All animal studies described in this report were approved by the Institutional Animal Care and Use Committee at New York University School of Medicine. Studies were performed using 6–10 week old, female ICR mice (Taconic Farms, Hudson, NY).
For injection of MPIOs, mice were anesthetized using isoflurane and placed in a stereotaxic frame. The head was shaved and a ~1 cm incision was made to expose the skull. Using a microdrill, a 1–2 mm hole was produced to allow an injection into the anterior portion of the SVZ (stereotaxic coordinates 0.7–0.9 mm lateral to bregma, 1.1–1.3 mm rostral to bregma and 2.4–2.6 mm deep from the pial surface). All injections were performed with glass microcapillary tubes (0.5 mm ID, 1.0 mm OD, 10 cm length, Sutter Instruments, Novato, CA), pulled and cut to produce a 100 µm diameter needle tip and mounted on a microinjector (Narishige International, East Meadow, NY). MPIO solution (50 nl, ~1.5 × 106 particles) was injected as provided by the manufacturer (encapsulated, fluorescent, magnetic beads, nominal 1.63 µm diameter, 520/480 nm excitation/emission, Bangs Laboratories, Inc, Fishers, IN). Following surgery, the skin over the skull was sutured closed and animals were placed in separate, heated cages and monitored until fully recovered.
Magnetic resonance imaging
All imaging experiments were performed on a 7T 200-mm bore Magnex magnet equipped with a Bruker Biospec Avance II console (Bruker Biospin MRI, Ettlingen, Germany) and actively shielded gradients (BGA9S, Bruker Biospin MRI). Images were collected using a quadrature, transmit/receive Litzcage volume coil (25 mm inner diameter, Doty Scientific, Columbia, SC, USA). For visualization of MPIO distribution, a multiple gradient-echo image acquisition sequence was employed (TR = 40 ms, TE = 4.0, 8.3 ms, 12° excitation angle and 8 repeats). Image dimensions included a field-of-view of 25.6 × 19.2 × 12.8 mm and matrix size 256 × 192 × 128 to yield 100 µm isotropic voxels in an imaging time of 2 h, 12 min. Image reconstruction with retrospective self-gating was used to minimize artifact due to respiration and motion as described previously (Nieman et al., 2009
). A magnitude image was generated for each echo and then the two images were averaged to yield a single image for subsequent analysis. The averaged image provided ~30% higher signal-to-noise and contrast-to-noise ratio than either single image, while retaining a short echo-time image for clearer visualization of the injection site. Prior to imaging, animals were prepared in an induction chamber with 4–5% isoflurane in compressed air and then transferred to the imaging holder and coil assembly where they were maintained under anesthesia with 1.0–1.5% isoflurane.
Labeled image voxels were isolated by statistically testing for hypointense voxels (). Images were first registered together into a common space to align ipsilateral voxels across all images and subsequently to align the contralateral voxels as well. This process, including each of the steps described below, was performed separately for analyses of RMS migration rates and OB distribution. For the migration rate analyses, image time points included Days 0 (~2 h post-injection), 1 and 2. The OB distribution analyses included images from Days 0–1 (comprised of images collected 3 to 24 h post-injection), 4, 7 and 21. In each case, the image processing began by positional alignment and intensity normalization of all images. The latter included correction for nonuniformity (Sled et al., 1998
) and histogram normalization. All images were then registered together to generate an unbiased space into which all images were resampled. This process has been described in several previous publications (Kovacevic et al., 2005
; Lerch et al., 2008
; Nieman et al., 2006
; Woods et al., 1998a
). The unbiased space was defined through a process of pairwise linear registrations and then refined through an iterative nonlinear registration of the individual images. For these analyses, the region of the MPIO injection was excluded from the registration process with a mask and the final nonlinear registration step was limited to a coarse resolution (300 µm). This permitted alignment of the olfactory bulbs without fine adjustment of isolated hypointense voxels, which appear somewhat stochastically through the OB at late stages. All registrations were performed using software tools from the Montreal Neurological Institute (Collins et al., 1994
Fig. 1 Intensity-based analyses of labeled MR images. To isolate hypointense, labeled voxels in MR images, images were first aligned to a reference (removing translation and rotation differences) and intensity-normalized (a). The set of all images–only (more ...)
With images registered together, individual voxels were determined to be hypointense—andtherefore “labeled”—by comparison of the intensity value in the ipsilateral side to the corresponding values in the equivalent contralateral voxel. This comparison was performed for each voxel in the forebrain and OB. To isolate the relevant voxel on the contralateral side of the brain, all images were mirrored left to right and then registered to the common space. Subsequently, the set of contralateral intensity values at the voxel of interest and from all acquired images was used as reference values to assign a p
-value to each ipsilateral voxel using a one-sided Student’s t
-test. Statistical thresholds were then applied using a false discovery rate (FDR, q
= 0.05) to correct for multiple comparisons (Benjamini and Hochberg, 1995
). For visualization of the distribution of particles, thresholded binary maps of hypointense voxels were overlaid on the average images.
In order to facilitate visualization of the RMS in three dimensions, two image planes were generated for viewing. The first image plane was a conventional parasagittal plane running through the center of the OB ipsilateral to the injection site. The second plane was a curvilinear plane perpendicular to the parasagittal plane and running along the curved path of the RMS, enabling clear visualization of the lateral distribution of MPIOs about the RMS and in the OB. This second plane was generated by manually identifying points on the RMS path in the average images and then using tricubic interpolation to generate signal intensities at equidistant sites on the curved path.
To estimate the speed of MPIO movement from MR images, the location of hypointense voxels was compared on consecutive days in images from individual mice. The distance between the detected voxels furthest along the RMS was computed along the curved path of the RMS. An average was determined for the maximum speed between Days 0 and 1 and between Days 1 and 2 by calculating the speeds in each mouse and then averaging the results. This method tended to detect the fastest migrating cells and therefore produced an average of the maximum RMS migration rate.
On Days 2, 7 or 21 after MPIO injection, selected mice were prepared for histological analysis. For this purpose, mice received an intraperitoneal injection of nembutal (400 mg/kg), and were perfusion fixed through the left ventricle with 4 °C, 4% paraformaldehyde (PFA, Sigma-Aldrich, St. Louis, MO). After extraction from the skull and post-fixation (overnight in 4 °C, 4% PFA), brains were cryoprotected in 30% sucrose in phosphate buffered saline, mounted in embedding compound (Tissue-Tek OCT, Sakura Finetek USA Inc., Torrance, CA) and frozen into blocks. Sections were cut serially at 20 µm thickness on a cryostat (CM3050S, Leica Microsystems, Bannockburn, IL).
The Day 2 specimens were used to verify the MR detected particle distribution with histological counts of the distribution. In cryostat sections, the rostral–caudal distribution of particles was characterized by counting particles prior to any staining. Particles were visualized using a fluorescence dissection scope (Leica MZ16F, Leica Microsystems) and the total particle count in each section was recorded. Where sections were damaged or missing, a count was assigned based on the averages of the nearest 5 available sections. In this manner, a complete rostral–caudal particle count distribution was determined. In order to compare with MRI results (in which the voxel size was 100 µm), the counts from every five consecutive (20 µm) slices were summed to estimate the expected total particle count in serial 100 µm thick sections. A comparable metric was generated from the MRI images by counting the MRI hypointense voxels from planes running in the same rostral–caudal dimension. The image voxel and particle count distributions were then aligned by maximizing the correlation between the two distributions. By virtue of the registration of separate MR images, separate histological particle count distributions were also brought to a common space and, therefore, an average particle count distribution based on the histological sections was calculated and compared to the average hypointense voxel counts by MRI. For visualization of the histology sections, the histogram of the resultant images was manipulated to highlight the auto-fluorescent signal from the tissue, thus putting the much brighter MPIO fluorescence into an anatomical context.
Selected mice were injected with bromodeoxyuridine (BrdU, 100 mg/kg, Sigma) once per day for 3 days immediately following MPIO injection. BrdU is incorporated into cells during cell division, and marks dividing cells in the SVZ. Sections from these mice were visualized after staining for BrdU and for Nissl. Evaluation of these sections permitted visualization of the particle distribution in the SVZ at the ventricles and qualitative comparison of MPIO and BrdU distribution.
Serial sections were stained at different time points to identify which cell types contained MPIOs. For Day 2 images, Glial fibrillary acidic protein (GFAP, an astrocyte marker) was detected using a mouse anti-GFAP primary antibody (Chemicon, Millipore Corp., Billerica, MA). Likewise, doublecortin (Dcx, marker of migrating neuroblasts) was detected with goat anti-Dcx (Santa Cruz Biotechnology, Santa Cruz, CA). At later time points, staining was also performed with primary antibodies for Class III β-tubulin (Tuj1, immature neuronal marker) (Covance, Inc., Princeton, NJ) and Ionized calcium binding adaptor molecule 1 (Iba1, microglial marker) (Wako Chemicals, Inc., Richmond, VA). All immunostaining used IgG Cy3 secondary antibodies (Jackson ImmunoResearch Laboratories Inc., West Grove, PA), following protocols provided by the manufacturers.
Following staining, three independent readers scored each section for labeled cells by counting the total number of MPIOs per section and the subset localized within stained cells. Cell localization was determined by inspection based upon the proximity of MPIOs to stained cells. An average reader count for each section was computed and then the total for all sections was tabulated. Cell types were assessed at both Day 2 and Day 21. All counts included at least 3 different mice and a total of at least 500 and 85 MPIO particles per stain type for Day 2 and Day 21, respectively.