Introduction

In the past years, stem cell biology has emerged as one of the most active areas in biomedical research. Stem cells possess two remarkable characteristics: they are capable of self-renewal, while retaining the ability to differentiate into mature daughter cells [74]. Two categories of stem cells are discussed here, embryonic stem cells (ES cells) and somatic stem cells. ES cells are derived from the inner cell mass of the blastocyst stage of embryo [30, 69, 96, 111]. ES cells are pluripotent: they can give rise to most, if not all, tissue types of the body [91]. On the other hand, somatic stem cells persist in various niches throughout postnatal life and are critical to maintaining tissue homeostasis by replacing cell losses due to turnover, damage, and degeneration [77]. Generally somatic stem cells are restricted to differentiation into lineage(s) that comprise the tissue of origin. However, in some cases somatic stem cells can differentiate into multiple lineages, showing more developmental potential than previously thought [92, 119]. This multi- or pluri-potent developmental potential of stem cells has spurred great interest in their potential implications in cell replacement therapy.

The microenvironment that stem cells reside in is critical to their maintenance, and communications between neighbouring cells play an important part in determining cell fate. While most studies of the stem cell niche focus on paracrine or juxtacrine cell interactions, intercellular communication through gap junctions represents an understudied area. This review aims to summarize the current understanding of gap junctions in both somatic stem cells and ES cells.

Gap Junctions

The term ‘gap junction’ comes from histological studies performed in the 1960s, where a distinct ‘gap’ between two plasma membranes was observed by electron microscopy [97]. Gap junctions are hydrophilic channels made up of two hemichannels termed connexons, each of them localized in the membrane of adjacent cells. In turn, each connexon consists of six integral membrane proteins termed connexins [53, 104] Connexons can be assembled from either a single type of connexins (homomeric) or multiple types of connexins (heteromeric). Subsequently, the gap junction can be either homotypic, consisting of two identical homomeric connexons, or heterotypic, consisting of two different heteromeric or homomeric composition (Fig. 1

Fig. 1
figure 1

Organisation of connexins into gap junctions and molecules capable of diffusing through gap junctions

Full size image

). For a detailed discussion on connexin trafficking and gap junction assembly, please refer to an excellent review by Martin et al. [70].

Connexin genes belong to a highly conserved multigene family [103, 125]. To date, 21 connexins have been identified in human and 20 in the mouse. Connexins generally consist of four transmembrane regions, two extracellular loops and three intracellular domains [104]. The extracellular and transmembrane domains are conserved among connexin family members, whereas their intracellular sequences are more diverse [35]. The cytoplasmic C-terminal tail contains multiple serine and tyrosine residues which allow phosphorylation of connexins, resulting in regulation of molecular diffusion through gap junctions [57]. The interaction of other cellular proteins with connexins is the subject of an exhaustive review [33]. Connexins can be regulated at the transcriptional, translational and post-translational levels. This in turn can affect biogenesis, assembly, intracellular transport, docking, channel gating, internalization and degradation of gap junctions [56, 57, 79, 84, 103].

Traditionally, connexins are believed to be the only proteins capable of forming gap junctions in vertebrates, while innexins form gap junctions in invertebrates [60]. Recently three innexin orthologs, termed pannexins PANX1, PANX2 and PANX3, have been described in the mouse and human [86, 87]. At least one pannexin, PANX1 is able to form functional gap junctions in mammalian cells, rendering pannexin as the second class of gap junction proteins in vertebrates [43, 55]. However, it remains controversial whether pannexins play an overlapping role with connexins in mediating gap junctional intercellular communication (GJIC).

Gap Junctional Intercellular Communication

Gap junctions are the only intercellular junctions that allow direct transfer of signaling molecules and metabolites to adjacent cells [1, 53]. Atomic force microscopy studies have demonstrated that gap junctions can exist in an ‘opened’ conformation that allows molecular diffusion through their pore, or a ‘closed’ conformation that restricts molecular diffusion [28]. Moreover, gap junctions made up from different connexins can have different pore sizes, giving rise to different permeability for the transfer of molecules [103]. However, it is generally accepted that only molecules less than 1–1.5 kDa with a maximum diameter of ∼1.5 nm can diffuse through gap junctions in the ‘opened’ state [22, 31].

Gap junctional intercellular communication (GJIC) refers to the diffusion of intracellular molecules through gap junctions to the neighboring cell. Molecular movement through the gap junctions is believed to occur by passive diffusion [53]. As illustrated in Fig. 1, numerous cytoplasmic molecules can diffuse through gap junction channels, including small ions (Na+, K+, Ca2+, H+, Cl−), second messengers (cyclic nucleotides, inositol triphosphate), amino acids (glycine, glutamate), metabolites (glucose, glutathione, adenosine, AMP, ADP, ATP), short interfering RNA and peptides involved in cross-presentation of major histocompatibility complex class I molecules [1, 52, 80, 115]. GJIC is involved in various cellular mechanisms, including intercellular buffering of cytoplasmic ions, electrical synchronization, control of cell migration, cell proliferation, cell differentiation, metabolism, apoptosis and carcinogenesis [22, 52, 75, 88, 112, 121].

The functional state of GJIC can be determined experimentally by demonstrating gap junction-mediated transfer of either small biochemical dyes (e.g. Lucifer yellow, 6-carboxyfluorescein) or ions between adjacent cells (microelectrode impalements to detect electrical voltage across adjacent cells). It is important to recognize that GJIC should not be considered a simple all-or-none phenomenon. In some situations, gap junctions can close to a pore size that allows ions to pass through but not relatively larger biochemical dyes [54, 63]. Furthermore, different connexins can form gap junctions with dramatically different permeability and selectivity to specific molecules. For example, gap junctions comprised of Cx43 mediate transfer of glucose metabolites much better than Cx32 channels, while Cx32 channels transfer adenosine with greater efficiency than Cx43 channels [1]. The basis of this selectivity of molecular transfers is thought to be related to the ability of different connexins to assemble channels with different sizes, charges and binding affinities to specific molecules [1]. Gap junctions made of multiple types of connexins (heterotypic or heteromeric) can have different permeability and selectivity to diffused molecules compared to homotypic gap junctions, in a manner that is not yet fully understood.

Finally, recent evidence has also suggested that unpaired connexon hemichannels can also mediate intercellular communication without forming gap junctions [25, 31, 36]. Under resting conditions, connexon hemichannels in the plasma membrane are closed. However, in certain situations, connexon can be stimulated to open and release intracellular signalling molecules. Please refer to Goodenough and Paul for an extensive review on this topic [36].

Gap Junctions in Growth Control

Gap junctions have an important role in cellular growth control. Early studies led to a simple view that up-regulation of GJIC inhibits cellular proliferation, while down-regulation of GJIC stimulates proliferation [68, 114]. The diffusion of putative growth inhibitory factors through the gap junction is thought to cause this effect [68]. Previous studies showed that GJIC become undetectable as the cells enter mitosis [34, 106]. Decreasing GJIC is also observed during normal cell cycle transit and is disrupted during neoplastic growth [101]. In the 1960’s, some pioneering works from Loewenstein and Kanno demonstrated that rat hepatomas do not communicate via gap junctions [66, 67], contrary to normal hepatocytes which possess functional GJIC [90]. This work led to the hypothesis that gap junctions are involved in growth control, where deficiency in GJIC results in the deregulated growth of cancer cells [65]. This hypothesis is supported by the fact that GJIC deficiency is described in many cancer cell types as opposed to their normal counterpart [75, 76, 121]. Other evidences supportive of this hypothesis in cancer cells include the following:

  1. I.

    In many types of tumour cells, GJIC is not functional as a result of deficient in connexin expression, or inability to assemble functional gap junctions into the membrane [121].

  2. II.

    Cx32−/− knockout mice have a higher incidence of liver cancer and leukemia, and Cx43+/− knockout mice have a higher susceptibility to lung cancer [3, 38, 39, 110].

  3. III.

    Tumour-promoting agents (e.g. phorbol 12-myristate 13-acetate) and oncogenes (e.g. Src) can regulate gap junctional communication, further suggesting a role of GJIC in carcinogenesis [33, 130].

  4. IV.

    A direct relationship between gap junction expression and the malignant phenotype is demonstrated by a number of transfection studies, where transfection of cancer cells with certain connexins is followed by the subsequent restoration of GJIC and growth suppression [121, 130]. In addition, isolated HeLa clones with Cx43 expression and GJIC restoration exhibit marked decrease in cellular growth and tumorigenicity compared to normal HeLa that lack Cx43 and GJIC [49].

Thus, early studies of gap junctions in cancer cells supported a view where down-regulation of GJIC promotes cell proliferation, while up-regulation of GJIC inhibits cell proliferation [65]. This view is almost certainly an oversimplification, since cell proliferation and functional GJIC are correlated in some cell types. Certain cancer cells possess functional GJIC, such as embryonal carcinoma cells and rat bladder carcinoma cells [4, 6, 51]. Moreover, ES cells and early embryonic cells which proliferate rapidly also form functional GJIC. Increasing evidence also hints that tumour suppression by connexin transfection may act via a GJIC-independent pathway [95]. It appears that the relationship between the functional state of GJIC and cell proliferation may be more complicated than previously thought. Nevertheless, the data do indicate that gap junctions serve as a powerful regulator of growth control in a cell type-dependent fashion.

Gap Junctions in Mediating Cellular Differentiation in the Early Embryo

The early mouse embryo provides an excellent model to study the role of gap junctions in mediating cellular differentiation. The pre-implantation embryo expresses multiple connexin transcripts, including Cx30, Cx30.3, Cx31, Cx31.1, Cx36, Cx40, Cx43, Cx45 and Cx57, but not Cx26 and Cx32 [21, 42, 81]. In the human blastocyst, Cx26, Cx31, Cx43 and Cx45 transcripts are detected, however connexin expression patterns can be inconsistent between different embryos [12, 37].

GJIC between blastomeres appears from the eight-cell stage during compaction in the mouse [64], or much later at day 5–6 blastocyst stage in human [20]. At the blastocyst stage of the mouse embryo, GJIC was observed between cells within the inner cell mass [64]. Upon in vitro implantation of mouse blastocysts (equivalent to embryonic day 6.5, E6.5), dye coupling is progressively lost between the inner cell mass and the trophectoderm, but ionic coupling remains between these two populations [63]. By the egg cylinder stage (E7.5), ionic coupling also ceases and GJIC is completely lost between the cells that originate from the inner cell mass (ectoderm, endoderm, mesoderm and extraembryonic endoderm) and those originate from the trophectoderm (extraembryonic ectoderm and ectoplacental cone), thus giving rise to two different ‘communication compartments’ [45, 46]. Such changes in GJIC are consistent with the spatial re-organisation of connexin expression. For example, Cx43 and Cx31 are expressed in both the inner cell mass and trophectoderm in the pre-implantation mouse embryo, but Cx43 expression becomes restricted to derivatives of the inner cell mass while Cx31 expression is confined to derivatives of trophectoderm after implantation (E6.5) [19] Since Cx43 and Cx31 are not able to form functional heterotypic gap junctions [27], their distribution patterns effectively divide the derivatives of the inner cell mass and trophectoderm into two communication compartments. Further subdivision into smaller communication compartments occurs within both the embryo proper and extra-embryonic tissues, with each germ layer comprising a separate communication compartment [45, 46]. Curiously, in some cases, ionic coupling may persist between some germ layers [45, 46]. In summary, the current theory is that during early post-implantation development, GJIC is maintained within cells of the same lineage but lost between tissues whose fate diverges [62, 122]. Thus the formation of gap junction communication compartments is proposed to mediate specific intercellular signalings between cells with similar cell fates.

Despite many attempts to dissect the functional significances of GJIC during pre-implantation embryo development, to date whether GJIC is a functional requirement remains controversial [41]. Inhibition of GJIC by antibodies targeting multiple connexins [5, 58] or antisense RNA [9] prevents compaction and further development in mouse embryos. However, pharmacological inhibition of GJIC using α-glycyrrhetinic acid (α-GA) does not affect development of the blastocyst, suggesting GJIC may be dispensable during pre-implantation development in mouse [42, 118]. More studies are needed to elucidate the role of gap junctions in the early embryo development. In particular, knockout study of multiple connexins will help address this question in the future.

Gap Junctions in Embryonic Stem Cells

ES cells are traditionally derived from the inner cell mass of the blastocyst stage of embryo [30, 69, 96, 111]. Since then, recent reports have demonstrated successful derivation of ES cells from different stages of embryo, including single blastomere from eight–ten cell stage [18], morula and late stage embryo [107, 108]. ES cells are pluripotent and capable of proliferating indefinitely in vitro, thus potentially serve as an unlimited source of healthy tissue for cell replacement therapy [91].

Mouse Embryonic Stem Cells

Similar to cells from the inner cell mass, undifferentiated mouse ES cells (mESC) express transcripts of Cx43 and Cx45 and display functional GJIC [81, 85]. Furthermore, it was recently shown that pannexin-1 and five additional connexin transcripts are present in mESC: Cx26, Cx30.3, Cx31, Cx32 and Cx37. Curiously, only Cx31, Cx43 and Cx45 proteins were detected in mESC [129], suggesting that a regulatory system may exists in mESC to suppress translation of many connexin isoforms.

Several studies have attempted to study the function of gap junctions in mESC by knocking out or knocking down specific connexins. Cell proliferation of Cx43-knock down mESC is reduced significantly, but cell survival remains unchanged upon Cx43 knockdown [112, 129]. Moreover, Cx43 knock-down mESC exhibit down-regulation of several stem cell markers as well as up-regulation of differentiation markers [112]. Interestingly, these mESC are also unable to form embryoid bodies [112]. This suggests that GJIC is essential for the regulation of mESC to maintain undifferentiated as well as initiating differentiation. In addition, gene expression profiling indicates that Cx43-null mESC undergoing neural differentiation exhibit a deficiency in oligodendrocyte differentiation and an increase in astrocyte differentiation [88]. Although more detailed studies are needed to confirm these data, this study is the first to indicate an essential role of gap junction for proper neuroectodermal specification. Together, these results indicate an essential role of gap junction in mESC biology.

In contrast, Cx45-null mESC are morphologically similar to the wildtype and can readily differentiate into cells of the three germ layers following embryoid body formation [26]. One explanation for this lack of phenotype is that Cx43 is still expressed in Cx45-null mESC and may compensate for the loss of Cx45 [26]. Interestingly, when Cx45-null mESC were injected into the blastocyst, they failed to form chimeric mice or post-implantation embryos. Subsequent studies of chimeric blastocysts cultured in vitro showed that Cx45-null mESC were unable to incorporate into the recipient inner cell mass [26]. Cx45+/− heterozygous mutant mESC also show a decreased efficiency in forming chimeric mice [26]. From this study, it is proposed that Cx45 may play a distinct role in the incorporation of ES cells into inner cell mass by mediating intercellular communication.

Human Embryonic Stem Cells

Many transcriptome studies of human embryonic stem cells (hESC) are now available. These studies initially aimed to search for ‘stemness’ genes in hESC, and interestingly their results hinted at a role of gap junctions in maintaining hESC pluripotency. Multiple studies have demonstrated that Cx43 is enriched in undifferentiated hESC when compared to embryoid bodies [10, 61] or a pooled sample of adult human tissues [11, 98]. Moreover, the pluripotent factors Oct4, sox2 and nanog can co-occupy the upstream region of Cx43 to activate its expression, suggesting its role in the maintenance of hESC pluripotency [13]. Other microarray analysis have also shown that Cx43 and Cx45 mRNA are both highly enriched in hESC compared to a range of somatic tissues or spontaneously differentiated hESC [7, 105]. Interestingly, Cx45 is up-regulated in hESC compare to human embryonal carcinoma cells [105]. Furthermore, Enver et al. showed that Cx45 is also up-regulated in SSEA-3 positive undifferentiated hESC compare to its SSEA-3 negative differentiated counterpart [29]. Finally, Assou et al. summarized 38 previous reports of microarray studies in hESC and devised a ‘consensus list’ of undifferentiated hESC markers, including Cx43 and Cx45 [2]. Together, these transcriptome studies hinted an understudied role of connexins in the maintenance of hESC pluripotency.

Our previous study demonstrated that hESC express at least two connexins, Cx43 and Cx45 [128], a result consistent with others [15]. Two phosphorylated forms of Cx43 were observed in undifferentiated hESC, although the role of connexin phosphorylation in regulating GJIC is not clear [128]. Moreover, a recent study demonstrated the mRNA expression of an additional 16 connexin types in hESC, including Cx25, Cx26, Cx30, Cx30.2, Cx30.3, Cx31, Cx31.1, Cx31.9, Cx32, Cx36, Cx37, Cx40, Cx46, Cx47, Cx59 and Cx62 [43]. It appears that hESC express almost all transcripts of human connexin isoforms, except Cx40.1 and Cx50 [44]. Furthermore, GJIC was observed in undifferentiated hESC, as determined by dye coupling [15, 128, 132] and ionic coupling [44]. Our previous studies showed that such GJIC can be inhibited by BMP stimulation, protein kinase C activation and Erk1/2 inhibition [127, 128]. However, exogenous calcium does not modulate GJIC in hESC. In addition, hESC cultured in serum-containing, serum-free or feeder-free conditions possess similar gap junction properties [127]. Together, these studies suggest that functional GJIC is a common characteristic of hESC maintained in different culture conditions [127, 128]. We also showed that hESC do not communicate with mouse embryonic fibroblast feeder cells through gap junctions, a result consistent with studies in mESC [85]. Finally, the functional role of gap junctions in hESC was studied using the gap junction blocker α-GA. We utilized a serum-free culture system for hESC previously established in our laboratory using S1P and PDGF [89]. In hESC cultured in S1P and PDGF, inhibition of GJIC by α-GA resulted in reduced colony growth and increased apoptosis [127]. This result indicates an important role of gap junctions in the maintenance of hESC. However, it must be noted that α-GA had no significant effect on hESC cultured in fetal calf serum, suggesting that under certain conditions the effect of GJIC inhibition may be compensated by other unidentified signals.

Future studies that utilise RNAi techniques to knockdown specific connexins will be helpful in determining the role of specific connexins in regulating cell pluripotency, proliferation and survival in hESC. Research in this area will prove useful in maintaining hESC in an undifferentiated state or enhancing hESC differentiation along specific lineage. In addition, hESC survive poorly as single cell, which is vital to many genetic modification procedures. It is speculated that cell–cell interaction between hESC is important to their survival, thus a better understanding of the role of gap junctions may lead to improved methodology for cloning of hESC.

Gap Junctions in Somatic Stem Cells

In contrast to ES cells, a hallmark feature of somatic stem cells is that they exist in specific niches where they remain in a quiescence state [82]. Somatic stem cells can undergo symmetric division to self-renew, or asymmetric division to generate a differentiated daughter cell [50]. GJIC is often important in regulating the cell fates of somatic stem cells, a phenomenon possibly conserved from lower organism to mammalians. For example, Ovieda et al. recently demonstrated that the inhibition of GJIC in somatic stem cells prevents regeneration of the planarian flatworm, suggesting a conserved role of gap junctions in regulating stem cell fate [83]. Here we will discuss the current understanding of gap junctions in a number of somatic stem cells.

Hematopoietic Stem Cells

Hematopoietic stem cells (HSC) are the most extensively studied somatic stem cell population in both humans and mice [73]. HSC are capable of giving rise to all blood cell types, including red blood cells, B lymphocytes, T lymphocytes, neutrophils, natural killer cells, basophils, eosinophils, monocytes, macrophages and platelets [113]. During adulthood, HSC reside primarily within niches in bone marrow, where cell–cell communication with the surrounding stromal cells is critical for regulating HSC maintenance [48]. In particular, it has been proposed that GJIC may be involved in this process [94, 100]. Recent studies suggested that Cx32 is vital to HSC differentiation. Cx32 expression can be readily detected in Lin−c-kit+ HSC-enriched cells [39]. Interestingly, Cx32 knockout mice exhibit more undifferentiated HSC and fewer progenitor cells, suggesting a role of Cx32 in maturation of HSC to progenitor cells [38, 39]. In addition, Cx43 is also implicated in hematopoiesis. During the quiescent state, undifferentiated HSC (Lin−, Sca1+, C-kit+) do not express Cx43 mRNA [78]. However, Cx43 expression can be massively up-regulated in adult mouse bone marrow upon forced stem cell division [99]. Cx43 deficient mice also demonstrate clear defects in blood cell formation [78]. However, it is not clear whether this effect is mediated by functional gap junctions or by connexon hemichannels. Moreover, it remains unclear whether GJIC among HSC or between HSC and stromal cells is important in defining the stem cell niche. Thus, further research is needed to provide a definitive role of GJIC in HSC.

Mesenchymal Stem Cells

Mesenchymal stem cells (MSC) have been isolated from various tissues, including bone, umbilical cord blood and adipose [8, 123]. They can readily differentiate into adipose tissue, tendon, cartilage and bone [93]. Human MSC express Cx40, Cx43 and Cx45, and can communicate among themselves via gap junctions [59, 116]. However, a rare population of MSC was shown to be deficient in GJIC [59]. The exact identity of this MSC population remains unknown. Moreover, human MSC were demonstrated to form Cx43-mediated GJIC with umbilical vein endothelial cells, where this communication is important for osteogenic differentiation of MSC [120].

Neural Stem Cells

Neural stem cells exist in the developing or adult nervous system and can differentiate into all neural cell types, including neurons, astrocytes and oligodendrocytes [32]. Neural stem cells isolated from the rat developing brain express Cx43 and Cx45 and are capable of forming GJIC, where GJIC is essential for survival and proliferation in these cells [14]. Moreover, Cx43 and Cx32 levels were observed to increase temporarily during differentiation in these cells [131]. Future studies on human neural stem cells will help address whether this role of gap junctions is conserved among species. Similarly in mouse fetal neural progenitors, the closure of gap junctions decreases cell proliferation and diminishes cell survival [17, 24]. On the other hand, overexpression of Cx43 stimulates cell proliferation in these cells [17]. In addition, neural progenitors from other species have been demonstrated to express connexin and communicate via GJIC, possibly suggesting a conserved role of gap junction in neural stem cell biology. The astroglial progenitor cell line TB2, isolated from the fish brain, expresses Cx43 and Cx35 [124]. In the turtle, the spinal cord neural progenitors express Cx43 and are electrically and metabolically coupled via gap junctions [102]. Together, it appears that gap junctions are critical to maintaining neural stem cell homeostasis, both in self-renewal and early stages of differentiation.

Other Somatic Stem Cells

Although GJIC seems to play a role in regulating many somatic stem cell types, gap junctions are absent in some somatic stem cells. This included two somatic stem cells that give rise to the epithelium, keratinocyte stem cells [71] and corneal epithelial stem cells [72], both of which lack connexins and functional GJIC. A number of presumptive somatic stem cells were also demonstrated to lack functional GJIC, including pancreatic ductal epithelial stem cells [109], neural–glial stem cells [23], bovine mammary gland progenitor cells [40], human breast epithelial stem cells [47] and human kidney epithelial stem cells [16]. However, in some cases the identification of the cell populations studied as stem cells is questionable, and further characterization is needed to confirm their potential for self renewal and differentiation. Altogether, it appears that intercellular communication via gap junctions is not a universal feature of all somatic stem cells. Although the function of gap junctions seems to be highly dependent on the cell type, previous reports demonstrated that functional GJIC plays a critical role in regulating the cell fates of many somatic stem cells.

Summary

As the reader can appreciate, the role of gap junctions in stem cells requires further research and remains an important research question. Emerging evidence suggests that gap junctions seem to play an important role in regulating the cell fate of ES cells, neural stem cells, mesenchymal stem cells and possibly hematopoietic stem cells. However, the precise role of gap junctions appears to be highly dependent on the type of stem cells. This is not surprising, given the biological differences between somatic stem cells and ES cells. To date, several key questions remain to be addressed: Is GJIC a defining feature of the stem cell niche that maintains homeostasis of somatic stem cells? During commitment and differentiation, do stem cells form ‘communication compartments’ where GJIC is restricted to only cells with similar cell fate? If so, what is the role of the specific connexin isoforms? Does the formation of communication compartment enhance cellular commitment and differentiation? Which intracellular molecules that diffuse through gap junction are responsible for such effects? One interesting candidate is microRNA that diffuses through gap junctions [117], given the increasing appreciation of the important role of microRNA in directing stem cell fate [132]. Indeed, recently it was demonstrated in hESC that siRNA are able to move through gap junctions to affect gene expression of the neighbouring cells [126]. The identification of molecules that mediate intercellular communication through gap junctions will be an important step towards the elucidation of regulatory mechanisms of stem cell survival, differentiation and proliferation. Studies on gap junctions in stem cells can potentially lead to the development of novel methods in expanding stem cells in vitro, directing their differentiation into functional mature cells or cancer treatment that target particular cancer stem cells.