N2 and B27 culture supplements were purchased from Gibco (Paisley, UK). Epidermal growth factor (EGF) and fibroblast growth factor-2 (FGF-2) were obtained from R&D Systems (Abingdon, UK). Accutase was purchased from Millipore (Chandlers Ford, UK). The integrin β1-stimulatory antibody, TS2/16, was purified from the hybridoma culture supernatant (LGC-Promochem, Teddington, UK) by incubation with Protein G beads (GE Healthcare, Little Chalfont, UK) overnight at 4°C, before being eluted according to the manufacturer's instructions. Unless already sodium azide-free, antibodies used in functional studies were dialysed overnight at 4°C in phosphate-buffered saline (PBS) using sterile Slide-a-lyzer cassettes (Pierce, Cramlington, UK). The sterile mouse IgG₁ isotype control (clone W3/25) was purchased from Serotec (Oxford, UK). Human placental laminin (hereafter referred to as "laminin") was purchased from Sigma-Aldrich, as were all other reagents (Poole, UK).

Human fetal tissue (8–10 weeks postconception) was collected in accordance with the arrangements for informed consent recommended by the Polkinghorne Committee (1989) and the UK Department of Health guidelines (1995), and with the approval of the Addenbrooke's Hospital Medical Research Ethics Committee (Cambridge, UK). The hNSC from the cortex were propagated as free-floating aggregates (neurospheres) and cellular expansion prior to use in the studies described here was achieved using a passaging method described in detail elsewhere 25. In brief, fresh tissue was initially dissociated in trypsin and seeded at a density of 200,000 cells/ml in a T75 flask containing Dulbecco's Modified Eagle's Medium (DMEM)/HAMS F12 (3:1), supplemented with N2 (1:100), EGF, FGF-2 (both at 20 ng/ml) and heparin (5 μg/ml). Cultures were fed every 4–5 days by replacing half the medium and spheres were passaged after 14 days expansion by cutting into 200 μm sections using a McIlwain tissue chopper.

Human neurospheres were dissociated with accutase before live cells were sorted using a MoFlo flow cytometer (DakoCytomation, Ely, UK) into 96-well plates (Nunc, Roskilde, Denmark) at a density of 500 per well (1,500 cells per cm²). In the experiments using a limiting dilution assay to determine the frequency of neurosphere-forming cells, progressively decreasing numbers of cells were plated per well. Neurosphere medium (200 μl), containing EGF, FGF-2 (both at 20 ng/ml), heparin (5 μg/ml) and 2% B27 together with laminin or antibodies at the concentrations indicated in the figure legends, was added per well. Prior experiments examining the effects of different laminin concentrations on mouse neurosphere numbers showed no significant increase above 10 μg/ml (see Additional file 1 and Additional file 2), so this concentration was used throughout for the studies on human cells. Cells were fed every 4–5 days by replacing half of the medium, and the number of neurospheres formed was counted after 21 days in culture. For the purpose of this paper, these are termed primary neurospheres.

Secondary neurospheres were grown by collecting together primary neurospheres from a particular treatment group, dissociating them with accutase before re-plating at the same density using the flow cytometer in fresh medium in the absence of any antibodies or laminin. All other details are as for primary spheres.

Conditioned medium had been exposed for 5 days to neurospheres growing in a T75 flask (see 'human cell culture'), before being passed through a 0.45 μm filter to remove cell debris. The percentage refers to the amount of conditioned medium added; fresh DMEM/Hams F12 constituted the remaining volume. An appropriate quantity of growth factors (20 ng/ml EGF and FGF-2, 5 μg/ml heparin) and supplements (1% N2 or 2% B27, as indicated in the figure legends) were added according the final volume.

Human neurospheres were dissociated using accutase and grown for 24 hours in medium ± laminin (10 μg/ml). In order to examine apoptosis, cells were incubated with fluoroscein isothiocyanate (FITC) conjugated annexin V for 15 min at room temperature in binding buffer (Annexin V-FITC apoptosis detection kit I, BD Pharmingen), before being analysed on at FACScan II flow cytometer (Becton Dickinson, Oxford, UK), with at least 10000 events being collected per data file. Further analysis was performed using Summit 3.0 software (DakoCytomation).

For all neurosphere assays, 8 experimental replicates were performed for each tissue sample. The n value recorded in the figure legends refers to the number of biological replicates. The data are presented as mean ± standard error. All statistics were calculated according to the test described in the figure legends, with p < 0.05 considered significant.

To test our hypothesis as to the role of exogenous laminins, we needed to develop an assay that enables single dissociated cells to reveal any stem cell properties in culture. Neurospheres provide a very widely used method to grow NSC, with each neurosphere representing a 3D aggregate generated from a single NSC and containing a mixture of NSC and more differentiated cells 26. Human NSC are commonly grown in N2 medium containing 20 ng/ml FGF2 and EGF, and are passaged by chopping the spheres into segments to be placed in fresh medium. However this approach in which large fragments are passaged does not allow accurate quantification of the effect of laminins or other manipulations on single NSC. First, therefore, we dissociated spheres passaged by chopping into single cell suspensions and re-plated them in the same medium at 1 cell/well. No neurospheres formed in these cultures (not shown). Moreover, even when higher cell plating densities of 2000, 1000 or 500 per well (10, 5 or 2.5 per μl) were examined, extremely low neurosphere yields were observed, with only 0.036 ± 0.009%, 0.029 ± 0.007% or 0.063 ± 0.050% of the plated cells forming neurospheres, respectively (not shown).

To ask whether these dissociated human NSC had the potential to form larger numbers of spheres given more appropriate growth conditions, we attempted to mimic the much higher cell densities present in chopped spheres by the addition of conditioned medium to the culture. 50% conditioned medium added to N2/FGF2/EGF led to a 22-fold increase in the number of primary neurospheres which formed when cells were plated at 500/well (Fig. 1A; p < 0.001, n = 2). None of the other concentrations of conditioned medium caused any significant change in neurosphere generation.

We conclude from this that the potential of the dissociated human cells to behave as NSC is much greater than as revealed by N2 medium, and that N2 does not therefore represent an appropriate choice of medium supplement for assays of human NSC behaviour and laminin effects. B27 supplement has been shown to increase the survival in culture of a range of CNS cell types, including rodent neural precursors, and is necessary to establish human neurosphere cultures from primary tissue, whereas N2 supplement is sufficient to maintain them 2527. Therefore, we next examined the effect of B27 supplement on the human neurosphere assay (Fig. 1A). When compared with N2 supplement, B27 led to a large and significant increase in primary neurosphere formation, with 3.775 ± 0.320% versus 0.013 ± 0.013% of cells forming primary neurospheres, respectively (p < 0.0001, n = 2). Critically, neurosphere-conditioned medium did not increase neurosphere formation in the presence of B27, with 50% conditioned medium actually leading to a decrease in neurosphere numbers (2.838 ± 0.149% versus 3.775 ± 0.320%, respectively. p < 0.05, n = 2). Consequently, these results show that B27-supplemented medium does enable human NSC to show stem cell properties in dissociated culture, and this was therefore used in all subsequent assays.

Prior to using neurospheres as a quantitative assay of stem cell properties, it was necessary to show that the numbers of spheres formed reflected the growth of single cells and not aggregation of two or more cells after plating. To do this, we used a limiting dilution assay (LDA) as an alternative method to the neurosphere assay for determining the frequency of the neurosphere-initiating cell in a population. By measuring the likelihood of sphere formation in progressively greater dilutions of dissociated cells, the LDA calculates the number of individual cells able to form spheres. The neurosphere-initiating frequency determined by the LDA was 2.91 ± 0.23% (Fig. 1B), which was not significantly different from the frequency calculated by the neurosphere assay (Fig. 1A; 3.78 ± 0.32%; p = 0.19, n = 3 and n = 2 for the LDA and the neurosphere assay, respectively). Therefore, as the two assays gave equivalent results, neurosphere assays were used in future experiments.

Having established the conditions for the human neurosphere assay, the effect of soluble laminin was investigated. After 21 days in culture, two-fold more primary neurospheres had formed in the presence of laminin compared to medium alone (Fig. 2A; p < 0.05, n = 2). When these primary neurospheres were dissociated and replated at the same clonal density but now in the absence of added laminin, no difference in the number of secondary neurospheres produced was observed (Fig. 2B; 8.63 ± 1.00 versus 7.00 ± 0.82 neurospheres per 500 cells for cells originally grown in medium or medium+laminin, respectively). Therefore, laminin increased primary neurosphere formation, but had no effect on secondary sphere growth. This shows that the relative numbers of hNSC was not increased by growth in laminin, suggesting that the effect of the laminin responsible for increased numbers of primary spheres is on NSC proliferation/survival.

The loss of attachment to the ECM usually leads to cells undergoing anoikis 28. Consequently, laminin in the culture medium may prevent cells from undergoing apoptosis, leading to increased primary neurosphere formation. In order to investigate this, human neurosphere cells were treated with laminin for 24 hrs before being stained for the presence of phosphatidylserine in the external leaflet of the plasma membrane, a marker of apoptosis, by FITC-conjugated annexin V (Fig. 2C) 29. The resulting data were normalised to the values seen with medium alone. Laminin led to a decrease in annexin V staining, with 73.69 ± 4.70% of cells being annexin V⁺ relative to medium alone (p < 0.01; n = 5). As such, laminin decreased apoptosis in cells dissociated from human neurospheres.

Integrins are the main receptor family for the ECM, with α3β1, α6β1, α7β1 and α6β4 mostly responsible for binding laminin 10. The role of integrin subunit β1 in mediating the effect of laminin on cell survival was investigated further for two reasons (Fig. 2C). Firstly, integrin subunit β4 is not expressed by human neurospheres 30. Secondly, in parallel experiments using the much faster-growing murine cells we found that integrin β1-blocking antibodies negated the laminin-induced increase in neurosphere numbers (see Additional file 3 and Additional file 1). In the absence of laminin, adding an isotype control or integrin β1-blocking antibody to the human cell cultures did not significantly alter the percentage of annexin V⁺ cells relative to medium alone (99.52 ± 9.44% and 115.90 ± 11.81%, respectively). However, in the presence of laminin, the integrin β1-blocking antibody led to a significant rise in the percentage of annexin V⁺ cells, increasing from 80.74 ± 2.03% in the presence of the isotype control antibody to 96.18 ± 0.94%, relative to medium alone (p < 0.05, n = 3). Therefore, integrin β1 is necessary for the laminin-mediated increase in survival.

Human laminins are commonly isolated from placenta 31. This presents three problems to its use by the biotechnology sector. First is the question of how to guarantee sufficient supply. Secondly, laminin placental preparations are heterogeneous. Depending on the purification protocol, they contain different laminins that are degraded to varying degrees, and in addition may contain fibronectin 32. Thirdly, there may be significant variation between batches from the same manufacturer 32. In contrast, the mouse Engelbreth-Holm-Swarm (EHS) tumour produces relatively pure and un-degraded type of laminin (laminin-111) 33. However, in light of the expression of non-human antigens by human embryonic stem cells exposed to animal-derived culture supplements 34, it is not suitable for growing cells to be used in therapies. An alternative, more homogeneous and scaleable solution is to use a monoclonal antibody to stimulate the laminin receptor. Given the finding that laminin-dependent survival was mediated by integrin β1, the question of sufficiency was addressed using the integrin β1-activating antibody TS2/16 35. The concentration of TS2/16 used (10 μg/ml) was sufficient to activate integrin signalling, as demonstrated by augmented integrin β1-dependent platelet adhesion and the increased phosphorylation of tyrosine 397 on focal adhesion kinase in human neurosphere cells (see Additional file 4 and Additional file 1) 3637. However, when added to the hNSC culture medium, no significant difference was observed in the number of primary (Fig. 3A) or secondary (Fig. 3B) neurospheres that formed, compared to an isotype control antibody. Therefore, β1 integrin activation alone was not sufficient to increase neurosphere formation.

Growth factor and integrin signalling demonstrate considerable cross-talk 36. For example, high levels of platelet-derived growth factor (PDGF) negate any differential effects of ECM substrates on oligodendrocyte proliferation 38. As all cultures thus far had been supplemented with 20 ng/ml EGF and FGF-2, we examined whether a lower concentration (2 ng/ml) would allow any effects on neurosphere formation of activating integrin signalling to be observed (Fig. 3C, D). However, no significant effects of the activating antibody were found at these lower growth factor concentrations, either for the number of primary (Fig. 3C) or secondary neurospheres produced (Fig. 3D).

The increase in human neurospheres in the presence of conditioned medium supports the observation that astrocyte-conditioned medium augments the proliferation of mouse NSC 394041. The presence of astrocytes within human neurospheres suggests that they could play a similar role 42. Whilst the identity of the secreted factors in human neurosphere-conditioned medium remains to be elucidated, candidates identified from rodent studies include cystatin C, galectin-1 and pigment-derived epithelial factor (PEDF) 4344.

We show here that the addition of B27 supplement also enables hNSC to generate neurospheres when plated at low, clonal densities. B27 supplement is chemically defined, containing the all ingredients present in N2 supplement, as well as additional factors. However, it is difficult to assess whether the effect of B27 is due to an individual factor, or the synergy of several. Interestingly, B27 also contains the antioxidant enzymes superoxide dismutase and catalase, which reduce oxidative stress and promote cell survival 4546. Although human NSC have been grown at low densities in other studies without B27 supplementation, this has been achieved by using the additional mitogen, leukaemia inhibitory factor (LIF), as well as EGF and FGF-2 47. LIF has been demonstrated to have a powerful positive effect on hNSC proliferation via the JAK-STAT pathway and it would be interesting to see if compounds in B27 acted through similar mechanisms 48.

Stem cell maintenance involves two issues: firstly, a proliferative/survival component, and, secondly, the preservation of the undifferentiated state 49. In this study, laminin led to an increase in primary neurosphere numbers, but had no effect on secondary sphere formation. This suggests that the effect of laminin was on NSC proliferation or survival, rather than promoting the undifferentiated (stem cell) state, as the latter would be predicted to increase the number of secondary neurospheres formed form dissociated primary spheres. In agreement, laminin was found to decrease apoptosis in neurosphere cells in an integrin β1-dependent manner. These findings support work done by Leone and colleagues, who used Cre-LoxP technology to excise integrin β1 from neural cells in vitro 50. They found that stem cell maintenance was unaffected by the loss of the integrin, whilst cell death was increased and proliferation was reduced. Preliminary proliferation data indicates that laminin does not alter the proportion of cells in S or G₂/M phase (as determined by propidium iodide incorporation and flow cytometry), whereas the addition of EGF/FGF-2 does increase this proportion. Therefore, whilst laminin-integrin β1 interactions mediate hNSC survival, a different ECM-integrin combination appears to regulate proliferation.

At least two waves of apoptosis occur during CNS development. Cell death is prominent amongst proliferating neural stem/precursor cells and neuroblasts present in the ventricular and intermediate zones in both rodents and humans 51. At later developmental stages, cell death mainly happens in post-mitotic neurons that have failed to innervate their targets 52. The importance of cell death in the stem/precursor populations is shown by knock-outs of various components of the apoptotic pathway in mice. These result in the malformation of the CNS, with enlarged ventricular zones and disruption of the cytoarchitecture 535455. As these alterations are apparent from E9.5–E10.5, before neurogenesis commences, they must reflect a requirement for regulated cell death in the neural stem/precursor cells as part of normal development.

The function of cell death at this early stage of CNS development is uncertain, although the correction of spatial/temporal errors, control of cell numbers or the selection for particular neuronal phenotypes have all been suggested (reviewed in 5657). The control of cell numbers is particularly important, as mutations that lead to the early loss of NSC during development cause microcephaly 58. Likewise, the upstream pathways are not widely understood, although there are links with known regulators of CNS development. For example, the receptor Patched signals for cell death in neuroepithelial cells unless it is bound by its ligand, Sonic hedgehog (Shh) 59. Similarly, signalling from receptor tyrosine kinases, via mitogen-activated protein kinase (MAPK) and Akt family members, also regulate apoptosis in the developing neuroepithelium 6061626364. This provides a potential link between the extracellular matrix and apoptosis, as integrins can activate MAPK pathways in rodent NSC 8.

Integrin β1 activation failed to reproduce the increased human NSC growth seen with laminin. There are three potential explanations. Firstly, integrin β1 can form heterodimers with subunits α1–11, and αV, which together can bind many ECM molecules including laminins, collagens, tenascins and fibronectin 10. The different ECM molecules have a different effect on NSC behaviour 2365. Therefore, it is possible that activating different integrin β1 heterodimers results in opposite effects, leading to no net increase in NSC growth.

Secondly, integrin signalling is determined by both affinity regulation, through conformational changes in the subunits, and by avidity regulation, in which integrins are clustered. Laminin may form a polymer, presenting multiple integrin binding sites and thus leading to both affinity and avidity changes 28. In contrast, it is likely that the activating β1 integrin antibody increased affinity only and did not cause integrin clustering. As it has been shown that both processes are required for effective outside-in signal transduction 66, the activating antibody may not result in as complete a signalling response as seen with laminin even though the blocking antibody shows the integrin to be necessary for this signalling.

Thirdly, laminins have been shown to bind to a range of receptors, including dystroglycan and syndecans 2867. Dystroglycan is expressed in both the VZ and cortical plate of the embryonic brain 9. However, mutant mice have defects in neuronal migration, consistent with the cobblestone (type II) lissencephaly phenotype seen in human congenital muscular dystrophy 6869. Therefore, it is unlikely that dystroglycan is involved in NSC regulation. However syndecans are also expressed in the VZ and co-operate with integrins in the adhesion of embryonic fibroblasts to fibronectin 97071. It remains to be examined whether such an interaction occurs in the response to laminin by neurosphere cells and whether syndecans and integrins together promote signalling in NSC in response to ECM.

The neurosphere assay we have used here represents a rapid and convenient method for assessing the number of neural stem cells within a population. However, it is not without its caveats. The main limitation is that cell types other than neural stem cells may proliferate and form spheres, based on the finding that hippocampal progenitors are able to form neurospheres which can be maintained for up to three passages 7273. Similarly, transit-amplifying type C cells in the adult brain can also produce neurospheres 7475. Consequently, counting the number of neurospheres is not necessarily the same as counting the number of stem cells present in a population. Therefore, the absolute numbers reported in this study may be an over-calculation, with an effect of laminin on the progenitors also possibly contributing to the increased numbers of spheres seen. In order to distinguish the growth of progenitors from NSC, maintaining neurosphere cultures for at least five passages has been suggested, although the exact number of passages required remains debatable 76. The best way to prove that any effect is directly on the stem cells would be an in vivo method such as the repopulating assay used for haematopoietic stem cells, but this has yet to be adapted for the nervous system.

The assay we have used also does not address the impact on multipotentiality of growing hNSC in laminin. The priming of hNSC with FGF-2/heparin/laminin prior to in vitro or in vivo differentiation resulted mainly in neurons forming, whereas the fate of unprimed hNSC was biased towards an astrocytic phenotype 77. This is consistent with the findings of other groups that laminin induces a neuronal fate 78. Moreover, laminin has also been implicated in altering the subtype of neuron formed. Human NSC primed as described above mostly formed cholinergic neurons, in addition to smaller numbers of glutamatergic and gamma-aminobutyric acid (GABAergic) neurons 77. Similarly, laminin has been found to increase the ratio of GABAergic to glutamatergic neurons compared to fibronectin 79. Therefore, it would be expected that the culture of hNSC in laminin would bias the cells towards a neuronal fate.