We have identified an arrangement of five residues, which are perfectly conserved in a subclass of MPN domains, while none of them shows appreciable conservation outside of this subclass. The highly correlated conservation of these residues suggests that they participate in a common structural element and/or a common function that is critical for the proteins in which they are found. This idea is corroborated by the polar nature of the conserved MPN+ residues: glutamate, histidines, serine and aspartate are all amino acids frequently found in the active site of enzymes or as the coordinating ligands in metal-binding proteins. Certainly, these two possibilities are not mutually exclusive. Several classes of enzymes, particularly metal containing hydrolases and proteases, harbor bound metal ions such as Zn2+
as part of their catalytic center [30
]. However, while the motif does bear some resemblance to that found in metalloproteases, the specific organization of E–HxH–S–D residues does not correspond to the metal ligands in any of the known Zinc proteases [30
], thus positive identification of Rpn11 as a metalloprotease awaits definite proof.
The proteasome complex from a number of sources is known to contain a ubiquitin hydrolyzing activity [31
], while the CSN complex is known to cause the hydrolytic removal of the Nedd8/Rub1 ubiquitin-like molecule from the cullin subunit of the SCF ubiquitin ligase (E3) complex [40
]. Thus, the MPN+ containing subunits of the proteasome lid (Rpn11) and the COP9 signalosome (Csn5) would be prime candidates for such a hydrolytic function. The architecturally related eIF3 complex, for which no such enzymatic activity has been described, is conspicuously devoid of MPN+ proteins. That the CSN from yeast lacks the plain MPN protein (Csn6) retaining only the MPN+ subunit (Csn5), while the plain MPN proteins found in eIF3 from eukaryotes appear to be missing altogether in the yeast complex, emphasizes that the MPN+ residues are likely to be the catalytic residues, with plain MPN subunit playing a redundant structural role in complexes in which they are found. Even though purified recombinant Rpn11 does not appear to exhibit DUB capabilities (data not shown), it is possible that once incorporated into their respective complexes, Rpn11 and Csn5 confer the documented hydrolase activities onto the 19S RP and CSN. In this case, Rpn11 would belong to a unique class of enzymes, as all other known DUBs are cysteine proteases. Of note, the conserved cysteine residue common to both Rpn11 and Csn5, which is not part of the MPN+ motif, is not important for the function of Rpn11.
Single site substitutions in various Rpn11 MPN+ residues exhibit similar phenotypes supporting the identification of MPN+ as a discrete functional motif. Due to the severe growth phenotypes and attenuated ability of the proteasome to proteolyze polyubiquitinated substrates in these mutants, it appears that the MPN+ motif defines the role carried out by Rpn11 in the lid. As we show, these defects arise from an intrinsic activity of Rpn11 within the context of the proteasome, and not due to a gross structural effect upon incorporation of mutated Rpn11. As mentioned above, the lid, where Rpn11 is situated, is critical for proteolysis of polyubiquitinated substrates. So far, and somewhat surprisingly, all ubiquitin binding activity has been mapped to the Base of the 19S RP. Two subunits in the base can interact directly with ubiquitin chains, Rpt5 and Rpn10 [5
]. A number of proteins, such as Rad23 and Dsk2, can also bind ubiquitin and interact with the proteasome, presumably with the base, thus they are thought to serve as shuttles of polyubiquitinated substrates to the proteasome [45
]. It is possible that the lid rather than bind ubiquitin, serves to cleave or trim polyubiquitin chains once attached to the Base.
While this manuscript was under review, two independent papers substantiated our findings by characterizing a novel deubiquitinating activity in the proteasome, and showing that the MPN+ motif residues of Rpn11 are largely responsible for this activity [47
]. These independent studies identified DUB activity associated with the 19S RP of the proteasome, which is lacking upon lethal substitutions in Rpn11. Interestingly, all substitution mutations that we studied in the MPN+ motif of Rpn11 are viable. Quite possibly, Rpn11 is not the sole proteasome-associated DUB, and a number of DUBs play partially overlapping functions. For instance, Doa4/Ubp4 interacts weakly and substoichiometrically with the proteasome and may serve to release ubiquitin and regenerate the proteasome for the next catalytic cycle [37
]. The Ubiquitin-like domain (UBL) containing deubiquitinating enzyme, USP14, has been found to interact with the proteasome from mammalian sources [38
]. Ubp6, the budding yeast homolog of USP14, interacts with the proteasome as well [49
], and plays a role in proteasome-associated deubiquitination [39
]. It has also been reported that the Drosophila
DUB p37a [31
] and its homologs UCH37 (H. sapiens
) and Uch2 (S. pombe
), may be responsible for the polyubiquitin chain editing function associated with purified proteasomes [31
]. However, as budding yeast lacks an obvious ortholog of UCH37/p37a, other DUBs must play a greater role in proteasomes from this organism.
Finding MPN+ motif proteins in prokaryotes will help elucidate the origins of proteasome evolution and the function of Rpn11 in particular. The proteasome probably evolved from self-compartmentalized macromolecular proteases found in prokaryotes. Thus, proteasomal subunits, or proteins with motifs related to proteasomal subunits, are present in archaea and certain eubacteria. These "lidless" prototypes include the 20S CP subunits, and homologs of the base ATPases (Rpt subunits) [50
]. However, proteins relating to the ubiquitination process, such as ubiquitin itself, ubiquitin activating conjugating or ligating enzymes, and even lid subunits, are missing from prokaryotes. Identification of a motif from Rpn11 – a protein that is linked to processing of polyubiquitinated proteins – in prokaryotic proteins is an interesting development. It is possible that during evolution, the proteasome recruited an existing enzyme as the lid was forming into a regulatory module of the proteasome; likewise for the analogous CSN complex. Studying the prokaryotic MPN+ motif proteins should aid in elucidating what this motif does. The fact that these proteins are the shortest MPN proteins and correspond to the structural core region of the domain should greatly aid in enzymatic and structural studies, especially in comparison to members of the family such as Rpn11 and Csn5 that are naturally found only incorporated into complexes.
Interestingly, all identified prokaryotic MPN domains, both from archaea and from eubacteria, also contain the MPN+ motif. It thus appears likely that the ancestral MPN domain was of the MPN+ type, and that the extant 'plain' MPN proteins have lost this motif later on. It is possible that the plain MPN proteins play a structural role in complexes that additionally contain a MPN+ subunit. So far all known MPN proteins are incorporated into complexes that also contain PCI proteins [51
]. Interestingly, several of the prokaryotic organisms harboring MPN+ proteins appear to lack PCI domain proteins and therefore probably could not form large PCI/MPN complexes related to the lid and CSN particles. An interesting case is the tail assembly protein 'K' of the bacteriophage lambda (vtak) and its many homologs from other phages and prophages (Fig. ). Little is known about the specific function of the protein but none of the other tail assembly factors contain a PCI or MPN domain (K.H unpublished results). Our work suggests that the K-protein plays an enzymatic role in tail assembly rather than a merely structural one, although whether the analogy to Rpn11 is relevant to vtak is still enigmatic.
A similar situation exists for the orphan MPN proteins of eukaryotes. It is noteworthy that complex eukaryotic organisms have far more orphan MPN protein than orphan PCI proteins. However, the known PCI/MPN complex particles have stoichiometries that require more PCI components than MPN components. This apparent paradoxon can be resolved by a number of different but non-exclusive assumptions. i) Additional PCI proteins have eluded detection due to high sequence divergence allowing for more than three PCI/MPN complexes to exist, ii) alternative MPN proteins can interact with existing PCI complexes, iii) some MPN proteins have (enzymatic) functions outside of PCI/MPN complexes. Currently, there is no data available supporting the first two possibilities, although there is some promiscuity in the PCI protein interactions [52
]. The third possibility is likely to be true in prokaryotes and could also explain why all eukaryotic orphan MPN proteins belong to the MPN+ class. This could also explain how supra-stochiometric amounts of Rpn11 exhibit dominant phenotypes; possibly it can function outside of the proteasome as well.