BBRC Biochemical and Biophysical Research Communications 342 (2006) 1088–1097 www.elsevier.com/locate/ybbrc
Molecular regions responsible for differences in activation between heag channels Min Ju, Dennis Wray
*
Faculty of Biological Sciences, University of Leeds, Leeds LS2 9JT, UK Received 7 February 2006 Available online 21 February 2006
Abstract The ether-a-go-go potassium channels heag1 and heag2 are highly homologous; however, the activation properties between the two channels are different. We have studied the molecular regions that determine differences in activation properties by making chimeras between the two channels, expressing them in oocytes and recording currents with two-electrode voltage-clamp. The activation time course has an initial sigmoidal component dependent on the Cole-Moore shift, followed by a faster component. We show that not only is the extreme N terminus involved in differences between heag1 and heag2 channels, but also the PAS domain itself. Also multiple regions of the membrane-spanning part of the channel appear to be involved, with different regions involved for the early and late time courses, reflecting their different mechanisms. The later time course involved S1 and P-S6 regions. Taken together, our data show that activation involves multiple regions of the N terminal region and membrane-spanning regions of the channel. Ó 2006 Elsevier Inc. All rights reserved. Keywords: Ion channels; Potassium channels; Channel activation; Electrophysiology
Voltage-gated potassium channels play key roles in the control of electrical activity and proper function of excitable cells, such as setting membrane potentials, modulating action potential duration and firing frequency. These channels are also very important as targets of drug action and the sites for inherited disorders [1]. Numerous voltagegated potassium channels have been cloned and they share a common structure comprising four a subunits, each of which consists of six membrane-spanning segments (S1– S6) and the pore P region, with intracellular N and C termini [2,3]. The selectivity filter, through which potassium ions pass, is formed from the P region, and the rest of the channel is lined by S6, with S5 close by. Segments S1–S4 comprise the voltage sensor domain which senses the membrane depolarisation that opens the channel. The S4 segment with positively charged conserved residues is the major part of the voltage sensor, although S2 and S3 also contribute to voltage sensing [4–7]. *
Corresponding author. Fax: +44 011 334 34228. E-mail address:
[email protected] (D. Wray).
0006-291X/$ - see front matter Ó 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2006.02.062
The ether-a-go-go (eag) potassium channels share this membrane-spanning architecture, but also possess long intracellular regions, with a ‘‘Per-Arnt-Sim’’ domain (PAS) within the N terminal region and a cyclic nucleotide binding domain (cNBD) within the C terminal region. The N terminal regions have been shown to be mainly involved in modulating deactivation; this has been mainly studied in herg [8–14], but also in eag1 [15]. Furthermore, the extreme N terminal region (up to residue 12) has been implicated as interacting with the S4/S5 linker. For the herg channel, the cyclic nucleotide binding domain (cNBD) in the C terminal region may have a modulating influence on channel function via its binding of cAMP [16]. There are eight mammalian members of the ether-a-gogo family, and in this paper we focus on two of these channels, heag1 and heag2 [17–19]. These latter two eag channels are highly homologous; both display outward rectifying, non-inactivating currents, and the activation kinetics are dependent on the holding potential (‘‘Cole-Moore shift’’). The Cole-Moore shift is due to the presence of more than one voltage-dependent closed state such that,
M. Ju, D. Wray / Biochemical and Biophysical Research Communications 342 (2006) 1088–1097
for a depolarising pulse from more hyperpolarised potentials, more than one closed state is traversed through before the channel reaches its open state. For heag channels, the effect has been explained by the proposal of two main gating modes: a slower sigmoidal transition at hyperpolarised potentials, and a faster transition at more depolarised potentials [20]. Despite the homology between the two channels, there are striking differences between heag1 and heag2 in their activation properties [18]. Thus the activation kinetics are different for heag2 than heag1, and the extent of the Cole-Moore shift is more marked for heag1. Also, the conductance–voltage curve for heag2 is shifted to the left when compared with heag1 and the voltage sensitivity is less steep [18]. Here, we have systematically investigated the role of molecular regions in determining these differences in activation properties between the two channels. For this, we have constructed chimeras between heag1 and heag2 channels, expressed them in oocytes and recorded currents using two-electrode voltage-clamp.
using restriction enzyme digestion and subsequent ligation with T4 DNA ligase or by the PCR overlap method, using methods as previously described [21]. The positions of the joins for the chimeras are shown in Fig. 1 and listed in Table 1. Chimeras were named according to the sequence replaced (e.g., ‘‘1N’’ is a chimera with the N terminus in heag2 replaced by heag1 and ‘‘2N’’ is the reverse chimera). All chimeras and mutants were confirmed by automatic dideoxy sequencing. After linearisation with NotI, capped cRNA was transcribed in vitro using the T7 promoter (Ambion MEGASCRIPT). Electrophysiology. Xenopus oocytes were injected with 0.5–10 ng RNA for chimeras, mutants or wild type channels in 50nl, and electrophysiological recordings were made at room temperature (22–24 °C) as previously described [21]. Oocytes were perfused with solution containing 2 mM KCl, 115 mM NaCl, 10 mM HEPES, 1.8 mM CaCl2, pH 7.2. Twoelectrode voltage-clamping was carried out using a Geneclamp500 amplifier (Axon Instruments), and depolarizing steps were applied (500 ms duration, 0.1 Hz). All currents were leak-subtracted using twenty 10 mV hyperpolarizing steps. Conductance–voltage (G/V) curves were obtained from I/V curves using a reversal potential of 98.5 mV, and fitted to a corrected Boltzmann function G = A + Gmax/(1 + exp( (V V1/2)/k)), with G conductance, Gmax maximal conductance, V voltage, V1/2 voltage at half maximal activation, k slope parameter, and A a constant. The Boltzmann parameters that we obtained as above for heag2 were very similar to those obtained using the formula previously described [18]. Values for Boltzmann parameters for chimeras were always compared with the appropriate wild type controls. The initial sigmoidal time course of the current (from the beginning to 20% of maximum current) was fitted with: I = C(1 exp ( t/si))4 where si is the time constant, t is time and C a constant. For the later phase of activation, activation times, t20–80%, were taken as the time for currents to rise from 20% to 80% of maximum current. Means ± SEM are shown in the figures, and Student’s t test (as compared with the appropriate wild type controls in the same batch of cells) was used to test for statistical significance.
Methods Preparation of chimeras and mutants. Human eag cDNA clones in pGEM-HE used in this study were heag1 (Accession No. AJ001366, [17]), and heag2 (Accession No. AF472412, polymorph 2233A, [18]). Restriction enzymes were obtained from Helena, New England Biolabs and Invitrogen. Chimeric channels between heag1 and heag2 were generated either by
A
5 6
C heag1
8
heag1 PAS 2
3
1
1089
4
10
-60mV
heag2
cNBD
9
7
heag2
-60mV
-130mV
-130mV
N C 1
B
PAS
PAS
3
4 S3
2
S1
5
6
S4
8
S2 7
*
S5
P
S6
10
9
cNBD
Fig. 1. Wild type channels and chimeric constructs used in this study. (A) The positions of the joins (1–10) are shown schematically. (B) The positions (arrows) of the joins are shown on an alignment of heag1 and heag2 sequences (only residues 1–568 are shown). The position of the point mutation is indicated by an asterisk. (C) Examples of wild type currents are shown for heag1 and heag2. Currents for steps from 80 mV to 0 mV for heag1 and heag2 are shown (left) normalised to the same maximum (horizontal bar: 50 ms). Sample currents are also shown for heag1 (centre) and heag2 (right) for steps to +40 mV from holding potentials of 130 mV to 60 mV (vertical and horizontal calibration bars: 2 lA and 20 ms, respectively).
1090
M. Ju, D. Wray / Biochemical and Biophysical Research Communications 342 (2006) 1088–1097
Table 1 Joins for chimeras with heag2 replaced by heag1 Chimera
Position of join 1
Position of join 2
1C 1NL 1CL 1N 1M 1NT 1PAS 1PP 1S1/3 1S5/6 1S1 1S2/3 1S5P 1PS6
10 3 9 Start 3 Start 1 1 3 7 4 5 7 8
End 4 10 3 10 1 3 2 7 10 5 6 8 9
The numbers refer to the positions shown in Fig. 1. The joins for the reverse chimeras were at the same positions.
Results and discussion To explore activation properties of the chimeras, we used expression in oocytes followed by two-electrode voltage–clamp recordings. Fig. 1C illustrates the differing time course of activation between the two wild type channels [18]. We first investigated the dependence of activation kinetics on holding potential (Cole-Moore shift). Chimeras were constructed by swapping various regions between the heag1 and heag2 wild type channels, as shown in Figs. 1A and B, and Table 1, and the activation time course at different holding potentials was compared with results for wild type channels. For wild type channels, the most obvious feature of the Cole-Moore shift is the dependence of the initial sigmoidal time course (si) on holding potential. This is seen for both heag1 and heag2 channels in the presence of magnesium. In the absence of magnesium, the Cole-Moore shift is only prominent for the heag1 channel but is only prominent for heag1 in the absence of magnesium. We have studied this difference between heag1 and heag2 (marked at 130 mV, Fig. 2A) in the absence of magnesium for our chimeric study; in what follows we have normally compared values of si at a hyperpolarised potential of 130 mV for the chimeras used. At this hyperpolarised potential, the initial time course is much slower for heag1 than for heag2. All the chimeras described in this section gave si versus holding potential curves either like heag1 or heag2 wild type curves (examples for chimeras are shown in Fig. 2A). There was very little difference between heag1 and heag2 initial time constant at less hyperpolarised potentials (Fig. 2A). Swapping the C-terminus (i.e., from the cNBD to the end) (chimeras 1C and 2C) did not change the initial time course at this hyperpolarised potential, as compared with the respective wild types (Fig. 2B), indicating that the entire C terminus is not involved. Also, swapping the linkers at
the N and C termini (chimeras 1NL, 2NL,1CL, and 2CL) did not affect the initial time course (Fig. 2B), indicating that these linkers are also not involved. On the other hand, replacing the N terminus (i.e., from the N terminus to the end of the PAS domain) in heag2 with that of heag1 led to a channel with initial time constant like heag1 (chimera 1N, Fig. 2B), indicating that this N terminal region is involved. Replacing the central region of heag2 with heag1 produced a channel with time course like heag1 (chimera 1M, Fig. 2B), indicating a contribution from the central region also in the time course. Therefore, both the N- terminal region and the membrane-spanning region are involved in determining the differences in initial time course between the two channels at hyperpolarised potentials. This result is confirmed by chimeras 2N and 2M which also behave like heag1 (Fig. 2B), which can readily be understood because the only differences from chimeras 1M and 1N, respectively, are the C termini (which do not contribute). Taken together the data fits a consistent pattern whereby a channel with properties like heag2 results only if the N terminus is heag2 and the membrane-spanning part is also heag2. On the other hand, if either or both, of these regions is heag1, a channel with properties like heag1 results. The data can best be explained as due to interactions between these regions. Thus, for example the reverse chimera of 1N (chimera 2N) still behaves like heag1, rather than heag2 (as might have been expected) because of interactions between these regions. As we have shown that the N terminal region (comprising the extreme N terminus and the PAS domain) is involved in determining differences between heag1 and heag2, we attempted to narrow the region down by making chimeras with either the extreme N terminus (chimera 1NT) or the PAS domain (chimera 1PAS) of heag2 replaced by heag1 (Fig. 2B). However in both cases, the initial time course remained like heag2, suggesting that the N terminus/PAS domain cannot be functionally subdivided, so that the whole N terminal/PAS region is required to transfer properties like heag1. Thus in this respect at least, the whole of the N terminal region and the PAS domain behave as a functional entity. Similar experiments were carried out attempting to subdivide the membrane-spanning region by making chimeras with either S1–S3 or S5–S6 swapped. However, the membrane-spanning region could also not be functionally subdivided as far as its effect on initial time course is concerned, because replacing heag2 by heag1 in either the S1–S3 region or the S5–S6 region still produced channels (chimeras 1S1/3 or 1S5/6) with properties like heag2 (Fig. 2B). Thus also the whole of the membrane-spanning region behaves as a functional entity, since replacement of the whole membrane-spanning region of heag2 by heag1 gave (as shown above) a channel with properties like heag1. There is one amino acid that is different in the S4/S5 linker between heag1 and heag2, but point mutation of this residue in heag1 (I345L) or mutation in heag2 (L342I) did
M. Ju, D. Wray / Biochemical and Biophysical Research Communications 342 (2006) 1088–1097
1091
A
* 10
10
* *
*
-130
-120
0
B heag1
0 -110
-130
-120
-110
N PAS NL S1 S2/3 S4 S5P PS6 CL cNBD-C
heag2 1C
*
2C
+
1NL
*
2NL
+
1CL
* +
2CL
+
1N
+
1M
+
2N
+
2M 1NT
*
1PAS
* *
1S1/3
*
1S5/6
+
I345L
*
L342I 0
50
100
150
τi(%) Fig. 2. Initial activation time course of the chimeras. (A) Examples of results for two chimeras are shown for the time constant, si, of the initial time course of the currents, plotted against the holding potential for wt heag1 (j, n = 13), wt heag2 (d, n = 8), and chimera (m) (left, chimera 1N, n = 10; right, chimera 1C, n = 8). Significant difference from wild type, *P < 0.05. (B) The figure shows schematically (left) the chimeras used (not drawn to scale), together with (right) the values for the initial time course (normalised to the value for heag1) (n = 6–20 for the chimeras, n = 69–77 for the wild types). Arrows show the position of the point mutations. Significant difference from heag1, *P < 0.05. Significant difference from heag2, +P < 0.05.
not change initial activation time (Fig. 2B). For the heag1 mutation this fits in with the previous results where the N terminus alone (like chimera 1N) determines activation time, while for the heag2 mutant, the activation time would be like heag1 only if the S4/5 linker were the sole determinant within the membrane part, which it obviously is not. In summary, the data show that, for the initial time course at hyperpolarised potentials, multiple regions within the N terminal/PAS region and the membrane-spanning region are involved in determining the differences in initial time course between heag1 and heag2 channels. The data
suggest interactions between the N terminal and transmembrane regions as determining these Cole-Moore shifts. The later activation time course (parametrised here as time from 20% to 80% maximum amplitude, t20-80%) also depends on holding potential (Fig. 3A), but for this part of the time course heag2 is slower than heag1 (rather than the reverse as was the case for the initial time course), presumably because different mechanisms underlie the later time course as compared with the initial time course. For this component, large differences between heag1 and heag2 were observed at 80 mV holding potential with test
1092
M. Ju, D. Wray / Biochemical and Biophysical Research Communications 342 (2006) 1088–1097
A
B
C
N PAS NL S1 S2/3 S4 S5P PS6 CL cNBD-C
heag1 heag2 1C
* +
2C 1NL
* +
2NL 1CL
* +
2CL 1N
+ +
1M
+ +
2N 2M
+
1NT
+ +
1PAS 1PP
*+
1S1
*
1S2/3
*
1S5P
*+
1PS6
+ +
2S1 2S2/3
+
I345L
*
L342I 0
50
100
150
t20-80% % Fig. 3. Later activation time course of the chimeras. (A) Examples of results for two chimeras are shown for the time constant, t20–80%, of the later time course of the currents, plotted against the holding potential for wt heag1 (j, n = 7), wt heag2 (d, n = 6) and chimera (m) (left, chimera 1NL, n = 11; right, chimera 2NL, n = 10). (B) Examples are shown for the time constant, t20–80%, of the later time course of the currents, plotted against test potential for wt heag1 (j), wt heag2 (d) and chimera (m) (left, chimera 1NL, n = 10; right, chimera 2NL, n = 10). (C) The figure shows schematically (left) the chimeras used, together with (right) the values for the later time course (normalised to the value for heag1) (n = 6–20 for the chimeras, n = 90–96 for the wild types). Arrows denote position of point mutations. Significant difference from heag1, *P < 0.05. Significant difference from heag2, +P < 0.05.
M. Ju, D. Wray / Biochemical and Biophysical Research Communications 342 (2006) 1088–1097
potential steps to, e.g., 0 mV (Fig. 3B) and so our chimeric analysis of this later time course was made for steps from this potential to 0 mV. As for the initial time course, we found that the N terminal region and the membrane-spanning region are involved in determining the later time course. Thus, swapping the C terminal region had no effect on activation, again indicating lack of involvement of the C terminus (chimeras 1C and 2C, Fig. 3C). Similarly, replacing the linkers at the N and C termini also did not affect later time course (chimeras 1NL, 2NL, 1CL, 2CL), also showing lack of involvement of these regions (Fig. 3C). Replacing the N terminus in heag2 with heag1 gave a channel with later time course like heag1 (chimera 1N), again showing involvement of the N terminal region. Replacing the membrane-spanning region of heag2 with heag1 (chimera 1M) gave a channel with a later time course like heag1 (Fig. 3C), showing that the membrane-spanning region is also involved. Again as for the initial time course, interactions between the N terminal region and the membrane-spanning region seem to occur because both N terminal and membranespanning parts need to be heag2 sequence for the channel to remain slow; if either or both regions have the heag1 sequence, a channel with activation kinetics like heag1 results (chimeras 1N, 2N, 1M, 2M, and 2C, Fig. 3C). To determine more precisely the region involved, we subdivided the N terminal region. Replacing the extreme N terminus (amino acids 1–27) in heag2 by heag1 produced a channel (chimera 1NT) with later time course like heag1 (Fig. 3C), indicating a contribution from this region in determining t20–80%. Also, replacing the PAS domain of heag2 with that from heag1 gave a channel (chimera 1PAS) with later time course like heag1, indicating the involvement of this region too. Most, or all of this effect appeared to be due to the first 28–93 amino acids of the PAS domain, because substituting this domain of heag2 with that from heag1 produced a channel (1PP) with later time course like heag1. To narrow down the important region(s) within the membrane-spanning part, we replaced S1, S2–S3, S5-P, and P-S6 regions separately in heag2 with the corresponding counterparts from heag1. The data (Fig. 3C) are consistent with effects on later time course due to the S1 region and the P-S6 regions. Thus replacement of the S1 region or the P-S6 region in heag2 with heag1 produced channels (chimeras 1S1 or 1PS6) with intermediate properties between the two wild types, whereas replacement of the other regions was without effect (chimeras 1S2/3 and 1S5P) (Fig. 3C). The reverse chimeras were both fast like heag1, presumably because the N terminal part is heag1 (which as seen above determines the overall properties of the channel independently of the composition of the membrane-spanning part). Point mutations in the S4/S5 linker did not alter the later time course; mutation heag1 I345L remained like heag1, while heag2 L342I remained like heag2 (Fig. 3C). These
1093
results are consistent with the above findings from chimeras implicating just S1 and P-S6. Thus the data for the later time course indicates that, as for the initial time course, both the N terminal and membrane-spanning parts are important (and again appear to interact). However, for this component it was possible to narrow down the regions of importance further to the extreme N terminus, the proximal part of the PAS domain, and the S1 and P-S6 regions. The two components of time course appear to represent different molecular mechanisms, since the pattern of involvement of specific regions within the N terminal and membrane-spanning regions are different. The steady-state activation properties also differ between the two channels (Fig. 4A). The conductance/voltage (G/V) curve for wild type heag2 channel is shifted to the left, and is less voltage sensitive, than for wild type heag1 [18]. We have therefore also investigated the molecular regions underlying these differences using the same chimeras. Replacing the N terminus or the C terminus in heag2 by heag1 (chimeras 1N and 1C) produced channels with Boltzmann parameters of the G/V curves like heag2 (Fig. 4B). On the other hand, for the reverse chimeras (2N and 2C), Boltzmann parameters were like heag1. Therefore the data suggest that the N and C terminal regions are not involved in determining G/V characteristics. Furthermore, replacing the linkers in the N and C regions of heag2 with heag1 (chimeras 1NL and 1CL) gave channels with properties like heag2, while replacing these regions in heag1 with heag2 (chimeras 2NL and 2CL) gave channels with Boltzmann parameters like heag1 (Fig. 4B). Thus these linker regions are also not involved in determining G/V characteristics. Since only the membrane-spanning region appears to be involved, we have examined chimeras with replacements in this region. Replacing the membrane-spanning part of heag1 with heag2 (chimera 2M) gave a channel with Boltzmann parameters like heag2 (Fig. 4B) (although V1/2 was even more negative than for heag2), indicating the involvement of the membrane spanning region in steady-state activation properties. The reverse chimera (1M), with the membrane-spanning part of heag2 replaced by heag1, had Boltzmann parameter k like heag1 (Fig. 4B) (although in this case V1/2 was only shifted a little towards the heag1 value), again indicating the membrane-spanning region in steady-state activation. However, attempts to narrow down the region of interest by replacing the S1–S3 region or the S5–S6 region of heag1 with heag2 (chimeras 2S1/3 and 2S5/6) gave channels with Boltzmann parameters closest to heag1, which is more difficult to interpret. Again a likely explanation is the functional involvement of multiple regions in the membrane-spanning part. The functional involvement of multiple regions within the membrane-spanning part was reinforced by consideration of point mutations in the S4/S5 linker. Although the mutation I345L in heag1 gave Boltzmann parameters like heag1, the mutation L342I in heag2 gave k parameter
1094
M. Ju, D. Wray / Biochemical and Biophysical Research Communications 342 (2006) 1088–1097
A 1.0
G/Gmax
G/Gmax
1.0
0.5
0.0 -80
0.5
0.0 -40
0
40
80
-80
-40
0
V (mV)
B heag1
40
80
V (mV)
N PAS NL S1 S2/3 S4 S5P PS6 CL cNBD-C
heag2
*
1N
*
*
1C
* +
2N
+
+
2C
+ *
*
1NL 1CL
*
* +
+
+
+
2NL 2CL
2M
*
*
1M
+
*
2S1/3
+
2S5/6
+
+
I345L
+
+
L342I
+ +
* -80
-40
0 V1/2 (mV)
40 0
40
80
k (mV)
Fig. 4. Steady-state activation properties of the chimeras. (A) Examples of G/V curves are shown for two chimeras (m): (left) chimera 1NL, n = 10; (right) chimera 2NL, n = 10, together with wild type curves (heag1 j, n = 7, heag2 d, n = 6). (B) The figure shows schematically (left) the chimeras used, together with (right) the results obtained for the Boltzmann parameters V1/2 and k (n = 6–20 for the chimeras, n = 66–67 for the wild types). Arrows denote position of point mutations. Significant difference from heag1, *P < 0.05. Significant difference from heag2, +P < 0.05.
like heag1 but V1/2 like heag2 (Fig. 4B), implicating this linker but also suggesting the involvement of other regions. Therefore to summarise the data on steady-state activation, the molecular region that determines differences between heag1 and heag2 channels appears to be located in multiple parts of the membrane-spanning region rather than the N or C terminal parts. This work is the first to report involvement of the PAS domain (amino acids 28–134) in the Cole-Moore shift for the eag channels. Much previous work has been carried out using the herg channel, showing that the PAS domain is involved in deactivation kinetics [9,22,23], and some Long Q-T mutations also affect activation kinetics of herg
[9,23,24], but the herg channel does not exhibit a ColeMoore shift. It seems clear from our data that there are functional interactions of the PAS domain with the membrane-spanning part of the channel, and indeed the PAS domain is tightly bound to the channel protein [8]. However, it is not at all clear to which part of the channel the PAS domain binds. The C terminal cNBD domain hangs centrally below the membrane, most likely in a similar fashion to that for the cNBD of HCN2 [25]; our homology model for heag2 is shown in Fig. 5A. Because of its size, it seems clear from the model that the PAS domain cannot reach the S5/S6 pore region, and even may not reach the S5/S6 linker, although it may bind to the cNBD or the S1-S4
M. Ju, D. Wray / Biochemical and Biophysical Research Communications 342 (2006) 1088–1097
A
S5-S6
S1-S4
PAS
cNBD
B
Fig. 5. Homology models. (A) The figure shows a homology model of the heag2 channel (side view). The model was constructed using the KcsA pore region [37], the KvAP voltage sensor [38], the C terminal cNBD from HCN2 [25] and the N terminal PAS domain from herg [8]. The position of the PAS domain relative to the rest of the channel has not been determined; one possibility is shown here. The extreme N terminus (27 residues) and the remainder of the C terminus (337 amino acids) are not shown. (B) View of both sides of PAS domain homology model for heag2. Residues that are different between heag1 and heag2 in the proximal part of the PAS domain (some of which must affect activation kinetics) are shown shaded. Black residues are conserved residues already identified as affecting activation kinetics [9,19].
region to exert its effects, or it may even exert its effects via the extreme N terminal part (as follows). We also showed a role for the extreme N terminal region (amino acids 1–27) in the Cole-Moore shift. Such an involvement of this region in the Cole-Moore shift has been previously reported [15] for rat eag for amino acids 7–12. However, amino acids 8–12 are identical between the two channels, likely implicating the involvement of other residues within the region 1–27 (see below). For herg, the extreme N terminal region has been widely implicated for its role in deactivation kinetics, though it too may affect activation rate [9]. For the extreme N terminus, evidence has been put forward that it may interacts with the S4/S5 linker to produce its functional effects [9,10,13,15]. Our data show that there are (at least) two distinct components of the activation time course with different molecular regions determining their properties. The two
1095
components map onto the two gating modes of the heag channels discussed by Schonherr et al. [20]. These authors showed that the initial slow time course prevalent at hyperpolarised potentials is due to mode switching whereby the environs of the S4 first slowly change in conformation following depolarisation from hyperpolarised potentials, and this is followed by a faster activation step leading to channel opening. The S4 region is identical between heag1 and heag2 channels, and so our chimeric approach cannot give further information about the S4 region, although our results do show the involvement of different molecular regions of the channel for the two processes (fast and slow components), consistent with the idea of two different mechanisms for the two time courses. It is also worth mentioning that our results (carried out in the absence of magnesium) extend the concept of two gating modes, not only caused by magnesium binding to the S2/S3 region [20], but also in its absence. It is noteworthy that although the ColeMoore shift described by Schonherr et al. [20] occurs in the presence of magnesium, the Cole-Moore shift can still occur in its absence. Besides the N terminal regions, the molecular regions involved in determining the later time course were the S1 and P-S6 regions. For heag channels, such regions have not been previously determined as influencing activation although the involvement of the S1 region may be expected due to its contribution to the S1–S4 voltage sensor, or it may also be due to interactions with the N terminal domain. For the P-S6 region, a direct interaction with the N terminal region seems unlikely because it is shielded by the cNBD hanging directly below it. The role of S5–S6 regions in determining activation has been shown for other channels (e.g. [26,27]), perhaps by mainly interrupting interaction with the voltage-sensor. For the initial sigmoidal time course, it was not possible to dissect out molecular regions of the membrane-spanning part of the channel as being particularly important in determining differences in activation properties between heag1 and heag2 channels. As previously found [28], the membrane-spanning part gives a major contribution to the time course, and the pore domain has also been implicated. In fact, one of the main features of our results is that the channel appears to gate/ activate as a whole, involving perhaps conformational changes throughout the whole channel protein. Similar activation changes involving conformational changes of the whole channel have also been proposed for other families of Kv channels [21,29]. However, for the steady-state activation properties it was the membrane-spanning region alone that is functionally involved; again our data suggested the involvement of the whole (or most) of the membrane-spanning region since attempts to dissect it further with chimeras gave an inconsistent picture (as previously found by Gessner et al. [28]), unless functional interactions between the various regions is postulated. Within the S1 region, there are four residues that are different between heag1 and heag2 (including one residue in the S1/S2 linker, Fig. 1), and some of these residues must
1096
M. Ju, D. Wray / Biochemical and Biophysical Research Communications 342 (2006) 1088–1097
contribute to the observed differences in function between the two channels. Effects of the S1 segment on activation kinetics have also been reported for some other channels, e.g., Kv1.1 [30–32] and HCN [33,34]. For the P-S6 pore region, eight of the residues are different, and most of them are located at the extracellular region. The gate itself (located deeper within the pore [3]) seems not to be involved. Effects of the pore region have also been reported, e.g., for Kv2.1 [26]. For the extreme N terminus, 4 residues are different at the beginning of the N terminus and another four next to the PAS domain. For the PAS domain itself, the proximal part (residues 28–93) is particularly important in the later time course. In this region, there are seven amino acids that are different between heag1 and heag2 channels. Using a homology model for the heag2 PAS domain (Fig. 5B), most of the residues that are different between heag1 and heag2 lie on the surface of the domain; some at least of these residues must be involved in activation kinetics. These residues lie scattered throughout the central part of the surface of the domain, which may be an important interaction surface with the rest of the channel protein. Previous work on eag and erg channels has shown that the S4/S5 linker region is important and may interact with the extreme N terminus to affect deactivation and also activation properties [9,10,13,15]. The mutation in this linker that we studied was not tested in previous studies. For activation kinetics, we found no evidence for involvement of this linker, but our data confirmed the involvement of the S4/S5 linker in steady-state activation properties. Our data extend this picture by showing that multiple molecular regions (both in the membrane-spanning part and the N terminal region) are also involved besides simply between extreme N terminus and the S4/S5 linker. Our results did not suggest a role for the involvement of the C terminal region in determining differences between heag1 and heag2 channel activation properties. This is at first sight a surprising unexpected result, because there are many amino acids that are different in the long C terminal region. Indeed C termini have been implicated in functional effects in other channels [21,35,36]. However, using the homology model shown in Fig. 5A for the cNBD, it appears that almost all of the residues that differ between heag1 and heag2 lie inside the cNBD or underneath it, rather than on the surfaces at the top and sides. Therefore, if the main interacting surfaces of the cNBD with the rest of the heag channel structure is with the top and sides of the cNBD, it is perhaps not so surprising that there is indeed little role for the C terminus in determining functional differences between heag1 and heag2 channels. In summary, our data show that differences in activation kinetics between heag1 and heag2 channels are mediated by both N terminal and membrane-spanning regions. Within each of these stretches of amino acids, multiple regions are involved. Our data show the presence of interactions not simply between the extreme N terminus and the S4/ S5 linker, but also between multiple regions in the N terminal region (including the PAS domain) and between
multiple regions of the membrane-spanning part (including S1 and PS6). Steady-state activation properties again involved multiple regions of the membrane-spanning part, but did not involve the intracellular regions. Acknowledgment We thank the Biotechnology and Biological Sciences Research Council for support. References [1] K. Bracey, D. Wray, Inherited disorders of ion channels, in: VoltageGated Ion Channels as Drug Targets, Wiley-VCH, New York, 2006. [2] D. Wray, How to communicate with cell, Sci. Spectra 23 (2000) 64–71. [3] G. Yellen, The voltage-gated potassium channels and their relatives, Nature 419 (2002) 35–42. [4] S.P. Yusaf, D. Wray, A. Sivaprasadarao, Measurement of the movement of the S4 segment during the activation of a voltage-gated potassium channel, Pflugers Arch. 433 (1996) 91–97. [5] F. Bezanilla, Voltage sensor movements, J. Gen. Physiol. 120 (2002) 465–473. [6] C.S. Gandhi, E.Y. Isacoff, Molecular models of voltage sensing, J. Gen. Physiol. 120 (2002) 455–463. [7] C.J. Milligan, D. Wray, Local movement in the S2 region of the voltage-gated potassium channel hKv2.1 studied using cysteine mutagenesis, Biophys. J. 78 (2000) 1852–1861. [8] J.H. Morais Cabral, A. Lee, S.L. Cohen, B.T. Chait, M. Li, R. Mackinnon, Crystal structure and functional analysis of the HERG potassium channel N terminus: a eukaryotic PAS domain, Cell 95 (1998) 649–655. [9] J. Chen, A. Zou, I. Splawski, M.T. Keating, M.C. Sanguinetti, Long QT syndrome-associated mutations in the Per-Arnt-Sim (PAS) domain of HERG potassium channels accelerate channel deactivation, J. Biol. Chem. 274 (1999) 10113–10118. [10] J. Wang, M.C. Trudeau, A.M. Zappia, G.A. Robertson, Regulation ˜ -goof deactivation by an amino terminal domain in human ether-A go-related gene potassium channels, J. Gen. Physiol. 112 (1998) 637–647. [11] J. Wang, C.D. Myers, G.A. Robertson, Dynamic control of deactivation gating by a soluble amino-terminal domain in HERG K+ channels, J. Gen. Physiol. 115 (2000) 749–758. [12] R. Schonherr, S.H. Heinemann, Molecular determinants for activation and inactivation of HERG, a human inward rectifier potassium channel, J. Physiol. 493 (1996) 635–642. [13] M.C. Sanguinetti, Q.P. Xu, Mutations of the S4-S5 linker alter activation properties of HERG potassium channels expressed in Xenopus oocytes, J. Physiol. 514 (1999) 667–675. [14] P.S. Spector, M.E. Curran, A. Zou, M.T. Keating, M.C. Sanguinetti, Fast inactivation causes rectification of the IKr channel, J. Gen. Physiol. 107 (1996) 611–619. [15] H. Terlau, S.H. Heinemann, W. Stuhmer, O. Pongs, J. Ludwig, Amino terminal-dependent gating of the potassium channel rat eag is compensated by a mutation in the S4 segment, J. Physiol. 502 (1997) 537–543. [16] J. Cui, A. Kagan, D. Qin, J. Mathew, Y.F. Melman, T.V. McDonald, Analysis of the cyclic nucleotide binding domain of the HERG potassium channel and interactions with KCNE2, J. Biol. Chem. 276 (2001) 17244–17251. [17] T. Occhiodoro, L. Bernheim, J.H. Liu, P. Bijlenga, M. Sinnreich, C.R. Bader, J. Fischer- Lougheed, Cloning of a human ether-a-go-go potassium channel expressed in myoblasts at the onset of fusion, FEBS Lett. 434 (1998) 177–182. [18] M. Ju, D. Wray, Molecular identification and characterisation of the human eag2 potassium channel, FEBS Lett. 524 (2002) 204–210.
M. Ju, D. Wray / Biochemical and Biophysical Research Communications 342 (2006) 1088–1097 [19] R. Schonherr, G. Gessner, K. Lober, S.H. Heinemann, Functional distinction of human EAG1 and EAG2 potassium channels, FEBS Lett. 514 (2002) 204–208. [20] R. Schonherr, L.M. Mannuzzu, E.Y. Isacoff, S.H. Heinemann, Conformational switch between slow and fast gating modes: allosteric regulation of voltage sensor mobility in the EAG K+ channel, Neuron 35 (2002) 935–949. [21] M. Ju, L. Stevens, E. Leadbitter, D. Wray, The Roles of N- and Cterminal determinants in the activation of the Kv2.1 potassium channel, J. Biol. Chem. 278 (2003) 12769–12778. [22] B. London, M.C. Trudeau, K.P. Newton, A.K. Beyer, N.G. Copeland, D.J. Gilbert, N.A. Jenkins, C.A. Satler, G.A. Robertson, Two isoforms of the mouse ether-a-go-go-related gene coassemble to form channels with properties similar to the rapidly activating component of the cardiac delayed rectifier K+ current, Circ. Res. 81 (1997) 870–878. [23] I. Splawski, J. Shen, K.W. Timothy, M.H. Lehmann, S. Priori, J.L. Robinson, A.J. Moss, P.J. Schwartz, J.A. Towbin, G.M. Vincent, M.T. Keating, Spectrum of mutations in long-QT syndrome genes. KVLQT1, HERG, SCN5A, KCNE1, and KCNE2, Circulation 102 (2000) 1178–1185. [24] A. Paulussen, A. Raes, G. Matthijs, D.J. Snyders, N. Cohen, J. Aerssens, A novel mutation (T65P) in the PAS domain of the human potassium channel HERG results in the long QT syndrome by trafficking deficiency, J. Biol. Chem. 277 (2002) 48610–48616. [25] W.N. Zagotta, N.B. Olivier, K.D. Black, E.C. Young, R. Olson, E. Gouaux, Structural basis for modulation and agonist specificity of HCN pacemaker channels, Nature 425 (2003) 200–205. [26] A. Scholle, R. Koopmann, T. Leicher, J. Ludwig, O. Pongs, K. Benndorf, Structural elements determining activation kinetics in Kv2.1, Receptors Channels 7 (2000) 65–75. [27] Y. Li-Smerin, D.H. Hackos, K.J. Swartz, A localized interaction surface for voltage-sensing domains on the pore domain of a K+ channel, Neuron 25 (2000) 411–423. [28] G. Gessner, M. Zacharias, S. Bechstedt, R. Schonherr, S.H. Heinemann, Molecular determinants for high-affinity block of human EAG
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
1097
potassium channels by antiarrhythmic agents, Mol. Pharmacol. 65 (2004) 1120–1129. S.J. Cushman, M.H. Nanao, A.W. Jahng, D. DeRubeis, S. Choe, P.J. Pfaffinger, Voltage dependent activation of potassium channels is coupled to T1 domain structure, Nat. Struct. Biol. 7 (2000) 403–407. P. Zerr, J.P. Adelman, J. Maylie, Episodic ataxia mutations in Kv1.1 alter potassium channel function by dominant negative effects or haplo insufficiency, J. Neurosci. 18 (1998) 2842–2848. F. Bretschneider, A. Wrisch, F. Lehmann-Horn, S. Grissmer, Expression in mammalian cells and electrophysiological characterization of two mutant Kv1.1 channels causing episodic ataxia type 1 (EA-1), Eur. J. Neurosci. 11 (1999) 2403–2412. P. Imbrici, A. Cusimano, M.C. D’Adamo, A. De Curtis, M. Pessia, Functional characterization of an episodic ataxia type-1 mutation occurring in the S1 segment of hKv1.1 channels, Pflugers Arch. 446 (2003) 373–379. T.M. Ishii, M. Takano, H. Ohmori, Determinants of activation kinetics in mammalian hyperpolarization-activated cation channels, J. Physiol. 537 (2001) 93–100. J. Stieber, A. Thomer, B. Much, A. Schneider, M. Biel, F. Hofmann, Molecular basis for the different activation kinetics of the pacemaker channels HCN2 and HCN4, J. Biol. Chem. 278 (2003) 33672–33680. E. Aydar, C. Palmer, Functional characterization of the C-terminus of the human ether-a-go-go-related gene K+ channel (HERG), J. Physiol. 534 (2001) 1–14. A. Scholle, T. Zimmer, R. Koopmann, B. Engeland, O. Pongs, K. Benndorf, Effects of Kv1.2 intracellular regions on activation of Kv2.1 channels, Biophys. J. 87 (2004) 873–882. D.A. Doyle, J. Morais Cabral, R.A. Pfuetzner, A. Kuo, J.M. Gulbis, S.L. Cohen, B.T. Chait, R. MacKinnon, The structure of the potassium channel: molecular basis of K+ conduction and selectivity, Science 280 (1998) 69–77. Y. Jiang, A. Lee, J. Chen, V. Ruta, M. Cadene, B.T. Chait, R. MacKinnon, X-ray structure of a voltage-dependent K+ channel, Nature 423 (2003) 33–41.