The authors wish to note the following: "The contrast of our final projection map was inverted, so that we interpreted the background density rather than the actual protein density in terms of structural features of the potassium channel-Fv complex. In addition, we indexed the 2D crystals with unit cell parameters of a = b = 175 Å, while the correct indexing would be a = b = 124 Å. Given these analysis errors, the resulting density map and our interpretation of the structural features are not correct. Accordingly, we would like to retract this paper. We acknowledge Yoshinori Fujiyoshi, Rod MacKinnon, Kazutoshi Tani, and Tom Walz for identifying the errors and pointing them out to us."