Keeping the Rhythm: hERG and Beyond in Cardiovascular Safety Pharmacology

Clemens Möller


Expert Rev Clin Pharmacol. 2010;3(3):321-329. 

In This Article

Novel Assays for Early Testing the Cardiac Liabilities of Compounds

Novel test systems for assessing cardiac liabilities of compounds are continuously being developed and have reached fairly high levels of validation. Some of the areas where significant recent advances have been made to predict and/or understand potential cardiac liabilities of compounds early include in silico cardiac ion channel modeling, automated patch clamp electrophysiology on cardiac ion channels, electrophysiology on stem cell-derived cardiac myocytes and drug tests in a cardiac zebrafish model.

Ion Channel Modeling

Numerous in silico models of the hERG ion channel and prediction algorithms for interactions between compounds and the hERG ion channel have been published over the past years. Until recently, the majority of published in silico hERG models have, however, suffered from a large fraction of false positives. A common pharmacophore model for compounds interacting with the hERG channel has the 'hydrophobe/acceptor–linker–positive charge–linker–hydrophobe' motif,[25,34] which is contained in many structures of common drugs, including those showing only very minor interactions with the hERG channel. Taking recent knowledge of the structures of ion channels into account, often by homology modeling[26,34–36] and incorporating the charges of the hERG channel pore into the prediction, these models could be significantly refined. In silico modeling still delivers a number of false positives and false negatives, and thus may, not be considered a contribution to the actual cardiac safety testing. The important impact of hERG models to a number of drug discovery programs comes from the benefit that they not only deliver highly significant predictions of hERG channel interactions, but also give the medicinal chemists involved, in an optimization program, evidence on which structural elements of a molecule would require modification to reduce hERG channel affinity.

Automated Electrophysiology

While the predictive power of in silico hERG models has increased, patch clamp electrophysiology remains the gold-standard method for assessing the effects of compounds on the hERG ion channel. Traditional manual patch clamp analysis, however, is a slow and time-consuming process and requires a highly skilled operator. Actual throughput of a patch clamper obviously depends on the protocols used, but a typical throughput common to manual patch clamping is in the order of one compound per day. This compound turnover is not compatible with typical iteration cycles during medicinal chemistry optimization of the lead series. Other higher-throughput ion channel analysis methods have, until recently, suffered from inadequate voltage control or other constraints and, thus, have not delivered the quality, physiological relevance and consistence of patch clamp electrophysiology data. With the recent development of automated patch clamp electrophysiology robots, based on planar patch clamping, this major bottleneck appears much alleviated. Following the high demand in the pharmaceutical industry for a highly predictive, higher-throughput hERG channel assay, the hERG ion channel was one of the first ion channels to be extensively studied on these instruments. A careful calibration of the required protocols and procedures associated with running the patch clamp robots allowed the establishment of validated systems correlating very well with manual electrophysiology, and several companies have launched, and are continuing to launch, automated patch clamp robots with increasing throughput and advanced capabilities.[37–40] Such instruments are capable of achieving a throughput in the order of tens or hundreds (e.g., PatchLiner [Nanion], PatchXpress [MDS], QPatch 16 [Sophion]) to thousands (e.g., IonWorks Quattro [MDS], QPatch HTX [Sophion]) of compound tests per day. It needs to be stressed, however, that higher-throughput automated patch clamp electrophysiology data have the same shortcomings and potential weaknesses as any other high-throughput method. These are mainly attributed to physicochemical properties of compounds, and include potentially missing active compounds because of poor solubility or adsorption to the material, which is due to the unfavorable surface-to-volume ratio of miniaturized screening. Consequently, automated patch clamp data should not be expected to replace GLP standard, manual, patch clamp electrophysiology data of ion channel effects in the close future.

Human Stem Cell-derived Cardiac Myocytes

In contrast to molecularly characterized and heterogeneously expressed ion channels, patch clamp electrophysiology of cardiac myocytes has the benefit of delivering information on the effects of compounds on the cardiac action potential composed of all cardiac ion channels, within a more native environment of the channels. Until recently, there had been a lack of homogenous, high-quality human cardiac cells for in vitro research. With an improved understanding of how to differentiate stem cells into cardiac myocytes, a novel source of high-quality functional myocytes appears to be evolving,[41] and validation of these cells for use in safety pharmacology has reached a fairly high level.[42,43] Figure 2 shows a series of action potentials recorded by manual patch clamp electrophysiology from spontaneously beating stemcell-derived cardiac myocytes. The relevant phases of the action potential, as discussed earlier, are clearly visible and the effects of compounds on cardiac action potential recordings have been demonstrated. Recently, the first stem cell-derived cardiac myocytes have been applied to automated patch clamp electrophysiology[44] and to multielectrode array systems, recording surface potentials from spontaneously beating cells.[45,46] A recent study has shown a good correlation of extracellular field potential recordings of stem cell-derived cardiomyocytes with QT values from patients receiving 12 different drugs.[47]

Figure 2.

Action potential recordings from stem cell-derived cardiac myocytes from mouse. Cells were maintained in confluent layers, and a spontaneous, continuous beating was observed, corresponding to the action potentials elicited in the electrophysiological patch clamp recordings.

Difficult ethical issues are associated with the generation and access to suitable human embryonic stem cell lines. Landmark investigations demonstrated an ability to induce the reversal of normal somatic tissue-derived cell lines into stem cell populations, termed induced pluripotent stem cells (iPS), thus negating potential ethical issues.[48] While iPS cells do show subtle differences in their gene-expression profiles,[49] they have been shown to be very similar in their phenotype and epigenetic status to embryonic stem cell lines, and cardiomyocyte cell lines derived from iPS cells have been shown to be tools for safety screening.[50,51] Development of further cell lines originating from iPS cells is ongoing and considerable interest, including that from the commercial side, is placed on stem cell-derived cardiomyocytes.[52,53]

Zebrafish Screening

Entering a higher level of organization than individual cells and tissue preparations, there has been recent excitement concerning the use of zebrafish (Danio rerio) as a simple in vivo model system for drug discovery and development. This has also led to the development of assays reported to be useful as early screens in safety pharmacology.[54–56] Indeed, the zebrafish has a number of features that makes it an attractive model system for use in drug discovery, genetics and other lifescience research. The little tropical fish grows from larvae of few millimeters in size, making it relatively easy to breed and to be assayed in medium throughput format. It has many of the organs present in mammals, and a number of genes have been shown to possess a high level of homology to the corresponding human genes. The transparency of the larvae permits the different organs of the fish to be visually observed, including the beating heart, although it is composed of two chambers, accounting for significant differences to human heart physiology. Nevertheless, using an optical readout that tracks ventricular and atrial contraction independently, many physiological responses of the human cardiovascular system can be reproduced in zebrafish heart. These include effects of IKr blockers, such as terfenadine (which, in zebrafish, cause bradycardia and atrial:ventricular decoupling[55]), but also responses arising only at higher levels of organization, such as compound effects on other cardiac ion channels and drug–drug interactions. A major challenge in the zebrafish model is the control of drug uptake. For a number of drugs, discrepancies between zebrafish responses and hERG channel inhibitions have been reported, which are caused by poor absorption of some drugs into the fish.[56] Therefore, a meaningful zebrafish study requires an accompanying analysis of the drug concentrations actually present in the relevant zebrafish organ, and this is how zebrafish studies are controlled in the author's laboratory. Currently, the safety pharmacology community appears interested in using the zebrafish as a model system for a higher-throughput analysis of cardiotoxicity; however, a more thorough validation of the model appears to be required before zebrafish would be accepted as a model system more widely.[57]