Author contributions: D.M.: collection of data, data analysis and interpretation, manuscript writing; M.R., R.D., and W.R.: collection of data, data analysis and interpretation; N.S.: provision of study material, collection of data, data analysis and interpretation; J.H.: data interpretation; M.P.: provision of study material, data interpretation; W.B. and B.K.F.: conception and design, data interpretation, financial support, manuscript writing, final approval of the manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSExpress October 16, 2008.
Laminins form a large family of extracellular matrix (ECM) proteins, and their expression is a prerequisite for normal embryonic development. Herein we investigated the role of the laminin γ1 chain for cardiac muscle differentiation and function using cardiomyocytes derived from embryonic stem cells deficient in the LAMC1 gene. Laminin γ1 (−/−) cardiomyocytes lacked basement membranes (BM), whereas their sarcomeric organization was unaffected. Accordingly, electrical activity and hormonal regulation were found to be intact. However, the inadequate BM formation led to an increase of ECM deposits between adjacent cardiomyocytes, and this resulted in defects of the electrical signal propagation. Furthermore, we also found an increase in the number of pacemaker areas. Thus, although laminin and intact BM are not essential for cardiomyocyte development and differentiation per se, they are required for the normal deposition of matrix molecules and critical for intact electrical signal propagation. STEM CELLS2009;27:88–99
Basement membranes (BM) are thin layers of extracellular matrix (ECM) that surround individual cells and cell layers. Laminins, collagen IV, and nidogens are considered critical structural elements of the BM , and an intact BM formation is a key prerequisite for normal tissue development and function [2, 3]. Laminin is a heterotrimeric ECM glycoprotein consisting of α, β, and γ chains. Its fundamental importance was unequivocally proven by targeting the LAMC1 gene encoding the γ1 chain, resulting in homozygous embryos with very early embryonic lethality at embryonic day 5.5 . The particular biological relevance of laminins for muscle development and differentiation was underscored by the observation that mutations in the LAMA2 and LAMA4 genes cause different types of severe skeletal muscle pathologies, such as the congenital muscle dystrophy or the milder limb girdle type 2I in humans [5–7]. A LAMA4 mutation was also found to result in alterations of endothelial and cardiac cell morphology . Furthermore, the LAMA4 gene appears to be responsible for changes in heart function due to abnormal cardiovascular ECM, and the resulting phenotype is reminiscent of cardiac ischemia . Studies in C2C12 cells addressed the mechanism of action of laminins in skeletal muscle and revealed that its polymerization and interaction with cell surface receptors were necessary and sufficient to create a cortical cellular architecture wherein the other components of the BM were integrated . Because of these findings, we wondered whether the development and differentiation of cardiac muscle cells, similar to those of skeletal muscle cells, could be greatly altered by the deletion of the predominant laminin subtype. For this purpose we have chosen deletion of laminin γ1, as this chain is ubiquitous in BM and particularly strongly expressed in heart muscle tissue compared with other types of striated muscle . Indeed, this laminin chain is an obligatory component in most trimeric laminin forms. Because of the early embryonic lethality of the LAMC1-deficient mice, we have chosen the embryonic stem (ES) cell in vitro differentiation technique [11, 12] as a suitable model to study the cell biological and functional consequences of the laminin γ1 deletion on cardiomyocyte development and differentiation. This approach appeared preferable to a cardiomyocyte-specific knockout strategy because the other cell types present in the heart (i.e., endothelial cells and fibroblasts) also contribute to laminin formation.
Herein, we demonstrate that cardiomyocyte development and differentiation are preserved in laminin γ1 (−/−) cardiomyocytes despite the complete absence of BM formation. We also show that disturbed BM formation causes defective ECM assembly, matrix deposits, and secondary disturbances in the propagation of the electrical signals.
MATERIALS AND METHODS
ES Cell Preparation
Heterozygous (+/−) and homozygous (−/−) laminin γ1 chain ES cells, generated on an R1 background [4, 13], were cultured and differentiated into spontaneously beating cardiomyocytes as previously described . Single cardiomyocytes were isolated from clusters of spontaneously beating areas . The laminin γ1 (−/−) deletion was produced by two sequential targeting events in cells. Thus, (+/−) and (−/−) cells have the same genetic background, and therefore heterozygous ES cells were used as control .
Microelectrode Array Mapping Technique
Extracellular recordings on beating laminin (+/−) and (−/−) ES cell clusters were performed using a microelectrode array (MEA) data acquisition system (Multi Channel Systems, Reutlingen, Germany, http://www.multichannelsystems.com) [16, 17]. Standard measurements were performed at 2 kHz (bandwidth, 1–5 kHz) in normal Tyrode solution at 37°C. Data were analyzed off-line with a customized toolbox programmed with MATLAB (MathWorks, Natick, MA, http://www.mathworks.com) to detect and characterize field potentials (FPs) as described earlier . The analysis of the interspike intervals, measured as readout for the beating frequency, was used to study the hormonal regulation of the spontaneous electrical activity (also described in [17, 18]). Propagation velocities were calculated as reported earlier  (supporting information data).
For whole-cell patch-clamp recordings only spontaneously beating, single cardiomyocytes were used. Patch-clamp experiments applying the current- or voltage-clamp mode of the whole-cell configuration were performed as previously reported [20, 21]. Briefly, for measuring peak L-type Ca2+ current (ICa-L), cardiomyocytes were held at a holding potential of −80 millivolt (mV), and depolarizing voltage presteps to −40 mV for 50 milliseconds were applied to inactivate INa. Thereafter, 100-millisecond voltage steps to 0 mV were applied at a frequency of 0.2 Hz. The pipette solution contained the following (in mM): 120 CsCl, 1 MgCl2, 5 Mg-ATP, 10 EGTA, and 5 Hepes (pH 7.4; CsOH). The extracellular solution was as follows (in mM): 120 NaCl, 5 KCl, 3.6 CaCl2, 20 tetraethylammonium-chloride, 1 MgCl2, and 10 Hepes (pH 7.4; tetraethylammonium-hydroxide). The stimulation or inhibition of ICa-L is reported in terms of the percentage of the increase or decrease of ICa-L density, respectively. Muscarinic effects on cardiomyocytes were calculated in terms of percentage of variation with respect to isoprenalin (ISO) stimulation. Action potential (AP) recordings were obtained in the current clamp configuration. The pipette solution contained the following (in mM): 50 KCl, 1 MgCl2, 3 Mg-ATP, 10 EGTA, 80 K+ L-aspartate, and 10 Hepes (pH 7.4; KOH). The extracellular solution was as follows (in mM): 140 NaCl, 5.4 KCl, 1.8 CaCl2, 2 MgCl2, 10 Hepes, 10 glucose (pH 7.4; NaOH).
Antibodies included rat anti-mouse actinin (1:500; Sigma-Aldrich, Munich, Germany, http://www.sigmaaldrich.com), rabbit polyclonal anti-collagen IV (1:500; Acris Antibodies GmbH, Herford, Germany, http://www.acris-antibodies.com), mouse anti-rat monoclonal perlecan (1:500; Biotrend, Cologne, Germany, http://www.biotrend.com), rabbit polyclonal anti-nidogen-1 and -2 (1:1,000 ), rabbit polyclonal anti-β1 integrin (1:500; Chemicon, Temecula, CA, http://www.chemicon.com), rabbit polyclonal anti-connexin 43 (Cx43) (1:400; Alpha Diagnostics, San Antonio, http://www.4adi.com), monoclonal anti-laminin γ3 (1:1,000, a gift from Dr. Manuel Koch, University of Cologne), rabbit polyclonal anti-N-cadherin (1:250; Zymed Laboratories, South San Francisco, http://www.invitrogen.com), monoclonal antibody for mouse reticular fibroblasts, BM4018 (1:500; Acris Antibodies), rabbit anti-hyperpolarization-activated cyclic nucleotide gated potassium channel 4 (HCN4) (Alomone Labs, Jerusalem, http://www.alomone.com), and rabbit polyclonal collagen VI (1:2,000; kindly provided by the late Dr. Rupert Timpl, Max Planck Institute for Biochemistry, Martinsried, Germany).
Morphological Analysis, Deconvolution, and Confocal Microscopy
The distribution pattern of some proteins was analyzed by deconvolution fluorescence microscopy (Axiovert 200 M; Carl Zeiss, Göttingen, Germany, http://www.zeiss.com) or with an apotome (Carl Zeiss) in laminin γ1 (+/−) (7 + 6 days of differentiation) and laminin γ1 (−/−) (7 + 6 days of differentiation) EBs. Alternatively, where indicated, antibody distribution was observed by confocal microscopy using the LSM 510 META Zeiss microscope.
Proteins from 10 EBs were incubated with rabbit polyclonal antiserum against β1 integrins (Chemicon), rabbit antiserum against fibronectin (raised against highly purified human plasma fibronectin), rabbit collagen I (Quartett, Berlin, http://www.quartett.com), collagen III (Abcam, Cambridge, U.K., http://www.abcam.com), collagen IV (Acris Antibodies), collagen VI (Fitzgerald Inc., Concord, MA, http://www.fitzgerald-Fii.com), a rat monoclonal antibody against the nidogen-1 (entactin) G2 domain (MAB 1884; Chemicon), a rabbit antiserum against Cx43 , or a rabbit Cx45 antibody .
Statistical significance of differences between groups was determined using the paired Student's t test for the effects of drugs on ICa-L, AP, and FP in the same cells and/or EBs and with the analysis of variance with Bonferroni post hoc and/or Student's t test for unpaired data between the control and laminin γ1 (−/−) groups (SPSS 12.0, SPSS Inc., Chicago, IL, http://www.spss.com). A Fisher exact test was applied for the analysis of pacemaker cells with GraphPad Prism 5 (GraphPad Software, Inc., San Diego, http://www.graphpad.com). A p value below.05 was considered statistically significant.
Cardiomyocyte Development and Differentiation
To investigate the influence of BM on the development of cardiac muscle cells and their differentiation, we used laminin γ1 (+/−) or laminin γ1 (−/−) ES cells for the in vitro differentiation [14, 25]. Both ES cell lines developed spontaneous beating areas in EBs, although laminin γ1 (−/−) EBs displayed a lower plating efficiency (approximately 60%). The in vitro differentiation pattern was similar in laminin γ1 (+/−) and laminin γ1 (−/−) EBs, start of beating was regularly observed 3–4 days after plating, and beating was seen to continue at least for 7–8 days thereafter. Next, we analyzed the sarcomeric organization of cardiomyocytes in plated whole EBs by immunostaining with cardiac α-actinin. In both control and mutant EBs, normal spindle-shaped and triangular cardiomyocytes with intact sarcomeric cytoarchitecture and myofibrillar orientation were observed (Fig. 1A). These findings indicated that the development and differentiation of cardiomyocytes were preserved despite lack of the laminin γ1 chain. In addition, we also investigated whether quantitative differences in cardiomyocyte differentiation could be found. For this purpose 20 EBs were dissociated, and the cells were replated and stained with cardiac α-actinin after fixation. We could not detect significant differences in the number of cardiomyocytes, as 18.9 ± 0.3 (−/−) (n = 7 experiments) and 21.3 ± 1.5 (+/−) (n = 6 experiments) of the 50,000 cells plated from the dissociated EBs for each experiment were cardiac muscle cells.
As γ1-containing laminins are important for BM formation and architecture , we next determined whether BM assembly and organization occurred normally in laminin γ1 (−/−) cardiomyocytes. Since collagen IV is a reliable indicator of the presence and formation of BM, we studied its distribution in EBs in vicinity of cardiomyocytes using double immunofluorescence labeling for cardiac α-actinin and collagen IV. As expected, collagen IV surrounded cardiomyocytes in control laminin γ1 (+/−) EBs but was absent in laminin γ1 (−/−) EBs (Fig. 1B). This finding was corroborated by Western blotting on isolated beating areas where no collagen IV was detected (Fig. 1B, middle panel). Moreover, staining for the BM components nidogen-1 and -2 was disrupted in the laminin γ1 (−/−) EBs (Fig. 1C), and this correlated well with Western blotting, which revealed a marked reduction in the nidogen-1 (49% ± 6.1% vs. 100% in control) extractable from the matrix surrounding the cardiomyocytes (Fig. 1C, middle panel). We also quantified other ECM molecules in Western blots and found, in accordance with defective BM formation and ECM assembly, an upregulation of fibronectin (150% ± 4.5% vs. 100% in control), collagen I (100% ± 14%), and collagen VI (94% ± 1.1%) (supporting information Fig. 2), as well as the above-reported downregulation of other components. To rule out clonal aberration and to directly correlate the absence of BM formation with deletion of the laminin γ1 chain, we performed reconstitution experiments by adding 25 μg/ml laminin 111 protein to laminin γ1 (−/−) EBs. In the reconstituted EBs a partial recovery of collagen IV and of BM deposits surrounding the cardiomyocytes was detected (Fig. 1D), and this is clearly illustrated in the accompanying stack animations (supporting information Videos 1, 2). We next analyzed whether compensatory mechanisms could be responsible for the surprising finding of the apparently normal cardiac development and differentiation and intact cytoarchitecture of the laminin γ1 (−/−) cardiomyocytes. One candidate molecule was the laminin γ3 chain, known to be present in laminin 213, an isoform reported to be expressed in the heart [26, 27]. Double immunostaining for laminin γ3 and cardiac α-actinin revealed a more organized structure of the γ3 chain surrounding laminin γ1 (−/−) cardiomyocytes, suggesting that γ3-containing laminin molecules could act as a new binding partner between cell surface molecules and hence induce cytoskeletal reorganization. However, this rescue was not sufficient to restore intact BM assembly (Fig. 2A, 2B); the γ3 chain expression could not be quantified, as the antibody did not work under reducing conditions. Besides the γ3 chain, we also investigated N-cadherin expression and found more intense staining at the border of the cardiomyocytes in laminin γ1 (−/−) EBs compared with controls (Fig. 2C, 2D). These immunohistological findings suggest a morphological redistribution of cytoskeletal components. Thus, deletion of the laminin γ1 chain results in a complete lack of BM formation, whereas the development and structure of cardiomyocytes remained intact, presumably because of the presence of γ3-containing laminins and a compensatory upregulation of cadherins [28, 29].
Expression and Distribution of Integrins and Associated Cytoskeletal Components
Lack or disruption of BM could also affect the distribution and function of β1 integrins, which we reported to be important for heart muscle development  and function . In fact, our earlier studies in β1 integrin (−/−) ES cell-derived cardiomyocytes revealed that specifically muscarinic signaling was missing. Therefore, we first analyzed the expression of these key molecules with immunocytochemistry and Western blotting. In control cardiomyocytes a close association of β1 integrins with the z-line was observed, whereas a less defined distribution pattern was seen in the laminin γ1 (−/−) cardiomyocytes (Fig. 2E, 2F). This was caused by a homogeneous redistribution of β1 integrins in the absence of intact BM. The densitometric analysis of β1 integrin expression revealed only a modest reduction in laminin γ1 (−/−) EBs (75.6% ± 4.3% β1 integrin expression vs. 100% in control [n = 2; p = .03; Fig. 2G]). The distribution of talin, a focal adhesion-associated molecule and member of coronary cytoskeletal proteins, revealed an intact and stable cytoskeletal organization (Fig. 2H). This differs from our findings in β1 integrin (−/−) cardiomyocytes  and suggests that laminin assembly and BM organization are independent processes.
Differentiation into Cardiomyocyte Subtypes
Our findings suggest that cardiomyocyte development is intact in laminin γ1 (−/−) ES cells, and we therefore investigated next whether their differentiation into the different cardiomyocyte subtypes was preserved. For this purpose we determined key parameters of AP shape and duration using the patch-clamp technique: maximum diastolic potential (MDP), maximum rate of rise of the AP (dV/dt), and action potential duration at 50% and 90% of repolarization. We tested isolated cardiomyocytes obtained from seven (−/−) and six (+/−) EB preparations (details also given in Fig. 3A, 3B; Table 1). As established earlier [14, 30], we could classify the ES cell-derived cardiomyocytes on the basis of AP shape and duration into three subtypes: pacemaker-, atrial-, and ventricular-like cells. Interestingly, we found in the laminin γ1 (−/−) EBs significantly (p = .0129) more pacemaker-like cells (Table 1; Fig. 3A, 3B, 3D, left panel) compared with control EBs. To corroborate this important finding with a different methodological approach, we performed double immunohistochemical stainings on slices of whole EBs with the cardiac pacemaker marker HCN4 and with cardiac α-actinin as well as nuclear counterstaining (Fig. 3C). These experiments yielded a significant (p = .0003) increase of the ratio of HCN4-positive cells versus the total number of nuclei in cardiac-α-actinin-positive areas of mutant EBs (0.25 ± 0.03 cells; n = 25 EBs) (supporting information data) compared with the ratio in control EBs (0.08 ± 0.03 cells; n = 35 EBs) (Fig. 3D, right panel). Higher numbers of pacemaker cells were also observed with anti-HCN4 3–3′-diaminobenzidene (DAB) staining (supporting information Fig. 2B).
Table 1. Cell type and AP morphology
Hormonal Regulation of Cardiomyocyte Function at the Single-Cell and Multicellular Level
The lack of BM and the accompanying changes of β1 integrin distribution prompted us to analyze the hormonal regulation in laminin γ1 (−/−) cardiomyocytes. This was determined by measuring the frequency of APs as well as the modulation of the ICa-L at the single-cell level using the patch-clamp technique. Carbachol (CCh) (1 μM) slowed spontaneous beating by 49.0% ± 0.50% (n = 14) in laminin γ1 (−/−) cardiomyocytes, similar to laminin γ1 (+/−) cardiomyocytes (48.5% ± 0.50%; n = 9) (Fig. 4A). The known ISO-induced increase of the AP frequency  was observed to a similar degree in both laminin (+/−) (61.1% ± 1.4%; n = 10) and laminin (−/−) (66.6% ± 1.0%; n = 14) cardiomyocytes. In addition, after ISO pretreatment led CCh to a reduction of the beating frequency in both laminin (+/−) (24.1% ± 0.5%; n = 10) and laminin (−/−) (24.0% ± 0.6%; n = 5) cardiomyocytes, respectively (supporting information Table 1). As a standard read-out for muscarinic and β-adrenergic signaling [15, 21, 31], we further investigated the regulatory effects of these hormones on ICa-L amplitude. This yields information concerning the physiological integrity of cardiomyocytes, since ICa-L is a key determinant of the excitation-contraction machinery. Our experiments showed that the CCh-mediated inhibition of ICa-L after prestimulation with ISO was present in laminin γ1 (−/−) (Fig. 4B) and laminin γ1 (+/−) (data not shown) cardiomyocytes. Statistical analysis revealed that the percentage of the inhibition of ICa-L by CCh was almost identical in heterozygous (−28.7% ± 4.8%) and homozygous (−24.5% ± 3.7%) cells. The ISO-induced stimulation was comparable in both groups, with 52.6% ± 13.6% (n = 8) and 44.1% ± 12.9% (n = 8) for heterozygous and homozygous cardiomyocytes, respectively. Thus, we found at the single-cell level that adrenergic and muscarinic regulation is preserved in laminin γ1 (−/−) ES cell-derived cardiomyocytes. Interestingly, a small but statistically significant (p = .05) upregulation of the density of ICa-L (picoampere/picofarad) was noticed in the laminin γ1 (−/−) cells (12.1 ± 2.2, n = 8, for (+/−) and 17.0 ± 5.8, n = 8, for (−/−), respectively) (Fig. 4C; suppporting information Table 2).
Next we used the MEA system to monitor the hormonal modulation in intact EBs without enzymatic treatment. For this purpose, cardiac clusters were plated and differentiated on MEAs, and the modulation of chronotropy was tested upon CCh (10 μM) and/or ISO (1 μM) application. Spontaneously occurring FPs showed a typical physiological response to CCh and ISO, similar to the single-cell AP measurements described above and in earlier work [16, 32]. The percentage of EBs that responded to CCh was 71.4% for the laminin γ1 (+/−) (n = 5) and 77.7% for the laminin γ1 (−/−) (n = 7) preparations. Moreover, laminin γ1 (−/−) EBs showed a negative chronotropic response to CCh, and this was reversed after wash-out. CCh reduced the FP frequency to 68% ± 10.5% (n = 5) in (+/−) EBs and to 77% ± 10.4% (n = 7) in (−/−) EBs. Similarly, muscarinic inhibition slowed after β-adrenergic prestimulation spontaneous activity to a similar degree in (+/−) and (−/−) EBs (supporting information Fig. 1; supporting information Table 3). Thus, the hormonal regulation of chronotropy of cardiac clusters was fully intact.
Electrical Signal Propagation in Beating Clusters
To understand the mechanism(s) whereby laminin disruption and consequent breakdown of the ECM network change conduction and crosstalk between cardiomyocytes, we performed a detailed analysis on EBs by measuring electrical impulse generation and electrical propagation between cardiomyocytes using MEA recordings. In Figure 5A, two original FP traces from different electrodes (26 and 73) show a representative example of the laminin γ1 (+/−) EBs (n = 22). The gray shaded areas on the left (Fig. 5A–5C) indicate the total area of the plated EBs, and this proved comparable in size between control and the mutant EBs. Interestingly, some of the laminin γ 1 (−/−) EBs (∼40%) did not adhere stably to MEAs; for MEA analysis, only the ones with good attachment providing adequate electrical signals could be used. In a representative control EB, a single pacemaker area was found in vicinity of electrode 26, and the signal was propagated in the direction of electrode 73, as indicated by the delay between the two electrodes (dotted line). Electrical coupling was proven by the identical frequency and the stable delay. On the contrary, isolated and competing pacemaker regions were found in the laminin γ1 (−/−) EBs (n = 14; Fig. 5B): the two representative traces show two electrically active areas around electrode 23 (areas 1 and 2), one serving as the pacemaker and one as the driven area as indicated by the stable delay. However, none of these two signals correlate with the FPs around electrode 87 (3), suggesting a second pacemaker area, which was electrically not coupled to the other pacemaker center. Moreover, we conducted functional rescue experiments by plating laminin γ1 (−/−) EBs pretreated with laminin 111 (25 μg/ml). In some (two of four) of the reconstituted EBs a recovery of the wild-type phenotype was evidenced by the presence of only one pacemaker region (region 1) demonstrated by the MEA recordings from electrodes 36 and 83 (Fig. 5C). The videos of these recordings reveal homogeneous contraction in the entire (+/−) rescued EB, whereas several independently beating areas were seen in the (−/−) EB (supporting information Videos 3-5). Our experiments demonstrated that significantly (p < .05) more pacemaker areas were found in laminin γ1 (−/−) (2.9; n = 14) EBs compared with controls (1.27; n = 22) EBs (Fig. 5D); pacemaker regions were calculated according to the number of independently beating areas of the EB. Next, we normalized this number with the size of the EBs by dividing the number of pacemaker regions by the number of electrodes being covered by the EBs. This analysis again yielded significantly more pacemaker regions per electrode in laminin γ1 EBs (0.16 ± 0.03; n = 14) compared with controls (0.06 ± 0.007; n = 22; p < .05) EBs. This finding is depicted in detail in Figure 5D (right panel), where the approximate path of the spreading of the electrical excitation was determined using FP delay analysis and is marked with a black line. In the laminin γ1 (+/−) EB (left) a functional syncytium can be identified, whereas in the laminin γ1 (−/−) EB (right) only some areas were electrically coupled (black circles and line, Fig. 5D left panel), which was proven by the fixed delay between the pacemaker-related electrodes (also described in Fig. 5A, 5B). In addition, a higher incidence of isolated pacemaker areas (black triangles, Fig. 5D left panel) was detected in the (−/−) EB, as indicated by changes of the delay between the respective electrodes.
We also found that the conduction of the signal between individual electrodes was not as homogeneous in the laminin γ1 (−/−) as in the laminin γ1 (+/−) EBs. The time course shown in Figure 5E depicts strong changes of the delay in the (−/−) EB (right panel), whereas the delay between respective electrodes remained stable in the (+/−) EB (left panel). This finding suggests a high presence of electrically isolated and/or partially coupled pacemaker areas in the laminin γ1 (−/−) EBs. In fact, the pacemaker regions proved to some extent isolated in the laminin γ1 (−/−) EBs, when using the FP delay analysis. In addition, the apparent conduction velocity in cardiomyocyte clusters of laminin γ1 (−/−) EBs was significantly (p = .014) lower (0.0139 ± 0.0022 m/s; n = 11) versus controls (0.025 ± 0.0019 m/s; n = 12). Slowing of the apparent conduction velocity in cardiomyocyte clusters of laminin γ1 (−/−) EBs could be due to changes of intracellular conduction, which is governed by Na+ and/or Ca2+ channels. These are reflected in the upstroke velocity of APs, and we therefore analyzed dV/dt max (Table 1). This revealed similar values for cardiomyocytes derived from mutant and control EBs. However, since the average maximal diastolic potential of our cells was found to be relatively depolarized and since this could result in voltage-dependent inactivation of Na+ channels, we determined dV/dt max in cardiomyocytes with relatively negative (less than −55 mV) MDP. This analysis yielded an average dV/dt max of 14.7 ± 4.4 V/second (n = 5; average MDP, −59 ± 4.4) and 13.5 ± 8.1 V/second (n = 5; average MDP, −58 ± 6.3) for control and mutant cardiomyocytes, respectively, ruling out prominent differences of Na+ currents. Furthermore, differences in K+ channel expression also appear unlikely because of the similar MDPs (Table 1).
Besides ion channels being responsible for intracellular conduction, connexins play a key role for intercellular conduction and spreading of the electrical signal. We have therefore also investigated the most relevant connexin isoforms in developing heart muscle cells. There was no obvious difference in the distribution of Cx43 in laminin γ1 (+/−) and (−/−) EBs (Fig. 6B, arrows); this analysis is somewhat difficult, as the anisotropic alignment of the cardiomyocytes is not preserved in the EBs. Nevertheless, our findings are also supported by the quantitation of connexins using Western blotting, which revealed no differences in expression for Cx43 (Fig. 6A, left). Similarly, the expression of Cx45 (Fig. 6A, right), the predominant isoform during early cardiac development , was also found to be unaltered, making it unlikely that changes in the expression and/or distribution of connexins underlie the observed alterations. Specific staining with BM4018 revealed that fibroblasts and reticular fibroblasts were similarly distributed in wild-type and knockout cardiomyocytes (data not shown), excluding the possibility that interdispersion of electrically isolating cells caused the observed conduction defect.
To understand whether lack of laminin can mechanistically account for the observed disturbances of the electrical signal propagation in the laminin γ1 (−/−) EBs, Van Gieson staining was performed. This demonstrated deposition of ECM between cells (−/−) in EBs (Fig. 6C), presumably due to the lack of an intact BM. This could lead to the isolation of electrically active areas, as these deposits correlated with sites of disturbed electrical spreading on representative (+/−) (left) and (−/−) (right) EBs (from Fig. 5D) plated on the MEAs (Fig. 6D). Since it was not possible to quantitate the matrix interdispersed between cardiomyocytes, we further underscored our observation by ultrastructural analysis. Electron microscopy depicted collagen-like extracellular structures (Fig. 6E, bigger panel), which caused spacing between adjacent laminin γ1 (−/−) cells. Such intercellular gaps prevent intercellular coupling via gap junctions despite normal expression of connexins. Furthermore, DAB staining confirmed enhanced extracellular deposits of matrix components, particularly of collagen VI in the knockout EBs (Fig. 6E, inset), and this finding was further corroborated by Western blotting analysis (Fig. 6F). This increased ECM deposition was also shown for other matrix proteins (supporting information Fig. 2). Thus, altered BM formation results in disorganized ECM deposition, leading to disturbed electrical pulse propagation.
The aim of our work was to investigate cardiac cells development and function upon deletion of the laminin γ1 chain. The biological relevance of laminins for development has been unequivocally evidenced by targeting experiments, as deletion of the laminin α1, β1, and γ1 [4, 34, 35] chains results in BM defects and early embryonic lethality. Since laminin plays a key role for skeletal muscle biology and since mutations in laminin genes result in prominent muscle pathologies in humans, we wondered about the cell biological and functional consequences on cardiac muscle development and differentiation upon deletion of the laminin γ1 chain. We show that this results in a complete absence of BM (collagen IV, nidogen) around cardiomyocytes, whereas cardiomyocyte development and differentiation remain surprisingly unaffected. The observed lack of BM formation is clearly due to the deficiency of laminin, since our rescue approach with soluble laminin led to a partial recovery of BM structure/formation around cardiomyocytes. Our data also demonstrate that the cytoskeleton of cardiomyocytes is in contrast to skeletal muscle [1, 2, 7], relatively independent of ECM structures. This is most likely due to compensatory effects of the laminin γ3 chain. This chain is broadly expressed in several tissues , has the same modular structure as the laminin γ1 chain, and was found to be more organized around laminin γ1 (−/−) cardiomyocytes, thereby forming a partial molecular substitute. We also observed a higher N-cadherin concentration at the border zones of laminin γ1 (−/−) cardiomyocytes, indicating its contribution to the maintenance of cytoskeletal anchorage and cell-cell contacts.
Laminin polymerization is essential, first, to induce reorganization of laminin itself into a BM, and then to form a structured network of the components of the cortical cytoskeleton, such as β1 integrin, vinculin, and talin. We showed in our earlier work that β1 integrin (−/−) cardiomyocytes did only differentiate into primitive cardiac muscle cells  and that muscarinic signaling was entirely absent because of spatial displacement of Gαi . However, although the lack of laminin affected the expression and the distribution of β1 integrins, the function of laminin γ1 (−/−) cardiomyocytes at the single-cell level and in whole EBs was preserved [20, 36]. This suggests that the presence of β1 integrins, even if not entirely correctly distributed, suffices to preserve an orderly spatial organization of the receptor-dependent actinin-talin clustering, presumably thereby maintaining the cortical cytoskeleton in cardiomyocytes.
Alterations in certain laminin genes result in muscular pathologies but also in severe changes of heart function (i.e., disturbances of heart rate variability), impaired conduction resulting in arrhythmias, and left ventricular failure accompanied by progressive accumulation of connective tissue [37–39]. In line with these observations our MEA results show that the electrical signal is not appropriately propagated in laminin γ1 (−/−) EBs, and this is supported by the reduced apparent conduction velocity within cardiomyocyte clusters. There are two mechanisms potentially responsible for this disturbed conductance, intracellular and intercellular. We could not detect a significant lowering of Na+ (dV/dt max measurements), ICa-L, and/or K+ (MDP) channels, thereby ruling out the possibility that changes of the intracellular conduction caused the observed slowed conduction in the laminin γ1 chain (−/−) cardiomyocyte clusters. The intercellular conduction is mediated by connexins, and we therefore investigated cardiac connexins with Western blotting and immunostaining. The experiments yielded no apparent differences in their expression and localization. Furthermore, we could not find an increase in potentially electrically isolating cells interspersed between the cardiomyocytes, a finding supported by the relatively little reduction in apparent conduction velocity in the mutant cardiomyocyte clusters. In fact, in the case of cardiomyocyte-fibroblast coupling the apparent conduction velocity would strongly drop . Rather, we found deposits of ECM between adjacent cells, which presumably caused altered spacing between the cells (also described in Fig. 6E, right) resulting in defective intercellular coupling via gap junctions despite their normal expression pattern. Furthermore, we found in single-cell experiments and in EBs more pacemaker-like cardiomyocytes and ectopic centers, respectively. This suggests that the maintenance of spontaneously active, pacemaker-like cells is enhanced in electrically silent and/or uncoupled areas and/or that there is a compensatory mechanism to overcome the uncontrolled ECM deposition and the alteration of the propagation of electrical excitation. We propose that this mechanism underlies the increased number of pacemaker-like cells rather than the lack of laminin γ1 expression directly modulating the subtype differentiation of early cardiomyocytes.
The clearly higher deposits of several ECM proteins (collagen I and VI and fibronectin, among others) in laminin γ1 (−/−) EBs because of disturbed BM formation (downregulation of collagen IV, nidogen-1, and β1 integrins), fit nicely into the pathophysiological sequelae of cardiomyopathies, where remodeling of the heart is followed by deposition of ECM molecules or, eventually, calcifications as well. Our data demonstrate that such deposits can hinder normal spreading of the electrical signal and consequently result in the in vivo situation, with a propensity toward potentially life-threatening ventricular arrhythmias [41–43].
Our study demonstrates that the differentiation and function of laminin γ1 (−/−) cardiomyocytes are intact despite the complete absence of BM. However, disruption of normal BM formation and the resulting deposits of ECM hinder physiological spreading of the electrical propagation wave, leading to slowing of the apparent conduction velocity and an increase of the number of pacemaker areas. Thus, even though laminin is not involved directly in heart muscle development and cardiomyocyte signaling, it plays a key role in the organization of the ECM and hence in the orderly propagation of electrical signals.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (BL 419/2-2 and SM 65/1-3 within Priority Program 1086 (to B.K.F., W.B., and N.S.). R.D. was supported by a grant (SFB 645, B2) of the Deutsche Forschungsgemeinschaft to K. Willecke. Parts of this study have been reported as abstracts at the annual meeting of the German Physiological Society. We thank Dr P. Sasse for helpful and critical discussion.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.