Gap Junctional Coupling with Cardiomyocytes is Necessary but Not Sufficient for Cardiomyogenic Differentiation of Cocultured Human Mesenchymal Stem Cells§

Authors

  • Arti A. Ramkisoensing,

    1. Department of Cardiology,Laboratory of Experimental Cardiology, Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
  • DaniëL A. Pijnappels,

    1. Department of Cardiology,Laboratory of Experimental Cardiology, Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
  • Jim Swildens,

    1. Department of Molecular Cell Biology,Laboratory of Experimental Cardiology, Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
  • Marie José Goumans,

    1. Department of Molecular Cell Biology,Laboratory of Experimental Cardiology, Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
  • Willem E. Fibbe,

    1. Laboratory of Experimental Cardiology, Department of Hematology, Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
  • Martin J. Schalij,

    1. Department of Cardiology,Laboratory of Experimental Cardiology, Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
  • Antoine A.F. de Vries,

    Corresponding author
    1. Department of Cardiology,Laboratory of Experimental Cardiology, Leiden University Medical Center, Leiden, The Netherlands
    2. Department of Molecular Cell Biology,Laboratory of Experimental Cardiology, Leiden University Medical Center, Leiden, The Netherlands
    • Department of Cardiology, Leiden University Medical Center, Leiden 2300 RC, The Netherlands
    Search for more papers by this author
    • Telephone: +31-715262020; Fax: +31-715266809

    • contributed equally to this article.

  • Douwe E. Atsma

    Corresponding author
    1. Department of Cardiology,Laboratory of Experimental Cardiology, Leiden University Medical Center, Leiden, The Netherlands
    • Department of Cardiology, Leiden University Medical Center, Leiden 2300 RC, The Netherlands
    Search for more papers by this author
    • contributed equally to this article.

    • Telephone: +31-715262020; Fax: +31-715266809


  • Author Contributions: A.A.R. and D.A.P.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; J.S.: provision of study material, data analysis and interpretation, manuscript writing, and final approval of manuscript; M.J.G. and W.E.F.: provision of study material, administrative support, financial support, and final approval of manuscript; M.J.S.: financial support, data analysis and interpretation, manuscript writing, and final approval of manuscript; A.A.F.V.: conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript; D.E.A.: conception and design, financial support, data analysis and interpretation, manuscript writing, and final approval of manuscript.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLSEXPRESS Month 00, 2012.

Abstract

Gap junctional coupling is important for functional integration of transplanted cells with host myocardium. However, the role of gap junctions in cardiomyogenic differentiation of transplanted cells has not been directly investigated. The objective of this work is to study the role of connexin43 (Cx43) in cardiomyogenic differentiation of human mesenchymal stem cells (hMSCs). Knockdown of Cx43 gene expression (Cx43↓) was established in naturally Cx43-rich fetal amniotic membrane (AM) hMSCs, while Cx43 was overexpressed (Cx43↑) in inherently Cx43-poor adult adipose tissue (AT) hMSCs. The hMSCs were exposed to cardiomyogenic stimuli by coincubation with neonatal rat ventricular cardiomyocytes (nrCMCs) for 10 days. Differentiation was assessed by immunostaining and whole-cell current clamping. To establish whether the effects of Cx43 knockdown could be rescued, Cx45 was overexpressed in Cx43↓ fetal AM hMSCs. Ten days after coincubation, not a single Cx43↓ fetal AM hMSC, control adult AT MSC, or Cx43↑ adult AT mesenchymal stem cell (MSC) expressed α-actinin, while control fetal AM hMSCs did (2.2% ± 0.4%, n = 5,000). Moreover, functional cardiomyogenic differentiation, based on action potential recordings, occurred only in control fetal AM hMSCs. Of interest, Cx45 overexpression in Cx43↓ fetal AM hMSCs restored their ability to undergo cardiomyogenesis (1.6% ± 0.4%, n = 2,500) in coculture with nrCMCs. Gap junctional coupling is required for differentiation of fetal AM hMSCs into functional CMCs after coincubation with nrCMCs. Heterocellular gap junctional coupling thus plays an important role in the transfer of cardiomyogenic signals from nrCMCs to fetal hMSCs but is not sufficient to induce cardiomyogenic differentiation in adult AT hMSCs. STEM CELLS2012;30:1236–1245

INTRODUCTION

Gap junctional coupling is essential in establishing electrochemical communication between cardiomyocytes (CMCs) [1]. Such coupling of cytoplasmic compartments is mediated by gap junctions, consisting of two hexameric assemblies of connexin (Cx) proteins embedded in the plasma membranes of neighboring cells thereby forming so-called hemichannels or connexons. Gap junctions permit the bidirectional passage of small molecules and ions between cells and play an important role in the regulation of both physiological and pathophysiological processes.

During embryonic development, spreading of signals across tissues through gap junctions contribute to the migration and specialization of cells [2]. In the developing heart, connexin40 (Cx40), connexin43 (Cx43), and/or connexin45 (Cx45) deficiency result in serious cardiac malformation [3–7]. Besides cardiac development, gap junctional communication also affects the therapeutic and hazardous potential of cardiac cell therapy [8–10]. However, the role of gap junctional coupling in cardiomyogenic differentiation of stem cells remains unclear, although gap junction-mediated processes, such as spread of microRNAs (miRs) and hyperpolarization, have been implicated in the induction of cardiomyogenesis [11, 12].

Mesenchymal stem cells (MSCs) are a population of mononuclear stromal cells that can be harvested from a wide variety of tissues, are easily expandable in vitro, have immunomodulatory properties, secrete paracrine factors that stimulate tissue regeneration, and can differentiate into various types of mesodermal and nonmesodermal cells [13, 14]. In addition, these cells were shown to improve cardiac function upon transplantation in diseased rodent, pig, and human hearts [15]. However, to what extent MSCs can undergo cardiac differentiation and which factors are involved in this process are still unclear [16–18]. Although donor age and heterocellular interactions were found to affect the cardiomyogenic differentiation potential of MSCs [19], more research is needed into their biological properties and the modification thereof to improve the therapeutic potential of MSC-based cardiac cell therapy.

Accordingly, in this study, we specifically investigated the role of Cx43, the major gap junction protein of the working myocardium, in cardiomyogenic differentiation of naturally Cx43-rich fetal and Cx43-poor adult human MSCs (hMSCs) [19] in cocultures with neonatal rat ventricular CMCs (nrCMCs). To this end, Cx43 expression levels in hMSCs were either downregulated by short hairpin RNA (shRNA)-mediated RNA interference (RNAi) or upregulated by recombinant human Cx43 (hCx43) gene delivery. To control for possible off-target effects of hCx43-specific shRNAs, an RNAi rescue experiment based on the forced expression of human Cx45 (hCx45) in Cx43 knockdown cells was included. The consequences of these genetic interventions on the cardiomyogenic differentiation capacity of the hMSCs were assessed using immunocytological and patch-clamp analyses.

This study shows that gap junctions are directly involved in the transfer of cardiomyogenic signals from nrCMCs to fetal hMSCs. Downregulation of Cx43 gene expression in these hMSCs prevented their cardiomyogenic differentiation in cocultures with nrCMCs. Overexpression of hCx45 in the Cx43 knockdown cells, on the other hand, restored their ability to respond to cardiomyogenic stimuli provided by neighboring nrCMCs.

MATERIALS AND METHODS

The role of gap junctional coupling in the transfer of cardiomyogenic signals from nrCMCs to hMSCs was investigated by suppressing Cx43 expression in naturally Cx43-rich fetal amniotic membrane (AM) hMSCs using two lentiviral vectors (LVs) encoding different hCx43-specific shRNAs (Cx43↓(1) fetal AM hMSCs and Cx43↓(2) fetal AM hMSCs) and by LV-mediated overexpression of hCx43 in intrinsically Cx43-poor adult adipose tissue (AT) hMSCs (Cx43↑ adult AT hMSCs). To facilitate the identification of genetically modified cells, the LVs also directed the synthesis of the green fluorescent protein (GFP) of Renilla reniformis (hrGFP). Fetal AM hMSCs transduced with an LV encoding an shRNA directed against the Aequorea victoria enhanced GFP (eGFP) gene and adult AT hMSCs transduced with an LV coding for hrGFP alone served as control cells (control fetal AM hMSCs and control adult AT hMSCs, respectively). For an optimized analysis of the adult AT hMSCs, these cells were transduced in a separate experiment with an LV-encoding Cx43 and puromycin N-acetyltransferase. In this experiment, adult AT hMSCs transduced with an LV coding for puromycin N-acetyltransferase alone served as a negative control. Immunocytological analysis, quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR), and Western blot analysis were applied to determine Cx43 expression levels, while dye transfer assays and whole-cell current-clamp recordings were used to assess the level of functional coupling. To induce cardiomyogenic differentiation, hMSCs were coincubated with nrCMCs for 10 days. Differentiation was assessed by immunocytological staining and whole-cell current clamping. LV-encoded hCx45 was overexpressed in Cx43↓(1) fetal AM hMSCs (Cx43↓(1) + Cx45↑ fetal AM hMSCs) to establish whether the effects of Cx43 knockdown could be reversed using the techniques described above. Cx43↓(1) fetal AM hMSCs transduced with an LV-encoding eGFP were used as control cells (Cx43↓(1) + eGFP fetal AM hMSCs) in these experiments. For a detailed description of Materials and Methods, the reader is referred to Supporting Information and previous papers of our research group [19–21].

RESULTS

Characterization of Fetal AM hMSCs and Adult AT hMSCs

Both fetal AM hMSCs and adult AT hMSCs were characterized according to established criteria. To this purpose, their cell surface marker profile and adipogenic and osteogenic differentiation capacity were studied. Both types of hMSCs were negative for CD31 (endothelial cell marker), CD34, and CD45 (both hematopoietic cell markers), whereas they were positive for CD73, CD90, and CD105 (MSC markers) (Fig. 1A--1C). In vitro differentiation assays showed that the two types of hMSCs were able to differentiate into adipocytes (Fig. 1D1, 1D2) and osteoblasts (Fig. 1E1, 1E2) confirming their multipotency.

Figure 1.

Characterization of fetal AM hMSCs and adult AT hMSCs. (A, B): Flow cytometric analyses showed for both types of hMSCs abundant surface expression of the MSC markers CD90, CD105, and CD73 and hardly any surface expression of the hematopoietic cell markers CD34 and CD45 or the endothelial cell marker CD31 (black lines). Isotype-matched control antibodies were included to determine background fluorescence levels (gray lines). Percentages are means of ≥ 4 measurements. (C): Only minor differences in the expression levels of the surface marker proteins were present between fetal AM hMSCs and adult AT hMSCs. (D1, D2): Adipogenic differentiation was visualized by the presence of Oil Red O-stained fat vacuoles. (E1, E2): Calcium depositions after osteogenic differentiation were visualized by Alizarin Red S staining. Abbreviations: AM, amniotic membrane; AT, adipose tissue; hMSCs, human mesenchymal stem cells; MSC, mesenchymal stem cell; ND, not detected.

Evaluation of Cx43 Expression Levels

The impact of the different genetic interventions on the Cx43 expression levels in fetal AM hMSCs and adult AT hMSCs was investigated by three different methods. Immunocytological analysis showed that Cx43↑ adult AT hMSCs (Fig. 2A2) contained higher levels of Cx43 than control adult AT hMSCs (Fig. 2A1), while the protein was also more abundant in control fetal AM hMSCs (Fig. 2B1) than in Cx43↓(1) fetal AM hMSCs (Fig. 2B2) or in Cx43↓(2) fetal AM hMSCs (Fig. 2B3) (for each genetic modification, n = 4 isolates of each hMSC type were assessed). These results were validated by Western blot analysis (n = 4 hMSC isolates per experimental group), which showed that Cx43↑ adult AT hMSCs contained 1,798% ± 146% more Cx43 protein than control adult AT hMSCs (p < .001) (Fig. 2C1, 2C3). Western blot analysis also confirmed the presence of significantly less Cx43 in Cx43↓(1) fetal AM hMSCs (79.4% ± 3.0% reduction) and in Cx43↓(2) fetal AM hMSCs (77.4% ± 2.6% reduction) than in control fetal AM hMSCs (p < .001) (Fig. 2C2, 2C4). In agreement with these results, qRT-PCR revealed that Cx43↑ adult AT hMSCs contained 9.77 ± 0.63-fold (p < .001) (Fig. 2D1) more Cx43 transcripts than control adult AT hMSCs, while the amount of Cx43 RNA in Cx43↓(1) fetal AM hMSCs and in Cx43↓(2) fetal AM hMSCs was, respectively, 3.67- ± 0.14-fold and 3.94- ± 0.25-fold (p < .001) lower than in control fetal AM hMSCs (Fig. 2D2).

Figure 2.

Analysis by immunofluorescence microscopy, Western blotting, and qRT-PCR of Cx43 expression in monocultures of genetically modified fetal AM hMSCs and adult AT hMSCs. (A1, A2): Immunocytological analyses revealed low Cx43 levels in control adult AT hMSCs, while high levels of Cx43 were detected in Cx43↑ adult AT hMSCs. (B1): High Cx43 levels were present in control fetal AM hMSCs. (B2, B3): After knockdown of Cx43 gene expression, low Cx43 levels were detected in Cx43↓(1) fetal AM hMSCs and in Cx43↓(2) fetal AM hMSCs. Nuclei were stained with the DNA-binding fluorochrome Hoechst 33342. (C1, C2): Pictures of representative Western blots showing that Cx43↑ adult AT hMSCs have large amounts of Cx43 in contrast to control adult AT hMSCs and that control fetal AM hMSCs contain large amounts of Cx43 in contrast to Cx43↓(1) fetal AM hMSCs and Cx43↓(2) fetal AM hMSCs. The glyceraldehyde 3-phosphate dehydrogenase (GAPDH)-specific immunostaining was included for normalization purposes. (C3, C4): Bar graphs of the quantification by Western blotting of Cx43 levels in the different populations of hMSCs corrected for differences in GAPDH expression. *, p < .001 versus Cx43↑ adult AT hMSCs or versus Cx43↓(1) fetal AM hMSCs and Cx43↓(2) fetal AM hMSCs. (D1, D2): Bar graphs of the assessment by qRT-PCR of Cx43 mRNA levels in control adult AT hMSCs, Cx43↑ adult AT hMSCs and in control fetal AM hMSCs, Cx43↓(1) fetal AM hMSCs, and Cx43↓(2) fetal AM hMSCs. *, p < .001 versus control adult AT hMSCs or versus Cx43↓(1) fetal AM hMSCs and Cx43↓(2) fetal AM hMSCs. Abbreviations: AM, amniotic membrane; AT, adipose tissue; Cx43, connexin43; hMSCs, human mesenchymal stem cells; nrCMCs, neonatal rat ventricular cardiomyocytes; qRT-PCR, quantitative reverse transcription-polymerase chain reaction.

Assessment of Functional Heterocellular Gap Junctional Coupling Via Cx43

The effects of modulating Cx43 expression levels in fetal AM hMSCs and in adult AT hMSCs on their functional heterocellular coupling with nrCMCs were investigated by dye transfer experiments using the gap junction-permeable fluorochrome calcein red-orange AM (calcein). Cx43↑ adult AT hMSCs showed a significantly higher dye intensity than control adult AT hMSCs (67.6% ± 4.7% vs. 8.4% ± 1.6% of that in adjacent nrCMCs) (p < .001) (Fig. 3A1, 3A2, 3C1). Furthermore, the fraction of hMSCs that took up calcein from neighboring nrCMCs was much higher for the Cx43↑ adult AT hMSCs than for the control adult AT hMSCs (71.6% ± 7.6% vs. 28.3% ± 4.6% of cells) (p < .001) (Fig. 3C3). The relative dye intensity in control fetal AM hMSCs (63.2% ± 3.8%) was higher than in Cx43↓(1) fetal AM hMSCs (5.9% ± 1.4%) and in Cx43↓(2) fetal AM hMSCs (6.0% ± 1.9%) (p < .001) (Fig. 3B1--3B3, 3C2). Moreover, significantly more of the control fetal AM hMSCs (75.9% ± 9.2%) had taken up calcein than of the Cx43↓(1) fetal AM hMSCs (13.2% ± 2.4%) or of the Cx43↓(2) fetal AM hMSCs (13.0% ± 2.1%) (p < .001) (Fig. 3C4).

Figure 3.

Study of the influence of Cx43 levels in hMSCs on their functional coupling with nrCMCs. (A1, A2): Calcein transfer from nrCMCs to Cx43↑ adult AT hMSCs was much more efficient than to control adult AT hMSCs. (B1--B3): Dye transfer from nrCMCs to fetal AM hMSCs was strongly inhibited by shRNA-mediated downregulation of Cx43 gene expression in the hMSCs. (C1, C2): Bar graphs of the quantification of dye intensity in GFP-positive hMSCs. Dye intensity in the GFP-positive hMSCs was expressed as percentage of the dye intensity in the surrounding nrCMCs. (C3, C4): Bar graphs of the assessment of the percentage of GFP-positive hMSCs that had taken up calcein from neighboring nrCMCs. *, p < .001 versus control adult AT hMSC or versus Cx43↓(1) fetal AM hMSCs and Cx43↓(2) fetal AM hMSCs. (D): Average MDPs in hMSC monocultures (indicated as MSC) as measured by whole-cell current clamping were similar for the different experimental groups. However, the average MDP in hMSCs became more negative in the presence of adjoining nrCMCs (indicated as MSC-CMC), with Cx43↑ adult AT hMSCs and control fetal AM hMSCs reaching the most negative values. Abbreviations: AM, amniotic membrane; AT, adipose tissue; Cx43, connexin43; GFP, green fluorescent protein; hMSCs, human mesenchymal stem cells; MDPs, maximal diastolic membrane potentials; nrCMCs, neonatal rat ventricular cardiomyocytes; shRNA, short hairpin RNA.

To further assess functional coupling between hMSCs and nrCMCs, maximal diastolic membrane potentials (MDPs) were measured in each group of hMSCs in the absence and presence of adjacent nrCMCs. Similar average MDPs were found in all hMSC monocultures (n = 6) (Fig. 3D, black bars). These values were much less negative than the average MDP of nrCMCs (−67.9 mV) (n = 10). However, the MDPs of hMSCs became more negative when they were surrounded by nrCMCs, although the degree of hyperpolarization differed between the various groups (n = 8) (Fig. 3D, white bars). Cx43↑ adult AT hMSCs and control fetal AM hMSCs showed the most negative MDPs (−38 ± 4 mV and −43 ± 5 mV, respectively; p = .68) in the presence of nrCMCs. On the other hand, control adult AT hMSCs, Cx43↓(1) fetal AM hMSCs, and Cx43↓(2) fetal AM hMSCs, which all contain low levels of Cx43, showed less negative MDPs.

Assessment of Cardiomyogenic Differentiation Ability of hMSCs with Genetically Altered Cx43 Levels

Human-Specific Immunocytological Evaluation

At day 10 of coculture with nrCMCs, 2.7% ± 0.4% GFP/human lamin A/C-double-positive control fetal AM hMSCs were positive for the sarcomeric protein α-actinin with a cross-striated staining pattern like that of the nrCMCs (Fig. 4A1, 4A2, 4C). However, GFP/human lamin A/C-double-positive Cx43↓(1) fetal AM hMSCs and Cx43↓(2) fetal AM hMSCs did not contain detectable amounts of α-actinin (Fig. 4A3, 4A4, 4C). Furthermore, neither Cx43↑ adult AT hMSCs nor control adult AT hMSCs stained positive for α-actinin (Fig. 4B1--4B3, 4C). These results were confirmed with Cx43↑ adult AT hMSCs and control adult AT hMSCs that were transduced with LVs encoding puromycin N-acetyltransferase instead of GFP (Supporting Information Fig. S1A, S1B). Also, only in cocultures of control fetal AM hMSCs with nrCMCs, some GFP-positive cells were detected to be positive for the cardiac transcription factors Nkx2.5 and GATA4 (Supporting Information Figs. S2A1, S3A1, respectively). In the other hMSC groups, all GFP-positive cells were negative for these cardiac transcription factors (Supporting Information Figs. S2A2, S2B2, S3A2, S3B2).

Figure 4.

Immunocytological assessment of cardiomyogenic differentiation of genetically modified adult AT hMSCs and fetal AM hMSCs after 10 days of coculture with nrCMCs. (A1): Upon coculture with nrCMCs, a fraction of the GFP- and human lamin A/C-double-positive control fetal AM hMSCs became positive for sarcomeric α-actinin (indicated as Act). (A2): Intense punctate Cx43 (indicated as Cx) immunostaining of the interfaces between a control fetal AM hMSC and two adjacent nrCMCs (white arrows). (A3, A4, B1, B2): GFP-labeled Cx43↓(1) fetal AM hMSCs, Cx43↓(2) fetal AM hMSCs, Cx43↑ adult AT hMSCs, and control adult AT hMSCs in coculture with nrCMCs did not stain positive for α-actinin. (B3): Presence of Cx43 plaques at the interfaces between Cx43↑ adult AT hMSCs and two bordering nrCMCs (white arrows). (C): Quantitative analysis of the cardiomyogenic differentiation ability of the genetically modified hMSCs using immunopositivity for sarcomeric α-actinin as read out. The graph is based on a minimum of 5,000 cells analyzed from four separate hMSC isolates per experimental group. Abbreviations: AM, amniotic membrane; AT, adipose tissue; Cx43, connexin43; GFP, green fluorescent protein; hMSCs, human mesenchymal stem cells; ND, not detected; nrCMCs, neonatal rat ventricular cardiomyocytes.

hMSC-Specific Intracellular Electrophysiological Measurements

To assess cardiomyogenic differentiation at a functional level, the ability of GFP-positive hMSCs to generate spontaneous action potentials was studied after gap junction uncoupling (Fig. 5A, 5B). A fraction of the control fetal AM hMSCs showed spontaneous action potentials (n = 6), with MDPs similar to those of native CMCs (n = 9) (Fig. 5B, 5C). In contrast, both Cx43↑ adult AT hMSCs and control adult AT hMSCs (n = 7) showed more depolarized MDPs of −15 ± 4 mV and −15 ± 3 mV, respectively. Also, Cx43↑ adult AT hMSCs stayed inexcitable (n = 5). Importantly, knockdown of Cx43 in fetal AM hMSCs rendered these cells incapable of generating spontaneous action potentials after 10 days of coculture with nrCMCs (n = 9 for both shRNAs) and left them with an MDP comparable to those of adult AT hMSCs (−15 mV vs. −12 and −14 mV for Cx43↓(1) fetal AM hMSCs and Cx43↓(2) fetal AM hMSCs, respectively).

Figure 5.

Electrophysiological assessment, after gap junctional uncoupling, of cardiomyogenic differentiation of hMSCs. (A1, A2): Bright-field (A1) and fluorescence (A2) image of a GFP-positive hMSC with adjacent nrCMCs and patch-clamp electrode. (B): Current-clamp recordings in nrCMCs and in GFP-positive hMSCs from the different experimental groups. Action potentials were measured in nrCMCs and in some control fetal AM hMSCs, while the other cells displayed stable membrane potentials. (C): Average MDPs of nrCMCs and in the different groups of GFP-positive hMSCs. For the control fetal AM hMSCs, only the average MDP of cells showing action potentials is presented. Abbreviations: AM, amniotic membrane; AT, adipose tissue; Cx43, connexin43; GFP, green fluorescent protein; hMSCs, human mesenchymal stem cells; MDPs, maximal diastolic membrane potentials; nrCMCs, neonatal rat ventricular cardiomyocytes.

Evaluation of Cx45 Expression Levels

To check whether the inability of Cx43↓ fetal AM hMSCs to undergo cardiomyogenic differentiation in cocultures with nrCMCs was not caused by some off-target effect(s) of the hCx43-specific shRNAs, a rescue experiment was carried out. To this end, the Cx43↓(1) fetal AM hMSCs were transduced with an LV encoding GFP and hCx45 or with a control LV directing the synthesis of eGFP only. Immunocytological analysis of the resulting cell populations showed that Cx45 was much more abundant in Cx43↓(1) + Cx45↑ fetal AM hMSCs than in Cx43↓(1) + eGFP fetal AM hMSCs (Fig. 6A1, 6A2). Western blot analysis confirmed these results by showing 1,163% ± 57% higher levels of Cx45 in Cx43↓(1) + Cx45↑ fetal AM hMSCs than in Cx43↓(1) + eGFP fetal AM hMSCs (n = 4 for both sample types) (p < .001) (Fig. 6B1, 6B2). Consistently, qRT-PCR showed that Cx45 transcript levels were 12.5- ± 1.9-fold higher in Cx43↓(1) + Cx45↑ fetal AM hMSCs than in Cx43↓(1) + eGFP fetal AM hMSCs (p < .01) (Fig. 6C).

Figure 6.

Assessment of Cx45 expression and functionality. (A1, A2): Immunocytological analyses showed abundant Cx45 expression in monocultures of Cx43↓(1) + Cx45↑ fetal AM hMSCs but not of Cx43↓(1) + eGFP fetal AM hMSCs. A part of the Cx45 signal was detected at the interfaces between Cx43↓(1) + Cx45↑ fetal AM hMSCs. Nuclei were stained with Hoechst 33342. (B1): Representative picture of a Western blot confirming high-level expression of Cx45 in Cx43↓(1) + Cx45↑ fetal AM hMSCs and the presence of very low amounts of Cx45 in Cx43↓(1) + eGFP fetal AM hMSCs. The GAPDH-specific immunostaining was included for normalization purposes. (B2): Bar graph of the assessment by Western blotting of Cx45 levels in Cx43↓(1) + Cx45↑ fetal AM hMSCs and in Cx43↓(1) + eGFP fetal AM hMSCs corrected for differences in GAPDH expression. *, p < .001 versus Cx43↓(1) + eGFP fetal AM hMSCs. (C): Bar graph of the quantification by (qRT-PCR) of Cx45 mRNA levels in Cx43↓(1) + Cx45↑ fetal AM hMSCs and in Cx43↓(1) + eGFP fetal AM hMSCs. *, p < .01 versus Cx43↓(1) + eGFP fetal AM hMSCs. (D1, D2): Cx43↓(1) + Cx45↑ fetal AM hMSCs much more readily take up calcein from adjacent nrCMCs than Cx43↓(1) + eGFP fetal AM hMSCs. (E1): Quantitative analysis of the dye intensity in Cx43↓(1) + Cx45↑ fetal AM hMSCs and in Cx43↓(1) + eGFP fetal AM hMSCs. Dye intensity in GFP-positive hMSCs was expressed as percentage of the dye intensity in the surrounding nrCMCs. (E2): Bar graph of the assessment of the percentage GFP-positive hMSCs that had taken up calcein from neighboring nrCMCs. *, p < .001 versus Cx43↓(1) + eGFP fetal AM hMSCs. (F): The average MDPs of Cx43↓(1) + Cx45↑ fetal AM hMSCs and of Cx43↓(1) + eGFP fetal AM hMSCs in monocultures are less negative than in cocultures with nrCMCs. The effect of adjacent nrCMCs on the average MDP was bigger for fetal AM hMSCs in which the knockdown of Cx43 expression was compensated by Cx45 overexpression than for Cx43↓(1) + eGFP fetal AM hMSCs. Abbreviations: AM, amniotic membrane; Cx43, connexin43; eGFP, enhanced green fluorescent protein; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; hMSCs, human mesenchymal stem cells; MDPs, maximal diastolic membrane potentials; nrCMCs, neonatal rat ventricular cardiomyocytes; qRT-PCR, quantitative reverse transcription-polymerase chain reaction.

Assessment of Functional Heterocellular Gap Junctional Coupling Via Cx45

The extent of functional heterocellular coupling between nrCMCs and Cx43↓(1) + Cx45↑ fetal AM hMSCs or Cx43↓(1) + eGFP fetal AM hMSCs was determined in dye transfer assays. The dye intensity relative to that in neighboring nrCMCs was much higher in Cx43↓(1) + Cx45↑ fetal AM hMSCs (92.3% ± 3.2%) than in Cx43↓(1) + eGFP fetal AM hMSCs (6.0% ± 0.8%) (p < .001) (Fig. 6D1, 6D2, 6E1). Also, the proportion of hMSCs that took up calcein from adjacent nrCMCs was approximately fourfold higher for the Cx43↓(1) + Cx45↑ fetal AM hMSCs than for the Cx43↓(1) + eGFP fetal AM hMSCs (66.7% ± 7.6% vs. 16.0% ± 2.3%) (p < .001) (Fig. 6E2).

To study the impact of Cx45 overexpression on the electrophysiological properties of Cx43↓ fetal AM hMSCs, whole-cell current-clamp measurements were performed. Cx45 overexpression did not significantly alter the average MDP of Cx43↓(1) fetal AM hMSCs in monoculture (n = 6) (Fig. 6F). However, in coculture with nrCMCs, Cx43↓(1) + Cx45↑ fetal AM hMSCs (n = 6) showed more negative MDPs as compared to Cx43↓(1) + eGFP fetal AM hMSCs (−43 ± 5 mV and −25 ± 5 mV, respectively) (Fig. 6F), indicating that electrical coupling of Cx43↓ fetal AM hMSCs with neighboring nrCMCs was enhanced by Cx45 overexpression.

Assessment of Cardiomyogenic Differentiation Ability of Fetal AM hMSCs After Rescue of Cx43 Knockdown by Cx45 Overexpression

Human-Specific Immunocytological Evaluation

Ten days after coincubation with nrCMCs, Cx43↓(1) + Cx45↑ fetal AM hMSCs and Cx43↓(1) + eGFP fetal AM hMSCs were analyzed for α-actinin positivity. While Cx43↓(1) + eGFP fetal AM hMSCs did not contain detectable amounts of α-actinin, 1.6% ± 0.4% (n = 2,500) GFP/human lamin A/C-double-positive Cx43↓(1) + Cx45↑ fetal AM hMSCs stained positive for this sarcomeric protein indicating that these human cells had differentiated into CMCs (Fig. 7A1, 7A2, 7B). In the cocultures established with Cx43↓(1) + Cx45↑ fetal AM hMSCs, Cx45-positive gap junctional plaques were visible at the interfaces with neighboring nrCMCs (Fig. 7A3), which was not the case in those containing Cx43↓(1) + eGFP fetal AM hMSCs. Furthermore, only in cocultures of Cx43↓(1) + Cx45↑ fetal AM hMSCs with nrCMCs, GFP-labeled cells were detected that were positive for the cardiac transcription factors Nkx2.5 and GATA4 (Supporting Information Fig. S4A1, S4B1). Cx43↓(1) + eGFP fetal AM hMSCs in coculture with nrCMCs did not stain positive for these transcription factors (Supporting Information Fig. S4A2, S4B2).

Figure 7.

Investigation by immunocytological and patch-clamp analysis of cardiomyogenic differentiation of fetal AM hMSCs in coculture with nrCMCs after rescue of Cx43 knockdown by Cx45 overexpression. In the presence of nrCMCs, a fraction of the GFP- and human lamin A/C-double-positive Cx43↓(1) + Cx45↑ fetal AM hMSCs (A1) expressed α-actinin (indicated as Act) in a striated pattern typical of sarcomeric proteins, while Cx43↓(1) + eGFP fetal AM hMSCs (A2) did not. (A3): Punctate Cx45 immunostaining of the interface between a Cx43↓(1) + Cx45↑ fetal AM hMSC and an adjacent nrCMC (white arrows). (B): Quantitative analysis of the cardiomyogenic differentiation capacity of Cx43↓(1) + Cx45↑ fetal AM hMSCs and of Cx43↓(1) + eGFP fetal AM hMSCs using immunopositivity for sarcomeric α-actinin as read out. The graph is based on a minimum of 5,000 cells analyzed from four separate hMSC isolates per experimental group. (C1): Current-clamp recordings in pharmacologically uncoupled cocultures of nrCMCs and Cx43↓(1) + eGFP fetal AM hMSCs or Cx43↓(1) + Cx45↑ fetal AM hMSCs. Fetal AM hMSCs that had lost the capacity to produce action potentials due to knockdown of Cx43 expression regained this ability following forced Cx45 expression but not after eGFP overexpression. (C2): Average MDPs of nrCMCs, Cx43↓(1) + eGFP fetal AM hMSCs, and Cx43↓(1) + Cx45↑ fetal AM hMSCs. For the Cx43↓(1) + Cx45↑ fetal AM hMSCs, only the average MDP of cells showing action potentials is presented. Abbreviations: AM, amniotic membrane; Cx43, connexin43; Cx45, connexin45; eGFP, enhanced green fluorescent protein; hMSCs, human mesenchymal stem cells; MDPs, maximal diastolic membrane potentials; ND, not detected; nrCMCs, neonatal rat cardiomyocytes.

hMSC-Specific Intracellular Electrophysiological Measurements

To further study the cardiomyogenic differentiation capacity of Cx43↓(1) + Cx45↑ fetal AM hMSCs, these cells were subjected to whole-cell current-clamp measurements following pharmacological uncoupling. Cx43↓(1) + eGFP fetal AM hMSCs (n = 5) showed steady membrane potentials, while some of the Cx43↓(1) + Cx45↑ fetal AM hMSCs (n = 5) produced spontaneous action potentials with MDPs comparable to those of nrCMCs (n = 7) (−65 ± 4 mV and −69 ± 5 mV, respectively) (Fig. 7C1, 7C2). Thus, the loss of excitability in Cx43↓(1) fetal AM hMSCs could be overcome by Cx45 overexpression, at least in a subpopulation of cells.

DISCUSSION

The key findings of this study are (a) fetal hMSCs, which intrinsically express Cx43 at high levels, efficiently communicate with adjacent nrCMCs via gap junctions and can differentiate into functional heart muscle cells when cocultured with nrCMCs. (b) Adult hMSCs that contain low amounts of Cx43 by nature do not undergo cardiomyogenic differentiation in coculture with nrCMCs. (c) Overexpression of Cx43 in adult hMSCs does not lead to cardiomyogenic differentiation upon coculture with nrCMCs. (d) Cardiomyogenic differentiation of fetal hMSCs in coculture with nrCMCs is inhibited by Cx43 knockdown but is rescued by concurrent overexpression of Cx45.

Gap Junctions and hMSCs

The expression of the three major cardiac Cx (i.e., Cx40, Cx43, and Cx45) genes by hMSCs has previously been demonstrated [22, 23]. Also, the formation of functional gap junctions, important for integration of donor cells with the surrounding myocardium after cardiac stem cell therapy, has been shown to occur [9, 22, 23]. Recently, our group showed that hMSCs derived from fetal tissues contain much higher levels of Cx43 than those derived from adult tissues [19]. In addition, evidence was provided that hMSCs of embryonic or fetal origin can undergo cardiomyogenesis when cocultured with nrCMCs, while hMSCs derived from adult sources fail to do so. To directly investigate the role of gap junctional coupling in cardiomyogenic differentiation of hMSCs, in this study, Cx43 expression was downregulated in fetal AM hMSCs and upregulated in adult AT hMSCs through LV gene transfer. These genetic interventions resulted in a strong reduction in gap junctional communication between fetal AM hMSCs and nrCMCs and led to a substantial increase in the gap junctional coupling between adult AT hMSCs and nrCMCs. In addition, it was investigated whether the effects of Cx43 knockdown could be reversed by Cx45 overexpression. The reason to choose Cx45 for this purpose was that it can be engaged in the formation of functional homotypic as well as heterotypic gap junctions composed of homomeric or heteromeric connexins [24–27]. In this study, it was indeed established that after transduction with a Cx45-encoding LV, Cx43↓(1) fetal AM hMSCs express high levels of Cx45 mRNA leading to rescue of their efficient gap junctional communication with nrCMCs.

Role of Gap Junctional Communication in Cardiomyogenic Differentiation

Connexins have been reported to be involved in differentiation processes such as osteogenesis, neural differentiation, and hematopoiesis [28–32]. However, little is known about the role of gap junctional communication in cardiomyogenic differentiation. As already stated above, our group recently found that fetal AM hMSCs express Cx43 at high levels in contrast to adult AT hMSCs. Interestingly, these fetal AM hMSCs were able to differentiate, in coculture with nrCMCs, into functional CMCs, while adult AT hMSCs were not [19]. Other investigators using cocultures with CMCs to induce cardiomyogenic differentiation in stem cells have also proposed that intercellular communication through gap junctions might be essential in this differentiation process [33, 34]. However, the role of gap junctional communication in the transfer of cardiomyogenic signals from CMCs to other cell types has not yet been investigated. In this study, the role of Cx43 in cardiomyogenic differentiation of hMSCs in the presence of neighboring CMCs was further elucidated. Fetal AM hMSCs lost their ability to differentiate into functional CMCs after knockdown of Cx43 gene expression. However, overexpression of Cx45 not only restored the ability of these hMSCs to form functional gap junctions with nrCMCs but also rescued the cardiomyogenic differentiation capacity of these fetal hMSCs upon coincubation with nrCMCs. In contrast to these findings, adult AT hMSCs did not gain the ability to differentiate into functional CMCs after overexpression of Cx43. The inability of adult AT hMSCs to undergo cardiomyogenesis following coculture with nrCMCs could have an epigenetic explanation [35–37]. According to this hypothesis, genes important for cardiomyogenic differentiation could, in adult AT hMSCs, exist in a transcriptionally repressive state imposed by specific DNA methylation patterns and/or chromatin signatures. Also, qualitative and/or quantitative differences in the repertoire of transcription factors and miRs expressed by fetal AM hMSCs and by adult AT hMSCs might contribute to their differential responsiveness to cardiomyogenic signals produced by neighboring nrCMCs [38, 39].

Since gap junctions enable intercellular communication, induction of cardiomyogenesis after coincubation with nrCMCs may be caused by exchange of signals between nrCMCs and fetal hMSCs. Stimulation of cardiomyogenic differentiation by miRs-499, as previously described, has been inferred to be a gap junction-mediated process [11]. Moreover, intercellular communication makes [Ca2+] oscillations possible, which has been shown to increase the regenerative potential of human cardiac progenitor cells in a mouse myocardial infarction model [40]. In line with these findings are the observations by Muller-Borer et al. who found that in coculture with nrCMCs, rat liver stem cells obtain CMC-like properties and display [Ca2+] oscillations synchronous with those of adjoining CMCs. The [Ca2+] oscillations in the liver stem cells were dependent on gap junctional communication with neighboring CMCs and their inhibition led to a decrease in the expression of CMC-specific genes by the liver stem cells [34]. The recent observation that hyperpolarization is sufficient to induce cardiomyogenic differentiation of human CMC progenitor cells underlines the importance of gap junctional coupling in this process [12]. In this study, fetal AM hMSCs in coculture with nrCMCs were hyperpolarized in contrast to those that were transduced with hCx43-specific shRNAs. More importantly, Cx43↓(1) fetal AM hMSCs were also hyperpolarized in coculture with nrCMCs after overexpression of Cx45. As a matter of fact, the average MDPs of control fetal AM hMSCs and Cx43↓(1) + Cx45↑ fetal AM hMSCs in contact with nrCMCs were identical. This could explain why knockdown of Cx43 leads to an inability of fetal AM hMSCs to differentiate toward functional CMCs, while after subsequent Cx45 overexpression, their cardiomyogenic differentiation potential was restored. So while the precise composition of the gap junctions involved in the restoration of the cardiomyogenic differentiation capacity in Cx43↓(1) fetal hMSCs following Cx45 overexpression remains elusive, the factor(s) that exert their cardiomyogenic effects on fetal AM hMSCs via gap junctions can pass Cx45-containing channels in high enough amounts to set off the cardiomyogenic differentiation of these cells.

Limitations

It would have been of interest to conduct Cx43 knockdown experiments using a cell type with a higher propensity to differentiate into functional CMCs than fetal hMSCs. However, such a cell type was not available to us. Furthermore, it would have been clinically more relevant to coculture different hMSC subtypes with adult human CMCs, but obtaining these cells in the numbers needed to conduct these experiments was impossible. Also, adult human CMCs cannot be maintained ex vivo long enough in a differentiated state to perform some of the key experiments described in this article.

CONCLUSIONS

The results of this study indicate that efficient gap junctional coupling with adjacent CMCs is necessary to induce cardiomyogenic differentiation of naturally Cx43-rich fetal hMSCs in coculture with nrCMCs. However, adult AT hMSCs, which contain relatively low intrinsic levels of Cx43 and do not undergo cardiomyogenesis in the presence of nrCMCs, cannot be endowed with cardiomyogenic differentiation ability by overexpression of Cx43.

Acknowledgements

Zeinab Neshati, M.Sc., is gratefully acknowledged for the technical support. This research forms part of Project P1.04 SMARTCARE of the BioMedical Materials (BMM) program, which is cofunded by the Dutch Ministry of Economic Affairs, Agriculture, and Innovation. The financial contribution of the Dutch Heart Foundation (NHS) is gratefully acknowledged. D.E.A. is funded by the Smart Mix Program of The Netherlands Ministry of Economic Affairs and The Netherlands Ministry of Education, Culture, and Science. D.A.P. is recipient of Veni Grant 91611070 from The Netherlands Organization for Scientific Research (NWO).

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

No conflicts of interest.

Ancillary