Many cell types in the brain express chemokines and chemokine receptors under homeostatic conditions, arguing for a role of these proteins in normal brain processes. Because chemokines have been shown to regulate hematopoietic progenitor cell proliferation, we hypothesized that chemokines would regulate neural progenitor cell (NPC) proliferation as well. Here we show that chemokines activating CXCR4 or CCR3 reversibly inhibit NPC proliferation in isolated cells, neurospheres, and in hippocampal slice cultures. Cells induced into quiescence by chemokines maintain their multipotential ability to form both neurons and astrocytes. The mechanism of chemokine action appears to be a reduction of extracellular signal-related kinase phosphorylation as well as an increase in Reelin expression. The inhibitory effects of chemokines are blocked by heparan sulfate and apolipoprotein E3 but not apolipoprotein E4, suggesting a regulatory role of these molecules on the effects of chemokines. Additionally, we found that the chemokine fractalkine promotes survival of NPCs. In addition to their role in chemotaxis, chemokines affect both the survival and proliferation of human NPCs in vitro. The presence of constitutively expressed chemokines in the brain argues that under homeostatic conditions, chemokines promote survival but maintain NPCs in a quiescent state. Our studies also suggest a link between inflammatory chemokine production and the inhibition of neurogenesis.
Neural progenitor cells (NPCs) are capable of forming both neurons and glia upon differentiation, and are present both during development of the brain and in adulthood in a broad range of mammals including humans [1, 2]. The most primitive neural stem cells (NSCs) appear to undergo asymmetric cell division, forming a daughter stem cell (SC) and a more restricted progenitor cell . These progenitor cells are then capable of forming neurons and astrocytes . However, not all progenitor cells in the adult brain enter the cell cycle. Some progenitor cells exist in a quiescent state that is regulated by the local environment . Some molecules implicated in the regulation of NPC proliferation include cortisol , retinoic acid , opioids , and glutamate . The extent to which these molecules play a role in vivo is not well described. Because of the need to tightly control cell division in the adult, other molecules likely play a role in the regulation of cell division under homeostatic conditions.
One group of molecules that has been shown to regulate cell division in other systems is the chemokine group of cytokines. These are small peptides secreted by a variety of cell types, originally described because of their ability to induce chemotaxis in inflammatory cells. They have been shown to regulate diverse processes such as hematopoiesis  and angiogenesis . Interestingly, several cell types in the mammalian brain have been shown to possess chemokine receptors . This includes CCR3, CXCR4, CXCR2, and CX3CR1 expression on neurons, CXCR4 expression on astrocytes, and CCR3 and CCR5 expression on microglia. The actual function of these receptors in the developed nervous system is not known. However, several parallels between hematopoiesis and the development of NPCs have been noted, including the involvement of many of the same cytokines in both systems . Because quiescence is forced upon proliferating hematopoietic stem cells (HSCs) by the chemokine family of molecules , one possible role of chemokine on NPC development might be the control of cellular proliferation. However, no similar studies with chemokines have been undertaken with NPCs. Additionally, some chemokines support the survival of (HSCs)  but have unknown effects on NSC survival. Because the chemokines stromal derived factor-1 (SDF-1) and fractalkine are produced constitutively in the brain, and several others are produced during inflammation , we hypothesized that chemokines would affect neural SC survival and proliferation. To test this hypothesis, we studied the effects of chemokines on isolated colonies of human NPCs and slice cultures of human hippocampus. We found that indeed specific chemokines promote both survival and quiescence of NPCs.
Materials and Methods
Cells and Reagents
Normal human NPCs were purchased from Clonetics Corp. (San Diego, CA; http://www.informagen.com) or from Clonexpress, Inc. (Gaithersburg, MD; http://www.clonexpress.com). These cells were obtained from human fetal tissue between 14 and 21 weeks post conception, and were expanded by the manufacturer according to previously described methods . Cells have been quality controlled by the manufacturer to express microtubule associated protein-2 (MAP 2) and beta Tubulin III. Cells were grown according to manufacturer's directions in Dulbecco's modified Eagle's medium (DMEM):HAMS F12 (50:50), gentamicin 30 mg/ml, amphotericin B 15 μg/ml, human recombinant basic fibroblast growth factor (FGF-B) and epidermal growth factor (EGF) both at 20 ng/ml, and N2 1:100, (neural progenitor maintenance medium [NPMM]; Clonetics; San Diego, CA; http://www.informagen.com). Cells exhibited robust proliferation and formed neurospheres in culture, and formed both neurons and astrocytes when plated onto laminin-coated plates as assessed by neuronal nuclear antigen (NeuN) and glial fibrillary acidic protein (GFAP) staining. For monolayer cultures, DMEM:HAMS F12, gentamicin, amphotericin B, FGF-B and N2 were used. Cells were assayed by flow cytometry and determined to contain less than 2% microglial cells by staining for microglial markers CD14 and HLA class II (not shown). Chemokines were obtained from R&D Systems (Minneapolis, MN; http://www.rndsystems.com) and were used at a concentration of 100 ng/ml. Recombinant apolipoproteins were from Pan Vera (Madison, WI; http://www.invitrogen.com) and were used at a concentration of 20 nM. Heparan sulfate (HS) and heparin were from Sigma (St. Louis, MO; http://www.sigmaaldrich.com) and were both used at a concentration of 5 μg/ml. Fluorescein isothiocyanate (FITC)-conjugated anti-Reelin monoclonal was from Santa Cruz Biotechnology (Santa Cruz, CA; http://www.scbt.com) and was used at a 1:25 dilution.
Chemokine Receptor Detection
Chemokine receptors were detected using fluorescently labeled monoclonal antibodies to CCR3, CCR5 or CXCR4 (R&D Systems). Briefly, cells were washed and treated with a 1:100 dilution of the appropriate antibody. After incubating for 1 hour, cells were washed and flow cytometric analysis was performed using the FACScan analyzer (Becton Dickinson; San Jose, CA; http://www.bd.com). Triplicate determinations of receptor expression were performed. An irrelevant mouse monoclonal antibody of the same isotype served as control for non-specific staining.
Culture of NPC
For proliferation analysis, individual NPCs were grown by limiting dilution as described  in 96 well plates, with or without chemokines. Cultures were analyzed after 5 days for the formation of colonies. The proliferation of an average of 36 individual cells per chemokine was assessed by microscopy. The entire experiment was repeated at least twice. Proliferation was confirmed by growing NPCs as monolayers in 24-well plates with or without chemokines, then labeling cells with the addition of 5 μM bromodeoxyuridine (BrdU) for 24 hours on days 2 or 7 of culture. Each chemokine was tested in triplicate. BrdU incorporation was assessed by a BrdU flow cytometry kit (Pharmingen; San Diego, CA; http://www.bdbiosciences.com/pharmingen). The experiment was repeated at least once. Multipotential ability was assessed by transferring NPCs to laminin-coated plates, culturing for 7 days, then assessing for NeuN and GFAP expression. Cells were fixed in 50:50 acetone:methanol, washed, and antibodies to NeuN (Chemicon; Temecula, CA; http://www.chemicon.com) or GFAP (Pharmingen) were applied for 1 hour at 1:25 or 1:10 dilution, respectively. A secondary antibody conjugated to FITC (Jackson Immunoresearch Laboratories; West Grove, PA; http://www.jacksonimmuno.com) was then used at 1:500 dilution for detection. For survival analysis, NPCs were grown in 50:50 HAMS F12:DMEM with 10% fetal bovine serum (FBS) and antibiotics but without FGF-B, EGF and N2, testing each chemokine in triplicate. Viability was assayed at day 5 by trypan blue exclusion and confirmed by using a kit for fluorescent detection of Annexin V (Molecular Probes; Eugene, OR; http://www.probes.com). The entire experiment was performed at least twice. Fluorescence was then assayed by flow cytometry. For neurospheres analysis, an average of 100 neurospheres/well were grown with or without chemokines at 100 ng/ml (R&D Systems) for 7 days on non-coated 24-well plates. Each chemokine was tested in triplicate. Cells were then plated onto 24-well polyethylene-imine coated plates. The number of cell colonies with >20 cells with typical neuronal morphology was counted for each well after 7 days in culture. All chemokines were tested in triplicate; the entire experiment was performed at least four times on at least four different lots of NPCs.
Signaling was assayed by flow cytometry as described . Briefly, 105 isolated NPCs were incubated with chemokines at 100 ng/ml for 10 minutes, then fixed and permeabilized with saponin and stained with FITC-conjugated phosphospecific antibodies to extracellular signal-related kinase (ERK), or p38 (Santa Cruz Biotechnology). Each chemokine was tested in triplicate. Cells were then analyzed by flow cytometry for expression of the phosphorylated signaling molecules. Activity of PP2A was blocked by incubating isolated NPCs with okadaic acid (OA) (Calbiochem; San Diego, CA; http://www.calbiochem.com) 2.5 μM for 1 hour prior to the addition of chemokines. Cells were then analyzed for the degree of phosphorylation of (p)-ERK by FITC-labeled phosphospecific antibodies. All experiments were performed at least three times with identical results.
Hippocampal Slice Cultures
Hippocampal tissue was obtained from patients aged 21–50 years undergoing corrective surgery for seizures as previously described [20, 21], which contains abundant NPCs , and was done according to institutional guidelines for human subjects research. Tissue was obtained within 1 hour of removal and cultured as described . Briefly, tissue was sliced to 200 μm thickness by a Brendel-Vitron tissue slicer (Vitron, Inc; Tucson, AZ; http://www.vitron.com). Slices were placed in a Transwell insert (Costar; Corning, NY; http://www.corning.com/lifesciences) and cultured in NPMM with or without chemokines at 100 ng/ml. Chemokines were tested in triplicate; the entire experiment was performed at least twice with identical results. Slices were labeled with BrdU on day 5 of culture, and harvested for immunohistochemistry (IHC) on day 7 of culture. Slices of brain tissue were placed in OCT and frozen prior to sectioning into 5-micron sections. BrdU incorporation into cells was detected by a BrdU in situ kit following the manufacturer's instructions (Pharmingen). Double-label experiments were performed using an FITC-labeled anti-BrdU antibody together with an anti-Musashi-1 (Msi-1) antibody (Chemicon) at 1:200, anti-GFAP at 1:10 or anti-isolectin (Vector Laboratories; Burlingame, CA; http://www.vectorlabs.com) at 1:10 dilution detected by a Texas Red-labeled secondary antibody (Jackson Immunoresearch Laboratories). Apoptotic cells were detected by a TUNEL assay (Oncogene; Boston, MA; http://www.oncogene.com).
Expression of Chemokine Receptors on NPC
We first sought to establish which chemokine receptors are present on NPCs. Both hippocampal neurons and HSCs have been shown to express CCR5, CXCR4, and CCR3 [24–26]. We therefore analyzed isolated human NPCs by flow cytometry to determine whether they would similarly express CCR3, CCR5, or CXCR4. In multiple experiments, surface expression of CCR3 was seen on 55%–69% of cells, while CXCR4 was seen on 65%–85% of cells (Fig. 1). In contrast, CCR5 expression was detected on less than 5% of cells. These results show that like HSCs, NPCs express CCR3 and CXCR4 but unlike HSCs, NPCs do not express appreciable levels of CCR5.
Chemokines Promote Quiescence of NPC
To determine the role of these chemokine receptors on NPCs, we tested the ability of chemokines to affect the proliferation of NPCs. Isolated NPCs were cultured by limiting dilution with and without chemokines that bind to CCR3, CCR5 or CXCR4; namely, eotaxin, microphage inhibitory protein 1 alpha (MIP-1α), or SDF-1. We found that eotaxin and SDF-1 prevented cell division by NPCs (Fig. 2A) while MIP-1α allowed cells to proliferate and form colonies of tightly packed cells similar to controls. In order to determine whether cells treated with SDF-1 and eotaxin were quiescent, cells were washed and placed in fresh media without chemokines. After 5 days, individual cells showed evidence of proliferation (Fig. 2A). To confirm that cells ceased proliferation in response to chemokines, NPCs were cultured with or without chemokines and BrdU was added on day 2 or day 7 of culture. The incorporation of BrdU was analyzed 24 hours after addition using flow cytometric detection of BrdU. We found that on day 2, chemokines that can activate CCR3 or CXCR4 inhibited BrdU incorporation, while chemokines that activate CCR5 did not (Fig. 2B). By day 7, this inhibitory effect was no longer seen, and chemokine-treated cultures showed no difference in BrdU uptake compared to proliferating controls, indicating that this inhibition is transient (not shown). Thus chemokines that bind to the CCR3 or CXCR4 receptors present on NPCs inhibit cellular proliferation and induce cellular quiescence, but allow cells to maintain their proliferative potential.
To further confirm that chemokines were inducing NPCs to become quiescent, we tested whether NPCs exposed to chemokines maintained their multipotential ability. NPCs that had been treated with chemokines were washed and plated onto laminin-coated plates to induce differentiation. We found that once chemokines were removed, even SDF-1- or eotaxin-treated NPCs could still form neurons as determined by NeuN staining (Fig. 2C) and GFAP staining (Fig. 2E).
NPCs can also be grown as clusters of cells termed neurospheres. These spherules of cells have been used as models of NPC biology in vitro and in vivo [13, 27]. Growing cells as spherules maintains cell-to-cell contact, tight junctions, and prevents the trauma of passaging monolayers of cells. Thus these neurospheres may be more representative of cells in vivo. We tested whether chemokines would also affect NPCs grown as spherules, reasoning that if chemokines induced quiescence in these cells, the spherules would fail to attach and begin a program of differentiation into neurons and astrocytes. Counting the number of attached and differentiated colonies then allows a measure of the ability of each chemokine to induce cellular quiescence. We again tested chemokines that bind to CCR3 and CXCR4 receptors as well as several chemokines that do not. We found that incubation of NPCs with the chemokines MCP-4, eotaxin, and RANTES, all of which can bind to CCR3 , decreased the number of colonies formed by 77%, 68% and 79%, respectively (Fig. 3). The CXCR4-binding chemokine SDF-1 decreased the number of colonies formed by 61%. In contrast, the chemokines MIP-1α and MCP-1, which can bind to CCR1, CCR2, or CCR5, had no effect on colony formation. The ability of the CCR3 receptor to decrease colony formation was confirmed by using a monoclonal antibody to CCR3 in the NPC culture, which also decreased colony formation by 50%. To rule out the possibility that chemokines were cytotoxic, we removed unattached spherules from SDF-1 and eotaxin-treated cultures after plating. These spherules were washed and replated on coated plates in fresh media. We found that these cells were able to attach and differentiate (not shown), thus confirming that the inhibition by chemokines is transient. These studies show that chemokines that activate either the CCR3 or CXCR4 receptor can inhibit the growth and differentiation of NPCs in neurospheres.
Effects of Chemokines on Reelin Expression
Because Reelin has been shown to prevent attachment of NPCs to fibronectin , we reasoned that chemokines might be inducing Reelin expression in NPCs that would then prevent the attachment of neurospheres to coated surfaces. We tested chemokine-treated NPCs for Reelin expression and found that SDF-1 increased Reelin expression compared to controls, but MIP-1α did not (Fig. 4). This result provides further mechanistic evidence for how chemokines induce quiescence in NPCs.
Chemokines Reduce Phosphorylation of ERK
We next asked what signaling pathway might be activated by chemokines to result in reduced proliferation. Because the ERK inhibitor PD98059 has been shown to block the proliferation of NPCs , we hypothesized that the MAPK pathway is affected by chemokines. Therefore, NPCs were incubated with chemokines for 10 minutes then assayed for intracellular expression of the phosphorylated MAPK family member ERK. Consistent with ongoing proliferation in these cells, we found a high basal level of p-ERK (Fig. 5A, 5B) and p38 (not shown). We found that eotaxin reduced p-ERK by 42%. SDF-1 reduced p-ERK by 37%. In contrast, MIP-1α showed a non-significant reduction of 20%. Because a reduction of phosphorylation might imply an increase in phosphatase activity, the PP2A phosphatase inhibitor OA was added to the cultures prior to the addition of chemokines. The phosphatase inhibitor blocked the reduction in p-ERK induced by eotaxin and SDF-1 (Fig. 5B), suggesting that phosphatases are activated by these chemokines. Neither chemokine, however, had any effect on the level of p38 phosphorylation (not shown). We also tested the ability of OA to block the inhibitory effects of eotaxin in cultured NPCs. We found that OA completely reversed the inhibitory effects of eotaxin on cultured NPCs (Fig. 3). These results demonstrate that specific chemokines can decrease the p-ERK by activating cellular phosphatases.
Fractalkine Promotes Survival of NPC
Chemokines have been shown to act as survival factors for other types of SCs . In agreement with this role, we did not detect an increase in apoptosis in chemokine-treated, isolated NPCs (95% versus 96% viability by trypan blue exclusion), or in neurospheres. To determine whether chemokines would promote survival of NPCs during growth factor withdrawal, we tested SDF-1 and fractalkine, both of which are constitutively expressed in the brain, as well as the chemokines that failed to inhibit proliferation. We grew NPCs in media without FGF-B, EGF, and N2 to induce apoptosis, adding chemokines to some cultures to determine whether they would promote survival. As measured by trypan blue analysis, fractalkine increased viability of NPCs under these conditions from 50% in control cultures to 75% in fractalkine-treated cultures. The chemokines MCP-1, SDF-1, and MIP-1α showed 48%, 53%, and 44% viability in cultures, respectively, that were not statistically different than control cultures. We confirmed this activity of fractalkine by measuring Annexin V staining in chemokine-treated NPC cultures by flow cytometry. Fractalkine-treated cultures showed only 50% of cells expressing Annexin V, compared to 69.1% of cells in the control (Fig. 6). This demonstrates that chemokines also act as survival agents for NPCs, similar to the survival role ascribed to fractalkine on mature neurons.
Apolipoproteins and Heparan Sulfate Modulate the Effects of Chemokines on NPC
If chemokines promote quiescence of NPCs, then a mechanism must exist to counteract this inhibition under conditions requiring new neuron formation. Conditions such as development, injury or inflammation, where new neurons are needed, have been associated with alterations in HS expression  and apolipoprotein expression . We hypothesized that HS might bind directly to chemokines, while the apolipoproteins might compete indirectly with chemokines for HS on the cell surface. Thus alterations in either HS or apolipoproteins could abrogate the inhibitory effect of chemokines. We tested the ability of exogenous HS to prevent the inhibitory effects of chemokines on NPCs (Fig. 7). We found that HS prevented SDF-1 (17% of control with SDF-1 alone versus 86% of control with HS + SDF-1, p < 0.001) and eotaxin (29% of control with eotaxin alone versus 71% of control with HS + eotaxin, p < 0.001)-induced inhibition of NPC proliferation in isolated colonies as well as in hippocampal slice cultures with eotaxin (21% of control with eotaxin alone versus 207% of control with HS + eotaxin, p < 0.01). A closely related glycosaminoglycan, heparin, could not block the effect of eotaxin in isolated colonies (29% of control versus 23% of control with heparin + eotaxin, p = not significant [n.s.]) or in slice cultures (21% of control versus 29% of control with heparin + eotaxin, p = n.s.). We also tested recombinant apolipoproteins E3 and E4 for their ability to block the effects of chemokines on NPCs (Fig. 7). We found that while apolipoprotein E3 prevented chemokine inhibition of NPC proliferation by SDF-1 in isolated colonies (17% of control with SDF-1 alone versus 148% of control with SDF-1 + Apo E3, p < 0.001), apolipoprotein E4 did not (17% of control alone versus 42% of control with SDF-1 + Apo E4, p = n.s.). We conclude that HS and apolipoproteins can modulate the effects of chemokines on NPCs, although not all apolipoproteins share this ability.
Chemokines Promote Quiescence of NPC in Human Hippocampal Slices
The local environment has been shown to be an important factor in the behavior of SCs in vivo . We therefore sought to determine whether chemokines affect NPCs in the complex environment of the whole brain. Human hippocampal tissue was obtained from patients aged 21–50 years and grown as thin slices in tissue culture. Cultures were treated with a variety of chemokines, and pulsed with BrdU as a marker for proliferating NPCs. Cultures were harvested and IHC for BrdU-positive cells was performed. We found that the number of BrdU positive cells was decreased in cultures treated with eotaxin and SDF-1 by 79% and 71%, respectively, but not with MIP-1α or fractalkine (Fig. 8). To determine whether chemokines act only to induce quiescence of NPCs or are actively inducing apoptosis in slice cultures, a TUNEL assay was performed on the chemokine-treated hippocampal slices. No difference in apoptosis was seen in treated versus untreated cultures (MIP-1α, eotaxin, and SDF-1 showed 20.5, 29.5 and 29 apoptotic cells per high power field, respectively, versus 27.5 for control). To confirm that BrdU-stained cells were in fact NPCs, we tested cells for expression of the NPC-associated marker Msi-1. We stained slices with fluorescent monoclonal antibodies to BrdU and either Msi-1, GFAP, or isolectin B4. When slices were examined, the vast majority of BrdU-labeled cells were also Msi-1+ (Fig. 9) but were GFAP and isolectin B4 negative (not shown), thus confirming their status as NPCs and not astrocytes or microglia. Thus, chemokines act specifically on NPCs in whole brain cultures to decrease proliferation.
Our results suggest that chemokines play an important role in regulating NPC proliferation and survival. While one study suggested that SDF-1 alone had no effect on cerebellar granule cell proliferation but could enhance Sonic hedgehog-induced proliferation , studies with hematopoietic progenitors have shown opposing effects for chemokines and sonic hedgehog on proliferation [14, 35], which is consistent with our studies. It could also be that SDF-1 has different effects depending on the degree of SC differentiation, similar to the enhancing effect of MIP-1α on more mature hematopoietic progenitor cell proliferation but an inhibitory effect on more primitive hematopoietic progenitors . Other studies have examined the role of either SDF-1 or CXCR4 on granule cells in the dentate gyrus and found that in the absence of signaling through this chemokine receptor, granule cells either continue to proliferate aberrantly  or differentiate before reaching their target tissue . In light of our data, this could be explained by a lack of the induction of quiescence leading to continued proliferation or premature differentiation of NPCs. We also found that SDF-1 promotes Reelin expression, which could explain why the lack of SDF-1 or CXCR4 leads to a lack of proliferating progenitor cells in the dentate gyrus  that is similar to that seen in reeler mutants . Our studies demonstrate a new role for chemokines in the brain, and suggest that insults to the brain that induce chemokines may prevent NPC-dependent repair processes or NPC-dependent memory formation. Our studies may also explain why those with the apolipoprotein E4 genotype are more susceptible to various insults  since this apolipoprotein failed to prevent the inhibition of NPC proliferation by chemokines that may be produced during inflammation.
We gratefully acknowledge the technical help of Diane Trussoni in preparing the tissue sections, and the surgical expertise of Dr. Robert Maxwell in obtaining the hippocampal specimens. This work was supported in part by NIH grant K08 01544 and the Great Lakes Regional Center for AIDS Research Developmental Award.