News & Views | Published:

Cell biology

Shape-shifting protein channel

Newly made proteins are moved across cellular membranes through a protein channel. The crystal structure of this channel is now revealed and confirms expectations that it must change shape to allow proteins to pass.

All cellular proteins are synthesized in the body of the cell, the cytosol. But many of them must then be transported through phospholipid membranes to reach their final destinations, which might be intracellular compartments or even, following secretion, outside the cell1,2,3,4,5. The molecular mechanism of this 'translocation' process has been the subject of elegant biochemical1,4,5,6,7,8,9, genetic3 and biophysical10,11,12 studies. This body of work has shown that the main secretion pathway in all kingdoms of life involves a heterotrimeric protein complex, or translocase4,5, which forms a tightly controlled conduit that allows newly made proteins to pass through membranes before the proteins fold into their functional shape.

On page 36 of this issue, van den Berg and colleagues13 present the X-ray crystal structure of the translocase (the SecYEβ complex) from the single-celled archaeon Methanococcus jannaschii, providing an initial view of the atomic architecture of this universally conserved channel. This structure could be considered the latest triumph of genomics, because the choice of the best translocase for structure determination was made on the basis of empirical examination of the expression, purification and crystallization properties of a range of translocases from organisms with fully sequenced genomes.

Almost 30 years ago, as a corollary to his 'signal hypothesis', Günter Blobel1,2 proposed that the translocation of proteins across membranes would occur through a protein-aceous channel of the kind now crystallized. Blobel's research at that time had shown that newly made proteins ('preproteins') that are targeted for export from the cytosol have an extension at one end, called a signal peptide, that is removed during passage through the membrane1. The hypothesis proposed that different types of extension would function as signals, directing newly made proteins to different membrane-bounded compartments1,2, and the importance of this insight was rapidly accepted. The proposal that protein translocation occurs through a protein channel remained controversial for some time, until later experiments3,4,5,6,7,10 confirmed it.

Ion channels have received considerable attention because they show remarkable specificity in allowing one kind of ion through while preventing the passage of very similar molecular species. The protein-translocation channel faces what could be considered a more daunting task, in that it must allow the passage of chemically and sterically varied substrates — representing any segment from a translocating protein — without compromising the permeability barrier of the membrane2,4,5,6,11. Blobel's original answer to this conundrum was that translocation would occur at the same time that the protein was being made1 (co-translationally), with a tight seal between the protein-synthesis machinery (ribosomes) and the pore of the translocation channel maintaining osmotic integrity. But biochemical studies have shown that translocation occurs at least partially post-translationally4,5,6,7, and cryo-electron-microscopic studies show a gap of some 15 Å separating the ribosomal exit site from the translocase in detergent-solubilized samples performing co-translational translocation10 (as in Fig. 1, overleaf). So the pore of the channel seems likely to have variable but controlled conformational properties, expanding just enough to accommodate larger protein segments without compromising the integrity of the osmotic seal.

Figure 1: The SecYEβ channel13, with some of its binding partners and model substrates shown at the same scale.

The channel transports newly made proteins across cell membranes. Two cut-away views through its centre are shown at left and right, with the α-helix that blocks the exit channel moved into the proposed 'open' position13. Hydrophobic amino-acid side chains are in green, acidic groups in red, basic groups in blue, and other atoms in grey; yellow ribbons denote α-helices of SecY inside the surface. Space-filling representations of model substrates use the same colours. The channel's molecular dimensions are shown in the middle. The large ribosomal subunit synthesizes proteins; RNA is in grey and the subunit's protein backbone in yellow, with its exit channel positioned roughly as seen in ref. 10. A surface representation of an Hsp70 protein related to BIP shows the peptide-binding domain in yellow, bound peptide in red and ATP-hydrolysing domain in blue (with relative orientation roughly as in ref. 14); BIP prevents preproteins from slipping back into the channel. A surface representation of SecA from Bacillus subtilis shows the first (blue) and second (cyan) ATP/ADP-binding folds, amino-terminal (yellow) and carboxy-terminal (orange) preprotein-crosslinking domains, scaffold domain (dark green), wing domain (light green), and carboxy-terminal linker (red)15. This protein helps to push preproteins into the channel. The figure was created with DINO16. Protein Data Bank accession numbers are available from us on request.

The crystal structure of the SecYEβ channel presented by van den Berg et al.13 provides an initial atomic-level glimpse of this shape-shifting channel. The authors advance structural arguments to support the hypothesis that the channel's pore is located at the centre of a single SecYEβ heterotrimer. Previous work had found that biochemically well-characterized translocases, such as that from Escherichia coli, purify as higher-order multimers4,5,10, which led to the suggestion that the pore would be located at the interface between several heterotrimers. This possibility was supported by biophysical studies suggesting that the pore is too large to be accommodated in a single heterotrimer11. Although these observations suggest that several heterotrimers might fuse to form a larger channel under certain circumstances, the hypothesis that individual heterotrimers are active in protein translocation is supported by biochemical studies of the E. coli complex7,8.

The SecYEβ structure13 seems to be in a closed state — not surprisingly, given that the complex was crystallized in the absence of a preprotein substrate. With support from biochemical and genetic results, van den Berg et al. propose that a short α-helix (a canonical structural feature in proteins) that plugs the exit from the channel will move out of the way during protein trans-location. But removal of this plug would expose a pore whose molecular surface, at its narrowest point, has a diameter of only 3 Å (Fig. 1). The minimal requirement for channel function would be the ability to pass an arbitrary preprotein segment in an extended conformation, which would have a maximal width of around 12 Å. Moreover, several observations suggest that the channel can accommodate larger transport substrates, including α-helices destined to adopt a transmembrane conformation2,5,9, which have a diameter of some 14 Å. So the channel must dynamically expand and contract in a way that is coordinated with the passage of preprotein segments.

Surprisingly, van den Berg et al. find that the limiting constriction — the 3-Å pore in the channel — is lined by a ring of inflexible isoleucine amino acids. This suggests that the inferred dynamic resizing of the channel must be mediated by diaphragm-like movements of the constituent transmembrane α-helices of the SecYEβ heterotrimer, rather than by changes in side-chain conformation. Movements of the α-helices in SecYEβ that are considerably larger than those that allow preprotein passage are also inferred from evidence that hydrophobic transmembrane α-helices in translocating preproteins can be released directly into the membrane bilayer when their presence is sensed in the translocation channel9. Van den Berg et al. propose a pathway for this sideways release, involving disruption of the interface between two pseudosymmetric halves of the SecY subunit of SecYEβ, leading to opening of the central channel to the membrane.

Future experiments will need to explore how these inferred conformational changes are controlled by, and coupled to, preprotein translocation. In addition to regulation by translocating transmembrane α-helices, van den Berg et al. propose a model in which the channel's dynamics are modulated directly by the signal peptide of the preprotein being transported. Some of the required conformational changes could also be coupled to the protonmotive force (a transmembrane gradient in hydrogen-ion concentration and/or electrical potential), which enhances the efficiency of protein translocation in bacteria6.

Both these possibilities need to be addressed, as does the manner in which the channel is regulated by its known protein partners (Fig. 1). In bacteria, for instance, the SecA protein uses energy from ATP hydrolysis to push preprotein segments through the channel4,6,7. How does the binding of SecA to the channel's cytosolic surface control the expansion and contraction that allows preproteins to pass while maintaining osmotic integrity? Finally, in eukaryotic (nucleated) cells, the Hsp70-family chaperone BIP is believed to act like a ratchet that prevents translocated preprotein segments from slipping back into the channel. Given that BIP and the channel interact12, BIP might also regulate channel dynamics. The crystal structure of the SecYEβ channel13 will allow mechanistic issues such as these to be addressed with greater structural specificity and sophistication.


  1. 1

    Blobel, G. & Dobberstein, B. J. Cell Biol. 67, 835–851 (1975).

  2. 2

    Blobel, G. Proc. Natl Acad. Sci. USA 77, 1496–1500 (1980).

  3. 3

    Schatz, P. J. & Beckwith, J. Annu. Rev. Genet. 24, 215–248 (1990).

  4. 4

    de Keyzer, J., van der Does, C. & Driessen, A. J. Cell. Mol. Life Sci. 60, 2034–2052 (2003).

  5. 5

    Rapoport, T. A., Jungnickel, B. & Kutay, U. Annu. Rev. Biochem. 65, 271–303 (1996).

  6. 6

    Schiebel, E., Driessen, A. J., Hartl, F. U. & Wickner, W. Cell 64, 927–939 (1991).

  7. 7

    Yahr, T. L. & Wickner, W. T. EMBO J. 19, 4393–4401 (2000).

  8. 8

    Duong, F. EMBO J. 22, 4375–4384 (2003).

  9. 9

    Do, H., Falcone, D., Lin, J., Andrews, D. W. & Johnson, A. E. Cell 85, 369–378 (1996).

  10. 10

    Beckmann, R. et al. Cell 107, 361–372 (2001).

  11. 11

    Hamman, B. D., Chen, J. C., Johnson, E. E. & Johnson, A. E. Cell 89, 535–544 (1997).

  12. 12

    Hamman, B. D., Hendershot, L. M. & Johnson, A. E. Cell 92, 747–758 (1998).

  13. 13

    van den Berg, B. et al. Nature 427, 36–44 (2004).

  14. 14

    Zhu, X. et al. Science 272, 1606–1614 (1996).

  15. 15

    Hunt, J. F. et al. Science 297, 2018–2026 (2002).

  16. 16

Download references

Author information

Correspondence to John F. Hunt.

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.