The protein for the swelling-activated chloride conductance is one of the few ion channels left for which we do not have a universally acceptable clone. In this brief note, I hope to describe why chloride channels in general, and ICl.swell in particular, have been difficult to pin down. Dr. Strange describes why he believes the pICln protein is not ICl.swell. I agree. In 1995 we demonstrated that pICln is unlikely to be a channel itself (Krapivinsky et al., 1994). Voets et al. (1996) also concludes that, although oocyte expression of pICln activates a chloride current, this chloride current is not ICl.swell. pICln is related to chloride channel activation, but how or why is not understood. In an attempt to place the arguments in a broader perspective, Table I lists most of the proposed chloride channel proteins that have been isolated and cloned (see also Jentsch and Gunther, 1997).
Although at least eight proteins have been proposed to comprise chloride channels, only three have been established. When a new protein is proposed to comprise a channel function, it faces more of an uphill battle if it has no homology to known channels. For this reason, I believe it is best to keep an open mind about the phospholemman, p64, and Ca-CC proposed channel types. Of the eight proteins listed, only ClC-2 and ClC-3 are viable candidates for the swelling-activated chloride channel itself, although other proteins cannot yet be excluded from participating in activation of the swelling current.
Why has there been difficulty in nailing down chloride channels?
So far, no chloride channel sequence resembles a known voltage-gated channel, the most well-established group of channel proteins. Thus, at least initially, homology to known channels was not helpful in identifying these channels. Second, if one accepts the results of numerous mutagenesis studies on chloride channels, one must conclude that the chloride pore either involves the whole protein or there is more than one way to make a chloride-conducting pore. From studies on GABAA, glycine, CFTR, and ClC proteins, no chloride-selective consensus domain has emerged. Third, the most common expression cloning system, Xenopus oocytes, has numerous background chloride channels, some of which seem to be activated by expression of almost any protein (Tzounopoulos et al., 1995). Finally, the lipid bilayer method is too sensitive to contaminating proteins to use as a reliable assay for identification of novel protein function. In a picogram of 99.9% pure protein, there are >10,000 contaminating protein molecules. Even one of these molecules can be detected if it inserts into the membrane, and this has led to numerous false identifications of various proteins as ion channels. The initial methods that have led to successful identification of chloride channels involved (a) ligand binding, purification, and microsequencing (glycine, GABAA), (b) genetic approaches (CFTR), and (c) a modified method of expression cloning in Xenopus oocytes involving hybrid depletion (ClC-0).
Why has there been difficulty in finding the swelling-activated chloride conductance?
Most of the difficulties are related to the problems mentioned above. But it is not clear that ICl.swell is represented by a single channel type, given the range of properties that have been described for these currents (Okada, 1997). Finally, it is likely that ICl.swell is regulated by other proteins that are involved in cell swelling, such as the cytoskeleton. Swelling, like temperature, has fairly broad repercussions in cells. This means that expression of all kinds of proteins may trigger the activation of the swelling current. This activation may be direct—or very indirect.
We expression-cloned ICln (Paulmichl et al., 1992). Other groups confirmed that its expression leads to activation of a chloride conductance (Abe et al., 1993; Buyse et al., 1997). When we published our expression cloning of pICln, we stated that “the assumption that the protein is a chloride current is the simplest conclusion, but more complex interpretations are feasible” (Paulmichl et al., 1992). After we purified the protein, made antibodies, and studied it for 2 yr, we decided it was unlikely to be the channel itself since it was soluble, mainly cytoplasmic, and abundant (Krapivinsky et al., 1994). We thought it likely the channel was linked somehow indirectly to endogenous oocyte ICl.swell, that it somehow regulated ICl.swell (Krapivinsky et al., 1994). Gschwentner et al. (1996), however, disagreed and maintain that the protein does comprise the channel itself. Voets et al. (1996) and Buyse et al. (1997) have presented evidence that ICln does not evoke ICl.swell, but does evoke a swelling-insensitive Cl channel with similar sensitivity to nucleotide block. Overall, the weight of the evidence is that pICln is indirectly related to chloride current activation in expression systems, but that it is probably not ICl.swell, nor even a chloride channel itself. Since our finding that pICln is not an integral membrane protein and is associated with several other cytoplasmic proteins (Krapivinsky et al., 1994), my opinion is that pICln's function has yet to be discovered.
One way science is done is to formulate a hypothesis, and then subject it to scrutiny. In my opinion, the hypothesis that either P-glycoprotein or pICln comprise chloride channels in themselves has been rejected by experimentation. The current evidence that ClC-2 or ClC-3 are themselves chloride channels is strong. Whether they will turn out to be forms of ICl.swell as currently defined will require more experiments. Certainly it will be interesting to discover how cell swelling is translated into gating of a presumed integral membrane protein. The swelling-sensing and channel-gating mechanism will likely require several proteins, and pICln may turn out to play a role in one of these steps. But the only comment I can make with certainty is that we have not seen the end of proteins proposed to comprise ICl.swell.