Magnetic resonance imaging of the migration of neuronal precursors generated in the adult rodent brain

Abstract

Neural progenitor cells (NPCs) reside within the subventricular zone (SVZ) in rodents. These NPCs give rise to neural precursors in adults that migrate to the olfactory bulb (OB) along a well-defined pathway, the rostral migratory stream (RMS). Here we demonstrate that these NPCs can be labeled, in vivo, in adult rats with fluorescent, micron-sized iron oxide particles (MPIOs), and that magnetic resonance imaging (MRI) can detect migrating neural precursors carrying MPIOs along the RMS to the OB. Immunohistochemistry and electron microscopy indicated that particles were inside GFAP⁺ neural progenitor cells in the SVZ, migrating PSA-NCAM⁺ and Doublecortin⁺ neural precursors within the RMS and OB, and Neu-N⁺ mature neurons in the OB. This work demonstrates that in vivo cell labeling of progenitor cells for MRI is possible and enables the serial, non-invasive visualization of endogenous progenitor/precursor cell migration.

Introduction

The subventricular zone (SVZ) is the largest germinal layer in the adult rodent brain. It is localized next to the lateral wall of the lateral ventricle. Primary progenitors in this region correspond to type B cells, which have properties of astrocytes (Garcia-Verdugo et al., 1998; Doetsch et al., 1999). These cells divide to generate transit amplifying type C cells, which generate new neurons that migrate to the olfactory bulbs (OBs) (Garcia-Verdugo et al., 1998; Doetsch et al., 1999) along a well-defined pathway, the rostral migratory stream (RMS) (Doetsch and Alvarez-Buylla, 1996). Once in the OB, these new neurons differentiate into granule cell neurons and periglomerular neurons (Lois and Alvarez-Buylla, 1993). In mice, many new neurons reach the OB by two days and a majority arrive in the OB by day 6 (Lois and Alvarez-Buylla, 1994). Experimentally induced traumatic brain injury (Salman et al., 2004) and stroke (Zhang et al., 2004) induce new SVZ progenitor cell proliferation and migration, suggesting that these cells may also have a function in repair mechanisms. Additionally, analogous neural progenitor cells have recently been discovered in non-human primates (Kornack and Rakic, 2001; Pencea et al., 2001) and in humans (Sanai et al., 2004).

To date, fluorescence imaging and electron microscopy (EM) have been principally used to identify migrating cells derived from cells in the SVZ. Due to the number of antibodies specific to these cells, immunohistochemistry has been used to detect groups of cells in well-defined areas. Bromodeoxyuridine has also been used to specifically label these cells because they divide in the SVZ before migrating. Additionally, transgenic cells expressing green fluorescent protein have been successfully used to study cell migration (Suzuki and Goldman, 2003). However, none of these techniques enable migration throughout the brain within living animals to be studied.

Magnetic resonance imaging (MRI)-based neural cell tracking in intact animals was first described in 1990 (Ghosh et al., 1990) and has been further developed in a number of laboratories (Tang et al., 2003; Hoehn et al., 2002; Bulte et al., 2001; Franklin et al., 1999; Yeh et al., 1995; Hawrylak et al., 1993; see Bulte and Kraitchman, 2004, for a recent review of this field). In nearly every demonstration of cell tracking by MRI, cells were labeled with an MRI contrast agent in culture, injected into the animal, and imaged either in vivo or in vitro. Only blood borne cells such as macrophages have been consistently labeled with MRI contrast agent in vivo (Zhang et al., 2000). The MRI contrast agent most often used is a nanometer-sized dextran-coated iron oxide nanoparticles (commonly referred to as USPIOs or MIONs, 30–200 nm diameter). An advantage of iron oxide particles as an MRI contrast agent is that iron oxide disturbs the otherwise homogeneous magnetic field used for MRI, and as such cells harboring iron oxide generate dark spots in MRI (Lauterbur et al., 1986). The large quantity of iron oxide that can be loaded into cells, primarily by endocytosis, enables sensitive detection of cells. A major advantage of using MRI to study cell migration is that the migration can be directly mapped onto the anatomy available from MRI, as well as functional activity available from fMRI or neuronal tract tracing (Pautler et al., 1998).

Recently, it has been demonstrated that micron-sized iron oxide particles (MPIOs) have some advantages for cell labeling studies by MRI (Shapiro et al., 2005). Very stable particles are readily available, which are impregnated with various fluorescent dyes. Most cells studied to date readily endocytose MPIOs. Each particle contains 1–10 pg of iron, enabling very high levels of iron loading. Indeed, even single MPIO particles have been detected in single cells in culture (Shapiro et al., 2005) and in fixed mouse embryos (Shapiro et al., 2004). Furthermore, single cells labeled with MPIOs have been detected, in vivo, by MRI. This was accomplished in mouse livers with transplanted hepatocytes (Shapiro et al., 2006) and in rat hearts with individual macrophages (Wu et al., 2005). Thus, MPIOs open the possibility of sensitive labeling and detection of cells for MRI.

In principle, as few as a single MPIO is needed to be taken up by cells to be labeled for MRI. Therefore, it should be possible to directly label cells in vivo. The goal of the present work was to label neural progenitor cells in the SVZ, in vivo, by direct injection of MRI contrast agent into the lateral ventricles, near the SVZ. Due to the small size of the SVZ in adult rats and subsequent difficulty in consistently delivering injections there, the strategy was to target the neural progenitors via the ventricle. As only single MPIOs are required for detection of cells (Shapiro et al., 2004), inefficient labeling could be tolerated. Following endocytosis of the particles from the ventricle, the particles were incorporated into daughter neural precursors through cell division. When the neural precursors migrated along the RMS, the presence of one or more MPIOs generated dark spots in the MRI revealing the locations of the migratory cells. Immunohistochemistry and Prussian blue iron staining confirmed the presence of particles in SVZ cells close to the lateral ventricle and along the RMS, and in mature neurons in the granule cell and periglomerular layers. EM confirmed the intracellular location of the particles in granule cell neurons in the OB.

Materials and methods

Animal injections

Twenty, 6-week-old Sprague–Dawley rats (Harlan, Indianapolis, Indiana) were stereotactically injected with 5–50 μl (1.5 ×10⁷ to 1.5 × 10⁸ particles) of 1.63 μm diameter MPIOs (encapsulated, fluorescent, magnetic beads, Bangs Laboratories, Fishers, IN) into the anterior right lateral ventricle. The coordinates chosen were 2 mm caudal from bregma, right 2 mm, down 3 mm, and were based on prior mapping of rat cerebral anatomy in similar sized rats. The MPIOs were styrene/ divinylbenzene-coated iron oxide microparticles with a green fluorescent dye (480 excitation, 520 emission, standard green fluorescent protein filters) impregnated into the shell. The MPIOs were COOH–functionalized and contained 45% w/w iron oxide. In some cases, 400 Al MPIOs was soaked overnight with 100 μg EGF (Sigma) prior to injection. As the RMS exists proximal to a white matter band, control (n = 3) injections were also performed by injecting 50 μl (1.5 × 10⁹ particles) of MPIOs into the corpus callosum. This served to test for background migration along white matter tracks. Six animals were also injected with 50 μl (1.5× 10⁹ particles) of MPIOs into the cortex near the SVZ, serving as a control for gray matter migration. All experiments were carried out in compliance with guidelines set by the National Institute of Neurological Disorders and Stroke Animal Care and Use Committee.

MRI

Immediately following injection, animals underwent a baseline MRI investigation. MRI was performed either on an 11.7-T or a 7.0-T system (Bruker Biospin, Billerica, MA, Paravision software 3.0.1). Separate transmit-only volume (35 mm birdcage coil) and receive-only surface coils (30 mm diameter) were used. Under 3% isoflurane anesthesia in 90% oxygen/10% medical air, animals were placed in a custom-built MR compatible stereotactic head frame. The high oxygen concentration lowers the amount of deoxygenated blood, lessening the appearance of dark veins in the MR images (Dunn et al., 2002). The surface coil was placed directly on the head of animals. Animals were orally intubated and mechanically ventilated at 65 breaths/min and end tidal CO₂ and respiratory patterns were monitored. Body temperature was maintained at 37° C by use of a circulating water bath and rectal temperature feedback. 3D gradient echo MRI was performed using the following image parameters: FOV 2.56 cm³, matrix 256³ (100 μm isotropic voxel size), TR = 70 or 30 ms (75 or 32 min acquisition), TE = 8 ms, 12.5 kHz acquisition BW. Following MRI, animals were revived and returned to the animal facility. Subsequent MRI was performed either weekly for 4 weeks (n = 7) or once at 5 weeks post-injection (n = 13).

Following the last MRI session, animals were transcardially perfused with 0.9% saline, followed by 10% formalin containing 1 mM Gd-DTPA. The Gd-DTPA reduces the T₁ of the sample to allow rapid imaging. Brains, with intact olfactory bulbs were removed and post-fixed overnight in the same Gd-DTPA containing fixative. Brains were then placed in saline and imaged in vitro in a 35-mm birdcage coil with the following image parameters: gradient echo MRI, FOV 2.56 × 1.28 × 1.28 cm³, matrix 512× 256 × 256 (50 μm isotropic voxel size), TR = 70 ms, TE = 8 ms, 12.5 kHz acquisition BW, 16 averages.

Histology

For immunohistochemistry, brains were processed for frozen sections. Frozen sections preserve the fluorescence on the particles; paraffin embedding procedures using organic solvents do not. Following a two-day soak in 30% sucrose, brains were bisected sagittally and individually embedded in TissueTek® embedding compound. Sagittal 16-μm sections, encompassing the entire OB, RMS, and SVZ, were cut using a cryostat. After washing 2 × in 0.1 M PBS, sections were incubated with primary antibodies overnight at 4°C in 0.1 M PBS + 10% normal goat serum and 0.1% Triton X-100. The following primary antibodies were used: mouse monoclonal anti-GFAP (1:500, Chemicon, Temecula, CA), anti-PSA-NCAM (1:1000, AbCys, France), and anti-NeuN (1:500, Chemicon, Temecula, CA); rabbit polyclonal anti-S100β (1:500) (Dako A/S, Denmark); and anti-Iba-1 (1:500, Wako, Richmond, VA). Then, tissue sections were washed 4 times in PBS and incubated with the appropriate Alexa 594-conjugated secondary antibodies (all 1:400, Molecular Probes, OR, USA). Controls in which primary antibodies were omitted resulted in no detectable staining. Confocal fluorescence microscopy was performed using a Zeiss LSM 510 (Zeiss, Thornwood, NY). Electron microscopy was performed as previously described (Doetsch et al., 1997). Prussian blue staining with nuclear fast red counterstaining was carried out on 20-Am-thick sections using standard procedures (Histoserve, Germantown, MD) and visualized with a stereomicroscope (Leica Microsystems, Bannockburn, IL).

Results

MRI of migrating neuroblasts within the rostral migratory stream

Five weeks following intraventricular injection, MRI detected dark contrast along the RMS and in the olfactory bulb. Fig. 1 shows an example of MRI immediately before injection of MPIOs (row 1, A–C), immediately after injection of MPIOs (row 2, D– F), five weeks post-injection (row 3, G–I), and MRI from the same brain fixed, 5 weeks post-injection (row 4, J–L). Columns A, B, and C show the sagittal, horizontal, and oblique slice along the RMS. Fig. 1A is a sagittal view of the rat brain prior to MPIO injection. A white dashed line is superimposed over the image, from the tip of the lateral ventricle to the olfactory bulb, depicting the RMS. Fig. 1B is a horizontal slice at the level of the lateral ventricles (superimposed black dotted line in Fig. 1A). A typical injection site is depicted by the superimposed white circle. Fig. 1C shows a view along the RMS and through the OB (superimposed black diagonal dashed line in Fig. 1A). Two faint streaks in the MR image (arrows) are visible extending into the OB. These are the white matter tracks along which the dorsal RMS lies.

Fig 1.

MRI of endogenous rat adult neural progenitor cells. Panel A is a sagittal slice along the right RMS, in vivo, prior to the injection of contrast agent. The white dashed line is overlayed on the position of the RMS. Panel B is a slice along the dotted horizontal line in panel A. Panel C is an oblique slice along the dashed diagonal line in panel A, along the RMS. Panels D–F are in vivo MRI slices of the same animal 90 min after injection of the contrast agent. Panels G–I are in vivo MRI slices of the same animal post five weeks. Panels J–L are in vitro MRI slices of the same brain after perfusion and fixation. All white scale bars are 5 mm.

Fig. 1, row 2, shows the same three slice orientations for a rat 90 min after injection of 50 Al MPIOs into the right lateral ventricle. Within this time period, the entire ventricular system became filled with particles. The MPIOs caused the well-known effect of dark contrast in a gradient echo, T*₂-weighted MRI. Figs. 1D and E show that the right lateral and third ventricles were filled with particles and that the third ventricle drained particles into the subarachnoid space above and below the brain. Because the ventricles were initially filled with so much contrast material, the dark contrast area was much larger than the actual ventricles. This “blooming” effect is well known and key for visualizing migratory cells but can also distort heavily labeled areas such as the ventricles.

Fig. 1, row 3, shows results from the same animal as shown in Fig. 1, row 2, five weeks after injection of MPIOs, and Fig. 1, row 4, shows the MRI from the fixed brain. After five weeks, a trail of dark contrast from the ventricle to the OB, along the RMS, was clearly visible (arrows in Figs. 1G and J). The contrast was confined to a 300-μm-thick strip within the RMS; however, inside the OB, the contrast was highly punctate and fanned out towards the edges of the OB. In the case shown, the contralateral side of the brain from the injection site showed a similar pattern, however, with fewer dark contrast areas (data not shown). Figs. 1H and K are horizontal views, in vivo and in vitro, respectively, showing that five weeks later, the ventricles still contained particles; however, the contrast area had shrunk, indicating that there were less MPIOs than there were immediately following the injection. Also, there was dark, punctuated contrast within the central granule cell layer of both OBs (arrow). Fig. 1I shows the view along the RMS, from the tip of the ventricle to the middle of the OB. The dark contrast stream curved laterally at first, then medially once inside the OB (arrows). It was present in the contralateral side as well, although more faint. Fig. 1L, acquired at twice the resolution of the in vivo MRI, shows the position of the RMS with respect to the white matter track running into the OB (arrows). The RMS tracked medially to the white matter track and followed closely its curve into the OB. The dark contrast on the outside of the brain is due to particles that adhered to the outside of the brain from drainage of the CSF into the arachnoid space.

Immunohistochemical analysis and electron microscopy

Histology was performed on the brains to determine what type of cells were labeled by MPIOs along the entire migratory pathway, from the ventricle through the RMS, into the OB. Fig. 2 shows images from sections stained for iron with Prussian blue and counterstained with nuclear red, a stain for cell nuclei. Fig. 2A shows a wide field sagittal section of a rat brain, from the lateral ventricle to the olfactory bulb. The RMS can be seen as a thin, dark pink strip, beginning at the tip of the ventricle, and diving diagonally down, before making a sharp turn and heading into the OB. The darker pink color comes from the high density of cells in the RMS. Figs. 2B–G are magnified from the corresponding areas labeled in Fig. 2A. Fig. 2B shows the SVZ where labeled cells were present, indicative of iron particles that left the ventricular space and entered the SVZ. Two general sized stained areas are visible. Large, globular stained areas (dotted arrow) and smaller, isolated stained areas (solid arrow). One area containing smaller isolated dark stains is shown in Fig. 2C and is an expansion of the dotted boxed area in Fig. 2B. Here can be seen small, dark-stained areas, associated with cells arranged in a chain like fashion (arrow). Neural precursors migrate from the SVZ to the OB along themselves in a chain like manner (Lois et al., 1996). Thus, this is likely a chain of migrating neural precursors carrying particles to the OB from the SVZ. A similar chain can be seen in the right side of Fig. 2B.

Fig 2.

Prussian blue-stained histology with nuclear red counterstaining of rat brain five weeks post-injection of particles. Prussian blue-stained iron is blue, cell nuclei are stained dark pink. Panels B–G are expansions of the areas shown in panel A.

Figs. 2D and E show Prussian blue-stained sections of the beginning of the RMS and at the area of the sharp turn, respectively. Immediately entering the RMS, there was a high density of large globular dark stain (arrow) with many isolated dark stains as well. The dark contrast was almost entirely contained within the RMS. The dark stain also tapered off quickly inside the RMS to where only isolated dark stain is observed within the RMS. Fig. 2E shows many small, isolated dark spots associated with the RMS (arrows). Prussian blue normally stains iron as blue, and this can be seen in the images. However, the stain often appears black, possibly due to a lot of iron locally or an effect of the polymer coating of the particle.

Although the fate of most of the migrating neural precursors is to become granule cell layer interneurons, some cells do travel to the periglomerular layer (Doetsch et al., 1999). Figs. 2F and G show the OB and an expanded view of a few glomeruli. Visible in 2F are numerous blue spots surrounding glomeruli in the OB (arrows). Fig. 2G allows one to count individual blue spots in this view, in three cells in the periglomerular layer (arrows). Blue staining of iron was not seen anywhere else in this region except in the periglomerular layer.

To investigate the full range of cell types that incorporated the fluorescent particles into their cytoplasm, immunohistochemistry analyses were performed on counterstained sections using antibodies for GFAP (astrocytes-like progenitor cells, “B” cells), S100β (ependymal cells, “E” cells), PSA-NCAM (migrating neuroblasts, “A” cells), doublecortin (migrating neuroblasts, “A” cells), and NeuN (mature neurons). Five weeks after the injection, green-fluorescent MPIOs were observed in the GFAP⁺ processes of the SVZ astrocytes. Some of these processes were contacting the lateral wall of the ventricle (Fig. 3B). In the same region, many S100β⁺ ependymal cells were found containing the green-fluorescent MPIOs (Fig. 3A). At the dorsal part of the SVZ and throughout the RMS, several PSA-NCAM⁺ and Doublecortin⁺ migrating neuroblasts contained the particles in their cell bodies (Figs. 3D and E). Here, the typical chain organization of migrating cells can be seen, with cells within the chains carrying single MPIOs. Within the granule cell layer of the olfactory bulb, NeuN⁺ cells were commonly observed with intracytoplasmic fluorescent particles, suggesting that migrating neuroblasts tagged with the fluorescent particles can differentiate into mature granular interneurons (Fig. 3H). Because many of the migrating cells die en route and presumably release their particles, some sections were stained with Iba-1 antibody to detect whether microglia cells incorporated free green fluorescent MPIOs. A small population of Iba-1⁺ cells contained the fluorescent MPIOs. These cells were found in the dorsal SVZ (Fig. 3C), RMS (Fig. 3F), and olfactory bulb. Most of the fluorescent-tagged Iba-1⁺ cells were observed surrounding the blood vessels in the striatum and in periventricular regions. In general, few particles (1 to 5) were found in astrocytes and migrating neuroblasts; however, microglia cells were detected harboring as many as 20 or more particles.

Fig 3.

Immunohistochemistry and electron microscopy of MPIO-labeled cells in the SVZ, RMS, and OB. (A) S100B⁺ ependymal cells, (B) GFAP⁺ neural progenitor cells, (C) Iba-1⁺ microglia, (D) PSA-NCAM⁺ migrating neuroblasts, (E) Doublecortin⁺ migrating neuroblasts, (F) Iba-1⁺ microglia, (G) Doublecortin⁺ migrating neuroblasts, (H) NeuN⁺ mature neurons, (I) granule cell neuron with MPIOs, inset is magnification of particles. All immunohistochemistry stains are red. MPIOs are fluorescent green and are shown with arrows, where necessary.

To confirm that the MPIOs were intracellular, electron microscopy was performed in the same brain regions. Five weeks after injection, electron microscopy detected intracellular MPIOs in astrocytes, microglia, ependymal, and migratory cells. Of all the cellular types that were labeled with MPIOs, astrocytes and microglia showed greater amounts of iron particles in electron micrographs, a fact that is probably related to their phagocytosing capacity. In the RMS, MPIOs were observed in migrating cells and microglia. In the latter, the labeling was particularly intense. In the OB, MPIOs were found in the cytoplasm of granular interneurons, as shown in Fig. 3I. In the cell shown, eleven MPIOs were detected. The ultra-structural features of these neurons were consistent with those of granular neurons, with scarce organelle, and high nucleus-to-cytoplasm ratio. The labeling did not seem to produce any apparent effect on the morphological features of these neurons.

As the RMS lies adjacent to a white matter track leading from the ventricle to the OB, control experiments were performed to test whether the particles can simply migrate along white matter tracks, independent of cellular uptake. Fifty microliters of particles was injected into the corpus callosum of three animals. This location was chosen over a direct injection into the RMS to largely eliminate contaminating the result with endocytosis of particles by the migrating neural precursors within the RMS. MRI and histological evaluation were performed five weeks post-injection, identically as for the experimental animals. No detectable expansion of the initial injection site or signs of migration were observed within the corpus callosum either across hemispheres or within hemispheres for any animal (data not shown). Furthermore, injections in the cortex above the SVZ resulted in no observable expansion of the injection site or signs of cell migration (data not shown).

Serial MRI of neuroblast migration

MRI should allow one to follow the migration of SVZ cells along the RMS in individual animals. Thus, we investigated the distribution of contrast agent by MRI in animals that received injections as above. Fig. 4 shows in vivo MRIs of the frontal portions of rat brains for two different rats, displaying the evolution of contrast within the RMS and OB for time points at the time of injection, and for three weeks post-injection. It can be seen in both rats that immediately post-injection the RMS and OB had no dark contrast (Figs. 4A and E). Contrast appeared in the RMS by week 2 in rat 1, as shown in Fig. 4C, with little contrast at week 1 (Fig. 4B) and intensified at week 3 (Fig. 4C). In rat 2, contrast within the RMS and OB appeared by week 1 (Fig. 4F), which increased in weeks 2 (Fig. 4G) and 3 (Fig. 4H). In all, at week 1, two rats exhibited contrast only in the RMS, three rats had strong contrast in the RMS with intermediate contrast in the OB, and two rats had extensive contrast within the RMS and OB. Six of the seven rats subjected to weekly MRI post-injection displayed significant contrast by two weeks, with all seven displaying significant contrast at week 3.

Fig 4.

MRI time course measurement of NPC migration. Each column is a sagittal MRI through the right RMS, from one animal. Row 1 is MRI 90 min post-injection. Rows 2–4 are MRI from 1, 2, and 3 weeks post-injection, respectively. White scale bars measure 0.25 cm.

One important factor to getting good migration was the place of injection. Only injections within the ventricle, close to the SVZ, led to migration. Injections into the ventricle far from the SVZ were much less successful at generating contrast related to cell migration. Fig. 5 demonstrates that even within successful injections there was variability of contrast in the RMS and OB between rats, likely due to the exact spot of injections. Shown here are in vivo (first column) and in vitro (second column) MRIs for four rats, acquired at five weeks post-injection. Rats 1 (Figs. 5A and B) and 2 (Figs. 5C and D) displayed “spotty” hypointense contrast within the RMS and OB while rats 3 (Figs. 5E and F) and 4 (Figs. 5G and H) displayed hypointense contrast that was more continuous within the RMS and became “spottier” within the OB.

Fig 5.

Uniqueness of each MRI experiment. Each row is MRI from one animal. The first column is in vivo MRI 5 weeks post-injection. The second column is MRI of the same brain, perfused and fixed and imaged in saline. Each image is a sagittal slice through the right RMS. White scale bars measure 0.25 cm.

Discussion

The migration of SVZ cells along the RMS and into the OB has been clearly established in adult rodents. Although much has been learned about the molecular and cell biology of these cells, it has been difficult to study the role of these cells, in vivo, in intrinsic function of the brain. One reason may be the lack of an in vivo method to follow these cells. The methods previously used to study migration of these neural precursors largely relied on immunohistological analysis of excised tissues. There is one recent report that used a catheter confocal fluorescent microscope to visualize neural progenitor cell migration in vivo, but only within a restricted 200 Am field of view (Davenne et al., 2005). Here we demonstrate that MRI can detect migration of endogenous cells after intraventricular injection of MPIOs. Furthermore, this report demonstrates that endogenous, migrating cells can be detected by MRI in an animal, in cell types other than blood cells. Fluorescent MPIOs directly injected into the ventricle were incorporated into SVZ progenitors and migrating neuroblasts. These cells migrated into the OB, where they could be localized in both granule layer and periglomerular cells. Although it is clear that particles are endocytosed by GFAP⁺ neural progenitors residing behind the ependymal cell layer, the mechanism of neuroblast labeling is unclear. It is also unclear whether neuroblasts directly incorporated MPIOs or inherited the label from the progenitor cells.

Once inside the OB, new neurons have three options: cells can (1) die, (2) graft within the granule cell layer, or (3) migrate outwards to the periglomerular layer. It is apparent that as the neural precursors traveled along the RMS to the OB, some died, releasing their particles that were taken up by microglia, explaining why labeled microglia were only visualized along the RMS. It is unlikely that microglia were labeled at the ventricle as well and migrated along with neural precursors to the OB because labeled microglia were found nowhere else in the brain but the RMS and OB, and there is little evidence for long-distance migration of microglia. Prussian blue staining captured several periglomerular cells containing blue iron stain, with no other iron staining anywhere within the slice. Doublecortin and NeuN immunohisto-chemistry demonstrated labeled neuronal precursors and mature neurons, respectively, in the granule cell layer. Electron microscopy demonstrated that MPIOs were intracellular.

In mouse studies, Lois and Alvarez-Buylla (1994) showed by bromodeoxyuridine injections that neural precursors arrive at the OB from the ventricle within a few days, with significant accumulation by day six. In the present study, some animals clearly displayed significant hypointense contrast in the OB by seven days post-injection, with all displaying significant hypointense contrast by day fourteen. The slower rate at which MPIO-labeled neural precursors arrived at the OB versus bromodeoxyuridine-labeled neural precursors in the Lois study may be due to slow uptake of significant numbers of particles from the ventricle as opposed to bromodeoxyuridine labeling. Another factor is that the Lois study was performed in mice whereas the present study was performed in rats. It should now be possible to study the detailed time course of uptake and migration with MRI.

There are a number of limitations to this method. First is the large number of injected particles necessary for adequate cell labeling. Preliminary recent results indicate that similar labeling can be accomplished with fewer particles. Another approach to reducing the number of injected particles needed for labeling may be conjugating antibodies against the NPCs (anti-GFAP) to the particles, effecting greater affinity of the particles to the NPCs. A second limitation is the inability to distinguish live, migrating cells from dead cells or particles within microglia. As the rate of migration is ~40–80 μm per hour (Davenne et al., 2005), it should be possible to acquire serial images to distinguish which MRI spots are moving (i.e., live, migrating cells) from those that are completely stationary (i.e., microglia). A final limitation is that not all the cells that migrated were labeled. Thus, only a fraction of cells can be followed.

In vivo magnetic cell labeling of endogenous neural stem/ progenitor cells with an MRI contrast agent and subsequent monitoring of migration should open up a number of experimental possibilities. It will enable following the migratory pathway of the precursor cells in vivo, without the need to sacrifice the animal. This can be done in the same animal at time different time points, whereas before different animals needed to be used for different time points. In vivo cell labeling, as opposed to labeling stem cells in culture, bypasses the tedious and possibly confounding steps of maintaining the undifferentiated stem cell state in culture. Furthermore, the niche environment that the stem cells exist in can be preserved. More interestingly, the distribution of cells in the olfactory bulb in response to different odor stimulations, or the ability to cause migratory deviations from the RMS in response to injury, should be amenable to study using these techniques, possibly to the point of detecting single cells (Shapiro et al., 2006;Wu et al., 2006;Heyn et al., 2006). The migratory pathway and endpoints can be readily co-registered with a wide variety of other anatomical and functional information available from MRI.