Cell biology

Destruction deconstructed

Article metrics

Correctly dismantling a structure can be as challenging as assembling it. The architecture of the yeast proteasome reveals this enzyme's intricate machinery for protein degradation. See Article p.186

The ensemble of proteins that makes up a cell is constantly changing. The malleability of a cell's protein content on a short timescale is largely due to the proteasome — a complex member of the protease family of enzymes, which break down proteins. The proteasome, found in eukaryotic organisms (such as plants, animals and yeast), is composed of two main parts: the 20S core particle, which is a 28-subunit protein-destruction complex, and the 19S regulatory particle, a 19-subunit complex that mediates substrate selection. On page 186 of this issue, Lander et al.1 report the complete subunit structure of the yeast 19S regulatory particle. The results include many surprises and some puzzles that will challenge the proteasome field.

Unlike extracellular proteases (Fig. 1a), many of which break down proteins almost indiscriminately, the major intracellular proteases are large and highly selective protein complexes. The eukaryotic proteasome seems to have evolved from a protease known as PAN (or something comparable to this enzyme), which is found in microorganisms called archaea (Fig. 1b). The proteolytic sites of both PAN and the proteasome are hidden in an internal compartment, which a substrate protein can reach by passing through a narrow pore; the pore prevents the entry of properly folded proteins. Movement and unfolding of the substrate require energy, which the enzymes derive from the hydrolysis of ATP molecules2. The ATP-hydrolysing components, called ATPases, form a ring directly atop the pore. Any protein that passes through the pore is unlikely to survive the journey.

Figure 1: The evolution of proteases.

a, Trypsin, an example of a typical extracellular protease, with its proteolytic active site shown in red (model taken from ref. 13). b, Central cross-section of a model4 of the archaeal PAN complex showing the six-subunit ATPase with its N-ring and the 28-subunit core particle, together with its proteolytic active sites (red). To model the structure, the ATPase, N-ring and core particle were manually placed in proximity. The substrate-translocation channel of PAN (yellow) has the entry port at its top. c, Surface of the eukaryotic proteasome, as obtained by Lander et al.1, showing the core particle and the ATPase (Rpt) ring. Subunits comprising those parts of the regulatory particle that are specific to the eukaryotic proteasome are shown in orange.

In simple ATP-dependent proteases such as PAN, the ATPases select substrates for degradation from the pool of cellular proteins. In eukaryotes, however, chains of a protein called ubiquitin 'tag' those proteins that are destined for degradation3. Ubiquitin is then recognized by the regulatory particle of the proteasome, and the substrate is channelled into the core particle. A large family of ubiquitin ligase enzymes, which probably numbers more than 500 in humans, can thus condemn specific proteins by attaching ubiquitin groups to them. This dramatic evolutionary elaboration of the protein-degrading machinery is reflected in the fact that the proteasome has assumed regulatory functions in virtually all aspects of eukaryotic cell biology.

The evolution of ubiquitin tagging also coincided with a transformation of the proteasome's structure. In the PAN complex, the regulatory particle is composed solely of the ATPase ring4,5 (Fig. 1b). By contrast, the eukaryotic proteasome contains 13 additional subunits (Fig. 1c), nine of which make up the proteasome lid, with the other four, in combination with the ATPase (Rpt) ring, forming the base3.

Lander and colleagues' study1, together with three other recent papers6,7,8, provides a comprehensive picture of the proteasome components that are specific to eukaryotes (Figs 1c and 2). Of particular interest are the ubiquitin receptors and deubiquitinating enzymes, as well as the placement of these elements with respect to the substrate-translocation channel of the ATPase ring. At the top of the channel is the N-ring, which is thought to be the entry port (Fig. 2a) for the substrate4,5. Deep within the channel are the pore loops of the Rpt proteins, which are proposed to make contact with substrate proteins and use ATP-driven movements to inject them into the core particle, as has been shown for simple ATP-dependent proteases2.

Figure 2: Subunit architecture of the yeast proteasome regulatory particle1.

a, Tilted view over the Rpt ring (blue) between the Rpt4/5 (right) and Rpt1/2 (not visible) coiled coils. The core particle is shown in green. The centre of the N-ring constitutes the presumed substrate entry port. Highlighted in various colours are the ubiquitin receptors Rpn13 and Rpn10, with its ubiquitin-binding element (Rpn10-UIM), and the deubiquitinating enzymes Rpn11 and Ubp6 (Ubp6 model taken from ref. 14). The position of Ubp6 is approximate and was set manually on the basis of Lander and colleagues' structure1. All other subunits are in grey. Free ubiquitin (upper right) is shown for comparison. b, Lateral view of the proteasome with the lid (grey) turned to the right. Also visible are the base subunits Rpn10, Rpn13, Rpn1, Rpn2 and, in blue, the Rpt ring (note that Rpn10 straddles the base and lid).

One of two deubiquitinating enzymes in the yeast proteasome, Rpn11, forms part of the lid assembly and hovers above the channel1,6 (Fig. 2a). By removing the poly-ubiquitin chain, which does not readily fit into the narrow translocation channel, Rpn11 promotes substrate degradation. This process is ATP dependent9,10, although Rpn11 lacks an ATP binding site. Lander and colleagues' structure1 suggests that, when ATP hydrolysis by the Rpt ring threads the substrate through the channel, it brings the attached ubiquitin chain close to the active site of Rpn11 (refs 1, 9, 10). This arrangement could explain how Rpn11 couples deubiquitination to substrate degradation. In striking contrast to this, the other deubiquitinating enzyme, Ubp6, is located far from the entry port (Fig. 2a), and its activity is accordingly not linked to substrate degradation.

Rpn11 also restricts the accessibility of the entry port (Fig. 2a), a feature not seen in PAN or in any other ATP-dependent protease. This design idiosyncrasy may explain why the proteasome is poor at attacking some protein aggregates and complexes.

Lander et al. also show that the bulk of the lid unexpectedly straddles the side of the regulatory particle, with individual subunits extending like fingers to grip and presumably stabilize various elements of the base (Fig. 2b). Moreover, two of the lid subunits, Rpn5 and Rpn6, project as far down as the core particle, thereby stabilizing the interface between it and the regulatory particle. The authors of a recent paper7 reporting the crystal structure of Rpn6 drew this same conclusion.

In both PAN and the eukaryotic proteasome, structures called coiled-coil elements project like spokes from the N-ring1,4,5. In the proteasome, numerous subunits form contacts with these elements1, suggesting that the coiled coils link the Rpt ring with the rest of the regulatory particle. This is markedly different from the case for PAN, which, apart from its ATPase ring, has no additional subunits for these coiled coils to contact. It is possible that these elements have a completely different function in PAN, perhaps in substrate recognition.

The largest proteasome subunits, Rpn1 and Rpn2, are thought to be scaffolds for ubiquitin receptors, ubiquitin ligases and deubiquitinating enzymes3,11. The new structures1,6 show Rpn1 and Rpn2 situated to the side of and above the Rpt ring, respectively (Fig. 2b). Because the exact position of Rpn1 is variable1,6, the factors that reversibly attach to it may not have fixed positions in the proteasome.

Intriguingly, the proteasome's two ubiquitin receptors, Rpn10 and Rpn13, sit on opposite sides of the port1,8 (Fig. 2a). This configuration could mean that substrates can reach the port by more than one pathway, which may be an advantage given the tremendous variety of protein structures on which proteasomes must act. It might also help to explain why proteasomes prefer to bind ubiquitin chains of four or more units — longer chains would be able to engage both receptors. Lander and colleagues' structural model should allow this possibility to be tested.

The structure1 also provides information on the coordination between the six ATPases of the Rpt ring, which work together to drive substrates into the core particle. Some studies have suggested that ATPases act randomly, whereas others indicated a more defined pattern of activity12. Surprisingly, Lander et al. show that the pore loops of the Rpt proteins form a staircase structure, which is suggestive of a rotary mechanism for ATP hydrolysis and substrate engagement. However, whether this structure represents the ATPases in an idle or active state is unclear.

Given the dynamic nature of the Rpt ring, resolving the mechanism of coordination among the ATPases may require 'trapping' the proteasome in multiple ATP-bound states for analysis by electron microscopy. Because ATP is small and hard to resolve, the visualization of ubiquitin chains bound to the proteasome might be a more promising option. This and other objectives for future work will build on the structural model presented by Lander and colleagues, which provides a platform for answering questions about the proteasome that were previously beyond reach.


  1. 1

    Lander, G. C. et al. Nature 482, 186–191 (2012).

  2. 2

    Sauer, R. T. & Baker, T. A. Annu. Rev. Biochem. 80, 587–612 (2011).

  3. 3

    Finley, D. Annu. Rev. Biochem. 78, 477–513 (2009).

  4. 4

    Zhang, F. et al. Mol. Cell 34, 473–484 (2009).

  5. 5

    Djuranovic, S. et al. Mol. Cell 34, 580–590 (2009).

  6. 6

    Lasker, K. et al. Proc. Natl Acad. Sci. USA http://dx.doi.org/10.1073/pnas.1120559109 (2012).

  7. 7

    Pathare, G. R. et al. Proc. Natl Acad. Sci. USA 109, 149–154 (2012).

  8. 8

    Sakata, E. et al. Proc. Natl Acad. Sci. USA http://dx.doi.org/10.1073/pnas.1119394109 (2012).

  9. 9

    Verma, R. et al. Science 298, 611–615 (2002).

  10. 10

    Yao, T. & Cohen, R. E. Nature 419, 403–407 (2002).

  11. 11

    Elsasser, S. et al. Nature Cell Biol. 4, 725–730 (2002).

  12. 12

    Smith, D. M., Fraga, H., Reis, C., Kafri, G. & Goldberg, A. L. Cell 144, 526–538 (2011).

  13. 13

    Walter, J. et al. Acta Crystallogr. B 38, 1462–1472 (1982).

  14. 14


Download references

Author information

Correspondence to Daniel Finley.

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.