Bromodeoxyuridine Induces Senescence in Neural Stem and Progenitor Cells

Abstract

Bromodeoxyuridine (BrdU) is a halogenated pyrimidine that incorporates into newly synthesized DNA during the S phase. BrdU is used ubiquitously in cell birthdating studies and as a means of measuring the proliferative index of various cell populations. In the absence of secondary stressors, BrdU is thought to incorporate relatively benignly into replicating DNA chains. However, we report here that a single, low-dose pulse of BrdU exerts a profound and sustained antiproliferative effect in cultured murine stem and progenitor cells. This is accompanied by altered terminal differentiation, cell morphology, and protein expression consistent with the induction of senescence. There is no evidence of a significant increase in spontaneous cell death; however, cells are rendered resistant to chemically induced apoptosis. Finally, we show that a brief in vivo BrdU regimen reduces the proliferative potential of subsequently isolated subependymal zone neurosphere-forming cells. We conclude, therefore, that BrdU treatment induces a senescence pathway that causes a progressive decline in the replication of rapidly dividing stem/progenitor cells, suggesting a novel and uncharacterized effect of BrdU. This finding is significant in that BrdU-incorporating neural stem/progenitor cells and their progeny should not be expected to behave normally with respect to proliferative potential and downstream functional parameters. This effect highlights the need for caution when results based on long-term BrdU tracking over multiple rounds of replication are interpreted. Conversely, the reliable induction of senescence in stem/progenitor cells in vitro and in vivo may yield a novel platform for molecular studies designed to address multiple aspects of aging and neurogenesis.

Introduction

The brominated thymidine analog 5-bromo-2′-deoxyuridine (BrdU) incorporates into DNA during the S phase of the cell cycle and can therefore be used to detect DNA synthetic events, including cell division, DNA repair, and cell cycle reentry [1–3]. BrdU is commonly used to quantitate the proliferative index [4, 5] or to birthdate, identify, and track cycling endogenous and transplanted central nervous system (CNS) stem/progenitor cells [6–12]. However, quantitative comparison of independent studies of neurogenesis is difficult because BrdU does not exclusively label dividing cells, and there is no standardized dosing schedule (see [13] for review). In addition to these technical issues, some data suggest that detrimental downstream functional effects result from BrdU exposure. For example, early work showed that exposure to BrdU before maturation of the blood-brain barrier is toxic to neuronal cells [14, 15]. BrdU has also been shown to modulate the growth and differentiation of cultured neuroblastoma cells [16] and to be selectively toxic to cultured neuronal precursor cells in an extracellular signal-regulated kinase-dependent manner [17]. Finally, it has been shown that, in a variety of primary and transformed mammalian cells, BrdU administration results in the upregulation of several senescence-associated mRNAs and proteins [18, 19].

Senescence was originally described by Hayflick and Moorehead in 1961 [20] in studies showing that normal human fibroblasts have a finite ability to proliferate. These cells remain viable and bioactive, but the culture in its entirety experiences an irreversible loss of the ability to divide. Senescence is characterized most typically by a flat, enlarged cell shape, the induction of senescence markers such as senescence-associated β-galactosidase (SA-β-Gal), and resistance to apoptosis [21–23]. Senescence can be induced by a variety of stimuli and is currently understood to be stimulus-, cell type-, and species-specific to varying degrees (see [24] for review). Furthermore, the initiating events leading to senescence, the resulting molecular signature, and the sequence of altered gene/protein expression are not completely understood. Importantly, despite its ubiquitous use in stem cell biology, the implications of BrdU incorporation on the long-term function of multipotent stem and progenitor cells have not been systematically investigated.

Our present studies were designed to test the hypothesis that single-pulse BrdU exposure induces senescence in neural stem and progenitor cells. We show that BrdU exerts a strong anti-proliferative effect on cultured murine stem/progenitor cells that becomes more pronounced with age. Reduced proliferation is concurrent with the onset of a senescent phenotype in BrdU-treated cells that alters cell morphology, differentiation potential, and susceptibility to apoptosis. We further show increased expression of senescence markers and downstream signaling/cell cycle proteins known to be activated in senescent cells. Finally, we show altered proliferation and differentiation of neurosphere-forming cells (NFC) isolated from animals administered an experimentally relevant BrdU treatment regimen. Together, these data uncover a novel and uncharacterized effect of BrdU administration on stem/progenitor cells that has profound implications for the interpretation of results obtained with this thymidine analog. Specifically, long-term in vivo BrdU administration may alter the biology of stem and precursor cell pools, severely limiting downstream applications. Conversely, the reliable induction of a senescent-like phenotype in expandable, multipotent neural stem cells may represent a novel platform for molecular studies designed to address questions regarding multiple aspects of neurogenesis and aging.

Materials and Methods

Animals, Cell Culture, and Reagents

Primary neurospheres (NS) and neurogenic astrocytes were derived from neonatal C57BL/6 mice (postnatal day [P] 4–P8). In vivo studies were conducted using young C57BL/6 mice (~P21). Animals were housed at the University of Florida’s Department of Animal Care Services, and all procedures were in compliance with the regulations of the institutional animal care and use committee. NS were generated as described [25]. In brief, brains were removed from euthanized animals, incubated in trypsin, and dissociated into a single-cell slurry with a series of decreasing bore glass pipettes. The slurry was plated overnight in growth media (Dulbecco’s modified Eagle’s medium [DMEM]/Ham’s F12, 5% fetal bovine serum [FBS], 10 ng/µl basic fibroblast growth factor [bFGF], and epidermal growth factor [EGF]). To isolate NFC, the slurry was aspirated, pelleted by centrifugation, and incubated in trypsin for 2 minutes. Cells were gently triturated, washed, and resuspended. NFC were then plated in nonadherent flasks at clonal density (10,000 cells/cm²) in growth medium. Neurogenic astrocytes were generated as described [26] either from nestin-GFP or C57BL/6 animals. In brief, cells adherent to culture flasks at the time of slurry removal (see above) were washed and fed with new growth media. Cells were allowed to reach confluence and passaged two to three times before assay. Secondary NS were generated as described [27]. In brief, NS from embryonic day 14 (E14) or neonatal animals with an average diameter of approximately 150 µm (usually achieved within 4–5 days after initial plating of NFC) were incubated in 0.25% trypsin/EDTA at room temperature for 2 minutes and then gently triturated through a 1,000-µl pipette tip. Single cells derived from the dissociated primary NS were then counted with a hemocytometer and reseeded at 7.5 × 10⁵ cells/ml. NS were serially analyzed in this manner for two to five passages, and the total number of cells generated at each passage was recorded. Mesenchymal stem cell (MSC) cultures were derived from adult C57BL/6 mice as we described previously [28]. In brief, animals were given a lethal dose of isoflurane and sacrificed by cervical dislocation. Tibiae and femurs were removed, and an 18-gauge needle was used to flush the bone marrow compartment (phosphate-buffered saline [PBS plus 1% antibiotic/antimycotic (Ab/Am)]). Recovered marrow cells were gently triturated, incubated in PBS +1% Ab/Am for 15 minutes, and plated in growth media (DMEM, 10% FBS, and 1% Ab/Am) until confluence.

BrdU, iododexoxyuridine (IdU), and chlorodeoxyuridine (CldU) were purchased from Sigma-Aldrich (catalog nos. B5002, I7125, and C6891, respectively; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com). Ethynyldeoxyuridine (EdU) was purchased from Invitrogen (catalog nos. C35002; Invitrogen, Carlsbad, CA, http://www.invitrogen.com).

In Vitro and In Vivo BrdU Administration Regimens

Two in vitro BrdU administration regimens were used. For most in vitro NS culture conditions, a single BrdU pulse was administered at the time of plating of the original NFC, immediately after isolation from tissue (referred to as “BrdU treatment”). This procedure was used to conservatively assess long-term effects of experimentally relevant BrdU dosage, pulse time, and analysis time point after longer term labeling experiments. Alternatively, a single 24-hour pulse of BrdU was administered to preformed NS immediately before cell fixation (referred to as “BrdU labeling”) (Fig. 2) and used to recapitulate an experimental dosage, pulse time, and analysis time point within the range represented in the literature for short-term labeling experiments or those designed to assess the proliferative index. For adherent neurogenic astrocyte monolayers, a single BrdU pulse was administered 24 hours after flasks were seeded (thus bypassing the lag phase of initial growth). After 24 hours the BrdU was removed, and fresh growth medium was applied. Cells were harvested for analysis at the indicated time points. The time points (i.e., 96 hours, 120 hours, etc.) thus reflect time after BrdU wash-off or “post-pulse.” For in vivo administration, animals were given 50 or 100 mg/kg BrdU i.p. four times over 24 hours, and animals were sacrificed 2 hours after the last injection. Vehicle control animals were given an equivalent number of i.p. injections consisting of 0.9% saline.

Figure 2.

Timing of bromodeoxyuridine (BrdU) administration results in differential morphological and antigenic changes in treated cells. Preformed, attached NS subjected to a BrdU-labeling regimen (50 µM BrdU for 24 hours just prior to fixation) show BrdU incorporation in some cells (A) and exhibit typical process extension and expression of neural phenotype markers (B). Neurosphere-forming cells subjected to a “BrdU treatment” regimen (one pulse of 50 µM BrdU 24 hours after dissociation from tissue, 7 days before fixation) show BrdU retention in most cells (C) and exhibit large-cell morphology, diminished process elongation and reduced expression of mature neuronal/astrocytic markers (D). (B): Maximal projection of a 19-step, 1-µm z-stack obtained with a spinning disc confocal microscope. All other panels are single focal plane epifluorescence images. (A, C): green, BrdU; blue, 4,6-diamidino-2-phenylindole (DAPI). (B, D): green, glial acidic fibrillary protein; red, β-III tubulin; blue, DAPI. All scale bars = 50 µm.

Analysis of Neurosphere Frequency, Diameter, and Multipotency

NS were assessed for frequency and size as a measure of NFC number and proliferative capacity, respectively. NS frequency was determined by counting random triplicate 50-µl aliquots using a ×10 objective. Total NS number was extrapolated to total culture volume for each field counted. Because of the presence of large, hypertrophic cells (i.e., ≤30 µm in diameter) in these cultures (unpublished observation), only spheres ≥40 µm were included for analysis. NS size was determined by measuring the diameter of 10 randomly chosen spheres in each well (triplicate conditions) using calibrated measuring probes associated with digital image capture software (Spot Advanced; Diagnostic Instruments, Inc., Sterling Heights, MI, http://www.diaginc.com). To assess multipotency, NS were propagated for 7–10 days, selected by a hand-held pipette under a phase-contrast microscope, transferred to glass coverslips coated with poly-l-ornithine, and maintained in growth medium without the growth factors bFGF and EGF. NS were allowed to attach and differentiate for 2–3 days. Coverslips were then fixed for 1 hour with 4% paraformaldehyde and processed for immunolabeling to assess the expression of neuronal and astrocytic lineage markers.

Neurogenic Astrocyte Monolayer and Mesenchymal Stem Cell Growth Curves

Astrocytes or MSC were plated in triplicate at 50,000 cells/T-25 flask. Twenty-four hours later, cells received a single 24-hour pulse of BrdU (0, 1, 10, or 50 µM). For the thymidine analog growth curve (supplemental online Fig. 4), control cells were compared with cells that received a single 48-hour pulse of 10 µM BrdU, CldU, IdU, or EdU. Cells were then trypsinized and quantified using a Z2 Coulter Counter (Beckman Coulter, Fullerton, CA, http://www.beckmancoulter.com) at various postadministration intervals. Care was taken to quantify all cells before 100% confluence to avoid potential ceiling effects on cell density.

Immunocytochemistry and SA-β-Gal Stain

Cells plated on poly-l-ornithine-coated coverslips were fixed in 4% paraformaldehyde for 1 hour at room temperature (RT) and blocked for 1 hour at RT with PBS containing 0.01% Triton X-100 and 10% FBS). Primary antibody (rabbit anti-β-III tubulin 1:200 or rabbit anti-glial fibrillary acidic protein [GFAP] 1:2000) was applied overnight at 4°C. Coverslips were washed twice for 10 minutes each in wash buffer (PBS-0.01% Triton X-100) and incubated with fluorescence-conjugated secondary antibody (goat anti-mouse or goat anti-rabbit) for 3 hours at room temperature. Slips were washed twice for 10 minutes each in wash buffer, mounted on positively charged glass slides (Fisherbrand Superfrost/Plus; Fisher Scientific, Pittsburgh, http://www.fisherscientific.com) and cover-slipped in Vectashield containing, 4,6-diamidino-2-phenylindole counterstain (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). Fluorescence micrographs were obtained with either a Leica DMLB epifluorescence microscope equipped with a color Spot cooled charge-coupled device (CCD) digital camera or an Olympus IX81-DSU spinning disc confocal inverted microscope equipped with a color-cooled Hamamatsu CCD digital camera (Fig. 2B; supplemental online Fig. 1B).

SA-β-Gal activity was detected as described [22]. In brief, cells were incubated overnight at 37°C in buffer containing 1 mg/ml 5-bromo-4-chloro-3-indolyl β-d-galactoside (X-Gal), 40 mM sodium citrate (pH 6.0), 5% dimethylformamide, 5% potassium ferrocyanide, 5 mM potassium ferricyanide, 150 mM sodium chloride, and 2 mM magnesium chloride. Slides were viewed with a compound microscope and scored for the SA-β-Gal label as indicated by blue/green reactivation product over the cell soma. After BrdU administration (0, 1, 10, and 50 µM for a 24-hour pulse), cells were quantified by counting five random fields per coverslip (plated in triplicate).

Western Blot Analysis

Protein isolation and Western blot analysis were conducted as described previously [29]. In brief, cells were lysed using RIPA extraction buffer containing proteinase inhibitor cocktail. Total cellular protein was quantified and loaded (10–15 µg/lane) onto 2-[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)propane-1,3-diol (bis-Tris) 4–12% density gradient gels for SDS-polyacrylamide gel electrophoresis. Proteins were then transferred to nitrocellulose and incubated in blocking buffer for 1 hour at room temperature. Membranes were probed with the following primary antibodies overnight at 4°C: rabbit anti-gamma H2A.X (1:1,000; Upstate, Charlottesville, VA, http://www.upstate.com); rabbit anti-p16 (1:200; Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com); mouse anti-cyclin D1 (1:500; Santa Cruz Biotechnology, Inc.); rabbit anti-phosphorylated Rb (1:2,000; Oncogene, San Diego, http://www.oncogene.com); mouse anti-p53 (1:200; Calbiochem, San Diego, http://www.emdbiosciences.com); or rabbit anti-cyclophilin A (1:5000; Upstate). After incubation in secondary antibody, immunopositive proteins were detected by autoradiography using ECL reagents (Amersham Pharmacia Biotech, Piscataway, NJ, http://www.amersham.com) and quantitated with densitometry where indicated (ImageJ, http://rsb.info.nih.gov/ij/).

Statistical Analysis

Results are expressed as the mean ± SD or mean ± SE. Statistical analysis was performed using GraphPad Prism (GraphPad Software Inc., San Diego, http://graphpad.com) and consisted of one-way analysis of variance (ANOVA) with Tukey-Kramer post hoc analysis for single-factor multiple group comparison, two-way ANOVA with Bonferroni post hoc analysis for double-factor multiple group comparison or Student’s t test for two-group comparison.

Results

BrdU Incorporation into NFC Results in Morphological and Phenotypic Changes in Subsequent Primary NS

NFC were plated at clonal density, treated with BrdU (50 µM), and propagated in non-adherent conditions until multicellular neurospheres were apparent [25]. In cultures not receiving BrdU treatment, NS appear as cell clusters with a smooth and well-defined border (Fig. 1A). NS derived from NFC exposed to BrdU, however, are substantially smaller than control NS and usually have an uneven border (Fig. 1B). BrdU incorporation at the time of the neurosphere assay was verified by immunolabeling 3 days after treatment (Fig. 1B, inset). To assess multipotency, control and BrdU-treated NS were attached to coated glass coverslips and stained for lineage markers. Cells derived from control NS appear to radiate from a distinct sphere of cells and consist of both β-III-tubulin⁺ and GFAP⁺ cells (Fig. 1C). In contrast, BrdU-treated NS do not remain as an attached cohesive sphere once plated, and cells derived from them adopt large, flat morphologies with little process extension. No typical β-III-tubulin or GFAP labeling is seen (Fig. 1D). Preformed NS were exposed to BrdU (50 µM) for 24 hours immediately before the induction of differentiation by plating on a nonadherent substrate, similar to typical labeling paradigms reported in the literature (BrdU labeling). These were compared directly to NS derived from NFC that had been exposed to a single dose of BrdU (50 µM) for 7 days during sphere formation (BrdU treatment as above). Cells from BrdU-labeling conditions incorporate BrdU (Fig. 2A) and retain the ability to extend processes and express β-III-tubulin and GFAP (Fig. 2B). Comparatively, some cells from BrdU treatment conditions also incorporate BrdU (Fig. 2C) but display no processes and poor phenotype marker expression (Fig. 2D). The lack of process maturation and altered expression of markers for mature neurons and astrocytes suggest aberrant progression of cell differentiation pathways.

Figure 1.

Treatment of neurosphere-forming cells with bromodeoxyuridine (BrdU) alters morphology and immunophenotype of neurospheres (NS) and their progeny. (A, B): Representative morphologies of control NS (A) or NS treated with 50 µM BrdU (B). Inset in (B) shows immunolabeling for BrdU incorporation (green, scale bar = 50 µm). (C, D): Control NS (C) differentiate into neurons (red, β-III-tubulin) and astrocytes (green, GFAP). NS treated with 50 µM BrdU (D) exhibit large-cell morphology, diminished process extension, and reduced expression of neural phenotype markers. Blue is 4,6-diamidino-2-phenylindole nuclear counterstain; scale bar in (B) = 50 µm and applies to all panels.

BrdU Incorporation into NFC Alters Growth Rates of Primary and Secondary NS

Our observations led us to quantify NS proliferation dynamics by measuring both the number and diameter of spheres derived from control and BrdU-treated NFC. These parameters reflect the number of surviving NFC and the number of cells within each NS, respectively. NFC were treated with a single pulse of BrdU (0, 1, 10, and 50 µM), plated at clonal density, and allowed to form NS for 7–10 days. Subsequent analysis reveals no consistent or significant differences in NS number among groups (Fig. 3A). However, there is a significant, dose-responsive decline in NS diameter (Fig. 3B). These data suggest that in vitro administration of BrdU does not inhibit NS formation but impairs the growth of resulting NS. To more directly test proliferative capacity, NFC were treated with 50 µM BrdU and serially passaged to assess long-term self-renewal of NFC [27]. Whereas control primary NS from neonatal brain can be propagated through at least two passages, treatment of neonatal NFC with BrdU abolishes the ability to form secondary NS (Fig. 3C). Similarly, NFC obtained from E14 animals were subjected to the same serial passaging protocol. Whereas rederivation of control NS is consistent through passage 5, BrdU results in a dose-dependent decline in NS number, with NS from the highest dosage experiencing total proliferative arrest (Fig. 3D).

Figure 3.

BrdU suppresses the growth rate of primary and secondary neurospheres. (A, B): In BrdU-treated (1, 10, or 50 µM) neurospheres (NS) generated from neonates, no consistent differences in number of NS obtained are observed (A); however, there is a significant dose-responsive reduction in NS diameter (B). *, p < .05; **, p < .01; ***, p < .001. (C, D): NS obtained from neonates or embryonic day 14 (E14) pups were subjected to serial passaging into secondary NS. (C): NS derived from neonate neurosphere-forming cells (NFC) and treated with 50 µM BrdU experience a sharp decrease in the number of secondary NS observed whereas control NS display a growth curve indicative of cell population proliferation. #, p < .05. (D): NS derived from E14 NFC treated with BrdU (1, 10, or 50 µM) display a dose-responsive decrease in generation of secondary NS compared with control NS, with all treated groups lagging behind control at all passages. One-way analysis of variance with Tukey-Kramer post hoc analysis was used for (A), (B), and (D); Student’s t-test was used for (C). Abbreviation: BrdU, Bromodeoxyuridine; P, passage number.

We next wanted to determine whether NFC from adult animals would behave similarly to those from the embryo or neonate. Data pooled from three separate experiments quantifying NS obtained from 12-week-old animals demonstrate a 50% reduction in NS number and a 40% reduction in diameter of NS (supplemental online Fig. 1A). In addition, these cells exhibit alterations in morphology and β-III-tubulin/GFAP expression as seen in neonatal cells (supplemental online Fig. 1B, 1C). Together, these data reveal an antiproliferative effect of BrdU on NFC from embryonic, postnatal, and adult animals.

To determine whether BrdU exerts a similar effect on a non-neurogenic stem cell type, we treated mesenchymal stem cells (MSC) isolated from adult murine bone marrow with a single dose of BrdU (0, 1, or 50 µM for 48 hours) and assessed their expansion rate. As with neural-derived cell types, MSC exhibit a dose-responsive reduction in expansion after exposure to BrdU (supplemental online Fig. 2). This result supports the conclusion that in vitro BrdU administration can exert an antiproliferative effect on multiple stem cell types.

BrdU Induces a Senescent Phenotype in Neurogenic Astrocyte Monolayers

It has been shown previously that adherent monolayers of neurogenic astrocytes can be obtained from the subventricular zone brain slurry used to generate NS [26]. Harvested cells that are grown in growth media in the presence of mitogens (bFGF and EGF) consist of a mixed population monolayer containing primarily astrocytes and microglia (Fig. 4A). Upon withdrawal of mitogens, rosettes of neuroblasts (Fig. 4B, 4C) form on the top of the monolayer and express β-III-tubulin (Fig. 4D, red). Because these cells are adherent and easy to work with, they therefore represent an expandable multipotent cell source that is amenable to multiple avenues of quantitative and molecular analysis. Monolayers were propagated for three passages to eliminate residual postmitotic neurons and were treated with a single 24-hour pulse of BrdU (0, 1, 10, or 50 µM). Treated cells demonstrate altered morphology, and by 48 hours after plating a subtle reduction in expansion of BrdU-treated progenitors is observed (Fig. 5A). At 144 and 264 hours after plating, a dose-responsive reduction in expansion is seen, with cultures exposed to 10 and 50 µM nearing growth arrest. This reduced expansion is accompanied by altered cell cycle dynamics. Over time (24–120 hours after BrdU), a larger proportion of BrdUtreated cells are found in the G₁ and G₂ phases of the cell cycle, with a concomitant decreased proportion of cells in the S phase (supplemental online Fig. 3).

Figure 4.

Neurogenic subventricular zone (SVZ) astrocyte monolayers function as multipotent neural progenitors through inducible neurogenesis. Confluent monolayers of SVZ astrocytes grown in the presence of serum and the growth factors epidermal growth factor and basic fibroblast growth factor (A) can be induced to generate large numbers of neuronal cells by serum and growth factor withdrawal (B–D). Within 24 hours of withdrawal segregated clusters of small, phase bright cells become apparent on the surface of the astrocyte layer (B). Higher magnification (C) contrasts the astrocytic morphology of the deeper layer with the more fusiform morphology of the cells within the clusters. Immunofluorescence labeling with an antibody against β-III-tubulin (red) demonstrates that these clusters represent neuronal cells with an immature morphology characterized by small, rounded somas and short processes (D). In this example, the astrocyte monolayer was derived from a transgenic mouse in which the promoter for the intermediate filament protein gene, nestin, drives green fluorescent protein expression (green). Scale bar (A–C) = 50 µm. Scale bar (D) = 10 µm.

Figure 5.

BrdU-induced decrease in neurogenic astrocyte expansion coincides with increased SA-β-Gal and ϒ-H2A.X expression. Neurogenic astrocytes were treated with a single 24-hour pulse of BrdU (0, 1, 10, or 50 µM). (A): At 48 hours postpulse, no significant difference in population expansion is detected. However, at 144 and 264 hours postpulse, all treatment conditions demonstrate a dose-responsive reduction in expansion. (B–D): Treated cells were analyzed for expression of SA-β-Gal (96 hours postpulse). Compared with controls (B), BrdU-treated cells (50 µM) display a larger percentage of more intensely labeled SA-β-Gal⁺ cells (C). (D): Quantification reveals a dose-dependent increase in the number of SA-β-Gal⁺ cells. One-way analysis of variance with Tukey-Kramer post hoc analysis: ***, p < .001. (E): Western blot analysis shows ϒ-H2A.X expression increases over time after a single 24-hour pulse of 50 µM BrdU. These results were quantified by densitometric analysis and graphed to reflect the ratio of ϒ-H2A.X over cyclophilin A expression (F). Scale bar = 50 µm. Abbreviations: Arb units, arbitrary units; BrdU, bromodeoxyuridine; Ctrl, control; Cyclo A, cyclophilin A; hr, hour; SA-β-Gal, senescence-associated β-galactosidase.

To test whether the slowing of proliferation is unique to BrdU or is due to its properties as a halogenated thymidine analog, we measured the growth of control cells and those treated with BrdU, IdU, CldU, or EdU (10 µM for 48 hours) (supplemental online Fig. 4). At 7 days after plating (4 days after analog removal), all treated cells display a significant and near-identical reduction in growth compared with control cells (all conditions p < .001 compared to control).

Because of the observed large-cell morphology and reduced expansion, we hypothesized that the cells are undergoing senescence. We first tested this hypothesis by assaying the cells for SA-β-Gal reactivity. SA-β-Gal is widely recognized as a biomarker of senescence both in vitro and in vivo. Combined with morphological analysis and molecular markers, it is a distinguishing characteristic of senescent cells versus quiescent or apoptotic cells (see [24] for review). Cells treated with a single 24-hour pulse of BrdU (0, 1, 10, or 50 µM) were stained for SA-β-Gal reactivity 96 hours postpulse. Compared with control cells (Fig. 5B), SA-β-Gal staining is qualitatively altered in BrdU-treated cells, appearing more intense and distributed over a larger area of the soma (Fig. 5C). Quantification reveals a dose-dependent increase in the number of SA-β-Gal⁺ cells (Fig. 5D) in BrdU-treated conditions. Next, protein levels of the senescence marker, ϒ-H2A.X, and proteins known to be regulated after the induction of senescence were quantified in control and BrdU-treated neurogenic astrocytes. BrdU at 50 µM was added for a single 24-hour pulse, and total protein was isolated 72, 96, and 120 hours later. Western blot analysis was normalized to the expression of cyclophilin A and quantified using densitometry. An accumulation of ϒ-H2A.X protein, a DNA damage marker that localizes to double-strand breaks and is associated with senescence, is seen in BrdU-treated cells (Fig. 5E, 5F). We assayed known signaling proteins active in both stress-induced (p16/pRb-dependent) and replicative senescent (p53-dependent) pathways—two well-described but incompletely understood cascades leading to senescence (Fig. 6A, 6B). At 96 hours postpulse, BrdU-treated astrocytes show decreased cyclin D1 and phosphorylated (active) pRb, and induction of p53 expression, all consistent with the activation of the described signaling cascades. Unexpectedly, p16 levels are not appreciable until 96 hours, even though its upregulation is typically associated with events upstream of Rb hypophosphorylation.

Figure 6.

BrdU alters expression of senescence-associated proteins. Neurogenic astrocytes were treated with a single 24-hour pulse of BrdU (0 or 50 µM) and probed for senescence-associated markers and cyclophilin A. (A): BrdU-treated cells exhibit decreased cyclin D1 levels and a coordinate decrease in hypophosphorylated Rb. Also seen is an increase in p53. Unexpectedly, p16 was not detected before 96 hours and is expressed at higher levels in controls than in BrdU-treated cells. (B): Densitometry was performed, and results were graphed to reflect the ratio of cell markers over cyclophilin A expression. Abbreviations: arb units, arbitrary units; BrdU, bromodeoxyuridine; Ctrl, control.

Cells often undergo programmed cell death after exposure to toxic agents, with senescent cells showing increased resistance to such apoptotic stimuli [23]. Compared with control cells, BrdU-treated cells (50 µM for 24 hours) experience a statistically insignificant increase in terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) (supplemental online Fig. 5, gray series). To determine whether BrdU induces apoptosis resistance, we also conducted TUNEL analysis on either control cells or BrdU-treated cells that were then exposed to camptothecin (CPT). Control cells and BrdU-treated cells were treated with 0 or 2 µM CPT and assessed for TUNEL 72 hours later. More than 20% of control cells treated with CPT are TUNEL⁺, whereas BrdU-treated cells treated with CPT show no increase in TUNEL⁺ cells (supplemental online Fig. 3, black series). Together, these results reveal the induction of senescence after BrdU administration.

In Vivo BrdU Administration Results in a Morphologically Mixed Population of NS

To determine whether in vivo BrdU administration alters the growth or differentiation potential of NFC progeny, we administered 50 or 100 mg/kg BrdU (50 or 100 mg/kg) or 0.9% normal saline to animals i.p. four times during a 24-hour period. Two hours after the final injection, animals were sacrificed, and tissue was harvested for culture of primary NS. NS were then plated, counted, and measured as in the in vitro treatment experiments. NFC obtained from BrdU-treated and control animals yield similar numbers of primary NS (Fig. 7A). When cultures are qualitatively considered, it is clear that a wide variation in NS diameter exists in BrdU-treated cultures (Fig. 7D) compared with control NS (Fig. 7C). Rather than inducing a population of uniformly smaller NS, the BrdU-treated groups contain both large spheres that resemble those of control cultures and irregular small spheres that resemble in vitro-treated NS. When spheres are classified on the basis of diameter, (small = 40–80 µm; medium = 80–120 µm; or large = >120 µm), it becomes clear that control animals yield the fewest small NS and the greatest number of large NS (Fig. 7B, light gray series). Animals treated with 50 µM BrdU yield more small NS, an intermediate number of medium NS, and fewer large NS (Fig. 7B, dark gray series). This effect is stronger in cultures derived from animals treated with 100 µM BrdU, yielding far more small and intermediate NS and a drastically reduced number of large NS (Fig. 7B, black series). These results confirm a dose-responsive effect on NS growth after in vivo BrdU administration.

Figure 7.

In vivo bromodeoxyuridine (BrdU) administration results in a morphologically mixed population of neurospheres (NS). C57/BL6 pups (postnatal day 21) were injected i.p. four times with 50 mg/kg BrdU, 100 mg/kg BrdU, or 0.9% sterile saline over 24 hours and sacrificed 2 hours after the final injection. (A): BrdU-treated animals yield numbers of NS similar to those of vehicle controls. (B): The quantitation of NS diameter is depicted by sorting NS into one of three categories: small (40–80 µm); intermediate (80–120 µm); or large (>120 µm). Control animals generate near equal proportions of each size of NS. In contrast, 50 mg/kg BrdU-treated animals generate more small-sized NS and fewer large NS. Finally, 100 mg/kg BrdU-treated animals yield almost no large NS, and higher numbers of small- and medium-sized NS. (C, D): Phase micrographs of representative NS derived from control (C) and in vivo BrdU-treated (D) animals. Scale bar = 50 µm.

Discussion

We have demonstrated that neural stem and progenitor cells at a variety of developmental stages exhibit a dramatic truncation of proliferative capacity after single-pulse BrdU administration. In vitro, BrdU-treated NFC produce smaller primary neurospheres that also show an impaired capacity for generating secondary neurospheres. Similarly, neurogenic astrocytes exhibit reduced proliferative capacity, upregulation of senescence markers such as SA-β-Gal and ϒ-H2A.X, and resistance to apoptosis. This cluster of signs is indicative of cellular senescence (see [23] for review), and treated cells also show activation of downstream proteins involved in canonical senescence pathways, such as p53 and Rb. We show that these results are applicable to both neural- and bone marrow-derived stem cell populations. These results have profound implications for the interpretation of experimental outcomes based on BrdU, or similar halogenated pyrimidine incorporation and may limit downstream functional assessment of labeled stem/progenitor cells and their progeny as incorporating cells may have radically altered biology. Likewise, quantitative long-term in vitro and in vivo labeling paradigms may be expected to underestimate stem/progenitor cells and their progeny.

In vivo administration of BrdU leads to a dose-responsive reduction in the diameter of primary NS subsequently obtained from the neural stem cell niche. Because in vitro BrdU exposure leads to smaller NS, it is not surprising that a similar result would be obtained by pretreating animals with BrdU. The fact that we see a dramatic effect in the neurosphere assay, a surrogate test of stem/progenitor cell presence, suggests that relatively modest BrdU regimens may be a viable method of inducing senescence, either in experimental paradigms of agingassociated changes in neural stem cell function and neurogenesis or in the clinical setting as a potential antineoplastic approach.

Our present data support recent findings suggesting perturbed proliferation after BrdU exposure. Michishita et al. [18] showed that BrdU leads to inconsistent responses of important cell cycle regulatory proteins known to affect senescence and cancer pathways. Slowed proliferation resulting from BrdU incorporation has been described in a number of cancer cell lines [30], as well as in thymidine auxotrophic yeast [31], suggesting that the effect may be universal among eukaryotic cells. We have also found that all tested primary and cancer cell lines are susceptible to BrdU-induced senescence, regardless of species, telomerase activity, or status of cell cycle proteins such as p16, p21, and p53 [32].

Although BrdU is now most frequently used for birthdating and tracking proliferative cells, it was initially introduced as a mutagen to target rapidly dividing cancer cells [33–35]. A number of early reports attested to potentially negative consequences resulting from BrdU incorporation [36–39]. More recently, it has been shown [17] that BrdU may not affect all cells in the same way and is selectively toxic for neural progenitor cells at doses as low as 1 µM, which is well below the dose typically used in in vitro labeling paradigms. However, because most incorporating cells seem to maintain normal function [40], at least in the short term and in the absence of secondary stressors, BrdU is generally regarded as a benign substitute for thymidine.

Given the long history and ubiquitous use of BrdU, why is its link with senescence only now being recognized? Most likely this newly appreciated effect of BrdU administration is due to the initially subtle, but progressive nature of senescence in BrdU-incorporating cells. Multiple rounds of replication are required for dramatic effects on the proliferation rate to become manifest, given the tools typically used to assess normal cellular function. Thus, quantification of BrdU-treated cells shortly after exposure will not reveal the large divergence from normal control cells that is seen with longer postincorporation intervals by proliferative cells. Adult hippocampal neurogenesis, for instance, in which newly generated granule neurons do not continue dividing after genesis in the subgranular zone, would not be expected to show overt perturbation as a result of BrdU administration. However, such neurogenesis may eventually become impaired over time if the cell cycle kinetics of enough stem/progenitor cells are negatively affected by BrdU incorporation.

Our data also show that, in addition to its effect on the cell cycle, BrdU can perturb the normal differentiation of cells derived from neural stem cells. At first glance this result seems anomalous, given the large body of work using BrdU to label, for instance, newly generated neurons in neurosphere culture models. However, most antecedent studies applied BrdU at the time of neurosphere differentiation, whereas in the present study we exposed the neurosphere-forming cell to a single pulse of BrdU immediately after isolation from the primary tissue. In the latter approach, multiple rounds of replication occur before the cells are assessed for differentiation changes, again demonstrating that the most dramatic effects of BrdU on biological function are delayed substantially from the time of actual exposure. Thus, in short “chase” paradigms exemplified by our BrdU-labeling condition or in cells that do not divide frequently after exposure, BrdU would not be expected to substantially perturb differentiation.

Although there is strong evidence for senescence induction by BrdU, the mechanism of this induction remains enigmatic. Our analyses indicate that, at least in primary neural stem/progenitor cells, activation of p53 and proteins associated with the Rb pathway is a downstream consequence of BrdU incorporation; however, further work is required to temporally and sequentially identify these events. Prior microarray studies have identified a number of senescence-associated genes that are upregulated in response to BrdU administration [19], but their potential causative role has not been established. More recently, chromatin unpacking, regulated by BrdU incorporation into scaffold/nuclear matrix attachment region sequences, has been proposed as an initiating event for senescence induction [41, 42].

We conclude that BrdU incorporation leads to a delayed but progressive induction of senescence in neural stem/progenitor cells that manifests over multiple rounds of replication and is accompanied by perturbed differentiation of neural progeny. This effect is likely to be common to all stem cell pools and points out the need for caution when results based on long-term BrdU tracking over multiple rounds of replication are interpreted. Interestingly, the susceptibility to BrdU-induced senescence/growth perturbation appears to be more marked in aged animals, and future study is required to understand this relationship. The reliable induction of senescence in stem/progenitor cells in vitro and in vivo may yield a novel platform for molecular studies designed to address multiple aspects of aging and may also represent a therapeutic approach to slow the growth of cancer cells.