**3.2 Morphology and beating activity**

Directly after casting, FBMEs appeared as a soft fibrin-block fixed at the end of the silicone posts and exhibiting the dimensions of the casting moulds (12x3xx mm). Within this clot, the heart cells were round and amorphous but homogeneously distributed throughout the gel (Figure 3A). During the first days after casting, cells spread, elongated along force lines, started to beat as single cells and finally in the form of synchronously beating areas on day 4 to 5. The further development was characterized by matrix remodelling and degradation. This lead to a marked reduction of the size (from 3 mm to 1-2 mm width in the presence and 0.2-1 mm in the absence of tranexamic acid) and increased cell-density. Between day 5 and day 7 the cardiomyocytes formed small groups (Figure 3B through E). Coherent beating of the muscle construct started around day 7 to 9. By day 10 the generated force was sufficient to rhythmically deflect the silicone posts. Measurements were routinely performed between day 14 to 16. At this point the cardiomyocytes finally appeared as spindle-shaped cells with an approximately length of 100 to 200 µm and a diameter of 10 to 20 µm (Figure 3F).

Hematoxylin/eosin-stained paraffin sections of mature FBMEs (day 15) showed a dense network of longitudinally aligned cells throughout the gel (Figure 4A). No clear evidence for cell density gradient from peripheral to central areas of the gel could be found, arguing against a significant inhomogeneity of oxygen and nutrient supply.

Multi-Well Engineered Heart Tissue for Drug Screening and Predictive Toxicology 77

**A B** 

Fig. 3. Histological analysis of FBMEs. FBMEs were cultured with aprotinin plus tranexamic acid. For histological analysis they were fixed with formaldehyde at indicated time points, embedded with paraffin and the sections were HE-stained (10x magnification). A Day 0: cells are present as single, round and amorphous cell suspension in fibrin matrix; B day 3: cells spread out along force lines. C day 6, D day 9: degradation of extracellular matrix, cells spread and align with neighbouring cells. E day 12, F day 15: extended degradation of extracellular matrix, the cellular density is increased; cells align and show orientation along

**F** 

**D** 

Whole-mount FBMEs were also analysed by immunofluorescent staining. The pictures showed a dense network of regularly cross striated, α-actinin-positive, longitudinally aligned cardiomyocytes. The elongated cells had well developed sarcomeric structures reaching into the periphery of the cytoplasm. Cardiomyocytes were also characterized by connexion-43 positive structures, the gap junctions. In contrast to the *in vivo* situation connexion-positivity was mostly localized on lateral parts of the cell membrane, but not clustered at the border to connecting cells. Moreover, the immunfluorescence showed lectinpositive endothelial cells intermingled with cardiomycytes and forming primitive tube-like

force lines. Scale bar 100 µm (Hansen et al. 2010).

structures (Figure 4).

**E** 

**C** 

We developed a new technique for the generation and evaluation of contractile cardiac tissue from neonatal rat heart cells *in vitro* (Hansen et al. 2010). In this method, isolated heart cells are mixed with fibrinogen, thrombin and medium and pipetted into rectangular casting moulds made from 2%-agarose in ordinary 24-well cell culture plates. Due to the polymerisation of the fibrin, the gel is fixed to both silicone posts. After 2 h at 37 °C the constructs can be transferred to new culture dishes and maintained under cell culture conditions for several days. Figure 1C demonstrates this procedure in a schematic way. During cultivation, the cells inside the gel spread along the force lines and form extensive cell-cell contacts, the hydrogel is remodelled and degraded. These processes are accompanied by marked condensation of the constructs and deflection of the silicone posts towards each other. The initial length of a FBME directly after casting is 8.5 mm and the mean final length 6.5 mm. The post deflection differs between individual FBMEs, likely reflecting their degree of cardiac tissue development. In consequence, each FBME is exposed to an "individually optimized preload". With this simplified method, 48-72 FBMEs can be routinely generated out of one cell preparation (30 rat hearts). Figure 2B illustrates a typical 24 well plate with silicone posts racks and FBMEs. Fibrin is affected by proteases in the culture medium. To decrease degradation, the protease inhibitor aprotinin at a concentration of 33 µg/ml is added to the medium. This inhibitor markedly reduces fibrinolysis but cannot entirely stop it. To further protect the hydrogel from proteolysis tranexamic acid, another protease inhibitor, can be added to the medium. The combination of both inhibitors results in improved stability and allows longer cultivation periods. Tranexamic acid-treated FBMEs have a markedly increased diameter (final width 1.3 to

1.4 mm [Figure 2C and 8D] instead of 0.2 to 1.0 mm in its absence [Figure 4A]).

Directly after casting, FBMEs appeared as a soft fibrin-block fixed at the end of the silicone posts and exhibiting the dimensions of the casting moulds (12x3xx mm). Within this clot, the heart cells were round and amorphous but homogeneously distributed throughout the gel (Figure 3A). During the first days after casting, cells spread, elongated along force lines, started to beat as single cells and finally in the form of synchronously beating areas on day 4 to 5. The further development was characterized by matrix remodelling and degradation. This lead to a marked reduction of the size (from 3 mm to 1-2 mm width in the presence and 0.2-1 mm in the absence of tranexamic acid) and increased cell-density. Between day 5 and day 7 the cardiomyocytes formed small groups (Figure 3B through E). Coherent beating of the muscle construct started around day 7 to 9. By day 10 the generated force was sufficient to rhythmically deflect the silicone posts. Measurements were routinely performed between day 14 to 16. At this point the cardiomyocytes finally appeared as spindle-shaped cells with an approximately length of

Hematoxylin/eosin-stained paraffin sections of mature FBMEs (day 15) showed a dense network of longitudinally aligned cells throughout the gel (Figure 4A). No clear evidence for cell density gradient from peripheral to central areas of the gel could be found, arguing

**3. Results** 

**3.1 General aspects of the new technique** 

**3.2 Morphology and beating activity** 

100 to 200 µm and a diameter of 10 to 20 µm (Figure 3F).

against a significant inhomogeneity of oxygen and nutrient supply.

Fig. 3. Histological analysis of FBMEs. FBMEs were cultured with aprotinin plus tranexamic acid. For histological analysis they were fixed with formaldehyde at indicated time points, embedded with paraffin and the sections were HE-stained (10x magnification). A Day 0: cells are present as single, round and amorphous cell suspension in fibrin matrix; B day 3: cells spread out along force lines. C day 6, D day 9: degradation of extracellular matrix, cells spread and align with neighbouring cells. E day 12, F day 15: extended degradation of extracellular matrix, the cellular density is increased; cells align and show orientation along force lines. Scale bar 100 µm (Hansen et al. 2010).

Whole-mount FBMEs were also analysed by immunofluorescent staining. The pictures showed a dense network of regularly cross striated, α-actinin-positive, longitudinally aligned cardiomyocytes. The elongated cells had well developed sarcomeric structures reaching into the periphery of the cytoplasm. Cardiomyocytes were also characterized by connexion-43 positive structures, the gap junctions. In contrast to the *in vivo* situation connexion-positivity was mostly localized on lateral parts of the cell membrane, but not clustered at the border to connecting cells. Moreover, the immunfluorescence showed lectinpositive endothelial cells intermingled with cardiomycytes and forming primitive tube-like structures (Figure 4).

Multi-Well Engineered Heart Tissue for Drug Screening and Predictive Toxicology 79

**B** 

**d0 d3 d6 d9 d12 d15**

**<sup>140</sup>** \* \* \*

\*\*\* \*\*

**DNA content [µg]**

**d0 d3 d6 d9 d12 d15**

\*\*\* \*\*

**HEK 293**

test; Hansen et al. 2010).

**0**

**100**

**Extentsion (% of FBME d0)**

**200**

**FBME AraC**

**d0**

**d3**

**3.4 Cardiac marker gene expression over time** 

**d6**

**d9**

\* \*\*\*

**d12**

**d15**

remained several fold higher, indicating higher remodelling activity.

Fig. 5. DNA and total RNA content of FBMEs. A DNA content of FBMEs over time (n=4). B Total RNA content over time (n=4). \* p<0.05 vs. d0 (Student's t test). Bars show means +/- SD. C, D Concentration of phosphorylated histone H3 (C) and caspase-3 activity (D) in FBMEs over time of cultivation. Day 0 represents freshly solidified FBMEs 2 h after casting. Proliferating HEK293 cells and AraC-treated FBMEs served as positive and negative controls for proliferation, respectively. Doxorubicin-treated FBMEs served as positive controls for caspase-3 activity. Bars show mean +/-SEM, n=4. \* p<0.05 vs. d0 (Student's t

To get an idea about cardiomyocyte maturation in FBMEs, transcript levels of known cardiac marker genes (α-actinin, SR Ca2+-ATPase [SERCA], α- and β-myosin heavy chain [α-/β-MHC], Na+/Ca2+-exchanger [NCX], titin) were analysed over time. To avoid bias due to the effect of the drop of overall cell count after preparation, all values were normalized to the mRNA concentration of the cardiac myocyte-specific protein calsequestrin 2. Values were additionally compared to intact adult (ARH) and neonatal rat hearts (NRHT; Figure 6). In the first phase (day 0 to day 6), which represents the time when the single cells spread, formed clusters and started to beat, the transcript levels seemed to be relatively stable. In the second phase (day 9 to day 12) the expression levels reached their maximum concomitantly with the start of rhythmical deflection of the silicone posts by the FBMEs. In the third phase (day 12 to day 15) transcript levels generally decreased. Some cardiac markers (α-actinin, SERCA, α-MHC) reached a comparable level to native myocardium on day 15. Other (β-MHC, titin and NCX)

**FBME Dox**

**Caspase activity (% of FBME Dox)**

**D** 

**d0**

**d3**

**d6**

**d9**

**d12**

**d15**

**DNA content [µg]**

**A** 

**C** 

Fig. 4. Histological analysis of FBMEs (day 15, without tranexamic acid). A, HE-stained paraffin section. Note the almost complete absence of extracellular matrix and welldeveloped cardiac tissue structure. B, Merged immunofluorescence staining for α-actinin (green), lectin (red) and DRAQ5 (blue; 63x magnification). C, Lectin-positive structures alone (63x magnification). D, Connexin-43 (red), phalloidin (green) and DRAQ5 (blue; 63x magnification). Scale bar 50 µm (A) and 20 µM (B-D; Hansen et al. 2010).

#### **3.3 DNA/RNA content, histone H3 phosphorylation and caspase-3 activity**

To further investigate cell survival in FBMEs during culture, the histological data were supported by measurements on a molecular biological level. The DNA/RNA content, histone H3 phosphorylation as well as the caspase-3-activity were analysed. The DNA content dropped by 20% between day 0 and day 3. Thereafter it remained stable for at least two weeks (Figure 5A). Investigations of histone H3 phosphorylation as a marker of proliferative activity and caspase-3 activity as a marker for apoptosis in FBMEs were well in line with these observations. Histone H3 phosphorylation level was initially very low and further decreased over time (Figure 5C). Caspase-3 activity was high directly after cell preparation and dropped during culture even below detectable levels (Figure 5D). After an initial drop of the RNA-content during the first three days of 50% it remained, like the DNA content, more or less stable for at least two weeks (Figure 5B). In summary, these data suggest that some of the cells died within the first three days after casting, likely as a consequence of cell damage during the isolation procedure. Thereafter the cell population remained essentially stable in FBMEs.

**B** 

**D** 

Fig. 4. Histological analysis of FBMEs (day 15, without tranexamic acid). A, HE-stained paraffin section. Note the almost complete absence of extracellular matrix and welldeveloped cardiac tissue structure. B, Merged immunofluorescence staining for α-actinin (green), lectin (red) and DRAQ5 (blue; 63x magnification). C, Lectin-positive structures alone (63x magnification). D, Connexin-43 (red), phalloidin (green) and DRAQ5 (blue; 63x

magnification). Scale bar 50 µm (A) and 20 µM (B-D; Hansen et al. 2010).

remained essentially stable in FBMEs.

**C** 

**A** 

**3.3 DNA/RNA content, histone H3 phosphorylation and caspase-3 activity** 

To further investigate cell survival in FBMEs during culture, the histological data were supported by measurements on a molecular biological level. The DNA/RNA content, histone H3 phosphorylation as well as the caspase-3-activity were analysed. The DNA content dropped by 20% between day 0 and day 3. Thereafter it remained stable for at least two weeks (Figure 5A). Investigations of histone H3 phosphorylation as a marker of proliferative activity and caspase-3 activity as a marker for apoptosis in FBMEs were well in line with these observations. Histone H3 phosphorylation level was initially very low and further decreased over time (Figure 5C). Caspase-3 activity was high directly after cell preparation and dropped during culture even below detectable levels (Figure 5D). After an initial drop of the RNA-content during the first three days of 50% it remained, like the DNA content, more or less stable for at least two weeks (Figure 5B). In summary, these data suggest that some of the cells died within the first three days after casting, likely as a consequence of cell damage during the isolation procedure. Thereafter the cell population

Fig. 5. DNA and total RNA content of FBMEs. A DNA content of FBMEs over time (n=4). B Total RNA content over time (n=4). \* p<0.05 vs. d0 (Student's t test). Bars show means +/- SD. C, D Concentration of phosphorylated histone H3 (C) and caspase-3 activity (D) in FBMEs over time of cultivation. Day 0 represents freshly solidified FBMEs 2 h after casting. Proliferating HEK293 cells and AraC-treated FBMEs served as positive and negative controls for proliferation, respectively. Doxorubicin-treated FBMEs served as positive controls for caspase-3 activity. Bars show mean +/-SEM, n=4. \* p<0.05 vs. d0 (Student's t test; Hansen et al. 2010).

#### **3.4 Cardiac marker gene expression over time**

To get an idea about cardiomyocyte maturation in FBMEs, transcript levels of known cardiac marker genes (α-actinin, SR Ca2+-ATPase [SERCA], α- and β-myosin heavy chain [α-/β-MHC], Na+/Ca2+-exchanger [NCX], titin) were analysed over time. To avoid bias due to the effect of the drop of overall cell count after preparation, all values were normalized to the mRNA concentration of the cardiac myocyte-specific protein calsequestrin 2. Values were additionally compared to intact adult (ARH) and neonatal rat hearts (NRHT; Figure 6). In the first phase (day 0 to day 6), which represents the time when the single cells spread, formed clusters and started to beat, the transcript levels seemed to be relatively stable. In the second phase (day 9 to day 12) the expression levels reached their maximum concomitantly with the start of rhythmical deflection of the silicone posts by the FBMEs. In the third phase (day 12 to day 15) transcript levels generally decreased. Some cardiac markers (α-actinin, SERCA, α-MHC) reached a comparable level to native myocardium on day 15. Other (β-MHC, titin and NCX) remained several fold higher, indicating higher remodelling activity.

Multi-Well Engineered Heart Tissue for Drug Screening and Predictive Toxicology 81

Ion channels play an important role as targets of proarrhythmic drugs. To determine whether the principal ion channel subunits known from human hearts are expressed in rat FBMEs transcripts of 23 ion channel α-subunits (7 calcium channels, 6 sodium channels, 10 potassium channels) were amplified from total RNA of FBMEs and a nonfailing human heart sample by RT-PCR (35 cycles). 22/23 transcripts were amplified from both sources, one channel (CacnA1I) was neither amplified in FBMEs nor in the nonfailing human heart

Fig. 7. Agarose gel of the PCR-products of 23 ion channel subunits. The ion channel profile of FBMEs was compared to the expression profile of a nonfailing human heart. Potassium channels (A), calcium channel (B) and sodium channels (C) showed qualitatively similar results (descriptions indicate the related gene for each channel subunit; for further

**B C** 

To determine robustness and reproducibility or the assay, we generated 6 independent series of FBMEs (1-2 24-well plates each, total number 192 FBMEs) and analysed them under standard conditions (culture medium with insulin, aprotinin and tranexamic acid; measurements done by video-optical recordings). Two FBMEs could not be transferred from the casting moulds, 4 FBMEs were not recognized by the software and 17 did not beat during the recording time (60 s) at day 15. Thus, the total success rate was 89% (169/192; Figure 8). Contraction parameters were examined and compared among different series. These results showed that the average force per series (day 15) was between 0.11 and 0.22 mN (series SD 7.6%), the contraction time (T1=time to peak) ranged between 66 and 81 ms (series SD 41%), relaxation time (T2=time to 80% relaxation) between 67 and 88 ms (series SD 25%), frequency between 162 and 20 beats per minute (series SD 109%), construct diameter between 1.3 and 1.4 mm (series SD 9.9%) and length between 5.6 and 6.7 mm (series SD 16.7%). The relatively large size in FBME diameter in these examinations could be

**3.5 Non quantitative ion channel expression profile** 

**A** 

information see table 1; Hansen et al. 2010).

attributed to the use of tranexamic acid.

**3.6 Robustness and reproducibility of the new method** 

(Figure 7).

Fig. 6. RT-qPCR of FBMEs in comparison to neonatal (NRHT) and adult rat heart (ART). ∆∆CT values were generated by normalisation to the mRNA of cardiac specific protein calsequestrin 2 (average CT values for normalisation were as follows: d0: 20.8, d3: 20.4, d6: 20.6, d9: 21.3, d12: 21.6, d15: 20.5). Figures show relative expression compared to day 0. Each bar represents results from 4 biological replicates (each measured 3 times). \* p<0.05 vs. day 15 (Student's t test). Bars show means +/- SEM (Hansen et al. 2010).
