Reply from Jamie I. Vandenberg, Adam P. Hill, Terence J. Campbell, Catherine E. Clarke

Professor Tseng has objected to the statement in the discussion of our paper: ‘… the majority of the S5P α-helix cysteine [mutants] that had marked changes in reversal potential… form intersubunit disulphide bonds… significantly disrupted outer pore structure’… and its implication that the data interpretation in Liu et al. (2002) was incorrect.

Firstly, it is important to note that in our paper (Clarke et al. 2006) we defined the S5P α-helix as occurring from W585-I593 (see Fig. 2, based on NMR spectroscopy studies from our laboratory: Torres et al. 2003). Clearly the S5P α-helix cysteine mutants have altered function (eliminated inactivation and loss of selectivity for K⁺ over Na⁺, as described in Liu et al. (2002) and resummarized in the letter from Professor Tseng). All nine of the S5P α-helix cysteine mutants had markedly reduced selectivity for K⁺ over Na⁺ and in 7 of the 9 mutants (all but H587C and Q592C), the markedly reduced selectivity for K⁺ over Na⁺ was not reversed by the addition of DTT. Subsequently, Jiang et al. (2005) showed that all nine of the S5P α-helix cysteine mutants formed disulphide bonds. There is no doubt that the formation of disulphide bonds would significantly disrupt the structure of the outer pore region of hERG channels. Yet, even in the absence of disulphide bonds (following the addition of DTT), 7 of the 9 S5P α-helix cysteine mutants still have markedly reduced selectivity for K⁺ over Na⁺. This is the real issue at stake – how to interpret, from a structural perspective, data obtained from mutants that have markedly perturbed function (even in the absence of disulphide bonds).

In any mutagenesis study one must be wary that there are multiple mechanisms by which any mutation may affect protein function. There may be a direct effect on function if the residue mutated is critical for that function. Alternatively, effects may be indirect due to disturbances to interactions with other residue(s) that affect protein function or the mutation may cause a sufficiently gross perturbation to structure that the protein is in effect defunct. In the case of hERG inactivation, does loss of selectivity for K⁺ over Na⁺ (when measured in the presence of both Na⁺ and K⁺, as was the case in the Liu et al. study) constitute a sufficient disruption to function to make one suspicious that the protein is ‘defunct’ and therefore should make one wary about speculating on the structural significance of the mutants?

In her letter, Professor Tseng argues that the recent paper from Gang & Zhang (2006) demonstrating that the p-inactivated state of hERG conducts Na⁺, supports the hypothesis that mutants which affect inactivation of hERG can also affect its ability to discriminate between Na⁺ and K⁺ ions. However, the data in the Gang and Zhang paper show that it is only the p-inactivated state (in wild type and a series of mutant channels that still inactivate), that conducts Na⁺ and this can only be observed when intracellular K⁺ is removed. Gang and Zhang went to great lengths to show that channels were not ‘defunct’ and that the open state was still highly selective for K⁺ over Na⁺. This is very different to the situation with the S5P α-helix cysteine mutants that have a reduced selectivity for K⁺ over Na⁺ permeation through the open state (measured in the presence of high intracellular K⁺ and variable extracellular K⁺ and Na⁺).

We agree that hERG K⁺ channel inactivation is a complex phenomenon, and involves at least two distinct conformational states (as clearly shown by Zhang & Gang, 2006). It is therefore not surprising that different mutants can have different effects on inactivation with there being at least two clearly distinct phenomena: (1) altered voltage dependence, e.g. S631A (Zou et al. 1998), N588K and Q592K (Clarke et al. 2006) cause ∼+100 mV shift and N588E ∼–30 mV shift (Clarke et al. 2006) in the voltage dependence of steady state inactivation, and (2) inability to inactivate (e.g. S620T, Ficker et al. 1998). In all of these mutants, however, selectivity for K⁺ over Na⁺ is retained. We believe that any mutant that results in significantly reduced selectivity for K⁺ over Na⁺ (through the open state) should raise concerns about disrupted pore structure. If those mutants, such as the majority of the S5P α-helix cysteine mutants, also have eliminated inactivation one must suspect disruption of native protein structure.

In summary, we do not have any dispute with the data presented in either the Liu et al. (2002) or Jiang et al. (2005) studies. These studies clearly show that the S5P linker region, and the S5P α-helix in particular, is important for inactivation and it has a highly dynamic structure. We also agree with Tseng and colleagues that when studying mutational effects on hERG inactivation it is important to determine changes in P_K:P_Na. However, we urge great caution about using data obtained from any mutants that have markedly reduced selectivity for K⁺ over Na⁺ and complete loss of inactivation, to draw conclusions about the structural basis of inactivation. Equally and great caution should be exercised when using such mutants to interpret the significance of altered binding affinities for toxins that interact with the outer pore domain of hERG channels (Pardo-Lopez et al. 2002; Zhang et al. 2003).