Engineered Ionizable Side Chains 15
the transmembrane pore proper). In fact, the latter may well be the reason why ly-
sines engineered at positions lying outside of the membrane thickness (such as po-
sitions − 5ʹ, − 6ʹ and − 7ʹ in the intracellular M1–M2 linker and positions 21ʹ–25ʹ in
the extracellular M2–M3 linker) had no impact on the single-channel conductance.
Also, we think that this is likely the reason why protonation and deprotonation
events involving the large number of naturally occurring ionizable side chains in
the extracellular domain and in the cytoplasmic M3–M4 linker of the muscle AChR
are electrically silent. On the other hand, a lack of effect of ionizable side chains
engineered close to the pore domain would be expected from a basic group that is
largely deprotonated (as a result of a down-shifted pKa) or from an acidic group that
is largely protonated (as a result of an up-shifted pKa), as would be the case for side
chains located in highly hydrophobic regions of the protein or in close proximity to
other charged moieties of the same sign.
pKa Values In the same way as the quantification of the extent of block provides
information about the proximity of an engineered side chain to the axis of ion per-
meation, analysis of the occupancy probabilities of the different levels of single-
channel current leads to the pKa of the side-chain’s ionizable group (Fig. 4 ). Indeed,
inasmuch as sojourns in the different levels of open-channel current represent inter-
vals in the protonated and deprotonated states of the engineered side chain, all that
is needed to calculate the pKa of the ionizable moiety ( pKa = − log Ka, where Ka is the
acid-dissociation equilibrium constant) is the ratio of deprotonated-to-protonated
occupancies and the solutions’ pH. Single-channel recordings, however, allow us
to go one step further and estimate the rates at which protons bind and unbind; to
this end, a kinetic model is needed. Although identifying a proper kinetic model can
be difficult, the rates we are interested in extracting from the data are those of the
interconversion between two open-channel levels, and hence, there is no need for
a model where all shut states and their connectivities make physical sense. Essen-
tially, the idea is that a ligand-gated ion channel with an engineered ionizable side
chain can occupy the different conformational states (closed, open and desensitized)
and ligand-binding states (unliganded, partially liganded, fully liganded) with or
without a proton bound to the engineered side chain (Fig. 9 ). Thus, a sojourn in
the “deprotonated open” state may be terminated by a shutting (that is, a channel
closure or a transition to the desensitized conformation) or by a protonation event.
Similarly, a sojourn in the “protonated open” state may be terminated by a shutting
or by a deprotonation event. This is the only portion of the kinetic scheme that needs
to reflect physical reality; whether shutting leads to a closed or a desensitized state,
and whether there are multiple or just one shutting rate leading away from each
open state, for example, is inconsequential. But even when simplified, the appropri-
ate kinetic model need not be simple. Certainly, the data may require that multiple
open states—each one with its deprotonated and protonated forms—be included in
the model, as would be the case for a ligand-gated ion channel that can give rise to
openings of different mean duration depending on the number of ligand-molecules
bound; if this were the case, then each type of open state would allow for the esti-
mation of a separate, eventually different, pKa. Also, the distributions of protonated