|Home | About | Journals | Submit | Contact Us | Français|
Stroke is a leading cause of permanent disability and death. It is well accepted that the principal mammalian estrogen (E2), 17-β estradiol, provides robust neuroprotection in a variety of brain injury models in animals of both sexes. E2 enhances neurogenesis after stroke in the subventricular zone; however, it is unknown if these cells survive long-term or enhance functional recovery. In this study, we examined stroke-induced neurogenesis in male, gonadally intact female, and ovariectomized female mice 2 and 6 weeks after stroke. Treatment with 17-β estradiol increased 5-bromo-2′-deoxyuridine-labeled cells at both time points in both the dentate gyrus and subventricular zone; the majority were colabeled with doublecortin at 2 weeks and with NeuN at 6 weeks. Stroke-induced neurogenesis was reduced in estrogen receptor knockout mice, as well as in mice lacking the gene for aromatase, which converts testosterone into E2. Improved behavioral deficits were seen in E2-treated mice, suggesting that E2-induced increases in poststroke neurogenesis contribute to poststroke recovery.
It is well accepted that the principal mammalian estrogen, 17-β estradiol (E2), provides robust neuroprotection in a variety of acute experimental brain injury models (McCullough and Hurn, 2003). In contrast, large clinical trials have shown no benefit of chronic estrogen replacement on stroke prevention (Anderson et al, 2004). Differences in the duration of treatment, timing of administration, and use of prevention regimens (chronic exposure) versus neuroprotection (acute dosing) may explain some of the differences in the experimental and clinical literature (reviewed in Sherwin (2009)). Regardless, as stroke is now the leading cause of long-term disability in the United States, and incidence continues to rise (Rosamond et al, 2008), increased emphasis should be placed on investigating neurorestorative agents for use once a stroke has occurred.
It is now well documented that ongoing neurogenesis occurs in specific regions of the adult mammalian brain (Altman and Das, 1965), and that this can be induced by a variety of pathological stimuli, including stroke (reviewed in Lindvall and Kokaia (2010)). A major challenge is to determine whether enhanced stroke-induced neurogenesis can lead to functional improvement and how to potentially amplify this response to improve behavioral outcomes and hasten recovery. E2 is a potent modulator of physiological neurogenesis, especially in the hippocampus. Rodent models have also provided definitive evidence for a role of E2 in ameliorating acute neuronal damage (McCullough and Hurn, 2003), but much less is known regarding the potential of E2 to enhance neurogenesis after an injury. In the central nervous system, E2 can be directly synthesized by hippocampal neurons in adult rats (Hojo et al, 2004) by the actions of P450 aromatase, which catalyzes the final step in E2 biosynthesis from C19 steroids such as testosterone (Garcia-Segura, 2008). Aromatase is rapidly upregulated after injury, and via the local production of E2, may be an important contributor to its proliferative and reparative effects (Garcia-Segura, 2008; Roselli et al, 2009). Female mice with a targeted deletion of aromatase (ARKO) have significantly more tissue damage after middle cerebral artery occlusion (MCAO) than ovx wild-type (WT) mice, suggesting that there is also a contribution from nongonadal sources of E2 to neuroprotection (McCullough et al, 2003).
Several receptors have been described that are specifically activated by E2 binding: the classical nuclear estrogen receptor (ER) subtypes, ERα (α) and ERβ (β) in addition to several membrane-associated receptors, including GPR30 (Heldring et al, 2007). Both embryonic and adult neural stem cells express ERα and ERβ, and treatment with E2 increased neural stem cell proliferation in vitro, an effect that was partially ameliorated by administration of the nonsubtype-specific ER antagonist ICI-182,780 (Brannvall et al, 2002). The ERα- and ERβ-positive cells have been identified in the ventricular wall of both embryonic and adult rat brain and mRNA for both ERα and ERβ are present in the dentate gyrus (DG) (Merchenthaler et al, 2004). Both ERα and ERβ appear to contribute to early cell proliferation (e.g., 72hours) after stroke in mice (Suzuki et al, 2007), but whether these cells survive past 3 days or if there were corresponding changes in functional recovery is not known. As the peak of stroke-induced neurogenesis occurs well after this period (Kuge et al, 2009), we aimed to (1) evaluate the effect of E2 on stroke-induced neurogenesis in the male and female brain, (2) assess maturation and long-term survival of newborn cells, (3) determine whether E2-induced neurogenesis led to enhanced behavioral recovery, and finally (4) determine whether classical ERs or aromatase has a role in mechanisms by which E2 induces neurogenesis.
The present study was conducted in accordance with the National Institute of Health guidelines for the care and use of animals in research and under protocols approved by the Center for Lab Animal Care at the University of Connecticut Health Center. In initial experiments comparing male and female WT mice, we used five groups of 3-month-old C57Bl/6 mice (Jackson Laboratory, Bar Harbor, ME, USA; n=8 to 9/stroke and 5/sham/group): (1) gonadally intact females (IF), (2) ovariectomized females supplemented with 17-β estradiol (Ovx+E2), (3) ovariectomized females supplemented with oil placebo (Ovx+oil), (4) males supplemented with E2 (ME2), or (5) males supplemented with oil placebo (M+oil) pellets. There were two cohorts within each of these five treatment groups, one that was evaluated at 2 weeks after stroke, and the other that was evaluated 6 weeks after stroke (see Figures 1A and 1B). Animals were housed in cages with a 12-hour light/dark schedule, and provided with food and water ad libitum. In the latter experiments using female ERα−/− (ERKO and ERβ−/− BERKO) and aromatase (ARKO) knockout animals, animals were left gonadally intact, and all animals were evaluated at one time point, 2 weeks after MCAO (Figure 1C). Female ERKO and BERKO mice, back crossed onto a C57BL/6J genetic background (Dubal et al, 2001; reviewed in Hewitt et al (2005)) and aromatase-deficient (ArKO on SV129) (McCullough et al, 2003) mice were compared directly to their respective gonadally intact age-matched female WT littermates.
Female WT mice were ovariectomized and implanted subcutaneously with a pellet containing either 180μg/mL of 17-β estradiol (E2) in sesame oil (0.062-in. ID/0.125-in. OD) or sesame oil vehicle 10 days before MCAO as previously described (McCullough et al, 2003, 2005). Male animals were also given exogenous E2 or oil pellets to assess for hormonal effects on neurogenesis independent of biological sex. Hormone implants were replaced at 3 weeks in all mice of the 6-week survival cohort to ensure continuous physiological hormone levels. All hormonal manipulations were performed by an independent masked investigator (MY). In the second experiment examining ERKO, BERKO, and ARKO mice and their WT littermates, all mice were left gonadally intact. Plasma E2 levels (enzyme-linked immunosorbent assay) and uterine weights were assessed at kill, as described (McCullough et al, 2003, 2005).
Focal cerebral ischemia was induced by 90minutes of reversible MCAO as previously described (McCullough et al, 2003, 2005) Laser Doppler Flowmetry was evaluated throughout MCAO and early reperfusion in all cohorts as described previously (McCullough et al, 2003, 2005). Previous studies have shown no physiological differences between groups (Dubal et al, 2001; McCullough et al, 2003, 2005). At either 14 or 42 days of reperfusion, the brain was harvested for histological examination.
At days 2, 7, 12, 21, 28, 35, and 42 after stroke, animal behavior was scored using the neurologic deficits score as follows: 0, no deficit; 1, forelimb weakness and torso turning to the ipsilateral (IL) side when held by tail; 2, circling to affected side; 3, unable to bear weight on affected side; and 4, no spontaneous locomotor activity as previously described (McCullough et al, 2005). All behavioral tests were performed by a masked investigator at the beginning of the light cycle. Animals were weighed daily for the first week after stroke then weekly thereafter. All animals were pretested to ensure absence of turning and paw preference before MCAO.
The cylinder test (9 × 15cm2) was used to assess forelimb use and rotational asymmetry as previously described (Li et al, 2004). A total of 20 movements were recorded during the 10-minute test. The final score=(nonimpaired forelimb movement–impaired forelimb movement)/(nonimpaired forelimb movement+impaired forelimb movement+both movement).
The mouse was placed between two cardboard pieces (30 × 20 × 1cm3), which were gradually moved closer together to encourage the mouse into a 30° corner with a small opening as described in Li et al (2004). Once deep in the corner the vibrissae were stimulated, the mouse reared forward and upward, and then turned to face the open end. Only turns involving full rearing along either board were recorded. Right turn percentage was calculated as in Li et al (2004).
In all experiments, mice were injected three times intraperitoneally with 50mg/kg of the thymidine analog 5-bromo-2′-deoxyuridine (BrdU; Sigma-Aldrich, St Louis, MO, USA) as a marker for dividing cells at 8-hour intervals for 24hours (three total injections), 7 days after stroke. The WT mice were randomly assigned to treatment group (see Figure 1).
Animals were killed under anesthesia at either 2 or 6 weeks after stroke (Figure 1) and perfused transcardially with cold PBS followed by 4% paraformaldehyde; the brain was postfixed for 6hours and placed in 30% sucrose for ~48hours. Tissue was cut into 25μm free-floating sections on a Cryotome (Thermo Electron, Waltham, MA, USA) and placed in antifreeze buffer before staining.
Frozen sections were mounted onto slides in 1 × PBS. Every eighth slice was stained for evaluation of ischemic cell damage. Slides were fixed in a 1:1 concentration of chloroform and 100% ethanol for 30minutes followed by washes in 95%, 70%, 50% ethanol, and distilled water for 3minutes before staining with cresyl violet. Sections were digitized by a CCD camera and analyzed with photoimaging software (Jandel Scientific, San Rafael, CA, USA). Cerebral atrophy was used as an indirect measure of cell loss as in Li et al (2004). Neuronal death was measured by determining the volume of tissue atrophy by measuring both hemispheres and lateral ventricles and transformed to mm3. Percent atrophy was computed by dividing the ischemic (right) hemisphere from the intact (left) hemisphere, then multiplying by 100 as previously described (Li et al, 2004).
For histochemical staining, sections were initially washed in 1 × PBS followed by a distilled water rinse. DNA was denatured by incubating in 2N HCl for 45minutes at 37°C followed by neutralization with immersion in 0.1mol/L borate buffer. After rinsing with washing buffer (10 × PBS+0.5% Tween 20 diluted 1:10), the sections were blocked in 2% Normal Donkey Serum and incubated in the appropriate primary antibodies, BrdU (Accurate Chemical and Scientific Corp., Westbury, NY, USA; 1:50 anti-rat) as a marker of newborn cells, NeuN (Chemicon International, Temecula, CA, USA; 1:100 anti-mouse) as a marker for mature neurons, doublecortin (DCX; Santa Cruz Biotechnology, Santa Cruz, CA, USA; 1:100 anti-goat) for immature neurons, or glial fibrillary acidic protein for astrocytes (GFAP; DAKO Corporation, Via Real, Carpinteria, CA, USA; 1:1,200 anti-rabbit) for 72hours at 4°C as described (Jin et al, 2001; Liu et al, 1998). The sections were rinsed and incubated with secondary antibody (1:200 Texas red donkey anti rat IgG; Jackson Immunoresearch Laboratories, West Grove, PA, USA and 1:1,000 Alexa fluor 488: Molecular Probes, Eugene, OR, USA) for 2hours, rinsed and mounted. DAB staining (VECTASTAIN ABC kit, Vector Labs, Burlingame, CA, USA) was performed on select slices as detailed by the manufacturer to identify and confirm BrdU staining in cresyl violet sections. No BrdU staining was seen in control animals that received no BrdU injections.
Confocal images were obtained using an Axiovert 200mol/L microscope (Carl Zeiss, Thornwood, NY, USA). All images were acquired in dual-scan mode with a z-scan interval of 1μm (~25 scans/stack). Stacks were converted to eight-bit images, and the stacks for each color were combined, resulting in one RGB stack. The SVZ (1.18 to 0.26 from Bregma) and DG (−1.56 to −2.20 from Bregma) were evaluated in both the IL (stroke) and contralateral (CL) (nonstroke) hemisphere. Individual single-labeled cells were examined in each scan of the RGB color-combined stack and counted within the dorsal SVZ, ventral SVZ, and DG as indicated in Figure 1A, with Metamorph image analysis software (version 6.0; Universal Imaging Corporation, Downingtown, PA, USA) and confirmed by formal cell counting in a predetermined 20 × field by an investigator masked to the treatment group. Each brain had five contiguous sections analyzed. A mean cell count was obtained. Colocalization was assessed in the same area, and verified by analyzing each scan of the RGB color-combined stack for individual cells. Each brain was analyzed by two independent masked investigators (MS and LF), and cell counts were all within a 7% range variation. Images were formatted in Adobe Photoshop 7.0 (Adobe, San Jose, CA, USA).
All data are expressed as means±s.e.m. One-way ANOVA was used to analyze the neurogenesis cell count data with a post hoc Bonferroni correction. All behavioral data were analyzed by ANOVA (all % data) with repeated measures by investigators masked to the treatment group. The neurologic deficit score (a nonparametric value) was analyzed by the Mann–Whitney U test. Bonferroni post hoc analyses were conducted to correct for multiple comparisons where appropriate. P<0.05 was considered statistically significant. Correlations were assessed with Pearson coefficients (two-tailed, SPSS, Chicago, IL, USA) comparing individual infarct size with the amount of neurogenesis in the dSVZ or DG.
Stroke induced a significant increase in BrdU+/DCX+-labeled cells in the dorsal SVZ in both male and ovx female mice treated with oil (Figure 2A). Interestingly, oil-treated females had significantly higher numbers of BrdU/DCX+ cells in both the IL (#female oil; 76.65±9.8 versus male oil; 47.5±6.7, bar 1 versus bar 7) and CL SVZ (##female oil 48.6±7.1 versus male oil; 32.2±5.8, bar 2 versus bar 8, n=8) after stroke compared with oil-treated males after MCAO (see Figure 2A; #P<0.05). Treatment with physiological levels of E2 abolished this sex difference. E2 had a potent proliferative effect in the ischemic brain that was significantly greater than that seen with stroke alone, and this was evident in both male and ovx female mice (Figure 2A, n=8, *P<0.01). No proliferative effect was seen in E2-treated sham mice. We also assessed a cohort of IF mice. Female mice with intact gonadal function had equivalent levels of stroke-induced neurogenesis as female ovx mice supplemented with E2 (data not shown). Stroke-induced enhancement of neurogenesis in IF was seen in both the CL and IL SVZ after MCAO; levels were significantly higher than in oil-treated ovx females and oil-treated males (as shown in Figure 2A) (intact females; BrdU+/DCX+ in IL=123±11 versus ovx 76.65±9.8 and intact CL=59±4.3 versus oil 48.6±7.1, P<0.01 versus ovx females, n=8, 9) but were not significantly different than E2-supplemented females or E2-treated males, suggesting that the presence of either endogenous or exogenous E2 is sufficient to enhance neurogenesis after MCAO. The majority of BrdU+ cells were colabeled with DCX (Figure 2B). No colabeling was seen with GFAP (not shown). No significant changes were seen in neurogenesis in the ventral SVZ (vSVZ) in any treatment group, regardless of hormone or stroke exposure (data not shown).
Stroke significantly increased the number of BrdU+/DCX+ cells in the IL DG compared with sham-treated mice (#P<0.001; Figure 2C, stroke effect) 2 weeks after MCAO. Absolute cell numbers were lower than that seen in the SVZ. Similar to what was observed in the dSVZ, E2 treatment significantly enhanced stroke-induced neurogenesis in ovx females as well as in males (*P>0.01; Figure 3, E2 effect). Stroke-induced neurogenesis was again higher in ovary intact randomly cycling females (IF) compared with oil-treated ovx females and no different than that seen in E2-supplemented ovx females (intact; 36.8±4.2 versus ovx females 23±3.9, data not shown). In agreement with previous work (Suzuki et al, 2007) and as in the dSVZ, supplementation of physiological levels of E2 did not affect the basal proliferation of neurons or astrocytes in sham-operated mice in the DG (Figures 2A and 2C), suggesting the proliferative effect of E2 is limited to ischemic brain.
E2-treated females had smaller total hemispheric infarction at 2 weeks, as measured by hemispheric atrophy, compared with oil-treated females (E2; 12.7%±2.9% versus oil; 21.9%±2.7%, P0.05), an effect that was also seen in E2-treated male mice (E2; 14.5%±3.6% versus oil; 23.1%±3.1%, P<0.05, n=9/group) and IF mice (intact; 11.8%±2.1% versus ovx oil 21.9%±2.7%, P0.05). Correlation analyses (Pearson coefficients, two-tailed, SPSS) comparing individual infarct size and the amount of neurogenesis in either the dSVZ or DG were performed. There was no significant correlation between infarct size and dSVZ neurogenesis (although a trend was seen; P=0.09) and no significant correlation between infarct sizes and the amount of DG neurogenesis (P>0.05). Therefore, the amount of neurogenesis was independent of the degree of ischemic injury.
To determine whether the early proliferative effects of E2 on stroke-induced neurogenesis seen at 2 weeks were sustained, we evaluated the number of cells colabeled with BrdU and the mature neuronal maker NeuN 6 weeks after stroke. Stroke enhanced the number of BrdU+/NeuN+ cells in the DG and E2 treatment further increased this neurogenic response (Figure 2D). No significant differences were seen in E2-treated versus oil-treated sham animals, confirming that the robust proliferation induced by hormone exposure is limited to the response to a pathological state such as stroke, even after chronic exposure. We found BrdU staining exclusively in cells that labeled with NeuN+, but not in GFAP+ cells, suggesting a selective effect of E2 on neurogenesis rather than gliogenesis, at least with the dosing administration used here (Figure 2E).
The number of BrdU+ cells in the SVZ could not be reliably quantified 6 weeks after stroke. There were scattered DCX+ cells in the SVZ, but none of these cells colabeled with the BrdU given 5 weeks previously. Very few mature neurons were present in our predefined area of the SVZ. Rare scattered BrdU+/NeuN+ cells were seen in the IL cortical areas, but their presence was too variable to reliably quantify. E2 treatment reduced infarct damage at 6 weeks as measured by hemispheric atrophy in both female (E2; 11.7%±3.8% versus oil; 19.9%±2.6%, P0.05), and male mice (E2; 12.5%±3.1% versus oil; 21.2%±2.1%, P<0.05, n=9/group).
Examination of mice in the cylinder test revealed that although E2-treated and oil-treated female ovx MCAO mice demonstrated similar overall recovery by 6 weeks (day 42), E2-treated ovx females recovered significantly faster than did oil-treated ovx females (Figure 3A). Even at day 35, E2-treated ovx females had significantly better performance on the cylinder test compared with oil-treated ovx females (Figure 3A). Similar to previous reports (Li et al, 2004), there was no sex difference in this effect; the slope of recovery was similar in E2-treated males and IF (data not shown). All groups, regardless of E2 treatment status, recovered to sham levels by 6 weeks, although recovery was significantly faster in the groups with enhanced neurogenesis. Consistent with previous studies (Li et al, 2004), behavioral recovery was significantly slower and was incomplete, even at 6 weeks, on the corner test (Figure 3B). This was seen in all cohorts of MCAO mice compared with their respective sham groups (only male and ovx female±E2 shown for clarity), regardless of sex or E2 treatment. However, E2 treatment significantly enhanced recovery on this task in both ovx females and males (Figure 3B) as compared with oil-treated male and ovx females (*P<0.05 for E2 versus oil, and #P<0.01 for sham versus oil, n=8/9 MCAO, 5 sham). The IF also improved faster than oil-treated ovx females (data not shown), but not significantly faster or more completely than ovx E2-treated females. This suggests that E2's enhancement of stroke-induced neurogenesis contributes to more rapid and complete behavioral recovery.
In an attempt to investigate the mechanism by which E2 enhances stroke-induced neurogenesis, we examined brain-derived neurotrophic factor (BDNF) levels 72hours after stroke in oil-treated males, E2-treated males, ovx oil-treated females, and ovx females treated with E2. E2 treatment significantly enhanced BDNF levels in sham mice (P<0.05, n=6/group stroke versus 4/group sham). Stroke itself was a potent stimulus for BDNF, and although there was a trend for an additive effect of MCAO and E2, this did not reach statistical significance (Table 1). Stroke-induced increases in BDNF were significantly more robust in females compared with males, independent of E2 treatment.
To determine the contribution of the ER to poststroke neurogenesis, we examined IF mice with selective deletion of either ERα (ERKO) or ERβ (BERKO) and compared them to their respective WT littermates BrdU was given at 7 days to capture the peak of poststroke neurogenesis and BrdU+/DCX+ cells were quantified 2 weeks after MCAO. Interestingly, there was a significant reduction in basal neurogenesis in the dSVZ of ERKO female mice (see Figure 4A, #P<0.05, sham) compared with both WT and BERKO sham mice. After stroke, only WT females demonstrated an increase in dSVZ neurogenesis with minimal response in either the IL (*P<0.001) or CL (**P<0.01) dSVZ in both ERKO and BERKO females (Figure 4A; individual strain-specific WT littermates collapsed for clarity, as no significant strain differences in WT were seen), as noted in a representative IHC (Figure 4C) in ERKO mice (BERKO not shown).
There were no significant differences in basal neurogenesis in the DG of WT, ERKO, or BERKO mice (Figure 4B; IHC Figure 4C, g–l). However, as was seen in the dSVZ, there was a significant reduction in stroke-induced neurogenesis in both ERKO and BERKO female mice as seen by the lack of BrdU+/DCX+ staining compared with WT in both the IL (*P<0.001) and CL hemisphere (**P<0.01, n=8/group 5/group in sham WT versus ERKO and BERKO; WT collapsed over one group, as no differences were present between WT littermate strains, see Figure 4C). Therefore, both isoforms of the ER must be present for stroke-induced neurogenesis to occur. Infarct size, as measured by CV, was significantly higher in ERKO and BERKO females compared with WT. Infarct was higher in ERKO mice compared with the BERKO females (WT 11.3%±3.2%, n=15; ERKO 28.9%±4.2% BERKO 19.9%±4.2%, P<0.05, n=8). E2 levels were elevated in ERKO mice compared with WT (88±9.7pg/mL versus 31.4±5.8pg/mL, P<0.01), but were not different in BERKO mice (36±7.9 versus 33.3±6.4, P=n.s).
Gonadally intact female ARKO and WT mice were subjected to MCAO, and the brains were examined 2 weeks later. ARKO females had a significant reduction in both ipsilateral (*P<0.001, n=7/group stroke versus 4/group sham) and CL (**P<0.01 n=7/group stroke versus 4/group sham) dSVZ compared with WT females (Figure 5A). Similar results were found in the DG, with ARKO mice having significantly lower stroke-induced NG in both the IL (*P<0.01) and CL (**P<0.001) DG (Figure 5B) compared with WT female littermates. ARKO mice had the lowest levels of stroke-induced neurogenesis of all three knockout strains (P<0.05, n=7 to 8/group stroke and 4 to 5/group sham) and significantly lower levels of neurogenesis compared with WT ovx mice (P<0.01, n=8/group; as in Figure 2). DAB staining showed a striking reduction in BrdU+ cells in both the SVZ and the DG of ARKO mice compared with their WT littermates (Figure 5C; brown). Fluorescent microscopy confirmed the reduction in neurogenesis with a dramatic reduction in BrdU+/DCX+ cells in both the SVZ and DG (Figure 5D). ARKO females also had significantly more severe infarct damage than IF WT littermates (WT 13.2%±2.9% ArKO 39.7±4.3; P<0.01, n=7 to 8/group stroke) as well as significantly larger infarcts than that of ERKO (27.9%±4.2%) and BERKO (19.9%±4.2% P<0.001 females, n=7 to 8/group stroke). There was no significant correlation between infarct size and stroke-induced neurogenesis.
To assess behavioral deficits, the corner test was performed throughout the 2 weeks of stroke survival in ERKO, BERKO, and ARKO mice and their corresponding littermates. ARKO, ERKO, and BERKO mice all showed minimal to no recovery over the 14-day testing period, whereas WT female littermates (collapsed over three WT littermate strains for clarity) demonstrated significant behavioral improvement by day 14 of testing although even WT females did not return to sham levels (Figure 6, P<0.05) (data shown for ERKO mice only as there was an identical lack of recovery in both the ARKO and BERKO groups). Similar results were seen in the cylinder test (data not shown), with little recovery in any of the three KO strains.
These experiments show several important new findings. First, E2 increases stroke-induced neurogenesis. This is not a sex-specific effect as both male and ovx female mice responded to E2. The proliferative effect of E2 was only unmasked under pathological conditions, in this case, stroke, as no increase was seen in E2-treated sham mice. Although E2 robustly increased ischemic neurogenesis, stroke itself remained a potent stimulus, as oil-treated mice of both sexes had significantly increased neurogenesis compared with sham animals. This may be secondary to increased BDNF levels after stroke. Second, BrdU+-labeled cells were present as long as 6 weeks after stroke and acquired a mature neuronal phenotype as indicated by NeuN colabeling. Third, E2-mediated enhancement of neurogenesis in the ischemic brain occurred in both neurogenic niches in the adult brain, the SVZ and DG. Stroke-induced neurogenesis was ameliorated in ER-deficient mice, demonstrating the importance of ER signaling. Fourth, local production of E2 in the brain may contribute to injury-induced neurogenesis, as loss of aromatase led to even more striking deficits in neurogenesis and even greater impairments in behavioral recovery than ovariectomy in WT mice, an effect independent of infarct size. Finally, E2 enhanced behavioral recovery after stroke in both male and female WT mice. Recovery was also slower and less complete in female mice lacking ERα, ERβ, and aromatase, demonstrating an association between reductions in neurogenesis, ER signaling, and behavioral recovery. Given the striking lack of recovery in the knockouts, small reductions in NG may have large, unanticipated effects on behavioral recovery.
It is now well established that neurogenesis occurs throughout adulthood in the mammalian brain, primarily in two areas, the DG of the hippocampus and the SVZ (Altman and Das, 1965; Lindvall and Kokaia, 2010). Global and focal ischemic insults induce rapid proliferation within these regions in both animal models and in humans (Nakayama et al, 2010), and redirect neuroblast migration toward the injured cortex and striatum. These early neurons can form synapses with neighboring cells, although the contribution of these cells to neurologic recovery is not yet known (Hou et al, 2008). Most studies of postischemic neurogenesis have been performed in male animals, yet the principal mammalian estrogen, E2, has a major role in physiological neurogenesis within the hippocampus. Females in the high estrogen, proestrous state have higher levels of neurogenesis than males (Galea, 2008; Hojo et al, 2004), a characteristic that is reversed by ovariectomy and in part by treatment with the nonselective ER antagonist ICI 182,780 (Perez-Martin et al, 2003). Both ERα and ERβ agonists increase cell proliferation in the DG (Mazzucco et al, 2006). The ERβ knockout mice show regional neuronal hypocellularity, especially in the cerebral cortex, suggesting that ERβ has a role in developmental and physiologic neurogenesis (Wang et al, 2003). Both ERα and β are expressed in embryonic and adult rat neural stem cells (Brannvall et al, 2002; Galea, 2008; Merchenthaler et al, 2004) derived from the adult SVZ; however, the role of ERs in physiological NG in vivo remains a subject of debate.
Our results convincingly show that both ERα and β have an important role in stroke-induced neurogenesis. Both receptor subtypes are required, as loss of either ER subtype dramatically abolished ischemia-induced neurogenesis in both the SVZ and DG. This is consistent with recent pharmacological studies demonstrating that administration of a selective ERα agonist, propyl-pyrazole triol, or ERβ agonist, diarylpropionitrile, increased cell proliferation in ovx adult female rats (Mazzucco et al, 2006). However, neither agonist was as effective as E2, consistent with a requirement for engaging both receptor subtypes. It has been well documented that ERα is upregulated after ischemic injury, and this isoform mediates many of the vascular and acute neuroprotective effects of E2 (Dubal et al, 2001). We now extend this concept to show that ERα likely has a major role in injury-induced neurogenesis, but that it does not likely operate alone. Recent work has shown epigenetic modification of the ER after murine stroke; injury induces demethylation of the ER promoter allowing for E2-induced gene transcription (Westberry et al, 2008). This, in conjunction with the known upregulation of aromatase in injured areas of brain (Garcia-Segura, 2008; Roselli et al, 2009), may lead to enhanced E2 signaling, development of both an early protective and later proproliferative environment, with enhancement of growth and survival genes such as BDNF, and increased functional recovery. However, as estrogen levels were not directly manipulated in our knockout studies, caution is warranted that the assumption that estradiol is working via a ligand-dependent effect on the ER has not been directly tested in these studies.
We examined ARKO mice in part to address if endogenous E2 production specifically mediates poststroke neurogenesis, rather than a secondary downstream effect triggered by nonligand activation of the ER. Aromatase is present in the embryonic neocortex, within radial glia, and is upregulated with injury (Garcia-Segura, 2008; Roselli et al, 2009). Treatment of rat hippocampal cultures with the aromatase inhibitor, letrozole, decreased the number of proliferative cells and increased apoptosis, an effect that was reversed by E2 (Fester et al, 2006). This suggests that local central nervous system E2 could contribute to stroke-induced neurogenesis and subsequent recovery. We have previously shown that female ARKO mice have significantly more damage after MCAO than WT ovx mice, demonstrating a role for local brain-synthesized E2 in acute neuroprotection (McCullough et al, 2003). As ARKO mice had lower levels of neurogenesis than any other strain, including ovx WT females, this suggests that extragonadal synthesis of E2 contributes to stroke-induced neurogenesis. However, these results must be interpreted with caution, as we compared ARKO mice with WT ovx females rather than ARKO littermates. Another caution to be noted is that the ARKO mice had larger infarcts than any other strain. However, the extent of injury did not significantly correlate with the amount of stroke-induced neurogenesis. Our data appear to be consistent with the study by Arvidsson et al (2001) that showed the amount of stroke-induced neurogenesis, at least in MCAO models, is independent of the amount of cell death. However, the possibility that the extensive ischemic damage in the KO strains led to glial scarring or destruction of the ‘neurogenic niche' remains a possibility.
The downstream mechanism by which E2 acts to enhance neurogenesis is not yet known, but likely involves numerous protective, antiapoptotic, and growth-promoting pathways. One candidate growth factor is BDNF, as several studies have shown that expression of BDNF mRNA and protein is modulated by E2 (Singh et al, 1995; Sohrabji et al, 1995). The BDNF promoter is known to have an estrogen response element-like sequence (Sohrabji et al, 1995). BDNF enhances the number and survival of new neurons derived from the SVZ both in vitro and in vivo (Pencea et al, 2001) under physiological and pathological (Ploughman et al, 2009) conditions. In this study, BDNF levels were elevated 72hours after stroke, and were significantly higher in intact females compared with males (Table 1). Surprisingly, exogenous E2 administration did not significantly increase BDNF levels over that seen with stroke alone in either ovx females or males. There are several possible explanations for this finding. We only examined one time point after stroke, and either shorter or longer assessments may have led to different results and unmasked effects of E2 on BDNF. Alternatively, it is possible that the dramatic stroke-induced BDNF masked any possible E2 effect, or that injury-induced upregulation of aromatase had already led to significant local production of E2 and a maximal BDNF response. Astrocytes and endothelial cells secrete BDNF and induce neurogenesis in adult neural stem cells (Song et al, 2002) and may provide a scaffold for migration of newborn cells to injured areas. As our experiments focused on stroke-induced neurogenesis in vivo, we chose to examine brain homogenates for BDNF; therefore, the specific cell type responsible for stroke-induced increases in BDNF is unclear.
In this study, both male and ovx females pretreated with E2 showed a significant increase in BrdU+/DCX+ colabeling 2 weeks after stroke in the SVZ and DG. By 6 weeks after stroke, most BrdU+ cells developed a mature phenotype, suggesting that these cells mature and survive. Our findings are consistent with those of Suzuki et al (2007) who documented that pretreatment with physiological doses of E2 enhanced poststroke neurogenesis in the dorsal SVZ of female mice. However, this paper evaluated acute changes (i.e., 96hours after injury), so functional recovery could not be assessed. As it is well documented that the peak of stroke-induced neurogenesis can occur as late as 3 weeks after injury in rodents (Kuge et al, 2009), we deliberately avoided early dosing BrdU to avoid uptake by dead or dying cells, or those in the process of repair. Our data show that E2 selectively enhances neurogenesis in the injured brain, as no increase was seen in E2-treated sham mice even after 6 weeks of exposure.
There are several technical caveats in this study. BrdU staining as a means to identify ‘newborn' cells has limitations; especially in models of brain injury such as stroke (Taupin, 2007). The BrdU incorporation is not invariably a marker of cell proliferation; it is only a marker of DNA synthesis. The BrdU incorporation ordinarily occurs in neurons after cell cycle reentry or after cellular repair, and does not detect the physiological status of the BrdU+ cell. Furthermore, attributing the behavioral benefits seen in E2-treated mice directly to enhanced neurogenesis would be premature. Recent work indicates that ischemia induces newborn neurons that display tetrodotoxin-sensitive Na(+) action potentials and spontaneous excitatory postsynaptic currents by 8 weeks of injury (Hou et al, 2008), suggesting that some newborn neurons are functional. However, it remains unclear if these neurons successfully integrate into neuronal networks and directly lead to improved recovery. The improvements in our behavioral studies may not be reflective of neurogenesis per se, presumably because the animals recovered too quickly for improvement to be secondary to ‘repopulation' of damaged circuitry. We speculate that the recovery is more likely secondary to enhanced growth factor signals or local support of damaged, but not irreversibly injured cells that remain in the penumbra. Our BDNF data support this hypothesis. E2 is robustly protective in most preclinical models (McCullough and Hurn, 2003), and we confirm here that stroke damage was significantly less in both male and female mice treated with E2; however, infarct size was independent of the amount of neurogenesis. Interestingly, all mice, regardless of their treatment status, sex, or genotype had similar behavioral deficits at day 2. This suggests that the behavioral tests used do not reflect initial infarct size but are more sensitive for assessing the rate of behavioral recovery, which was closely related to the amount of poststroke neurogenesis.
The major translational limitation of this work is that animals were treated with E2 before injury. A more relevant approach would be to evaluate the beneficial effects of E2 given poststroke on recovery to avoid possible confounding neuroprotective effects on behavior. Delayed E2 treatment (2 days after stroke) does not improve behavioral recovery in rats, suggesting that the treatment window may be short (Farr et al, 2006). Poststroke neurogenesis occurs in humans (Nakayama et al, 2010), but whether enhancing this endogenous response would improve recovery is unclear. Recent studies examining selective ablation of neuronal cell precursors indicate that short-term recovery (24hours) is linked to the neurogenic response, at least in animals (Jin et al, 2010). Another limitation of our study is that we did not use older mice. Aging animals exhibit stroke-induced neurogenesis, but this response is ‘muted' compared with young animals (Gao et al, 2009). As BDNF levels are known to drop with age (Jezierski and Sohrabji, 2001), this may have major implications if E2-induced neurogenesis is mediated in part by BDNF. Translation to clinical populations could also involve some risk, as the Women's Health Initiative has shown that stroke incidence increased in healthy postmenopausal women treated with E2 although there is evidence that the timing of replacement may be a critical factor (reviewed in Sherwin (2009)).
In conclusion, this work shows that E2 contributes to behavioral recovery via enhancement of neurogenesis, suggesting that further work to examine dose and duration of therapy are warranted. As over 80% of the neurons generated after stroke die within the first 2 weeks after their formation (Arvidsson et al, 2002), strategies to enhance survival and function of these cells are urgently needed. As the incidence of stroke continues to rise in our aging population, despite improved prevention and acute treatments, the mechanisms involved in recovery after stroke and enhancing function will be major goals in the future.
The authors declare no conflict of interest.
This work was supported by the AHA (FTF to LDM), NINDS (RO1 NSO50505 and NSO55215 to LDM), The Hazel Goddess Fund (to LDM), and NIH (NS 33668, NS49210, and NR03521 to PDH).