**Multi-Well Engineered Heart Tissue for Drug Screening and Predictive Toxicology**

Alexandra Eder, Arne Hansen and Thomas Eschenhagen *Department of Experimental Pharmacology and Toxicology, University Medical Centre Hamburg-Eppendorf, Germany* 

## **1. Introduction**

70 Toxicity and Drug Testing

Thomas, W.D.E. & Potter, L. (1975). Solution/Air Interfaces I. An oscillating jet relative

Trukhin, D.V.; Sinyachenko, O.V.; Kazakov, V.N.; Lylyk, S.V.; Belokon, A.M. & Pison, U.

van Oss, C.J.; Good, R.J. & Chaudhury, M.K. (1986). The role of van der Waals forces and

energy surfaces. *Journal of Colloid and Interface Science*, Vol. 111, pp. 378–390. Vasir, J.K.; Tambwekar, K. & Garg, S. (2003). Bioadhesive microspheres as a controlled drug delivery system. *International Journal of Pharmaceutics*, Vol. 255, pp. 13–32. Vejnovic, I.; Simmler, L. & Betz, G. (2010). Investigation of different formulations for drug

Warszunski, P.; Wantke, K.-D. & Fruhner, H. (1998). Surface elasticity of oscillating spherical

Xue, H.T.; Fang, Z.N.; Yang, Y.; Huang, J.P. & Zhou, L.W. (2006). Contact angle determined

Zelkó, R.; Orbán, Á.; Süvegh, K.; Riedl, Z. & Rácz, I. (2002). Effect of plasticizer on the

Zhang, L.-H. & Zhao, G.-X. (1989). Dynamic surface tension of the aqueous solutions of

Zhou, G.; Ge, W. & Li, J. (2008). A revised surface tension model for macro-scale particle

*Science*. Vol. 50. pp. 397-412.

185-194.

pp. 353-361.

*and Surfaces B: Biointerfaces*, Vol. 21, pp. 231-238.

interfaces, *Colloids and Surfaces A*, Vol. 139, pp. 137–153.

theory. *Chemical Physics Letters*, Vol. 432, pp. 326–330.

*Journal of Pharmaceutics,* Vol. 244, pp. 81-86.

methods. *Powder Technology*, Vol. 183, pp. 21–26.

method for determining dynamic surface tensions. *Journal of Colloid and Interface* 

(2001). Dynamic surface tension and surface rheology of biological liquids. *Colloids* 

hydrogen bonds in ''hydrophobic interactions'' between biopolymers and low

delivery through the nail plate, *International Journal of Pharmaceutics*, Vol. 386, pp.

by spontaneous dynamic capillary rises with hydrostatic effects: Experiment and

dynamic surface tension and the free volume of Eudragit systems. *International* 

cationic-anionic surfactant mixtures. *Journal of Colloid and Interface Science*, Vol. 127,

Drug development is time- and cost-intensive and, overall, inefficient. Only one out of an estimated 10.000 new chemical entities (NCEs) finally enters the market. The later the failure occurs, the higher are the costs. It is for this reason that preclinical development aims at identifying the potential for failure as early as possible and with high sensitivity. On the other hand, high sensitivity generally also means low specificity, suggesting that many potentially successful NCEs are currently excluded from further development. Common reasons for exclusion are adverse drug reactions (ADR). Among the various ADRs, cardiac toxicities and arrhythmias play an important role, because they represent about 21% of all ADRs (Lasser et al., 2002) and are frequently lethal. The single most important mechanism in this context is the prolongation of cardiac repolarization bearing a proarrhythmic potential. These interferences can be visualized by standard ECGs as a prolongation of the QT-interval. Such a prolongation is called "long QT-syndrome" (LQT-syndrome) and is associated with *Torsade-de-Pointes* (TdP) arrhythmias and sudden cardiac death. In the past, several prominent drugs had to be withdrawn from the market due to TdP in humans, e.g. astemizole, terfenadine, cisapride, sparfloxacin, grepafloxacin and recently clobutinol (Silomat). Moreover, numerous drugs are still on the market that are associated with the potential to cause LQT and TdP, including widely prescribed drugs such as the antibiotic erythromycin.

Given the fatal consequences of LQT and TdP in healthy patients without any cardiac disposition, the regulatory bodies (FDA, EMEA and others) have decided some years ago to require testing for LQT to be an obligatory part of preclinical development of any NCE. Several tests have been developed and some of them are routinely used. The three major (but by far not exclusive) tests in the field are the HERG test, rabbit Purkinje fibers and telemetry in dogs. These tests have different advantages and disadvantages and are generally employed subsequently. The HERG test can be considered an obligatory test for all NCEs and it is unlikely that any company further develops a compound that showed major inhibitory activity in this test (of a single ion channel activity). However, examples exist of successful drugs on the market that are potent inhibitors of the HERG current without ever giving rise to TdP arrhythmia (e.g. verapamil, azithromycin). Thus, the predictive value of the HERG test is limited. Reasons lie, among others, in its inability to

Multi-Well Engineered Heart Tissue for Drug Screening and Predictive Toxicology 73

softness in comparison to other filamentous biopolymers (Janmey et al., 2009). The final properties of fibrin are mainly governed by the concentrations of fibrinogen and thrombin. Additionally fibrin properties can be affected by the introduction of bonds by plasma transglutaminase (factor XIII; Janmey et al., 2009). In contrast to other extracellular matrices, the *in vitro* polymerisation of fibrin is very close to the *in vivo* fibrin polymer. Fibrin gels are fully degradable by fibrinolytic enzymes like plasmin. All together, the mechanical and biological properties, its availability from autologous sources in addition to the possibility to covalently bind growth or other factors (Hubbell 2003) make fibrin an interesting compound

Total heart cells (excluding the atria) were isolated from neonatal Wistar rats (postnatal day 0 to 3) by a fractionated DNase/Trypsin digestion protocol as previously described (Eschenhagen & Zimmermann, 2005). The resulting cell populations were immediately subjected to FBME generation. Experimental procedures were reviewed and approved by

For the generation of fibrin-based mini-EHTs (FBMEs), Teflon spacer and silicone post racks were used. The Teflon spacers were important for the casting molds. They had the following geometry (Figure: 1B): length 12 mm, width 3 mm, height 13.5 mm. Sylgard 184 silicone elastomer (Dow Corning) was used for the production of silicone post racks, which were needed for culturing the FBMEs. The silicone post racks were made in custom-made Teflon casting molds. According to the manufacture's instructions, the 2-component Sylgard 184 was degassed under vacuum conditions before casting. The final silicone post racks consisted of 4 pairs of posts, having a little plate at their end. The racks had the following geometry (Figure 1A): length/width of rack: 79x18.5 mm, length of posts 12 mm, diameter 1 mm, plate diameter 2 mm, distance (center-center) 8.5 mm. They were initially self-made and currently industrial-

The reconstitution mixture for the generation of fibrin-based mini-EHTs was prepared on ice as follows (final concentration): 4.1x106 cells/ml, 5 mg/ml bovine fibrinogen (Sigma F4753, stock solution: 200 mg/ml in 0.9% NaCl supplemented with 0.5 µg/mg aprotinin), 3 U/ml bovine thrombin (Sigma T7513, stock solution: 100 U/ml). To ensure isotonic conditions, one additional fibrinogen and thrombin volume of 2x DMEM was added. Ordinary 24-well cell culture plates were used as casting molds. After 1.6 ml of sterile 2% agarose (Invitrogen 15510-027) in PBS was pipetted into each well, the Teflon spacers could be placed. After the agarose was solidified, the Teflon spacers were removed. The silicone posts racks were placed onto the cell culture dish with each pair of silicone posts reaching into one of the preformed casting molds (geometry: 12x3x4 mm). The reconstitution mix was carefully resuspended. For each FBME 100 µl of the mixture was mixed briefly with an appropriate volume of thrombin and pipetted into an agarose slot. To ensure complete polymerization of the fibrinogen, the constructs were placed into a humidified cell culture incubator (37 °C, 7% CO2, 40% O2) for 2 hours. Before transferring the silicone posts racks to

for tissue engineering approaches.

Ethics Committee, Hamburg University.

**2.3 Generation of fibrin-based mini-EHTs** 

**2.2 Manufacturing teflon spacers and sylgard posts racks** 

made. Silicone post racks can be autoclaved and reused for several times.

**2. Methods** 

**2.1 Cell isolation** 

give an integrated readout of effects of drugs on the electrophysiology of the intact cardiac myocyte or the intact heart as a multicellular organ consisting of a functional syncytium of cardiac myocytes and all other cardiac cell types that make up normal heart tissue (e.g. fibroblasts, endothelial cells and smooth muscle cells).

#### **1.1 Cardiac tissue engineering**

Over the past decade, techniques have been developed to generate cardiac tissue-like 3 dimensional constructs *in vitro* (Eschenhagen & Zimmermann, 2005). The field of cardiac tissue engineering opened the possibility for many applications. Artificial hart constructs may serve as means for cell-based cardiac repair and as improved *in vitro* models for predictive toxicology and target validation, taking advantage of a more physiological cellular environment. Previous studies used different approaches to construct engineered tissues: Cell seeding onto solid, preformed scaffolds (Carrier et al., 1999; Engelmayr et al., 2008; Leor et al., 2000; Li et al., 2000; Ott et al., 2008; Radisic et al., 2004), matrix-free generation of tissues from stackable cell sheets (Shimizu et al., 2002) or the generation of constructs in preformed casting moulds using hydrogels such as collagen I, matrigel, fibronectin or fibrin (Bian et al., 2009; Eschenhagen et al., 1997; Huang et al., 2007; Naito et al., 2006; Zimmermann et al., 2002). The hydrogel technique has been shown to be suitable for both, cardiac repair *in vivo* (Zimmermann et al., 2006) and target validation *in vitro* (El-Armouche et al., 2007). Circular engineered heart tissues (EHTs) were made by casting neonatal rat heart cells, collagen I and matrigel into circular casting moulds and develop a high degree of cellular differentiation, longitudinal orientation, intercellular coupling and force generation (Zimmerman et al., 2002). It turned out that several factors improve tissue quality and force generation of EHT such as phasic (Fink et al., 2000) or auxotonic stretch, increased ambient oxygen concentration during culture and supplementation with insulin (Zimmermann et al., 2006). Others demonstrated beneficial effects of electrical stimulation (Radisic et al., 2004). The possibility to generate cardiac myocytes from human embryonic stem cells (Kehat et al., 2001) or induced pluripotent stem cells (Zhang et al., 2009) have opened the realistic and exciting perspective to use these techniques for the validation of hypotheses and testing drugs in healthy and diseased human heart muscles (Zimmermann & Eschenhagen 2007).

The current techniques to generate engineered cardiac tissues are either not suitable for this purpose (stacked cell sheet technique) or exhibit drawbacks that limit their usefulness. Extensive handling steps preclude routine execution of large series in an at least medium through put scale and are always a source of variability. Furthermore, the EHT technique in the ring format requires relatively high numbers of cells and turned out to be difficult to miniaturize.

In this chapter we describe a new EHT technique that was driven by the intention to miniaturize the EHT-format for multi-well-testing and automated evaluation and to determine the suitability of EHTs for drug screening and predictive toxicology. The main results have been published in a recent original paper (Hansen et al. 2010). An essentiel change was to use fibrin(ogen) instead of collagen I as a matrix. Fibrinogen is part of the blood clotting cascade. It is a glycoprotein with a size of 340 kDa. Physiologically it achieves plasma concentrations of 1.5 to 4 g/l and can be relative easily purified from different species. An important mechanical property is its nonlinear elasticity. Due to this, fibrin polymers have a high elastic modulus under shear stress combined with a beneficial softness in comparison to other filamentous biopolymers (Janmey et al., 2009). The final properties of fibrin are mainly governed by the concentrations of fibrinogen and thrombin. Additionally fibrin properties can be affected by the introduction of bonds by plasma transglutaminase (factor XIII; Janmey et al., 2009). In contrast to other extracellular matrices, the *in vitro* polymerisation of fibrin is very close to the *in vivo* fibrin polymer. Fibrin gels are fully degradable by fibrinolytic enzymes like plasmin. All together, the mechanical and biological properties, its availability from autologous sources in addition to the possibility to covalently bind growth or other factors (Hubbell 2003) make fibrin an interesting compound for tissue engineering approaches.
