RESEARCH SUMMARIES

The research of our group is concerned with ion transport processes, in particular chloride and potassium channels, chloride-proton exchangers, as well as potassium-chloride cotransporters. We investigate their structure-function relationship and study their cell biology, physiology, and role in human disease. Investigations of these latter aspects are greatly facilitated by a number of knock-out mouse models generated in our laboratory.

A main focus of our work is on CLC chloride channels and transporters , a gene family we discovered in 1990. Originally thought to represent only plasma membrane Cl^- channels, it is now clear that many of these rather function as electrogenic Cl^-/H⁺-exchangers in mainly intracellular membranes. We have  disrupted most CLC genes in mice  which led to important new insights. For instance, the disruption of the chloride transporter ClC-7 unexpectedly led to osteopetrosis. This phenotype is due to a dysfunction of osteoclasts, the cells that are responsible for  bone degradation. ClC-7 currents may balance the electrogenic transport of the H⁺-ATPase that acidifies the resorption lacuna. Stimulated by these findings, we found mutations in either the ClC-7 chloride channel or the a3 subunit of the H⁺-ATPase in human patients with severe juvenile osteopetrosis. ClC-7 is normally present in a lysosomal compartment, but is inserted into the plasma membrane of osteoclasts. Similarly, our results show that ClC-3 through ClC-6 reside normally in membranes of intracellular compartments, where they may contribute to their acidification by electrical shunting of the proton ATPase. More recently, we have demsontrated that ClC-7 requires the β-subunit Ostm1 for function. OSTM1 mutations also cause osteopetrosis in mice and man. ClC-5, a Cl^-/H⁺ exchanger mutated in the kidney stone disorder Dent’s disease, is present in renal endosomes. Our knock-out mouse model revealed that an impaired endosomal acidification led to a broad defect in proximal tubular endocytosis. This entails secondary changes in calciotropic hormones, eventually resulting in the hyperphosphaturia and hypercalciuria (and kidney stones) in Dent’s disease. The related ClC-3 putative Cl^-/H⁺ exchanger is also present in endosomes, as well as in synaptic vesicles. As the uptake of neurotransmitters into synaptic vesicles is coupled to the proton gradient, this may have interesting consequences for synaptic transmission. Surprisingly, the knock-out of ClC-3 led to nearly complete degeneration of the hippocampus.
The further elucidation of the roles of the five intracellular CLC chloride transporters e.g. in trafficking, sorting, and synaptic transmission is a major goal of our future research. Our knock-out mouse models, which we cross to yield multiple knock-outs, will greatly facilitate this task.

In another area, we focus on KCC K-Cl cotransporters, in particular on their role in the regulation of intracellular Cl^--concentration in neurons. This concentration determines whether the response to the neurotransmitters GABA and glycine is inhibitory (as in most adult neurons) or excitatory (as early in development). ClC-2 is thought to play a role in this regulation, but the disruption of this Cl^- channel led to male infertility, blindness  and leukodystrophy rather than to CNS hyperexcitability that may have been expected. By contrast, the disruption of the neuronal K-Cl cotransporter KCC2 led to perinatal death due to the inability to breathe and to a spastic phenotype. This was indeed due to an increase in intracellular chloride in motoneurons. The knock-out of KCC4 led to a rapid hearing loss that was associated with renal tubular acidosis, whereas the disruption of KCC3 led to a degeneration of the nervous system, deafness, and hypertension. Finally, we have shown that KCC1 and KCC3 co-operate in regulating the volume of red blood cells. Genetic elimination of both transporters partially alleviate the symptoms of sickle-cell anemia.

Finally,  our third area of research focuses on KCNQ potassium channels. We have shown that mutations in KCNQ2 and KCNQ3, which can form heteromeric channels, may lead to neonatal epilepsy. KCNQ2, KCNQ3, and KCNQ5 mediate neuronal M-currents, a highly regulated potassium current that is already active at resting potentials and that sensitively regulates neuronal excitability. We  also identified KCNQ4 and showed that it is mutated in a form of dominant deafness. It is expressed in hair cells of the inner ear, and surprisingly also in the central auditory pathway. We generated mouse models for KCNQ4 to elucidate its role in human deafness. Some KCNQ K⁺ channels can associate with ß-subunits of the KCNE family. We provided evidence that KCNQ1/KCNE3 heteromers are important for Cl- secretion in the intestine.

1.  CLC CHLORIDE CHANNELS AND CHLORIDE/PROTON EXCHANGERS

1.1 Structure and Function of CLC Chloride Channels and Chloride/Proton Exchangers

The Torpedo channel ClC-0 was postulated by Chris Miller to be a double-barrel channel (havin two pores) based on reconstitution studies of the native protein. Once we had cloned ClC-0, we demonstrated that the single Torpedo cDNA was sufficient to give the full 'double-barreled' picture in single channel recordings. We subsequently used concatemers linking wild-type and mutant subunits to demonstrate  that the resulting channel had one normal (wild-type) and one mutant pore. Using concatemers between ClC-0 and ClC-2, we showed that each permeation pathway is contained within a single subunit of a dimeric protein. This was beautifully confirmed by the crystal structure of bacterial CLC proteins by Dutzler and MacKinnon.

Whereas the gating of typical cation channels like Kv K channels depends on charged amno-acids that are located within the lipid bilayer, we proposed a totally different mechanism for CLC channels. In this simple model ('gating by the permeant anion'), chloride  enters the pore in a voltage-dependent manner (as it 'sees' the voltage that drops across the bilayer) and by binding to some site shifts the equilibrium to an open state. This model is now interpreted and refined on the background of the cyrstal structure of CLC proteins: the negative side-chain of a certain 'gating glutamate' blocks the external access of anion to the narrowest part of the pore, suggesting that it moves out when competed for by anion. Indeed, neutralizing this glutamate by converting it to neutral amino-acids abolishes gating in several CLC channels.
Starting from a chimeric approach that exploits the difference in inhibitor sensitivity between ClC-1 and ClC-2, we have identified an inhibitor binding site in ClC-1. Using the crystal structure of the bacterial CLC to guide mutagenesis, we have demonstrated that the inhibitor 9-AC binds in a hydrophobic pocket located close to the narrowest part of the pore. Other structure-function studies from our lab include the identification of structures in ClC-2 that are necessary for the activation of that channel by hyperpolarization  and cell swelling, as well as the identification of a PY-motif  in the carboxyterminus of ClC-5. This motif interacts with certain ubiquitin ligases, leading to the endocytosis of ClC-5. We also showed that certain structures in the carboxyterminus of CLC channels, including the so-called CBS domains, influence a gate that closes both pores simultaneously.

The discovery (by Accardi and Miller) that the E. coli CLC is not a Cl channel, but rather a Cl^-/H⁺ exchanger, prompted us (and M. Pusch) to investigate whether mammalian CLCs may also function as exchangers. Indeed, we demonstrated that ClC-4 and ClC-5 (in contrast to e.g. ClC-0, which IS a channel) are electrogenic Cl^-/H⁺ exchangers. Like in the bacterial protein, neutralizing the 'gating glutamate' uncoupled Cl- from H⁺ fluxes, leading to a pure Cl-conductance. In contrast to the E. coli protein, mutating a so-called 'proton glutamate' at the internal surface of the protein did not uncouple transport, but abolished it totally. (Uncoupled) transport could be rescued by neutralizing additionally the 'gating glutamate'. This suggests that the 'proton glutamate' is necessary to supply protons for the Cl^-/H⁺ exchange at the exchange site that comprises the 'gating glutamate'. With an obligatory exchange, transport is stalled if H⁺ supply is stalled. This can be overcome by uncoupling Cl^- from H⁺ transport by mutating the 'gating glutamate'.

It is fascinating that members of the same gene family can function as Cl channels or Cl^-/H⁺-exchangers. Together with the available crystal structures of CLC proteins, this offers excellent opportunities for structure-function studies.

1.2 The Function of ClC-2 Revealed by a Knock-Out Mouse Model

ClC-2 is an ubiquitously expressed plasma membrane Cl- channel that is activated by hyperpolarization, external acidic pH, and cell swelling. It was hypothesized to play an important role in lung development, gastric acid secretion, and in the determination of intracellular chloride in neurons. However, knock-out mice did not show conspicuous abnormalities in these organs. There was neither a reduced threshold to seizure induction that might have been expected when changing neuronal Cl- concentration. By contrast, we observed a degeneration of male germ cells and photoreceptors. Intriguingly, both cell types depend on the transepithelial transport and nutrition by epithelia (formed by Sertoli cells and retinal pigment epithelial cells, which both express ClC-2 normally) that form a blood-organ barrier. Most recently, we found that ClC-2 KO mice also suffer from leukodystrophy, in which oligodendrocytes display severe vacuolation. The blood-brain barrier of these mice is also disrupted. Nonetheless, except for the blindness, the neurological phenotype of these animals is mild. We hypothesized that ClC-2 plays an important role in regulating extracellular ion concentrations. In the brain, this is similar to the process of potassium-siphoning, in which glia take up K that leaves neurons upon their repolarization.

1.3 Intracellular Cl Transporters ClC-3, ClC-5, ClC-6 and ClC-7: Lessons from Mouse Models and Human Inherited Disease

The role of ClC-3, a broadly expressed Cl- channel, was highly controversial. Whereas we and others did not detect currents upon heterologous expression, four groups published contradictory results on associated currents. ClC-3 was claimed by Duan and Hume to represent the long-sought swelling activated Cl- channel. However, we have now shown that swelling-activated currents are not affected in ClC-3 KO mice. ClC-3 is present in endosomal compartments as well as in synaptic vesicles. This localisation is mutually consistent, as synaptic vesicles are recycled at the synaptic terminal in part via endosomal intermediates. The rate of synaptic vesicle acidification is reduced in the KO, strongly suggesting that ClC-3 provides a conductive pathway to compensate for the charge transfer by the proton ATPase. As neurotransmitter uptake into synaptic vesicles is driven by the electrochemical potential for protons, ClC-3 may play important roles in synaptic transmission. As synaptic vesicle acidification remained Cl- dependent in the absence of ClC-3, we suspect that other CLC channels may have similar roles. Surprisingly, the disruption of ClC-3 led to a drastic degeneration of the hippocampus and the retina. Even in the virtual absence of the hippocampus, Clcn3^-/- mice survived for more than a year and were able to learn motor skills.

ClC-3 is highly expressed in neuroendocrine cells, e.g. in chromaffin cells and in pancreatic β-cells. Using capacitance measurements and amperometry, we have shown that the knock-out of ClC-3 leads to an impaired exocytosis of large dense core vesicles (LDCVs) from chromaffin cells. Because ClC-3 is not detectable on LDCVs, but rather on endosomes and synaptic-like microvesicles, the effect on exocytosis is probably an indirect effect of disturbed intracellular trafficking.

ClC-6 is a late endosomal member of the CLC family. Its disruption also led to lysosomal storage disease, which - in contrast to the one observed with a loss of ClC-7 - did not cause a severe neuronal phenotype. Whereas the storage material in ClC-7 KO mice is distributed over the neuronal cell body, it specificall localized to initila axon segments in mice lacking ClC-6.

We now know that five CLC proteins (ClC-3 to ClC-7) reside primarily in intracellular organelles. We believe that all these proteins function as chloride/proton exchangers rather than chloride channels. The availability of several knock-out mouse models, which we cross to obtain double and multiple knock-outs, will enable us to address important issues of cell biology such as endocytosis from a new perspective.

1.4 Barttin, a ß-Subunit of ClC-K Channels, and Ostm1, a ß-Subunit of ClC-7

Barttin is   a small protein with two transmembrane domains that is mutated in Bartter syndrome type 4, a human inherited diesease associated with massive renal salt loss and deafness. We have shown that Barttin is a ß-subunit of ClC-Ka  and ClC-Kb. Both ClC-Ka and ClC-Kb need barttin for functional expression, an effect that depends at least in part on the abilitiy of barttin to enhance surface expression.
In the ear, barttin is expressed exclusively in basolateral membranes of the stria vascularis, where it co-localises with both ClC-Ka and ClC-Kb. ClC-Ka/barttin and ClC-Kb/barttin heteromers provide a recycling pathway for chloride that is transported into the cell by the basolateral NaK2Cl co-transporter. As a consequence, mutations in barttin lead to a defect in potassium secretion by the stria, leading to deafness. By contrast, mutations in ClC-Ka or ClC-Kb alone do not lead to deafness, as the other channel suffices for secretion.
In the kidney, ClC-Ka and ClC-Kb are expressed in different segments of the nephron. In all these segments, they co-localize with barttin. ClC-K/barttin channels are responsible for the basolateral efflux of chloride. Thus, its disruption leads to a severe salt loss. As predicted, ClC-Kb mutations found in Bartter type 3, as well as barttin mutations found in Bartter type 4 almost always led to a severe loss of ClC-K/barttin currents.

We have generated a knock-out mouse model in which barttin is deleted in the inner ear, but not in the kidney. These mice are congenitally deaf because the epithelium of the stria vascularis can no longer generate the large positive voltage in the scala media. As a consequence, the electromechanical reponse of sensory outer hair cells to sound is abolished. Besides clarifying the pathogenesis of deafness in human Bartter syndrome type IV, this study has revealed an unexpected role of chloride channels in generating the endocochlear potential.

We also identified Ostm1 as a hitherto unknown β-subunit of ClC-7. Ostm1 had been identified by J. Vacher as the gene underlying ostepetrosis in the spontaneous osteopetrotic mouse mutant grey lethal, as well as in a few human patients. We have shown that ClC-7 and Ostm1 form a complex that is transported to lysosomes and the ruffeld border of osteoclasts. Either protein is unstable without its partner. As a consequence, ClC-7 protein levels are severely down in the absence of Ostm1, explaining the osteopetrosis and lysosomal storage in mice and men lacking this subunit.

2. KCC POTASSIUM-CHLORIDE COTRANSPORTERS

2.1 The K-Cl Co-Transporter KCC2 is an Important Regulator of Cl-Dependent Synaptic Transmission

As ClC-2 was expected to play a role in the regulation of neuronal intracellular chloride concentration, we decided to disrupt KCC2 as well. KCC2 is a neuronal isoform of the K-Cl co-transporter gene family. Under most circumstances, K-Cl cotransport will lower intracellular Cl^- below its electrochemical equilibrium. The intracellular Cl^--concentration determines the response to the neurotransmitters GABA or glycine. In early development, this response shifts from excitatory to inhibitory as a consequence of a lowering of the intraneuronal chloride concentration. In the hippocampus of rodents, this shift occurs a couple of weeks after birth.

When we disrupted KCC2 in mice, however, we observed severe deficits already at birth. This is due to the fact that KCC2 is upregulated in the brainstem already before birth. Neonatal mice died from the inability to breathe and displayed a spastic phenotype. Patch-clamping of spinal cord motoneurons revealed that the loss of KCC2 entailed an increase in intracellular chloride concentration that led to an excitatory response to GABA and glycine.

2.2 Loss of K-Cl Co-Transporter KCC3 Leads to Neurodegeneration, Deafness, and Hypertension

KCC3 is expressed in several tissues, including the CNS. When we disrupted KCC3 in mice, we observed a neurodegeneration in the central and peripheral nervous system, which resembles the phenotype of patients with Andermann syndrome in which KCC3 is mutated. The loss of KCC3 led to a moderate increase in intraneuronal Cl-concentration, which, however, is mainly lowered by KCC2. We additionally observed a slowly progressive hearing loss, which we attributed to a defect in inner ear potassium recycling. Additionally, KCC3 KO mice have high blood pressure, which is most likely of neurogenic origin. We demonstrated in cultured neurons and renal proximal tubular cells that KCC3 has a role in regulatory volume decrease. This is also true for red blood cells, where KCC3 operates in parallel to KCC1 (see below).

2.3 Disruption of the K-Cl Co-Transporter KCC4 Leads to Deafness and Renal Tubular Acidosis

With the exception of the CNS, KCC4 is a broadly expressed KCC isoform. Surprisingly, its disruption led to deafness with renal tubular acidosis (28). In the cochlea, KCC4 expression is limited to the supporting cells of sensory hair cells. Our data suggest that it takes up the potassium that leave outer hair cells and delivers it to the gap junction recycling pathway. KCC4 is highly expressed in basolateral membranes of acid secreting intercalated cells. It is crucial for recyling chloride that is exchanged basolaterally for bicarbonate. As a consequence, its disruption impairs renal acid secretion. Together with KCC3, KCC4 plays a role in regulating the volume of renal proxiaml tubular cells.

2.4 Red Blood Cell Volume Regulation Depends on Both KCC1 and KCC3

Red blood cells (RBCs) need to actively regulate their volume during oxygenation/deoxygenation and during their passage through capillaries. Whereas a role of K-Cl cotransport in this process has since long been postulated, the underlying isoforms had not been identified. In sickle cell anemia, K-Cl cotransport acitivity is secondarily increased, which leads to cell shrinkage and more severe sickling. It was therefore suggested that pharmacological inhibition  of K-Cl cotransport might be useful to tret the disease. We generated KCC1 KO mice (which lack immediately visible phenotypes) and crossed them with our KCC3 KO mice. Both KCC isoforms are expressed in red blood cells. Eliminating both isoforms increased RBC volume. We crossed these mice to SAD mice, a transgenic mouse model for human sickle cell disease. While some RBC parameters in SAD mice improved, our study suggests that even a total elimination of K-Cl cotransport activity would not be sufficient to phenotypically 'cure' human sickle cell disease.

3.  KCNQ (Kv7) POTASSIUM CHANNELS

3.1 KCNQ2, -3 and –5 Potassium Channels: a Molecular Basis for Neuronal M-Currents

We had previously shown that KCNQ2 and KCNQ3 can form heteromeric channels, and that their loss of function in a form of neonatal epilepsy (BFNC) leads only to a moderate loss of function. These heteromers yield currents with properties of the important and highly regulated M-current. The increase in current observed upon KCNQ2/KCNQ3 co-expression is mainly due to an increased surface expression. We showed that the novel KCNQ5 protein is also expressed in brain, where it may form heteromers with KCNQ3 (10). In a structure-function study, we have identified a C-terminal domain that determines the subunit-specificity of assembly. Using mutageneisis and biophysical analysis, we have mapped the binding site for retigabine, an opener of KCNQ channels that is in clinical trials for epilepsy treatment. Both KCNQ5 and KCNQ3/5 heteromers have properties of M-currents. Our findings suggested the possibility that KCNQ5 might also be mutated in some forms of epilepsy. However, a first screen of appropriate patient’s DNA for KCNQ5 mutations was negative.
In a novel syndrome, neonatal epilepsy (BFNC) combined with myokymia (spontaneous muscle movements), we identified and analysed a new KCNQ2 mutation. It drastically slowed the activation of KCNQ2 by voltage. The disease, together with the localization of KCNQ2 and KCNQ3 in the spinal chord, revealed an important role of M-currents in motor control.

3.2 KCNQ4, a Potassium Channel Critical for Hearing

The KCNQ4 K⁺ channel is mutated in DFNA2, a dominantly inherited form of progressive hearing loss. In the ear, KCNQ4 is expressed predominantly in outer hair cells of the organ of Corti and in type I vestibular hair cells. KCNQ4 may be required for the basal exit of K⁺ from outer hair cells, which is then recycled to the stria vascularis for a new round of secretion. Surprisingly, KCNQ4 is also expressed rather specifically in the central auditory pathway, pssobly implying  a central component in DFNA2 deafness. We have generated and analyzed two KCNQ4 mouse models: a constitutive KO, as well as a knock-in (KI) carrying a a dominant negative KCNQ4 mutation that we had identified in DFNA2 patient. Both mouse models are able to hear at first, but then develop a hearing loss of about 60 dB. Whereas this occurs within a few weeks in the total KO, heterozygous KI mice became deaf over a couple of months, resembling the slowly progressive hearing loss of DFNA2 patients. The hearing loss can be largely explained by the loss of sensory outer hair cells. We patched these cells before they had degenerated and showed that they had totally lost their M-type K-current. The ensuing  depolarization may lead over the long run to their degeneration by an influx of calcium.

3.3 KCNE3: a ß-Subunit for KCNQ1

KCNE3, a protein with a single transmembrane segment that is homologous to minK (KCNE1), can associate with KCNQ1. Its effect differs dramatically from that of KCNE1: while KCNE1 drastically slows KCNQ1 activation by depolarization and shifts the voltage-dependence to more positive voltages, KCNE3 renders KCNQ1 into a constitutively open channel. Our studies suggest that KCNQ1/KCNE3 underlie the cAMP-stimulated basolateral K⁺ conductance in colonic crypt cells. This conductance is indirectly necessary for Cl^- secretion as it provides a pathway for the recycling of K⁺ that is taken up by the basolateral NaK-2Cl co-transporter.