NMR Structure and Functional Characteristics of the Hydrophilic N Terminus of the Potassium Channel β-Subunit Kvβ1.1*

Rapid N-type inactivation of voltage-dependent potassium (Kv) channels controls membrane excitability and signal propagation in central neurons and is mediated by protein domains (inactivation gates) occluding the open channel pore from the cytoplasmic side. Inactivation domains (ID) are donated either by the pore-forming α-subunit or certain auxiliary β-subunits. Upon coexpression, Kvβ1.1 was found to endow non-inactivating members of the Kv1α family with fast inactivation via its unique N terminus. Here we investigated structure and functional properties of the Kvβ1.1 N terminus (amino acids 1–62, βN-(1–62)) using NMR spectroscopy and patch clamp recordings. βN-(1–62) showed all hallmarks of N-type inactivation: it inactivated non-inactivating Kv1.1 channels when applied to the cytoplasmic side as a synthetic peptide, and its interaction with the α-subunit was competed with tetraethylammonium and displayed an affinity in the lower micromolar range. In aequous and physiological salt solution, βN-(1–62) showed no well defined three-dimensional structure, it rather existed in a fast equilibrium of multiple weakly structured states. These structural and functional properties of βN-(1–62) closely resemble those of the “unstructured” ID from Shaker B, but differ markedly from those of the compactly folded ID of the Kv3.4 α-subunit.


Rapid N-type inactivation of voltage-dependent potassium (Kv) channels controls membrane excitability and signal propagation in central neurons and is mediated by protein domains (inactivation gates)
occluding the open channel pore from the cytoplasmic side. Inactivation domains (ID) are donated either by the pore-forming ␣-subunit or certain auxiliary ␤-subunits. Upon coexpression, Kv␤1.1 was found to endow non-inactivating members of the Kv1␣ family with fast inactivation via its unique N terminus. Here we investigated structure and functional properties of the Kv␤1.1 N terminus (amino acids 1-62, ␤N-(1-62)) using NMR spectroscopy and patch clamp recordings. ␤N-(1-62) showed all hallmarks of N-type inactivation: it inactivated non-inactivating Kv1.1 channels when applied to the cytoplasmic side as a synthetic peptide, and its interaction with the ␣-subunit was competed with tetraethylammonium and displayed an affinity in the lower micromolar range. In aequous and physiological salt solution, ␤N-(1-62) showed no well defined three-dimensional structure, it rather existed in a fast equilibrium of multiple weakly structured states. These structural and functional properties of ␤N-(1-62) closely resemble those of the "unstructured" ID from Shaker B, but differ markedly from those of the compactly folded ID of the Kv3.4 ␣-subunit.
Fast N-type inactivation of voltage-gated potassium (Kv) 1 channels shapes the action potential, governs the firing rate (spiking), and controls signal propagation in central neurons (1). Biophysically, N-type inactivation has long served as the model for gating transitions in ion channels and is realized by a "ball plug-in" mechanism. In this mechanism a protein domain termed "inactivation gate" or "inactivation ball" binds to its receptor at the inner vestibule of the open channel and thereby occludes the ion pathway (2)(3)(4)(5). Such inactivation gates have been localized in the N terminus of various Kv␣ subunits and were shown to be functional entities, i.e. they conferred rapid inactivation to "ball-less" Kv␣ subunits when applied to the cytoplasmic side of the channels as synthetic peptides (5)(6)(7)(8). Identical to protein-harbored inactivation domains, the synthetic gates interacted with channels in the open state, blocked the pore with low voltage dependence, and were competed with the channel blocker tetraethylammonium (TEA) (9 -11). Recently, the structures of the inactivation domains (ID) from Kv1.4 and Kv3.4 were determined with NMR spectroscopy. Both IDs were found to exhibit well defined and compact folding in aequous solution (12). In contrast, the ID from Shaker B showed no unique, folded structure (13,14).
Besides with Kv␣ subunits owning an N-terminal ID, fast inactivation was observed for a subset of non-inactivating Kv␣1 channels when coexpressed with certain ␤-subunits (15)(16)(17). These auxiliary subunits constitute a family of cytoplasmic proteins (subdivided into subfamilies Kv␤1, -2, and -3) that are made up of two distinct regions: a highly conserved core region that shows homology to the superfamily of aldo-keto reductases and a variable N terminus (18,19). Fig. 1A, depicts the mechanism for inactivation of a non-inactivating Kv␣1 subunit by the coexpressed ␤-subunit, Kv␤1.1. Kv␤ constitutively (and specifically) binds to the Kv␣1-subunit via the conserved core region (20,21) and mediates fast inactivation via its unique N terminus (15,16,18). The latter should contain a "ball-like" domain that occludes the channel pore from the cytoplasmic side similar as known for IDs derived from ␣-subunits. This view is mainly supported by two findings. First, ␤-induced inactivation was absent when the N terminus of Kv␤1.1 was deleted, and second, inactivation was sensitive to oxidation, similar to that reported for inactivation in fast inactivating Kv channels (16). Although this suggested an N-type mechanism of inactivation, the "␤-ball" was never tested for competition with TEA. This competition is considered indicative for N-type inactivation (9,10). Moreover, there is puzzling controversy on the functionality of a peptide that comprises the N-terminal 24 amino acids of Kv␤1.1 and is supposed to constitute the ␤-ball. This peptide induced inactivation in some experiments (16) but failed in others (22), although the coexpressed Kv␤1.1 subunit inactivated the Kv␣1 subunit in either case.
To get insight into the molecular mechanism of fast ␤-mediated inactivation, we investigated structure and function of the hydrophilic Kv␤1.1 N terminus using NMR spectroscopy and giant patch clamp recording on a synthetic peptide (␤N-(1-62)) that covers the entire hydrophilic domain N-terminal to the well conserved core region of Kv␤1.1. * This work was supported by Grant Fa332/2-1 from the Deutsche Forschungsgemeinschaft (to B. F. and C. A.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

MATERIALS AND METHODS
Electrophysiology-Kv1.1 channels were heterologously expressed in Xenopus oocytes as described elsewhere (23). Giant patch recordings were made at room temperature (approximately 23°C) 3-7 days after injection of capped Kv1.1-specific cRNA. Pipettes used were made from thick walled borosilicate glass, had resistances of 0.3-0.6 megaohm (tip diameter of about 20 m), and were filled with 5 mM KCl, 115 mM NaCl, 10 mM HEPES, and 1.8 mM CaCl 2 (pH 7.2). Currents were sampled at 10 kHz and corrected for capacitative transients with an EPC9 amplifier (HEKA Electronics, Lamprecht, Germany) with an analog filter set to 3 kHz (Ϫ3 decibel). The fast application system used is described elsewhere (8,24) and allowed for a complete solution exchange in less than 2 ms (24). ␤N-(1-62) or TEA were dissolved in K int solution and applied via one barrel of the application system; K int was composed as follows: 120 mM KCl, 10 mM HEPES, 10 mM EGTA (pH 7.2).
Rates of inactivation were determined as described previously (8,24). All values are given as mean Ϯ S.D. of 3-5 experiments.
Peptide Synthesis and Sample Preparation-The ␤N-(1-62) peptide was made by conventional solid-phase synthesis and high pressure liquid chromatography-purified (25) Structural Analysis-NMR spectra were recorded at 283 K with a Bruker DRX 600 and Bruker DMX 800 spectrometer. Spin system identification and sequential assignments were accomplished by twodimensional homonuclear NOESY (26), total correlation spectroscopy (27), and DQF-COSY (28) experiments. NOESY spectra were acquired with mixing times of 250 ms; total correlation spectroscopy experiments were acquired with isotropic mixing times of 80 ms. 3 J H N H ␣ coupling constants were measured by high-resolution DQF-COSY spectra. The digital resolution of the time domain data used in the t 2 direction was 0.77 Hz/point. The time proportional phase increments method was used for signal detection in the 1 -direction; for water suppression a binomial pulse sequence with gradients as well as presaturation technique was employed. 1 H chemical shifts were referred to the internal standard of 2,2-dimethyl-2-silapentane-5-sulfonate.
NMR data were processed with XWINNMR (Bruker). Spectral assignment was performed by AURELIA (29) according to standard procedures using total correlation spectroscopy and DQF-COSY spectra for identification of spin systems and NOESY spectra for sequence-specific assignment. Amide-H ␣ coupling constants 3 J H NH␣ were obtained by fitting the corresponding DQF-COSY peaks to two antiphase Lorentzian functions. Proton-proton distances were calculated from the volumes of the NOESY spectra using ␦-proline protons with a known separation of 0.18 nm as reference. NOESY cross peaks were integrated by the automated segmentation procedure, which is part of the AURE-LIA software package.
Prior to first order H/D exchange rate determination the peptide was lyophilized of a 90% H 2 O, 10% D 2 O solution. To evaluate the exchange rates, the peptide was dissolved in 99.9% D 2 O, and a one-dimensional NMR spectrum was recorded within less than 3 min.  (Fig. 1B), similar to IDs derived from Kv␣ subunits (6,10). The latter are known to interact with a receptor that becomes accessible upon opening of the channel and that is competed by the "open channel blocker" TEA (9, 10). Therefore, ␤N-(1-62)-induced inactivation was investigated in the presence of 1 mM TEA, a concentration which blocked about twothirds of open Kv1.1 channels. As shown in Fig. 1, C-E The interaction between ␤N-(1-62) and the channel ␣-subunit was more closely investigated by the "fast application" technique. This technique (see "Materials and Methods") allows solution exchange at giant inside-out patches in less than 2 ms and enables separate determination for on and off rates of channel-peptide interaction (8,24). Fig. 2, A and B receptor, exhibited a time constant of Ϸ100 ms (99.9 Ϯ 16.8 ms, n ϭ 3) and could be well fitted with a monoexponential (Fig. 2B,  right). This indicated that interaction between ␤N-(1-62) and its receptor on the ␣-subunit is a first order reaction, with on and off rates (k on , k off ) calculated as 1.7 ϫ 10 6 mol Ϫ1 s Ϫ1 and 10.3 s Ϫ1 (Fig. 2C). These values were close to the rates determined for the "␣-ball" from Shaker B channels (k on , 4.4 10 6 mol Ϫ1 s Ϫ1 , k off 13.8 s Ϫ1 , (5)), but differed from those found for the ID from Kv3.4 channels (Kv3.4-ID: k on , 10.2 ϫ 10 6 mol Ϫ1 s Ϫ1 ; k off , 2.6 s Ϫ1 , Fig. 2C). The binding affinity of ␤N-(1-62) was about 25-fold lower than that of Kv3.4-ID, but almost equals that of the Shaker B-inactivation domain (5,8).

␤N
␤N-(1-62) Does Not Exhibit a Defined Structure in Solution-The reason(s) for the largely different kinetics observed with the IDs from Shaker B and Kv3.4 channels are not well understood but might be related to their different tertiary structure. Kv3.4-ID shows a well defined and compactly folded three-dimensional structure, whereas the Shaker B ID does not have a unique structure in solution (13,14,30).
To get more insight into the structure-function basis of Ntype inactivation in general and ␤1.1-mediated inactivation in particular, we investigated the structural properties of ␤N-(1-62) with NMR spectroscopy. Experiments were performed under various conditions, in aequous solution at low pH and in a physiological salt solution at neutral pH (see "Materials and Methods"). The spin systems were completely assigned by twodimensional NMR methods in the low pH solution and verified under physiological salt and pH conditions. As illustrated in Fig. 3, only sequential NOE contacts were found, with the pattern of NOEs dominated by d␣N, the NOE between the ␣-proton of an amino acid and the amide proton of the following ones. Stretches of consecutive amide-amide NOEs (dNN), indicative for formation of local structures, were absent as were intermediate or long range NOE contacts. All of the experimentally determined coupling constants between amide and ␣-protons showed values very close to those reported for random coil peptides (31) (Fig. 3).
This view was further supported by the deviations of amide and ␣-proton chemical shifts (⌬␦H N and ⌬␦H ␣ ) from the random-coil values (32). As depicted in Fig. 4, some of the shifts slightly deviated from random-coil values. However, analysis of the data with the chemical shift index by Wishard et al. (33) did not lead to identification of a pattern typical for canonical secondary structures (Fig. 4, A and C). Nevertheless, upfield shifts (characteristic for helical structures) and downfield chemical shifts (characteristic for ␤-pleated structures) can be observed and are distributed nonuniformly over the sequence. This suggests transient formation of these structures in the thermal equilibrium. In addition, there are three stretches of downfield shifted amide resonances (between amino acids 2-10, 44 -52, and 56 -61) indicating a helical propensity in these regions. These results were essentially independent of whether the chemical shifts were determined under low pH or physiological salt and pH conditions (Fig. 4, B and D). Furthermore, experiments performed to determine the rates of H/D exchange of amide protons showed that all these protons were exchanged in Ͻ3 min., indicating that none of these protons are protected by formation of hydrogen bonds within an "internal structure." Rather all amide protons are easily accessible from the solvent, as is typical for not compactly folded peptides. In summary, these results indicated that ␤N-(1-62) does not exhibit a well defined structure in solution, but rather behaves like a flexible peptide showing only transient formation of local structures.

DISCUSSION
The results presented here show that Kv␤1.1 mediates fast inactivation of Kv1␣ subunits via an N-type mechanism of inactivation. Accordingly, the unique N terminus of Kv␤1.1 comprises a ball-like domain that blocks the channel via interaction with a receptor site that becomes accessible upon channel opening. This interaction is competed with TEA, very similar to what is known from ␣-derived IDs. As determined from NMR experiments, the hydrophilic N terminus of Kv␤1.1 does not exhibit a well defined, unique three-dimensional structure. Rather it can be described by a fast conformational equilibrium of weakly structured substates.
With the structural and functional properties described above, the ID of Kv␤1.1 closely resembles the ID from Shaker B, but clearly differs from that of Kv3.4 (5, 8, 12-14, 30, 34). Thus, on rates of inactivation were significantly slower for the unstructured IDs compared with the structured Kv3.4-ID, whereas the respective off rates were considerably faster for the IDs from Shaker B and Kv␤1.1 than for Kv3.4-ID. Inter-estingly, on and off rates of inactivation mediated by the well ordered Kv3.4-ID changed considerably when its structure was disturbed by protein phosphorylation (8). Partial unfolding of the N-terminal hemisphere of the Kv3.4-ID resulted in a significant slowing of the on rate, whereas destabilization of the C-terminal hemisphere lead to an increase in the off rate of inactivation (8).
These correlations suggest that the structural properties of the IDs in solution affect the characteristics of their receptor binding. Thereby, a compactly folded ID exhibits faster access to its receptor than an unfolded domain, even if the latter is more positively charged (net charges are ϩ2, ϩ6, and ϩ4 for IDs from Shaker, Kv␤1.1, and Kv3.4, respectively). Unbinding is much faster for the nonstructured domains compared with the folded. This may either be because of the higher number of molecular contacts (hydrogen bonds, etc.) formed between the folded domain and the receptor or because of the higher flexibility of the unfolded IDs that destabilizes the ID-receptor interaction. Taken together, the overall structural stability may represent a major determinant of the ID-channel interaction in addition to overall charge and the presence of hydrophobic domains, which were reported previously to govern on and off rates of this interaction (5,34).
The results about inactivation of Kv1.1 channels with the "core-free" N terminus of Kv␤1.1 differs from the results obtained from coexpression of both subunits. ␤N-(1-62) induced almost complete inactivation, whereas inactivation mediated by the entire Kv␤1.1 protein was found to be considerably less effective (15,16,35,36). This discrepancy might reflect some variation in binding of Kv␤1.1 to the N terminus of Kv1.1 or might be because of interaction of the ␣-␤ complex with other proteins affecting the mobility of the ␤-inactivation domain. Thus, it was recently suggested that the Kv1.1-Kv␤1.1 complex may interact with the G␤␥ dimer, thereby increasing the degree of ␤-mediated inactivation (35).