Distinctive effects of arachidonic acid and docosahexaenoic acid on neural stem /progenitor cells

Authors

  • Nobuyuki Sakayori,

    1. Division of Developmental Neuroscience, United Centers for Advanced Research and Translational Medicine (ART), Tohoku University School of Medicine, 2-1 Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan
    Search for more papers by this author
  • Motoko Maekawa,

    1. Laboratory for Molecular Psychiatry, RIKEN Brain Science Institute, 2-1 Hirosawa, Wako 351-0198, Japan
    Search for more papers by this author
  • Keiko Numayama-Tsuruta,

    1. Division of Developmental Neuroscience, United Centers for Advanced Research and Translational Medicine (ART), Tohoku University School of Medicine, 2-1 Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan
    Search for more papers by this author
  • Takashi Katura,

    1. Department of Cellular Signaling, Tohoku University Graduate School of Pharmaceutical Sciences, 6-3 Aoba, Aramaki, Aoba-ku, Sendai 980-8578, Japan
    Search for more papers by this author
  • Takahiro Moriya,

    1. Department of Cellular Signaling, Tohoku University Graduate School of Pharmaceutical Sciences, 6-3 Aoba, Aramaki, Aoba-ku, Sendai 980-8578, Japan
    Search for more papers by this author
  • Noriko Osumi

    Corresponding author
    1. Division of Developmental Neuroscience, United Centers for Advanced Research and Translational Medicine (ART), Tohoku University School of Medicine, 2-1 Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan
    Search for more papers by this author

  • Communicated by: Hideyuki Okano

osumi@med.tohoku.ac.jp

Abstract

Arachidonic acid (ARA) and docosahexaenoic acid (DHA), which are the dominant polyunsaturated fatty acids in the brain, have crucial roles in brain development and function. Recent studies have shown that ARA and DHA promote postnatal neurogenesis. However, the direct effects of ARA on neural stem/progenitor cells (NSPCs) and the effects of ARA and DHA on NSPCs at the neurogenic and subsequent gliogenic stages are still unknown. Here, we analyzed the effects of ARA and DHA on neurogenesis, specifically maintenance and differentiation, using neurosphere assays. We confirmed that primary neurospheres are neurogenic NSPCs and that tertiary neurospheres are gliogenic NSPCs. Regarding the effects of ARA and DHA on neurogenic NSPCs, ARA and DHA increased the number of neurospheres, whereas neither ARA nor DHA had a detectable effect on NSPCs in the differentiation condition. In gliogenic NSPCs, DHA increased the number of neurospheres, whereas ARA had no such effect. In contrast, ARA increased the number of astrocytes, whereas DHA increased the number of neurons in the differentiation condition. These results suggest that ARA promotes the maintenance of neurogenic NSPCs and might induce the glial differentiation of gliogenic NSPCs and that DHA promotes the maintenance of both neurogenic and gliogenic NSPCs and might lead to the neuronal differentiation of gliogenic NSPCs.

Introduction

Neural stem/progenitor cells (NSPCs) are defined as immature cells with self-renewing and multipotent abilities. During embryonic development, NSPCs lining the neural tube rapidly divide to produce cells that constitute the entire central nervous system (CNS) in an orderly fashion, with an initial wave of neurogenesis followed by gliogenesis. NSPCs are located not only in the developing brain but also in the adult brain, in which they are especially abundant in the anterior subventricular zone of the lateral ventricle and subgranular zone (SGZ) of the hippocampal dentate gyrus [reviewed by Kriegstein & Alvarez-Buylla (2009)]. Neurogenesis in the adult brain may be required for certain forms of brain function, such as the olfactory function and learning and memory functions related to the hippocampus [reviewed by Zhao et al. (2008)]. The maintenance and differentiation of NSPCs are known to be dynamically regulated by various extrinsic and intrinsic factors under physiological and pathological conditions [reviewed by Ming & Song (2005)]. The identification of signaling molecules that regulate NSPC activity may contribute not only to the understanding of neurogenesis, but also to the comprehension of brain functions under physiological and pathological conditions.

The existence of NSPCs in the mammalian nervous system was clearly shown by Reynolds and Weiss in vitro (Reynolds et al. 1992). These isolated NSPCs were shown to proliferate in the presence of growth factors to form spheres of proliferating, undifferentiated neural cells that were subsequently termed ‘neurospheres’. These neurospheres could be serially passaged to form more neurospheres or differentiated into both neuronal and glial cell types after the removal of growth factors from the medium and plating on an adhesive substrate. Although other methods are available to culture NSPCs (Ray et al. 1993; Conti et al. 2005), the neurosphere culture remains the most frequently used method to enrich, expand and even calculate the frequency of NSPCs. The neurogenic-to-gliogenic fate switch can be observed even within the clones of single NSPC in culture (Qian et al. 2000). Recently, many articles using neurosphere culture have focused on both neurogenic and gliogenic NSPCs (Naka et al. 2008; Okada et al. 2008; Hirabayashi et al. 2009).

Various molecules are specifically expressed, although not always exclusively, in NSPCs in vivo and in vitro, and brain lipid-binding protein (BLBP), or brain-type fatty acid–binding protein (B-Fabp, Fabp7), is one such molecule (Feng et al. 1994). Fabps are a family of 15-kDa cytosolic proteins that have high affinity to long-chain free fatty acids and shuttle them to specific intracellular organelles [reviewed by Furuhashi & Hotamisligil (2008)]. Several members of the Fabp family have been identified in the CNS with distinct spatial and temporal distributions [reviewed by Owada (2008) and references therein]. We have previously shown that Fabp7 is expressed in embryonic and hippocampal NSPCs and is necessary for the maintenance of NSPCs (Arai et al. 2005; Watanabe et al. 2007). Fabp7 is also required for the differentiation of neuronal and glial cells (Feng et al. 1994). In addition, Fabp5, another member of the Fabp family, is expressed in the CNS and promotes neurite extension during the differentiation of PC12 cells (Liu et al. 2008). These results have led us to suggest that long-chain fatty acids, the ligands of Fabps, might regulate the maintenance and/or differentiation of NSPCs.

The adult mammalian brain contains approximately 50–60% of its dry weight as lipids, mostly structural lipids in the form of phospholipids [reviewed by Sastry (1985)]. These include large amounts of long-chain polyunsaturated fatty acids (LCPUFAs), mainly arachidonic acid (ARA, 20:4n-6) and docosahexaenoic acid (DHA, 22:6n-3), and very small amounts of linoleic acid (18:2n-6) and α-linolenic acid (18:3n-3). It is well known that the amount of LCPUFAs, especially ARA and DHA, increases in the brain during development (Green et al. 1999). A dietary or metabolic deficiency in these molecules during brain development has long-lasting effects on later functional adaptation, such as motor and mental functions, suggesting that these molecules are important in brain development [reviewed by Ryan et al. (2010)]. We have previously shown that dietary ARA promotes hippocampal neurogenesis in the rat (Maekawa et al. 2009). Another group has shown that DHA promotes the neuronal differentiation of NSPCs in vitro and adult neurogenesis in the hippocampus in rats fed with a fish-oil-deficient diet over three generations (Kawakita et al. 2006; Katakura et al. 2009). However, the direct effects of ARA on NSPCs and the effects of ARA and DHA on NSPCs at the neurogenic and subsequent gliogenic stages are not well understood.

Based on the aforementioned background, we analyzed the effects of ARA and DHA on NSPCs in proliferation and differentiation conditions using neurosphere assays. Here, we show that ARA promotes the maintenance of neurogenic NSPCs and also increases the number of astrocytes derived from gliogenic NSPCs. In contrast, DHA promotes the maintenance of both neurogenic and gliogenic NSPCs and also increases the number of neurons derived from gliogenic NSPCs.

Results

ARA and DHA promote the maintenance of neurogenic NSPCs

First, we confirmed whether NSPC marker molecules were expressed in culturing neurospheres. Twelve hours after culturing in media hormone mixture (MHM) with epidermal growth factor (EGF) and fibroblast growth factor-basic (bFGF), the plated cells from primary neurospheres were positive for the NSPC markers nestin (98.0 ± 1.0%) and Sox2 (99.0 ± 0.6%) (Fig. S1A, B in Supporting Information). There were only a few cells that were positive for the neuronal marker β-III tubulin (1.1 ± 0.8%) or for the astrocyte marker glial fibrillary acidic protein (GFAP) (0.5 ± 0.3%) in this condition (Fig. S1C, D in Supporting Information), whereas more cells were immunoreactive for β-III tubulin and GFAP in the differentiation condition (see below). The plated cells from tertiary neurospheres were also positive for nestin (96.1 ± 0.3%) and Sox2 (98.6 ± 0.3%), and no cells were positive for β-III tubulin (Fig. S1E–G in Supporting Information). These cells were also positive for GFAP (94.5 ± 0.7%) (Fig. S1H in Supporting Information). GFAP is another marker for NSPCs and is expressed in late neurospheres, whereas early neurospheres do not express GFAP in proliferation condition [reviewed by Pevny & Rao (2003)]. These results show that NSPC markers are actually expressed in neurospheres. Consequently, we considered these cultured cells to be NSPCs.

Neurons and astrocytes in the cortex are derived from common multipotent NSPCs that subsequently pass through phases of proliferation, neurogenesis and gliogenesis. The neurogenic-to-gliogenic fate switch can be observed even within the clones of single NSPC in culture (Qian et al. 2000). In our neurosphere culture system, we also observed that primary neurospheres differentiated mainly into neurons (Fig. 2B, C, E, F) and that gliogenesis was activated in subsequent generations of neurospheres (Fig. 4B, C, E, F). In this study, we regarded primary neurospheres as neurogenic NSPCs and tertiary neurospheres as gliogenic NSPCs and assessed the effects of ARA and DHA on both types of NSPCs using primary and tertiary neurospheres.

Figure 2.

 Effects of arachidonic acid (ARA) and docosahexaenoic acid (DHA) on neurogenic neural stem/progenitor cells in the differentiation condition. Primary neurospheres from an E16.5 rat cortex were dissociated and cultured for 4 days on coated glass slides without growth factors in the presence of various concentrations of ARA or DHA followed by immunostaining. (A–C) ARA does not change the number of β-III tubulin+ cells or that of GFAP+ cells ( 3 for each concentration) compared with vehicle control. (D–F) DHA at 10−9 and 10−7 m does not change the number of β-III tubulin+ cells or GFAP+ cells. DHA at 10−5 m decreases the number of β-III tubulin+ cells but does not change the number of GFAP+ cells ( 3 for each concentration). ***< 0.001. Scale bar: 20 μm.

Figure 4.

 Effects of arachidonic acid (ARA) and docosahexaenoic acid (DHA) on gliogenic neural stem/progenitor cells in the differentiation condition. Tertiary neurospheres from an E16.5 rat cortex were dissociated and cultured for 4 days on coated glass slides without growth factors in the presence of various concentrations of ARA or DHA followed by immunostaining. (A–C) ARA at 10−9 to 10−5 m does not change the number of β-III tubulin+ cells. ARA at 10−9 and 10−7 m does not change the number of GFAP+ cells, but ARA at 10−5 m increases the number ( 6 for each concentration). (D–F) DHA at 10−9 m does not change the number of β-III tubulin+ cells or that of GFAP+ cells. DHA at 10−7 m increases the number of β-III tubulin+ cells but does not change the number of GFAP+ cells. DHA at 10−5 m decreases the number of β-III tubulin+ cells and GFAP+ cells ( 6 for each concentration). *< 0.05, **< 0.01, ***< 0.001. Scale bar: 20 μm.

To examine the effects of ARA and DHA on the maintenance of neurogenic NSPCs, cells from primary neurospheres were cultured with various concentrations of ARA or DHA. ARA increased the number of secondary neurospheres at 10−6 m (< 0.05) but decreased their number at 10−4 m (< 0.001) (Fig. 1A). Similarly, DHA increased the number of secondary neurospheres at a low concentration (10−7 m, < 0.05) but decreased the number of secondary neurospheres at a lower concentration than ARA (10−5 m, < 0.001) (Fig. 1B). These findings show that both ARA and DHA promote the maintenance of neurogenic NSPCs at 10−6 and 10−7 m, respectively, whereas at concentrations higher than 10−4 and 10−5 m, ARA and DHA, respectively, have toxic effects on the survival of neurogenic NSPCs.

Figure 1.

 Effects of arachidonic acid (ARA) and docosahexaenoic acid (DHA) on the maintenance of neurogenic neural stem/progenitor cells. (A, B) Primary neurospheres from an E16.5 rat cortex were dissociated to single cells and cultured for 7 days with growth factors to make secondary neurospheres in the presence of various concentrations of ARA (A) or DHA (B). ARA at 10−6 m and DHA at 10−7 m increase the number of secondary neurospheres, but ARA at 10−4 m and DHA at 10−5 m decrease the number compared with the vehicle control ( 9 for each concentration). (C, D) The size of the secondary neurospheres was quantified. The sizes were separated into three classes: small (50–150 μm), medium (150–250 μm) and large (>250 μm) diameters. ARA at 10−7 to 10−5 m significantly increases the number of neurospheres of a medium diameter ( 9 for each concentration) (C). DHA at 10−7 m significantly increases the number of neurospheres of a small diameter compared with vehicle control ( 9 for each concentration). *< 0.05, **< 0.01, ***< 0.001.

There was no significant difference between the effects of ARA and DHA on the number of neurospheres. To further analyze the effects of ARA and DHA on the maintenance of neurogenic NSPCs, we examined the size of the neurospheres. At concentrations from 10−7 to 10−5 m, ARA increased the number of neurospheres with a diameter of 150–250 μm (< 0.05 at 10−7 m, < 0.01 at 10−6 m, < 0.05 at 10−5 m) (Fig. 1C). In contrast, DHA at 10−7 m increased the number of smaller neurospheres with diameter of 50–150 μm (Fig. 1D). These results suggest that the effects of ARA and DHA on the maintenance of neurogenic NSPCs are different from each other.

ARA and DHA have no detectable effects on neurogenic NSPCs in the differentiation condition

Next, we examined the effects of ARA and DHA on neurogenic NSPCs in the differentiation condition using primary neurospheres with various concentrations of ARA or DHA without growth factors. ARA at 10−9 to 10−5 m did not increase or decrease the number of β-III tubulin-positive (+) cells (Fig. 2A, B) or GFAP+ cells (Fig. 2A, C). However, ARA at 10−5 m increased the proportion of active caspase 3+ cells (< 0.01) and decreased the plating efficiency (< 0.01) (Fig. S2A–C in Supporting Information). Similar to ARA, DHA at 10−9 and 10−7 m did not increase or decrease the number of β-III tubulin+ cells (Fig. 2D, E) or GFAP+ cells (Fig. 2D, F). However, DHA at 10−5 m decreased the number of β-III tubulin+ cells (< 0.001) (Fig. 2D, E), but this reduction was not observed for GFAP+ cells (Fig. 2D, F). This may be because DHA at 10−5 m increased the proportion of active caspase 3+ cells (< 0.001) and decreased the plating efficiency (< 0.001) (Fig. S2D–F in Supporting Information). Therefore, neither ARA nor DHA has any detectable effect on neurogenic NSPCs in the differentiation condition. In addition, high concentrations of both ARA and DHA had toxic effects in neurogenic NSPCs and in cells differentiated from them.

ARA and DHA have no detectable effects on the neurogenic-to-gliogenic transition of neurogenic NSPCs

We showed that ARA at 10−6 m and DHA at 10−7 m increased the number of secondary neurospheres (Fig. 1). To further examine whether ARA and DHA at these concentrations keep secondary neurospheres in the neurogenic state or whether they accelerate the neurogenic-to-gliogenic transition, secondary neurospheres grown with ARA at 10−6 m or DHA at 10−7 m were dissociated and replated in the differentiation condition. The numbers of β-III tubulin+ cells and GFAP+ cells obtained from neurospheres grown with ARA at 10−6 m or DHA at 10−7 m were no different from those obtained from neurospheres grown with the vehicle control (Fig. S3 in Supporting Information). These results showed that neither ARA nor DHA have a detectable effect on the neurogenic-to-gliogenic transition of neurogenic NSPCs, although both fatty acids promote the maintenance of neurogenic NSPCs.

DHA promotes the maintenance of gliogenic NSPCs

We further analyzed the effects of ARA and DHA on the maintenance of gliogenic NSPCs. ARA at concentrations higher than 10−5 m decreased the number of quaternary neurospheres (< 0.001) (Fig. 3A). In contrast to ARA, DHA at 10−10 and 10−8 m increased the number of quaternary neurospheres (< 0.05), and there was a near significant increase at 10−9 m (= 0.113). A toxic effect of DHA on the maintenance of NSPCs was also observed at 10−5 m (< 0.001) (Fig. 3B). Thus, at low concentrations (10−10 and 10−8 m), DHA, but not ARA, promotes the maintenance of gliogenic NSPCs, whereas both ARA and DHA have toxic effects on the maintenance of gliogenic NSPCs at concentrations higher than 10−5 m.

Figure 3.

 Effects of arachidonic acid (ARA) and docosahexaenoic acid (DHA) on the maintenance of gliogenic neural stem/progenitor cells. (A, B) Tertiary neurospheres from an E16.5 rat cortex were dissociated into single cells and cultured for 7 days with growth factors to make quaternary neurospheres in the presence of various concentrations of ARA (A) or DHA (B). DHA at 10−10 and 10−8 m increases the number of quaternary neurospheres but ARA at 10−5 and 10−4 m and DHA at 10−5 m decrease the number as compared to the vehicle control ( 19 for each concentration). (C, D) The size of the quaternary neurospheres was also quantified. The sizes were separated into three classes: small (50–150 μm), medium (150–250 μm) and large (>250 μm) diameters. ARA at 10−10 to 10−6 m does not change the number of neurospheres with each diameter ( 19 for each concentration) (C). DHA at 10−10 and 10−8 m significantly increases the number of smaller neurospheres (= 15 for each concentration) (D). *< 0.05, **< 0.01, ***< 0.001.

Regarding the effects of ARA and DHA on the size of the neurospheres, ARA at concentrations <10−6 m did not increase the number of spheres of every size, and ARA at 10−5 m or higher decreased the number of them (< 0.01) (Fig. 3C). However, at lower concentrations (10−10 and 10−8 m), DHA increased the number of spheres with a diameter of 50–150 μm (< 0.05) (Fig. 3D). These results support that DHA, but not ARA, promotes the maintenance of gliogenic NSPCs.

ARA and DHA have distinct effects on gliogenic NSPCs in the differentiation condition

Finally, we analyzed the effects of ARA and DHA on gliogenic NSPCs in the differentiation condition. ARA did not increase the number of β-III tubulin+ cells at any of the concentrations examined but did increase the number of GFAP+ cells at 10−5 m (< 0.01) (Fig. 4A–C). Similar to the results observed in neurogenic NSPCs, ARA at 10−5 m increased the proportion of active caspase 3+ cells (< 0.05) and decreased the plating efficiency (< 0.001) (Fig. S4A–C in Supporting Information). Thus, ARA increases the number of astrocytes derived from gliogenic NSPCs but at higher concentrations has toxic effects on the survival of gliogenic NSPCs or of cells differentiated from them.

A distinct effect was observed for DHA. DHA did not increase the number of GFAP+ cells at any of the concentrations examined but did increase the number of β-III tubulin+ cells at 10−7 m (< 0.05) (Fig. 4D–F). Similar to the results obtained in neurogenic NSPCs, DHA at 10−5 m increased the proportion of active caspase 3+ cells (< 0.05) and decreased the plating efficiency (< 0.001) (Fig. S4D–F in Supporting Information). Combined with the aforementioned ARA results, it is clear that ARA and DHA have differential effects on gliogenic NSPCs in the differentiation condition; ARA increases the number of astrocytes, and DHA increases the number of neurons.

Discussion

In the present study, we examined the effects of ARA and DHA, the main components of the brain LCPUFA profile, on NSPCs in proliferation and differentiation conditions in vitro. Using a neurosphere assay, we showed that ARA and DHA had multiple roles in NSPCs. ARA promoted the maintenance of neurogenic NSPCs and increased the number of astrocytes derived from gliogenic NSPCs. We also found that DHA promoted the maintenance of both neurogenic and gliogenic NSPCs and increased the number of neurons derived from gliogenic NSPCs.

ARA and DHA at various concentrations regulate the maintenance of NSPCs

Arachidonic acid and DHA are the dominant LCPUFAs in the brain and have crucial roles in brain development and function. Although we showed that both ARA and DHA regulated the maintenance of NSPCs, their effects and effective concentrations were different from each other. Regarding the effects on neurogenic NSPCs, the optimal concentrations for increasing the number of neurospheres were very close, 10−6 and 10−7 m for ARA and DHA, respectively (Fig. 1A, B). However, ARA increased the number of neurospheres with a diameter of 150–250 μm, whereas DHA increased the number of neurospheres with a diameter of 50–150 μm (Fig. 1C, D). The differential effects of ARA and DHA on the maintenance of NSPCs were more apparent in gliogenic NSPCs. ARA did not increase the number of neurospheres from gliogenic NSPCs (Fig. 3A), whereas DHA at low concentrations increased the number of gliogenic spheres (Fig. 3B). These results may imply a distinctive mode of action of ARA and DHA in the maintenance of NSPCs.

The effective concentrations of ARA and DHA were different between neurogenic and gliogenic NSPCs. ARA decreased the number of neurospheres in neurogenic NSPC cultures at 10−4 m, and it decreased the number of neurospheres in gliogenic NSPC cultures at a lower concentration (10−5 m) (Figs 1A, 3A). Thus, gliogenic NSPCs are more sensitive to the toxic effects of ARA compared with neurogenic NSPCs. In addition, ARA increased the number of neurogenic spheres, but not that of gliogenic spheres (Figs 1A, 3A). It is still unclear how the mechanism by which ARA promotes the maintenance of NSPCs is lost in gliogenic NSPCs. Similar to ARA, DHA increased the number of neurogenic spheres at 10−7 m and that of gliogenic spheres at approximately 10−9 m (Figs 1B, 3B). We suspect that differences in gene expression between neurogenic and gliogenic NSPCs are the reason for these phenomena.

The present in vitro results on the effects of ARA are partially consistent with our previous report (Maekawa et al. 2009), which showed that in the rat, ARA promotes the maintenance of hippocampal NSPCs that show a neurogenic phenotype (Seaberg & van der Kooy 2002). In contrast, our results on the effects of DHA seem to be inconsistent with a previous report that showed that DHA inhibits the proliferation of NSPCs (Katakura et al. 2009). However, these authors examined the effects of DHA at concentrations >10−7 m on the proliferation of NSPCs in which gliogenesis had been activated, i.e., the proportion of GFAP+ cells was approximately 8% in the 4-day in vitro (DIV) differentiation culture (Fig. 1 in Katakura et al. (2009)). Therefore, the discrepancy between their results and ours could be explained by the interpretation that only the lower concentration of DHA can promote maintenance only of gliogenic NSPCs.

The clonality of the neurosphere culture was previously confirmed by the observation that cells separately prepared from neurospheres constitutively expressing enhanced green fluorescence protein (EGFP) and β-galactosidase generated a considerable population of chimeric neurospheres even when cultured at low cell densities (0.5 cells/μL) (Singec et al. 2006). In the present study, we cannot exclude the possibility that the fusion of neurospheres occurs. However, the fusion of neurospheres would decrease the number of smaller neurospheres and increase the number of larger neurospheres. Instead, we observed no significant decrease in the number of neurospheres with a diameter of 50–150 μm or an increase in the number of neurospheres with a diameter of >250 μm upon treatment with ARA at 10−6 m. Similarly, we observed no significant decrease in the number of neurospheres with diameters of 150–250 μm or with diameters of >250 μm upon treatment with DHA at 10−10, 10−8 and 10−7 m. Thus, we concluded that we can evaluate the effects of ARA and DHA on the maintenance of NSPCs under our conditions.

ARA and DHA at various concentrations regulate NSPCs in the differentiation condition

Similar to the effects on the maintenance of NSPCs, the effects of ARA and DHA on NSPCs in the differentiation condition are also different in neurogenic and gliogenic neurospheres. Neither ARA nor DHA had any detectable effects on the numbers of neurons and astrocytes derived from neurogenic NSPCs (Fig. 2), whereas with gliogenic NSPCs, ARA and DHA increased the number of astrocytes and neurons, respectively (Fig. 4). Although it is known that DHA promotes neuronal differentiation (Kawakita et al. 2006; Katakura et al. 2009), our results can be interpreted in different ways: (1) ARA and DHA promote astrocytic and neuronal differentiation; (2) ARA and DHA promote the proliferation of astrocytic and neuronal progenitor cells; (3) ARA and DHA promote the survival of astrocytes and neurons; and (4) ARA and DHA enhance the adhesion of astrocytes and neurons to the culture slides used in the present study. Regarding (3), we examined the proportion of active caspase 3+ cells of total cells (Figs S2, S4 in Supporting Information), but we admit that it is desirable to separately analyze the susceptibility of neuronal or astrocytic progenitor cells to ARA and DHA. Further analyses are needed to discriminate the effects of ARA and DHA.

Interestingly, the astrocytic differentiation of NSPCs in the central canal of the spinal cord is activated following spinal cord injury (Barnabe-Heider et al. 2010). In addition, the level of ARA in the spinal cord tissue is increased after injury (Pantovic et al. 2005). It is thus possible that the formation of glial scars is attributable to ARA produced upon inflammation caused by injury. Understanding the response of NSPCs to injury could allow the development of pharmacological strategies to modulate NSPCs in situ to produce the desired progeny after injury.

As mentioned earlier, here and in previous in vivo and in vitro studies, we observed a promoting effect of DHA on the number of neurons derived from gliogenic NSPCs (Kawakita et al. 2006; Katakura et al. 2009), but this effect was not observed with neurogenic NSPCs. This suggests that the function of DHA in neuronal differentiation may be to elicit neurogenic competence from non-neurogenic cells. Supporting evidence for this is that DHA can induce the neuronal differentiation of mesenchymal stem cells (Kan et al. 2007). All these results may be explained by the fact that neurons have unique fatty acid composition in their membranes, namely a high proportion of DHA (Innis 1991).

Another interesting difference between ARA and DHA is their effects on cell death. Although cell death was induced with ARA at the high concentration in NSPCs (Figs 1A, 3A), but not in neurons nor in astrocytes (Figs 2B, C, 4B, C), cell death was induced in NSPCs (Figs 1B, 3B), neurons and astrocytes (Figs 2E, F, 4E, F) with the same concentration of DHA. This means that DHA has more toxic effects at higher concentrations compared with ARA. The toxicity of DHA could be attributed to the production of an oxidative metabolite of DHA because DHA has more double bonds in its structure than ARA. Because we did not add an antioxidant, DHA at higher concentrations may easily become oxidized under the culture conditions used in these studies. The detrimental, toxic effects of oxidative metabolites of LCPUFAs on brain development and neurogenesis have actually been reported in vivo (Tozuka et al. 2009).

Mode of action of ARA and DHA in the maintenance and differentiation of NSPCs

It is known that prostaglandin E2 (PGE2) has an important role in neurogenesis (Uchida et al. 2002). 15-Deoxy-Δ12, 14-prostaglandin J2, an endogenous metabolite of prostaglandin D2 (PGD2), also regulates the maintenance of NSPCs via its electrophilic nature, which enables covalent binding to other molecules (Katura et al. 2010). Similar to the prostaglandins, stimulation of CB2, a receptor for endocannabinoids, promotes the maintenance of NSPCs (Molina-Holgado et al. 2007). PGE2, PGD2 and endocannabinoids are metabolites of ARA. Thus, it is possible that ARA could promote the maintenance of NSPCs via prostaglandins. Along this line of reasoning, several metabolites of DHA, including D-series resolvins and neuroprotectin [reviewed by Calder (2009)], might regulate the maintenance of NSPCs.

A remaining question is how ARA and/or DHA regulates the maintenance and differentiation of NSPCs. Regarding the effect of DHA on neuronal differentiation, a possible mechanism is that DHA promotes cell cycle exit and decreases the expression of nestin as has been shown in retinal neural progenitor cells (Insua et al. 2003). In addition, DHA has been shown to promote neuronal differentiation from NSPCs by suppressing the Hes1 repressor, which in turn activates p27kip1 to arrest cell cycle (Katakura et al. 2009).

The positive effect of ARA on the astrocytic differentiation of gliogenic NSPCs could be carried out through the activation of retinoid X receptor (RXR) because ARA also binds to and activates RXR (Lengqvist et al. 2004). Moreover, it was recently reported that retinoic acid (RA) facilitates the astrocytic differentiation of NSPCs through the epigenetic modification of the gfap promoter and that a putative RA response element in the gfap promoter was essential for the activation by RA (Asano et al. 2009).

In addition to the astrocytic differentiation of NSPCs by ARA, it is also possible that ARA and DHA augment the maintenance and neuronal differentiation of NSPCs by changing gene expression through nuclear receptors. ARA and DHA can be transferred from Fabps to nuclear receptor proteins [reviewed by Veerkamp & Zimmerman (2001)], thereby indirectly controlling the transcription of genes related to the maintenance and differentiation of NSPCs. This scenario is analogous to that in which RA is transferred from a cellular RA-binding protein (structurally similar to the Fabps) and Fabp5 to the RA receptor and peroxisome proliferator–activated receptor (PPAR)β/δ, respectively (Schug et al. 2007). ARA and DHA actually bind to and activate RXRα, PPARα and PPARγ (Wolfrum et al. 2001; Lengqvist et al. 2004). Fabps also enhance the transactivation of PPARs (Tan et al. 2002). Moreover, activation of RXR by DHA enhances the expression of Fabp5 (Volakakis et al. 2009), which induces neuronal differentiation (Liu et al. 2008). All these genes are expressed in NSPCs (Fig. S5 in Supporting Information). Identifying the precise mechanisms for the role of ARA and DHA in promoting the maintenance and differentiation of NSPCs warrants further studies.

Differential effects of ARA and DHA on neurogenic and gliogenic NSPCs

Multipotent NSPCs isolated from the early mouse embryonic cortex can sequentially generate specific types of neurons and glial cells in vitro in the proper in vivo order (Qian et al. 2000). Here, we observed differential effects of ARA and DHA on neurogenic and gliogenic NSPCs. One of the possible explanations for this phenomenon is the difference in gene expression between neurogenic and gliogenic NSPCs. The expression levels of various genes are changed during the neurogenic-to-gliogenic transition of NSPCs, and the chicken ovalbumin upstream promoter transcription factor (COUP-TF) I/II has an important role in achieving gliogenic competence in NSPCs (Naka et al. 2008). Although COUP-TFs are orphan nuclear receptors, it is known that they form heterodimers with RXR (Kliewer et al. 1992). As mentioned earlier, ARA and DHA bind to and activate RXR (Lengqvist et al. 2004). Thus, it would be interesting to know whether the functions of COUP-TFs are directly or indirectly regulated by ARA and/or DHA.

Experimental procedures

Animals

A large colony of wild-type Sprague–Dawley rats (Charles River Japan, Japan) was maintained at the Tohoku University School of Medicine. All animal experiments were carried out in accordance with the National Institutes of Health guidelines for the care and use of laboratory animals and were approved by the university’s Committee for Animal Experiments (animal experiment 22–306).

Neural stem/progenitor cell culture

The generation of spheres from the embryonic forebrain was performed as previously described (Reynolds et al. 1992) with minor modifications. Briefly, cortices were dissected from rat embryos at embryonic day (E) 16.5 and were collected into Tyrode’s solution and then transferred into the MHM composed of DMEM-F12 (1 : 1) (Invitrogen, Carlsbad, CA, USA), 0.6% glucose, HEPES buffer (5 mm), insulin (25 μg/mL) (Wako Chemicals, Osaka, Japan), transferrin (100 μg/mL) (Sigma-Aldrich, St. Louis, MO, USA), progesterone (20 nm) (Sigma-Aldrich), putrescine (60 μm) (Sigma-Aldrich) and selenite (30 nm) (Sigma-Aldrich). Tissue pieces were mechanically dissociated with a micropipette and then passed through a 40-μm nylon mesh (BD Biosciences, San Jose, CA, USA). Cells were seeded at 2.0 × 105 cells/mL (total of 12 mL) into the MHM, which also contained 20 ng/mL EGF (PeproTech, Offenbach, Germany), 10 ng/mL bFGF (PeproTech) and 2 μg/mL heparin (Sigma-Aldrich), in an uncoated T75 culture flask (culture area is 75 cm2) (BD Biosciences), and were maintained in a humidified incubator at 37 °C with 95% atmospheric air and 5% CO2. Fresh media containing 20 ng/mL EGF and 10 ng/mL bFGF were added every other day. Cells were cultured for 5 DIV and formed floating cell clusters, called neurospheres, of which the diameter was within 50–150 μm. Cells were subsequently collected by centrifugation and passaged after mechanical dissociation by pipetting followed by passing through a nylon mesh as described elsewhere. Cells were cultured for 3 DIV to form secondary neurospheres, of which the diameter was within 20–50 μm. Tertiary neurospheres were produced similarly.

Maintenance assay

Mechanically dissociated neurospheres were suspended at a density of 1.0 × 104 cells/mL in MHM containing EGF and bFGF, and 200 μL of the cell suspension was transferred to each well (0.32 cm2) of a 96-well plate. After 12-h culture, the cultures were treated with ARA or DHA (Cayman Chemical, Ann Arbor, MI, USA), which was dissolved in ethanol and diluted 10 000-fold into MHM. MHM containing 0.01% ethanol was used as the vehicle control, and the culture medium was changed at 4 DIV. The number of neurospheres was counted after culture for 7 DIV (Fig. S6A in Supporting Information).

Differentiation condition assay

Neurospheres were mechanically dissociated and plated onto poly-l-ornithine- and laminin-coated Lab-Tek Chamber Slides™ (Thermo Fisher Scientific, Waltham, MA, USA) at a density of 5.3 × 104 cells/cm2 with neither EGF nor bFGF. After 12-h culture, the cultures were treated with ARA or DHA as described earlier. After culture for 4 DIV, the number of β-III tubulin+ cells and GFAP+ cells was counted following immunocytochemistry (Fig. S6B, Table S1 in Supporting Information).

Immunocytochemistry

Cells were fixed in 4% paraformaldehyde for 30 min at room temperature. After rinsing with PBS, cells were incubated in 3% bovine serum albumin and 0.3% Triton X-100 for 1 h at room temperature. Cells were then incubated with primary antibodies overnight at 4 °C. After rinsing with PBS, cells were incubated with the appropriate secondary antibodies for 1 h at room temperature. Cell nuclei were counterstained by DAPI staining. After rinsing with PBS, cells were observed by fluorescence microscopy (IX71; Olympus, Tokyo, Japan).

The primary antibodies used in this study were as follows: mouse monoclonal anti-nestin IgG (1 : 20; Millipore Corporate Headquarters, Billerica, MA, USA), mouse monoclonal anti-Sox2 IgG (1 : 50; R & D Systems, Minneapolis, MN, USA), mouse monoclonal anti-β-III tubulin IgG (1 : 2000; Covance, Berkeley, CA, USA), rabbit polyclonal anti-GFAP antibody (1 : 800; Dako, Glostrup, Denmark) and rabbit monoclonal anti-active caspase 3 IgG (1 : 800; BD Biosciences). The secondary antibodies used in this study were as follows: Alexa 488-conjugated goat anti-mouse IgG (1 : 400; Invitrogen) and Cy3-conjugated goat anti-rabbit IgG (1 : 400; Jackson ImmunoResearch, West Grove, PA, USA).

For the quantification of the double-labeled cells in vitro, the pictures of seven visual fields for each condition were taken at random. The numbers of the labeled cells and the total number of cell nuclei stained with DAPI were counted.

RT-PCR

Total RNA was obtained from primary and tertiary neurospheres using TRIzol (Invitrogen) following the manufacturer’s instructions. Reverse transcription (RT) was performed with 2 μg of total RNA, oligo d(T)12–18 primers (Invitrogen) and SuperScript™ III Reverse Transcriptase (Invitrogen). cDNA was amplified by 27 (gapdh), 30 (others) or 33 (pparγ) cycles of PCR using Ex Taq (Takara Bio, Ohtsu, Japan) in a thermal cycler. The cycles were as follows: denaturation at 94 °C for 15 s, annealing at 55 °C for 30 s and extension at 72 °C for 60 s. The sense and antisense primers used were as follows: fabp7, sense 5′-aaggatggcaaaatggttg-3′ and antisense 5′-taattttctgcctccacacc-3′; fabp5, sense 5′-agtgggaagggaaagaaagc-3′ and antisense 5′-tccaggatgacgagttagcc-3′; pparα, sense 5′-gacaaggcctcaggatacca-3′ and antisense 5′-gtcttctcagccatgcacaa-3′; pparβ/δ, sense 5′-tcaacaaagacggactgctg-3′ and antisense 5′-tcttcagccactgcatcatc-3′; pparγ, sense 5′-ccctggcaaagcatttgtat-3′ and antisense 5′-actggcacccttgaaaaatg-3′; rxrα, sense 5′-gctcaccaaatgaccctgtt-3′ and antisense 5′-gtgagcgctgttcctgtgta-3′; and gapdh, sense 5′-tgaacgggaagctcactgg-3′ and antisense 5′-tccaccaccctgttgctgta-3′. All of these pairs of primers were designed to encompass at least one intron to avoid false-positive amplification from contaminating genomic DNA.

Statistical analysis

Statistical analysis was carried out using one-way analysis of variance (anova) or using an unpaired two-tailed Student’s t-test if applicable. The results are expressed as the mean ± standard error of the mean (SEM). A one-way anova followed by Dunnett’s test was used to compare with the control group. < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS software (IBM, Armonk, NY, USA).

Acknowledgements

We thank Dr Michio Hashimoto for critically reading the manuscript and Ms Sayaka Makino for animal care. We are also grateful to all of the members of our laboratory for fruitful discussions. This work was supported by the Core Research for Evolutional Science and Technology from the Japan Science and Technology Agency, the Tohoku Neuroscience Global COE program ‘Basic & Translational Research Center for Global Brain Science’ from MEXT of Japan and KAKENHI (21300115) from MEXT of Japan.

Ancillary