Classical experiments on N-type inactivation have suggested that an optimal ball peptide is a hydrophobic structure with closely associated positive charges that moves into a negative electrical potential environment to bind into the pore of the channel in what is essentially a single step block-unblock reaction 
. Our results here further refine this hypothesis by providing evidence that the main N-type inactivation ball structure in Kv1 channels found throughout most of the animal kingdom (Kv1AnID) contains two highly conserved negative charges at positions 2 and 9. The presence of these negative charges clearly does not disrupt N-type inactivation rather they serve to shape the inactivation process in many important ways.
While substitution of uncharged residues into positions 2 and 9 can increase the apparent affinity of the ball peptide for the channel, the manner in which this occurs does not correspond to an increased efficiency for pore block. In all cases tested, uncharged substitutions into positions 2 and 9 resulted in excess binding affinity that did not correspond to increased pore block. Indeed, the native residue at position 2, glutamate, is found to be an amazingly optimal choice for N-type inactivation. Not only does E2 produce pore block more effectively than any other residue tested, it also does so in a manner that highly efficient. The position of E2 on the predicted curves in suggests that for this residue all binding energy is being converted into pore block. While this is level of efficiency is essentially matched by E2D and E2K, both substitutions are relatively ineffective in producing pore block. Residues E2T and E2A show components of pore block that are highly efficient, but they also contain a second inefficient component. For E2A, 65% of the channels do recover in an efficiently blocked manner, possibly explaining why this residue is a common choice in insect Kv1 inactivation ball peptides, such as ShB 
. For position 9, blocking efficiency is nearly optimal with either E9 or E9K suggesting that highly efficient pore block only requires a charge at this position rather than a single specific residue. This observation is supported by the evolutionary data showing that in addition to glutamate at position 9 some channels in this evolutionary family substitute in aspartate and lysine (). Our hypothesis then is that evolution is selecting for residues at positions 2 and 9 of the Kv1AnID that produce optimally effective and efficient pore block.
Our studies also show that residue 2 moves deeply into the channel pore in the inactivation Block state, experiencing 35% of the applied electric field. Although residue 2 is the only charged residue in the initial N-terminus, there appears to be a hidden positive charge that is experiencing a similar fraction of the electric field as residue 2 at both the ON Threshold step and the Bound state. The most likely explanation for this effect is that the N-terminus is unmodified in the Kv1AnID sequence and thus has a free, positively charged N-terminal amino group. Given that a large number of metazoan proteins are N-terminally modified, it seems possible that some of the sequence conservation in the Kv1AnID is to prevent interactions with enzymes that modify N-termini, rather than to block the pore, per se.
Given our hypothesis that residue 2 is negatively charged and near a positively charged N-terminus in the Bound state, it seems likely that significant electrostatic interactions might be occurring between these charges since the environment around this site is expected to be largely non-polar. If so, this could easily explain the 4.2 kT lower pore blocking affinity for E2D in the pore block site (KI
(0 mV): E2
51.3±0.3 (11); E2D
0.79±0.3 (3)) since this shorter side chain would be expected to be at least 1 Å further from the N-terminal charge. Rough structural modeling of the N-terminus suggest that the charge on E2 could easily be within 4.5 Å of the N-terminal charge, and possibly much closer, with the charge on E2D being ~1 Å further. If we use these numbers as a first estimate, the energy difference in water (εr
80) would be only 0.8 kT for the E2D substitution. However in a more hydrophobic environment, such as the channel core (εr
10) the energetic cost for E2D is 6.2 kT. This difference in energy is sufficient to explain the entire effect of substituting E2D into the channel. Importantly, this effect occurs without invoking any specific interactions between E2 and the channel core. For E2Q and E2N we can imagine similar, but weaker, hydrogen bond energy differences between these polar R-groups and the hypothesized N-terminal charge to explain the difference in apparent greater blocking affinity for E2Q at positive potential.
In fact, all the effects ascribed to residue 2 substitutions could be explained by a competition between two locations, one deep within the pore (Site 1) and one near the opening of the pore exposed to water (Site 2) (). Site 1 would be the pore blocking site, 35% of the way into the transmembrane electric field, which has a largely hydrophobic environment. Site 2 could be the ON Transition state site for residue 2, which experiences much less of the transmembrane electric field (15%), has a more negative electrostatic potential than Site 1 (−15.9 mV compared to −1.4 mV) and binds non-charged residues better than charged by about 1.5 kT. Weak blocking of the channel would be caused by a preference for the intermediate Site 2 compared to Site 1, also likely producing the vertical shift in the Linear Brønsted plot for mutations at residue 2 other than E2D (). Voltage-dependence to block would be produced by the greater electric field felt at Site 1 compared to Site 2, shifting the relative equilibrium between these two sites. Non-polar substitution at position 2 would be expected to produce delayed recovery with multiple kinetics due to binding at Site 2. Indeed the very slow recovery for E2N and E2Q could be explained if channel closing partially traps the N-terminus at Site 2, in a manner that does not occur for other position 2 substitutions. Channel reopening would then appear to produce a non-inactivating current since the N-terminus is already bound to a site beyond the normal ON Transition state. For E2K, weak block is likely due to the energetic costs of moving 2 positive charges into Site 1, a non-polar pore location, compared to sitting at the water accessible, electrostatically negative Site 2. E2K likely recovers efficiently, however, since the side window (Site 3) potential is even more negative and a Site 2 binding interaction apparently does not occur with charged residues at position 2 ().
Model of Important Sites and Reaction Steps during N-type Inactivation.
Position 9 residues appear to be located at or near Site 2 in the pore blocked state: a negative electrostatic environment that prefers interactions with non-charged residues by about 1.5 kT. However, mutations that enhance binding of residue 9 to Site 2 also reduce pore block efficiency. If the N-terminal structure between positions 2 and 9 were rigid in the Bound state, then the binding of position 9 with the channel core would be associated with highly efficient pore block, even though its interaction with the channel would technically be outside the actual pore block site. The decreased efficiency seen with the uncharged position 9 substitutions is strong evidence that the N-terminus retains sufficient flexibility between positions 2 and 9, to allow some independence of movement between these two residues and thus at most only partially convert the additional binding affinity of uncharged residue 9 substitutions into more efficient pore block. Sequence conservation shows that small, flexible residues are highly conserved between residues 2 and 9 in agreement with this hypothesis ().
It therefore seems likely that the properties we are ascribing to Site 2 are in fact a general description of much of the environment within the channel between the side window opening and the pore block site, rather than a single competitive site. In the Blocked state, residue 9 would then be located at a more peripheral position in the Site 2 region and residue 2 would be jumping between a more axial location in Site 2 and Site 1 to produce pore block (). An important unanswered question however is whether any of the other uncharged residues in the N-terminus normally bind to Site 2 in a manner that promotes efficient pore block. It is possible that negative charges inserted in other residues could disrupt other interactions with Site 2 that promote efficient pore block, potentially explaining the L7E mutation and the effects of phosphorylation.
The model we have converged upon is very similar to the Pre-Inactivation site model proposed by Zhou et al, 2001 
; however, we propose multiple potential interaction sites leading from the side window openings to the pore block site where the important of any specific site depends upon the types of residues present in the N-terminus 
. The strong negative charge at Site 3, largely produced by EDE161-3, is critical for bringing the chain up to the side windows, by attracting chain positive charges R18, R26, R30, R38 and R42, and passing the N-terminal inactivation domain into the pore. E2 moving past the most negative region in Site 3 to Site 2 is the key ON Transition step (explaining the reversed coupling with charge at 161–3), while residue 9 is moving into Site 3, increasing its interaction with residues 161-3. Based on the similarities of the residue 9 environment in the Transition state to the residue 19 environment in the Bound state, it appears likely that as residue 2 moves into Site 1, residue 9 moves into Site 2 and residue 19 into site 3.
Generally speaking high levels of amino acid conservation are considered indicative of tight structural requirements for optimal interaction. In the case of the Kv1AnID, the conservation of charged residues at positions 2 and 9 is at least partly accounted for by their lack of interaction with the channel core site 2. This lack of interaction produces efficient block by ensuring that the interactions that do occur are causing pore block and not diverting the N-terminus from its optimal block site. We have furthermore hypothesized that E2 is highly conserved because it can move a negative charge closer to the amino terminal positive charge in the Bound State than E2D, lowering the energy for the N-terminus to occupy Site 1, rather than because it has optimal interactions with the non-polar residues lining the pore. Some studies have proposed a beta hairpin bend in the inactivation domain in the pore blocked state, with residue 1 at a relatively shallow location in the pore 
. Our studies do not support such a model for the Kv1AnID. Residue 2 is 35% of the way across the electric field in the Bound state whereas residue 9 is 1%, meaning that a beta hairpin would likely have to be pushed entirely up through the selectivity filter to place residue 2 in such a deep location without also putting residue 9 in the pore. Given that the entire structure of the channel is designed to stabilize the selectivity filter within narrow limits, this structure seems entirely implausible. We therefore suggest that the extended N-terminal structural model proposed by Zhou et al. 
is probably much more representative of the Kv1AnID structure in the Bound state than a beta hairpin.