Neurogenesis occurs in discrete regions of adult mammalian brain, including the subgranular zone of the hippocampus. Hippocampal neurogenesis is enhanced by different classes of antidepressants, but screening for neurogenic actions of novel antidepressants has been inefficient because of limitations of 5-bromo-2′-deoxyuridine labeling techniques. We describe an efficient in vivo method for measuring hippocampal neurogenesis involving incorporation of the stable isotope, 2H, into genomic DNA during labeling with 2H2O (heavy water). Male rodents received 8 to 10% 2H2O in drinking water; DNA was isolated from hippocampal progenitor cells or neurons. Label incorporation into progenitor cells of Swiss-Webster mice revealed subpopulation kinetics: 16% divided with t1/2 of 2.7 weeks; the remainder did not divide over 1 year. Progenitor cell proliferation rates in mice were strain-dependent. Chronic antidepressant treatment for 3 weeks, with 2H2O administered during the final week, increased progenitor cell proliferation across all the strains tested. Fluoxetine treatment increased 2H incorporation into DNA of gradient-enriched neurons or flow-sorted neuronal nuclei 4 weeks after 2H2O labeling, representing the survival and differentiation of newly divided cells into neurons. By screening 11 approved drugs for effects on progenitor cell proliferation, we detected previously unrecognized, dose-dependent enhancement of hippocampal progenitor cell proliferation by two statins and the anticonvulsant topiramate. We also confirmed stimulatory activity of other anticonvulsants and showed inhibition of progenitor cell proliferation by isotretinoin and prednisolone. In conclusion, stable isotope labeling is an efficient, high-throughput in vivo method for measuring hippocampal progenitor cell proliferation that can be used to screen for novel neurogenic drugs.
In rodents, adult neurogenesis occurs in discrete regions of the brain, particularly in the subventricular zone, giving rise to granule cells in the olfactory bulb, and in the subgranular zone, generating new granule cells in the dentate gyrus of the hippocampus (Altman and Das, 1965; Kaplan and Hinds, 1977; Kuhn et al., 1996). Hippocampal neuronal cells in the adult are formed through replication and differentiation from pluripotent neural progenitor cells (Gage et al., 1998). Immature neurons migrate to the granule cell layer and mature over a period of weeks into granule cells that form functional connections (van Praag et al., 2002).
Hippocampal neurogenesis has emerged as a central therapeutic target for antidepressant agents based on accumulating evidence supporting the neurogenic theory of depression (Jacobs et al., 2000; Kempermann, 2002). All the known classes of clinical antidepressant drugs, including tricyclics, monoamine oxidase inhibitors, and selective serotonin reuptake inhibitors (SSRI), have been shown to increase cell proliferation in the hippocampus (Malberg et al., 2000; Santarelli et al., 2003). Moreover, Santarelli et al. (2003) provided evidence, by irradiation of the hippocampus, that hippocampal neurogenesis is required to achieve the behavioral effects of antidepressants in animal models. Hippocampal cell proliferation is decreased under conditions of chronic stress, and this effect is reversed by antidepressant treatment (reviewed by Dranovsky and Hen, 2006; Warner-Schmidt and Duman, 2006). Imaging studies have also shown that depressed human subjects exhibit volume loss and atrophy of the hippocampus (Sheline, 2003). In addition to a role in mood disorders, increased neurogenesis may be a repair mechanism in stroke (Lichtenwalner and Parent, 2006) and traumatic brain injury (Emery et al., 2005) and may also contribute to learning and memory (Shors et al., 2001).
A number of factors and intersecting pathways influence the drug-induced stimulation of neurogenesis in the hippocampus. For example, antidepressant drugs activate intracellular second messenger systems, leading to the activation of transcription factors and neurotrophic factors, finally culminating in increased numbers of new neurons in the hippocampus (Warner-Schmidt and Duman, 2006). The complexity of neurogenic regulation opens the possibility that multiple therapeutic targets may exist, on the one hand, but also makes predicting the effect of any agent acting on a particular target difficult and complicates the interpretation of in vitro screening approaches based on receptor binding or modulation of enzyme activities. Therefore, it is important to validate in vivo the neurogenic activity of agents identified through in vitro screens; moreover, in vivo screening may uncover novel mechanisms that contribute to neurogenesis arising from unanticipated connectivity relationships in the whole organism.
Measurement of neurogenesis in vivo has been problematic, however. Early studies used [3H]thymidine to label dividing cells (Altman and Das, 1965; Kaplan and Hinds, 1977), whereas the most commonly used method currently for measuring cell proliferation in the hippocampus involves injecting animals with 5-bromo-2′-deoxyuridine (BrdU) 24 h before sacrifice (Kuhn et al., 1996; Cameron and McKay, 2001). The hippocampus is then serially sectioned for immunohistochemical detection of BrdU, as well as double-immunohistochemical labeling with neuronal markers, to establish the phenotype of BrdU-positive cells. BrdU labeling has limitations, however, such as a short half-life and rapid clearance of BrdU from the brain, variable efficiency of BrdU entry into cellular precursor pools, and the requirement for high doses to accurately estimate the number of proliferating cells (Cameron and McKay, 2001; Gould and Gross, 2002). Moreover, immunohistochemical enumeration of BrdU-labeled cells in the entire hippocampus is labor-intensive, so that throughput with BrdU labeling is not sufficient for use in broad screening or testing of potential neurogenic agents.
Here, we describe the application of heavy water (2H2O) as a DNA label for quantifying the proliferation of hippocampal progenitor cells, as well as their survival and maturation into neurons. We have previously shown in humans and experimental animals that deuterium (2H) from 2H2O is incorporated constitutively into the deoxyribose moiety of purine deoxyribonucleotides in genomic DNA through the de novo nucleotide synthesis pathway (Fig. 1) and exhibits several advantages as a labeling approach compared with the use of pyrimidine nucleoside labels like BrdU (Neese et al., 2002). Heavy water is easily administered at a constant level for extended periods, without toxicity, and thereby allows quantitative measurement of cell proliferation over the period of label administration. Because hippocampal cell proliferation is increased by antidepressant treatments, we validated the method by measuring the effects of different classes of anti-depressants. In addition, we tested retinoids and glucocorticoids as potential inhibitors of hippocampal progenitor cell proliferation. We then used this biomarker as a relatively high-throughput in vivo screening tool to discover previously unrecognized neurogenic stimulatory actions of approved drugs.
Materials and Methods
Animals. All of the animal studies were carried out within National Institutes of Health guidelines for the care and use of laboratory animals and received approval from the institutional animal use committee. Ten- to 12-week-old male C57Bl/6, Swiss-Webster, or BALB/c mice were obtained from Charles River Laboratories (Wilmington, MA), and 10- to 12-week-old male 129SvEv mice were obtained from Taconic (Oxnard, CA). The outbred Swiss-Webster strain of mice was used for the initial drug-screening experiments. Subsequent experiments were performed in the inbred 129SvEv strain because these showed the least interanimal variability. Male Sprague-Dawley rats (250 g) were obtained from Charles River Laboratories. Animals were housed in a climate-controlled environment and fed standard rodent chow and water ad libitum.
2H2O Labeling. For 2H2O labeling, animals received a priming i.p. bolus of 49 ml/kg >99% 2H2O (Spectra Stable Isotopes, Columbia, MD) containing 0.9% NaCl and were maintained on 10% 2H2O in drinking water for the duration of the labeling period. In previous studies (reviewed in Jones and Leatherdale, 1991), intake of up to 20% 2H2O has no apparent phenotypic or behavioral effects. For protocols involving label incorporation in hippocampal progenitor cells, mice were labeled continuously for up to 1 year. In a separate study, total hippocampal tissue was isolated after 3, 7, or 14 days of labeling. Animals that received drug treatment were labeled during the final 7 to 10 days of treatment. For studies that assessed proliferation of mature neurons, animals were labeled with 10% 2H2O for 3 weeks, after which time the label was discontinued, and animals were sacrificed 4 weeks later.
Drug Treatments. Male rats and mice were treated with anti-depressants of different classes, such as an SSRI (fluoxetine), a tricyclic (imipramine), or a serotonin-norepinephrine reuptake inhibitor (venlafaxine), administered in drinking water. The drug solutions were dissolved at a concentration of 100, 200, and 100 mg/l for fluoxetine, imipramine, and venlafaxine, respectively. These concentrations were calculated to achieve a dose of 10 mg/kg/day for fluoxetine and venlafaxine and 20 mg/kg/day for imipramine, based on the average cage consumption of water (3 ml/30-g mouse or 25 ml/250-g rat). Treatment with antidepressant drugs was continued for 3 to 5 weeks.
Various approved drugs were screened for neurogenic activity (Table 1). These agents were selected based on long-standing use in humans and on the recognition of having several therapeutic actions (i.e., pleiotropic effects). This effort was intended to be a proof-of-concept study to show “indications discovery,” i.e., ability to find unexpected actions of agents with potential use in a new indication. The doses selected were based on use in published preclinical studies with these agents, and the concentrations in food or water were calculated based on the average cage consumption of food (3 g/30-g mouse) and water (3 ml/30-g mouse). As a follow-up study after initial screening, a potential “class” effect for one of the agents (topiramate) tested in the initial screen was investigated by treating mice p.o. for 3 weeks with 1 of 11 other anticonvulsants: valproate (1 g/kg), clonazepam (3 mg/kg), gabapentin (100 mg/kg), carbamazepine (30 mg/kg), ethosuximide (300 mg/kg), levetiracetam (30 mg/kg), oxcarbazepine (100 mg/kg), phenytoin (100 mg/kg), primidone (100 mg/kg), tiagabine (30 mg/kg), or zonisamide (50 mg/kg).
Dose-response studies were also done for topiramate (10, 30, 100, and 150 mg/kg p.o.), atorvastatin (1, 3, 10, and 30 mg/kg in diet), and simvastatin (1, 3, 10, and 30 mg/kg p.o.). In addition, mice received chronic treatment with potential inhibitors of neurogenesis, such as isotretinoin (1 and 3 mg/kg i.p.) or prednisolone (5 and 40 mg/kg in diet).
Isolation of Hippocampal Progenitor Cells and Neurons. Animals were euthanized by CO2 asphyxiation; brains were immediately removed; and the hippocampus was dissected out. Progenitor cells and neurons were isolated by a modification of methods described previously (Palmer et al., 1999). In brief, tissues were finely minced and digested in a solution of papain (4 U/ml) (Worthington Biochemical Corporation, Lakewood, NJ) and DNase (250 U/ml) (Roche Applied Science, Indianapolis, IN) dissolved in Hibernate-A (BrainBits LLC, Springfield, IL). The digested tissue was then mechanically triturated and thoroughly mixed with an equal volume of Percoll solution, made by mixing nine parts of Percoll (Amersham Biosciences, Piscataway, NJ) with one part 10× phosphate-buffered saline (PBS). The cell suspension was fractionated by centrifugation for 30 min, 18°C at 20,000g. Density beads were run in parallel, and the progenitor cells were fractionated between 1.064 and 1.075 g/ml, whereas neuronal cells fractionated at densities ≤1.035 g/ml. The progenitor and neuronal cell fractions were collected, washed free of Percoll, and stored frozen at –20°C until isolation of DNA. A flow chart of the method is shown in Fig. 1A.
Characterization of Progenitor Cells and Neurons by Flow Cytometry. The gradient-purified progenitor cells were immunofluorescently stained for intracellular markers, nestin and vimentin, and analyzed by flow cytometry. The cells were fixed and permeabilized with IntraCyte buffers (Orion Biosolutions, Vista, CA) and incubated overnight at 4°C with mouse antinestin (1:50, Rat401; BD Pharmingen, San Diego, CA) or mouse antivimentin (1:50; BD Pharmingen) primary antibodies. Gradient-purified hippocampal neurons were stained for tetanus toxin C fragment (TTX), a cell surface marker for neurons. The cells were fixed with 4% paraformaldehyde and incubated with TTX followed by anti-TTX mouse monoclonal antibody (Roche Applied Science). After washing, cells were incubated with Alexa 488-conjugated goat anti-mouse IgG secondary antibody (Molecular Probes, Eugene, OR). Propidium iodide (PI) was used to stain DNA, and Alexa 488 staining of PI-positive single cells (gated on plots of forward scatter peak area versus peak height) was analyzed on a Coulter EpicsXL cytometer (Beckman Coulter, Fullerton, CA).
Isolation of Nuclei from Mature Neurons and Sorting NeuN-Positive Nuclei by Flow Cytometry. Neuronal nuclei were isolated from frozen hippocampal tissue by a modification of a recently described method (Spalding et al., 2005). In brief, tissue was homogenized in 1 ml of lysis buffer (0.32 M sucrose, 5 mM CaCl2, 3 mM magnesium acetate, 0.1 mM EDTA, 10 mM Tris-HCl, pH 8.0, 0.1% Triton X-100, and 1 mM DTT). Homogenized samples were gently suspended in 1.8 ml of sucrose solution (1.8 M sucrose, 3 mM magnesium acetate, 1 mM DTT, and 10 mM Tris-HCl, pH 8.0), layered onto a cushion of 1 ml of sucrose solution, and centrifuged at 30,000g for 2.5 h at 4°C. The isolated nuclei were resuspended in 1 ml of PBS and stored overnight at 4°C.
To identify neuronal nuclei, anti-NeuN antibodies were directly conjugated with Zenon mouse IgG labeling reagent (Alexa 488; Molecular Probes) by mixing 10 μl of Alexa 488 conjugate in 100 μl of blocking buffer (PBS/0.5% bovine serum albumin/10% normal goat serum) with 300 μl of NeuN antibody (1 mg/ml; diluted 1:250 in PBS) and incubating at room temperature for 5 min. The suspension of nuclei (1 ml) was added and incubated at 4°C for 1 h. The nuclei were then washed twice with 3 ml of PBS by centrifuging at 1000g for 10 min. PI was used to stain DNA, and neuronal nuclei were sorted as a homogeneous population of NeuNbright cells using a Coulter Epics Elite sorter (Beckman Coulter) with gates set for PI-positive single nuclei.
Measurement of Cell Proliferation. Gas chromatography/mass spectrometry (GC/MS) analyses were performed as described previously (Neese et al., 2002; Busch et al., 2004) to measure 2H incorporation from 2H2O into purine deoxyribonucleotides in genomic DNA. In brief, DNA was isolated from isolated progenitor cells, sorted neurons, or hippocampal tissue using a DNeasy tissue kit (Qiagen, Valencia, CA) and hydrolyzed enzymatically to free deoxyribonucleosides. The deoxyribose moiety of purine deoxyribonucleosides was converted to the pentafluorobenzyl tetraacetate derivative and analyzed by GC/MS in the negative chemical ionization mode using an Agilent (Palo Alto, CA) model 5973 mass spectrometer and a 6890 gas chromatograph fitted with a db-225 column. Selected ion monitoring was performed with mass/charge ratios (m/z) of 435 for the M0 and 436 for M1 mass isotopomer, respectively. Incorporation of 2H from 2H2O into purine deoxyribose was quantified as the molar excess fraction M1 (EM1), i.e., the increase over natural abundance (background) defined as the fractional M1 value for an unlabeled DNA standard from calf thymus.
The fraction of newly labeled cells was calculated as the ratio of excess 2H enrichment (EM1) in isolated cells to the corresponding enrichment in bone marrow DNA (an essentially fully turned-over tissue after 7–10 days of labeling, thereby representing an asymptotic enrichment value for comparison with other tissues) as described previously (Neese et al., 2002). For studies involving shorter labeling periods than 7 to 10 days, 2H incorporation in fully turned-over tissue DNA was estimated from body water 2H enrichments at sacrifice based on previously established relationships between body water 2H enrichments and asymptotic labeling in tissues (Neese et al., 2002).
Statistical Analysis. For label incorporation curves, the data were fit by nonlinear regression analysis (SigmaPlot). Student's t test was used with a 95% confidence interval for comparison between two groups. For comparison between multiple groups, one-way analysis of variance was used with a post-hoc Student-Newman-Keul test for all the pairwise multiple comparisons or Dunnett's test for comparisons with a negative control (SigmaStat). Data were considered significant at p < 0.05.
Flow Cytometric Analysis of Gradient-Enriched Hippocampal Progenitor Cells. Measurement of 2H incorporation into the DNA of hippocampal progenitor cells requires isolation of this cell population from intact tissue. To this end, progenitor cells were isolated from the hippocampus of Swiss-Webster mice by Percoll gradient fractionation (Fig. 2A); the isolated cells were fixed and permeabilized, stained for specific intracellular markers of progenitor cells (nestin and vimentin), and analyzed by flow cytometry after gating on nucleated (PI-stained) cells and excluding doublets. A majority of cells stained positively for either nestin (65% positive in Fig. 2B) or vimentin (72% positive) compared with isotype controls (2.1% positive). A broad distribution of fluorescence intensity was observed, and the cells remaining in the negative region of the dot plot appeared to be weakly stained.
Label Incorporation Kinetics in Progenitor Cells. To characterize the proliferation kinetics of gradient-enriched hippocampal progenitor cells, Swiss-Webster mice were labeled continuously with 10% 2H2O in drinking water starting at 10 weeks of age; hippocampal progenitor cells were isolated after various labeling times, and 2H incorporation into DNA, analyzed by GC/MS, was used to determine the fraction of cells that had incorporated the label through cell division (Fig. 2C). Approximately 16% of cells incorporated label at plateau, with a half-life within this dividing population of approximately 2.7 weeks; the majority of cells continued to be unlabeled over the course of 1 year. This is consistent with a “kinetic subpopulation” pattern wherein a preponderance of nondividing precursors is present with a subset of actively dividing cells that enters and exits the pool (by differentiation or death). The baseline rate of progenitor proliferation was strain-dependent (Fig. 2D). The initial rates of labeling, measured after 1 week of 2H2O intake, varied approximately 3-fold among four mouse strains tested and were significantly (p < 0.001) different from each other. The highest proliferation rate was observed in C57Bl/6 mice, whereas the 129SvEv mice had the lowest rate of proliferation.
The fraction of new cells in whole hippocampal tissue of C57Bl/6 mice after 3, 7, or 14 days of labeling was 0.7 ± 0.01, 1.3 ± 0.3, and 2.9 ± 0.6%, respectively. Thus, gradient isolation enriched for proliferating cells, and the consistency of the labeling results indicated that the cell isolation method was highly consistent in this regard.
Effects of Antidepressant Drugs on Progenitor Cell Proliferation in Rodent Hippocampus. Male Swiss-Webster mice (Fig. 3A) or Sprague-Dawley rats (Fig. 3B) were treated for 3 weeks with antidepressant drugs of different classes: an SSRI (fluoxetine, 10 mg/kg/day), a tricyclic (imipramine, 20 mg/kg/day), or a serotonin-norepinephrine reuptake inhibitor (venlafaxine, 10 mg/kg/day). After labeling with 2H2O during the last week of treatment, antidepressant-treated animals from all the groups showed a significant increase in the progenitor cell proliferation rate in the hippocampus (Fig. 3, A and B). Both baseline proliferation and the magnitude of the drug effects were somewhat different between mice and rats, however. Fluoxetine treatment produced a significant (p < 0.01) increase in the hippocampal progenitor cell proliferation in C57Bl/6, 129SvEv, Swiss-Webster, and BALB/c mice (Fig. 3C). The percent stimulation of progenitor cell proliferation by fluoxetine was similar across strains.
Effects of Antidepressant Treatment on Survival and Differentiation of Newly Divided Cells. A “pulse-chase” protocol was used to assess the survival and differentiation of progenitor cells into neurons in vehicle- and fluoxetine-treated animals. Male 129SvEv mice were treated with fluoxetine (10 mg/kg/day) or vehicle for 5 weeks and labeled with 10% 2H2O for the last 3 weeks of treatment. A longer labeling time was used in this protocol to allow greater numbers of newly divided progenitor cells to be labeled, allowing for some label loss as the result of death of labeled cells and differentiation into non-neuronal progeny. The antidepressant treatment and 2H2O intake were then discontinued, and the animals were followed for another 4 weeks, after which time intact neurons or neuronal nuclei were isolated for analysis of DNA labeling. Figure 4A shows that >80% of gradient-enriched hippocampal neurons stained positively for TTX, a surface marker for neuronal cells. However, further purification of neurons by flow cytometric sorting proved difficult because of the large size range of neuronal cell bodies and large amount of axonal debris present in these gradient fractions (data not shown). Successful resolution of NeuNbright and NeuNdim nuclei obtained from frozen mouse brain was possible (Fig. 4B), allowing sorting of brightly stained nuclei of mature neurons to >98% purity on reanalysis (not shown). GC/MS analysis of extracted DNA from both preparations showed that fluoxetine treatment produced a significant increase in the fraction of 2H-labeled neurons 4 weeks after the end of a 3-week labeling period (Fig. 4, A and B). 2H incorporation into DNA of gradient-enriched neurons exceeded that of NeuNbright neuronal nuclei; although Percoll gradients efficiently resolve proliferating neural progenitors from neuronal cells, the gradient-enriched population may include a greater proportion of immature neurons derived from dividing precursors that do not yet express high levels of NeuN. Taken together, these results confirm that differentiation of recently divided precursors into mature neurons is enhanced by fluoxetine and show that this antidepressant drug effect is detectable by 2H2O labeling.
Effect of Retinoid or Glucocorticoid Administration on Progenitor Cell Proliferation. We also evaluated the activity of a retinoid and a glucocorticoid on hippocampal progenitor cell proliferation. Male 129SvEv mice were treated with isotretinoin (1 or 3 mg/kg i.p.) for 3 weeks and labeled with 10% 2H2O in the final 10 days of treatment. There was a dose-dependent decrease in hippocampal progenitor cell proliferation following retinoid treatment (Fig. 5A). Treatment with a 3-mg/kg dose of isotretinoin produced a 38% reduction over baseline, which was statistically significant (p < 0.05) compared with vehicle-treated controls.
Male Swiss-Webster mice were treated with prednisolone (5 or 40 mg/kg in diet) and labeled with 8% 2H2O for 4 weeks. These mice received a slightly different treatment and labeling protocol because they were part of a study investigating the effects of glucocorticoid treatment on several metabolic pathways concurrently. Prednisolone treatment produced a dose-dependent significant (p < 0.01) decrease in hippocampal progenitor cell proliferation (Fig. 5B). Progenitor cell proliferation was significantly (52%) reduced at the 40-mg/kg dose compared with untreated controls and showed a trend toward reduction at the 5-mg/kg dose.
Use of the Hippocampal Progenitor Cell Proliferation Assay to Screen Drugs. Next, we explored the feasibility of using hippocampal progenitor cell proliferation as a screen for candidate neurogenic agents. To this end, a panel of approved drugs with known or suspected pleiotropic actions (Table 1), but not previously known to have neurogenic activity in normal murine hippocampus, was tested for their effects on proliferation of hippocampal progenitor cells in Swiss-Webster mice (Fig. 6). Two of these agents were found to increase the rate of progenitor cell proliferation. The hydroxymethyl-glutaryl CoA reductase inhibitor (statin) atorvastatin (10 mg/kg p.o.) and the anticonvulsant topiramate (100 mg/kg in diet), each given daily for 3 weeks, significantly increased progenitor cell proliferation as measured by 2H2O labeling during the 3rd week. The increases were of a similar magnitude as observed with the positive controls, fluoxetine and imipramine (Fig. 6). No stimulatory effects were seen for nine other drugs given by various routes (diet, drinking water, oral gavage, or i.p.) at daily doses reported in the literature to be pharmacologically active for other actions (Table 1).
In follow-up studies, we assessed dose-response relationships of the compounds with stimulatory activity and tested other agents in their classes (statins and anticonvulsants). Topiramate produced a dose-dependent increase in the proliferation of hippocampal progenitor cells of 129SvEv mice (Fig. 7A). Treatment with 30 mg/kg/day p.o. produced an approximately 70% increase over baseline proliferation, approaching a plateau at higher doses (100 and 150 mg/kg/day). We also screened 11 other anticonvulsant drugs at single doses known to have anticonvulsant activity in animals (Fig. 7B). Treatment of Swiss-Webster mice for 3 weeks with either valproate (1 g/kg in diet) or oxcarbazepine (100 mg/kg p.o.) significantly increased progenitor cell proliferation. Treatment with valproate doubled the number of proliferating cells compared with vehicle, whereas oxcarbazepine treatment produced a 67% increase over baseline proliferation. A valproate stimulatory effect on neurogenesis in the hippocampus has been previously reported (Hao et al., 2004), but the activity of oxcarbazepine is a novel finding. The other anticonvulsants either had no effect on progenitor cell proliferation or were slightly inhibitory at the single doses tested. We conclude that a subset of anticonvulsants is neurogenic, as judged by DNA labeling of hippocampal progenitors.
The stimulatory action of statins on hippocampal progenitor cell proliferation was also confirmed and extended in dose-response studies using atorvastatin or simvastatin. Treatment of 129SvEv mice with increasing doses of atorvastatin (1, 3, 10, and 30 mg/kg in diet) produced a dose-dependent increase in hippocampal progenitor cell proliferation, with the maximal (68% increase over baseline) effect seen at the 3-mg/kg dose (Fig. 8A). Chronic p.o. treatment with another hydroxymethyl-glutaryl CoA reductase inhibitor, simvastatin, also produced a dose-dependent increase in the proliferation of precursor cells (Fig. 8B) in the hippocampus of 129SvEv mice. This more lipophilic statin showed significant stimulatory activity at the 10- and 30-mg/kg doses.
Neurogenesis has been difficult to exploit as a drug target for therapeutic actions such as antidepression or cognitive enhancement because of methodological and practical limitations in measuring proliferation of brain cells in vivo by BrdU immunohistochemistry (cf. Introduction). Besides other methodological and technical concerns, perhaps the most important practical limitation of the BrdU approach for in vivo screening is the need for manual counting of multiple tissue sections, which limits throughput. In contrast, mass spectrometric analysis of DNA labeling after p.o. 2H2O labeling has advantages of automated sample processing and data analysis. Measurements are objective, reproducible, quantitative, and amenable to rigorous quality control, thus overcoming the limited throughput of counting BrdU+ cells on serial sections and subjectivity of visual scoring of positive cells. We previously developed a mass spectrometric technique for quantifying proliferation of slowly dividing cells based on cumulative 2H incorporation into DNA during continuous labeling with 2H2O in rodents and humans, and we have applied this approach to a variety of cell types (Neese et al., 2002). We have also shown a significant correlation between BrdU labeling index and fractional synthesis of DNA measured by 2H2O method in the same animal (Kim et al., 2005). Here, we have for the first time applied this technique to rodent brain and have shown that it affords sufficient sensitivity, reliability, and throughput (processing of 40–50 samples from cell isolation to GC/MS analysis can be done by a technician in 5 days) to enable the discovery of novel neurogenic agents by in vivo screening.
Overall, our 2H2O labeling results are consistent with measurements of hippocampal neurogenesis by BrdU labeling. Assuming that the hippocampus has ≈2 × 106 total cells (Abusaad et al., 1999), the labeling rate of 0.2% per day in total C57Bl/6 hippocampus that we observed with 2H2O would be equivalent to ≈4000 labeled cells/day, a value comparable with estimates by 12- to 24-h saturation labeling with BrdU (Hayes and Nowakowski, 2002). Palmer et al. (1999) reported that 0.7% of progenitor cells were labeled after daily BrdU injection for 6 days in a confocal analysis of BrdU incorporation into gradient-isolated progenitor cells from rat hippocampus, a similar value as we measured by 2H2O labeling on similarly isolated rat hippocampal progenitors (1% per week). In mice, the hierarchy of baseline progenitor cell proliferation across different mouse strains in our study agreed well with strain effects on total hippocampal cell proliferation obtained by BrdU labeling (C57Bl/6 > BALB/c > Swiss-Webster > 129/Sv) (Kempermann et al., 1997; Hayes and Nowakowski, 2002) and further supports the view that genetic background strongly influences hippocampal neurogenesis. Interestingly, genetic differences in hippocampal neurogenesis have been shown to correlate with hippocampal function (Kempermann et al., 1998), and recently, using recombinant inbred strain of mice, Kempermann et al. (2006) have identified several genes that control adult neurogenesis. In terms of survival and differentiation of proliferated cells, 1% new NeuN+ve neurons were observed in the hippocampus of 129SvEv mice 1 month after a 3-week 2H2O labeling protocol, similar to the estimation of BrdU-labeled hippocampal granule cells in the closely related 129/SvJ strain of mice (Kempermann et al., 1997). Most importantly for drug studies, 2H2O labeling qualitatively and quantitatively reproduced the known enhancing effects of antidepressants of various classes on hippocampal progenitor cell proliferation (Malberg et al., 2000; Santarelli et al., 2003), as well as the known inhibitory effect of isotretinoin on hippocampal progenitor cell proliferation (Crandall et al., 2004), and the proneurogenic activity of valproate (Hao et al., 2004). We conclude that 2H2O labeling generates similar results and reveals the same neurogenic drug actions as BrdU labeling.
For screening purposes, we chose to analyze progenitor cells rather than mature neurons, even though labeling of both was increased by antidepressants. Progenitor cells comprise the majority of proliferating cells in the hippocampus, and sufficient 2H label for quantification is incorporated into DNA of gradient-enriched progenitors after only 7 to 10 days of 2H2O exposure, whereas label detection in mature neurons takes an additional 4 weeks, and the new neurons are diluted into a much larger number of nondividing cells. Because standard antidepressants exhibit a lag period of approximately 10 days before stimulating neurogenesis in rodent hippocampus (Santarelli et al., 2003), we designed our screening experiments for 3 weeks of treatment, with 2H2O labeling for the last 7 to 10 days. The lag between drug treatment and labeling can be shortened for detecting early onset of action of novel agents.
Using 2H2O labeling, we were able to explore aspects of hippocampal cell dynamics that would not have been readily accessible by BrdU labeling. First, analysis of gradient-enriched hippocampal progenitors after up to 1 year of in vivo 2H2O labeling revealed kinetic heterogeneity in this population: only ∼16% of these cells turned over during this time. This finding suggests that high-density hippocampal gradient fractions harbor a mixture of proliferating progenitors and resting cells. 2H2O labeling may be valuable in searching for cellular markers associated with the proliferating progenitor cell subset.
Second, we were able to exploit the increased throughput of the 2H2O progenitor cell proliferation assay to screen a sizable panel of approved drugs for previously unknown neurogenic effects in the hippocampus. Two of the 11 agents tested initially, atorvastatin and topiramate, increased progenitor cell proliferation, and these findings were robust in follow-up dose-response studies. Moreover, the proliferative activity of these agents appeared to be shared by other drugs of their class: a second statin, simvastatin, showed stimulatory activity, as did another 2 of 12 structurally diverse anticonvulsants tested, oxcarbazepine and valproate. This surprisingly high “hit” rate supports our notion, discussed in detail elsewhere (Turner and Hellerstein, 2005), that unanticipated cross-talk of drug actions on apparently off-target pathways in vivo might be more common than previously thought because of unappreciated connectivity relationships in complex metabolic networks. In the case of neurogenesis, this makes sense in view of the recognized complex regulation of neurogenesis through sensory input, neurotransmitters, hormones, and neurotrophic factors such as brain-derived neurotrophic factor, vascular endothelial growth factor, insulin growth factor (Warner-Schmidt and Duman, 2006), and several other known and unknown pathways.
The drug activities that we uncovered also seem plausible, after the fact, based on published literature. Atorvastatin has been reported to boost neurogenesis in the hippocampus and subventricular zone after middle cerebral artery occlusion, a model for stroke (Chen et al., 2003), but drug effects in uninjured brain were not measured in that study. Increased expression of vascular endothelial growth factor and brain-derived neurotrophic factor may mediate the neurogenic effect of atorvastatin in both stroke-induced (Chen et al., 2005) and in unmanipulated animals. Moreover, statins also alter expression of genes associated with apoptosis, cell growth, and signaling (Johnson-Anuna et al., 2005). The proliferation-stimulating activities of oxcarbazepine and topiramate are plausible, given published results with valproate, although these results would have been difficult to predict because many anticonvulsants did not increase progenitor cell proliferation. Although the exact mechanism of neurogenic activation by some anticonvulsants remains unknown, these drugs influence several neurotransmitter systems, as well as intracellular signaling cascades, such as the extracellular signal regulated kinase pathway, which has been shown to be activated by the neurogenic anticonvulsant valproate (Hao et al., 2004). The inhibition of hippocampal progenitor cell proliferation by the synthetic corticosteroid analog, prednisolone, confirms the suppressive effect of glucocorticoids because similar effects have been reported for corticosterone and dexamethasone (Cameron and Gould, 1994; Kim et al., 2004).
Intriguingly, the novel proneurogenic and antineurogenic drug effects that we discovered closely parallel the known behavioral effects of these drugs. Retinoid therapy may cause depression in humans as a side effect of treatment (Hull and D'Arcy, 2003). Likewise, conditions exhibiting hypercortisolism (Cushing's syndrome, stress, glucocorticoid therapy) are also associated with clinical depression (Mitchell and O'Keane, 1998). Conversely, anticonvulsants have been used as augmentation therapy in depression (Hantouche et al., 2005); differential neurogenic activity may guide the choice of anticonvulsants for this therapeutic use. Moreover, there is increasing evidence that statins may have utility in the treatment of neurological diseases (Menge et al., 2005). Long-term use of statins has been shown to be associated with reduced risk of anxiety, depression, and hostility (Young-Xu et al., 2003). In a trial for Alzheimer's disease, patients treated with atorvastatin for 1 year showed significant improvement of depression (Sparks et al., 2005). Together with our observations of progenitor cell stimulatory activity of these drugs, these findings provide further support for the neurogenic theory of depression and indicate that in vivo screens for neurogenesis may be useful in the discovery and preclinical validation of novel antidepressants.
Finally, hippocampal neurogenesis is involved in other biological processes besides depression. Because neurogenesis plays a significant role in synaptic plasticity, disorders of learning and memory may be amenable to discovery efforts using this as a biomarker. Neurogenesis may also be a therapeutic target for other conditions, such as traumatic brain injury, stroke, and Alzheimer's disease.
In conclusion, 2H2O labeling represents a quantitative, reproducible, and relatively high-throughput in vivo method for measuring hippocampal progenitor cell proliferation, which is closely linked to neurogenesis and is useful for screening and discovering novel neurogenic stimulatory drugs. Statins and certain anticonvulsant agents were discovered to have potent stimulating activity and may have therapeutic uses based on this activity.
We thank Michael Swenson, Kerstin Kiefer, Michelle Goldfinger, Chancy Fessler, and Holly Turner for technical assistance.
This study was supported in part by KineMed, Inc. and University of California-Industry Discovery Program (BioStar Grant No. Bio04-10445 to M.K.H.).
Part of this work was presented at the Annual Society of Neuroscience Meeting: Shankaran M, King C, Busch R, Gee T, Keifer K, Mahsut A, and Hellerstein M (2005) In vivo heavy water (2H2O) labeling: a novel method for measuring hippocampal cell proliferation and neurogenesis. Program 1024.9, in 2005 Abstract Viewer/Itinerary Planner; 2005 Nov 12–16; Washington, DC. Online, Society for Neuroscience, Washington, DC.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
ABBREVIATIONS: SSRI, selective serotonin reuptake inhibitor(s); BrdU, 5-bromo-2′-deoxyuridine; PBS, phosphate-buffered saline; TTX, tetanus toxin C fragment; PI, propidium iodide; GC/MS, gas chromatography/mass spectrometry.
- Received July 11, 2006.
- Accepted September 12, 2006.
- The American Society for Pharmacology and Experimental Therapeutics