Standard cultivation conditions for cell cultures comprise the use of 20% oxygen, nevertheless a number of studies have described an enhanced proliferation in lowered oxygen. Reducing oxygen can have a number of different effects such as the increase of proliferation as shown by Zhao et al.  and Studer et al.  for rat embryonic mesencephalic cells, or conversely a decrease of proliferation as described by Chen et al.  who showed that long-term proliferation in hypoxia was not beneficial for hESC with short splitting intervals. Studer et al.  investigated the proliferation and differentiation of embryonic mesencephalic rat cells and came to the conclusion that hypoxia was beneficial for the cells in culture and that EPO could mimic this effect under normoxic oxygen levels. Recently Santilli et al.  described an increased proliferation though the cell cycle remained unaffected as well as an increased neuronal differentiation and decreased cell death of human neural stem cells caused by mild hypoxia. The effects of lowered oxygen on the proliferation of stem and progenitor cells are not limited to the central nervous system [7, 19]. More physiological culturing conditions are also favoured by other cell types like bone marrow stromal cells  and mesenchymal cells .
As a first step in this study we verified the expression of HIF-1α (Figure 1) and the EpoR (Figure 2). The sensibility of the hNPCs to hypoxic conditions is indicated by the expression of HIF-1α (Figure 1). A similar effect was observed by Zhou and Miller , Zhao et al.  and Zhang et al.  ranging from 30 minutes to 24 hours after the onset of hypoxia. HIF-1 is activated under hypoxic conditions in a variety of cell types and the HIF-1 targeted genes play an important role in maintaining cellular homeostasis in response to hypoxia [22–24]. To investigate the EpoR we chose western blotting as the currently available antibodies lead to inconclusive results obtained by immunocytochemistry . The EpoR expression level was not altered by culturing the cells under EPO application or hypoxic conditions, the latter being in line with the absence of a hypoxic EPO effect. Even though this is contrary to Theus et al.  where hypoxia led to an increase in the EpoR expression, Milosevic et al.  likewise observed that hypoxia does not affect EPO signaling. This inconsistency could be due to different culturing conditions or cell types. The effect of EPO on the metabolic activity and apoptosis is independent from the regulation of expression of its receptor since the expression levels are not altered between different stages of proliferation or differentiation, as well as EPO treated cells. In summary, we conclude that the differentiation of the human NPCs used in this study as a model system is hypoxia-sensitive and EPO-responsive.
Both hypoxia and EPO have been reported to induce the proliferation of NPCs [3, 5, 7], though our own data support the idea that hypoxia does not change the proliferation rate or doubling times within three days of expansion (Figure 3). Similar results were obtained by Chen et al.  where lowered oxygen levels did not prove to be favourable, and by Milosevic et al.  who described a positive effect of hypoxia on the proliferation only after culturing NPCs for 1 month, but not prior to that. In addition, EPO did not affect proliferation although the EpoR could be detected in proliferating cells and 10 IU/ml EPO seems to lead to an increased proliferation though this effect was not significant compared to the control. However, higher amounts of EPO could be saturating and thus lead to no effect, either.
The differentiation of the hNPCs was investigated under various conditions. First, the metabolic activity of differentiating hNPCs was monitored with and without EPO treatment (Figure 4). An effect of EPO was detected early in 1 day differentiated cells. Remarkably at 3% oxygen, EPO was required at higher concentrations to produce an equivalent effect. This indicates that hypoxia acts only in part via the EPO pathway and that addition of EPO mimics the effect of lowered oxygen. Generally one can say that hypoxia increases the metabolic activity of hNPCs, which was highest at 1 d of differentiation, indicating the importance of early differentiation processes, as the effect at day 3 was not as high as at day 1. These data are in accordance with Studer et al.  where EPO mimicked the effect of hypoxia under normoxic conditions in embryonic mice NPCs.
For further investigation of the differentiation, the cell cycle of the hNPC was analysed under normoxic and hypoxic conditions (Figure 5). This analysis revealed that the cells needed around 20 h to enter G1 phase, and that this time frame is the same under normoxic and hypoxic conditions. These findings are in line with data about the cell cycle of murine midbrain NPCs where the cell cycle, the proliferation and neurosphere formation was not altered within 4 weeks of cell culture . Similar results were obtained by Santilli et al.  who likewise demonstrated no effect of hypoxia on the cell cycle of human NSCs. These results are of major importance to further interpret the expression levels of βIII-tubulin as a marker for neuronal differentiation.
In this study EPO did not alter neuronal differentiation in the hNPCs (Figure 5). This is in contrast to rat and human mesencephalic progenitors where EPO enhanced the number of neurons [3, 7]. A possible explanation for this discrepancy could be the fact that different model systems have been used. The percentage of neurons in our study was increased after culturing the cells under hypoxic conditions. This is in accordance with Zhang et al.  and Studer et al. , where hypoxic culturing conditions also led to a higher yield of neurons. Since the cells enter differentiation at the same time point under normoxia and hypoxia the higher yield of neurons is not due to an accelerated cell cycle, leading to the conclusion that hypoxia induced neuronal differentiation in the precursor cells.
An increase in neurogenesis could be obtained by two different mechanisms - one during proliferation and the other during differentiation - partially mimicked by EPO. First, culturing differentiating NPCs under lowered oxygen increased the number of neurons after 3 days of differentiation. In addition, proliferation of NPCs under hypoxia and differentiation of those cells under hypoxic or normoxic conditions raised the same amount of neurons, indicating a manipulation of the progenitor cell pool during proliferation. EPO partially mimicked the effect under normoxia and displayed anti-apoptotic effects under these culturing conditions (Figure 5E). Therefore we propose two different mechanisms of differentiation. One deals with the increase of neuronal cells by hypoxia during differentiation and the other one displays an increase of the progenitor pool of cells during proliferation under hypoxia. The two mechanisms result in the same effect, namely the increase of neuronal cells and the increase of the overall activity of differentiated cells. The first mechanism indicates that hypoxia induces differentiation and the second one indicates that hypoxia increases the pool of differentiating cells by changing the cell-fate of the progenitor cells. Proliferation was investigated at 3% O2 and the rate of differentiation did not change when cells were differentiated at 3% as well. These results demonstrate that 3% oxygen modifies the differentiation capability of NPCs.
The cell line used in this study showed a maximal number of neurons of around 6% (Figure 5), which can be interpreted as a limitation of this study, however reported levels of neurons in other NPC lines are similar. Nevertheless, this cell line also possesses advantages like the very fast differentiation potential and the easy accessibility, which enabled us to closely monitor changes in proliferation and differentiation. Therefore, those cells serve as a model to investigate differentiation mechanisms which then can be transferred to systems which allow for an engraftment into the CNS to cure neurodegenerative diseases like Parkinson's disease or stroke.
Concerning apoptotic cells, the number was reduced by 50% at day 4 of differentiation at 3% oxygen (Figure 6). This apoptotic effect was not in consensus with a neuronal cell death, as the number of neurons was not influenced which leads to the conclusion that the number of βIII-tub+ cells at 3 days of differentiation is not only an outcome of an anti-apoptotic effect. At the fourth day of differentiation the effect of EPO is anti-apoptotic, but numbers of neuronal cells are not altered by EPO and therefore EPO has no neuron-specific anti-apoptotic effect. We observed an increased apoptosis at day 4 in the cells that underwent proliferation and differentiation at 20% oxygen, however the underlying mechanism is not clear. Depending on the severity of hypoxia it can have differential effects on the apoptosis. On the one hand it was proven to be anti-apoptotic in a model for hypoxic-preconditioning , on the other hand it can be pro-apoptotic if it is lowered beyond levels of mild hypoxia [5, 18]. In our study, the anti-apoptotic effect of hypoxia was also indicated by the expression of the anti-apoptotic protein bcl-2. The western blot of bcl-2 revealed an increase between day one and two of differentiation, followed by a stable expression level (Figure 6). Shingo et al.  showed an increase of neurons induced by hypoxia. This enhancing effect was mimicked by EPO, as it promoted the production of neuronal progenitors. This is contrary to our results, as EPO could not manipulate the neuronal-producing effect of hypoxia, but did mimic other effects of hypoxia, like the anti-apoptotic effect during differentiation. The percentage of cells rescued by EPO (18.80 ± 2.27%) at 20% oxygen was not significantly different from the amount of cells rescued by hypoxia (15.82 ± 4.65%) proving that EPO has the potential to imitate hypoxic effects under normoxia.
Contrary to Studer et al.  and Shingo et al. , EPO did not completely mimick the actions of hypoxia in our study. In this study, a human fetal cell line was used whereas Studer et al. and Shingo et al.  used mouse embryonic stem cells. This leads to the conclusion that either the point in time (fetal vs. embryonic) or the origin (human vs. mouse) can account for the observed differences. In addition, the application of human recombinant EPO to murine cells might lead to different results than in the human system. And finally, the oxygen concentration can also influence the outcome as shown by Zhang et al.  and Horie et al. . Both tested varying oxygen concentrations ranging from 0% to 10% and found 2% to 3% oxygen to be most effective.
For translational and clinical research our findings are important because we provide further evidence of increased neurogenesis in hypoxic scenarios. The cell survival and ideal environmental oxygen after engraftment of hNSC remain yet unclear and our data supports the thesis that a hypoxic environment, as seen in stroke or other neurodegenerative diseases, are beneficial for engrafted hNSC. Furthermore we were able to provide evidence that hypoxia could induce neurogenesis during proliferation and differentiation, thus the engrafted cells would not have to be used at a certain point in time during the cell cycle and therefore making the engraftment process easier. Researchers have tried to profit from EPO as a neuroprotective agent  in patients with stroke  but it remains unclear how EPO acted neuroprotective. There are three main theories of EPO action in the human brain. The first presumes a better oxygenation of the brain through an elevation of red blood cells after EPO application, the second assumes EPO effects on astrocytes and blood vessels and indirectly affecting neurons and the third theory actually proposes a neuroprotective effect of EPO . We provide supporting evidence for the last theory, which encourages the use of EPO in stroke. As EPO mainly acts in hematopoiesis and can thus cause hematopoietic side effects, the neuroprotective effect we explored for hNSCs should be further and directly exploited by derivatives of EPO, which are non-hematopoietic, neuroprotective and able to pass the blood brain barrier easily. Such structural as well as functional variants of EPO  that fulfil these requirements, among them modified antibody fragments  and peptides , have been described recently.