Therapeutics

CHANNELLING DRUG DISCOVERY current trends in drug discovery research

Ion channels are proteins that span membranes thus forming conduits or ‘channels’ through which charged ions such as sodium and potassium can pass across a normally impermeant barrier such as the plasmalemma. By so doing, ion channels can mediate a wide variety of physiological functions from the generation of action potentials in nerve cells to immune cell function and more.

By Dr David Owen lthough the pharmaceutical industry has channels are fundamental in controlling the heart and Andrew been slow to recognise the true potential of beat, sensory transduction including pain and Silverthorne Aion channels due to a combination of disbe- brain function. In non-excitable cells, ion channels lief that ion channel dysfunction could cause disease are involved in hormonal secretion, immune cell and a lack of screening technology to keep up with responsivity, cell-cycling, ion distribution and other HTS assays, ion channels are currently very more. Why is it then that only now at the begin- much in vogue. The discovery of ‘Channelopathies’ ning of the 21st century that ion channels are sud- and exciting emerging ion channel screening tech- denly creating so much interest in the pharmaceu- nologies herald a new era of intensive ion channel- tical industry? Here we review the current trends in based drug discovery. Here we review the current the ion channel drug discovery business and how a trends in the ion channel drug discovery business. convergence of research and technology develop- ments may provide the answer. Why should drug companies be Ion channels are a super-family of proteins that interested in ion channels? span cell membranes and form conduits or chan- We have known about ionic currents in nerve cells nels through which charged ions such as sodium since 19521 and been able to electrically visualise and potassium can pass across what is normally an single ion channels in real time using the patch- impermeant barrier. We tend to think of the plas- clamp technique since 19812. The first sodium ion malemma as being home to most ion channels but channel was cloned in 1984 (Noda et al) and the intracellular organelles such as mitochondria and first potassium channel in 19873. Since then we endoplasmic reticulum also have ion channels. have realised that ion channels are physiologically Broadly speaking, ion channels can be divided into important in a huge variety of functions and in all those controlled by receptors and those opened cells. In excitable cells such as nerve and muscle, and closed (gated) by changes in the voltage of the ion channels generate and shape electrical signals cell. By their very nature voltage-gated ion chan- leading to action potential propagation, neuro- nels are particularly attractive targets, but also transmitter release and muscle contraction. Ion present a challenge for the industry particularly in

48 Drug Discovery World Spring 2002 Therapeutics

finding appropriate screening technology. This review will focus primarily on these so-called volt- AB age-gated ion channels. In 1984 in the first edition of his seminal book on Activation gate the ‘Ionic Channels of Excitable Membranes’, Hille4 recognised in his preface that ion channels went well Outside beyond nerve cells and were likely to be important in non-excitable cells such as ‘sperm, white blood cells and endocrine glands’. He also predicted that our Plasmalemma genome would probably code for more than 50 dif- ferent types of ion channel. Bearing in mind that the Inside first ion channel had yet to be cloned (it was later Voltage Binding sites for that same year), this was a bold statement to make. sensor small molecules He needn’t have worried though, 18 years later the K+ Inactivation gate Human Genome Project predicted that there are more than 300 human genes encoding ion channels5. Voltage-gated ion channels are turned-on and off (or gated) by movements of so-called ‘gates’ creat- ed by elements of its own protein structure. In chan- without doubt. Add to the variability of ion channel Figure 1 nels such as K+ channels, a change in the voltage of structure afforded by this heteromeric association A cartoon representation of a K+ channel showing activation the cell causes a segment of the channel (known as phenomenon, variations on the basic subunit caused and inactivation ‘gates’ and the voltage sensor) to move within the membrane by ‘splice variation’ and it easy to imagine that the possible binding sites for small thereby opening a channel through which K+ ions number of 300 could easily be increased by a factor molecules and water molecules can pass one at a time. Other of two or more. Overlayed on this variety are more B Many K+ channels are parts of the protein which make up the lining of the or less specific tissue distributions of ion channel composed of 4 subunits that together form the functional channel determine which ions can or cannot pass expression as for traditional targets such as neuro- channel through the channel and also act as receptors for transmitter receptors. Of course some ion channels small molecules and toxins which can modulate are more ubiquitous than others, but it is clear that these functions in various ways (Figure 1A). A func- there is real potential for selective modulation of ion tional voltage-gated K+ channel is composed of at channels both between tissues and within cell-types, least four subunits represented as cylinders, which an important consideration in any drug discovery assemble as a complex in the membrane (Figure context. At this point in time, we know that K+ cur- 1B). Na+ and Ca2+ channels have analogous fea- rents (that is currents carried by the flux of potassi- tures although the four subunits found in K+ chan- um ions across a membrane) can be generated by nels are contained within a single protein. one or more of around 70 different potassium-selec- tive -subunits. Na+ currents arise from around 10 Ion channel cloning different genes; there are around nine voltage-gated Ever since 1984 when the first was and another seven non-transmitter operated chlo- cloned, cloning of new ion channels has gathered ride channels and 13 voltage-gated calcium channel pace, culminating perhaps in completion of The -subunits. Other channel types have significant Human Genome Project. The combined effort of the numbers of family members as well. For example, HGP and Celera parallel project indicates that we there are around 20 TRP channels, 12 Deg/ENaC can expect around 300 ion channel genes divided channels, 13 connexins and so on. between the major ion channel families. We also See also: www.gene.ucl.ac.uk/nomenclature/ know that for K+ channels, for example, which are genefamily/KCN.html,www.gene.ucl.ac.uk/ composed of tetramers of a basic pore-forming sub- nomenclature/genefamily/CACN.html and 6. unit (-subunit), it is also possible to get functional Although traditionally one thinks of sodium and channels from permutations of -subunits. In addi- potassium channels and the generation of action tion, many pore-forming subunits associate with potentials in nerve cells, in fact all cells (as far as we auxiliary subunits which, while not necessarily pore- know) have ion channels of some type or other. The forming in themselves, can modify the properties of bewildering number of different types of ion chan- the ion channel, either biophysically (eg speed-up nels identified at a molecular level suggests that ion inactivation) or pharmacologically (eg increase sen- channels are important in a similarly large number sitivity to drugs). While many of these -subunits of physiological processes. Sure enough, ion chan- are known, many more remain to be discovered nel involvement ranges from action potential

Drug Discovery World Spring 2002 49 Therapeutics

threshold in nerve and heart to the resting potential their propensity to block cardiac ion channels such of immune cells and sperm motility. The richness of as hERG and the potential for drug-induced QT this pool of possible therapeutic targets is surely prolongation prior to first use in humans. If there irresistible. See also: www.neuro.wustl.edu/neuro- wasn’t a demand for ion channel screening resource muscular/mother/chan.html7,8. before aLQT there certainly was afterwards. This has had a number of important consequences: Relevance of ion channels to disease The profile and importance of the ion channel as (channelopathies) a target for drugs has been raised dramatically. For many years, ion channel modulators were seen Almost everyone in the R&D hierarchy of a drug as palliative at best. However, since the first ‘chan- company now knows about at least one voltage- nelopathy’ was identified in the cystic fibrosis activated ion channel: hERG. transmembrane regulator protein (CFTR) by Since the CPMP note there has been a rapidly Riordan et al in 19899, this has turned into a escalating requirement for hERG and other ion growth industry. There are around 30 chan- channel screens. In some cases these have been pro- nelopathies to date including major therapeutic vided in house but in many companies these assays areas such as diabetes, cardiac disease, deafness, are out-sourced to other specialist organisations. blindness and epilepsy. All can be caused by ion The realisation that activity at ion channels like channels that malfunction or are not expressed at hERG are best eliminated early in the drug discov- all. See also 10 and www.neuro.wustl.edu/neuro- ery process has heightened the need for screening muscular/mother/chan.html. techniques with appropriate information content Without doubt these links have reinforced the and throughput that will make this possible. The industry’s interest in ion channels and in some cases issues involved in screening complicated ion chan- provided the key rationale for drug discovery pro- nels such as hERG are now much better appreciat- grammes (cystic fibrosis being the prime example). ed across the pharmaceutical industry. As a result of the above, the specialist ion chan- Ion channel safety pharmacology nel screeners such as Channelwork hERG: bête noir or best thing since sliced bread? (www.cenes.com/channelwork), GENION By the 90s, although represented by programmes in (www.for-genion.com) and Chantest most of the big Pharma, ion channels were still not (www.chantest.com) enjoyed a booming business mainstream. However, that was about to change. in LQT-related screening while the Big Pharma Between 1995 and 1996 genetic linkages were estab- worked out their strategies. The established CROs lished between a family of inherited cardiac disorders have not been slow to catch on either. Now, MDS- known as LQT and inherited mutations in the cardiac Pharma (www.mdsps.com), Quintiles (www.quin- voltage-gated potassium channels, KvLQT1 and tiles.com) and Cerep (www.cerep.fr) all provide hERG and the voltage-gated cardiac sodium channel, hERG screening as well as the isolated cardiac SCN5A11-13. Furthermore a link was suggested Purkinje fibre assay, recommended in the original between drug induced, or acquired LQT (aLQT), and CPMP advisory note and now being cemented in block of the hERG potassium channel14. Next, an the ICH7A document (www.ifpma.org/ich1.html). advisory note issued by the Committee on Proprietary Furthermore, if any further encouragement was Medicinal Products (CPMP)15 sparked an enduring needed, the expanding need for ‘ion channel safety debate concerning the possible mechanisms by which screening’ has fuelled the need to develop automated drugs such as cisapride and terfenadine were appar- and high-throughput technologies for screening ently causing sudden death by cardiac arrest. The those channels that appear in the list of undesirables. CPMP note suggested that it would be desirable to conduct a preclinical ‘diagnostic’ test of candidate Ion channels as pharmaceutical targets drugs, prior to first use in humans, not only for their Existing ion channel drugs propensity to prolong the cardiac action potential Currently, potential for ion channel-directed drugs (which can lead to dysrhythmia) but also for their is relatively untapped. Only about 5% of the tar- potential to block cardiac ion channels. Meanwhile, gets of marketed drugs are classed as ion chan- the finger of blame was being pointed most emphati- nels17. Nevertheless, existing drugs that modulate cally at the hERG potassium channel16. ion channels already represent a valuable class of Arguably, hERG has done more to promote ion pharmaceutical agents with a total market value in channels to the current high level of prominence excess of $8 billion in 2000. than any other single development. What has fol- The global pharmaceutical market is worth more lowed is a de facto requirement to test new drugs for than $240 billion annually with the major markets

50 Drug Discovery World Spring 2002 Therapeutics

in the US, Europe and Japan where combined sales exceed $220 billion. The therapeutic areas in which ion channel modulators are most likely to be used include the largest categories of cardiovascular Open Na+ channel Inactivated Na+ channel (with annual sales of $48 billion) and CNS (with annual sales of $40 billion). In addition to these Activation gate Na+ Activation gate Na+ areas, ion channel modulators have application in a wide range of other high value areas such as pain (neuropathic pain is estimated to be worth approx- Outside imately $550 million per year)30. Pharma Projects lists around 100 launched ion Plasmalemma channel modulators. The majority of these are Ca2+ Inside channel blockers (60) aimed at cardiovascular sec- Binding sites for tor. Around 26 Na+ channel blockers (analgesics Binding sites for small molecules small molecules and anticonvulsants) and eight K+ channel modula- Inactivation gate New binding site tors (inter alia vasodialators such as minoxidil). To created by closed date no Cl- channel modulators have been launched. inactivation gate Some of these are very well known and include the Action anticonvulsant, lamotrigine (Na+ channel blocker); potential dihydropyridine antihypertensives such as nifedipine firing (ProcardiaTM), an L-type Ca2+ channel blocker; the diabetes drug, glyburide (DiabetaTM), which is an ATP-sensitive K+ channel blocker and local anaes- thetics such as lidocaine (XylocaineTM) which is also a Na+ channel blocker. ion channels after they have opened and inactivated, Figure 2 Many of these have been discovered serendipi- thus not interfering with basal activity (but reducing Smart drugs.A blocker targeted at a binding site tously and before the real explosion in ion channel frequency of activation – smart drugs (Figure 2). created in the ‘inactivated’ cloning, molecular genetics and screening technol- state of the channel will not ogy development that has taken place over the last Current ion channel drug discovery programmes prevent opening of the channel few years. Deliberate targeting of specific ion chan- Pharma Projects lists around 29 active Na+ chan- per se, but could delay nel sub-types in drug discovery programmes prom- nel projects, 28 active K+ channel programmes, recovery from inactivation which normally occurs ise more and better ion channel drugs. three active Cl- programmes and 19 Ca2+ pro- between action potentials.This grammes. This appears to show a trend away from results in fewer action New ion channel drugs: smart drugs? Ca2+ to Na+ (consistent with the launch of a large potentials being generated per What makes ion channels attractive targets for number of Ca2+-based cardiovascular-based drugs) unit time drug development? There are a number of features and a sustained emphasis on K+ channel modula- of ion channels and especially voltage-gated ion tors. It also appears to represent a significant effort channels that makes them attractive targets for by the industry as a whole. For comparison, novel drugs: Pharma Projects also lists 20 SRI, 20 tyosine As illustrated above, ion channels are fundamen- kinase, nine AchE-I and five statin programmes, tal elements in physiology playing an important role among many others of course. in all cells and across a wide range of organs and As well as exploring new ion channel families, functions from nerve cells to the immune system. we can probably expect to see companies revisiting Ion channels are under exploited and thus pro- areas such Ca2+ modulators in the light of new vide huge potential for novel drug targets. information (eg cloned sub-types) and cardiac ion The huge variety of subtypes of ion channels channels armed with additional information from and differential distribution within and between molecular genetics and the molecular mechanisms tissues supports the idea that selective drugs can underlying acquired LQT (discussed above). be developed. The inherent gating mechanisms of ion channels Patent activity offers the chance to develop state-dependent drug Patent activity is often a guide to the commercial actions and dynamic drug therapy which responds interests of the pharmaceutical industry being as it and is dependent on activity of ion channels and is, highly dependent on intellectual property for behaviour of cells and tissues. For example, ion recouping the massive investment that goes into channel blockers can be engineered only to block drug discovery and development. Figure 3 shows

Drug Discovery World Spring 2002 53 Therapeutics

the growth in granted US Patents from 1980 to the maceutical industry, it has effectively been written end of 2001. Note that the industry did not wait off as a screening platform because it is a manual for ion channels to be cloned eg ROMK1, the first and laborious procedure and hence slow (~100 K+ channel to be Patented, issued in 1994. Indeed dp/wk). Instead a good deal of effort has been much of the early activity reflected development of made in developing alternative methods of moni- Ca2+ channel blockers in the dihydropyridine fam- toring ion channel activity that can be integrated ily using binding assays and functional cardiovas- into industry-standard compound screening for- cular screens and it was not until the late 90s that mats with corresponding high throughput. Ca2+ channel sequences were granted Patents. However, in the K+ channel area, there has been an High throughout ion channel screening acceleration in granted Patents since the ROMK1 Fluorescence Patent issued and it does seem logical that molecu- The principle methods in ion channel programmes lar cloning and Patenting therein should encourage today are: receptor binding assays, flux measure- pharmaceutical activity. Not only is the trend ments and fluorescence detection techniques. The upward for Ca2+ and K+, Na+ and Cl--related principle advantage of these approaches is their Patents also show an upward trend. Although drug medium to high throughput (15-60K dp/wk) albeit companies have been relatively slow to enter the at the expense of information content. In particu- ion channel arena, there is a growing recognition lar ligand displacement assays say nothing about of the potential value in the sector. In part this functional activity of unknown compounds and seems to be driven by molecular cloning but it is and are unlikely to detect novel types of modula- also clear that a significant reason for this reticence tors by definition. Of the functional assays devel- has been the lack of suitable screening technology. oped, fluorescence has been seen as the most cut- However, this is about to change as can be seen in ting edge. This methodology exploits changes in the Figure and we can expect to see the number of fluorescence that occur either with changes in the ion channel-related filings accelerating. concentration of ions or changes in the membrane potential of the cell. For more details see Denyer et Screening technology: the key to al18 and Xu et al19. Since FLIPRTM (fluorometric Figure 3 success imaging plate reader) was developed by Noveltech Ion channel-related Patents issued in the US since 1980. A number of developments in R&D over the last (now manufactured by Molecular Devices Inc) for Source: www.delphion.com. + 10 years have converged. A major constraint on Upjohn in the USA to screen the ROMK1 K chan- Patents which mention developing new ion channel-based drugs has been nel, the art has been refined considerably. The voltage-gated ion channels in the difficulty in screening ion channels (voltage- VIPRTM (voltage ion probe reader) developed by the title or abstract were gated) at the throughputs required of the modern Aurora Biosciences to exploit its proprietary FRET selected.The cumulative number, which includes industry in a cost-effective way and with function- dye systems has improved both time-resolution composition of matter as well ally relevant screens. Effective exploitation of the and sensitivity of this approach to the best yet. as sequence, assay and use ion channel arena has been awaiting the develop- However, in spite of the undoubted significance of Patents, is plotted on the ment of new technology. the FLIPRTM in bringing the first multi-well (now y-axis against year (x-axis)

Patch-clamp: the gold standard Undoubtedly the definitive method for studying ion channel function is that of patch-clamping. In 250 the form that most practitioners know it, patch- clamping was developed in the late-70s and the 200 definitive text on the subject published in 1981 by Hamill et al2. The technique has gone practically 150 unaltered since then. As far as biophysically and Misc Chloride 100 pharmacologically-characterising ion channels, it Sodium has remained the gold standard. The method can Potassium 50 detect signals in the pA range and even measure the Calcium current passing through a single ion channel pro- 0 tein in real time. The time-resolution is in the tens Calcium

1980 1982 1984 Misc of microseconds range and, crucially, patch-clamps 1986 1988 1990 1992 1994 1996 allow the experimenter to fix the membrane poten- 1998 2000

2002 tial of the cell (voltage-clamp). However, with respect to the practical considerations of the phar-

Drug Discovery World Spring 2002 55 Therapeutics

Rethinking ion channel screening The trend in the pharmaceutical industry’s priori- ties has been to emphasise the screening of as many Throughput compounds as possible per unit time, often sacri- Cost per datapoint ficing informational content and physiological rel- Current ‘Smart drug’ evance of the screen in the process (Figure 4). priorities Physiological relevance priorities This is the mentality that has driven the FLIPR and Information content VIPR approaches to ion channel screening. However, there has been a consistent clamour from the scien- tists running ion channel drug discovery programmes for a viable electrophysiological screens to support, and in some cases replace, the state of the art repre- sented by FLIPR and VIPR. Some of the reasons for Figure 4 384-well based) ion channels screen to drug com- the dissatisfaction with fluorescence and other HTS Current trends emphasising panies, FLIPRTM is largely used for Ca2+ assays are as summarised below. quantity not quality could be flux/mobilisation measurements and has not reversed with development of data-rich screening technology proved to be a generic ion channel screening tool. Limitations of current HTS ion channel such as HTS patch-clamp While VIPR IITM (also 384-well based) clearly has techniques: superior time-resolution and suffers less from Many channels generate too small a signal or are quenching artifacts than the non-ratiometric too transient to record. For example the currents FLIPRTM, in common with FLIPRTM and other generated by the Nav1.3 Na+ channel ‘inactivate’ fluorescence methods it still suffers from relatively within 20ms following activation with a depolaris- high false positive and negative rates of up to ~5%. ing (activating) voltage step. HTS fluorescence methods can resolve down to the 1-10sec time Flux assays frame at best. Use of radioactive surrogate channel ions was a None of the non- methods are suited popular early solution to side-step the need for to controlling the voltage of the host cell and as such electrophysiological assays such as patch-clamp. cannot properly control the gating of the ion chan- Although arguably a more relevant measure than nel in question. This precludes precise targeting of the indirect measure of voltage detected by most modulators to specific states of the ion channel fluorescence assays, flux assays have lost out in which not only limits the scope for smart drugs that time resolution and the negative association with interact dynamically with ion channels but may radioactivity. However, flux is back in vogue with result in significant false hit rates. An example of a the development of atomic absorption methods target class that may require ‘state-aware’ screening (pioneered by Bayer) which allow use of non- is that of voltage-gated Na+ channels where is it radioactive surrogates such as Rb+ without the known that existing clinically-used blockers prefer- radioactivity ‘headache’. Commercial systems are entially block the inactivated form of the channel available from Thermo-Elemental UK (Cambridge) rather than either closed or open states. and Aurora Biomed (Canada) and this approach In any case, increased throughputs at the pri- has recently been launched as a commercial service mary screening level has lead to a bottleneck at the in a Quintiles plc (Scotland)/BioFocus validation of hits and lead optimisation stages that plc(Cambridge) joint initiative. Currently, the tech- do require techniques. nique is optimised for K+ channel screening using Figure 5 illustrates how automated patch-clamp Rb but it may also be possible to adapt the tech- systems could alleviate the bottleneck that is creat- nique to read other surrogate ions for other types ed where current HTS screens can be applied due of voltage-gated ion channels. As with fluores- the slowness of conventional patch-clamp used to cence, there is no ability to control the membrane validate hits and in lead optimisation. potential of the host cell and as such is suited to Furthermore, where fluorescence and flux assays primary screening in combination with hit valida- are inappropriate, HTS patch-clamp will be able to tion and lead optimisation using medium to high- replace existing approaches, perhaps in conjunc- throughout patch-clamp techniques. Throughputs tion with smaller focused libraries following a are claimed to be of the order of 30,000 trend exemplified by Arqule in the USA20. High- cpds/week8. Without adequate patch-clamp back- throughput patch-clamp systems which retain high up use of the technique, as with fluorescence, information content will dramatically increase the should be treated with caution. quantity and quality of the data fed back into

56 Drug Discovery World Spring 2002 Therapeutics

medicinal chemistry in the lead optimisation Interface-PatchTM Clamp technique which was process which should be reflected in better drugs. launched in 2000 in the AP1 AutoPatchTM. Possibly encouraged by these developments there is Automated electrophysiology a renewed vigour attached to the quest for patch- The result is that in spite of the great technical clamp devices with high throughput. It is illustra- advances made in fluorescence in particular, there tive of the significant progress being made in this has been a continuing demand for genuine electro- field that, in contrast to the 1998 review18, a com- physiological methods that can be automated and parable review published in 200119 cites no fewer Figure 5 hence accelerated. As recently as 1998, a review of than 10 alternative and competing approaches to Flow chart illustrating current HTS approaches to ion channel screening relegated automating patch-clamp with the aim of developing bottleneck in current ion patch-clamping to a low-throughout methodology HTS patch systems. Other electrophysiological channel screening and future screening options using and not viable for HTS, reflecting perhaps a resig- approaches are also being developed (see below). emerging automated and HTS nation in the drug industry that this was not to be. patch-clamp technologies. HTS Ingenious but limited approaches such as the Automated patch clamp systems patch-clamp should at least Cytostar-T scintillating microplate (Amersham) AutoPatch (CeNeS) keep up with a FLIPR-based were still being advanced. But things have changed. AP1: The first generation ‘interface patch-clamp’, screen (1% hit rate from a diverse compound library)- Patch-clamping is on the agenda again. Two com- this system revolutionised patch-clamping by patch- currently around 60K dp/wk panies have developed automated patch-clamp sys- ing on to cells held at a liquid-air interface. Reduced max, (ie ≥ 600 dp/wk). HTS tems, Sophion Biosciences (Ballerup, Denmark) has to a one-dimensional process without requiring patch-clamp promises robotised a traditional patch-clamp workstation in optics, the whole process is automated from GΩ-seal throughputs of up 30K dp/wk, its Apatchi and CeNeS Channelwork (Cambridge) formation through to whole cell recording and drug providing an alternative to current FLIPR-like primary 21,24 has invented a novel form of patch-clamping (the application . Although a single-cycle device, mul- screens, particularly in first departure from Hamill et al’s protocol pub- tiple machines can be operated by unskilled opera- conjunction with smaller lished 20 years ago) which utilises the so-called tors thereby scaling up throughout with reduced focused libraries

Ion Channel Focused Diverse Library Diverse Library Library (150K+) (150K+) (10-15K)

HTS screen VIPR/FLIPR/Flux Med Chem HTS screen 20-60K dp/wk & Parallel VIPR/FLIPR/Flux Chemistry 20-60K dp/wk

HTS Patch-Clamp Med Chem Parallel planar or & Parallel ‘interface patchTM’ Hit Validation Chemistry Conventional ~30K dp/wk Patch-Clamp 100 dp/wk Assay Development AP1 AutoPatch SAR Hit Validation & SAR 200-300 dp/wk Conventional Multi-cycle patch: Patch-Clamp AP2 AutoPatch (2-3K dp/wk) 100 dp/wk Apatchi (150-500 dp/wk)

Drug Discovery World Spring 2002 57 Therapeutics

Figure 6 Graphical comparison between ‘interface patch’ and ‘planar patch’ approaches Solution

Liquid-air interface Cell Planar electrode (chip) Glass patch-pipette

INTERFACE-PATCH PLANAR PATCH

human recourse requirements. Throughput is 200- be integrated into such devices. The challenge 300 dp/wk. Five AP1s are installed at GSK in the UK with parallel systems will be to retain as much of and USA and two systems at Wyeth in the USA. the power of conventional patch-clamp and AP2: A sequential recording device derived from automated systems such as the AP1 AP1 but which can automatically switch between AutoPatchTM system with the required through- up to 48 recording sites in a ‘patch plate’ without out at processing at an affordable price and in a reloading cells or pipettes. The system is integrated practical format. with the AP1’s 96-well plate based drug delivery Many design issues have to be addressed which system providing dramatically increased capacity include: and at least an order of magnitude higher through- How to achieve parallel patching from a signifi- put. In late development, the system has been pre- cant number of active sites without sacrificing sam- ordered by Wyeth in the US. Throughput is expect- pling frequency and voltage-clamp integrity. Overall ed to be ~2-3K dp/wk. throughputs of ~30K dp/wk should be sufficient. Individual control of recording sites from GΩ- Apatchi-1TM (Sophion) seal formation through to recording to maximise Originally developed at NeuroSearch (Denmark) recording hit rates and longevity of individual in conjunction with Pfizer, the Apatchi-1 is a robo- recordings (≥ 50% hit rate). tised patch-clamp workstation which utilises Facility for single or multiple drug solution motorised manipulator systems and sophisticated applications to individual recording sites. image recognition software to automatically place Maximise stability of recordings. patch pipettes on cells and thereby establish patch- Minimise cross-talk and distributed capacitance clamp recordings. The system also incorporates a between and across sites. fluid handling system and carrousel of eight patch How to handle the potentially vast amount of pipettes which permits up to eight recordings in data acquired. sequence without reloading by an operator. The How to fully-automate the process into a robust system is extremely reliant on the precision of its and deskilled process at an affordable price (1$ per hardware and software to avoid pipette crashes as data point?). the cells are presented in dishes in the traditional Broadly-speaking there are two approaches to fashion. Pfizer has one Apatchi system installed in the electrode design (Figure 6) currently being Sandwich, UK. explored by different groups:- Interface-PatchTM which uses glass patch Parallel patch-clamp systems pipettes as in conventional patch-clamp and the The quantum leap in ion channel screening AutoPatchTM. throughout will come from parallel processing Planar electrodes (‘chips’). In the planar elec- systems. As yet, no automated parallel patch sys- trode a GΩ-seal is formed between cell and a pore tems have been developed although a number of formed by microfabrication techniques in a chip companies and research groups are working on made of or other materials. novel recording electrodes, electronic amplifiers A number of companies and academic groups and fluid handling systems that are expected to are working toward parallel patch-clamp devices

58 Drug Discovery World Spring 2002 Therapeutics

References Table 1 1 Hodgin et al. Measurements of current-voltage relations in Main players in the development of HTS patch-clamp the membrane of the giant technology platforms of Loligo. Journal of Physiology, 1952. 116: p. 424- 448. COMPANY ELECTRODE DESIGN STATUS 2 Hamill, O et al. Improved patch-clamp techniques for high- Aviva Biosciences Silicon; field potential cell Single pore chip recordings from resolution current recording San Diego, USA positioning & suction control mammalian cells; overall whole cell from cells and cell-free recording success rate ~15-20%; membrane patches. Pfluger’s not automated22 Archives, 1981. 391: p. 85-100. 3 Tempel,BL et al. Sequence of Axon Instruments Sylgard cast chip; suction control Whole cell recordings established a probable potassium channel Union City, CA, USA with mammalian cells on single component encoded at Shaker pore chips; incidence of GΩ-seals locus of Drosophila. Science, ~50%; not automated23 1987. 237: p. 770-5. 4 Hille, B. Ionic Channels of Excitable Membranes. 1992, CeNeS Channelwork Glass patch pipette; suction control Interface Patch validated in Sunderland, Massachusetts: Cambridge, UK automated AP1 & AP2 devices; Sinauer Associates Inc. whole cell recording rate ~50-60% 5 Venter, C et al.The sequence in mammalian cells; parallel version of the human genome. Science, (AP3) in development; control 2001. 291: p. 1304-1351. software already developed for 6 Goldin,AL. Resurgence of AP1 & AP221, 24 sodium channel research. Annu. Rev. Physiol., 2001. 63: p. Ω Cytion (subsidiary of Molecular Planar silicon; control of seal G -seals on single pore chips 871-894. Devices) formation by electrical field reported for vesicles; no data 7 Chandy, KG et al. in Ligand 25 Lausanne, Switzerland reported for mammalian cells and voltage-gated ion channels. ed A. North. 1995, CRC Press: Cytocentrics Glass patch pipette/silicon; suction Glass pipette-based with a centring Florida. Reutlingen, Germany control technique using porated silicon 8 Ford, JW et al. Potassium structures; claim high success channels: gene family, therapeutic rates26 relevance, high-throughput screening technologies and drug Essen Instruments Not disclosed Whole cell data reported for discovery. Prog. Drug Res., 2002. (20% owned by Molecular Devices) mammalian cells from array of 20 58: p. in press. Ann Arbor, MI, USA recording sites; parallel voltage-clamp 9 Riordan, JR et al. claimed; hit rate not disclosed; not Identification of the cystic automated; commercial device in fibrosis gene: cloning and development27 characterization of complementary DNA. Science, Nanion Technologies Planar glass; suction control Whole cell data reported on single 1989. 245(4922): p. 1066-73. Munich, Germany pore chips for mammalian cells; 10 Ashcroft,AM. Ion channels ~50% hit rate; multi-pore arrays in and disease: channelopathies. development; not automated28, 29 2000:Academic. 11 Wang, Q et al. Positional Sophion Biosciences Planar silicon; suction control No data on chip performance cloning of a novel potassium Ballerup, Denmark reported; parallel array in channel gene: KVLQT1 development mutations cause cardiac arrythmias. Nature Genetics, 1996. 12: p. 17-23. 12 Curran, ME et al.A molecular basis for cardiac and a comparison of the approaches and progress Other electrophysiological approaches arrhythmia: HERG mutations cause long QT syndrome. Cell, made by the main commercial players is sum- Impedance probes 1995. 80(5): p. 795-803. Ω TM marised in Table 1. One of the fundamental chal- CeNeS Channelwork, Cambridge, UK. (T RM ) 13 Wang, Q et al. SCN5A lenges of planar electrodes has been reliably form- The ‘transepithelial resistance measurement tech- mutations associated with an ing the GΩ-seal which is critical for successful nique’ (TΩRM)31 exploits single mammalian cells inherited cardiac arrhythmia, patch-clamp. This of course is a well-established or confluent pseudo-epithelial layers of mam- long QT syndrome. Cell, 1995. 80(5): p. 805-11. phenomenon with glass fabricated patch-electrodes malian cells expressing the ion channel of choice. It as used in ‘interface patch’ clamp. is then possible to measure changes in the overall Continued on page 60

Drug Discovery World Spring 2002 59 Therapeutics

Figure 7 Key factors that have converged to catalyse and enable a new era in ion FLIPR/VIPR technology channel-related drug discovery Ion channel physiology ‘Channelopathies’

Ion channel cloning Barrier to voltage-gated ion channel drug discovery

Ion channel safety (‘hERG’) HTS Patch-Clamp Technologies

New ion channel drugs

Continued from page 59 resistance across this layer in response to various in throughput but long-term unattended operation 14 Sanguinetti, MC et al.A conditions which can include depolarising media in provides increased capacity per working day. mechanistic link between an the presence and absence of ion channel modula- Scion Pharmaceuticals (Boston, MA, USA) inherited and an acquired cardiac TM arrhythmia: HERG encodes the tors. Although not a voltage-clamp in its basic HTEP : drug application and experimental IKr potassium channel. Cell, form, the approach can provide good time resolu- design system. This system facilitates scheduling of 1995. 81(2): p. 299-307. tion and simple assay protocols. drug application and voltage-step protocols but 15 Points to consider: Adaptive Screening, Cambridge, UK (CytoFluxTM) does not constitute an automated recording sys- Assessment of the potential Probably measures transmembrane impedance. tem, is limited to oocytes and is comparable for QT interval prolongation by non-cardiovascular Although details are scant, this UK company throughout to manually-operated recording set- medicinal products. claims to have developed a 96-array parallel pro- ups. Committee for Proprietary cessing drug screening tool based around imped- Medicinal Products, 1996. ance measurement across membranes that may be Significant differences in the background experi- CPMP/986/96. combined with fluorescence imaging and field enced by cloned ion channels heterologously 16 Vandenburg, J et al. HERG K+ channels: friend and foe. potential stimulation. expressed in frog eggs can significantly affect the TIPS, 2001. 22. sensitivity of channels to modulators. This factor, 17 Science, 2000. 287: p. 1960- Frog oocyte voltage-clamp systems combined with the requirement to inject every sin- 1964. A number of two-electrode voltage-clamp based gle egg with RNA encoding the ion channel of 18 Denyer, J et al. HTS systems, based around the use of Xenopus (frog) choice and seasonal variation in expression limit approaches to voltage-gated ion channel drug discovery. oocytes, are being developed by several companies the usefulness in drug discovery. Nevertheless, as a DDT, 1998. 3: p. 323-332. including: workhorse for novel gene expression, the oocyte 19 Xu, J et al. Ion-channel Axon Instruments (Union City, CA, USA) remains a valuable tool and these technologies will assay technologies: quo vadis? OpusXpressTM: eight-channel (egg), semi-auto- undoubtedly prove useful in this context. It is not DDT, 2001. 6: p. 1278-1287. mated recording system launched at the end of known whether any customers have taken delivery 20 Gallion, S et al.Tilting toward targets: biased 2001. Increases throughput by parallel recording. of any of the commercial systems yet. compound sets. Current Drug Abbott Labs (Abbott Park, IL, USA) Discovery, 2002. January, 2002.: In-house system: parallel eight-egg screening system Ion channel drug discovery: coming of p. 25-27. with integrated drug delivery and analysis systems age 21 Owen, DG et al.Automated with increased throughput and hands-off capability. Wholehearted enthusiasm by the pharmaceutical ion channel screening by interface-patchTM clamping. Multichannel Systems (Reutlingen, Germany) industry for ion channel drug discovery has taken TM Society for Neuroscience RoboCyte : an automated sequentially recording its time coming. Wholesale exploitation of ion Meeting, 2001. San Diego, USA: system based around a 96-well array. The system also channels in drug discovery has been held back by Abstract 272.27. automatically injects oocytes with the appropriate concerns in a number of areas as discussed above: Continued on page 61 RNA for the required ion channel target. No increase Physiological relevance of ion channels.

60 Drug Discovery World Spring 2002 Therapeutics

Therapeutic relevance of ion channels (chan- (Channelwork, GENION [now EVOTEC-OAI], Continued from page 60 nelopathies). Chantest, Zenas). 22 Xu, J et al. Micropositioning Molecular correlates of physiological ionic cur- Ion channel technology developers (AVIVA enabled patch clamp recordings on a chip. 46th Annual rents and potential for specificity of ion channel Biosciences, Axon Instruments, Channelwork, Biophysical Society Satellite modulators (cloning). Essen Instruments, Cytion [now Molecular Devices Meeting, 2002. Satellite: Drug Screening technology development (FLIPR/VIPR Inc], Nanion Technologies, Sophion Bioscience). Discovery Technology for Ion and HTS patch-clamp). Many of these companies aspire to contract serv- Channels, II. Ion channel safety screening. ices/drug discovery on the back of their technolo- 23 Mathes, C et al.Whole-cell recordings from planar patch Ironically, an undesirable attribute of ion chan- gy. However, whether any of these companies will clamp electrodes: a step toward nel modulators, namely hERG block, has catalysed succeed in making the transition from instrument high-throughput much of this progress in recent years. The final developer to drug discoverer (à la Aurora via the electrophysiology. Society for missing ‘subunit’ (HTS Patch-Clamp Technologies) VERTEX acquisition) remains to be seen. CeNeS’ Biomolecular Screening, 2001. 7th of this particular drug discovery and development Channelwork is currently being positioned to spin- Annual conference, Baltimore, MD, USA:Abstract 5039. pipeline is about to be inserted (Figure 7). Will the out into a new ‘ion channel company’ and seems 24 Trezise, D et al. Rapid ion channel modulators really start to flow now? uniquely well placed to make the transition, having electrophysiology screening of A number of the big pharma recognised the developed both contract ion channel research and ion channels using the CeNeS potential for ion channels in the late 80s and 90s commercial automated patch-clamp technology AutoPatch system. Society for but did not necessarily have the required resources. businesses in parallel. Biomolecular Screening, 2001. 7th Annual conference, Baltimore, As a result a number of niche businesses were All of which seems to indicate that we are indeed MD, USA:Abstract 5016. established in the late 90s to cater for the excess on the verge of a golden age in ion channel drug 25 Schmidt, C et al.A chip- demand. These included ChannelworkTM the con- discovery. It remains to be seen who is best pre- based biosensor for the tract research division of CeNeS Pharmaceuticals pared to reap the rewards. DDW functional analysis of single ion (Cambridge, UK), GENION (Hamburg, channels.Angew. Chem. Int. Ed. Engl., 2000. 39: p. 3137-3140. Germany), ChanTest (St Louis, USA) and Zenas 26 Stett,A et al.Automated Technologies (New Orleans, USA), all of which David Owen obtained both a degree and a PhD in patch-clamp approach specialised to a greater or lesser extent in providing Pharmacology at Bristol University and University providing high content ion channel screening on a contract basis. College London respectively. He then went on to screening. Biophysical Society ICAgen was the first company formed expressly train as an electrophysiologist at the Laboratory of Meeting, San Francisco, 2002, 2002:Abstract B341. to exploit ion channels in drug discovery and has Neurophysiology, NINCDS at National Institutes 27 Schroeder, K. High- enjoyed an almost unique niche for a number of of Health, USA where he characterised a novel cal- throughout electrophysiology years, although companies like NeuroSearch cium-activated chloride channel and GABA-R – a reality. Society for (Denmark) have also had quite an ion channel modulation in spinal and hippocampal cells. He Biomolecular Screening, 2001. emphasis. Both ICAgen and NeuroSearch have then returned to the Pharmacology department at 7th Annual conference, Baltimore, MD, USA: p. relied very heavily on traditional patch-clamping UCL working under Professor David Brown Abstract 146. to support their programmes. However, like Big before joining Wyeth in 1988 to set up the electro- 28 Fertig, N et al. Stable Pharma, while making use of medium to high physiology section. With a spell as a freelance elec- integration of isolated cell throughout primary screening platforms such as trophysiologist, which included working with membrane patches in a FLIPR and flux, they also have lacked the killer Professor Olaf Pongs in Hamburg, he joined nanomachined aperture.Appl. Phys. Lett., 2000. 77: p. 1218-1220. ion channel screen required to really exploit ion CeNes Ltd in 1997 as Managing Director of the 29 Behrends, JC et al.The channel targets to the full. Now, reflecting the Channelwork group and is currently Director of NanoPatchClamp Array – a planar combined progress in the field summarised above, Technology Development. glass chip for high-throughput there are likely to be some new challengers to cellular electrophysiology. 46th these incumbents. Andrew Silverthorne has a BSc in Biological Annual Biophysical Society Satellite Meeting, 2002. Satellite: Drug SCION (Boston, USA) and IONIX (Cambridge, Sciences from Leicester University, UK and follow- Discovery Technology for Ion UK) are both betting on ion channels in a big way. ing post-graduate research work investigating the Channels, II. SCION has interests in the Kv1.1 channel role of growth factors on oral cancer at Bristol 30 IMS Health:World Drug (demyelinating disorders) among others while University, joined Xenova Ltd where he was Purchases to end Sept. 2001. IONIX, ostensibly a ‘pain’ company, will be focus- involved in drug discovery research, assay develop- 31 PCT application: PCT/GB99/01871. ing to a major extent on ion channels (inter alia the ment, HTS and preclinical research. He then went 32 Southan,AP et al.The SNS sodium channel) expressed in sensory neu- on to join the business development consultancy contrasting effects of rones. Both were founded in 2001. Other specialist Connect Pharma (now PharmaVentures) before dendrotoxins and other companies already working in the ion channel joining CeNes where he is currently Head of potassium channel blockers in domain may also enter the fray. These tend to fall Business Development with commercial responsi- the CA1 and dentate gyrus regions of rat hippocampal into one or both of two categories: bility for its ion channel contract research and slices. Brit. J. Pharmacol., 1997. Ion channel contract research providers technology development businesses. 122: p. 335-43.

Drug Discovery World Spring 2002 61