Macrophage migration inhibitory factor (MIF) promotes cell survival and proliferation of neural stem/progenitor cells

To date, one promising approach for the treatment of neurodegenerative diseases such as Parkinson's disease, e.g. amyotrophic lateral sclerosis, stroke, and spinal cord injury (SCI), is neural stem/progenitor cell (NSPC)-based therapy, which involves neural regeneration through NSPC transplantation and strategies for the activation of endogenous NSPCs. It is not yet clear which of these therapeutic approaches is the most effective (Okano, 2002; Okano et al., 2007; Okano and Sawamoto, 2008; Aboody et al., 2011). NSPCs are multipotent precursors present in both the embryonic and adult brains, which are capable of undergoing self renewal as well as of generating neurons, astrocytes, and oligodendrocytes (Reynolds and Weiss, 1992; Gage, 2000), thereby contributing to neurogenesis (van der Kooy and Weiss, 2000; Temple, 2001). A number of factors have been shown to participate in the regulation of embryonic and adult NSPCs (Shimazaki, 2003; Zhao et al., 2008), yet the identification of molecules capable of regulating NSPCs is required to better understand neural stem cell biology and its therapeutic potential.

In a previous study, we showed that the implantation of mouse splenic conventional dendritic cells (cDCs) into an injured site leads to functional recovery in a mouse SCI model (Mikami et al., 2004). To examine the effects of DCs on SCI, we characterized DCs derived from bone marrows of common marmosets (Ohta et al., 2008), and observed functional recovery from SCI by DC transplantation (Yaguchi et al., 2009) in our established SCI model (Iwanami et al., 2005), suggesting that DCs may be used for SCI therapy. In the mouse SCI model, cDCs have been shown to induce de novo neurogenesis in vivo and to increase NSPC proliferation in vitro and in vivo, although details of the mechanism underlying this induction remain unknown. Thus, the molecular mechanism by which cDCs are capable of inducing the proliferation and/or survival of NSPCs in the SCI model remained to be elucidated.

Based on these observations, we sought to identify molecules produced by cDCs that can increase the proliferation and/or survival of NSPCs in vitro. Using a lentivirus-mediated functional cloning system, we identified a candidate gene, CD74, which is a non-polymorphic type II integral membrane protein that is expressed on antigen presenting cells including DCs, B cells and macrophages. CD74 was initially considered a major histocompatability complex class II chaperone (Stumptner-Cuvelette and Benaroch, 2002). Nonetheless, macrophage migration inhibitory factor (MIF) was later demonstrated to be a ligand of CD74 (Leng et al., 2003), which can generate a complex with CD44 and act as a receptor for MIF (Shi et al., 2006). To date, MIF has also been identified as a non-cognate ligand for CXCR2 and CXCR4 (Schwartz et al., 2009). MIF is known to be a pro-inflammatory factor in many diseases including atherosclerosis and rheumatoid arthritis (Morand et al., 2006). Additionally, MIF has been shown to induce cell proliferation in macrophages (Shi et al., 2006), B lymphocytes, and prostate cancer cells (Meyer-Siegler et al., 2006; Starlets et al., 2006). However, the beneficial role of MIF through AMP-activated protein kinase (AMPK) activation in an ischemia model was recently reported (Miller et al., 2008), indicating that MIF may have multiple unknown functions in various diseases. In the central nervous system (CNS), MIF expression has been reported in the rat forebrain ventricular zone (Suzuki et al., 1999), yet the function of MIF in the CNS and in NSPSCs has yet to be elucidated. We therefore decided to investigate the roles of MIF in NSPCs. In the present study, we observed MIF secretion by NSPCs and cDCs. MIF could promote proliferation and/or survival of NSPCs. In addition, MIF secreted by cDCs partially supports the stimulation of NSPC proliferation and/or survival. This is the first report that shows the functions of MIF–CD74 signaling in NSPCs.

To identify mouse cDC-derived molecules that can induce NSPC proliferation, we first constructed a novel lentiviral expression vector, EFII-BSTXI-SV40-hrGFP, which contains a BSTXI cloning site and encodes GFP driven by a SV40 promoter (supplementary material Fig. S1A). A full-length cDNA library was generated from mouse splenic cDCs using an oligo-capping cDNA synthesis method (Maruyama and Sugano, 1994). The full-length cDNAs were cloned into the BSTXI site of EFII-BSTXI-SV40-hrGFP and the plasmids were amplified once in bacterial cells. Ninety-six randomly selected clones generated from the cDNA libraries were subjected to sequence analysis, and 92.4% of the clones containing the full-length cDNA were identified. The lentiviral cDNA expression plasmid library vectors were subjected to lentivirus production. Next, NSPCs cultured from mouse E14.5 GE were infected with the lentivirus-expressing cDC-derived cDNA. GFP-positive single cells dissociated from neurospheres were seeded onto 96-well plates at a low density (1 cell/µl). After 10 days of culture, some highly proliferative neurospheres were observed. Those single spheres were picked up, expanded, and subjected to a second round of screening based on their ability to form neurospheres compared to NSPCs infected with empty control vectors. Finally, genomic DNA PCR was performed to verify the cDNA inserted into the neurospheres (supplementary material Fig. S1B). DNA sequencing analysis revealed that two of the four neurosphere clones that were selected in the secondary screening encoded the same CD74 cDNA (NCBI accession number, AK154096.1), and the other two clones did not contain the full-length cDNA. Thus, we did not perform further analyses on the latter two clones.

To examine whether CD74 can induce the proliferation and/or promote the survival of NSPCs, CD74 was retrovirally expressed in NSPCs. Cell proliferation assay were performed on these cells using a Cell Titer-Glo Luminescent Cell Viability Assay kit. CD74-overexpressing NSPCs showed a significant increase in cell proliferation compared to NSPCs expressing the control virus at 4 days post-infection (1.3±0.02 fold of control, P<10⁻⁵; n = 3; Fig. 1A). This result was confirmed by cell number quantification using Trypan Blue staining at 5 days post-infection (1.47±0.26 fold of control, P<0.05; n = 3). Next, we performed a secondary neurosphere assay using the same CD74-expressing retrovirus and compared the results to cells infected with the control virus (pMX-Ig). At a low seeding density condition (1 cell/µl), a greater number of secondary neurospheres was generated from CD74-overexpressing primary neurospheres, indicating the accuracy of the expression cloning strategy (2.8±0.75 fold of control, P = 0.001; n = 8; Fig. 1B). These results show that activation of CD74 signaling in neurospheres can support the proliferation and/or survival of NSPCs without other stimulation in vitro.

Although CD74 overexpression in NSPCs supported cell proliferation and/or survival, the detailed mechanisms underlying these effects were unclear. CD74 was originally identified as a MHC Class II-associated molecule in antigen-presenting cells (APCs) (Stumptner-Cuvelette and Benaroch, 2002). To date, many studies have reported that MIF is a CD74 ligand that can activate various signaling pathways in some cells including tumor cells (Bucala and Donnelly, 2007). Therefore, we considered the possibility that MIF expressed endogenously in NSPCs can support the proliferation of NSPCs through CD74, which is expressed in NSPCs. If this were the case, CD74 overexpression in NSPCs could induce the proliferation and/or survival of NSPCs without the addition of exogenous MIF, consistent with the results above. In addition, the examination of soluble factors that can support NSPC proliferation and/or survival is desirable, especially when considering the therapeutic potential of MIF. We originally sought to find unknown factors secreted by cDCs that can support the proliferation and/or survival of NSPCs. Thus, we focused on studying MIF in NSPCs and first examined the amount of MIF secreted in the supernatants of splenic cDCs and of mouse GE- and spinal cord-derived neurospheres, which are cultured overnight, using ELISA. We found that both neurospheres derived from different embryonic tissues and cDCs secreted MIF, although cDCs secreted a higher amount of MIF compared to the neurospheres (Fig. 2A). We also observed MIF secretion in human NSPCs (1.87±0.1 ng/ml for NSPCs cultured overnight). In addition, we performed histological analyses. E14.5 GEs were observed using immunohistochemical techniques, and some cells expressed Nestin (a marker of NSPCs and hair-follicle bulge stem cells) (Amoh et al., 2005) and CD74 in the ventricular lining (Fig. 2B). We also observed MIF expression in the GE using immunohistochemistry and RT-PCR (Fig. 2C; supplementary material Fig. S2), which is consistent with a previous report that MIF is expressed in the ventricular zone of the fetal rat brain (Suzuki et al., 1999). In addition, CD74 expression in cultured neurospheres generated from E14.5 GE was also shown by immunocytochemical staining of the NSPC marker, Nestin (Fig. 2D). The expression of other MIF receptors (CD44, CXCR2, CXCR4 in addition to CD74) in neurospheres generated from E14.5 GE was examined using flow cytometry (Fig. 2E). The gene expression of MIF and its receptors in the GE, GE-derived neurospheres and cDCs was confirmed by RT-PCR analysis (supplementary material Fig. 2B).

To characterize the effects of MIF on NSPCs, we examined changes in cell NSPC proliferation with MIF treatment. The addition of MIF to single cells dissociated from neurospheres slightly increased cell viability at day 4. This increase was small but statistically significant (100 ng/ml, 1.1±0.05 fold of control, P = 0.001, n = 4; 400 ng/ml, 1.1±0.06 fold of control, P = 0.001, n = 4; Fig. 3A). We further examined the effects of MIF on NSPC proliferation by performing BrdU chase experiments in NSPCs with MIF treatment. Immunocytochemical analysis showed an increase in BrdU incorporation (1.24±0.12 fold of control, P = 0.034; n = 3; Fig. 3B), supporting the proliferation assay results. Next, we examined the effects of a MIF-specific inhibitor, ISO-1, on NSPCs and observed that ISO-1 treatment (100 µM) decreased cell viability (0.54±0.05 fold of control, P = 0.05; n = 3; Fig. 3C). In addition, the gene silencing effect of MIF on NSPC proliferation was tested using siRNA. First, we used two siRNAs to knock down MIF gene expression in NSPCs. A decrease in gene expression was observed in the transient transfection system, accompanied by a decrease in cell viability following treatment with both siRNAs (supplementary material Fig. S3). Since efficient gene silencing of MIF in NSPCs was expected, a retrovirally expressed short hairpin RNA (shRNA)-MIF vector was constructed and NSPCs were infected with the virus. Six days after infection with an shRNA-MIF-expressing virus, cell viability decreased (0.68±0.06 fold of control, P = 0.015; n = 3; Fig. 3C), supporting the results of the transient gene knockdown experiment. Cell cycle analysis also showed that gene silencing of MIF in NSPCs decreased the S-phase population, which was in accordance with the increase in BrdU incorporation observed with MIF treatment (Fig. 3D). Taken together, MIF contributes to NSPC proliferation, although the effect of MIF knockdown was greater than that of exogenous MIF addition. Finally, the effects of MIF treatment on NSPC apoptosis were examined. Gene silencing of MIF led to an increase in caspase 3/7 activity in shRNA-MIF-infected NSPCs compared to controls 5 days after infection (1.4±0.07 fold of control, P = 0.004; n = 4; Fig. 3E). Thus, MIF plays an important role in NSPC proliferation and survival.

To identify changes in NSPC's self-renewal capacity following MIF treatment, we performed a neurosphere-formation assay, the most commonly used method to measure the capacity of NSPC self-renewal. We first added MIF exogenously to single NSPCs plated at a low cell density (1 cell/µl) on 96-well plates and then evaluated the number of primary neurospheres that were generated. MIF increased the number of primary neurospheres in a dose-dependent manner (200 ng/ml, 2.2±0.87 fold of control, P = 0.02, n = 8; 400 ng/ml, 3.4±1.6 fold of control, P = 0.006, n = 8; Fig. 4A). This increase in primary neurosphere formation following exogenous MIF addition was attenuated by both anti-CD74 neutralizing antibody and ISO-1 treatment (supplementary material Fig. S4). Next, we performed primary sphere formation assays at a cell density of 4 cells/µl using three different conditions. First, we applied anti-CD74 neutralizing antibody in conditions conducive to primary sphere formation to block the interaction between endogenously expressed MIF and CD74. A decrease in the number of primary spheres was observed (0.38±0.09 fold of control, P = 0.006; n = 5; Fig. 4B). In addition to CD74, we also blocked CXCR4 and CXCR2 using neutralizing antibodies, and observed that the blockade of CD74 alone showed a significant decrease in the number of primary neurospheres (supplementary material Fig. S5). In the second condition, ISO-1 was added to the same culture condition, and this treatment also resulted in a decrease in the number of primary neurospheres (0.2±0.15 fold of control, P<10⁻⁵; n = 8; Fig. 4C). Finally, we examined the effects of MIF gene silencing using shRNA-MIF and observed a decrease in primary neurosphere formation (0.05±0.01 fold of control, P = 0.01; n = 8; Fig. 4D). Taken together, these data demonstrated that the MIF-CD74 ligand–receptor system can contribute to primary neurosphere formation, suggesting that MIF may support the proliferation and/or survival of NSPCs in vitro. To study the effects of MIF treatment on the self-renewal ability of NSPCs, a secondary neurosphere assay was developed. Exogenous addition of MIF increased the number of secondary neurospheres (2.0±0.59 fold of control, P = 0.003; n = 8; Fig. 4E). In addition, ISO-1 and shRNA-MIF treatment also attenuated the formation of secondary neurospheres compared to controls (ISO-1, 0.65±0.11 fold of control, P<10⁻⁵; n = 8; shMIF, 0.50±0.11 fold of control, P<10⁻⁵; n = 8; Fig. 4F,G). Thus, MIF may contribute to the self-renewal ability of NSPCs in vitro. As previously mentioned, one of the aims of this study was to identify molecules secreted by cDCs that support the proliferation of NSPCs. Thus, anti-CD74 neutralizing antibody was added to NSPCs and cDCs in a co-culture system established in a previous study (Mikami et al., 2004). The increase in primary neurosphere formation by cDCs was partially attenuated by antibody treatment, indicating that the MIF–CD74 pathway may contribute to DC-induced NSPC activations (supplementary material Fig. S6).

Next, we examined the effects of MIF on the differentiation potential of NSPCs in vitro. E14.5 GE-derived neurospheres were cultured with exogenous MIF (400 ng/ml) for 5 days and then differentiated into three neural lineages (neurons, astrocytes, and oligodendrocytes) in the absence of growth factors. There was no significant difference in the cell differentiation potential of NSPCs by MIF treatment compared to controls (Fig. 5). Thus, MIF did not change the cell fate of NSPCs in vitro, although it did induce the self-renewal ability of NSPCs.

MIF has been reported to activate many signaling pathways including the Erk and Akt pathways in many cells (Meyer-Siegler et al., 2006; Shi et al., 2006; Lue et al., 2007). In the present study, we examined whether MIF could induce the activation of some pathways that have been reported to be important for cell proliferation and/or survival, as well as for the maintenance of self-renewal ability and stem cell properties of NSPCs. MIF activated Erk, a well known cell proliferation marker and Akt, which contributes to NSPC survival (Otaegi et al., 2006; Kalluri et al., 2007), as shown by a higher Erk and Akt phosphorylation status after 30 minutes of MIF treatment, although Erk activity was attenuated at 60 minutes (Fig. 6A). In addition, STAT3 (Ser727) phosphorylation has been reported to be important for the maintenance of the stem cell properties of NSPC (Androutsellis-Theotokis et al., 2006; Nagao et al., 2007). Accordingly, MIF increased STAT3 (Ser727) phosphorylation, although no significant increase of STAT3 (Tyr705) phosphorylation was observed following MIF treatment (Fig. 6B). In contrast, gene silencing of MIF caused the down regulation of Bcl-2 and Bcl-xl, which are known as anti-apoptotic factors. This result is in accordance with the previous result that MIF knockdown in NSPCs increases caspase 3/7 activity, as shown in (Fig. 3E). It has been reported that AMPK is stimulated by MIF in rat cardiomyocytes under hypoxic conditions with an increase in the number of GLUT4 transporters on the cell surface (Miller et al., 2008). Additionally, AMPK agonist (AICAR) is known to increase the cell surface expression of GLUT1 in rat ventricular papillary muscle cells (Li et al., 2004). In our study, MIF treatment increased AMPKα/β phosphorylation (Fig. 6D) and upregulation of the Glut1 gene was also observed (1.83±0.46 increase of control; P = 0.001; n = 3; supplementary material Fig. S7A) along with an increase in GLUT1 translocation to the cell surface (supplementary material Fig. S7B). These results indicate that MIF may support NSPC survival through glucose metabolism. EGFR and VEGF-A are also important factors that support NSPC stem cell properties (Wada et al., 2006; Aguirre et al., 2010). Thus, we confirmed changes in Egfr and Vegf-A gene expression levels with MIF treatment. Egfr upregulation following MIF treatment was confirmed by qRT-PCR analysis (2.13±0.23 fold of control, P = 0.002; n = 3; Fig. 6E). Vegf-A gene expression also statistically increased by MIF treatment (1.37±0.23 fold of control, P = 0.05; Fig. 6F), which is consistent with the study in fibroblasts and hepatocellular carcinoma cells (Ren et al., 2003; Kim et al., 2007). In addition, we assessed whether MIF induces gene expression of Hes3, which is reported to be activated by Stat3-pSer727 in NSPCs (Androutsellis-Theotokis et al., 2006). As expected, Hes3 gene expression was upregulated by MIF (2.09±0.2 fold of control, P = 0.001; n = 3; Fig. 6G), although changes in Hes1 gene expression could not be observed through MIF treatment (1.05±0.1 fold of control, n = 3) under the conditions of this assay.

MIF has been shown to play a role in the migration of cells including monocytes. To extend these findings to NSPCs, we performed a cell migration assay in vitro using a xCELLigence SP system. MIF was added to the bottom chamber and single dissociated NSPCs were seeded onto the cell insert of the upper chamber. The number of cells that migrated through the membrane of the cell insert was defined as the cell migration index. MIF-treated NSPCs displayed increased migration compared to controls. In addition, we performed MIF receptor blockade experiments using CD74, CXCR4, and CXCR2 neutralizing antibodies in this system. CD74 blockade most effectively inhibited MIF-mediated migration among antibodies (Fig. 7A). To further confirm the results, in vitro slice cultures were performed. Lentivirally labeled GFP-positive neurospheres of similar size were plated on the ganglionic eminence (GE) of E14.5 mouse brain slices and migration of GFP-positive cells on the brain slices was observed. Ubiquitous MIF expression was observed by immunohistochemical analysis in the GE of E14.5 mouse brains (Fig. 2C) and MIF gene expression in the GE of E14.5 mouse brains was also confirmed by RT-PCR analysis (supplementary material Fig. S2), suggesting that MIF can act as a chemoattractant for NSPCs in vivo. MIF gene expression in the GE of E14.5 mouse brains was also confirmed by RT-PCR analysis (supplementary material Fig. S2). Thus, ISO-1 was added to the culture medium to block the effects of MIF on NSPC migration. ISO-1 treatment led to a significant reduction in cell migration compared to controls at zone II, as shown in Fig. 7D (0.37±0.06 fold of control, P = 0.05; n = 8; Fig. 7B–E). In addition, we blocked SDF1 using SDF1-neutralizing antibody in the same culture conditions, and we observed a significant reduction of NSPC migration by SDF1 blockade (supplementary material Fig. S8). Taken together, these results indicate MIF involvement in NSPC migration.

In the present study, we identified MIF as a factor secreted by cDCs and NSPCs that can act on its own to support the proliferation and/or survival, as well as the self-renewal of NSPCs in vitro. MIF receptors were expressed in the NSPCs, suggesting that MIF can maintain NSPCs through autocrine and/or paracrine mechanisms. MIF also showed chemokine-like characteristics. These findings indicate that MIF may be effective for the treatment of neurodegenerative diseases.

Many factors can contribute to the proliferation and/or survival of NSPCs, including EGF, FGF2, VEGF, CNTF/LIF, PACAP, and PEDF (Reynolds and Weiss, 1992; Kuhn et al., 1997; Chojnacki et al., 2003; Ohta et al., 2006; Wada et al., 2006; Andreu-Agulló et al., 2009). In this present study, we identified MIF as a proliferation and/or survival factor for NSPCs derived from the GE of E14.5 mice. MIF has been reported to activate ERK (Shi et al., 2006), and this activation was observed in NSPCs. In addition, gene silencing of MIF in NSPCs decreased Bcl2 and Bcl-xl expression and activated caspase 3/7, demonstrating that MIF can act as a survival factor NSPCs. Moreover, the Akt signaling pathway is also known to be important for cell proliferation and survival of NSPCs (Otaegi et al., 2006; Kalluri et al., 2007). Consistent with these previous studies, we showed that MIF treatment activated Akt and led to NSPC proliferation and/or survival. MIF treatment also led to an increase in the number of secondary neurospheres, indicating its ability to increase NSPC self-renewal. EGF and Notch signaling pathways are known to be essential for the maintenance of NSPC progeny (Doetsch et al., 2002; Mizutani et al., 2007; Aguirre et al., 2010). We observed an increase in Stat3-pS727 phosphorylation, which is downstream of the Notch signaling cascade and is important for the survival and/or self-renewal of NSPCs (Androutsellis-Theotokis et al., 2006; Nagao et al., 2007). A previous study showed increased Hes3 expression accompanied by Akt and Stat3-pS727 phosphorylation through Notch activation in NSPCs (Androutsellis-Theotokis et al., 2006), which is consistent with our findings. Furthermore, we observed increases in gene expression of Egfr in NSPCs, indicating that MIF can stimulate EGF signaling pathways, as well as pathways downstream of Notch to induce the self-renewal capacity of NSPCs. In addition, MIF was found to be secreted in the human embryonic-brain-derived-NSPC culture supernatant. This result indicates that MIF may exert the same function on human NSPCs as seen in murine NSPCs.

AMPK is an important regulator of both glycolysis and glucose uptake in NSPCs (Rafalski and Brunet, 2011). AMPK was shown to play an important role in brain development, as indicated by AMPKβ1 knockout mice, which showed defects in cell cycle regulation of NSPCs (Dasgupta and Milbrandt, 2009). In other cell types such as neurons and an immortalized cerebellar cell line, AMPK activation showed a protective effect against glucose deprivation and oxidative stress (Culmsee et al., 2001; Park et al., 2009). Additionally, in the ischemic heart model, MIF showed a protective effect accompanied by activation of AMPK signaling and glucose uptake (Miller et al., 2008). Intriguingly, the induction of Glut1 mRNA was enhanced in hypoxic MIF-overexpressing breast cancer cells (Oda et al., 2008), which coincided with our result that MIF can induce Glut1 gene expression in NSPCs. Thus, MIF may protect NSPCs against cell death through AMPK activation, which correlates with energy metabolism. In future studies, the protective effect of MIF on NSPCs should be examined based on energy metabolism, which is known as a key NSPC regulatory system (Rafalski and Brunet, 2011).

MIF has been reported to enhance the migration of hepatocellular carcinoma cells, melanoma cells, and monocytes (Shimizu et al., 1999; Ren et al., 2003). In addition, the inhibition of MIF binding to CD74 decreases prostate cancer cell invasion (Meyer-Siegler et al., 2006). SDF1, a CXCR4 ligand, has been shown to increase the motility of type A cells in the subventricular zone (SVZ) of the adult mouse (Kokovay et al., 2010) and to promote NSPC transmigration (Imitola et al., 2004). In fact, we observed that SDF1 blockade led to a significant decrease of NSPC migration in the brain slice culture system. In our study, MIF showed an effect similar to that of SDF1 through the enhancement of NSPC migration in vitro. Furthermore, NSPC migration on cultured brain slices was blocked by a MIF-specific inhibitor. Thus, it would be interesting to compare the chemoattractant abilities of MIF and SDF1 in an in vivo model using knock out mice. Taken together, MIF may contribute to NSPC migration in the developmental brain and in other inflammatory conditions seen in stroke and neurodegenerative diseases, as discussed below.

MIF has been shown to have pro-inflammatory effects and to modulate the immune system in diseases such as septic shock, rheumatoid arthritis, and atherogenesis (Bernhagen et al., 1993; Morand et al., 2006). Inflammation is associated with stroke and many neurodegenerative disorders, which include Alzheimer's disease, multiple sclerosis, and Parkinson's disease (Glass et al., 2010). MIF expression in the brain cytosol and cerebrospinal fluid is higher in Alzheimer's disease patients compared to age-matched controls (Bacher et al., 2010). Increased expression of CD74 in neurofibrillary tangles in Alzheimer's disease (Bryan et al., 2008) has been similarly reported, although the detailed function of the MIF–CD74 system in Alzheimer's disease related inflammation remains to be elucidated. Given the complexity and the multiple functions of inflammatory factors, which include cytokines, it is still difficult to understand the roles that these factors play in physiological conditions. Indeed, in some cases, inflammatory responses are beneficial (Wyss-Coray and Mucke, 2002). The effect of inflammatory factors on NSPCs has been studied and a variety of responses have been discovered (Ziv and Schwartz, 2008; Carpentier and Palmer, 2009; Russo et al., 2011). TNF-α, IL-1β, and SDF1 (CXCL12) are capable of increasing NSPC proliferation (Whitney et al., 2009; Wu et al., 2009). In this study, we showed that MIF can increase the proliferation and/or survival of NSPCs, as was seen with SDF1. It is noteworthy that a new role of NSPCs as a potent inducer of anti-inflammatory responses has been reported (Einstein et al., 2003), indicating that the increase in NSPCs number by MIF in neurodegenerative diseases may be a useful therapeutic strategy that is based on immunoregulation. These results define new models of MIF action and establish a link between NSPC maintenance and molecular pathways that are central to inflammation.

In a previous study, we showed functional recovery in a mouse SCI model through the administration of splenic cDCs, which activate endogenous NSPCs. In addition, cDCs increased the number of NSPCs in vitro (Mikami et al., 2004). In this study, we showed that the MIF–CD74 system can contribute in part to cDC activity, which increases the number of NSPCs in vitro, implying that MIF can support functional recovery from SCI through either NSPC activation as seen in cDCs or the recruitment of NSPCs to the injured site, which displays chemoattractant abilities. Although many therapeutic approaches for SCI treatment based on regenerative medicine have been designed, including cell therapy using NSPCs, oligodendrocyte progenitors, and multipotent hair follicle stem cells (Thuret et al., 2006; Amoh et al., 2008, Liu et al., 2011), the results in the present study may contribute to the development of a novel SCI therapy. In addition, MIF upregulation has been reported in stroke patients (Wang et al., 2009). Furthermore, hypoxia-induced MIF secretion by human endothelial cells has been reported (Simons et al., 2011). Thus, detailed analysis of MIF function as it correlates to NSPCs and examination of therapeutic potential of MIF in stroke may be an important issue to address in future.

In conclusion, MIF secreted from DCs and NSPCs was identified as a novel factor that can support the proliferation and/or survival of NSPCs in vitro. MIF also increased self-renewal of NSPCs. MIF activated many signaling pathways that support cell survival, proliferation and/or maintenance of NSPCs. In the present study, we showed that MIF stimulates DC-mediated NSPC proliferation. Taken together, MIF may be a new potential therapeutic factor for the treatment for degenerative brain disorders and SCI through NSPC activation.

Pregnant C57BL/6J and ICR mice were purchased from Sankyo Labo Service (Tokyo, Japan). ICR mice were used only in brain slice culture experiments. All experiments were carried out in accordance with the guidelines of the Experimental Animal Care Committee of Keio University School of Medicine.

NSPCs were isolated from the GE and spinal cord of embryonic day 14.5 (E14.5) mice as previously described, and the cells were cultured as neurospheres (Ohta et al., 2006). These neurospheres were cultured at a cell density of 50 cells/µl in a neurosphere culture medium (NSP medium) consisting of neurobasal medium (Invitrogen) supplemented with B27 (Invitrogen), EGF (20 ng/ml; Peprotech), and FGF2 (10 ng/ml; Peprotech). For the primary and secondary neurosphere formation assay, single, dissociated cells from neurospheres were sorted into a 96-well plate at a low cell density (1 cell/µl or 4 cells/µl), using a fluorescence activated cell sorter (FACS) (EPICS ALTRA; Beckman Coulter). The number of primary and secondary neurospheres in the NSP medium was counted between days 10 and 14 of culture. In the MIF blockade assay, rat anti-CD74 antibody (Santa Cruz Biotechnology), rat anti-CXCR2 antibody (R&D Systems), rat-anti-CXCR2 antibody (R&D Systems), and MIF inhibitor ISO-1 (Calbiochem) were used, and in the gene knockdown assay, retrovirally-expressed MIF shRNA was used. Human neural stem/progenitor cells (NSPCs) were cultured as neurospheres in NSP medium supplemented with EGF (20 ng/ml), FGF2 (10 ng/ml), LIF (10 ng/ml; Millipore), and heparin (5 µg/ml; Sigma) as previously described (Hattori et al., 2007).

cDCs were isolated from mouse splenocytes using anti-CD11c-antibody-conjugated magnetic beads (Miltenyi Biotec). Total RNA was prepared from the cDCs using TRIZOL (Invitrogen), and then subjected to the oligo-cap cDNA library construction procedure following the method developed by Maruyama and Sugano (Maruyama and Sugano, 1994). The full-length cDNAs generated through this method were cloned into a CSII-EF-BSTXI-SV40-hrGFP lentivirus expression vector (supplementary material Fig. S1A) constructed from an original CSII-EF-MCS-IRES-hrGFP vector (kindly given by H. Miyoshi). The lentivirus was produced following the method described by Miyoshi (Miyoshi, 2004) and neurospheres were infected with the lentivirus (MOI = 1). GFP-positive single cells dissociated from neurospheres 5 days post-infection were seeded into 96-well plates at a low density (1 cell/µl). After 10 days of culture, highly proliferated spheres were picked up. After expansion of the spheres, a second round of screening was performed based on their ability to form neurospheres compared to neurospheres infected with control lentivirus. Finally, the genomic DNA of neurospheres was extracted using a Wizard® SV Genomic DNA Purification System (Promega) and the inserted cDNA sequence was amplified using Bend-Taq polymerase (Toyobo) using the PCR primers listed in supplementary material Table S1, The PCR conditions were as follow, 1 cycle of 5 min at 95°C, followed by 35 cycles of 94°C for 30 sec, 58°C for 30 sec, and 72°C for 3 min in PCR reaction buffer containing 2.5% DMSO. For the functional assay, obtained the cDNAs were inserted into pMX-Ig vectors (a gift from T. Kitamura, The University Tokyo).

Total RNA treated with RNase-free DNaseI (Takara Bio) was isolated from tissues or cultured cells using a PureLink RNA Mini Kit (Invitrogen). Synthesis of cDNA was performed using 1 µg of total RNA using PrimeScript RT Master Mix (Takara Bio). The intron-spanning primers were designed as shown in supplementary material Table S1, and for Hes3, previously tested and optimized primer sets were used (Kobayashi et al., 2009). Quantitative RT-PCR analysis was performed with a FastStart Universal SYBR Green Master (Roche), using the ABI prism 7900 HT Sequence Detection System (Applied Biosystems). The PCR conditions were as follows: 1 cycle of 5 min at 95°C, followed by 40 cycles of 95°C for 30 sec, 60°C for 60 sec, and 72°C for 60 sec. Relative gene expression levels were determined using the ΔΔCt method. GAPDH mRNA levels were used as internal normalization control. For semi-quantitative PCR analysis, cDNA samples were amplified using Bend-Taq polymerase and the PCR products were resolved on a 2% agarose gel.

To construct short hairpin RNA (shRNA)-expressing retroviral vectors, oligonucleotides targeting the coding sequence of MIF (5′-GGGUCUACAUCAACUAUUA-3′) and luciferase (Clontech) were inserted into a pSIREN vector (Clontech). Recombinant retroviruses were produced through the transfection of retrovirus vectors and a pVSVG envelope vector into packaging cell line 293GP (kindly given by T. Kitamura).

Cell lysates were prepared using RIPA buffer (25 mM Tris-HCl, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, and 0.1% SDS, pH 7.6) containing protease inhibitors (Cocktail Tablet; Roche). Lysates were centrifuged at 14,000 g for 15 min at 4°C, and the protein concentrations of each sample were determined using a Bio-Rad protein assay kit (Bio-Rad) with bovine serum albumin as a standard. Identical amounts of proteins were electrophoresed in 10% SDS-PAGE gels and transferred to a nitrocellulose membrane. Blots were blocked with Blocking One (Nacalai Tesque) at RT for 1 hr, then incubated with primary antibodies overnight at 4°C as follows: Akt (1∶1000), phospho-Akt (Ser473; 1∶1000), AMPKα (1∶1000), AMPKβ (1∶1000), phospho-AMPAKα (Τhr172) (1∶1000), phospho-AMPAKβ (Ser108) (1∶1000), Bcl-xL (1∶1000), STAT3 (1∶1000), phospho-STAT3(Ser727) (1∶1000), phospho-STAT3(Tyr705) (1∶1000, Cell Signaling Technology), Bcl2 (1∶1000), MIF (1∶200; MBL), Erk (1∶200), phospho-Erk (1∶200, Santa Cruz Biotechnology), and actin (1∶4000; Sigma). After three washes in TBST (20 mM Tris-HCl, 150 mM NaCl, and 0.02% Tween-20, pH 7.4), the blots were incubated with the appropriate secondary antibodies conjugated with horseradish peroxidase (1∶4000, anti-rabbit and anti-mouse; Thermo Scientific) for 1 hr at room temperature. Signals were detected with an ECL-Plus Substrate (GE Healthcare) and exposed to Hyperfilm (GE Healthcare).

The sequences of the mouse and human-specific MIF siRNA, 5'- GGGUCUACAUCAACUAUUAdTdT-3' (MIF siRNA1), mouse-specific siRNA 5'-CCGCAACUACAGUAAGCUGdTdT-3' (MIF siRNA2) and control siRNA 5'-GUACCGCACGUCAUUCGUAUC-3' (Cntl siRNA) were used for gene knockdown experiments. Single cells dissociated from neurospheres were seeded on 6-well or 96-well plates and siRNAs were transfected at a concentration of 50 nM using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instructions.

Cell viability assay were performed using Cell Titer-Glo Luminescent Cell Viability Assay kits (Promega) and a luminometer (Wallac ARVO 1420 multilabel counter; WALLAC OY) according to the manufacturer's protocol. For the apoptosis assay, caspase 3/7 activity was measured using Caspase-Glo 3/7 Assay Kits (Promega) according to the manufacturer's instructions. In both assays, single cells dissociated from neurospheres were seeded onto 96-well plates at a density of 5×10³ cells/well, and caspase activity was assayed 3 days post-infection.

After transcardial perfusion with 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS), fetal brains were removed, fixed in 4% PFA for 2 hours at 4°C, incubated overnight at 4°C with 30% sucrose in PBS, then frozen in OTC compound (Sakura Finetek). In immunohistochemical studies, the coronal sections (14 mm) were incubated in blocking buffer (PBS containing 10% normal goat serum) for 60 min at room temperature. The following primary antibodies were used: rat anti-CD74 (1:100; Santa Cruz Biotechnology), rabbit anti-MIF (1:200; Biovision), and mouse anti-Nestin (1:5; DSHB) antibodies. Brain sections were incubated with primary antibodies overnight at room temperature, then with the appropriate Alexa Flour dye-conjugated secondary antibodies (1:1400; Invitrogen). Cell nuclei were counterstained with To-Pro-3 (Invitrogen) or 4',6-diamino-2-phenylindole (DAPI, Invitrogen). For immunocytochemical studies, cells were fixed with PBS containing 4% PFA for 20 min at room temperature, and the cells were subjected to a blocking procedure. Immunofluorescence staining was performed using the following primary antibodies: rat anti-CD74 (1:200; Santa Cruz Biotechnology), mouse anti-Nestin (1:5; DSHB), mouse anti-β-III-tubulin (1:1000; Sigma), rabbit anti-MBP (1:5; DAKO) and rabbit anti-GFAP (1:200; Biomedical Technologies) antibodies. After PBS washes, antibody binding was visualized using the appropriate Alexa Flour dye-conjugated secondary antibodies (Invitrogen), and the nuclei were stained with To-Pro-3 or DAPI. In the immunohistochemical and immunocytochemical studies, permeabilization was achieved through the addition of 0.3% TritonX-100 to the primary antibody solution (5% normal goat serum in PBS). In the differentiation assays, single dissociated cells of cultured neurospheres were plated on poly-L-lysine-coated glass slips at a density of 2×105 cells/cm² in NSP medium without growth factors for 5 days, then subjected to immunocytochemical analysis. In 5-bromo-2'-deoxyuridine (BrdU) chase experiments, neurospheres were treated with 10 mM BrdU (Sigma) for 24 hrs, then mechanically dissociated and plated onto poly-Lornithine-coated coverslips at a density of 1×105 cells/cm². The cells were processed for BrdU immunocytochemistry 2 hrs after plating. In order to enable the detection of BrdU-labeled cells, these cells were fixed with 4% PFA, pretreated with 2 M HCl for 15 min at 37°C to denature the DNA, and stained with rat anti-BrdU antibody (1:100; Abcam). At least 10 different viewing fields were counted for the analyses. All images were obtained using a Zeiss LSM-510 confocal microscope (Zeiss).

Flow cytometric analyses were performed using Flow cytometry EPICS-XL (Beckman Coulter). Mouse neurospheres were stained with the following antibodies: anti-CD44, anti-CXCR2 (BioLegend), anti-CXCR4, anti-GLUT1 (R&D Systems), and anti-CD74 (Santa Cruz Biotechnology). Cell cycle analysis for live cells was performed using flow cytometry. Cells were stained with Hoechst 33342 for 60 min at 37°C, then subjected to flow cytometry (EPICS-ALTRA, Beckman Coulter). Retrovirally GFP-labeled cells were gated and then analyzed using Multicycle for Windows (Beckman Coulter).

Human and mouse MIF ELISA kits (Sapporo Immuno Diagnostic Laboratory) were used to measure the amount of secreted MIF protein in culture supernatants. Mouse dendritic cells were isolated from spleens using anti-CD11c antibody-conjugated magnetic beads (Miltenyi Biotec).

Cell migration of the cells was tracked in vitro by impedance signal changes in CIM plates (Roche) measured on the back-side of the membrane using the CELLigence SP system (Roche), a real-time migration monitoring device. MIF-mediated NSPC chemotaxis was blocked by the antibodies described above. In the migration assays on brain slices (400 µm thickness), E14.5 ICR mouse brain slices were placed on cell inserts (Millipore). Lentivirally GFP-labeled neurosphere colonies (200 µm diameter) were placed on coronal brain slices and co-cultured in DMEM/F12 medium containing N2 supplement and T3 (30 µg/ml). GFP fluorescence intensity of cells migrating from the neurospheres to zones I and II (200×300 µm) was measured using ZEN lite 2011 software (Zeiss). The distance from the center of the neurosphere to the center of zone II (outer area) was 800 µm. Zone I is the contiguous inner area closer to the sphere. In the neutralization assay, mouse IgG and monoclonal anti-SDF1 neutralizing antibody (R&D) were added to the medium.

All values are expressed as means ± standard deviation (s.d.). Student's t-tests were used to determine the statistical significance of differences between groups. For the nonparametric multiple comparison procedure, Steel–Dwass tests were performed to compare levels between groups (*P<0.05, **P<0.01).

The authors would like to acknowledge T. Kitamura (The University of Tokyo) for providing the pMX-Ig vector, Hiroyuki Miyoshi (RIKEN) for providing the CSII-EF-MCS-IRES-hrGFP vector, Y. Mizue (Sapporo Immuno Diagnostic Laboratory), F. Renault-Mihara, R. Kuwahara and S. Teramoto (Keio University) for technical assistance.