To characterize neurosphere composition, we analysed marker expression in suspension and in cultures plated on a variety of adhesive substrates by immunofluorescence [see Additional file 1]. Although each neurosphere is likely derived from a single NPC, progeny spontaneously adopt distinct developmental lineages in culture (Fig 1a). We found neurospheres cultured in the absence of exogenous ECM were primarily composed of nestin⁺ cells (Fig 1i). Cells co-expressing the Type 1 NPC marker GFAP localized to the periphery of neurospheres (Fig 1b). Type 2a cells expressing nestin only were found in the core of cultures (Fig 1b). DCX⁺, NCAM⁺, and TuJ1⁺ neuroblasts and immature neurons were detected in clusters throughout the neurosphere structure (Fig 1c–e). NG2⁺ and PDGFαR⁺ oligodendrocyte progenitor cells (OPCs) and RIP⁺ oligodendrocytes were found at the periphery of neurospheres with rare cells detected in the core (Fig 1f–h). When cultures were plated on laminin, we observed a significant increase in the percentage of cells that retained a nestin⁺ NPC identity and a significant decrease in the percentage of cells that specified to neuroblasts, immature neurons, and/or OPCs (Fig 1i). No change in the number of astrocytes was detected but a small increase in RIP⁺ oligodendrocytes was observed in laminin-plated cultures (Fig 1i).

To identify the connexins expressed by postnatal progenitor cells in the absence of exogenous ECM we performed RT-PCR. Fourteen connexins (Cx26, Cx29, Cx30, Cx30.2, Cx31.1, Cx31.9, Cx32, Cx36, Cx37, Cx40, Cx43, Cx45, Cx47, Cx57) are expressed in the mammalian CNS 12131415. Transcripts for ten of these connexins (Cx26, Cx29, Cx30, Cx32, Cx36, Cx37, Cx40, Cx43, Cx45, and Cx47) were detected in neurosphere culture (Fig 2a–j). Cx29, Cx30, and Cx36 expression was further confirmed by analysis of the reporter gene products present in null-mutant lines. In Cx29^-/- and Cx30^-/- mice, the lacZ open reading frame replaces the connexin coding sequences 2021 whereas a bicistronic reporter cassette containing lacZ, an internal ribosome entry site, and placental alkaline phosphatase replaces the connexin coding sequence in Cx36^-/- mice 22. β-galactosidase expression was readily detected in neurosphere cultures from these three null-mutant lines (Fig 2b, c, e).

We used immunofluorescence to localize connexins to NPC subtypes grown in the absence of exogenous laminin (Fig 3, 5, 6). For all connexins except Cx26, cultures from null-mutant mice served as negative controls to ensure antibody specificity. For Cx26, where deletion is embryonic lethal 23, antibody specificity was determined by Western blot (Fig 4a, inset). Four distinct connexin protein profiles were detected:

(I) Cx30 (Fig 3a) and Cx43 (Fig 3d) were expressed by the majority of cells (> 60%) throughout the structure of the sphere. Punctate membrane-associated immunoreactivity characteristic of gap junction labelling was evident in nestin⁺ and GFAP⁺ cells (Fig 3b–f). We used flow cytometry to quantify Cx30 expression in nestin⁺/GFAP⁺ Type 1 NPCs, nestin⁺ Type 2a NPCs, and GFAP⁺ astrocytes [see Additional file 2]. We found that all of the Type 1 NPCs (93 ± 6%) and the vast majority of Type 2a NPCs (83 ± 2%) expressed Cx30. Surprisingly, few if any committed astrocytes were Cx30⁺ (< 0.1%) [see Additional file 2]. These results were further substantiated by triple-labelling for Cx30, Cx43, and GFAP [see Additional file 3]. Cx43 but not Cx30 was detected in all GFAP⁺ cell types suggesting ubiquitous localization of Cx43 in GFAP⁺ NPCs and astroglia but a more restricted distribution of Cx30 to subsets of multipotential Type 1 GFAP⁺ NPCs.

(II) Cx26 (Fig 4a), Cx29 (Fig 5a), Cx37 (Fig 5c), and Cx40 (Fig 5e) were expressed in a minority of cells (~10%) localizing to both the sphere core and periphery in the absence of laminin. Cx26 and Cx29 exhibited membrane-associated staining but rarely in apposition with neighbouring cells expressing the same connexin (Fig 4a, 5a arrows, 4b, 5b inset). Cx26 localized to nestin⁺ and GFAP⁺ Type 1 NPCs (Fig 4b, c) at the periphery of the sphere and nestin⁺ Type 2a NPCs within the core (Fig 4b) as well as GFAP⁺/nestin^- astrocytes (Fig 4b, c). Cx29 localized to PDGFαR⁺ OPCs (Fig 5b, inset). Cx37 and Cx40 labelling appeared to be largely intracellular (Fig 5c, e, arrows) suggesting these connexins are not forming gap junction channels at the plasma membrane. Both Cx37 and Cx40 were expressed primarily by nestin⁺ Type 2a cells (Fig 5d, f, insets).

(III) Cx45 (Fig 6a) and Cx47 (Fig 6c) were expressed by rare cells (< 10%) restricted primarily to the periphery of neurospheres cultured in the absence of laminin. These connexins were often expressed in adjoining cells (Fig 6a, c, large arrows) where contiguous labelling was detected at the plasma membrane (Fig 6a, c, small arrows). Cx45 localized to nestin⁺ Type 2a NPCs (Fig 6b, inset) and was not detected in GFAP⁺ cells (data not shown). Cx47 expression was restricted to RIP⁺ oligodendrocytes (Fig 6d, inset).

(IV) Both Cx32 and Cx36 were detected at the mRNA level (Fig 2d, e) but no immunodetection of protein was observed (Fig 6k, o). Antibody efficacy was confirmed by Western analysis (Fig 6k, o, insets).

To assess the effect of ECM on connexin expression, NPCs were expanded for 8 DIV in suspension before plating on laminin-coated glass coverslips [see Additional file 1]. Cultures exposed to laminin rapidly adhered to the ECM substrate. Cells at the periphery of the neurosphere migrated from the core as adherent monolayers over 6 DIV. Fewer Cx43⁺ (Fig 3j–l) and Cx40⁺ cells (Fig 5m–o) and no Cx45⁺ (Fig 6e–g) or Cx47⁺ (Fig 6h–j) cells were observed when NPCs were cultured on laminin substrate. Laminin increased the number of Cx26⁺ cells (Fig 4d–f) and induced Cx32 (Fig 6l–n) and Cx36 (Fig 6p–r) protein. Cx32⁺ cells exhibited small nuclei (Fig 6m, n, arrow) and were immunopositive for the OPC marker NG2 (data not shown). Cx36 was expressed in refractile cells with neuritic extensions (Fig 6r, s, arrow) and were immunoreactive for the neuronal marker Tuj1-βIII tubulin (data not shown). Although exposure to the ECM substrate did not alter the percentage of cells expressing Cx30 (Fig 3g), a change in subcellular localization was observed. Immunoreactivity was markedly less punctate (compare Fig 3a and 3i) with protein detected throughout the cytosol of laminin-treated cultures (Fig 3h, i, arrows). Contact with laminin did not change the number of cells expressing Cx29 (Fig 5g–i), Cx30 (Fig 3g–i), or Cx37 (Fig 5j–l).

An alternative explanation for these changes lies in the mechanical impact of NPC plating to an adhesive substrate 162425. To test this, we plated neurospheres on poly-L-lysine, laminin, or a Matrigel mixture containing the following ECM components (56% laminin, 31% collagen IV, 8% entactin) and assessed Cx43 expression (Fig 7). The same reduction in Cx43-expressing cells was observed following engagement with Matrigel as plating on laminin alone. Plating on poly-L-lysine had no effect on Cx43 expression. Taken together, these data provide converging evidence for a laminin-specific regulation of connexin expression as compared to a physico-mechanical influence of culture condition.

Transmembrane flux of the low molecular mass fluorescent dye, LY, was used to assess hemichannel activity in neurosphere cultures. RD was used to control for uptake resulting from plasma membrane damage. Hemichannel opening was induced by mechanical stimulation with glass microbeads as previously described 26. Cultures were expanded in suspension for 8 DIV, then plated and analyzed immediately after adherence (Fig 8a, DIV 1) prior to any detectable change in connexin protein expression (data not shown) or after 6 DIV in contact with laminin (Fig 8a, DIV 6) following the observed changes in connexin expression (Fig 3, 4, 5). Mechanical stimulation elicited a significant increase in dye uptake on DIV 1 (Fig 8a, DIV 1) that was largely inhibited by the dual-specificity chloride channel and connexin/pannexin-channel blocker FFA 2728 but not the chloride channel inhibitor DIDS (Fig 8a, DIV 1). Hemichannel activity was lost when NPCs were cultured on laminin for 6 DIV (Fig 8a, DIV 6). We cannot, however, rule out that these changes are due to an effect of ECM on pannexin channel formation as we have determined that NPCs cultured in suspension express both pannexin 1 and 2 mRNA (data not shown) yet we have not investigated impact of laminin on this expression at the protein level.

Biochemical coupling indicative of GJIC was assessed by scrape loading. Little to no dye transfer was observed on DIV 1 or DIV 6 (Fig 8b, Proliferative conditions, DIV 1 and DIV 6,). These data were surprising given the ubiquitous Cx30 and Cx43 expression detected in the majority of cells prior to laminin engagement. As a positive control, we differentiated NPCs to a predominantly nestin^-/GFAP⁺ astrocytic lineage [see Additional file 4]. Robust GJIC was detected in these cultures and was inhibited by gap junction channel blocker GRA but not its inactive analog GZA (Fig 8b). To provide mechanistic insight into the lack of LY transfer in NPC cultures, we assessed Cx43 phosphorylation status and found, in neurospheres cultured without laminin, that the P2, P3, and hyperphosphorylated forms of Cx43 predominated [see Additional file 5]. Previous studies have indicated that the degree of LY dye coupling is inversely correlated with phosphorylation state 29 suggesting that post-translation modification of Cx43 in NPC cultures may account for this lack of dye transfer. Taken together, these data indicate that postnatal NPC cultures do not establish gap junctions permeable to LY, but do exhibit hemichannel activity that is suppressed upon exposure to laminin.

The repertoire of connexins identified here is much larger than that reported by Wörsdöfer et al., 30 who demonstrated expression of seven connexin transcripts but protein for only Cx31, Cx43, and Cx45. However, these studies used ECM components and not suspension-grown neurospheres. Furthermore, they used embryonic stem (ES) cells rather than postnatal NPCs. Thus, the differences in connexin expression patterns likely reflect the unique connexin signature of the different stem and progenitor cell populations present in these cultures as well as the impact of ECM adhesion on connexin expression. Here, we show that Type 1 and Type 2a multipotential NPCs express Cx26, Cx30, and Cx43. Cx26 and Cx43 have been previously observed in embryonic NPC populations 3132 but this is the first demonstration that Cx30, exclusively expressed postnatally 833, is also expressed by multipotential Type 1 NPCs. We found that progression to a Type 2a lineage is associated with de novo Cx37, Cx40, and Cx45 protein expression. Further commitment to an oligodendrocyte lineage is accompanied by the loss of all of the connexins identified in multipotential NPCs and de novo expression of connexins associated with mature oligodendrocytes (Cx29, Cx32, Cx47) 34. Surprisingly, Cx32 transcript but not protein was detected in OPCs and oligodendroglia in the absence of laminin. The presence of Cx32 transcript but not protein is also observed in ES cells 30 with unknown consequence on cell fate. Taken together, these data provide evidence of intrinsic connexin signatures expressed by progenitor cell subsets over the course of commitment in vitro.

Exposure of NPCs to a laminin substrate altered connexin protein expression and/or stability, intracellular distribution, and functional channel activity. The simplest explanation for these changes is that this profile mirrors the effect of laminin on the differentiation of NPCs. Plating of NPCs on ECM is known to alter the rate of spontaneous specification and survival of embryonic stem cell cultures 353637. Indeed, we found more cells retained a multipotential Type 1/2 phenotype in laminin-treated cultures than those expanded in the absence of laminin. An alternative explanation is a direct effect of laminin/integrin signalling on the regulation of connexin protein expression or stability in NPCs. Certainly, laminin has been shown to alter connexin expression and function in other cell types 171819. The consequence of these changes have only begun to be appreciated. In CNS, the recent finding that gap junctional plaques formed by Cx26 or Cx43 are required for migration of neuronal progenitors along radial glial cells during cortical development in the mouse 7 suggests that ECM regulation of connexin expression is involved in cortical lamination and thus is likely to influence postnatal NPC migration. This hypothesis represents an unexplored role for ECM-connexin interactions in vivo.

The changes in connexin profile detected in the presence of laminin are not restricted to NPCs. OPC and neuronal profiles are also altered. Cx32 and Cx36 protein, not observed in the absence of exogenous ECM, are present following laminin engagement. Cx32 null-mutation has previously been shown to delay terminal differentiation of NG2⁺ OPCs in adult hippocampus 38. Here, we find that the induction of Cx32 protein in NG2⁺ OPCs by laminin is associated with a moderate increase in the frequency of RIP⁺ oligodendrocytes suggesting enhanced oligodendrogenesis. Surprisingly, these new oligodendrocytes did not express Cx47 in the presence of laminin suggesting that perhaps different functional subset of cells are generated. Recent studies indicate that Cx32 and Cx47 do not oligomerize and do not form the same heterotypic intercellular channels taken as evidence for functional differences between these connexins and perhaps between subsets of oligodendrocytes 3940. Finally, the induction of Cx36 protein in TuJ1⁺ neurons in the face of reduced neurogenesis was equally surprising. This observation may recapitulate the changes observed between early and late phases of embryonic neurogenesis. Cx36 mRNA is initially detected in multipotential neural stem cells during early but not late neurogenesis where expression is restricted to their neuronal progeny 841. It may be that the functional changes in channel activity observed following engagement of neurosphere cultures with laminin regulate NPC and OPC commitment.

At no point did we detect cell-cell coupling with respect to the passage of the anionic dye LY. While surprising given that embryonic progenitor populations are coupled during early neurogenesis 424344 our data agree with the findings of Duval et al. who demonstrated GJIC in astrocytes but not in other NPC-derived cells 45. Here, we show passage of LY only when NPC cultures are induced to differentiate towards a primarily astroglial lineage and likely corresponds with a shift in the phosphorylation status of Cx43. Conversely, robust transmembrane dye flux indicative of functional hemichannel activity is evident prior to ECM-induced changes in connexin protein profiles. Laminin engagement suppresses this activity. It is tempting to speculate that connexin (or pannexin)-mediated hemichannel activity promotes specification whereas channel-independent adhesion supports the migration through ECM-defined domains in postnatal hippocampus. However, care must be taken in directly extrapolating our results to the in vivo situation given the complex three dimensional character of the developing hippocampus where many different matrices and cellular scaffolds are present given evidence of different cell phenotypes within 2D and 3D culture systems 16. Clearly, it will be important to establish the impact of ECM control of connexin protein expression on postnatal NPC proliferation, migration, and specification in vivo.

Breeding pairs of Cx29^-/-, Cx36^-/-, Cx37^-/+, Cx40^-/-, Cx45^F/F, and Cx47^-/- animals were generated by our laboratories as described 223446474849. Cx32^-/- and Cx30^-/- animals 2050 were kindly provided by Dr. Klaus Willecke (University of Bonn). Nestin-cre recombinase transgenics 51 were kindly provided by Dr. Ruth Slack (University of Ottawa). Each strain was backbred for 3–12 generations into a C57BL/6 lineage. All null-mutant mice were compared to congenic wild-type (WT) littermates. Mice were kept on a 12 h light-dark cycle and allowed food and water ad libitum. Mice were genotyped using the primer pairs listed in [see Additional file 6]. All experimental protocols were approved by the Animal Care Committee of the University of Ottawa according to guidelines set forth by the Canadian Council on Animal Care.

All chemical reagents were obtained from Sigma-Aldrich (St.-Louis, MO, USA) and all cell culture reagents were obtained from Invitrogen (Burlington, ON, Canada) unless otherwise stated. NPCs were cultured as described in 52 with some modifications. Briefly, cells were isolated from the hippocampi of postnatal day 0 – 2 (P0–P2) mouse pups. Animals were sacrificed by lethal injection with Euthansol (Schering-Plough Canada Inc., Pointe-Claire, QC, Canada) and 500 μm sections between bregma -1.7 mm and -2.2 mm were prepared on a VT1000S vibratome (Leica Microsystems Inc.) in ice-cold artificial cerebral spinal fluid (26 mM NaHCO₃, 124 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1.3 mM MgCl₂, 10 mM D-glucose, 100 U/ml penicillin, and 100 μg/ml streptomycin). Hippocampi from n = 2–6 pups were pooled for each culture. Under a Leica MZ6 dissecting microscope, hippocampi were removed and dissected free of blood vessels and choroid plexus. Hippocampi were minced with a scalpel and enzymatically dissociated in: 26 mM NaHCO₃, 124 mM NaCl, 5 mM KCl, 0.1 mM CaCl₂, 3.2 mM MgCl₂, 10 mM D-glucose, 1% penicillin/streptomycin, 0.1% neural protease, 0.01% papain, and 0.01% DNase I for 45 minutes at 37°C. Single cells were resuspended in expansion media (Dulbecco's modified Eagle's medium F12 (DMEM/F12), 2 mM L-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 1× B27 supplement, 20 ng/ml human recombinant epidermal growth factor (EGF), and 10 ng/ml basic fibroblast growth factor (FGF-2)) in 60 mm Petri dishes (Fisher Scientific, Nepean, ON, Canada). Cell viability was established by Trypan Blue hemacytometer counts and cells plated at a density of 2.5 × 10⁵ viable cells/dish. This protocol typically yields 200 viable neurospheres/60 mm dish cultured in suspension. Cultures were maintained at 37°C in a 5% CO₂ atmosphere with fresh EGF and FGF-2 added every two days for 8 days in vitro (DIV).

On DIV 8, neurospheres were plated in 10 cm tissue culture dishes containing laminin (15 μg/mL), Matrigel (Invitrogen), or poly-L-lysine (100 μg/ml)-coated glass coverslips (Corning, NY, USA) and cultured in expansion media supplemented with EGF and FGF-2 or in differentiation media: DMEM/F12, 2 mM L-glutamine, 1 mM sodium pyruvate, 200 mM D-glucose, 100 U/ml penicillin, 100 μg/ml streptomycin, and 1× N2 supplement containing EGF and FGF-2 or 0.5 μM retinoic acid and 0.5% fetal bovine serum. The latter condition promotes astroglial specification 53. Cells were not dissociated to maintain the 3D topography of the cultures. Neurospheres adhered to each substrate and began to grow as monolayer colonies within 24 h of transfer. To maintain NPC proliferation, half the volume of media was removed and replenished every 2 days with fresh addition of EGF and FGF-2 to final concentrations of 10 and 20 ng/ml for 6 DIV (total 14 DIV). Some cultures were pulsed with 20 μg/mL 5'-bromo-2-deoxyuridine (BrdU, Roche Diagnostics, Laval, QC) 24 h prior to plating on laminin to determine the percentage of actively dividing cells in suspension that generated mature progeny. Cultures exposed to exogenous matrix or adhesive substrate were compared to cultures maintained in suspension in the same media supplemented with EGF and FGF-2. All immunocytochemistry and western analysis were performed on DIV 14.

Total RNA was isolated from WT, Cx29^-/-, Cx30^-/-, and Cx45^-/-neurospheres and from adult and/P2 WT, Cx32^-/-, Cx36^-/-, Cx37^-/-, Cx40^-/-, and Cx47^-/- mouse brain using Trizol Reagent (Invitrogen). RNA was collected from one dish per experiment. Each assessment was confirmed in duplicate or triplicate. Total RNA was treated with DNaseI (Promega, Madison, WI, USA). First-strand synthesis was performed using pdN₆ (Promega) random primers and Superscript II RT (BD Biosciences, San Jose, CA, USA) according to manufacturer's recommendations. PCR was performed using the primers and conditions listed in [see Additional file 6]. The PCR reaction contained, in a final volume of 25 μl: 1× PCR buffer, 0.8 mM dNTPs, 1× Advantage 2 Polymerase (Clontech, Cambridge, ON, Canada). All reactions were carried out using the following cycling parameters: 94°C for 5 min, 35 cycles of 94°C for 25 sec, 59°C for 50 sec, and 72°C for 1 min 45 sec, and a final step at 72°C for 7 min in a Whatman Biometra T-Gradient thermocycler (Montreal-Biotech Inc., Kirkland, QC, Canada).

Neurospheres were washed in PBS (10 mM PBS: 0.154 M NaCl, 0.0028 M NaH₂PO₄, 0.0072 M Na₂HPO₄, pH 7.2), pelleted, and resuspended in RIPA buffer (1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulphate (SDS), 1 mM sodium fluoride, 1 mM sodium orthovanadate, 50 μg/ml aprotinin, and 1 mg/mL phenylmethylsulphonylfluoride in PBS). Where Western analysis of cultures grown on laminin substrate is indicated, adherent cultures were washed with PBS and RIPA buffer added directly to plates. Cells were collected using a cell scraper (Fisher), incubated on ice for 30 min, then centrifuged at 13 400 × g for 30 min. Protein was isolated from mouse brain homogenized in RIPA buffer using a Tissue Tearor (Biospec Products Inc., Bartlesville, OK, USA). Protein concentration was determined using the Bio-Rad DC protein assay kit (Bio-Rad, Hercules, CA, USA) according to manufacturer's protocol. Protein samples of 30 μg were separated by SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes (Fisher). Membranes were blocked at room temperature for 1 h with 1% casein in PBS. Primary and secondary antibodies are listed in [see Additional file 6]. Immunoreactivity was visualized using the SuperSignal West Pico Chemilunescent Substrate kit (Pierce Biotechnology Inc., Rockford, IL, USA). All assessments were performed in duplicate.

Neurospheres expanded in suspension or grown on laminin were fixed in 3.7% formaldehyde solution in PBS for 20 min. Following extensive washes in PBS, neurospheres were cryoprotected for 24 h with 15% sucrose in PBS + 0.001% NaN₃, flash-frozen, and serially sectioned (10 μm thickness) using a cryostat (Leica CM1900), Immunocytochemistry was carried out as previously described 38. Antibodies are listed in [see Additional file 6] and were diluted in 3% bovine serum albumin + 0.3% Triton X-100 in PBS. Hoechst 33258 (1 μg/mL) was used as a nuclear counterstain. For suspension cells, the number of connexin⁺ cells was expressed as the percentage of Hoechst⁺ nuclei counted in serial sections of 10–50 neurospheres. For cultures plated on laminin, immunocytochemistry was performed directly on laminin-coated glass-coverslips. Connexin⁺ cells/Hoechst⁺ nuclei were established in five fields per neurosphere calculated for 5–10 neurospheres per connexin (n = 25–50 fields per condition). Because cell density in the sphere core was such that individual cells could not be quantified or verified to be in contact with the laminin substrate [see Additional file 7, area outlined in black], cell counts were performed over five defined peripheral cell fields shot at 40× magnification [see Additional file 7]. All cell images were taken with a Leica DMXRA2 epifluorescence microscope and analyzed using Openlab software v5.05 (Improvision, Lexington, MA, USA).

Cultures were expanded as neurospheres for 14 DIV before being suspended in 2% formaldehyde in 10 mM PBS and triturated to achieve a single cell suspension. Following two PBS washes and gentle centrifugation at 2000 rpm for 5 min, NPCs were suspended in PFN (2% fetal bovine serum, 0.1% NaN₃, 0.18% saponin in 10 mM PBS). Cell suspensions were separated into 1.5 ml microfuge tubes at a concentration of 1 × 10⁶ cells per tube. Primary or isotype control antibodies were added directly to cell suspensions at appropriate concentrations followed by incubation at room temperature for 30 min on a shaker. Each sample was washed twice by the addition of 0.8 ml of PFN and gentle centrifugation. Secondary antibodies were added to cell suspensions at appropriate concentrations followed by incubation at room temperature for 30 min on a shaker and washed as described above. Cells were resuspended in 0.5 ml of PFN and analyzed using a Beckman Coulter FC500 Flow Cytometer (Beckman Coulter Canada Inc., Mississauga, ON) and Beckman Coulter CXP software.

Dye uptake and dye transfer assays were assessed on 10 neurospheres/condition in two separate experiments per condition. For quantification, four fields were photographed per sphere. For dye coupling assays, photomicrographs represented two fields within the core of the neurosphere and two fields at the periphery where cells were in direct apposition and thus potentially coupled. To image cells within the core in direct contact with matrix, micrographs were obtained using a Leica DMIR epifluorescent inverted microscope equipped with a QICAM digital camera (Quorum Technologies, Guelph, ON, Canada) and captured using OpenLab software v5.05. For hemichannel assays, all four fields were shot along the periphery and were taken with a Leica DMXRA2 epifluorescence microscope and analyzed using Openlab software (v5.05). Cultures expanded in suspension as neurospheres for 8 DIV were plated on coverslips in expansion media containing (a) EGF and FGF-2 for 1 DIV to assess function prior to laminin-induced changes in connexin expression, (b) EGF and FGF-2 for 6 DIV to assess connexin function after laminin-induced changes, or (c) in differentiation media for 6 DIV supplemented with RA (0.5 μM) and FBS (0.5%) to promote glial differentiation. Hemichannel activity was assessed by comparing uptake of connexin channel-permeant Lucifer yellow (LY, 457 Da) and channel-impermeant rhodamine B isothiocyanate-dextran (RD, 10,000 Da) in the presence or absence of 50 μM flufenamic acid (FFA) or 100 μM 4,4'-diisothiocyanatostilbene-2,2'-disulfonic acid (DIDS) as previously described 26. Mechanical stimulation with glass microbeads was used to trigger hemichannel opening 2628. Cells positive for both LY and RD were excluded from the measurements to control for LY uptake resulting from loss of membrane integrity. For all assays, the number of LY⁺/RD^- cells is expressed as a percentage of the total number of cells per microscopic field ± standard error of the mean (SEM). Dye transfer between cells indicative of GJIC was assessed by scrape loading in the presence or absence of the intercellular channel blocker 18α-glycyrrhetinic acid (GRA, 100 μM) or its inactive analog glycyrrhizic acid (GZA, 100 μM) as previously described 26. Dye transfer was quantified by counting the number of LY⁺/RD^- cells emanating from a LY⁺/RD⁺ cell adjacent to the scrape line. Data are expressed as the mean number of LY⁺/RD^- cells ± SEM.

Data were analyzed by analysis of variance (ANOVA) followed by post hoc Tukey tests or Student's t test where appropriate.