In article <5vom6o$13aq at f1n1.spenet.wfu.edu> John Mihic, jmihic at bgsm.edu
writes:
>In case you're interested, here are some
>other papers which did NOT show an increase in conductance by
>benzodiazepines.
And what an increase! Ten-fold in some cases.
A few methodological comments:
In the recent paper (Eghbali, Curmi, Birnir and Gage, Nature 388,71-75),
a rather unusual drug application method was used. Most previous studies
have used outside-out patches. Instead, the authors usually used on-cell
patches, and *injected* GABA and diazepam into the recording pipette
during the experiment. It is not clear why this was done, and one might
worry about the possible effects of changing the intrapipette pressure
(e.g., activation of stretch sensitive chloride channels).
Along a similar line, I can recall no experiment in which they showed
that the large conductance was in fact sensitive to GABA-A receptor
antagonists (correct me if I missed it). This would seem to be a crucial
control.
The effect reported was highly variable from cell to cell, and showed a
"ceiling" effect. That is, channels that were already large were less
potentiated. Once again, one might worry that some of the observed
chloride channels (the larger ones) were something other than GABA-A
channels.
Drug washout was not shown. Because stationarity is such a great concern
in single channel studies, this also seems to be an important control.
The drug effect was "graded". That is, rather than the amplitude
histogram shifting in weight from low amplitude to high amplitude peaks,
the peak remained sharp but moved gradually to the right. This suggests
that there must be a "graded" response of *each* channel (or each
functionally coupled cluster of channels, if such things exist) to
diazepam, rather than a single binding site with an all or none action.
In fact, such a graded effect suggests that there must be an awful lot of
diazepam sites (more than five, say) per channel, each of which can
modulate conductance to some degree. Oddly, whole cell diazepam
dose-response curves usually give pretty low Hill slopes, which might be
interpreted as there being only one or two diazepam sites per channel. In
fairness, though, both the multibarreled pore idea and the binding
stoiciometry of GABA and benzos remain open questions, and are not easy
to address experimentally.
Obviously it's good to keep an open mind about such a potentially
important finding. But the potential importance is also a good reason to
show a lot of stringent controls.
Matt Jones
