Ion channels

Gate expectations

The opening and closing — gating — of ion channels in response to specific stimuli is crucial for cell function. The membrane-partitioning activities of two venom toxins give insights into the mechanisms involved.

The membrane surrounding a biological cell forms a highly selective barrier that allows the cell to control its internal environment. It consists of an assembly of lipids and proteins. Some membrane proteins form channels through which ions can flow. These ion channels are said to be ‘gated’ if they can be opened and closed, and the trigger for this can be electrical, chemical or mechanical stimuli. Ion channels are widely distributed throughout the human body, and in many cells different types of channel coexist. It is their interplay that tunes the cell's electrical activity, enabling it to function properly; even small changes in the balance of activity of the channel proteins can have significant physiological consequences, sometimes leading to severe medical conditions1. The development of therapeutic agents that target ion channels is therefore under way.

Now Lee and MacKinnon2 and Suchyna et al.3 (pages 232 and 235 of this issue) describe how two toxins isolated from the venom of tarantulas interact with two different types of ion channel. Remarkably, they find that the toxins' high-affinity block of these channels is not due to a complementary interaction between channel and toxin surfaces, but to the toxins' ability to partition into the membrane.

The deadly effects of spider venoms are generally thought to result from the high-affinity binding of their toxin components to membrane ion channels. Many toxins are peptides — small assemblies of amino acids — that specifically modulate ion-channel activity. Several peptides that modify the gating of ion channels have been identified. Members of this class interact with the families of voltage-gated channels for sodium, calcium or potassium ions, and with mechanosensitive channels, whose gating can be altered by mechanical forces4. Gating-modifier peptides seem to have a common structural feature in that one face is almost exclusively hydrophobic, which seems to be important for affinity5. The peptides are thought to access their receptor on the channel from the outer, extracellular, side of the membrane and to bind to regions on the protein that are normally involved in the conformational changes, triggered by a given stimulus, that lead to channel opening6.

But gating-modifier peptides have two characteristics requiring special consideration. First, whereas the kinetics of other channel-inhibitory peptides that bind at the outer, surface-exposed region of the channel pore are generally consistent with a diffusion-controlled process, gating-modifier peptides act far more slowly. And second, the interaction surface on the channel for gating-modifier peptides seems to be too small to account for their high affinity.

In an attempt to explain these discrepancies, Lee and MacKinnon2 investigated the interaction of a voltage-sensor toxin, VSTX1, with a voltage-dependent potassium channel known as KvAP. The structure of this channel has been determined from high-resolution X-ray crystallography by MacKinnon's laboratory7,8, making it a good study candidate. Suchyna et al.3, meanwhile, looked at the venom peptide GsMTx4, which inhibits stretch-activated ion channels.

Lee and MacKinnon found that if the KvAP channel is removed from the membrane using detergent, VSTX1 binds to the solubilized protein with a 10,000-fold lower affinity than when the channel is present in the membrane. Suchyna et al. show that GsMTx4 has the same efficacy as its enantiomer — its mirror-image structure — in modifying native stretch-activated channels. They obtained similar results with artificial membranes consisting of lipid bilayers and channels formed by a peptide called gramicidin. These results fly in the face of the traditional ‘lock-and-key’ model of ligand–receptor binding, and instead suggest that the peptide not only acts on the ion channel but also alters the lipid packing at the interface between the channel and the membrane. Thus, much of the binding energy of VSTX1 and GsMTx4 is derived from the nonspecific free energy of membrane partitioning, whereas the actual peptide–channel interaction is rather weak (Fig. 1).

Figure 1: Inhibition of ion channels by two gating-modifier peptides.

Schematic representation of a voltage-gated potassium channel (a) and a stretch-activated channel (b). The peptides VSTX1 (a, red) and GsMTx4 (b, purple) partition within the membrane. Lee and MacKinnon2 show that the gate in a is favoured in the closed conformation when VSTX1 reaches the positively charged voltage sensor through the membrane, whereas Suchyna et al.3 show that opening of the gate in b is blocked through alterations in lipid packing at the interface between the membrane and channel.

The high-resolution X-ray structure of KvAP has led to a controversial proposal for voltage-dependent gating that involves a ‘voltage-sensor paddle’, a positively charged voltage sensor (Fig. 1a). The paddle model is very different from conventional gating models9 in both the shape of the channel in the closed state and the mechanism by which membrane depolarization leads to opening of the ion pathway. Studies of another gating-modifier peptide called hanatoxin have shown that, when this peptide is applied extracellularly, it can bind to closed channels. This has been interpreted as evidence that the voltage sensor lies on the outer part of the membrane in the channel's closed state, thereby limiting its movement within the membrane during activation10. Furthermore, the slow kinetics of hanatoxin-induced block suggested that it remains bound to the channel during gating. However, these observations must be reconsidered in the light of Lee and MacKinnon's results2. If hanatoxin can diffuse into the membrane, previous results would be compatible with a paddle model of gating whereby the voltage sensor moves within the membrane during depolarization. Moreover, membrane partitioning combined with low-affinity binding could allow the peptide to dissociate and rebind rapidly during gating.

How do the findings in these two papers2,3 bear on our understanding of therapeutic agents that target ion channels? Traditional pharmacological methods provided the ion-channel drugs currently in use, but drug development is now based on a more rational approach. An ion channel is usually selected as a potential therapeutic target, and possible channel modulators are then screened for activity. Ideally, however, structural information would be used to increase the potency and specificity of lead candidates. Knowledge of the conformations that ion channels assume during gating, as well as the precise nature of drug–channel inter-actions, will lead to more effective drugs. Because gating-modifier peptides seem to bind to a region of the channel that differs between channel families and even within the same family, it was assumed that drug specificity might be achieved by using a small molecule that mimics the peptide's effect on its target channel. But these two papers suggest that such an approach might not work because nonspecific membrane partitioning contributes greatly to the high affinity of these peptides.

There is one piece of good news, though. The observation that the two enantioners of GsMTx4 are equally effective in blocking stretch-activated channels has major impli-cations for the use of peptides in drug therapy. This is because, generally, only one of the two enantiomers is susceptible to the body's breakdown enzymes, so the alternative form might be stable in the circulation. Finally, although other small organic molecules seem to modify gating by accessing the channel through the lipid membrane, it remains to be determined whether membrane partitioning is a common mechanism for all ion-channel gating modifiers.


  1. 1

    Ashcroft, F. Ion Channels and Disease (Academic, New York, 2000).

    Google Scholar 

  2. 2

    Lee, S. -Y. & MacKinnon, R. Nature 430, 232–235 (2004).

    ADS  CAS  Article  Google Scholar 

  3. 3

    Suchyna, T. M. et al. Nature 430, 235–240 (2004).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Ruta, V., Jiang, Y., Lee, A., Chen, J. & MacKinnon, R. Nature 422, 180–185 (2003).

    ADS  CAS  Article  Google Scholar 

  5. 5

    Wang, J. M. et al. J. Gen. Physiol. 123, 455–467 (2004).

    CAS  Article  Google Scholar 

  6. 6

    Winterfield, J. R. & Swartz, K. J. J. Gen. Physiol. 116, 637–644 (2000).

    CAS  Article  Google Scholar 

  7. 7

    Jiang, Y. et al. Nature 423, 33–41 (2003).

    ADS  CAS  Article  Google Scholar 

  8. 8

    Jiang, Y. et al. Nature 423, 42–48 (2003).

    ADS  CAS  Article  Google Scholar 

  9. 9

    Ahern, C. A. & Horn, R. Trends Neurosci. 27, 303–307 (2004).

    CAS  Article  Google Scholar 

  10. 10

    Lee, H. C., Wang, J. M. & Swartz, K. J. Neuron 40, 527–536 (2003).

    CAS  Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Garcia, M. Gate expectations. Nature 430, 153–155 (2004).

Download citation

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.