TISSUE-SPECIFIC STEM CELLS Noninvasive and Quantitative Monitoring of Adult Neuronal Stem Cell Migration in Mouse Brain Using Bioluminescence Imaging VEERLE REUMERS,a,d CHRISTOPHE M. DEROOSE,b,c,d OLGA KRYLYSHKINA,b JOHAN NUYTS,c,d MARTINE GERAERTS,b LUC MORTELMANS,c,d RIK GIJSBERS,b,d CHRIS VAN DEN HAUTE,a,d ZEGER DEBYSER,b,d VEERLE BAEKELANDTa,d Laboratories for aNeurobiology and Gene Therapy and bMolecular Virology and Gene Therapy, Division of Molecular Medicine, and cDivision of Nuclear Medicine, dMolecular Small Animal Imaging Center (MoSAIC), Katholieke Universiteit Leuven, Leuven, Flanders, Belgium Key Words. In vivo optical imaging • Neural stem cells • Gene transfer • Lentiviral vectors • Growth factors

ABSTRACT It is now generally accepted that continuous neurogenesis occurs in the adult mammalian brain, including that of humans. Modulation of adult neurogenesis can provide therapeutic benefits for various brain disorders, including stroke and Parkinson’s disease. The subventricular zone-olfactory bulb pathway is one of the preferred model systems by which to study neural stem cell proliferation, migration, and differentiation in adult rodent brain. Research on adult neurogenesis would greatly benefit from reliable methods for long-term noninvasive in vivo monitoring. We have used lentiviral vectors encoding firefly luciferase to stably mark endogenous neural stem cells

in the mouse subventricular zone. We show that bioluminescence imaging (BLI) allows quantitative follow-up of the migration of adult neural stem cells into the olfactory bulb in time. Moreover, we propose a model to fit the kinetic data that allows estimation of migration and survival times of the neural stem cells using in vivo BLI. Long-term expression of brain-derived neurotrophic factor in the subventricular zone attenuated neurogenesis, as detected by histology and BLI. In vivo monitoring of the impact of drugs or genes on adult neurogenesis is now within reach. STEM CELLS 2008;26:2382–2390

Disclosure of potential conflicts of interest is found at the end of this article.

INTRODUCTION In the adult rodent brain, neurogenesis occurs mainly in two different areas: the subgranular layer of the dentate gyrus [1] and the subventricular zone (SVZ) [2] of the lateral ventricle. In the SVZ, slowly proliferating astroglial-like type B cells generate actively proliferating type C cells, which in turn give rise to migrating neuroblasts or type A cells [3] (Fig. 1). These type A cells migrate tangentially in chains through tubular structures toward the olfactory bulb (OB), forming a well-defined migratory pathway called the rostral migratory stream (RMS). Upon arrival in the OB, the progenitor cells differentiate in functional interneurons and integrate into existing neural pathways [4]. The turnover of neurons in the OB is believed to play a role in olfactory memory formation [5]. In the human brain, the analogous migration of neuroblasts to the olfactory bulb via a lateral ventricular extension was only recently demonstrated [6], albeit without a clear functional role. However, it is known that the proliferation of progenitor cells in the SVZ and neuroblast migration are reduced in Parkinson’s disease (PD) animal models and in PD patients [7, 8]. Together, the data suggest that

adult neuronal stem cell proliferation and migration toward the OB play a physiological role in humans. It also follows that the rodent brain provides a relevant model by which to study migration, proliferation, and differentiation of the adult neuronal stem cell population. The rodent model has already been used to study the effects of different factors on adult neurogenesis. Infusion [9, 10] or transient overexpression [11] of brain-derived neurotrophic factor (BDNF) in rat SVZ, for example, was shown to increase the formation of new neurons in the OB and other brain regions. However, much less is known about the long-term impact of BDNF on adult neurogenesis. Noninvasive imaging of stem cell migration in rodent brain would facilitate this research considerably. We have shown before that lentiviral vectors (LVs) can be used for efficient and long-term in vivo marking of adult neuronal stem cells in rodent SVZ [12]. This enables histological detection of neural progenitor cells during proliferation, migration, and differentiation. Noninvasive imaging of reporter genes encoded by LVs would allow direct monitoring of stem cell migration in an individual animal over time. Bioluminescence imaging (BLI) has so far been used to follow the migration of exogenous, ex vivo-labeled stem cells toward glioblastomas [13,

Author contributions: V.R. and C.M.D.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; O.K.: collection and/or assembly of data, data analysis and interpretation, final approval of manuscript; J.N.: data analysis and interpretation, final approval of manuscript; M.G.: collection and/or assembly of data; L.M.: financial support; R.G. and C.V.D.H.: collection and/or assembly of data, data analysis and interpretation; Z.D. and V.B.: conception and design, financial support, administrative support, data analysis and interpretation, manuscript writing, final approval of manuscript; Z.D. and V.B. contributed equally to this work. Correspondence: Veerle Baekelandt, Ph.D., Kapucijnenvoer 33, B-3000 Leuven, Belgium. Telephone: 32-16-33-36-32; Fax: 32-16-33-63-36; e-mail: [email protected] Received December 13, 2007; accepted for publication June 11, 2008; first published online in STEM CELLS EXPRESS July 3, 2008. ©AlphaMed Press 1066-5099/2008/$30.00/0 doi: 10.1634/stemcells.2007-1062

STEM CELLS 2008;26:2382–2390 www.StemCells.com

Reumers, Deroose, Krylyshkina et al.

2383

RNA copies per milliliter). In vivo BLI imaging of gene transfer in rodent brain using eGFP-IRES-Fluc vector was previously validated [18].

Stereotactic Surgery

Figure 1. Adult rodent neurogenesis. A sagittal view on the rodent brain showing the two main sites of neurogenesis: the DG of the hippocampus and the SVZ of the LV. Neural progenitor cells from the SVZ migrate through the RMS and differentiate into interneurons in the OB. The inset shows the structural organization of the SVZ. Slowly proliferating astroglial-like type B cells generate actively proliferating type C cells, which in turn give rise to migrating neuroblasts or type A cells. These different cell populations can be distinguished on the basis of several immunohistochemical markers: type B cells are positive for GFAP, type C cells and type A cells are dcx-positive, and mature interneurons are NeuN-positive. Abbreviations: CC, corpus callosum; dcx, doublecortin; DG, dentate gyrus; E, ependymal cells; GFAP, glial fibrillary acidic protein; LV, lateral ventricle; NeuN, neuronal nuclear antigen; OB, olfactory bulb; RMS, rostral migratory stream; SVZ, subventricular zone.

14] and ischemic infarcts [15, 16]. Considerably less attention has been paid to detecting the fate of the different types of endogenous stem cells present in adult organisms. Nevertheless, recent evidence suggests that beneficial effects of exogenous stem cell therapy might be mediated by the endogenous stem cells in the brain [17]. We have previously validated BLI for efficient noninvasive imaging of LV-mediated gene transfer in mouse brain [18] using firefly luciferase as the reporter. Here, we have evaluated the feasibility of detecting and quantifying migration of the progeny of endogenous neuronal progenitor cells with BLI. To the best of our knowledge, this is the first report using whole-body imaging to monitor the migration of endogenous stem cells in adult animals.

MATERIALS

AND

METHODS

Lentiviral Vector Construction and Production We constructed a bicistronic lentiviral transfer plasmid encoding enhanced green fluorescent protein (eGFP) and firefly luciferase (Fluc) linked by an encephalomyocarditis virus (EMCV) internal ribosome entry sequence (IRES) [19] as previously described [18]. Highly concentrated human immunodeficiency virus type 1 (HIV1)-derived lentiviral vectors were produced by a standardized and upscaled three-plasmid transient transfection protocol [20]. Briefly, 293T cells were transfected with a second-generation packaging plasmid, a plasmid encoding the glycoprotein G of vesicular stomatitis virus (VSV-G), and the transfer plasmid encoding the genes of interest. The cell culture medium was harvested, filtered, and concentrated by tangential flow filtration. For gene transfer into rodent brain, vectors were further concentrated by ultracentrifugation. Quality control and quantification of functional vector particles was performed by three independent methods: titration by transduction of 293T cells with serially diluted vector and flow cytometry of transduced cells (unit: transducing units [TU]/ml); determination of the concentration of p24 antigen, an HIV-1 capsid protein, by enzyme-linked immunosorbent assay (HIV-1 p24 enzyme-linked immunosorbent assay kit; PerkinElmer Life and Analytical Sciences, Milan, Italy, http://www.perkinelmer.com) (unit: ng of p24 per milliliter); and quantitative reverse transcriptase polymerase chain reaction to determine the number of RNA molecules (unit:

www.StemCells.com

Adult female C57BL/6 and C57BL/6-Tyrc-2J/J mice were used. The animals were housed under a 14-hour-light/10-hour-dark cycle with free access to food and water. All animal experiments were approved by the bioethical committee of the Katholieke Universiteit Leuven. After anesthesia, the animals were placed in a stereotactic head frame (Stoelting, Wood Dale, IL, http://www.stoeltingco. com). A midline incision of the skin was made, and a small hole was drilled in the skull at the appropriate location, using the bregma as a reference. The coordinates used were as follows: SVZ, anteroposterior 0.5 mm, lateral 1.5 mm, and dorsoventral 3.0 –2.0 mm; OB, anteroposterior 4.7 mm, lateral 1.0 mm, and dorsoventral 1.5 mm. Two microliters (in the SVZ) or 1 ␮l (in the OB) of highly concentrated vector supplemented with polybrene (4 ␮g/ml) was injected at a rate of 0.25 ␮l/minute with a 30-gauge needle on a 10-␮l Hamilton syringe. After 4 minutes of injection (1 ␮l), the needle was raised slowly in the dorsal direction over the distance indicated by the dorsoventral coordinates for the SVZ. After the injection, the needle was left in place for an additional 5 minutes to allow diffusion before being slowly withdrawn from the brain. The animals were kept in individually ventilated cages after surgery.

In Vivo Bioluminescence Imaging The mice were imaged in an IVIS 100 system (Caliper Life Sciences, Hopkinton, MA, http://www.caliperls.com). Anesthesia was induced in an induction chamber with 2% isoflurane in 100% oxygen at a flow rate of 1 l/minute and maintained in the IVIS with a 2% mixture at 0.5 l/minute. Before every imaging session, the mice were injected in a lateral tail vein with 126 mg/kg D-luciferin (Xenogen) dissolved in phosphate-buffered saline (PBS) (15 mg/ ml). Subsequently, they were placed in prone position in the IVIS, and three consecutive 1-minute frames were acquired with a field of view of 10 cm. Each frame depicted the measurement of the bioluminescent signal as a pseudocolor image superimposed on a gray-scale photographic image. For quantification of the photon flux (photons per second), all imaging data were corrected for scatter, on the basis of an empirical gaussian model (supplemental online Fig. 1). Animals with a BLI signal from the SVZ smaller than 4.0 ⫻ 105 photons per second at day 7 were excluded from the study. Histological analyses confirmed that these animals were stereotactically injected in the ventricle instead of in the SVZ.

Ex Vivo Brain Bioluminescence Imaging Mice were placed in the IVIS and injected with D-luciferin as described above, and two consecutive scans of 1 minute were acquired. Immediately afterward, the mice were sacrificed by cervical dislocation and decapitated, and the brain was dissected. The OBs were dissected separately and imaged for 1 minute. The brain was placed in an acrylic brain matrix (Harvard Apparatus, Holliston, MA, http://www.harvardapparatus.com) and sliced in 1.0-mmthick coronal sections. These were also imaged for 1 minute in the IVIS. The unit of the scale of the bioluminescent images is expressed in photons per second per cm2 per steradian. The OBs were fixed overnight in 4% paraformaldehyde in PBS and subsequently analyzed as described below.

Histology and Stereological Counting To assess lentiviral vector transduction histologically, the mice were deeply anesthetized with pentobarbital (Nembutal; CEVA Sante´ Animale, Libourne, France, http://www.ceva.com) and perfused transcardially with saline followed by ice-cold 4% paraformaldehyde in PBS for 15 minutes. The brain was removed from the skull and postfixed overnight in the same fixing solution. Coronal or sagittal brain sections (50 ␮m thick) were cut with a vibratome and stored at 4°C in PBS buffer containing 0.1% sodium azide. A polyclonal antibody against eGFP (made in-house by immunization

Bioluminescence Imaging of Neuronal Stem Cells

2384

with purified recombinant eGFP [21]) was used for immunohistochemistry. The sections were treated with 3% hydrogen peroxide and incubated overnight with the primary rabbit anti-eGFP antibody (diluted 1:10,000) in 10% normal swine serum and 0.1% Triton X-100. The sections were then incubated in biotinylated swine anti-rabbit secondary antibody (diluted 1:300; Dako, Glostrup, Denmark, http://www.dako.com), followed by an incubation with streptavidin-ABC-horseradish peroxidase complex (Dako). Detection was with diaminobenzidine, using H2O2 as a substrate. The number of eGFP-positive cells in the OB was estimated by a random sampling stereological counting method, the optical fractionator [22], using the Stereo Investigator software (MicroBrightField Inc., Magdenburg, Germany, http://www.mbfbioscience.com) and a Leica DMR Biopoint 2 microscope (Leica Microsystems, Wetzlar, Germany, http://www.leica.com). For each animal, eight sections at 300-␮m intervals throughout the OB were used. Double and triple immunofluorescent stainings were done overnight at room temperature in PBS/10% horse serum (vol/vol)/0.1% Triton (vol/vol) with the following antibodies: goat anti-doublecortin for immature neurons (diluted 1:200; Santa Cruz Biotechnology, Heidelberg, Germany, http://www.scbt.com), mouse anti-neuronal nuclear antigen (anti-NeuN) for mature neurons (diluted 1:100; Chemicon, Temecula, CA, http://www.chemicon.com), mouse antiglial fibrillary acidic protein (anti-GFAP) for astroglial cells and type B stem cells (diluted 1:200; BD Biosciences Pharmingen, Erembodegem, Belgium, http://www.bdbiosciences.com), rabbit S-100␤ for mature glial cells (diluted 1:2,500; Swant, Bellinzona, Switzerland, http://www.swant.com), and chicken anti-eGFP (diluted 1:1,000; Aves Labs, Tigard, OR, http://www.aveslab.com). The next day, sections were incubated with the appropriate mixture of the following fluorescently labeled secondary antibodies at room temperature for 2 hours: donkey anti-mouse Alexa 555 (diluted 1:500; Molecular Probes, Eugene, OR, http://probes.invitrogen. com); donkey anti-goat Alexa 555 (diluted 1:500; Molecular Probes); donkey anti-chicken fluorescein isothiocyanate (diluted 1:400; Jackson Immunoresearch Laboratories, West Grove, PA, http://www.jacksonimmuno.com), and donkey anti-rabbit Alexa 647 (diluted 1:500; Molecular Probes). All secondary antibodies were adsorbed against several species to reduce cross-reactivity. Sections were mounted on glass slides, coverslipped with Moviol (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), and analyzed by laser scanning confocal microscopy (ConfoCor 2; Carl Zeiss, Oberkochen, Germany, http://www.zeiss.com).

Statistical Analysis and Curve Fitting Data were imported in SPSS (SPSS, Chicago, http://www.spss. com). Linear regression analysis was performed and the Pearson’s correlation coefficient was determined for the in vivo BLI, ex vivo BLI, and olfactory bulb cell number data. For comparison of different time points, one-way analysis of variance (ANOVA) was used with a post hoc Dunnett’s test. For comparison of treatment conditions one-way ANOVA was used with a post hoc Bonferroni corrected t test. Significance was determined at p ⬍ .05. A simple kinetic model is proposed (supplemental online Fig. 2a) to describe the accumulation of labeled stem cells in the OB. The model has three kinetic parameters and one parameter representing the number of labeled stem cells in the OB. A simultaneous fit to all available data was applied to determine the three kinetic parameters. Graphs are shown with SE bars.

RESULTS

Figure 2. In vivo long-term follow-up of stem cell migration with bioluminescence imaging (BLI). (A): At day 7 post-injection (pi) there was a detectable BLI signal only at the site of injection (SVZ). At 15 and 30 weeks pi an additional focus was detected at the OB projection site, as well as the original focus at the SVZ. A representative mouse is shown. (B): The graph shows the quantification of the in vivo OB BLI signal from all the animals followed for 30 weeks (n ⫽ 14). The OB photon flux at weeks 15 and 30 was significantly higher than that on day 7 (p ⫽ .002 and .045). Abbreviations: OB, olfactory bulb; p, photons; s, second; sr, steradian; SVZ, subventricular zone.

Fig. 3). Next, we injected 2 ␮l of a LV encoding eGFP and Fluc linked by an EMCV IRES sequence (LV-eGFP-IRESFluc) (8.0 ⫻ 104 TU, 180 ng of p24, 6.6 ⫻ 108 RNA copies) in the SVZ of 14 C57BL/6 mice at 8 weeks old. The animals were imaged at 7 days, 15 weeks, and 30 weeks post-injection (pi). At day 7 pi we observed a unifocal signal originating from the site of injection, which corresponds to the SVZ as previously determined [18]. When the animals were scanned at 15 weeks pi a distinct second focus, separate from the original SVZ focus, appeared between the eyes of the animal, exactly at the site where the signal from cells transduced in the OB projected. This signal from the right OB persisted at 30 weeks pi (Fig. 2A). Next, we attempted to quantify the signal originating from the OB. Since some of the photons emitted from the SVZ are scattered and project to the OB, we performed a scatter correction on the raw imaging data (supplemental online Fig. 1). After correcting for scatter, the photon flux (photons per second) originating from the OB was calculated. The absolute OB signal at weeks 15 and 30 was significantly higher than that at day 7 (p ⫽ .002 and .045) (Fig. 2B). We conclude that neuroblasts migrating from the SVZ to the OB can be detected by noninvasive in vivo BLI. Moreover, the BLI signal originating from the OB can be quantified after a correction for scatter.

Bioluminescence Imaging of In Vivo Stem Cell Migration

Ex Vivo Data Correlate with In Vivo-Acquired BLI Data

The aim of our study was to investigate whether the progeny of neural progenitor cells labeled with LV encoding Fluc could be detected and quantified in the OB using BLI. Therefore, we first determined that the BLI signal emitted from cells transduced in the OB is located between the eyes of the animal, slightly lateral from the midline (supplemental online

To validate our in vivo measurements, we performed an ex vivo analysis. For this purpose, we injected the same LVeGFP-IRES-Fluc in the SVZ of 20 C57BL/6 mice. Small groups of animals (n ⫽ 3–5) were sacrificed at days 3, 14, and 60 and weeks 15 and 30 pi. In vivo BLI was measured at 15 and 30 weeks pi. Since C57BL/6 mice develop hyperpig-

Reumers, Deroose, Krylyshkina et al. mentation of the skin after shaving, in vivo BLI at intermediate time points was technically not feasible (data not shown). We performed both ex vivo BLI and histology on the OBs of each animal. The ex vivo BLI images of the OB showed unilateral Fluc activity that increased over time (Fig. 3A). After the ex vivo BLI measurements, both right and left OBs were processed for histological analysis. Immunohistochemical staining confirmed the presence of eGFP-positive cells in the right OB (Fig. 3B). We then determined the number of eGFP-positive cells in each OB by stereological quantification. The numbers of positive cells at day 60 and at the later time points were significantly higher than at day 3 (p ⱕ .003). Both the ex vivo BLI signal and the number of eGFP-positive cells in the OB were in agreement with the proposed kinetic model (described below). Finally, we correlated the number of eGFP-positive cells in each individual animal with the values for in vivo and ex vivo BLI measured for the same animal (Fig. 3C). A strong linear correlation between the ex vivo OB flux and the number of eGFP-positive cells in the OB was obtained (p ⬍ .0001; R2 ⫽ 0.72; n ⫽ 20). More importantly, we observed an even stronger linear correlation between the in vivo OB photon flux and the number of eGFP-positive cells in the OB (p ⫽ .001; R2 ⫽ 0.85; n ⫽ 8). These strong correlations show that BLI allows appropriate quantification of the number of marked cells in the OB. We conclude that BLI of LV-marked neural stem cells allows qualitative and quantitative detection of the migration of endogenous neural stem cells toward the OB in live mice.

Characterization of eGFP-Positive Cells in the SVZ and the OB We identified the different cell types labeled with eGFP by performing double or triple immunofluorescent stainings and subsequent confocal analysis. In the SVZ, we detected eGFP-labeled type B stem cells since they were positive for GFAP, a marker for type B stem cells and glial cells, but negative for S100␤, a marker for mature glial cells (Fig. 4A). As expected for VSV-G pseudotyped lentiviral vectors, we also detected eGFP-labeled glial cells that costained for GFAP and S100␤ (Fig. 4B) and differentiated neurons (data not shown) [23]. In the OB, eGFP-positive cells colabeled with doublecortin, a marker for neural progenitor cells (Fig. 4C). Moreover, the eGFP-labeled cells arriving in the OB matured into neuronal cells, as evidenced by costaining with NeuN, a marker for adult neurons (Fig. 4D). These data are in complete agreement with a more detailed stem cell marker analysis of eGFP-labeled cells after LV injection in the SVZ previously reported by our group [12].

Short-Term Kinetics of Stem Cell Migration Revealed by BLI in Albino C57BL/6 Mice As mentioned above, hyperpigmentation of the skin of C57BL/6 mice interferes with repeated BLI measurements at early time points after stem cell labeling by LV. This jeopardizes the kinetic study of early effects of biological factors on stem cell migration. We reasoned that the use of a white furred or nude mouse strain would circumvent this limitation. Since different strains of mice appear to have differences in neurogenic potential [24], we decided to use C57BL/6-Tyrc-2J/J mice [25], which have the same genetic background but a mutation in the tyrosinase gene, rendering them albino. These mice indeed do not develop hyperpigmentation and allow measurement of BLI signal at a greater number of earlier time points. We injected 7 C57BL/6-Tyrc-2J/J mice and scanned the animals at 1, 2, 4, 8, 12, 15, 30, and 45 weeks after injection. We could detect a signal originating from the OB as early as 4 weeks pi, and this signal became more pronounced at 8 weeks. From 8 weeks pi www.StemCells.com

2385

on, the BLI signal from the OB was always significantly higher than that at 2 weeks pi (p ⱕ .019) (Fig. 5). In addition, the kinetics of the in vivo BLI signal was studied. Therefore we normalized the BLI data by dividing the photon flux of the OB by the photon flux of the SVZ, after scatter correction. Assuming that the number of labeled cells in the SVZ is constant, this normalization corrects for differential reporter gene expression, such as that due to transgene silencing. As previously described [18], the BLI signal originating from the site of injection decreases during the first weeks after surgery and thereafter remains constant. Different data sets were pooled: ex vivo BLI measurements of the OB (Fig. 3A), histology data (Fig. 3B; supplemental online Fig. 2b), and in vivo BLI of individual albino mice (Fig. 6). The ex vivo BLI and histology data are presented as three different curves, representing the average of the groups, whereas the in vivo BLI data are presented as seven plots, one for each individual animal followed over time. We propose a parametric model to describe the observed kinetics. We assume that per unit of time, a constant number of neuroblasts (A) start to migrate toward the OB. The cells arrive t1 days later in the OB (where they contribute to the BLI signal of the OB). They all live at least until day t2. The time (t2 ⫺ t1) represents the time needed for differentiation and integration in the OB neural pathways. Starting at t2, there is a constant probability ␰ per unit of time that a particular labeled cell dies. The accumulated BLI signal in the OB can be defined as the integral of this model over time. The mathematical formulas and typical graphs are shown in supplemental online Figure 2a. In the combined weighted least-squares fitting procedure, each curve was assigned its own value for A, whereas the values of t1, t2, and ␰ were constant for all curves (Figs. 3A, 3B, 6; supplemental online Fig. 2b). The best fit yielded a very short migration time (less than a day), a relatively short guaranteed survival time (1–2 days), and a half-life of approximately 1 month.

BLI Detects the Negative Impact of Long-Term BDNF Overexpression on Adult Neurogenesis Modulation of adult neurogenesis by externally supplied neurotrophins, cytokines, or growth factors is a topic of intense investigation. Short-term (1 month or less) stimulation of neurogenesis by BDNF has previously been demonstrated [9 –11]. We confirmed this effect by LV marking of the SVZ (supplemental online Fig. 4a). To investigate the effect of long-term expression of BDNF on neurogenesis in the SVZ using in vivo BLI, we injected five C57BL/6 mice with 3 ␮l of a mixture of LV-eGFP-IRES-Fluc and LV-BDNF (2.6 ⫻ 109 p24 per milliliter; 3.16 ⫻ 1011 RNA copies per milliliter) in a 2:1 ratio. Functionality of LV-BDNF was confirmed by Western blotting after transduction of 293T cells and by immunohistochemistry after stereotactic injection of LV in the brain of C57BL/6 mice (supplemental online Fig. 5). As a control group, we injected six C57BL/6 mice with 3 ␮l of a mixture of LV-eGFP-IRES-Fluc and LV-LacZ (1.7 ⫻ 106 p24 per milliliter; 8.14 ⫻ 1010 RNA copies per milliliter) in a 2:1 ratio. Functionality of LV-LacZ was confirmed by a luminometric ␤-galactosidase assay of transduced 293T cells (data not shown). These experiments were performed in C57BL/6 mice, because the albino C57BL/6-Tyrc-2J/J mice were not yet available in the laboratory at the start of the study. Surprisingly, at 15 and 30 weeks pi we detected a strong decrease (78% and 85%, respectively; p ⫽ .010 and .001) in BLI signal from the OB in the BDNF group in comparison with the control group (Fig. 7A, 7B). In addition, a similar reduction of BLI signal in the SVZ was evidenced, suggesting that long-term overexpression of BDNF directly affects the cells in the SVZ and not only the subsequent migration and differentiation process. In comparison with the control group, there was a decrease in BLI signal of 86% and 91% at 15 and 30 weeks, respectively (p ⫽ .003 and p ⫽ .001) (Fig. 7C). The average SVZ

2386

Bioluminescence Imaging of Neuronal Stem Cells

Figure 3. Ex vivo analysis and correlation with in vivo bioluminescence imaging (BLI) signal in the OB. (A): BLI images of the L and R OB of a representative animal for each time point are shown. The graph shows the quantification of the ex vivo BLI signal of the OB and the fitted kinetic model. (B): Stereological quantification of the number of eGFP-positive cells in the R OB. The graph shows the measured data points and the fitted model. The number of cells from day 60 on was significantly higher than that at day 3 (p ⱕ .003). Sections of a representative animal for each time point are shown. Scale bar ⫽ 950 ␮m. (C): Highly significant linear correlations were observed between the number of eGFP-positive cells in the OB and both the ex vivo OB photon flux (p ⬍ .0001; R2 ⫽ 0.72; n ⫽ 20) and the in vivo OB p flux (p ⫽ .001; R2 ⫽ 0.85; n ⫽ 8). Abbreviations: eGFP, enhanced green fluorescent protein; L, left; OB, olfactory bulb; p, photons; R, right; s, second; sr, steradian.

photon flux at 7 days pi in the BDNF group (5.1 ⫻ 106 ⫾ 3.0 ⫻ 106 photons per second) was not significantly different from the

photon flux in the control group (8.2 ⫻ 106 ⫾ 2.8 ⫻ 106 photons per second), excluding differences in transduction efficiency. All

Reumers, Deroose, Krylyshkina et al.

2387

Figure 4. Confocal analysis with stem cell markers in the subventricular zone (SVZ) and the olfactory bulb (OB). (A, B): Triple immunofluorescent staining with anti-eGFP (green), anti-GFAP (red), and anti-S100␤ (blue) on coronal sections of the SVZ. (A): The horizontal arrows show an eGFP-positive stem cell that costains with GFAP, a marker for type B stem cells and glial cells, but is negative for S100␤, which is a marker for mature glial cells. The vertical arrows show a glial cell that is GFAP- and S100␤positive. (B): The horizontal arrows show an eGFP-positive glial cell that costains with GFAP and S100␤. The vertical arrows show an untransduced glial cell that is GFAP- and S100␤-positive. (C): Double immunofluorescent staining with anti-eGFP (green) and anti-dcx (red) on coronal sections of the OB. eGFP-positive cells colabel with dcx, a marker for progenitor cells. (D): Double immunofluorescent staining with anti-eGFP (green) and anti-NeuN (red) on coronal sections of the OB. eGFP-labeled cells arriving in the OB matured into neuronal cells, as evidenced by costaining with NeuN, a marker for adult neurons. Scale bars ⫽ 20 ␮m. Abbreviations: dcx, doublecortin; eGFP, enhanced green fluorescent protein; GFAP, glial fibrillary acidic protein; NeuN, neuronal nuclear antigen.

animals were sacrificed at 30 weeks pi, and their brains were processed for histology (Fig. 7D). Stereological quantification of the number of eGFP-positive cells in the labeled OB showed a 70% decrease (p ⫽ .0003) in the BDNF group in comparison with the control group (Fig. 7C). In a separate group of animals injected with a combination of LV-eGFP and LV-BDNF, we confirmed by histological analysis a significant decline in the number of labeled neurons in the OB of BDNF-overexpressing animals from 2 months on compared with the control group (supplemental online Fig. 4b, 4c). These findings demonstrate the importance of investigating long-term effects of growth factors with therapeutic potential. We demonstrate that in vivo imaging is a powerful tool to facilitate such studies. Our data also indicate that the mechanism of action of BDNF in vivo needs careful investigation.

DISCUSSION We report the first successful noninvasive and quantitative imaging strategy to monitor the dynamic behavior of adult neuronal stem cells in mouse brain. These results extend our previous work [12] showing that lentiviral vectors outperform other methods for stable and efficient gene marking of stem cells in the mouse brain. We previously corroborated LV-mediated transduction of the true neuronal stem cell by the following findings: (a) eGFP-expressing cells in the SVZ colabel with www.StemCells.com

GFAP (a marker for the type B-stem cell but also for astrocytes) but not with S100␤ (a mature astrocyte-specific marker); (b) the number of eGFP-doublecortin (a marker of immature neurons) double-positive neurons increased over time by a factor of 2.5 from day 60 to day 210 after injection; (c) at all time points after injection, eGFP-positive cells remained present in the RMS; and (d) there was a continuous increase of eGFP-positive cells in the OB up to 7 months after injection. Similar findings from Consiglio et al. [26] support the claim of efficient labeling of true adult neural stem cells by LVs. An additional advantage of this method is the use of a genetic label rather than a chemical label such as 5-bromo-2⬘-deoxyuridine (BrdU). The latter can be transferred to other cells after cell death, leading to erroneous conclusions [27, 28]. We have also shown recently that LVs encoding firefly luciferase enable long-term, quantitative imaging of reporter gene expression in mouse brain using BLI [18]. By combining both technologies, we tested here the hypothesis that BLI can be used to detect and quantify the progeny of LV-transduced stem cells arriving in the OB, because of the long-term persistence of the BLI signal and because signals originating from the SVZ and OB project at distinct sites [18] (supplemental online Fig. 3). We therefore scanned C57BL/6 mice after LV injection in the SVZ. We observed a clearly distinct second BLI focus emerging from the OB at week 15 after LV injection. For imaging of stem cell migration at earlier time points, we used albino C57BL/6-Tyrc-2J/J mice, which do not develop hyperpig-

2388

Bioluminescence Imaging of Neuronal Stem Cells

Figure 5. Repeated bioluminescence imaging (BLI) measurements in albino mice. Seven C57BL/6-Tyrc-2J/J mice were scanned at 1, 2, 4, 8, 12, and 15 w after stereotactic injection. At 1 and 2 w, we could detect only a signal originating from the site of injection. At 4 w, a signal from the OB was detected, and this signal became more clearly distinguishable at 8 w. From 8 w post-injection (pi) on, the BLI signal from the OB was always significantly higher than at 2 w pi (p ⱕ .019). A representative animal is shown. Abbreviations: Max, maximum; Min, minimum; OB, olfactory bulb; p, photons; s, second; w, week(s).

Figure 6. In vivo kinetics of stem cell migration. Kinetics of the in vivo OB/SVZ ratio for each of seven C57BL/6-Tyrc-2J/J mice. The dotted lines connect the measured bioluminescence imaging signal in time for each individual mouse. The solid lines represent the curves obtained with the parametric kinetic model. Abbreviations: OB, olfactory bulb; SVZ, subventricular zone.

mentation of the skin. In these mice we could detect a signal from the OB as early as 4 weeks after LV labeling of the SVZ. To quantify the number of marked cells in the OB with BLI, we applied a scatter correction based on an empirical gaussian model. A small fraction of the photons coming from the SVZ will scatter from the SVZ into the region of interest of the OB. A

Figure 7. Bioluminescence imaging (BLI) detects the impact of long-term expression of BDNF on adult neurogenesis. (A): At 30 w post-injection (pi) we detected a decrease in BLI signal from the OB and the SVZ in the BDNF group (n ⫽ 5) compared with the control group (n ⫽ 6). Note that the units of the scale are different for the BDNF and control groups. (B): In the OB there was a 78% and 85% decrease at 15 and 30 w, respectively (p ⫽ .010 and .001), in the BDNF group versus the control group. Stereological quantification of the number of eGFP-positive cells in the OB showed a 70% decrease (p ⫽ .0003) in the BDNF group compared with the control group. (C): In the SVZ there was a decrease of 86% and 91% at 15 and 30 w, respectively (p ⫽ .003 and p ⫽ .001), in the BDNF versus the control group. (D): Immunohistochemical staining for eGFP on coronal sections of the SVZ and the OB from an animal from the BDNF group and an animal from the control group are shown (30 w pi). Scale bar ⫽ 950 ␮m. Abbreviations: BDNF, brain-derived neurotrophic factor; eGFP, enhanced green fluorescent protein; OB, olfactory bulb; p, photons; s, second; SVZ, subventricular zone; w, weeks.

correction for scatter allows quantification of the signal originating from the OB at a certain time point. At the earliest time points after surgery, a large variability in the quantified OB signal was observed. In these images, the light distribution is disturbed by the scar, which results in a very irregular light distribution. At later

Reumers, Deroose, Krylyshkina et al. time points, the scatter correction was found to be very effective. It should be stressed, however, that the impact of the scatter of BLI signal from the SVZ to the OB was in fact rather small (on average, 13% reduction in OB signal after scatter correction). The strong linear correlation between the in vivo OB photon flux and the number of eGFP-positive cells, confirms that BLI allows appropriate quantification of the number of marked cells in the OB. Interestingly, this correlation outperformed the linear correlation between the ex vivo OB photon flux and the histological quantification. This could be explained by the higher variability in the ex vivo BLI measurements due to technical reasons such as differences in the time needed to dissect the OB immediately after sacrifice, causing differences in oxygenation state of the tissue and enzymatic activity. In vivo BLI thus produces a more reliable measurement than ex vivo BLI. When investigating the kinetics of stem cell migration, we chose to use the ratio of the scatter-corrected signal in the OB to the SVZ signal. This was done on the basis of the assumption that the number of labeled cells (and the fraction thereof that are stem cells) in the SVZ is constant over time. Consequently, this normalization corrects for possible changes in BLI signal over time, due to differential reporter gene expression (e.g., caused by transgene silencing) [18]. The similarity of the kinetics of this ratio, the kinetics of the ex vivo BLI values, and the kinetics of the number of eGFP-positive cells lends further support to the use of this ratio. Using a parametric model for kinetic fitting, we estimated that the neural stem cells in normal adult mouse brain have a migration time of less than a day and a half-life of approximately 1 month. When looking at potential factors influencing endogenous neurogenesis, the OB/SVZ ratio will be most reliable to detect effects on migration, differentiation, and survival but more ambiguous for factors affecting proliferation in the SVZ. In our BDNF study, both the signal from the OB and the signal from the SVZ were severely reduced. This implies that BDNF already acts at the level of neuronal progenitor cell proliferation in the SVZ. A strong effect of a factor on the SVZ, as is the case with BDNF, complicates the kinetic analysis, since the OB/SVZ ratio itself will be disturbed. Nevertheless, the absolute OB photon flux at a certain time point allows highly sensitive detection of factors affecting one of the different steps in neurogenesis. To the best of our knowledge, this is the first report on quantitative whole-body imaging of the migration of endogenous neural stem cells. Pineda et al. have demonstrated the ability of BLI to track the migration of exogenous stem cells from the RMS to the OB [29]. These cells, however, were maintained in cell culture and were not exposed to the original stem cell niche and thus constitute a completely distinct cell population. Furthermore, these cells were derived from the external germinal layer of mouse cerebellum and immortalized with the v-myc gene [30]. Recently Shapiro et al. [31] reported on in vivo tracking of endogenous stem cells with magnetic resonance imaging in rats. Although this technique has the advantage of a tomographical data set and detailed anatomical information, there were important limitations in this study: the animals were followed for only 5 weeks, no discrimination between live and dead cells was made, the labeling iron oxide particles can be taken up by microglial cells when neuroblasts die, and the signal strength in the RMS and OB was quite variable between animals. Moreover, no quantification of cell labeling in the OB or in the RMS was presented. Other groups have used invasive optical fiber microscopic techniques [32] that allow detection of cells in the RMS. These are limited to a field of view of 200 ␮m and do not yield quantitative data. Mizrahi et al. [33] used two-photon microscopy in a transgenic mouse strain, Thy-1-GFP-K12, specifically created for eGFP expression in the juxtaglomerular neurons of www.StemCells.com

2389

the olfactory bulb. The two-photon microscopy allows penetration only to several hundred micrometers, and only a fraction of the dorsal surface of the OB can be evaluated. As a first application of our imaging strategy, we investigated the effect of long-term overexpression of BDNF in the SVZ. In parallel experiments a transient increase followed by a strong decrease in neurogenesis in the SVZ was observed by immunohistochemistry (supplemental online Fig. 4). At 15 and 30 weeks after injection we detected a similar significant decrease in the number of labeled cells arriving in the OB. The BDNF-mediated downregulation of neurogenesis was detected by immunohistochemistry, as well as by BLI. The larger decrease detected with BLI is likely due to the very low numbers of labeled cells in the OB of the BDNF group that can generate only a signal very close to the detection limit of the IVIS 100. The impact of BDNF overexpression on neurogenesis appears to be dependent on the mode, time, and location of BDNF delivery. In the adult brain, short-term administration of BDNF in the lateral ventricle or in the SVZ was shown to stimulate neurogenesis [9 –11]. However, long-term overexpression of BDNF suppressed ischemia-induced hippocampal neurogenesis in rats [34], and in BDNF knockdown mice decreased levels of BDNF were associated with enhanced recovery and increased neuroblast number following experimental stroke [35]. In agreement with the latter findings, we demonstrate here that longterm overexpression of BDNF may attenuate neurogenesis, in contrast with short-term treatment. A reduction of the BLI signal was detected in both the OB and the SVZ, which was confirmed by eGFP immunohistochemistry. At present we do not have a complete explanation of the mechanism of action of BDNF. We performed BrdU stainings in the OB (supplemental online Fig. 2), which show a trend similar to that of the eGFP stainings. An extensive characterization of stem cell fate in the SVZ after longterm expression of BDNF is under way. At this stage, our results are important for two reasons. (a) Whereas transient overexpression of BDNF was shown to increase the proportion of new neurons in the OB [11], our data (BLI and immunohistochemistry in the supplemental online figures) demonstrate that it is crucial to investigate the long-term kinetics of the effect of growth factors as well. In light of the fact that BDNF has been proposed as a gene therapeutic agent for brain repair, our results represent a serious caveat. (b) We demonstrate unambiguously that BLI provides a noninvasive imaging technology to study the dynamics of growth factors over time. Our BLI approach could be further improved by increasing the specificity of the LV for the endogenous stem cells. Several strategies are possible, including pseudotyping the LV with proteins that confer a neural stem cell tropism, such as the lymphocytic choriomeningitis virus envelope glycoprotein [36], or by using promoters that are specific for the type B cells. Preliminary results with floxed LVs in transgenic mice expressing Cre under control of a stem cell-specific promoter look promising (data not shown).

CONCLUSION BLI of neuronal stem cell migration opens new avenues for stem cell research. Animals can now be quantitatively followed over time without the need to sacrifice them. The effect of pharmacological, behavioral, cell therapeutic, or gene therapeutic intervention on endogenous stem cell proliferation and migration can be monitored over time in the same animal without need for histology, facilitating testing of biological hypotheses in the stem cell field.

Bioluminescence Imaging of Neuronal Stem Cells

2390

ACKNOWLEDGMENTS We thank Martine Michiels, Irina Thiry, and Frea Coun for excellent technical assistance with the vector production. We thank Kristel Eggermont for help with the stereotactic surgery and Sylvie De Swaef for excellent help with histology. We are grateful to Irina Thiry for breeding the C57Bl/6-Tyrc-2J/J mice. We acknowledge the Cell Imaging Core of the K.U.Leuven for access to the confocal microscope facility. R.G. is a Postdoctoral Fellow of the Flemish Fund for Scientific Research (FWO Vlaanderen). V.R. is funded by a grant from the Institute for the Promotion of Innovation through Science and Technology in

Flanders (IWT-Vlaanderen). Research was funded by the SBO grant (IWT-30238) of the IWT, the IDO grant (IDO/02/012) from the K.U.Leuven, FWO Grant G.0406.06, the European grants DIMI (LSHB-CT-2005-512146) and Strokemap (FP6 037186), and the K.U.Leuven Center of Excellence Molecular Small Animal Imaging Center (MoSAIC).

DISCLOSURE

The authors indicate no potential conflicts of interest.

20

2 3 4 5 6 7

8 9

10 11

12 13 14 15 16 17

18 19

Kaplan MS, Bell DH. Mitotic neuroblasts in the 9-day-old and 11-monthold rodent hippocampus. J Neurosci 1984;4:1429 –1441. Doetsch F, Caille I, Lim DA et al. Subventricular zone astrocytes are neural stem cells in the adult mammalian brain. Cell 1999;97:703–716. Gage FH. Mammalian neural stem cells. Science 2000;287:1433–1438. Lois C, Garcia-Verdugo JM, Alvarez-Buylla A. Chain migration of neuronal precursors. Science 1996;271:978 –981. Saghatelyan A, Carleton A, Lagier S et al. Local neurons play key roles in the mammalian olfactory bulb. J Physiol Paris 2003;97:517–528. Curtis MA, Kam M, Nannmark U et al. Human neuroblasts migrate to the olfactory bulb via a lateral ventricular extension. Science 2007;315: 1243–1249. Freundlieb N, Francois C, Tande D et al. Dopaminergic substantia nigra neurons project topographically organized to the subventricular zone and stimulate precursor cell proliferation in aged primates. J Neurosci 2006; 26:2321–2325. Ho¨glinger GU, Rizk P, Muriel MP et al. Dopamine depletion impairs precursor cell proliferation in Parkinson disease. Nat Neurosci 2004;7: 726 –735. Pencea V, Bingaman KD, Wiegand SJ et al. Infusion of brain-derived neurotrophic factor into the lateral ventricle of the adult rat leads to new neurons in the parenchyma of the striatum, septum, thalamus, and hypothalamus. J Neurosci 2001;21:6706 – 6717. Zigova T, Pencea V, Wiegand SJ et al. Intraventricular administration of BDNF increases the number of newly generated neurons in the adult olfactory bulb. Mol Cell Neurosci 1998;11:234 –245. Benraiss A, Chmielnicki E, Lerner K et al. Adenoviral brain-derived neurotrophic factor induces both neostriatal and olfactory neuronal recruitment from endogenous progenitor cells in the adult forebrain. J Neurosci 2001;21:6718 – 6731. Geraerts M, Eggermont K, Hernandez-Acosta P et al. Lentiviral vectors mediate efficient and stable gene transfer in adult neural stem cells in vivo. Hum Gene Ther 2006;17:635– 650. Tang Y, Shah K, Messerli SM et al. In vivo tracking of neural progenitor cell migration to glioblastomas. Hum Gene Ther 2003;14:1247–1254. Shah K, Bureau E, Kim DE et al. Glioma therapy and real-time imaging of neural precursor cell migration and tumor regression. Ann Neurol 2005;57:34 – 41. Kim DE, Schellingerhout D, Ishii K et al. Imaging of stem cell recruitment to ischemic infarcts in a murine model. Stroke 2004;35:952–957. Kim DE, Tsuji K, Kim YR et al. Neural stem cell transplant survival in brains of mice: Assessing the effect of immunity and ischemia by using real-time bioluminescent imaging. Radiology 2006;241:822– 830. Munoz JR, Stoutenger BR, Robinson AP et al. Human stem/progenitor cells from bone marrow promote neurogenesis of endogenous neural stem cells in the hippocampus of mice. Proc Natl Acad Sci U S A 2005;102:18171–18176. Deroose CM, Reumers V, Gijsbers R et al. Noninvasive monitoring of long-term lentiviral vector-mediated gene expression in rodent brain with bioluminescence imaging. Mol Ther 2006;14:423– 431. Liang Q, Gotts J, Satyamurthy N et al. Noninvasive, repetitive, quantitative measurement of gene expression from a bicistronic message by

POTENTIAL CONFLICTS

OF INTEREST

REFERENCES 1

OF

21 22 23 24 25 26 27

28

29 30 31 32 33 34 35

36

positron emission tomography, following gene transfer with adenovirus. Mol Ther 2002;6:73– 82. Geraerts M, Michiels M, Baekelandt V et al. Upscaling of lentiviral vector production by tangential flow filtration. J Gene Med 2005;7: 1299 –1310. Baekelandt V, Eggermont K, Michiels M et al. Optimized lentiviral vector production and purification procedure prevents immune response after transduction of mouse brain. Gene Ther 2003;10:1933–1940. West MJ, Slomianka L, Gundersen HJ. Unbiased stereological estimation of the total number of neurons in the subdivisions of the rat hippocampus using the optical fractionator. Anat Rec 1991;231:482–497. Baekelandt V, Claeys A, Eggermont K et al. Characterization of lentiviral vector-mediated gene transfer in adult mouse brain. Hum Gene Ther 2002;13:841– 853. Kempermann G, Kuhn HG, Gage FH. Genetic influence on neurogenesis in the dentate gyrus of adult mice. Proc Natl Acad Sci U S A 1997;94: 10409 –10414. Le Fur N, Kelsall SR, Mintz B. Base substitution at different alternative splice donor sites of the tyrosinase gene in murine albinism. Genomics 1996;37:245–248. Consiglio A, Gritti A, Dolcetta D et al. Robust in vivo gene transfer into adult mammalian neural stem cells by lentiviral vectors. Proc Natl Acad Sci U S A 2004;101:14835–14840. Coyne TM, Marcus AJ, Woodbury D et al. Marrow stromal cells transplanted to the adult brain are rejected by an inflammatory response and transfer donor labels to host neurons and glia. STEM CELLS 2006;24: 2483–2492. Burns TC, Ortiz-Gonzalez XR, Gutierrez-Perez M et al. Thymidine analogs are transferred from prelabeled donor to host cells in the central nervous system after transplantation: A word of caution. STEM CELLS 2006;24:1121–1127. Pineda JR, Rubio N, Akerud P et al. Neuroprotection by GDNF-secreting stem cells in a Huntington’s disease model: Optical neuroimage tracking of brain-grafted cells. Gene Ther 2007;14:118 –128. Snyder EY, Deitcher DL, Walsh C et al. Multipotent neural cell lines can engraft and participate in development of mouse cerebellum. Cell 1992; 68:33–51. Shapiro EM, Gonzalez-Perez O, Manuel Garcia-Verdugo J et al. Magnetic resonance imaging of the migration of neuronal precursors generated in the adult rodent brain. Neuroimage 2006;32:1150 –1157. Davenne M, Custody C, Charneau P et al. In vivo imaging of migrating neurons in the mammalian forebrain. Chem Senses 2005;30(suppl 1): i115–i116. Mizrahi A, Lu J, Irving R et al. In vivo imaging of juxtaglomerular neuron turnover in the mouse olfactory bulb. Proc Natl Acad Sci U S A 2006;103:1912–1917. Larsson E, Mandel RJ, Klein RL et al. Suppression of insult-induced neurogenesis in adult rat brain by brain-derived neurotrophic factor. Exp Neurol 2002;177:1– 8. Nygren J, Kokaia M, Wieloch T. Decreased expression of brain-derived neurotrophic factor in BDNF(⫹/-) mice is associated with enhanced recovery of motor performance and increased neuroblast number following experimental stroke. J Neurosci Res 2006;84:626 – 631. Stein CS, Martins I, Davidson BL. The lymphocytic choriomeningitis virus envelope glycoprotein targets lentiviral gene transfer vector to neural progenitors in the murine brain. Mol Ther 2005;11:382–389.

See www.StemCells.com for supplemental material available online.