Excitatory amino acid transporter 2 and excitatory amino acid transporter 1 negatively regulate calcium-dependent proliferation of hippocampal neural progenitor cells and are persistently upregulated after injury

Cloning

GltI cDNA was purchased in plasmid form (pFLCI–GltI) from Geneservice (Cambridge, UK; catalog no. 6330500I22), and Glast mouse cDNA (pCMV6–Glast–Myc) was purchased from Origene (Rockville, MD, USA; catalog no. MR208665). GltI cDNA was subcloned into pCMV–IRES–DsRed2 (Clontech, Mountain View, CA, USA; catalog no. 632420) in order to create the GltI overexpression plasmid pCMV–GltI–DsRed2. The Glast cDNA construct (pCMV6–Glast–Myc) was directly used to overexpress Glast.

Transient transfections

P7 dentate gyrus-derived neurospheres were dissociated into single cells and transfected with either siRNA compounds or overexpression plasmids driven by the cytomegalovirus (CMV) promoter. Lipofectamine RNAi Max (Invitrogen, Eugene, OR, USA; catalog no. 56531) was used to reverse transfect 10 nmGltI siRNA (Sigma-Aldrich, St Louis, MO, USA; catalog no. SASI_Rn01_00062046), 10 nmGlast siRNA (Sigma-Aldrich; catalog no. SASI_Rn02_00262083) or 10 nm scrambled siRNA (Applied Biosystems, Carlsbad, CA, USA; catalog no. AM4611). Lipofectamine 2000 (Invitrogen; catalog no. 52887) was used to transiently transfect 300 ng of pCMV–GltI–DsRed2 (final concentration, 150 ng/mL), 300 ng of pCMV–Glast–Myc (final concentration, 150 ng/mL), or 300 ng of pCMV–DsRed2 (Clontech; catalog no. 632420) (final concentration, 150 ng/mL). Seven days after cells were transfected, some neurospheres were taken from each experiment and misexpression of GltI and Glast was confirmed by PCR (QPCR) analysis. Only transfections with confirmed knockdown or overexpression were used for further experiments. The siRNA sequences were as follows: GltI sense, 5′-GAU AGU GAC UGU AAG CCU U-3′; GltI antisense, 5′-AAG GCU UAC AGU CAC UAU C-3′; Glast sense, 5′-CUG UCA UUG UGG GUA CAA U-3′; and Glast antisense, 5′-AUU GUA CCC ACA AUG ACA G-3′.

QPCR

Knockdown and overexpression of GltI and Glast were confirmed with methods adapted from previous publications (Gilley et al., 2011). Seven days after transfection, RNA was isolated from neurospheres and reversed transcribed into cDNA, and relative mRNA levels were measured. mRNA levels were normalized to either GltI or Glast gene expression. The primer information is as follows: glyceraldehyde-3-phosphate dehydrogenase forward, 5′-CTC AAC TAC ATG GTC TAC ATG TTC CA-3′; glyceraldehyde-3-phosphate dehydrogenase reverse, 5′-CCA TTC TCG GCC TTG ACT GT-3′; GltI forward, 5′-GGA AGA TGG GTG AAC AGG C-3′; GltI reverse, 5′-TTC CCA CAA ATC AAG CAG G-3′; Glast forward, 5′-ACG GTC ACT GCT GTC ATT G-3′; and Glast reverse, 5′-TGT GAC GAG ACT GGA GAT GA-3′.

Neurosphere growth and assays

Neurobasal A medium lacking glutamate and aspartate was used in the preparation and growth of neurospheres (Invitrogen; catalog no. 10888-022). The methods for preparation of activated papain, neural stem cell medium and semi-solid medium were adapted from previously described protocols. All neurospheres were derived from P7 dentate gyrus and grown in neural stem cell medium for 7 days before they were used for further experiments [see Gilley et al. (2011) for prior protocols].

Misexpression neurosphere assays were performed in glutamate-free medium on neurospheres transfected with either siRNA or overexpression plasmids (see ‘Transient transfections’). Briefly, P7 neurospheres were dissociated into single cells, transiently transfected with Lipofectamine, and grown in neural stem cell medium for 7 days. Transfected neurospheres were dissociated again, counted, and plated in semi-solid medium before they were quantified 7 days later. This experiment was repeated four times, and the results were analyzed for significance with one-way anova and Bonferonni post hoc analysis.

For the glutamate neurosphere assays, neurospheres were grown in glutamate-free medium for 7 days, mechanically dissociated into a single-cell solution, and plated in semi-solid medium supplemented with either 0 μm or 5 μm glutamate. Additionally, we treated some of the cells with 10 μm (2S)-2-amino-2-[(1S,2S)-2-carboxycycloprop-1-yl]-3-(xanth-9-yl)propanoic acid (LY341495), which serves as a non-selective metabotropic glutamate receptor (mGluR) antagonist at this concentration (Kingston et al., 1998). Another group of cells were treated with 1 μm 1,2-bis(2-amino-5-fluorophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis ester (BAPTA-AM), which serves as an intracellular calcium chelator (Santa Cruz, Santa Cruz, CA, USA; catalog no. sc-202488). After 7 days in culture, neurospheres were quantified, and the diameters of at least 20 neurospheres from each group were measured (Gilley et al., 2011). An unpaired t-test and Bonferonni post hoc analysis were used to calculate significance. The experiment was repeated at least four times for each condition.

Bromodeoxyuridine (BrdU) incorporation

Transiently transfected neurospheres were pulsed with 10 μm BrdU for 15 min, dissociated with activated papain, and triturated into a single-cell solution before they were fixed overnight with 100% ethanol; 2 m HCl/0.5% Triton X in phosphate-buffered saline (PBS) was used to denature DNA for 30 min at room temperature. The reaction mixture was then neutralized with 0.1 m NaB₄O₇ before being incubated in staining solution [1.3 μL of BrdU-APC (BD Pharmingen, San Diego, CA, USA; catalog no. 552598), 5 μL of RNase and 50 μL of 0.5% Tween/1% bovine serum albumin in PBS] overnight at 4 °C. Cells were then washed and resuspended in PBS with propidium iodide (PI), and analyzed for BrdU incorporation by flow cytometry. Results are displayed as the percentage of cells that were BrdU⁺ and PI⁻. The experiment was performed in quadruplicate. Statistical significance was determined with a one-way anova followed by Bonferonni post hoc analysis.

Intracellular calcium measurement

After misexpression was confirmed 7 days after transfection, all neurospheres were dissociated and stained with Fluo4-AM calcium indicator (Invitrogen; catalog no. F14201) after cells had been exposed to a variety of compounds, including an intracellular calcium chelator and an mGluR antagonist. Cells were incubated with 1 μm BAPTA-AM according to the manufacturer’s protocols. Additionally, 10 μm LY341495, which acts as a non-selective mGluR antagonist at high concentrations, was added to a proportion of dissociated progenitors (Kingston et al., 1998). Cells were then analyzed by flow cytometry. Prior to supplementation with glutamate, baseline intracellular calcium levels were measured for 30 s. Cells were then spiked with 5 μm glutamate, and calcium measurements were made for an additional 1.5 min. The percentage of cells containing increased intracellular calcium levels (as compared with baseline) is presented. The final percentage of cells expressing Fluo4-AM was normalized to the percentage of positive cells at baseline levels (% induced cells/% baseline cells). Intracellular calcium was measured in quadruplicate for each condition, and significance was calculated with a one-way anova and Bonferonni post hoc analysis.

Mice

All animal experiments were approved by the Institutional Animal Use and Care Committee at UT Southwestern Medical Center. The Animal Resource Center within UT Southwestern, which is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care, humanely housed and cared for all animals.

All experiments were performed with nestin–rtTA–enhanced GFP transgenic mice in a CD1 background, which have been well characterized (Miles & Kernie, 2006, 2008; Shi et al., 2007; Koch et al., 2008; Yu et al., 2008; Gilley et al., 2011). P28 transgenic mice were exposed to either hypoxic–ischemic injury (HI) or sham injury before being subjected to traumatic brain injury (TBI). Both groups of mice (sham + TBI or HI + TBI) were anesthetized, placed in a supine position, had their right carotid artery ligated, and were allowed to recover for 30–60 min (see below). Mice in the HI + TBI group were then exposed to 8% oxygen for 1 h in a hypoxic chamber before they were allowed to recover in room air. Mice in the sham + TBI group were not treated with hypoxia. At P60, mice from both groups were injured by controlled cortical impact (CCI) in order to induce TBI. From P64 to P67, mice were injected with a single daily dose of 100 mg/kg BrdU, and they were perfused and killed 2 h after injection on P67.

The Rice–Vanucci model was used to emulate HI, and has been described previously (Koch et al., 2008; Miles & Kernie, 2008). Briefly, mice were anesthetized and maintained on isoflurane mixed with oxygen and nitrogen. A midline incision was made, and the right carotid artery was isolated and ligated with sutures. In sham injuries, the artery was exposed but not ligated. After the incision was closed, animals were allowed to recover for up to 1 h. Only mice that fully recovered were used for further experiments. After recovery, mice were placed in a hypoxic chamber and treated with 8% oxygen for 1 h. Mice were then allowed to recover in their cage.

A CCI devise was used to induce TBI as previously described (Kernie et al., 2001). Thirty-two days after HI or sham injury (P60), mice were anesthetized and placed in a stereotactic frame, and part of the skull was removed before the brain was impacted with a steel-tipped device in order to generate the injury. The incision was closed, and the mice were allowed to recover. Daily injections of BrdU (100 mg/kg) were started 4 days after TBI at P64, and mice were killed and perfused at P67.

Immunostaining

One, 3, 7 and 45 days after HI, 50-μm vibratome sections were obtained and stained for the presence of GltI and Glast. After being blocked with normal donkey serum, sections were stained with chicken anti-GFP (1 : 500; Aves Labs, Tigard, OR, USA; catalog no. GFP-1020), mouse anti-GFAP (1 : 400; BD Biosciences, Rockville, MD, USA; catalog no. 556330), and guinea pig anti-GltI (catalog no. AB1783) or anti-Glast (catalog no. AB1782) (both 1 : 200; Millipore, Temecula, CA, USA). Fluorescent secondary antibodies were used at a concentration of 1 : 200 (Santa Cruz). See Gilley et al. (2011) for details.

Vibratome sections from HI + TBI and sham + TBI mice were blocked in normal donkey serum, stained with rat anti-BrdU (1 : 400; Abcam, Cambridge, MA, USA; catalog no. AB6326) and chicken anti-GFP 1 : 500; Aves Labs; catalog no. GFP-1020), according to previously described protocols (Gilley et al., 2011).

Only sections in which antibody and signal penetration were even were used for confocal imaging. Percentages of BrdU⁺/GFP⁺ cells were quantified (non-blinded) by colocalizing BrdU⁺ nuclei with GFP, which was entirely included in the section throughout the z-axis. The top and bottom of the sections were identified by the presence of the first or last nuclei that could be entirely visualized by the z-plane. Colocalization was determined by scanning serially through the z-axis of each cell with a depth of focus optical slice of 2.72 μm.

Imaging

Confocal images were acquired on a Zeiss LSM 510 (Zeiss, Jena, Gottingen, Germany) scanning confocal multicolor microscope with Ar 488-nm, He 543-nm and He 633-nm lasers (sp2, version 3.2, aim interface), as previously described (Gilley et al., 2011). A Zeiss Neofluar ×40/1.3 lens and ×63/1.3 oil DIC lens were used. A pinhole aperture of 3.46 Airy units was used.

Fluorescence-activated cell sorting

To determine the extent of GltI and Glast upregulation after injury, cells from the ipsilateral and contralateral dentate gyrus of injured transgenic mice were stained with mouse anti-polysialic acid neural cell adhesion molecule (PSA-NCAM) (1 : 200; Millipore, Billerica, MA, USA; catalog no. MAB5324) for 20 min on ice. After unbound primary antibody had been washed away, cells were stained with donkey anti-mouse Alexa Fluor 647 (1 : 200; Molecular Probes, Eugene, OR, USA; catalog no. A31571) secondary antibody for an additional 20 min on ice. Cells were then washed, and four different populations were sorted on a MoFlo (Beckman Coulter, Brea, CA, USA; catalog no. ML99030): ipsilateral GFP⁺/PSA-NCAM⁺ and GFP⁺/PSA-NCAM⁻, and contralateral GFP⁺/PSA-NCAM⁺ and GFP⁺/PSA-NCAM⁻. RNA was then extracted and normalized across samples before being analyzed by QPCR.

Glutamate transporter misexpression influences neurosphere formation and progenitor proliferation in vitro

To determine whether GltI and Glast contribute to the regulation of hippocampal neurogenesis, we performed gain-of-function and loss-of-function experiments on neurosphere cultures grown in the absence of glutamate. First, we performed neurosphere assays on cells previously transfected with GltI and Glast siRNA or overexpression plasmids. Seven days after transfection, neurospheres were assayed for misexpression by QPCR (n = 4). Relative mRNA levels were normalized to GltI and Glast expression in cells transfected with scrambled siRNA as a negative control (Fig. 1A and B). Relative levels of GltI expression were 0.9 in cells transfected with DsRed2, 0.3 in cells lacking GltI, and 10.3 in cells overexpressing GltI (Fig. 1A). Glast expression levels after transfection were 1.2 for DsRed2, 0.3 for the knockdown, and 3.7 for overexpression (Fig. 1B). These data suggest that both glutamate transporters are misexpressed appropriately in culture as compared with controls (cells transfected with scrambled siRNA or DsRed2).

After overexpression and knockdown levels were confirmed, the remaining neurospheres were dissociated and replated in semi-solid media containing neural stem cell media (lacking glutamate) and 1.6% methylcellulose at a density of 20 cells/μL. After 7 days in culture, neurospheres over 50 μm were quantified (Fig. 1C). The average numbers of spheres per well were 64 for DsRed2, 57 for scrambled siRNA, 99 for GltI knockdown, 68 for GltI overexpression, 108 for Glast knockdown, and 69 for Glast overexpression. Analysis by one-way anova suggested that these differences were significant (P < 0.0001, F = 13.4, n = 4). Representative neurospheres are shown for each condition, and suggest that sphere size is dependent on the expression level of each glutamate transporter (Fig. 1D–I). Moreover, these results show that knockdown of glutamate transporters results in increased neurosphere formation in vitro.

To confirm this phenotype, we pulsed transfected neurospheres with BrdU and analyzed its incorporation with flow cytometry. In addition, the fidelity of misexpression of GltI and Glast was analyzed (Fig. 1J and K) (n = 4). The average relative expression of both GltI and Glast in cells transfected with DsRed2 was 1.1. Knockdown with either Glast or GltI siRNA resulted in relative mRNA levels of 0.3 and 0.6, respectively, suggesting that both genes were knocked down by at least 44%. Cells transfected with overexpression plasmids demonstrated increased GltI (11.0) and Glast (3.5) expression. From this, we can conclude that glutamate transporter knockdown and overexpression results in altered gene expression of GltI and Glast. These results are similar to those obtained previously, and suggest that we were able to consistently manipulate GltI and Glast expression levels in vitro.

Once misexpression was confirmed, the remaining neurospheres were pulsed with BrdU and stained with an anti-BrdU-APC antibody. We used flow cytometry to assess the effect of glutamate transporter misexpression on proliferation, and the results are reported as the percentage of BrdU⁺/PI⁻ cells (Fig. 1L). Baseline BrdU incorporation was 1.8% for DsRed2 and 2.0% for scrambled siRNA (both negative controls); these values were not significantly different from each other. Knockdown of GltI resulted in 2.9% BrdU incorporation, whereas its overexpression decreased the percentage of BrdU⁺/PI⁻ cells to 0.5% when compared with negative controls. Similarly, Glast siRNA increased the percentage of BrdU⁺ cells to 2.6%; however, its overexpression decreased BrdU incorporation to 0.2%. Statistical analysis with one-way anova indicated that these results were statistically different (P < 0.0001, F = 51.2, n = 4). Taken together, these results indicate that GltI and Glast negatively regulate neural progenitor proliferation in vitro.

Glutamate-induced neurosphere proliferation is dependent on calcium and mGluRs

Although previous studies have shown that glutamate stimulates neurosphere proliferation in a number of different cell types (Luk et al., 2003; Brazel et al., 2005; Suzuki et al., 2006), there are no studies that have specifically demonstrated glutamate’s effect on neurospheres derived exclusively from the dentate gyrus. To control for the presence of glutamate in most media, we prepared our neurospheres in medium lacking glutamate (Neurobasal A) before supplementing them with either 0 μm or 5 μm glutamate prior to final plating. In addition to plating neurospheres under wild-type conditions (Fig. 2A–D), we repeated the assay in the presence of 1 μm BAPTA-AM, a cell-permeable calcium chelator that neutralizes intracellular calcium (Tsien, 1980; Dieter et al., 1993) (Fig. 2E–H). Finally, to determine the role of mGluRs in proliferation, we also supplemented neurospheres with 10 μm LY341495, which non-selectively inhibits all mGluRs at this concentration (Fig. 2I–L) (Kingston et al., 1998). After 7 days in vitro, we quantified the number of neurospheres over 50 μm and their diameter as shown in Fig. 2. Under wild-type conditions, the numbers of spheres present were 27 and 28 in the presence and absence of exogenous glutamate, respectively (Fig. 2C). However, in media supplemented with glutamate, the average neurosphere size was 270 μm, as compared with 66 μm in media lacking glutamate (Fig. 2D). Although there was no difference in sphere number (unpaired t-test, P = 0.96, F = 9.0, n = 7), the difference in the size of the neurospheres was statistically significant when analyzed with an unpaired t-test (P < 0.0001, F = 24.9, n = 20). When we supplemented the media with BAPTA-AM or LY341495 in the presence and absence of glutamate, we saw a similar number of neurospheres. Among cells exposed to 1 μm BAPTA-AM, those treated with glutamate formed 44 neurospheres per well that averaged 66 μm in diameter (Fig. 2E and G). In the absence of glutamate, there were 45 spheres per well, and their average diameter was 63 μm (Fig. 2F and H). Similarly, cells treated with LY341495 and 5 μm glutamate formed 41 neurospheres, with an average diameter of 64 μm (Fig. 2I and K), whereas those grown in medium lacking glutamate formed 38 spheres, with an average diameter of 63 μm (Fig. 2J and L). According to statistical analysis with an unpaired t-test, cells treated with BAPTA-AM were not significantly different in neurosphere formation (P = 0.81, F = 1.2, n = 4) or neurosphere size (P = 0.49, F = 1.1, n = 20). Similarly, neither neurosphere number (unpaired t-test, P = 0.57, F = 3.4, n = 5) nor diameter (unpaired t-test, P = 0.72, F = 1.3, n = 20) was significantly different in wells treated with LY341495. These results suggest that intracellular calcium and mGluRs are both required for glutamate-induced neurosphere proliferation in vitro.

Glutamate transporter misexpression affects intracellular calcium levels

Although the exact mechanisms underlying neurogenesis remain unclear, several studies have suggested that the activation of certain ionotropic (ligand-gated ion channels) and/or mGluRs (G-protein-coupled) might be required (Choi et al., 1988; Luk et al., 2003; Deisseroth et al., 2004; Brazel et al., 2005; Nacher & McEwen, 2006; Nacher et al., 2007). The evidence suggests that activation of these receptors can also be associated with increases in intracellular calcium (Deisseroth et al., 2004; Brazel et al., 2005). We therefore chose to use calcium flux to assay for pro-proliferative pathways. To determine the effect of GltI and Glast misexpression on glutamate-mediated calcium flux, we incubated transfected neurospheres with a Fluo4-AM calcium indicator and assayed for changes in intracellular calcium using flow cytometry. Prior to stimulation with exogenous glutamate, baseline levels of intracellular calcium were recorded for 30 s. Cells were then stimulated with 5 μm glutamate, and we immediately began recording the increased fluorescence indicating a proportional increase in intracellular calcium. The results are presented as the percentage of cells that were Fluo4-AM⁺, and are normalized to the percentage of positive cells at baseline levels prior to stimulation with glutamate (Fig. 3C).

Before we conducted the experiment, we once again confirmed misexpression in vitro (Fig. 3A and B) (n = 4). The relative expression levels of GltI and Glast were 1.1 and 0.9, respectively, in cells transfected with DsRed2. Cells transfected with either GltI or Glast siRNA had expression levels of 0.2 and 0.5, respectively. Overexpression experiments demonstrated relative mRNA levels of 11.5 for GltI and 3.5 for Glast. Once again, these results show that GltI and Glast mRNA can be reliably manipulated in vitro. About 4% of cells above baseline levels were Fluo4-AM⁺ among cells transfected with either DsRed2 or scrambled siRNA prior to their stimulation with glutamate. After glutamate was added to the cells, the ratio of induced/baseline cells was 1.5 for both DsRed2 and scrambled siRNA. Whereas knockdown of GltI resulted in a ratio of 2.0, its overexpression decreased the percentage of positive cells to 1.2. Glast overexpression also demonstrated a ratio of 1.2, whereas its knockdown resulted in a ratio of 2.0. Analysis with one-way anova indicated that these results were significantly different from each other (P < 0.0001, F = 24.3, n = 4). These results suggest that glutamate transporter knockdown leads to increased intracellular calcium, presumably through activation of glutamate receptors. This therefore allows the cells to proliferate more readily than when glutamate transporters are present at baseline or increased levels.

We next wanted to determine whether this proliferative phenotype was calcium-dependent. To test this, we repeated the experiment in the presence of BAPTA-AM, a cell-permeable calcium chelator that neutralizes intracellular calcium (Tsien, 1980; Dieter et al., 1993) (Fig. 3F). Prior to incubation with BAPTA-AM, misexpression was confirmed in transfected cells (n = 4). GltI relative mRNA levels were 1.0 for scrambled siRNA, 1.0 for DsRed2, 0.4 for knockdown, and 11.1 for overexpression. Glast relative mRNA levels were 1.0 for scrambled siRNA, 1.2 for DsRed2, 0.5 for knockdown, and 4.2 for overexpression (Fig. 3D and E). After misexpression was confirmed, we incubated transfected cells with BAPTA-AM prior to staining with Fluo4-AM. This allowed us to determine whether intracellular calcium is required for the induction shown in Fig. 3C. Once again, we recorded baseline levels of calcium before we stimulated the cells with glutamate, at which point we calculated the ratio of induced/baseline cells. The results are shown in Fig. 3F. In cells transfected with scrambled siRNA, the ratio was 1.0, and cells transfected with DsRed2 also had a ratio of 1.0. Cells knocked down with either GltI or Glast siRNA had ratios of 0.9 and 1.0, respectively. Finally, cells overexpressing GltI and Glast both had ratios of 1.0. One-way anova suggested that these results were not statistically significantly different from each other (P = 0.95, F = 0.21, n = 4). This suggests that intracellular calcium is required for the phenotype observed in Fig. 3C.

To confirm that mGluRs are required for the calcium flux shown in Fig. 3C, we repeated the experiment in the presence of 10 μm LY341495 to inhibit all mGluR subtypes. Prior to experimentation, cells to be treated with LY341495 were analyzed for misexpression (n = 4). Relative GltI expression levels for scrambled siRNA, DsRed2, knockdown and overexpression were 1.0, 1.0, 0.3, and 12.2, respectively, whereas mRNA levels for Glast were 1.0, 1.0, 0.5, and 4.3, respectively (Fig. 3G and H). After misexpression was confirmed, cells previously treated with LY341495 were analyzed for glutamate-induced calcium flux. The results obtained were similar to those shown in Fig. 3F. The ratios of induced/baseline calcium flow for scrambled siRNA, DsRed2, GltI siRNA, GltI overexpression, Glast siRNA and Glast overexpression were 1.1, 1.1, 1.0, 1.0, 1.2, and 1.4, respectively (Fig. 3I). Analysis by one-way anova showed that the differences were not statistically significant (P = 0.80, F = 0.46, n = 4). Therefore, inhibiting mGluRs stifles the phenotype shown in Fig. 3C. These results suggest that glutamate transporters regulate intracellular calcium-dependent proliferation of neural progenitors in an mGluR-dependent fashion.

Glutamate transporters are persistently upregulated after HI

We previously demonstrated that GltI and Glast are expressed on type I cells in vivo (Gilley et al., 2011). Moreover, injury-induced activation of type I cells and subsequent neurogenesis have been well documented (Kernie et al., 2001; Miles & Kernie, 2006, 2008; Shi et al., 2007; Koch et al., 2008; Yu et al., 2008; Blaiss et al., 2011). We therefore hypothesized that GltI and Glast would be expressed on activated progenitors after injury. Following unilateral HI in nestin–GFP-expressing transgenic mice, we performed immunostaining for GFP, GFAP and either GltI (Fig. 4) or Glast (Fig. 5) to determine their expression at various time points after injury (n = 6). In this transgenic line, GFP is a well-established marker of slowly dividing type I cells as well as of the more rapidly dividing type IIa population (Miles & Kernie, 2006, 2008; Yu et al., 2008). GFAP in conjunction with GFP marks the type I cells only, so expression of both can be used to distinguish type I from type IIa progenitors (Shi et al., 2007; Miles & Kernie, 2008; Yu et al., 2008). On the uninjured (contralateral) side of the brain, glutamate transporter expression remained at baseline levels at all indicated time points (4, 5). However, GltI and Glast were both upregulated on type I cells on the ipsilateral (injured) side of the brain for up to 45 days after injury (4, 5).

Cell-specific elevation of glutamate transporter expression in injured progenitor cells

There are two possible explanations for the results shown in 4, 5. Either increased numbers of GFP⁺ cells on the injured side of the brain result in higher glutamate transporter expression, or the expression of GltI or Glast is elevated at the cellular level. To distinguish between these two possibilities, we used fluorescence-activated cell sorting to distinguish early and late progenitors from transgenic mice that were injured 3 days previously. We used GFP and PSA-NCAM expression to distinguish early (type I and IIa: GFP⁺/PSA-NCAM⁻) and late (type III: GFP⁻/PSA-NCAM⁺) progenitors (Chumley et al., 2007). We collected both populations from the ipsilateral (injured) and contralateral (uninjured) sides. We then isolated the RNA and normalized the concentrations across all samples before synthesizing cDNA and analyzing relative expression levels with QPCR. The expression of a particular gene was normalized to its own expression on the uninjured side of the brain. The results are shown in Fig. 6.

We observed that the GltI mRNA level was 2.5-fold higher in early progenitor cells from the injured side of the brain than in those from the uninjured side (Fig. 6A) (two-way anova, P = 0.002, F = 25.2, n = 4). However, the relative expression level of GltI was unchanged in late progenitors from injured and uninjured dentate gyrus. The latter result is not surprising, considering that neither GltI nor Glast colocalizes with markers indicative of late type III progenitors (Gilley et al., 2011). This suggests that GltI is upregulated in early progenitors at the cellular level as a result of HI. Similar to what was found for GltI, Glast was also upregulated on early progenitors isolated from the injured side of the brain, but its expression was not significantly different in later progenitors (Fig. 6B) (two-way anova, P = 0.01, F = 14.01, n = 4). Together, these findings indicate that elevated expression of glutamate transporters on the injured side of the brain (4, 5) result from changes in gene expression at the single-cell level.

Recurrent injury impairs progenitor proliferation

As we observed long-lasting changes in the expression of GltI and Glast at the cellular level following injury, we next wanted to determine whether injured progenitors maintained their proliferative potential in the context of gene expression changes. One month after HI, a CCI model was used to induce TBI in a well-established model of increasing progenitor proliferation (Kernie et al., 2001; Shi et al., 2007; Yu et al., 2008; Blaiss et al., 2011). To analyze the subsequent effects on progenitor proliferation, we injected both groups with BrdU, performed immunostaining, and quantified the number of GFP/BrdU-double-labeled cells. In animals exposed to sham HI (ligation with no hypoxia), there were no observed differences in the percentage of proliferative cells following TBI (Fig. 7E) (unpaired t-test, P = 0.74, F = 2.2, n = 5). However, Fig. 7J shows there was a significant difference between colocalized cells within the contralateral side (15.1%) and ipsilateral (7.8%) side in mice with both HI and CCI injury (unpaired t-test, P = 0.003, F = 2.5, n = 4). These results suggest that repetitive brain injuries lead to a decreased capacity for injury-induced neurogenesis that may, in part, reflect changes in glutamate transporter expression.