Note to users. If you're seeing this message, it means that your browser cannot find this page's style/presentation instructions -- or possibly that you are using a browser that does not support current Web standards. Find out more about why this message is appearing, and what you can do to make your experience of our site the best it can be.
GE and Science Prize

Modeling at the Gates of the Cell

Sarel Fleishman

Membrane proteins are gatekeepers to what goes into or out of a cell, playing pivotal roles in cellular processes such as metabolite transport, intercellular signaling, and the extrusion of toxic compounds. They are therefore the focus of great biomedical interest, comprising more than half of current drug targets. Technical obstacles to protein expression and purification of membrane proteins, particularly from eukaryotes, have resulted in a very limited number of proteins for which we have atomic structures, and many of these exhibit surprising structural features (1). When we combine a small number of experimentally determined structures with a preponderance of novel structural features, it becomes clear that we have an unsatisfactory picture of the possible structural motifs found in membrane proteins. This limited viewpoint raises the question: Can we apply computational modeling to provide additional insight into the relationship between structure and function in membrane proteins?

My scientific drive is to understand the mechanistic relationship between molecular structure and function in human health and disease. I was drawn into structural biology as a graduate student, when I learned of the relevance of this research area to understanding key processes in the biology of the cell and after realizing that a huge effort from molecular biologists had spawned a wealth of biochemical data on specific proteins that can be interpreted in structural terms. We hypothesized that these experimental data along with evolutionary analysis may help us bridge the gap in our structural understanding of membrane proteins and provide structural models that can then serve to suggest mechanisms of action and in the design of experiments.

At the start of my studies on membrane proteins, I developed a methodology for predicting the structures of small systems consisting of pairs of tightly packed helices (2). I then used this methodology to study the association of the membrane-spanning segments of an erbB2 homodimer (3). ErbB2 is an oncogene that has been implicated in one-third of breast cancers. Rather than identifying one mode of interaction for the pair of helices, this methodology identified two, and a path by which the membrane spans could switch from one mode to the other. This unanticipated result was particularly exciting since it provided evidence for a hypothesis of a mechanism of receptor activation, whereby the membrane-spanning helices rotate between active and inactive states. The model also offered insight into clinical and laboratory data, explaining why a certain mutation promotes cancer, whereas another is protective. It was extremely satisfying that recent experimental results gathered from nuclear magnetic resonance (NMR) experiments confirmed part of our prediction by providing evidence for binding through the interaction mode that we suggested was active (4).

Encouraged by the usefulness of this approach, I next targeted larger membrane proteins that were already visualized using cryo-electron microscopy (cryo-EM), a technique that provides the locations and tilt angles of α helices but not those of their constituent amino acid residues. To supplement the information from the cryo-EM map, I developed a method to orient membrane helices around their principal axes based on evolutionary information. Previous studies noted that the cores of membrane proteins are much more evolutionarily conserved than their peripheries [reviewed in (5)]. The reason for this is simple: Mutation at the core of the protein is much more likely to disrupt the protein structure than one in a lipid-facing position, and would be eliminated by the forces of natural selection. Evolutionary conservation could therefore distinguish the parts of each helix that face the lipid from those that face the core of the protein.

EmrE is a bacterial multidrug efflux transporter, involved in antibiotic resistance, for which a cryo-EM map was published (6). At the start of my study of EmrE, two atomic-resolution x-ray crystal structures were available (7, 8). However, they were inconsistent with the majority of the existing biochemical evidence as well as with the cryo-EM map. Using this map and evolutionary conservation data on EmrE, I computed a model structure (9), which was in very good agreement with the existing biochemical information. The model helped us to frame a plausible mechanism for substrate translocation by EmrE. At the time that we published this structural model, it was nevertheless troubling that it was so different from the previous experimentally determined x-ray crystal structures. It was not until later that the x-ray structures were retracted along with several other structures of membrane proteins due to a software error that rendered them wrong (10). When this error was corrected, a third x-ray crystal structure of EmrE was published (11), which deviated by only 1.4 Å from the computed model -- a level of accuracy that is unprecedented for a model of a membrane protein.

The gap junction intercellular channel is another membrane protein for which a cryo-EM map is available (12). This channel links the cytoplasms of neighboring cells in mammalian tissues and allows the cells to transfer metabolites and signals. It is a critical component in cellular signaling of many tissues, and numerous mutations in its membrane domain have been implicated in hereditary hearing loss, neurodegenerative disease, and other genetic diseases. Using the cryo-EM map and evolutionary information, I generated a first model structure of the membrane domain of the gap junction (13). We then used this model to suggest experiments and to interpret their results (14). It was an especially exciting moment for me when I attempted to interpret clinically identified mutations using this model; in fact, virtually all of the amino acid positions that were shown to be sensitive to mutation were packed at the interfaces between helices in the model, where mutations would disrupt the protein’s structure (see the figure). This was the first time that a unified molecular mechanism had been suggested for so many point substitutions in the gap junction.


The model structure of the gap junction membrane domain helps to explain the differential effect of disease-causing and benign polymorphisms. Radical substitutions (e.g., from a small to a large amino acid residue) that do not cause disease are colored green, whereas conservative substitutions (between similar amino acid residues) that cause disease are colored red. All of the disease-causing mutations pack at the interfaces between helices, whereas two benign polymorphisms either face the large channel pore or the lipid environment. The disease-causing mutations include nonsyndromic hereditary deafness, erythrokeratodermia variabilis (EKV), and Charcot-Marie-Tooth.

Compared to the clear portraits of membrane proteins coming out of atomic-resolution crystal structures, the models that I described here are mere sketches. Yet from these sketches a more complete understanding has emerged of how our cells’ gatekeepers work and why they sometimes fail.

References

  1. S. J. Fleishman, V. M. Unger, N. Ben-Tal, Trends Biochem. Sci. 31, 106 (2006).
  2. S. J. Fleishman, N. Ben-Tal, J. Mol. Biol. 321, 363 (2002).
  3. S. J. Fleishman, J. Schlessinger, N. Ben-Tal, Proc. Natl. Acad. Sci. U.S.A. 99, 15937 (2002).
  4. E. V. Bocharov et al., J. Biol. Chem. 283, 6950 (2008).
  5. S. J. Fleishman, N. Ben-Tal, Curr. Opin. Struct. Biol. 16, 496 (2006).
  6. I. Ubarretxena-Belandia, J. M. Baldwin, S. Schuldiner, C. G. Tate, EMBO J. 22, 6175 (2003).
  7. C. Ma, G. Chang, Proc. Natl. Acad. Sci. U.S.A. 101, 2852 (2004).
  8. O. Pornillos, Y. J. Chen, A. P. Chen, G. Chang, Science 310, 1950 (2005).
  9. S. J. Fleishman et al., J. Mol. Biol. 364, 54 (2006).
  10. G. Chang et al., Science 314, 1875 (2006).
  11. Y. J. Chen et al., Proc. Natl. Acad. Sci. U.S.A. 104, 18999 (2007).
  12. V. M. Unger, N. M. Kumar, N. B. Gilula, M. Yeager, Science 283,1176 (1999).
  13. S. J. Fleishman, V. M. Unger, M. Yeager, N. Ben-Tal, Mol. Cell 15, 879 (2004).
  14. S. J. Fleishman, A. D. Sabag, E. Ophir, K. B. Avraham, N. Ben-Tal, J. Biol. Chem. 281, 28958 (2006).

To Advertise     Find Products

Science. ISSN 0036-8075 (print), 1095-9203 (online)