the ORFEOME library. They identified a host of CUL-3 interactors, all of which encoded proteins containing a BTB domain. However, only one of these BTB proteins — MEL-26 — had been previously characterized. Most intriguingly, a further two-hybrid screen with MEL-26 identified MEI-1. Subsequent biochemical studies demonstrated comprehensive and detailed insights into the structure–function relationships underlying the interactions of CUL-3, MEL-26 and MEI-1. Their studies highlight the conceptual similarity among cullin-based E3s and support the view that a host of cellular proteins may be targeted for ubiquitin-mediated proteolysis by distinct BTB-protein family members (Fig. 1). In keeping with this proposal, Geyer et al.5 demonstrate that several BTB proteins are associated with CUL-3 in Schizosaccharomyces pombe and that at least one is itself a target of the bound CUL-3.

The genetic and biochemical evidence provided in these reports argues that MEL-26 functions as part of an Cul3-based E3 to promote the poly-ubiquitination, and thus degradation, of MEI-1. Direct support for this proposal is provided by Furukawa et al.2. In an attempt to reconstitute MEI-1 ubiquitination by recombinant MEL-26–CUL-3 complexes, they immunopurified MEL-26–CUL-3 complexes from transfected human cells and used this complex as a source of E3 ligase for ubiquitination reactions in the presence of MEI-1 also produced in human cells. Ubiquitin conjugates of MEI-1 were readily detected, implying that in the nematode, MEL26 may function directly to eliminate MEI-1 through the recruitment of a CUL3-based E3.

A common feature of ubiquitination reactions driven by SCF or SCF-like E3s is that substrate recognition necessitates prior post-translational modification of the substrate by, for example, phosphorylation or even prolyl-hydroxylation7,8,10. The reconstitution of MEI-1 ubiquitination has not been achieved with purified components, but rather with components isolated from transfected human cells, making it likely that one or more of these components undergo modification or that potential essential cofactors are co-purified or both. Identifying potential cofactors for BTB protein-containing Cul3-based E3s and defining possible modifications required for substrate recognition will certainly be a major challenge for the future. Another major challenge will be to match targets and pathways with the multitude of BTB proteins and to uncover potential functions of BTB-driven ubiquitination in human disease.