Generation of inducible GR gene inactivation in adult Nestin+ cells
In order to examine how the GR in adult hippocampal NPCs contributes to neurogenesis, we used a transgenic approach allowing for temporal control of genetic recombination restricted to NPCs (Figure 1). First, mice expressing a tamoxifen-inducible form of Cre recombinase (CreERT2) under nestin (marker for NPCs) transcriptional control84 were bred to a R26R-YFP reporter line to generate mice with a surrogate marker for recombination (Nestin-CreERT2/R26R-YFP). Next, we mated mice carrying GRloxP/loxP alleles239 with the Nestin-CreERT2/R26R-YFP line for at least two generations to produce a GR(loxP/loxP);R26YFP+/-;Tg(Nestin-CreERT2) (thereafter called GR(NestinCreERT2)) mouse line with a YFP reporter (Fig 1a). DNA from mouse tails was carefully extracted and analyzed by PCR to verify genotype (Fig 1b). Both GRloxP and R26R-YFP alleles are sensitive to Cre recombinase, which catalyzes site-specific recombination between the loxP sites. Thus, upon tamoxifen administration, GR was selectively inactivated and YFP expression was selectively induced only in cells expressing nestin and all their progeny thereafter (Fig 1a). Mutant mice (i.e. tamoxifen-treated GRNestinCreERT2 mice) are denominated GRNPCKO. Untreated GRNestinCreERT2 mice were used as control animals.
Verification of the inducible GR gene inactivation in adult Nestin+ cells
We used two approaches to verify the efficacy of recombinase activity in vivo. Using immunohistochemistry, we quantified and analyzed YFP expression and co-labeling of GR and BrdU in the hippocampus of GR(NestinCreERT2) mice after administrating vehicle (control) or tamoxifen. Tamoxifen binds to the mutated estrogen receptor (ERT2) of Cre in cells expressing Nestin, allowing the Cre recombinase to excise the STOP codon in the YFP reporter and the exon 3 segment in GRloxP alleles. Thus, mice receiving vehicle (-tam) maintained an inactive YFP reporter (Fig 1c), while mice receiving tamoxifen (+tam) expressed YFP in nestin-expressing cells and their progeny (Fig 1d). Similarly, adult –tam mice maintained active GR expression in the proliferating (BrdU+) hippocampal cell population (Fig 1e), while in +tam mice, the GR gene was inactivated in the majority of this population (Fig 1f). Quantitative analysis revealed a significant difference between –tam and +tam mice for the percent of BrdU+ cells expressing GR (two-way ANOVA corticosterone x genotype, effect of genotype: p<0.0001, F(1.31)=72.59, cort: p=0.25, interaction: p=0.73) (Fig 1g).
Chronic cort-treatment does not affect NPC survival
To examine adult neurogenesis in this cell-specific GR(NestinCreERT2) model, the experimental design was as follows: GR(NestinCreERT2) mice were treated with tamoxifen or vehicle (controls) at 5-6 weeks of age and then allowed at least two weeks for recovery (Fig 2a). Afterward, half of each of the +tam and –tam cohorts were chronically treated with corticosteroids through their drinking water for the next 20-26 weeks (age and treatment time were counterbalanced across all groups). Four weeks prior to perfusion, 6 mice within each group were given BrdU for analysis of cell survival and fate (Fig 2a). Control mice had similar levels of cell survival regardless of corticosterone treatment (Fig 2b). This effect was not altered by GR ablation (two-way ANOVA, effect of cort: p=0.64; tam: p=0.96). Area measured for cell counts did not affect results as BrdU+ cells per m2 was not significantly different either (data not shown). GrloxP/loxP mice were used as controls to verify NPC survival was not affected by CreERT2 genotype or tamoxifen alone (3-way ANOVA, p>0.05; Fig 2b).
Cort-suppressed differentiation of NPCs may be indirectly mediated
Finally, we sought to determine whether chronic corticosterone treatment reduced neurogenesis by reducing neuronal differentiation, and if this effect was directly mediated through the GR in NPCs. Mice from each group were administered BrdU 4 weeks prior to perfusion. This allowed for cell lineage tracking from newborn precursor to mature neuron72. We measured the number of hippocampal cells that were BrdU positive (proliferating cell marker), and of those, which were co-labeled as NeuN positive (mature neuronal marker) in the DG and hilus (Fig 3). When comparing the percent of BrdU positive cells co-labeling for NeuN, a two way ANOVA of genotype x corticosterone revealed a significant effect of corticosterone (p=0.003, F(1,15)=12.8), but no effect of genotype nor interaction (Fig 3a). This suggests that chronic cort-treatment reduced neurogenesis overall in both control and GRNPCKO groups. Further post-hoc analysis revealed a significant effect of corticosterone between controls (p=0.0087) as well as between GRNPCKOs (p<0.05; Neuman-Keuls), but no significant differences between noncort-treated groups or between cort-treated groups (Fig 3a). Again, examining the number of co-labeled NeuN+ cells per area by two-way ANOVA showed a significant effect of corticosterone (p=0.018, F(1.15)=7.072) (Fig 3b). Neither tam treatment nor CreERT2 genotype alone affected differentiation (data not shown). All groups had a comparable number of BrdU positive cells (Fig 2b) and no significant differences in areas measured (data not shown). These results show that chronic corticosterone treatment reduces neuronal differentiation in vivo, and it is not reversed or attenuated by GR gene inactivation in NPCs.
We have generated mice using an inducible Cre-lox system to conditionally lose GR function in an adult population of NPCs. We were able to show that chronic corticosterone treatment in vivo did not affect the survival rate of NPCs in the mouse hippocampus, however it did inhibit neuronal differentiation. This effect was not blocked by loss of GR in NPCs. In the present study, we show that GR gene function in NPCs is not necessary for cort-induced suppression of neurogenesis, implying that this effect of corticosterone may be mediated indirectly.
This is the first mouse model of GR function that specifically targets adult NPCs and did not require surgery. Previous studies have investigated the function of the GR in vivo using pharmalogical agents and found they played a significant role in the survival169, and proliferation of NPCs167,168. These studies, however, looked at the general role of GR in the brain and were not specifically targeting any cell type making it difficult to assess whether the effect of corticosterone on neurogenesis is mediated by cell-autonomous GRs. Comparatively, Tronche and colleagues created a transgenic mouse with GR gene inactivation in NPCs, but since it was not inducible, all neural cells lost GR function from birth239, making it impossible to study the role of GR in adult neurogenesis. A recently published study used RNA-interference to knockdown GR within the adult neurogenic niche, however these mice had intrahippocampal injections of lentivirus, which is invasive and not specific to proliferating cells218. Here we report that we were able to isolate the direct role of GR function in a discrete population of proliferative cells.
In our study, chronic treatment of corticosterone does not appear to affect survival. Although this supports other results11,110,287, our results are unclear without more information on the effects on proliferation. Many studies have previously demonstrated that chronic stress, as well as chronic corticosterone treatment, significantly suppresses the proliferative nature of NPCs11,108,110,115,116,119,121–128,131,161,162,164,165. Thus, if in fact corticosterone treatment also inhibited proliferation in our study, since there was no difference in the number of NPCs at 4 weeks post-BrdU injection, this would mean that noncort-treated mice had greater pruning among NPCs. In other words, cort-treated controls had less cell death. If proliferation was not affected in the GRNPCKOs, this would imply that GRNPCKOs have greater cell death, regardless of corticosterone treatment. It is worth noting that proliferation was affected in the transgenic mice with GR gene inactivation in all brain cells (F. Tronche, unpublished data), however as mentioned, whether this is also the case for NPC-specific GR gene inactivation is currently unknown. Moreover, if corticosterone had a similar effect on proliferation of GRNPCKOs compared to controls, this would imply that GR gene inactivation in NPCs has no effect on NPC survival. Given these points, in addition to a larger sample size, it would be important to know the effect of GRNPCKOs on proliferation in order to conclude that GR in NPCs does not affect survival rates.
Although many studies report chronic corticosterone treatment reduces neurogenesis, this effect appears to be due to reduced proliferation119,131,161,162,164,165,288. It is important to examine each stage of NPC development to understand how glucocorticoids regulate the rate of neurogenesis. This information will allow for greater elucidation of cort-mediated pathways that affect hippocampal cytoarchitecture. In line with other studies11,165, we find that chronic corticosterone treatment can also inhibit the differentiation of NPCs into mature neurons. This cort-induced suppression appears to be mediated independently of GR activation in NPCs, however, a larger sample size would make these results more convincing.
If in fact GR ablation in NPCs does not block cort-induced suppression of neuronal differentiation, there could be several explanations. First, it could be due to experimental design. It is possible that in our mouse model, GR was not sufficiently knocked out of enough NPCs. Whereas tamoxifen-induced Cre recombinase is not 100% efficient and some NPCs (i.e. type 2b) do not express nestin and so their GR gene remains functional, thus, it remains possible that not enough of the population was affected to block corticosterone influence. Furthermore, it could be that our chronic cort-treatment was too extended that other cort-induced pathways compensated for the lack of GR. This latter explanation also suggests another interpretation of the results, however- that neurogenesis can be regulated indirectly.
Although NPCs can express GR, GR is not expressed ubiquitously in NPCs. Whereas GR was not found to be expressed at all in one earlier study289, a more recent finding showed that only 13% of newborn BrdU+ cells expressed GR37. According to this study, GR was only expressed in approximately half of NPCs at each developmental stage, with the exception of type 2b cells (0% express GR). It was not until cells reached a post-mitotic stage that they all started expressing the GR37. Although this study was done on female mice, there is no known reason to assume that male mice would have a greater percentage of NPCs expressing GR. Thus, if corticosterone does indeed suppress neurogenesis by directly acting through GRs in NPCs, it can only be affecting at most 50% of the NPC pool. This lack of GR in earlier stages of neurogenesis may imply two contrasting hypotheses: either GR expression during early stages is not as functionally active as it is during the post-mitotic stage and thus, corticosterone indirectly influences NPC behavior; or GR expression is always functionally active and it is such that corticosterone directly influences the fraction of GR-expressing NPCs, which is sufficient enough to cause overall suppressed differentiation. Since our results show that when mice have a significant inactivation of GR genes in NPCs, even further reducing the sub-population of GR-expressing NPCs, they still show similarly reduced differentiation, this implies that the former hypothesis is more accurate. That being, corticosterone might be indirectly influencing NPC behavior.
One of the cort-mediated pathways indirectly influencing NPC behavior could still be through GR activation, but through GR in mature neurons or astroglia. Since much of the literature has shown that GR plays a role in cort-induced suppression of neurogenesis, it may be that the GR in NPCs alone is insufficient to drive this effect. Many studies have demonstrated GR expression in mature neurons37,290, astrocytes174,185,291– 293, and oligodendrocytes as well291,294. Interestingly, it has been shown that GR activation in astrocytes can induce secretion of different factors that can mediate neurogenesis, such as basic fibroblast growth factor (FGF2)119 and cell cycle inhibitors293. Additionally, stress effects on neurogenesis were found concomitantly with increases in FGF2 mRNA in the dorsal hippocampus119. They examined this in vitro by treating NPCs with conditioned media from cort-treated astrocytes. Not only did this affect neurogenesis, but also treating NPCs with similar levels of FGF2 alone caused a change in proliferation. Furthermore, neutralizing the FGF2 blocked the effect119. These results suggest that corticosterone may regulate NPC behavior indirectly through activating GR in astrocytes.
Similarly, neurogenesis can be regulated when GR activation in either astrocytes or neurons dysregulates cell signaling by inhibiting both glutamate uptake and N-methyl-D-aspartate (NMDA) receptors162,295,296. It had been shown that GR activation reduced glucose transport into both neurons and astrocytes297. This effect resulted in a cascade of increased levels of damaging glutamate in the synapse, which overactivated NMDA receptors, and thus, increased free cytosolic calcium295. Not only does increased cytosolic calcium (Ca2+) signaling damage the postsynaptic neuron295, but it has also been demonstrated to regulate all stages of neurogenesis (for review298). Furthermore, blocking NMDA receptor activity prevented GR-induced suppression of neurogenesis162,296. Although more research is needed to elucidate these pathways in vivo, it remains another possible explanation for how corticosterone can indirectly influence neurogenesis.
Another way that corticosterone can influence neurogenesis indirectly is through serotonin (5HT1A) receptors. GR antagonists were shown to block the effect of corticosterone treatment on the upregulation of serotonin transporter protein levels in the hippocampus299, demonstrating that cort-induced GR activity promotes serotonin transporter production. Similarly, other studies have demonstrated corticosterone can regulate 5HT1A receptors in the hippocampus300,301. Both activation and inhibition of 5HT1A receptors and transporters can cause changes in proliferation and differentiation of NPCs302,303. These studies suggest that GR-mediated increases in 5HT1A transporter and receptor protein may regulate neurogenesis. Regardless of whether this effect is mediated directly through altered levels of 5HT1A receptors on NPCs, GR activity induced in mature neurons or astroglia may be the initial mechanism that indirectly affects NPC development.
Overall, we found that chronic corticosterone treatment did not disrupt survival rates of NPCs in the adult murine hippocampus, however it did reduce neuronal differentiation. Knocking out GR in the NPCs did not attenuate this effect of cort-suppressed differentiation. Since GR is implicated in regulating neurogenesis, our results suggest that cort-suppressed neurogenesis is mediated indirectly through GR in other cells of the stem cell niche. Furthermore, it would be interesting to know if cort-suppressed proliferation is also unaffected by GRNPCKO, further suggesting that the effects of corticosterone are indirectly mediated. Both corticosteroids and neurogenesis are implicated in psychopathologies and mood disorders5,8,197,276,304. Additionally, they are shown to play a role in regulating memory processes44,45,204,252. Our novel mouse model of GR gene inactivation specific to NPCs in the adult brain allows for a better understanding of the molecular mechanisms that are mediated by corticosterone to regulate adult neurogenesis. This is relevant both to endogenously elevated corticosterone as induced by stress, learning, and exercise, as well as exogenously administered for medicinal purposes. Furthermore, our transgenic mice will allow for a better understanding of the role that GR plays in newborn neurons and potentially contribute to the development of new neuropharmalogical therapies.
Materials and Methods
Nestin-CreERT2 and GRloxP/loxP transgenic mice. To selectively inactivate the GR gene in neural progenitor cells, we generated GRloxP/loxP; R26R-YFP; Tg(NesCreERT2) mice (thereafter denominated GRNesCreERT2), by mating ad-hoc animals, all on a C57BL/6 genetic background. The GRloxP allele contains loxP sites flanking exon 3 of the GR gene, the first zinc finger of the GR DNA-binding domain. The strategic placement of loxP around this specific part of the GR gene not only removes exon 3, but also causes a shift in the open reading frame of the gene. Potential mRNAs generated from the mutated allele would fail in translating any functioning GR protein239. The NesCreERT2 transgene expresses the CreERT2 recombinase gene under control of the Nestin promoter84. The R26R-YFP Cre-reporter allele (Jackson Laboratories) harbors a transcriptional STOP cassette preventing, in the absence of active Cre recombinase, the expression of the YFP open reading frame on the Rosa26 locus. Upon tamoxifen induction in GR(NesCreERT2) mice, Cre recombinase efficiently promotes the recombination between two head-to-tail oriented loxP sites, which leads to excision of intervening DNA, selectively in nestin-expressing cells, thus creating a nestin-specific loss of GR function and active YFP expression. Vehicle-induced GR(NesCreERT2) mice were used as controls. Vehicle- and tamoxifen-induced GRloxP/loxP mice were also tested alongside controls to verify no effects of Cre or tamoxifen. 60 mice (n=10 per group) were used in experiment. For analysis of gene inactivation efficiency, 10 mice per group (-tam/-corticosterone, -tam/+corticosterone, +tam/-corticosterone, +tam/+cort) were used, however, 5 could not be analyzed as a result of 1 death and 4 exclusions due to inefficient recombination in NPCs.
We restricted our analysis to male mice. Animals were bred and raised under a 12h light/dark cycle; temperature was 22±2˚C and humidity 60±5%. Food and water were supplied ad libitum. Experiments were performed in accordance with French (Ministere de l’Agriculture et de la Forêt, 87-848) and European Economic Community (EEC, 86-6091) guidelines for the care of laboratory animals.Tamoxifen induction. Tamoxifen (Sigma T5648) was light-protected and dissolved in 10% ethanol and suspended in 90% sunflower seed oil84. Mice received daily injections at 180mg/kg, i.p. for 5 days when they were 5-6 weeks old. Control mice received vehicle (sunflower oil). Mice receiving tamoxifen were only group-housed with other tamoxifen-receiving mice; likewise, only vehicle (oil)-receiving mice had vehicle-receiving littermates. This guarded against possible cross-contamination of tamoxifen.
Corticosterone treatment. Corticosterone (Sigma C2505) was prepared as described previously11. It was dissolved at 35 µg/mL in a water solution of 0.45% (wt/vol) hydroxypropyl-beta-cyclodextrine (Sigma 332593) by sonication for two hours. Corticosterone treatment was given to mice in opaque water bottles to protect it from light, changed twice a week, and available ad libitum.
BrdU labeling. To trace cell lineage, 31-32 days prior to perfusion, mice were administered with bromo-deoxyuridane (150mg/kg, i.p. dissolved in saline; Sigma B5002) twice daily for 3 days.
Immunohistochemistry. Mice anesthetized with pentobarbital euthanasia solution were perfused transcardially with ice cold 0.1M phosphate buffered saline (PBS) followed by 4% paraformaldehyde (PFA). Brains were removed, fixed overnight in PFA at 4˚C, and transferred to 0.1M PBS. Serial sections were cut coronally at 35 µm using a vibrotome (Vibratome 3000 Plus; Ted Pella, Inc). Staining consisted of 3 x 5min washes in 1xPBS, blocking and permeabilizing in a 1xPBS solution of 5% normal donkey serum (Jackson 017-000-121) and 0.3% tritonX for 30-60 minutes, followed by overnight incubation at 4˚C with primaries. Tissue was then washed 3x5min in PBS, treated with secondary antibody for two hours at room temperature, washed again, fixed in 4% PFA for 15minutes, then washed again. Tissue preparation for BrdU co-labeling then proceeded with treatment of 0.9% saline for 5 minutes, followed by an acid wash in 2N HCL at 37˙C for 30 minutes. Tissue was then washed 3x5minutes, blocked, and incubated with BrdU antibody overnight. The following primary antibodies were used on free-floating sections: rat monoclonal anti-BrdU (Abcam 6326; 1:500), rabbit polyclonal anti-GR (Santa Cruz 1004; 1:500), goat anti-gfp, FITC-conjugated (Rockland 600-102-215; 1:500), and rabbit polycolonal anti-NeuN (Millipore ABN78; 1:500). Secondary antibodies used were as follows: biotinylated donkey anti-rat immunoglobulin G (Jackson 712-065-153, 1:500), donkey anti-rabbit Alexa647 (Jackson 711-605-152, 1:500), and donkey anti-goat AF488 IgG (Jackson 705-545-147, 1:200). Visualization of BrdU was performed with StrepAlexa488 for 1 hour at room temperature. Tissue was treated with Dapi (Invitrogen D1306, 1:20000) when appropriate. Sections were mounted onto slides and coverslipped with DABCO.
Microscopy and quantification.
Confocal laser scanning microscopy (Zeiss) was used for counting fluorescent-labeled cells. It was performed using a 40X-oil objective on a 1 in 10 series of sections through the entire DG. In all cases, 8 hippocampi per animal of each experimental group were randomly selected within each series and analyzed. For survival and differentiation analysis, 32 mice (n=4-6 per group) were given BrdU 4 weeks prior to perfusion; of these, only 27 were analyzed due to 1 death, 1 damaged brain tissue, and 3 exclusions due to inefficient recombination in NPCs: -tam/-corticosterone, n=6; -tam/+corticosterone, n=5; +tam/-corticosterone, n=4; +tam/+corticosterone, n=4; GrloxP/loxP –tam/+corticosterone, n=4; GrloxP/loxP +tam/+corticosterone, n=4. BrdU+ cells were counted in 8 hippocampi of 2 series and multiplied by 10 to represent total numbers throughout DG and hilus. Area of DG and hilus were measured using Stereo Investigator software.
Results are expressed as means±s.e.m. Statistical analysis was performed using Student t-tests when comparing effect of corticosterone between controls, and two-way ANOVA for comparing all groups. Analyses were followed by Dunnett’s or Newman-Keuls post-hoc tests for pairwise comparisons, as appropriate. * indicates p<0.05.
Table of contents :
CHAPTER 1 Introduction
1.1 Scope of thesis
1.2 Stress and neurogenesis in the mammalian central nervous system
1.3 The role of glucocorticoids in hippocampal neuroplasticity
1.4 The role of neurogenesis in hippocampal function
1.5 Functional significance of stress-induced changes to neurogenesis
CHAPTER 2 Regulation of adult hippocampal neurogenesis by GR in newborn neurons
2.5 Materials and Methods
2.6 Figures and Legends
CHAPTER 3 Functional contributions of the glucocorticoid receptor in adult neurogenesis
3.5 Materials and Methods
3.6 Figures and Legends
CHAPTER 4 Introduction to general discussion
4.1 Introduction to general discussion
4.2 CreERT2-mediated GR gene inactivation in neural progenitor cells
4.3 Glucocorticoid receptor signaling in the neurogenic niche
4.4 GR signaling in adult-born granule cells: implications for stress-related brain disorders
4.5 Future directions
CHAPTER 5 References