Accessibility of Site 1304 and Fast Inactivation
The major result of this work is that the behavior of the Na+ channel fast inactivation gate is independent of the state of the slow inactivation gate. These observations were only possible because the ability to monitor the conformational rearrangement accompanying fast inactivation allowed us to observe electrophysiologically silent transitions between channel states. Thus, the first set of experiments we performed was directed at characterizing our putative conformational marker for the fast inactivation gate.
We selected F1304 as a candidate site because previous studies had identified it as an important residue for fast inactivation (West et al., 1992
), and because there was some evidence that site 1304 becomes less solvent-accessible during inactivation (Vassilev et al., 1988
). Kellenberger at al. (1996) recently examined the effects of different MTS reagents and Ag+
on the analogous mutation (F1489C) in rat brain IIA channels. A primary aim of that study was to examine the effects of substituting groups of different charge and size at position 1489 by determining on- and off-rates for the inactivation particle using single channel recording. The effect of cysteine substitution and MTS-ET modification were similar to what we measured in the muscle channel: the rate of MTS-ET modification of site 1489 in the brain IIA channel, 0.786 μmol−1
, agrees well with the rate we measured for site 1304 in μ1, 0.932 μmol−1
. Kellenberger et al. (1996)
were also able to demonstrate that site 1489 in the brain IIA channel exhibits voltage-dependent accessibility that corresponds to the degree of inactivation present and that the modification rate is not intrinsically voltage dependent. Without rapid solution exchange, however, they could not use a defined prepulse duration before exposure for precise comparison with the h∞•
curve, nor were they able to examine the effects of slow inactivation on site 1489 accessibility.
While much of the present paper confirms the findings in Kellenberger et al. (1996)
, our use of a fast solution exchange system allowed us to control the duration of MTS-ET exposure down to the 20-ms level, permitting a more precise comparison between the h∞•
curve and the voltage dependence of the modification rate. When we compared the accessibility of cys1304 after 200-ms prepulses to the h∞•
curve (also measured with 200-ms prepulses), we found that the two different measures of inactivation agree quite well. In addition to establishing the correlation between cys1304 accessibility and fast inactivation, this result also provides evidence that the inactivation from closed states that occurs during mildly depolarized prepulses involves burial of site 1304 just as inactivation from the open state does. The relatively small differences in V1/2
= 3.5; Δk = 0.9) between h∞•
and the fit to the modification data might reflect gating shifts that occur in macropatches over time since modification data were, on average, recorded slightly later than h∞•
data and took longer to acquire. It is also possible, though the resolution of the data does not permit us to draw such a conclusion, that the small left shift in the fit to the modification data could reflect an additional conformational transition required for channel inactivation after site 1304 burial.
The fit of our modification data to a Boltzmann yielded modification rates Rmax and Rmin of 0.932 and 0.208 μmol−1 s−1 for hyperpolarized and depolarized potentials, respectively. Even if cys1304 is completely inaccessible for fast-inactivated channels, Rmin has a predicted lower limit because fast inactivation is incomplete for F1304C. The ratio of Rmin to Rmax, 0.22, should (in theory) be greater than or equal to the fraction of channels that fails to inactivate at strongly depolarized voltages.
Precise estimation of the fraction of channels that are not fast inactivated at equilibrium is complicated by the fact that these channels can become slow inactivated. Close examination of h∞• for F1304C reveals a steep descent that ends at ~−65 mV, followed by a much shallower descent (Fig. B). This second shallow component is absent from the WT h∞• curve, where fast inactivation is nearly complete. We interpret the shallow descent as the superimposition of slow inactivation onto the fraction of non–fast-inactivating channels. The 200-ms prepulses used to measure h∞• allows some slow inactivation to occur at these voltages, but equilibrium is not reached, so the apparent voltage dependence is mild. The value of h at the plateau (i.e., before the onset of the shallow descent) gives a rough estimate of the fraction of channels that are not fast inactivated at equilibrium, ~15%. The two-pulse recovery data in Fig. are also consistent with this estimate. After a 20-ms conditioning pulse, 10–15% of the current recovers within 0.1 ms, suggesting that this fraction of channels did not fast inactivate for F1304C. Using this estimate, the theoretical minimum for Rmin/Rmax is 0.15, which is below our measured value of 0.22. This difference suggests that burial of cys1304 does not reduce the reaction rate with MTS-ET to zero, although the imprecision in our estimate diminishes the certainty of this interpretation.
Despite the uncertainty in our estimate, we can nevertheless calculate approximate values for the rate of modification for buried channels (Rb) and the rate for completely accessible channels (Ra). The measured apparent rate at any given voltage (Rapp) is an average of Ra and Rb weighted by the equilibrium value of the h parameter at that voltage: Rapp = (h)Ra + (1 − h)Rb. At −120 mV, h is nearly one, and Rapp = Rmax. Thus, Ra = Rmax = 0.932 μmol−1 s−1. For depolarized voltages, we estimated h at 0.15, and Rapp = Rmin (0.208 μmol−1 s−1). Thus, Rmin = 0.208 = (0.85)Ra + (0.15)Rb. Substitution of 0.932 for Ra gives Rb = 0.08 μmol−1 s−1; and therefore Rb/Ra = 0.09.
According to this calculation, the reaction rate when cys1304 is buried is ~10-fold slower than when it is accessible. The fact that Rb is nonzero implies that there is some degree of accessibility to MTS-ET modification, even when cys1304 is buried. This may be an artifact of the cysteine substitution for a normally inaccessible phenylalanine, or it may reflect an unexpected degree of exposure for site 1304 in the inactivated conformation; our data alone cannot distinguish between these possibilities.
Kellenberger et al. (1996)
obtained a ratio of Rmin
for F1489Q in brain IIA channels modified with MTS-ES (negatively charged and smaller than MTS-ET) of 0.04 as compared with our value of 0.22 for F1304C in μ1 modified with MTS-ET. Part of this discrepancy is probably attributable to the fact that fast inactivation is more complete in the brain IIA F1489C than in μ1 F1304C. Additional contributing factors might include other isoform differences between brain IIA and μ1 (such as the nature of the buried conformation), or the use of a different modification reagent. Since the data given in Kellenberger et al. (1996)
do not permit an estimation of the equilibrium value of h
at depolarized voltages, a precise comparison with our data is not possible.
was much faster than MTS-ET modification rates reported for other sites on the sodium channel: 1,000-fold faster than modification of the site of the outermost S4 charge in domain IV (Yang and Horn, 1995
), and 3,000-fold faster than the site of the third domain IV S4 charge (Yang et al., 1996
). Our rate is also ~1,000× faster than modification at sites of S4 charges in Shaker
channels (Larsson et al., 1996
), and ~50× faster than a site at the outer mouth of the Shaker
channel pore (Liu et al., 1996
). The relatively high sensitivity of cys1304 to MTS-ET suggests a great degree of solvent accessibility of the site at −120 mV, an interpretation consistent with its putative role as a relatively mobile inactivation particle.
Finally, we were also able to evaluate the time dependence of the accessibility change, though the temporal resolution of our method prohibited precise comparison with rates of entry to and recovery from fast inactivation. Cys1304 is buried after 7 ms at 0 mV, and cys1304 accessibility recovers within 12 ms of repolarization after a 20-ms conditioning pulse to 0 mV. Although these times are substantially longer than the rates for entry to and recovery from fast inactivation, the data do allow us to constrain the rates of burial and recovery of accessibility of cys1304 to within an order of magnitude of the fast inactivation rates measured for ionic currents. This result, along with the close agreement of the voltage dependence of cys1304 accessibility with the h∞• curve, allows us to conclude that cys1304 is indeed an excellent marker for the position of the fast inactivation gate.
Relationship between Fast and Slow Inactivation
In the first experiment designed to test the relationship between fast and slow inactivation, the position of the fast inactivation gate was measured at the tail end of different length depolarizations. If fast and slow inactivation are mutually exclusive, directly competing processes (considered in Featherstone et al., 1996
), then cys1304 would be accessible at the tail end of long conditioning pulses when most channels are slow inactivated. If, on the other hand, channels can exist in a state that is fast and slow inactivated simultaneously (suggested in Bezanilla et al., 1982
), then cys1304 would remain buried regardless of the length of the conditioning pulse. Our data demonstrate that cys1304 does remain buried regardless of depolarization length, proving that fast and slow inactivation are not mutually exclusive processes; that is, that they can occur simultaneously in the same channel.
These data also show that accessibility does not progressively decline during prolonged depolarizations. The fact that the modification rate does not decrease with longer conditioning pulses suggests that a certain fraction of channels in F1304C remain non–fast inactivated regardless of the length of the conditioning pulse used. This is consistent with the interpretation that the decay of the slow component in the F1304C current trace (Fig. A) is due to slow inactivation and not to fast inactivation, and supports the generally accepted hypothesis that fast and slow inactivation are structurally distinct. In other words, the equilibrium between fast and non–fast inactivated channels is achieved within a few milliseconds at 0 mV and is not measurably biased by the closure of the slow inactivation gate.
In the next experiment, the MTS exposure was placed shortly after the end of variable length conditioning pulses, allowing us to determine whether recovery from fast inactivation could precede recovery from slow inactivation. We found that the accessibility of cys1304 almost completely returned within 12–32 ms, regardless of the length of the conditioning pulse used. For the longest conditioning pulse used, 10 s, the interval between 12 and 32 s corresponds to just 5–15% availability (Fig. , shaded area). This result allows us to draw several conclusions. First, fast and slow inactivation must be structurally independent processes, because the fast gate can be almost completely open while the slow gate is nearly completely closed. Second, it rules out a sequential model for the development of and recovery from slow inactivation, according to which sodium channels could only recover from fast inactivation after recovering from slow inactivation. Third, it indicates that in the normal process of recovery from slow inactivation, a state is populated that is slow inactivated but not fast inactivated. It also suggests that fast and slow inactivation are coupled only weakly, if at all, since neither the equilibrium of fast- and non–fast-inactivated channels nor the kinetics of recovery from fast inactivation are altered by the closure of the slow gate. The data from both experiments would support a model in which fast and slow inactivation are independent processes with relatively little interaction.
Relation to Previous Work on Slow Inactivation of Na+ Channels
The relationship between fast and slow inactivation has been difficult to probe, primarily because transitions that occur between different inactivated states are electrophysiologically silent. One line of investigation on the subject has focused on the long-term immobilization of gating charge. For short depolarizations, recovery from inactivation closely parallels recovery of gating charge (Armstrong and Bezanilla, 1977
), lending support to the hypothesis that sodium channels must deactivate to recover from inactivation (Kuo and Bean, 1994
). In response to prolonged depolarizations, however, some groups have found that available gating charge recovers on the order of hundreds of milliseconds, but still well before recovery from slow inactivation (Bezanilla et al., 1982
; Ruben et al., 1992
); while another group reports that gating charge recovers with the same time course as slow inactivation (Meves and Vogel, 1977
). Our results demonstrate that the accessibility of the fast inactivation gate recovers much more rapidly than this, implying that some other process (perhaps slow inactivation) must be responsible for long-term immobilization of gating charge. Further experiments are needed to determine how recovery from fast inactivation and recovery of gating charge are decoupled during recovery from long depolarizations, and whether this process requires slow inactivation.
Additional data on the relationship between fast and slow inactivation has come from studies on the subtle changes in slow inactivation brought about by disruption of fast inactivation. Several groups have found that slow inactivation is more complete and its onset more rapid after fast inactivation has been removed or disrupted (Featherstone et al., 1996
; Hayward et al., 1996
; Rudy, 1978
). We also found this to be the case for F1304C compared with WT (data not shown). This finding implies either direct competition between the two inactivation gates, or an indirect coupling effect in which the rate of slow inactivation is faster from the open state than from fast-inactivated states (Featherstone et al., 1996
; Hayward et al., 1997
; Rudy, 1978
). Our results rule out direct competition for the same binding site because fast and slow inactivation gates can be closed simultaneously; however, they do not rule out a more subtle coupling between the two forms of inactivation.
channels, for example, exhibit coupling between two structurally independent forms of inactivation (rapid intracellularly mediated N-type inactivation, and slower extracellularly mediated C-type inactivation) that occurs indirectly through effects of the N-type inactivation particle on activation gating and on occupancy of an ion-binding site near the outer mouth of the pore (Baukrowitz and Yellen, 1995
). One recent study on the human heart Na+
channel mutant F1485Q found that increasing extracellular Na+
inhibits slow inactivation, while lowering of extracellular Na+
enhances it (Townsend and Horn, 1997
). Previous studies have noted effects of external Ca2+
on slow inactivation of Na+
channels in frog nerve (Khodorov et al., 1976
). These results have been taken as evidence that binding of a Na+
ion to a site near the outer mouth of the pore inhibits closing of the slow inactivation gate (Townsend and Horn, 1997
). If closure of the fast inactivation gate increased the dwell time of an ion at that site, it would explain why fast inactivated channels are slightly more resistant to slow inactivation than open channels.
The results of our study do not directly speak to the possibility of this kind of weak coupling of fast and slow inactivation; however, they do exclude the two extreme cases, that fast and slow inactivation are tightly linked in a sequential gating scheme (employed in Townsend and Horn, 1997
), and that fast and slow inactivation are mutually exclusive processes (considered in Featherstone et al., 1996
). In addition, our data restrict the number of possible recovery pathways from slow inactivated states: recovery from fast inactivation precedes recovery from slow inactivation.