The addition of exogenous ROS and a higher endogenous ROS state enriches for self-renewing neural stem and progenitor cells in clonal neurosphere cultures from different species and developmental ages
The addition of low, non-toxic concentrations of hydrogen peroxide (H2O2) to culture media produced a large increase in multipotent--capable of producing neurons, astrocytes and oligodendrocytes--clonal density neurosphere formation over multiple serial passages (; p=0.001). There was a more modest increase in overall cell proliferation (Supplemental Figure 1A
). Exogenous ROS had a similar stimulatory effect on clonal neurosphere formation in embryonic and adult mouse and fetal human neurosphere cultures (; p<0.001).
Hematopoietic stem cells have relatively low levels of endogenous ROS (Jang and Sharkis, 2007
). To determine whether neural stem cells were also low-ROS cells, we used FACS and the ROS-sensitive dye DCFDA to separate cells into ROShi (top 10%) and ROSlo (bottom 10%) populations and assessed their serial clonal density neurosphere forming capacity. The ROShi population contained almost all of the multipotent sphere-forming cells in primary and secondary clonal cultures (p<0.001; ). We replicated this finding using multiple different ROS-sensitive dyes (see Supplemental Figure 1B
). In addition, a high-ROS status provided an enrichment in clonal neurosphere formation compared to unselected (US), sorted cells from the same sample. ROSlo cells only formed primary clonal neurospheres and therefore displayed a limited capacity for self-renewal. Culture and resorting of sorted cells demonstrated that ROShi cells gave rise to both ROShi and ROSlo cells in secondary neurospheres but ROSlo cells were not capable of giving rise to ROShi cells (). Consistent with these results in murine cells, we also observed that the ROShi population in human ES-derived neural progenitors had a greater proliferative capacity compared to ROSlo or unselected cells from the same sample (p<0.001; ).
Elevated ROS levels and NOX signaling are associated with increased NSC enrichment
Serial clonal density neurosphere formation (self-renewal), sphere diameter (proliferation), and multipotency were assessed in the ROShi cells compared to unselected cells from the same samples over multiple passages. ROShi cells were highly enriched for clonal neurosphere forming cells at all passages although a gradual decrease in this enrichment was observed (p<0.001; ). There was also an initial significant increase in neurosphere diameter (p<0.05), but this returned to control levels with successive passages. ROShi spheres maintained a high level of multipotency over serial clonal passages. In agreement with our data utilizing exogenous ROS stimulation, a high endogenous ROS status was also associated with a greater positive effect on self-renewing divisions than on overall proliferation.
Neural stem cells are associated with a high-ROS status and NOX is a significant endogenous source of cellular ROS regulating NSC function in low oxygen conditions
We next sought to identify differences in cellular phenotypes between the ROShi and ROSlo populations since they displayed different capacities for long-term clonal self-renewal. Therefore, we sorted primary adult SVZ cells for 3 different neural stem cell-enriching marker sets: 1. EGFR+GFAP+CD24- cells (Pastrana et al., 2009
) 2. ID1+GFAP+ cells (Nam and Nam, 2009
), and 3. Lex (SSEA1)+ cells (Capela & Temple et al., 2002
) and evaluated their relative endogenous ROS levels. We found that the enriched populations maintained significantly elevated endogenous ROS levels compared to the negative, non-NSC enriched populations from the same samples in each case, indicating that ROShi fraction contains the neural stem cell fraction (p<0.001; ). The “stem cell astrocytes” (EGFR+GFAP+CD24- cells) had 48% higher ROS levels than the (EGFR+GFAP-CD24-) transit amplifying cells and approximately 200% more than the EGFR-negative niche astrocyte-containing fraction of cells. The ID1+GFAP+ cells also had 57% higher ROS levels than the GFAP-negative cells.
Clonal neurosphere formation was greatly enhanced by the addition of exogenous ROS to the stem cell-enriched fractions derived from mouse SVZ, while the stem cell-negative fractions had limited or no response, an effect that was inhibited by the NOX inhibitor apocynin. (Supplemental Figure 2A
When cells were sorted directly from the SVZ according to their ROS status and then analyzed for other markers, we found no differences in the expression of Dlx2 in ROShi compared to ROSlo cells, while Mash1 positive cells were enriched in the ROSlo fraction (). These data suggest that transient amplifying cells are not responsible for differences observed in neurosphere formation between the two populations. The ROShi population was significantly enriched for cells expressing nestin and doublecortin (DCX). No significant differences in Sox2 or GFAP expressing populations were observed.
The previous experiments were performed under room oxygen conditions. However, low oxygen conditions are known to stimulate NSC self-renewal (Studer et al., 2000
). We found that low, physiological oxygen conditions (4% O2) resulted in elevated endogenous ROS levels (Supplemental Figure 2B
), consistent with findings of others in different cell models (Guo et al., 2008
), increased clonal neurosphere formation (; p<0.001), and up-regulation of the NOX2 homologue (; p<0.01). Conversely, lowering endogenous ROS levels in the low oxygen cultures through NOX inhibition eliminated the positive effects of hypoxia and resulted in decreased clonal density neurosphere formation (). These data suggest that the enhancement of self-renewal by lower oxygen concentrations is at least partially mediated through enhanced NOX activity which in turn leads to elevated ROS levels.
ROS augments growth and trophic factor stimulation and is required for normal NSC self-renewal
We next wanted to determine if NOX-generated ROS played an important role in facilitating growth factor signal transduction. To do this we placed neurosphere-derived cells in low concentrations of EGF and bFGF which led to a marked reduction in neurosphere formation (). However, clonal neurosphere formation could be restored to levels observed with high growth factor concentrations by supplementing low growth factor conditions with exogenous ROS (p<0.001; ). The addition of exogenous ROS to the low-growth factor cultures elevated intracellular ROS levels to those observed in the high growth factor conditions (). No clonal neurospheres were formed in cultures without any growth factors even with the addition of exogenous ROS (data not shown), indicating that ROS on its own is not sufficient to replace growth factor-initiated signaling.
Reactive oxygen species are required for stimulation of normal neural stem cell self-renewal
Since the addition of small amounts of ROS resulted in a gain-of-function we next investigated the effects of a loss-of-function in NOX signaling. In growth factor-supplemented SVZ neurosphere cultures we found that NOX inhibition (DPI) significantly decreases clonal neurosphere formation but this inhibition can be rescued by adding exogenous ROS (H2O2) back to the culture medium. We also observed that cells derived from the SVZ of NOX2 mutant mice had significantly lower endogenous ROS levels (Supplemental Figure 3A
) and subsequently displayed a significantly diminished NSC self-renewal and multipotency over serial clonal passages (; p<0.01). Mutant neurospheres produce approximately 29% more glial-only spheres (astrocytes and oligodendrocytes) compared to wild-type cultures (Supplemental Figure 3B
). Clonal neurosphere formation and multipotency in the NOX2 mutants could also be significantly rescued by the addition of exogenous ROS (H2O2) in the mutant cultures ().
Because NOX has been implicated in both growth factor and neurotrophin signaling, we next examined whether it may play a role in the proliferative effects of brain derived neurotrophic factor (BDNF) on neural stem and progenitor cells (Islam et al., 2009
). In the presence of standard concentrations of NSC growth factors (EGF and bFGF), we observed that BDNF could significantly increase clonal neurosphere formation. Therefore, we used inhibition of NADPH oxidase or treatment with the antioxidant N-acetyl-cysteine (NAC) in order to determine that NOX signaling played a significant role in the positive effects of BDNF on clonal neurosphere formation (p<0.001; ). In addition, we demonstrated that endogenous superoxide (the ROS species produced by NOX) was increased upon BDNF treatment which could be blocked by NOX inhibition (p<0.001; ). However, BDNF was not able to stimulate NSC self-renewal in cells derived from NOX2 mutant mice but was stimulatory only to wild-type cells (p<0.01; Supplemental Figure 4A
), suggesting that NOX-dependent signaling plays a significant role in the stimulatory effects of BDNF on neural stem and progenitor proliferation.
ROS augment trophic factor signaling and are dependent on the PI3K/Akt signaling pathway for their effects
NOX regulation of neural stem and progenitor cells is dependent on PI3K/Akt/mTOR signaling
Previous studies have suggested that ROS can activate the PI3K/Akt/mTOR pathway through the reversible inactivation of the PTEN protein (Kwon et al., 2004
; Leslie, 2006
). Consistent with this we found in neurospheres that the addition of stimulatory concentrations of H2O2 induced direct oxidation of the PTEN protein (). To more directly assess the requirement for PTEN expression in the mechanisms underlying the stimulatory effect of ROS, we used cells derived from PTEN- deficient, PTEN heterozygous, and wild-type mice (Groszer et al. 2001
), demonstrating that the addition of exogenous ROS is not capable of stimulating the PTEN-deficient cells (P<0.001; ). As would be predicted from this model, ROS stimulated clonal neurosphere formation in heterozygous cells to a greater extent than it did wildtype cells, (; p<0.01). Likewise, inhibition of NOX resulted in dramatically reduced clonal neurosphere formation in WT but not in PTEN-deficient cells (Supplemental Figure 4B
). Finally, BDNF stimulation of clonal neurosphere formation was also only observed in wild-type but not in PTEN-deficient cells (see Supplemental Figure 4A
We examined activation status of key downstream nodes of the pathway. Exogenous ROS (H2O2 and Gox) enhanced, while inhibition of endogenous NOX-generated ROS with DPI inhibited the phosphorylation of Akt (). Furthermore we observed increased phospho Akt (pAkt) in the ROShi compared to the ROSlo population of cells, increased pAkt in BDNF-treated neurosphere cultures, and increased pAkt following the addition of H2O2 into low-growth-factor conditions media (). We observed similar results from flow cytometry analysis of S6 phosphorylation (). In addition, the ROShi population from human ES-derived neural cells also had elevated pAkt and pS6 activation (Supplemental Figure 1C
Pharmacological experiments also support a role for the PI3K pathway. The effects of exogenous ROS on neurosphere formation were abolished by the PI3K inhibitor LY294002 (LY), suggesting that exogenous ROS do not exert their effects by either bypassing the pathway or by providing enough stimulation downstream of PI3K to overcome this inhibition. Since ROS can also mediate effects via activation of the MAPK pathway, we compared the relative effects of LY and the ERK inhibitor U0126 in ROShi and unselected cells (). In both cases, pathway inhibition had a greater effect on the ROShi compared to unselected cells. However, LY had a much greater inhibitory effect on the ROShi cells than the U0126, indicating a greater dependence of these cells on the PI3K pathway, than on the MAPK pathway. Acute LY treatment inhibition significantly decreased endogenous cellular ROS levels (Supplemental Figure 4C
) in agreement with our hypothesized pathway for NOX signaling in neural stem cells ().
Cellular ROS levels influence neurogenic potential
Conditional deletion of PTEN results in both enhanced NSC self-renewal and a sustained increase in neurogenesis (Groszer, et al., 2001
; Gregorian et al., 2009
). Therefore, we determined whether ROS stimulation of PI3K/Akt signaling had similar effects on neurogenesis. Treatment of clonal density cultures with low, non-toxic levels of exogenous ROS during sphere formation produced significantly higher numbers of neurons as a percentage of total cells when differentiated in the presence of standard conditions (p<0.001; ). However, treatment of cells with the same exogenous ROS during differentiation resulted in increased cell death and few, if any, neurons were produced (data not shown). Conversely, inhibition of NOX or inhibition of PI3K (LY294002) prior to differentiation significantly reduced neuron numbers (P<0.01; ). In combination with exogenous ROS stimulation, inhibition of the PI3K pathway (LY294002) eliminated the positive effects of ROS on neurogenesis (p<0.001; ). In agreement with our data demonstrating that NOX inhibition decreased neurogenesis we found that neurosphere cultures derived from NOX2 mutant mice produced significantly fewer neurons as well (p<0.01; ).
ROS stimulation during mitogenic expansion enhances neurogenesis in a PI3K-dependent manner
NOX-generated ROS regulates SVZ proliferation and neurogenesis in vivo
We next tested whether our ex vivo findings extend to an in vivo stem cell system. To this end, we tested the effects of the NOX inhibitor apocynin (Apo) on SVZ proliferation. We first assessed the effects of Apo treatment on endogenous ROS levels using the in vivo ROS-sensitive dye, hydroethidine (HEt). Even in control (vehicle-treated) animals, the SVZ had significantly higher ROS levels than surrounding brain tissues such as the striatum and cortex (p<0.01; ). The SVZ also had approximately 8-fold enriched expression for the NOX2 homologue compared to neighboring cortical tissue (p<0.001; ). The 3 week Apo treatment resulted in a significant reduction in SVZ ROS levels (p<0.01; ) and in the number of Ki67 (proliferative) cells within the SVZ (p<0.02; ). Cells acutely dissociated from the SVZ of mice similarly treated with Apo in vivo produced significantly fewer clonal neurospheres in primary cultures compared to vehicle-treated mice (p<0.01; ), indicating decreased neural stem or progenitor cell numbers. However, this deficit recovered in subsequent serial clonal passages, demonstrating that although APO administration acutely inhibited proliferation in vivo, the competency for self-renewal in the SVZ-derived cells was not affected.
In vivo inhibition of NADPH oxidase by Apocynin decreases SVZ proliferation, endogenous ROS levels, and NSC self-renewal
Consistent with our observations on apocynin-treated animals, we found that NOX2 mutant mice also had diminished numbers of Ki67 (proliferating) cells within the SVZ compared to wildtype mice (p<0.03; ). NOX2 mutant and wild-type mice were pulsed with BrdU followed by a 4 week wash-out period during which time labeled SVZ cells would be expected to leave the SVZ and migrate through the rostral migratory stream to the olfactory bulb where they normally differentiate into post-mitotic neurons. We found that a larger number of BrdU positive cells remained within the SVZ of mutant mice, whilst there were also fewer BrdU+ cells in the olfactory bulb of mutant mice and fewer new neurons (BrdU+/NeuN+) produced there (p<0.01; ). As a result of this defect in cell proliferation and possibly also in migration we observed that the granule cell layer of the olfactory bulb in mutant mice was smaller than those of wildtype mice (p<0.05; ).
In vivo SVZ proliferation and neurogenesis are significantly impacted by changes in cellular ROS
Using flow cytometry analysis of acutely dissociated SVZ, we found that the NOX2 mutants have more immature progenitor cells (nestin+ and Sox2+) and fewer cells expressing markers for neuroblasts (DCX) or transit amplifying cells (Mash1 and Dlx2; ). Although these data suggest an increase in some progenitor cells, our in vitro findings indicate a diminished capacity for the generation of clonal, serially passagable neurospheres, suggesting a diminished number of neural stem cells in NOX2 mutants. Therefore, the ex vivo cell phenotypes we have observed indicate that there may also be defects in cell maturation and differentiation.
In addition to the negative effects on NSCs caused by decreased NOX activity, we have also conversely demonstrated that increased NOX activity in vivo
can have stimulatory effects. Systemic administration of a low, non-toxic dose of the neuroinflammatory stimulus, lipopolysaccharide (LPS), resulted in a significant enhancement in SVZ proliferation (p<0.001; ) whilst inhibition of NOX activity by Apo co-treatment eliminated the stimulatory effects of LPS on SVZ proliferation (p<0.03; ). Although neuro-inflammatory cells are likely play a role in this effect in vivo
, low dose LPS stimulates NSC self-renewal in vitro
which is also blocked by NOX inhibition and antioxidant treatment (Supplemental Figure 5