**3. Discussion**

In this paper, a combined approach using thermal denaturation and homo-FRET assays was used to characterize the effects of TOA+ binding to KcsA. The addition of TOA+ to the channel protein induces a concentration-dependent decrease in protein thermal stability, opposite to the stabilizing effect observed of a shorter-chain QA, TBA+ [35,39]. When comparing the binding of these two QAs at pH 7.0 and pH 4.0, we observed that the closed-channel state at pH 7.0 exhibits a similar KD for both TOA+ and TBA+. On the other hand, when the inner gate is open by pH 4.0, the affinity for TOA+ remains unaffected, while that for TBA+ decreases five orders of magnitude [35]. Since the only difference between these QA blockers is the length of their alkyl chains, it is concluded that the four extra carbon atoms in TOA+ are critical for better hydrophobic interaction with the protein channel wall. There is no crystallographic information on QA–KcsA complexes at pH 4.0, but based on related evidence [42–44], it seems reasonable to expect that the widening of the channel cavity accompanies the acidic-pH-induced untangling of the cytoplasmic α-helical bundle and the bending of the TM2 segmen<sup>t</sup> away from the symmetry axes of the channel. Such a widening of the cavity could diminish the hydrophobic component in the binding of the shorter TBA+, making it prompt to dissociate from the complex. However, the longer acyl chains of TOA+ extend further so as to traverse the channel protein wall completely [33]. This should keep it firmly associated with the hydrophobic residues even in the open conformation at pH 4.0, thus explaining why it maintains a high binding affinity. The X-ray crystallographic data at pH 7.0 also revealed that while TBA+ establishes van der Waals interactions with I100 and F103 residues from the cavity wall (TM2 helix), TOA+ adds interactions with L36 (from the TM1 helix), T74 (near the SF), and G99 and S102 (from the TM2 helix) [33]. These additional interactions between the protein and the QA should be involved in providing a higher affinity for the binding of TOA+ to the open state of the channel compared to TBA+, thus preventing its dissociation from the complex, as suggested by the earlier electrophysiological studies [5,6]. Another difference in the interaction between these two QAs with the channel consists of a change on the side-chain

rotamer of the F103 residue in the TOA+–bound complex. The possible relevance of such observation is discussed below.

A thermal denaturation assay was also used to characterize the interaction of WT KcsA with Na+ and K+ in the presence of TOA+, in terms of both the number of binding events detected and their respective affinities. In the case of Na+, a single binding event with a slightly lower affinity than the control, in the absence of TOA+ (see Table 1), was observed, which corresponds to Na+ binding to its only available site, the extracellular S1 binding site. These observations on Na+ binding were quite similar at pH 7.0 and pH 4.0, or in the presence of TBA+, as indicated by the results. In contrast, the K+ binding studies indicated that the presence of bound TOA+ specifically affects the binding of the permeant species and is sensitive to the inner gate opening by the acidic pH. Thus, it is concluded that the effect of TOA+ on the ion–protein interactions specifically affects the binding to the channel of the permeant K+, and, therefore, monitoring of K+ binding becomes a useful tool to detect both the acidic pH-induced conformational change of the SF to an inactivated state and the changes induced by TOA+. As mentioned in the Introduction section, binding of K+ to WT KcsA is described by two consecutive binding events, with dissociation constants in the μM and mM range, respectively. Here, it is shown that in the presence of TOA+ bound to the channel cavity, the two K+ binding events still remain, suggesting that the SF, rather than collapsing, retains the ability to accommodate K+ at the stack of K+ binding sites and to undergo the K+-concentration-dependent conformational transition. Nonetheless, bound TOA+ induces a decrease in the affinity for K+ in both binding events at pH 7.0 when the inner gate is in the closed conformation. This is similar to that observed in the absence of TOA+ upon acidic-pH-induced channel inactivation. Therefore, this suggests that TOA+ binding by itself causes inactivation at pH 7.0 when the inner gate is closed. Indeed, a similar decrease in the affinity of the channel for K+ at pH 7.0 was also detected in mutant channels where the inactivation process is favored [35,45]. Furthermore, in apparent agreemen<sup>t</sup> with such conclusion, the presence of bound TOA+ in the pH 4.0 samples has only modest effects on the affinity of the two binding events for K+, likely because the channel is already inactivated. We have no evidence to propose a molecular mechanism to explain how TOA+ induces channel inactivation at neutral pH, but it could be speculated that the change on the side-chain rotation of the F103 residue in the TOA+–bound complex indicated above could be involved. The reason to speculate on such a possibility is that F103 is believed to be an essential residue in the allosteric crosstalk between the inner and outer channel gates [46–48] involved in the regulation of the channel's functional cycle.

In order to gain structural information on the KcsA–TOA+–K+ complex, we use the quadruple mutant KcsA W26, 68, 87, 113F, which carries a single tryptophan (W67) as a fluorescent reporter of the SF conformation and dynamics. In this WT–like mutant channel, the homo-FRET process between the W67 residues from each subunit allows us to estimate the changes in steady-state anisotropy and the intersubunit lateral distances according to the type and concentration of cations within the SF [34]. Here, we first characterize in detail the acidic-pH-induced inactivated state in the absence of QAs. Even though the thermal denaturation assay from above detected two consecutive binding events for K+ at pH 4.0, the homo-FRET process is only sensitive to the transition from the nonconductive to the inactivated state. The analysis of this latter event shows a clear decrease in K+ binding affinity at intermediate concentrations of the cation, although the final conformation at saturating amounts of K+ is almost identical to that observed in the closed-conductive state.

As to the effects of TOA+ binding on the W67 mutant channel, we observed that TOA+ bound at the cavity allosterically modifies the conformation of the pore helices, leading to longer W67–W67 intersubunit distances at any K+ concentration at both pH 7.0 and pH 4.0. This loosening in the outer mouth packing seems consistent with the observed decrease in the thermal stability of the protein. The changes in the pore helix conformation, along with the decreased affinity for K+ at pH 7.0 caused by TOA+, seen in both homo-FRET and thermal denaturation experiments, are very similar to those effects caused by inactivation

at pH 4.0. Therefore, as in the WT channel, it is concluded that TOA+ binding at pH 7.0 also causes channel inactivation in the W67 KcsA mutant.

The intersubunit distances determined from the time-resolved anisotropy decays can be compared to those calculated from the published X-ray data, which are usually obtained in the presence of Fab fragments bound to the channel to improve crystal resolution. Even though there are no X-ray data on WT KcsA at pH 4.0, some constitutively open mutant channels were successfully crystallized in different conditions [24,47,49]. Table 2 summarizes the W67–W67 distances calculated from the time-resolved anisotropy decays and compares them to the W67–W67 Cδ2-Cε2 lateral distances derived from X-ray data.

It is observed that the W67–W67 lateral distances calculated from the anisotropy decays of the open/inactivated W67 KcsA channel at pH 4.0 and high K+ concentrations (15.3 ± 0.1 Å) are almost identical to those W67–W67 Cδ2-Cε2 distances determine from Xray data obtained in the presence of an intracellular Fab fragment (15.2 Å; PDB 3PJS), but not when an extracellular Fab was used to form the crystals (17.7 Å; PDB 3F5W). These results highlight how the binding of the extracellular Fab fragment alters the conformation of the extracellular loop and the SF dynamics. In fact, the same Fab fragment has been described to have a profound effect on KcsA inactivation [17]. In this respect, it should be noted that the available X-ray data on the TOA+–KcsA complex was obtained in the presence of the extracellular Fab fragment [33]. Based on such data, the authors concluded that the SF conformation in the TOA+–inactivated channel is a collapsed structure, similar to that detected in KcsA alone at low K+ concentrations [50] (PDB 1K4D), where the inner S2 and S3 K+ binding sites are absent. In contrast to such a conclusion, we find that the stack of K+ binding sites, although with a lower affinity, remains accessible in the TOA+–KcsA complex and that the W67–W67 intersubunit distances are very much like those found in the resting channel in the absence of TOA+. We attribute such discrepancies to the perturbing effects of the extracellular Fab fragment and/or to the C-terminal deletion on the X-ray data and conclude that rather than being collapsed, the inactivated TOA+–bound state of the SF at pH 7.0 and high K+ has a "resting-like" conformation. This is not unique to the inactivated TOA+–induced state as it is shared by several inactivated models of KcsA [35,51,52], and it seems, therefore, a general feature of the inactivated SF of KcsA.

In summary, the results obtained by the combination of the thermal denaturation assay and the analysis of the homo-FRET process among the W67 residues of each subunit in KcsA reinforce the argumen<sup>t</sup> that the long chain QA TOA+ stabilizes an inactivated conformation of the SF, as suggested by earlier electrophysiological studies [5,6], which is characterized by a lower affinity for K+ without affecting the interaction with Na+. However, in contrast to the conclusion from the X-ray studies, this inactivated state is not collapsed but, rather, in a "resting-like" conformation, where the differences between the conductive and inactivated SFs are more subtle. However, if the inactivated SF is "resting-like", what makes it nonconductive? Earlier electrophysiological work concluded that inactivation is associated with a loss of K+ from the selectivity filter in potassium channels [53,54] and that the presence of the cations inside the selectivity filter is fundamental to stabilizing it in the conductive conformation [55–57]. In this respect, the drop in K+ affinity detected in our thermal denaturation and homo-FRET experiments would increase the probability of partial K+ depletion from the filter, thus hampering ion conduction.

#### **4. Materials and Methods**
