The major enzyme system catalysing the degradation of intracellular proteins is the proteasome system. A central inner chamber of the cylinder-shaped 20 S proteasome contains the active site, formed by N-terminal threonine residues. The 20 S proteasomes are extremely inefficient in degrading folded protein substrates and therefore one or two multisubunit 19 S regulatory particles bind to one or both ends of the 20 S proteasome cylinder, forming 26 S and 30 S proteasomes respectively. These regulatory complexes are able to bind proteins marked as proteasome substrates by prior conjugation with polyubiquitin chains, and initiate their unfolding and translocation into the proteolytic chamber of the 20 S proteasome, where they are broken down into peptides of 3–25 amino acids. The polyubiquitin tag is removed from the substrate protein by the deubiquitinating activity of the 19 S regulator complex. Under conditions of an intensified immune response, many eukaryotic cells adapt by replacing standard 20 S proteasomes with immuno-proteasomes and/or generating the proteasome activator complex, PA28. Both of these adaptations change the protein-breakdown process for optimized generation of antigenic peptide epitopes that are presented by the class I MHCs. Hybrid proteasomes (19 S regulator–20 S proteasome–PA28) may have a special function during the immune response. The functions of other proteasome accessory complexes, such as PA200 and PI31 are still under investigation.
A prerequisite for cell viability is the presence of a set of thousands of precisely working proteins. Since there is no storage form of proteins, the synthesis of each protein is based on specific demands, depending on the developmental and environmental situation of a given cell. This implies that each protein is eliminated from a cell as soon as it is no longer needed, resulting in the phenomenon that each cellular protein has a specific half-life. While synthesis of all proteins is regulated on the level of gene transcription and mRNA translation, the mechanisms of their specific rate of degradation are less well understood; the discovery of the UPS (ubiquitin–proteasome system) as the major pathway for intracellular protein degradation was a big step forward. The UPS works almost hierarchically: the ubiquitinating enzymes select and mark the proteins committed for degradation, by conjugation with polyubiquitin chains; the proteasomes, generally designated 26 S proteasomes but appearing in various forms (such as proteasomes 26 S, 30 S and hybrid), fulfil this degradation. As protein degradation is irreversible, it is devastating to the cell if it proceeds in an uncontrolled manner. Therefore most proteolytic enzymes are regulated by protease inhibitors and activators, or are spatially separated from their potential substrates. With regard to proteasomes, both of these control mechanisms occur and enable all eukaryotic cells, as well as several archaeal and some eubacterial cells, to eliminate selectively proteins that are no longer needed to maintain cellular homoeostasis [1–4].
20 S, 26 S and 30 S proteasomes
The proteolytic enzyme that recognizes proteins committed for degradation by conjugation with polyubiquitin tags is called the 26 S proteasome. This high-molecular-mass complex is composed of a central cylindrical or barrel-shaped 11 nm × 15 nm core complex, the 20 S proteasome, which is built up by four stacked rings containing, in its inner hole, the active sites for catalysis of peptide-bond hydrolysis. The outer rings are composed of seven α-subunits, and the internal two rings each consists of seven β-subunits, resulting in an overall configuration of α1–α7/β1–β7/β1–β7/α1–α7 (Figure 1). Since the 20 S proteasome alone is unable to recognize and bind polyubiquitinated substrate proteins, 19 S Rps (19 S regulatory particles) bind to one or both α-rings of the 20 S proteasome and thus build up 26 S and 30 S proteasomes respectively (Figure 1). Whether 26 S proteasomes exist in vivo, or result from dissociation of one 19 S Rp from 30 S proteasomes during the purification process, is unknown. The mechanism of assembly of the 20 S proteasome and 19 S Rp is largely unknown; in addition to its ATP-dependency, the chaperone Hsp90 (heat-shock protein 90) seems to play a positive regulatory role, and PAAF-1 (proteasomal ATPase-associated factor-1) a negative role within this process [5,6]. The 19 S Rp has the ability to bind ubiquitinated as well as several non-ubiquitinated substrates and to pass them over to the 20 S proteasome for degradation .
The 19 S regulator complex
The 19 S regulator, also called cap, is composed of about 17 subunits, which assemble into a complex . This is subdivided into two subcomplexes, one of which, the base, consists of six proteins with ATP-hydrolysing activity, designated Rpt1–Rpt6 (Figure 1). These proteins belong to a family of about 100 different cellular ATPases containing one or two copies of a conserved amino acid motif called AAA (ATPase associated with various cellular activities) . Rpt1–Rpt6 are essential for the survival of yeast cells and form a hexameric ring that associates with the outer α-rings of the 20 S proteasome. Two additional components of the base are non-ATPase subunits, Rpn1 and Rpn2, the largest proteins of the 19 S Rp. The second subcomplex, called the lid, is added on top of the base. The lid is composed of eight different non-ATPase subunits (Rpn3, Rpn5–Rpn9, Rpn11 and Rpn12) and might be partly anchored to the base by Rpn10, because deletion of this subunit in yeast cells results in lability of the 19 S Rp. However, additional proteins seem to stabilize the lid–base interaction.
Subunit Rpn10 has been found to bind polyubiquitin conjugates, but as it is not necessary for the survival of a yeast cell, recognition and binding of ubiquitinated substrates cannot be confined to this subunit. One would expect a protein escorting substrates to the proteasome to have two essential properties: on the one hand, it must be able to recognize the substrate, for instance by binding polyubiquitin via a UIM (ubiquitin-interacting motif) or a UBA (ubiquitin-associated) domain, and, on the other hand, it should be able to bind to the 19 S Rp, e.g. via a UBL (ubiquitin-like) domain. Several proteasome-interacting proteins (proteins that are non-essential components of the purified 26 S/30 S proteasomes but are known to interact transiently with them) have been found to contain both UBA and UBL domains, e.g. Rad23, Dsk2 and Ddi1. These and other proteins may carry polyubiquitinated substrates to the 19 S Rp by binding to subunit Rpn1 or Rpn2, or to both  (Figure 2). Another possible acceptor site in the 19 S Rp is Rpn6, as it was found to interact with the ubiquitinated pro-transcription factor NF-κB (nuclear factor κB)/p100. Considering the tremendous number of different substrate proteins that are degraded by proteasomes, multiple carriers [e.g. Cdc48 (cell-division cycle 48)/valosin, Bag1 and several other proteins] may work in parallel, delivering their cargos to different acceptor sites depending on the length of the conjugated polyubiquitin chain . The ATPase subunit Rpt5 is also able to bind polyubiquitin chains, suggesting that all substrates are funnelled to a final gateway in the base . The proteins that have been found to be degraded by 26 S proteasomes without being tagged by polyubiquitin chains seem to be recognized by the same elements in the 19 S Rp lid that recognize ubiquitin conjugates .
Substrate unfolding and translocation
Tethering of a protein–polyubiquitin conjugate at one or several acceptor sites of the 19 S Rp allows its further processing (Figure 2). Since final degradation of substrate proteins takes place in the innermost central chamber of the 20 S proteasome, which is accessible only through a narrow pore in the outer α-ring and a channel leading through an antechamber between the α- and β-rings, substrate proteins have to be unfolded before they are transportable through this bottleneck . This sequence of events was actually shown to take place when the ATP-dependent proteasomal degradation of a folded, globular protein was studied in the presence of an archaeal 19 S base-homologous complex, PAN (proteasome-activating nucleotidase) . Unfolding is efficiently initiated when the substrate contains an unstructured sequence that is recognized as an engagement site by the 19 S Rp . It has been suggested that solvent-exposed hydrophobic β-strands of subunits Rpn1 and Rpn2 act as a binding platform for unfolded proteins . As the 19 S Rp, and especially the base, has been found to exert chaperone-like activity , the Rp may also work as a reverse chaperone or ‘unfoldase’, depending on how a protein is delivered and what happens subsequently to the unfolded protein. Since breaking of a hydrogen bond in a folded protein is energy-consuming, the unfolding process is probably one of the major ATP-dependent steps in proteolysis and there is experimental evidence to suggest that some of the ATPase subunits (Rpt1–Rpt6) co-operate to catalyse this process .
Although it has not yet been investigated in detail in eukaryotic proteasomes, it is conceivable that this unfolding process is closely connected to the mechanical translocation of the unfolded part of a substrate protein, from the base complex through a (presumptive) coaxial pore formed by the ATPase hexameric ring, and then through the central pore in the annulus, formed by the juxtaposed α-subunits of the 20 S proteasomes. Experiments with PAN and archaeal 20 S proteasomes have shown that translocation of substrates is also dependent on ATP hydrolysis  (Figure 2).
For many substrate proteins, the polyubiquitin chains not only are essential for their tethering to the 19 S Rp, but also might act as a support during the unfolding and translocation process. During or just after this process, ubiquitin is detached from the substrate protein, catalysed by subunit Rpn11 in the 19 S Rp lid (Figure 2). Rpn11 contains an intrinsic Zn2+-dependent metallopeptidase activity that hydrolyses the isopeptide bond between the substrate and the adjacent ubiquitin molecule and thus releases the entire polypeptide chain. The process is dependent on ATP hydrolysis . Some other DUBs (deubiquitinating enzymes) [e.g. Ubp6 (ubiquitin-binding protein 6) and UCH37(ubiquitin C-terminal hydrolase 37)] might have similar and additional functions; for example, deubiquitination of certain protein–ubiquitin conjugates to prevent their degradation, or depolymerization of the released polyubiquitin chain to fill up the free ubiquitin pool in the cell .
The 20 S proteasome
Molecular structure and gating of the proteolytic chamber
α- and β-subunits of the proteolytic core particle, the 20 S proteasome, evolved from the same predecessor. The most primitive archaeal proteasomes contain just one α- and one β-subunit and their subunit stoichiometry is therefore α7,β7,β7,α7. During evolution, each of these two subunits diverged into seven different α- and β-subunits, resulting in a stoichiometry of α1–α7/β1–β7/β1–β7/α1–α7 in eukaryotic 20 S proteasomes (Figure 1) . In principle, the molecular structure of all 20 S proteasome subunits consists of two layers of five-stranded β-sheets sandwiched by two helices at either side (Figure 3). Within the cylinder-shaped complex, the helices mediate the interactions between α- and β- as well as between β- and β-subunits. A major structural difference between α- and β-subunits is found in their N-terminal parts. The N-termini of α-subunits form an extra helix that fills the cleft between the two layers of β-sheets. The extreme N-terminal amino acids of the α-subunits form a lattice that gates a central pore within the ring of α-subunits. Although in archaeal 20 S proteasomes this pore is wide enough to allow entry of small peptide substrates but not proteins , free uncapped eukaryotic 20 S proteasomes are latent with regard to their peptide- and protein-degrading activity. In vitro, their gate can be opened by low concentrations of SDS, long chain fatty acids, hydrophobic peptides and other compounds . In vivo, the 19 S Rp unlocks this gate by docking to the 20 S proteasome in an ATP-dependent process (probably concurrent with phosphorylation of Rpt6) and thus putting the AAA-ATPase base-subunit Rpt2 in a position to replace the N-terminal tail of subunit α3, which interacts with the N-termini of all other α-subunits and forms the closing mesh [22,23]. Thus opened, the entryway allows passage of single-stranded unfolded proteins from either the C- or N-terminus as well as double-stranded stretched polypeptides . Substrates can enter through both α-pores of a proteasome at the same time and degradation products are released the same way .
Mechanism of protein fragmentation
β-Subunits lack this N-terminal extra helix. Instead, their N-terminal threonine residue is positioned just at the open cleft between the two layers of β-sheets and is exposed to the inner surface of the central chamber (Figure 3). This single threonine residue forms the proteolytically active site, because its serves as the catalytic nucleophile and as a primary proton acceptor (Figure 4). Therefore proteasomes belong to the group of Ntn (N-terminal nucleophile) hydrolases.
To reach this proteolytic cavity, an unfolded substrate has to travel the 8 nm-long distance from the α-pore through the antechamber by a mechanism that is not yet understood. Small synthetic chromogenic peptide substrates easily cover this distance by diffusion, and by use of these substrates it has been found that archaebacterial proteasomes preferentially hydrolyse peptide bonds at the C-terminal side of large hydrophobic amino-acid residues, thus exhibiting chymotrypsin-like specificity. However, within protein substrates, peptide-bond cleavage occurs after almost every amino acid.
During evolution, the number of active sites was reduced from 14 in archaeal proteasomes to six in eukaryotic proteasomes, since only subunits β1, β2 and β5 still contain functional active sites. Although they all use their N-terminal threonine for the nucleophilic attack on peptide bonds, their substrate specificity is determined by different substrate-binding channels, containing binding sites for peptides with a length of 7–9 amino acids. Their cleavage-site specificity can be roughly characterized as chymotrypsin-like (catalysed by subunit β5), trypsin-like (by β2), and post-glutamylpeptide hydrolysing or caspase-like (by β1) . When tested with protein or polypeptide substrates, proteasomes produce cleavage fragments with an average length of 7–8 amino acids, but their length ranges from 3 to 25 amino acids. These features are found in all proteasomes, independent of the content of their active sites. The length of the degradation products is most probably determined by the affinity of the substrate for the binding channels, the rate of peptide-bond cleavage, fragment release and its diffusion rate out of the complex. Many small products obtained in in vitro digestion experiments may result from re-processing of initial substrate fragments [23,27].
Proteasomes can be classified as endoproteases, since a double strand of an unfolded part of a protein that is threaded through the α-ring-gate into the proteolytic hole is attacked, via hydrolysis of an internal peptide bond, and cut into two fragments . This property, and possibly stop-transfer mechanisms, enables proteasomes to process proteins by limited proteolysis. Among the few examples known, the most well-studied example is p105, the C-terminal part of which is cleaved off to release the N-terminal part, p50, which is a subunit of the transcription factor NF-κB . Since it remains undigested, the N-terminal part leaves the proteasome at the same site as that where it entered. Degradation products have been found to take the same route, at least in 20 S proteasomes , a process possibly facilitated by the 19 S Rp-induced enlargement of the pores . Results recently obtained with yeast cell extracts suggest that substrate degradation and fragment release by 26 S proteasomes triggers an ATP hydrolysis-dependent release of 19 S Rp from 20 S proteasomes as well as disassembly of 19 S Rp into sub-complexes and Rpn10 . Whether this ‘chew and spew’ model, necessitating a continuous re-assembly of the 26 S complex, reflects the basic working mechanism of the 26 S proteasome awaits further experimental substantiation.
Biogenesis of 20 S proteasomes
Knowledge about the regulation of proteasome subunits at the level of transcription is relatively poor and confined only to yeast cells. The promoters of 26 out of 32 genes for subunits of the 20 S proteasome and 19 S Rp contain a PACE (proteasome-associated control element) sequence, to which Rpn4 binds and acts as a transcriptional activator. Because Rpn4 is degraded by the proteasome, a feedback regulation exists for transcription of most proteasomal genes . Neither PACE sequences nor an Rpn4 homologue have been detected so far in mammalian cells.
Our present knowledge about the biogenesis of the 20 S proteasome is more precise, especially with regard to archaeal proteasomes. Co-expression of α- and β-subunits in an Escherichia coli protein-expression system yields complete, active 20 S proteasomes. Since archaeal proteasome α-subunits, but not β-subunits, have been found to contain the intrinsic property of self-formation of heptameric rings, immature β-subunits need the α-ring for autocatalytic processing of their N-terminal propeptides and ring formation .
Biogenesis of eubacterial 20 S proteasomes that are composed of either one or two species of α- and β-subunits probably proceeds via the formation of α/β-dimers that assemble to inactive half-proteasome precursors (α7,β7), followed by autocatalytic processing of the N-terminal pro-peptide of the β-subunits, unveiling the N-terminal threonine residue and yielding active holo-proteasomes. In contrast to biogenesis of archaeal proteasomes, the presence (either cis or trans) of the β-subunit pro-peptide is indispensable for formation of eubacterial proteasomes .
While complete 20 S archaeal and eubacterial proteasomes can be generated by means of an E. coli expression system, the generation of eukaryotic proteasomes requires a eukaryotic Ump1/POMP (proteasome-maturation protein, also known as proteassemblin) and chaperones (e.g. Hsc73) and thus cannot be performed successfully using bacterial protein expression systems. The earliest natural intermediate identified so far is an α-ring associated with subunits β2, β3 and β4. The next stable intermediate is a half-proteasome composed of an α-ring, a ring of all N-terminally unprocessed β-subunits and Ump1/POMP. Dimerization of two half-proteasomes results in encasement of Ump1/POMP, which thereupon interacts with the (pro-)subunit β5 and induces (auto)-catalytic processing of all pro-β-subunits. This processing step allows subsequent formation of the mature 20 S proteasome, the first substrate of which is Ump1/POMP .
Special proteasomes for special tasks
As oligopeptides are the major reaction products of proteasome action, Nature has taken advantage of these degradation fragments by using them as a source of antigenic epitopes of cellular proteins, and presenting them (by class I MHCs) to T-cells for immune surveillance in higher eukaryotes. When cells present epitopes, which may result from the proteasomal breakdown of a foreign, e.g. virus-encoded, protein, these are recognized as non-self and the cell is eliminated by the T-lymphocyte. Although proteasomes that are constitutively expressed in most somatic cells are able to generate these antigenic peptides (or at least their precursor fragments), cytokines (especially interferon-γ), released by activated T-lymphocytes, induce the expression of three alternative active-site-containing 20 S proteasome subunits, designated β1i, β2i, and β5i, that replace the standard subunits in nascent proteasomes, resulting in the formation of ‘immuno-proteasomes’ (Figure 5). Biogenesis of immuno-proteasomes is faster than that of standard proteasomes owing to a favoured interaction between Ump1/POMP, the expression of which is also enhanced by interferon-γ, and subunit β5i (as compared with β5). Incorporation of β5i favours the subsequent integration of β2i and β1i, thus enabling an infected cell to process antigens in a rapid and efficient manner. Since immuno-proteasomes have a half-life of about 1 day and standard proteasomes one of about 5 days, interferon-γ might trigger a transient and limited change to the type of proteasomes within a cell . Spleen cells and other cells involved ‘professionally’ in immune surveillance, however, constitutively express immuno-proteasomes. Although some of the cleavage preferences of immuno-proteasomes are the same as those of standard proteasomes, generation of certain virus-derived antigenic epitopes is improved or is even dependent on immuno-proteasomes. The substrate-binding sites of immuno-proteasomes probably facilitate a cleavage preference for certain peptide bonds that results in the production of a critical concentration of an antigenic epitope (and/or its precursor) that is high enough to induce a T-lymphocyte response. On the other hand, it has also been found that immuno-proteasomes fail to produce certain self-antigens that are more easily generated by standard proteasomes .
Although constitutively present in many mammalian cells, interferon-γ causes an intensified expression of another protein, called PA28 or 11 S regulator. This is an approx. 200 kDa complex of two different proteins, PA28α and PA28β, the expression of which can be differentially regulated owing to the presence of an NF-κB binding site that is present only in the promoter sequence of PA28β . X-ray crystallographic studies with the homomer of PA28α revealed a heptameric barrel-shaped structure and the most likely composition of native PA28 is α4β3 or α3β4. The openings of the barrel are 2 and 3 nm in diameter. With the latter opening ahead, the complex docks on to one or both of the outer α-rings of the 20 S proteasome — a process executed without ATP consumption (Figure 5). This interaction presses the N-terminal tails of the proteasome α-subunits upwards, resulting in an opening of the pore. Thus, in vitro PA28 activates latent 20 S proteasomes by facilitating the entry of small peptide substrates and the release of products . The in vivo function of PA28 is less clear. It has been shown that certain cells generate certain antigen epitope precursors only in the presence of PA28 . Whether this results from a facilitated release of longer degradation products owing to the opened α-pores, or whether PA28 induces allosteric changes within the proteasomes, allowing a co-ordinated dual- or singly-modified cleavage within a substrate protein, is still a matter requiring investigation .
As determined in HeLa cells, a remarkable number (approx. 40%) of total cellular proteasomes are free 20 S proteasomes, which allow the binding of 19 S Rp (approx. 15%), PA28 (approx. 20%) or both (Figure 5). In fact, about a quarter of all proteasomes in these cells were calculated to have bound 19 S Rp at one side and PA28 at the other, and their cellular concentration increases when the cells are treated with interferon-γ, probably owing to the induction of PA28 expression. These enzymes are called hybrid proteasomes . They degrade the protein ornithine decarboxylase in a ubiquitin-independent, but ATP-dependent, manner and generate a pattern of degradation fragments different from that generated by 26 S proteasomes . The exact biological function of hybrid proteasomes is still unknown.
Additional accessory parts
Since proteasomes degrade substrates from almost all cellular compartments except mitochondria, they have been detected not only in the cytosolic compartment but also attached to the ER (endoplasmic reticulum) and within cell nuclei (except nucleoli) . Even without being polyubiquitinated, several substrates seem to be degraded by the 26 S or 20 S proteasomes . In certain organisms, e.g. Ascidia, proteasomes perform special, even extracellular, functions such as fertilization .
To fulfil all these diverse functions, proteasomes have acquired several additional attachments . For example, a protein called PA28γ that seems to be the ancestor of PA28α and PA28β, but is not inducible by interferon-γ, forms homo-heptamers and binds to 20 S proteasomes, but activates predominantly the trypsin-like activity . It is present in high concentrations in the nuclei of brain cells and might function as an adaptor of proteasomes to other complexes, e.g. apoptosis regulators .
Also present in the nuclei of mammalian cells is a single-chain protein, called PA200. This protein has an asymmetric dome-like shape and binds to one or both α-rings of 20 S proteasomes (Figure 5). It has been shown to activate the hydrolysis of small peptide substrates by 20 S proteasomes, but its homologue (Blm3) in yeast cells may not. A possible function of PA200 to recruit proteasomes to enzyme complexes catalysing DNA repair has been questioned .
A proline-rich 30 kDa molecule has been found in vitro to inhibit the chymotrypsin-like and caspase-like activities of 20 S proteasomes, but is less effective towards the trypsin-like activity. It is designated proteasome inhibitor PI31 (Figure 5). The inhibitor, which is not present in yeast cells, is located predominantly at the nuclear/ER membranes and was found not to impair the overall cellular ATP/ubiquitin-dependent protein breakdown, but was suggested to be involved in the biogenesis of immuno-proteasomes .
Whether other complexes, such as PC530 (found in starfish oocytes)  and COP9 signalosome , can replace the 19 S Rp or its lid-subcomplex respectively, to enable proteasomes to fulfil such special functions, is still under investigation.
Although only discovered about 20 years ago, intensive studies of proteasomes from archaeal, eubacteria and eukaryotes have lead to deep insights into the structure and function of these nano-machines. In particular, the fact that proteasomes, as the molecular ‘mincers’ of the UPS, are involved in the regulation of many basic cellular functions and pathological malfunctions has attracted the inquisitiveness of investigators from biomolecular, biomedical and pharmaceutical research areas. Nevertheless, detailed knowledge about the function of most of the 19 S regulator Rpt- and Rpn-subunits is still lacking, as are the detailed mechanisms of substrate unfolding and translocation. On the basis of results obtained by two-hybrid screens and other investigations, it is not unreasonable to suggest that a plethora of proteasome-interacting proteins deliver substrates to proteasomes and direct, as well as furnish, the machinery for special functions at various subcellular locations [29,46–48]. Even basic knowledge about the exact functions of PA28, PI31 and PA200 and the molecular mechanism of how immuno-subunits change the enzymatic properties of 20 S proteasomes have yet to be elucidated, not to mention the the co-operation of the proteasomal system with other parallel-working and post-proteasomal proteinases in the cell. These and many other questions will attract the interest of hosts of investigators to proteasomes in the future.
20 S proteasomes are intracellular, multi-subunit, cylinder-shaped complexes with an interior cave containing proteolytically active sites mechanistically belonging to the N-terminal threonine hydrolases.
A major part, but not all, of their substrates are protein–polyubiquitin conjugates. These are trapped by the 19 S Rps, which are added to one or to both ends of the 20 S proteasome cylinder, to form 26 S and 30 S proteasomes respectively.
Driven by the energy of ATP hydrolysis, 19 S Rps unfold, deubiquitinate and translocate substrate proteins into the proteolytic cavity of 20 S proteasomes.
Proteolytic products of proteasomes are oligopeptides of 3–25 amino acids, a small fraction of which are used by higher eukaryotic cells as protein epitopes (or their precursors) for presentation to the immune surveillance system.
For the generation of immuno-epitopes, cytokines can induce many eukaryotic cells to generate 20 S immuno-proteasomes that contain special subunits with optimized substrate-binding pockets. Additionally, and alternatively, docking of a proteasome activator, PA28, can affect the type and length of degradation fragments.
Several other activating and inhibiting accessory molecule complexes modulate the action of proteasomes.
The author thanks Dr. R. Preissner and Dr. A. Goede for preparing the schematic drawing of the structure of 20 S proteasome subunits. The author's own contributions to the exploration of proteasomes were, and still are, supported by grants (SFB 113; DA146/1-6; SFB 507) of the Deutsche Forschungsgemeinschaft (Bonn, Germany). I apologize for not citing many other relevant publications owing to publisher's limitations.
- © 2005 The Biochemical Society