|Home | About | Journals | Submit | Contact Us | Français|
The cystic fibrosis transmembrane conductance regulator (CFTR) anion channel represents the rate-limiting step for chloride and fluid secretion in most epithelial tissues in the body. More recently, CFTR activity has also been shown to regulate muscle contraction, neuroendocrine function, and cartilage formation, implicating the channel in many important physiological activities from diverse systems. A major interest in the channel stems from the fact that loss of function mutations in the gene encoding CFTR result in the inherited disease cystic fibrosis, one of the most common, life threatening, diseases found in the Caucasian population. At the other end of the spectrum, and affecting far more people globally, over active CFTR causes clinically important secretory diarrhoea induced by toxins from pathogenic bacteria like cholera. Therefore, it is not surprising that much research has focussed on understanding how CFTR channel activity is regulated and what goes wrong in disease states. For the channel to open, it must be first phosphorylated by PKA, and then ATP must also bind to CFTR’s cytoplasmic domains. Now a recent Nature Cell Biology paper has shown that CFTR can also be activated by increases in membrane tension (or stretch), through a phosphorylation and ATP- independent mechanism. This unexpected and novel finding identifies CFTR as a mechanosensitive ion channel. This work could have major implications for our understanding of the biological control of CFTR as well identifying new roles for this channel in mechanosensitive tissues and processes such as regulatory volume decrease and muscle contraction.
Since the CFTR gene was identified in 1989 (Riordan et al.1989) an incredible amount of information has been published about the functional properties of the CFTR channel and its physiological regulation. More recently CFTR has been studied as a major drug target to treat cystic fibrosis (CF) and secretory diarrhoeas (Verkman and Galietta, 2009; Becq, 2010). From these studies we know that CFTR is a large (~170kDa) multidomain protein belonging to the ATP binding cassette (ABC) family of membrane transporters (Gadsby et al. 2006; Mendoza & Thomas, 2007). It is composed of two repeating units, each consisting of a membrane spanning domain (MSD) comprising of six alpha helices, followed by a nucleotide binding domain (NBD). The two units are joined together by a unique R or regulatory domain, which contains multiple PKA and PKC phosphorylation sites and which interacts physically with NBD1 (Gadsby et al., 2006; Baker et al., 2007; Hwang & Sheppard, 2009; Figure 1). Uniquely, and unlike other members of this large ABC transporter family, as well as any other known ion channel, CFTR uses the energy of ATP binding and hydrolysis to drive ligand-induced conformational changes in the protein that lead to the regulated opening and closing (gating) of the channel ‘pore’. In this way, CFTR gating tightly ‘regulates’ the flow of chloride ions in and out of the cell, with the anion being transported down its prevailing electrochemical gradient by a purely facilitative diffusional process. Ion channels are extremely efficient and when open CFTR conducts millions of chloride ions per second. However, there is still no universal consensus on the precise kinetic scheme for CFTR gating (Alexsandrov et al., 2009; Csanady et al., 2009; Tsai et al., 2010), nor an exact picture of the structural changes that occur in the protein during channel opening and closing, despite intense research in the field (Zhang et al, 2009). Nonetheless, biochemical, electrophysiological, structural and molecular modelling studies (Serohijos et al 2008; Mornon et al., 2009) support a scheme whereby following PKA phosphorylation of the R domain, CFTR gating is induced by ATP binding and dimerisation of its two NBDs (Mense, et al., 2006). This event is transmitted to several transmembrane segments (TMS) that form the ‘pore’ of the channel through a complex and dynamic interaction between their cytoplasmic linker loops and the dimerised NBDs (He et al., 2008). ATP (ligand) binding therefore drives the structural rearrangement of the TMS that result in the opening of the channel pore (Figure 1). CFTR is also an ATPase (Cheung et al., 2008); hydrolysis of ATP and the release of ADP and Pi, disrupt the NBD dimer thereby terminating channel opening. Further rounds of channel opening can continue if ATP again binds but the gating cycle can be stopped by the removal of phosphate groups from the R domain by intracellular phosphatases (Gadsby et al., 2006; Hwang & Sheppard, 2009). It is apparent then that CFTR gating is a complex process involving a variety of intracellular factors that regulate multiple inter- and intra-domain domain interactions within the protein. The new results of Zhang et al (2010) now show that the normal cellular machinery that controls CFTR activity can be completely bypassed simply by changes in membrane tension induced by direct application of negative pressure to the cell surface. On the face of it this is quite an astonishing result that potentially moves the whole field of CFTR gating into a new era.
The authors first showed that CFTR is mechanosensitive by studying the behaviour of the channel in a living cell to a short burst of negative pressure (suction). This was delivered through the back of a glass micropipette (patch pipette) that was sealed onto the surface of a CFTR-expressing epithelial cell. The patch pipette also enabled CFTR channel activity to be measured electrically. When 15 mmHg negative pressure was applied to the micropipette CFTR-like anion channel activity was activated in ~ 60% of membrane patches. This success rate exceeded that normally observed in membrane patches exposed to a cAMP agonist alone under normal (atmospheric) pressures. The effect of suction was not prevented by protein kinase inhibitors and the activated channels were blocked by CFTR-specific inhibitors. Mechanosensitivity was also confirmed in cell-free, excised, membrane patches in the complete absence of ATP, showing that neither an intact cell nor ATP was required for the effect. However, membrane tension was markedly less potent at activating CFTR in these cell-free experiments, which suggests that an intracellular factor or the underlying cytoskeleton contributes to mechanosensing. This is interesting because the C-terminus of CFTR contains a canonical PDZ- binding motif which can be used to link CFTR to a number of scaffold proteins to generate large macromolecular protein complexes (Li & Naren, 2010). In particular, CFTR binds to NHERF1 and NHERF 2, PDZ- binding proteins that can also attach to actin and thereby link CFTR to the cytoskeleton. It is conceivable therefore that a change in membrane tension could cause structural re-arrangement of the underlying cytoskeleton, which is then relayed to CFTR via these scaffold proteins, to cause channel opening. Future studies that investigate the role of the cytoskeleton will be very important to address this issue.
Studies were also performed on intact layers of epithelial cells which represent a more physiological situation. Using a novel approach, whole intact sheets of cells were exposed to suction to increase membrane tension, and anion secretion rate measured as the short circuit current. It should be remembered that it is net rate of transepithelial anion secretion that CFTR normally controls in epithelial tissues. Results from cultured airway cells grown as monolayers or intact intestinal tissue derived from normal and CFTR knock-out mice showed that membrane stretch increased anion secretion in a CFTR-dependent manner, consistent with the patch pipette studies. Moreover, and unlike other mechanosensitive channels (Hamill, 2006), the response to an increase in negative pressure was sustained for many minutes, and was fully reversible.
These exciting findings of Zhang et al (2010) generate many new questions that require answers: (1) What structural changes, and importantly which domains of CFTR, underlie the gating transition from the closed (nonconducting) channel to the open (conducting) state in response to changes in membrane tension (Figure 1)? In the E. Coli mechanosensitive channel, MscS, changes in the density of lipids in the inner leaflet of the membrane are suggested to decrease upon a rise in outward turgor pressure. This change in lipid environment is thought to cause rotation and tilting of MscS transmembrane segments which drives separation of pore lining helices sufficient to cause opening of the channel (Wang et al., 2008; Anishkin & Sukharev, 2009). In this way lipid-protein interactions lead to channel opening. It will be very illuminating to investigate if a similar mechanism of activation underlies the effect of tension on CFTR. Rather intriguingly, and before publication of the Zhang paper, Kevin Kirk’s group proposed a simple ‘spring model’ that may explain how gating signals from CFTR’s cytoplasmic domains are transmitted to the channel pore, based on results from cytoplasmic loop mutants (Wang et al., 2010). In this scheme the cytoplasmic loops (which link the NBDs and transmembrane helices) are modelled as compression springs which act to prevent (resist) spontaneous channel openings in the absence of ATP binding. They suggest that ATP binding to the NBDs ‘allosterically favours channel opening by compressing the spring’. In this context and taking the ‘spring model’ analogy further, could changes in membrane tension lead to channel gating via alterations in the resistance of the ‘spring’ hypothesised by Wang et al? This could be tested by studying the response of these cytoplasmic loop mutants to stretch. (2) Since the effect of membrane tension and cAMP/PKA were not additive, which of the two mechanisms is the most important physiologically? As noted by the authors CFTR does operate in tissues that experience normal periods of membrane stretch such as the airways. Perhaps during the breathing cycle, and especially during exercise, dynamic changes in membrane stretch may augment cAMP/PKA stimulation (or vice versa) to regulate CFTR activity and lung function. Furthermore, the observation that CFTR is expressed in different kinds of muscle (Michoud et al., 2009; Sellers et al., 2010) surely warrants further investigation of whether (and how) its mechanosensitive properties potentially contribute to muscle function. Notably, CFTR appears to be important in skeletal muscle metabolism during exercise (Lamhonwah et al., 2010), hinting that stretch-induced changes in CFTR activity could be physiologically relevant. (3) Is the fact that tension activated CFTR channels had ~ 40 % greater conductance than cAMP/PKA activated channels physiologically important? This finding means that even at maximal cAMP stimulation total ion flow through CFTR could be further enhanced by an increase in tension, which could be physiologically beneficial. It should also be remembered that several CF-causing mutations specifically reduce channel conductance through an unexplained mechanism (Sheppard et al., 1996). A better understanding of the structural changes in CFTR caused by membrane stretch may help to shed new light on how some CFTR mutations affect ion flow through CFTR. (4) Can membrane stretch activate CF gating mutations that are refractory to normal intracellular cues, such as G551D (Drumm et al., 1991; Bompadre et al., 2007)? If such gating mutants are shown to be mechanosensitive then it may be possible to design small molecule drugs that mimic the activating effect of tension. Such ‘mechanopharmaceutical’ drugs have already been described for some cation-selective mechanosensitive channels, (Hamill, 2006). Moreover, since switching on CFTR by cAMP/PKA can also change the activity of other ‘CFTR-dependent’ transporters, such as SLC26 Cl-/ HCO3− exchangers (Gray, 2004; Ko et al., 2004; Rakonczay, et al. 2008), then it will be vital to investigate if membrane tension alone has any effect on the function of these other proteins. (5) Is the channel still mechanosensitive after purification and reconstitution into an artificial membrane? This ultimately will be the final test to demonstrate that the CFTR channel is intrinsically mechanosensitive, as previously demonstrated for bacterial channels (Martinac et al., 1990).
The assertion that CFTR is a mechanosensitive ion channel certainly will stimulate much debate and head-scratching in the ion channel field. Why has it taken this long for someone to demonstrate this? Many electrophysiologists will be looking back at their old ion channel ‘traces’ to see if they really missed this. However, if it does turn out to be ‘fact’ after all, then the work by Zhang et al (2010) will surely lead to many new avenues of research in this topical area. Since CFTR is a member of the ABC transporter superfamily and shares many structural features with other members, such as the multi-drug resistant transporter P-glycoprotein, it may turn out that other ABC proteins are also mechanoresponsive, which opens up some very exciting possibilities. Ultimately, and very importantly, a better understanding of how the CFTR channel is gated is likely to lead to the development of new therapies for the treatment of CF and secretory diarrhoeas, a goal well worth pursuing.
My thanks to Peying Fong (Kansas State University, USA) for helpful suggestions and critical reading of the manuscript.