Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Aug 15.
Published in final edited form as: Neuroimage. 2017 Aug 16;197:657–667. doi: 10.1016/j.neuroimage.2017.08.046

Dominance of layer-specific microvessel dilation in contrast-enhanced high-resolution fMRI: Comparison between hemodynamic spread and vascular architecture with CLARITY

Alexander John Poplawsky a, Mitsuhiro Fukuda a,†, Bok-man Kang b,c, Jae Hwan Kim b, Minah Suh b,c, Seong-Gi Kim b,c,†
PMCID: PMC5815958  NIHMSID: NIHMS901298  PMID: 28822749

Abstract

Contrast-enhanced cerebral blood volume-weighted (CBVw) fMRI response peaks are specific to the layer of evoked synaptic activity (Poplawsky et al., 2015), but the spatial resolution limit of CBVw fMRI is unknown. In this study, we measured the laminar spread of the CBVw fMRI evoked response in the external plexiform layer (EPL, 265 ± 65 μm anatomical thickness, mean ± SD, n = 30 locations from 5 rats) of the rat olfactory bulb during electrical stimulation of the lateral olfactory tract and examined its potential vascular source. First, we obtained the evoked CBVw fMRI responses with a 55 × 55 μm2 in-plane resolution and a 500-μm thickness at 9.4 T, and found that the fMRI signal peaked predominantly in the inner half of EPL (136 ± 54 μm anatomical thickness). The mean full-width at half-maximum of these fMRI peaks was 347 ± 102 μm and the functional spread was approximately 100 or 200 μm when the effects of the laminar thicknesses of EPL or inner EPL were removed, respectively. Second, we visualized the vascular architecture of EPL from a different rat using a Clear Lipid-exchanged Anatomically Rigid Imaging/immunostaining-compatible Tissue hYdrogel (CLARITY)-based tissue preparation method and confocal microscopy. Microvascular segments with an outer diameter of <11 μm accounted for 64.3% of the total vascular volume within EPL and had a mean segment length of 55 ± 40 μm (n = 472). Additionally, vessels that crossed the EPL border had a mean segment length outside of EPL equal to 73 ± 61 μm (n = 28), which is comparable to half of the functional spread (50 – 100 μm). Therefore, we conclude that dilation of these microvessels, including capillaries, likely dominate the CBVw fMRI response and that the biological limit of the fMRI spatial resolution is approximately the average length of 1 – 2 microvessel segments, which may be sufficient for examining sublaminar circuits.

Keywords: layer-specific fMRI, olfactory bulb, cerebral blood volume, MION, capillary, pericyte

Introduction

The ability of functional magnetic resonance imaging (fMRI) to precisely map functional regional differences in the brain relies on how finely the blood supply is regulated by neuronal activity. Successful mapping of submillimeter-scale circuits in the brain, including functional modules and laminar connections, with hemodynamic-based functional imaging methods, such as optical imaging of intrinsic signal and fMRI, suggests that microvessels are responsible for their highly-specific functional signals (Cheng et al., 2001; Frostig et al., 1990; Fukuda et al., 2006; Goense et al., 2012; Hess et al., 2000; Moon et al., 2007; Poplawsky et al., 2015; Poplawsky and Kim, 2014; Sheth et al., 2004; Vanzetta et al., 2004; Wang and Roe, 2012; Xu et al., 2000; Xu et al., 2003; Zhao et al., 2006). However, few studies correlate such spatially specific functional signals to the underlying vasculature (Harrison et al., 2002). Therefore, it remains to be elucidated whether the underlying vascular architecture is consistent with the fMRI signal spread and whether increasing the spatial resolution of fMRI (i.e., smaller voxel sizes) has practical gains and neurophysiological meaning.

We have previously demonstrated that electrical stimulation of the lateral olfactory tract (LOT), which contains the axon bundles of the olfactory bulb output neurons, evokes dendrodendritic synaptic activities in the external plexiform layer (EPL) of the bulb and increases blood volumes that are well localized to the same layer using contrast agent-based cerebral blood volume-weighted (CBVw) fMRI (Poplawsky et al., 2015). In the present study, we determined the peak location of the LOT-stimulation evoked blood volume response in the bulb layers and its full width at half maximum (FWHM) to quantify the spread of the fMRI signal from the edges of EPL. Further, the vascular architecture of this layer was obtained from a different rat using Clear Lipid-exchanged Anatomically Rigid Imaging/immunostaining-compatible Tissue hYdrogel (CLARITY)-based methods to investigate its relationship to the CBVw laminar response. The number of segments, diameter, length, and volume of vessels in EPL were determined to examine whether the spread of the evoked CBVw fMRI responses were constrained by the dilation of microvessels specific to this layer.

Materials and Methods

General experimental design

We performed two separate experiments. First, with contrast agent-based fMRI, we measured the spatial spread of the CBVw fMRI response relative to the stimulated anatomical EPL, where synaptic activity is preferentially evoked during electrical stimulation of LOT. Second, we characterized the vasoarchitecture of EPL using CLARITY-based methods (Chung and Deisseroth, 2013; Chung et al., 2013) to determine the anatomical characteristics of the vessels that are likely dilating to produce the localized hemodynamic response.

Experimental Design #1: Laminar spread of evoked CBVw fMRI signals

Complete procedures on animal preparation, fMRI acquisition, and data processing can be found in (Poplawsky et al., 2015). Below is a brief description.

Animal preparation and stimulating electrode placement

Five male Sprague-Dawley rats (280 – 510 g, Charles River Laboratories International, Inc., Wilmington, MA, USA) were studied with approval from the University of Pittsburgh Institutional Animal Care and Use Committee. Rats were induced with 5% and maintained with 2% isoflurane gas during surgery. First, an Ag/AgCl reference electrode and a tungsten stimulating electrode were implanted to electrically activate the right LOT (Fig. 1A). Next, the right femoral artery and vein were catheterized for mean arterial blood pressure (MABP) monitoring (MP150, BioPac Systems Inc., Goleta, CA) and administration of 5% dextrose, anesthetic and contrast agent, respectively. Carprofen (4 – 5 mg/kg, s.q.) and atropine (0.05 – 0.07 mg/kg, i.m.) were administered at the beginning and end of surgery. Next, isoflurane was discontinued and α-chloralose was administered (45 mg/kg i.v. induction, 40 mg/kg/h i.v. maintenance) for functional studies. A 0.9% saline and 5% dextrose supplemental fluid was administered intravenously at 1.0 mL/kg/h. The MABP and the rectal temperature were maintained between 70 – 130 mmHg and at 37 ± 1°C (mean ± SD; all values are reported as mean ± SD unless otherwise specified hereafter), respectively. The rat freely breathed a mixture of 0.95 L/min medical air and 0.05 L/min O2 gases and the breathing rate was monitored with a pneumatic pillow sensor.

Figure 1. Experimental design.

Figure 1

Open in a new tab

A. We stereotaxically implanted a tungsten electrode and electrically stimulated (−200 μA, 200 μs pulse duration, 40 Hz) the lateral olfactory tract (LOT) at the anteroposterior position of Bregma (>6 mm posterior to the olfactory bulb (OB) where the fMRI study was performed). B. In the left panel, the six main layers of the bulb are clearly identified in an example fast-spin echo anatomical image (55 × 55 × 500 μm3). The glomerular layer (GL) and mitral cell layer (MCL) are identified as hypo-intense concentric bands and the bulb core as a hyper-intense strip; while the olfactory nerve layer (ONL), external plexiform layer (EPL), and granule cell layer (GCL) are identified by their spatial relations to these layers. In the right bulb, dashed lines represent the inner and outer boundaries of GL, the outer boundary of MCL, and core, respectively. The right panel shows a simplified cross-section of the main olfactory circuit taken from the white rectangle in the left panel. LOT contains the axons of mitral cells and is the main output of the bulb. Antidromic stimulation of LOT primarily evokes synaptic activity in EPL via dendrodendritic synapses with the apical dendrites of granule cells. In this study, we evaluate the evoked fMRI signal spread relative to the preferentially evoked anatomical layer EPL. CC, corpus callosum; AC, anterior commissure; D, dorsal; V, ventral; L, left; R, right; A, anterior; P, posterior; ORN, olfactory receptor neuron.

MRI data acquisition

MRI experiments were performed on a 9.4-T/31-cm MR system interfaced by a DirectDrive console (Agilent Tech, Santa Clara, CA) and an actively shielded gradient coil with 40-G/cm peak gradient strength and 120-μs rise time (Magnex, UK). The rat head was fixed in a non-magnetic head restraint with a bite bar and ear plugs. A custom-built 1-cm inner diameter surface coil was positioned dorsal to the olfactory bulb for radio-frequency excitation and reception.

Anatomical MRI

We acquired anatomical images using a fast spin-echo sequence with a 128 × 128 matrix size, 7 × 7 mm2 FOV (55 × 55 μm2 in-plane resolution), 3.0 s repetition time (TR), train of 4 echoes, 40.7 ms effective echo time (TE), 3 slices with a 0.5-mm thickness, and 24 averages.

CBVw fMRI data acquisition

We used a single i.v. bolus of Feraheme (15 mg Fe/kg, ferumoxytol, AMAG Pharmaceuticals, MA) for CBV-weighting and a compressed-sensing gradient recalled echo sequence to increase the temporal resolution by four times (Zong et al., 2014). No BOLD correction was applied to the CBVw fMRI signals since BOLD contributions are minimal (Poplawsky et al., 2015). We examined the spread of the CBVw signal at two spatial resolutions: 1) 64 × 64 matrix size, 110 × 110 μm2 in-plane, TE = 8 ms, 3 slices, TR = 125 ms, 2 s temporal resolution, and 64 dummy scans, and 2) 128 × 128 matrix size, 55 × 55 μm2 in-plane, TE = 10.18 ms minimum, 3 slices maximum, TR = 62.5 ms, 2 s temporal resolution, and 128 dummy scans; hereafter referred to as the 110-μm and 55-μm resolutions, respectively. We used the same FOV (7 × 7 mm2), slice thickness (0.5 mm), and slice positions as the anatomical images. We chose to acquire the three slices in the posterior bulb, where EPL is the thickest, to reduce partial volume effects caused by the limited resolution of fMRI. The phase encode under-sampling pattern for the 110-μm and 55-μm resolutions consisted of a constant sampling of the 6 and 12 center k-space lines, respectively, plus a random sampling of the remaining lines for each image (Zong et al., 2014). We determined the optimal flip angle by acquiring the baseline fMRI images at increasing flip angles and selecting the one that had the maximal signal within the olfactory bulb for each TR. Images were reconstructed using a k-t FOCUSS algorithm with a Karhunen-Loeve sparsifying transform (Jung et al., 2009; Jung et al., 2007; Lustig et al., 2006).

During LOT stimulation, we delivered a pulse train (-200 μA amplitude and 200 μs pulse width at 40 Hz) to the implanted monopolar stimulation electrode tip using an isolator (Isoflex, AMPI, Israel) equipped with an electrical pulse generator (Master 8, AMPI, Israel). A single fMRI block consisted of 60 baseline (120 s), 32 stimulus-evoked (64 s) and 60 recovery images (120 s), which were sequentially repeated six and 12 times for the 110-μm and 55-μm resolutions, respectively.

Functional maps

We spatially realigned, linearly detrended and calculated the normalized difference of the fMRI series ([St – S0]/S0, where S0 is the mean of repetition numbers 1–60 and 148–152) using home-written Matlab code (MathWorks, Natick, MA). Percent signal change and t-value maps were calculated from the concatenated CBVw fMRI blocks using a general linear model (GLM) analysis in SPM8 (Wellcome Trust Centre for Neuroimaging, London, UK), which used a predicted hemodynamic response function calculated by convolving the concatenated block-stimulation paradigm with a previously reported CBVw impulse response function (Silva et al., 2007). To determine the degree to which the partial volume effect was reduced at the higher resolution, the concatenated fMRI time series of the 55-μm resolution were spatially downsampled prior to the GLM analysis. Downsampling was achieved by averaging the four neighboring 55 × 55 μm2 voxels that corresponded to each 110 × 100 μm2 voxel; hereby referred to as the downsampled 110-μm resolution. The subsequent analysis of the downsampled fMRI data was then identical to the 110-μm resolution data. For each rat, these functional maps were co-registered to the individual anatomical images and linearly interpolated to a 128 × 128 matrix size to match the anatomical space, if needed.

Layer identification and line profiles

The following analyses were performed on an individual rat basis. MRI identification of bulb layers was previously verified with histological staining methods (Poplawsky et al., 2015); where the glomerular layer (GL) and mitral cell layer (MCL) were defined by the outer and inner bands of hypo-intense signal, respectively, and EPL by its spatial relationship between these bands (Fig. 1B). We manually drew the layer boundaries based on the high-resolution anatomical images of individual rats to: 1) measure the anatomical thickness of EPL, 2) identify the laminar location of the fMRI response peaks evoked by LOT stimulation, and 3) measure the FWHM of the fMRI peaks to determine the spread of the blood volume response beyond EPL. For these laminar analyses, straight lines were drawn from the left bulb through the right at a dorsoventral depth (approximately halfway between the dorsal and ventral extremes) where the layers ran orthogonal to the line. We chose this location because EPL is the thickest at this location and, thus, has reduced partial volume effects. Points were then averaged in the dorsoventral direction corresponding to a line width of 330 μm before plotting the mean percent signal change against distance. We finally identified the medial and lateral functional peaks and calculated their individual FWHM relative to a zero-baseline using linear interpolation between neighboring points.

Experimental Design #2: Vascular imaging using CLARITY

In the second experiment, we imaged the vasoarchitecture of an olfactory bulb section at a location similar to the fMRI line profiles to indirectly assess whether the source of the fMRI signal spread can be explained by characteristics of the microvasculature. We transcardially perfused one male Sprague-Dawley rat (400g, Orient Bio, Gyeonggi-do, South Korea) with saline at 10 mL/min and fixed it with 4% paraformaldehyde, which was not used in the previous fMRI experiments. We extracted the right olfactory bulb from the skull and post-fixed it in 4% paraformaldehyde at 4°C overnight with gentle agitation. Next, we chose a 1-mm thick coronal section at approximately the same anteroposterior position as the fMRI studies for tissue clearing using a modified CLARITY process (Chung and Deisseroth, 2013; Chung et al., 2013; Treweek et al., 2015). Specifically, we submerged the section in a hydrogel monomer solution (4% paraformaldehyde, 4% acrylamide, 0.25% azo-initiator in PBS) at 4°C overnight with gentle agitation to diffuse the hydrogel into the tissue. Hydrogels were polymerized and hydridized with the tissue using an Easy-Gel system (Live Cell Instrument, Seoul, South Korea) by heating the sample to 37°C while shaking it (60 rpm) under a negative pressure of −80 kPa. We ended the polymerization process when the viscosity of the hydrogel monomer solution increased to be honey-thick (~1.0 – 1.5 h) and removed the excess gel around the tissue section by rubbing with Kimwipes. Next, we placed the section in a 5-ml container with a clearing solution (4% sodium dodecyl sulfate, 20 mM tris-base, 200 mM boric acid, pH 8.5) and incubated it in a shaking water bath (100 rpm) at 37°C until the section was transparent. It took two weeks for clearing, during which time we replaced the clearing solution once per week. We washed the cleared section with 0.1% Triton X-100 in PBS (PBS-T) for one day at room temperature, while changing the bath three times with fresh PBS-T. To stain the vessels and cellular nuclei, we incubated the section with 0.5 mL of Tomato lectin (10 μg/mL in PBS-T, Vector laboratories, CA, USA) and Hoechst 33258 (20 μg/mL of PBS-T, Sigma-Aldrich, MO, USA), respectively, for three days at 450 rpm and room temperature using a ThermoMixer C (Eppendorf, Hamburg, Germany). We then washed the tissue for one day in PBS-T, while changing the bath three times with fresh PBS-T. To match the refractive index of the cleared tissue, we submerged the section in Easy-index solution (Live Cell Instrument, Seoul, South Korea) and rotated it by a tube rotator for over three hours. Finally, we imaged the 1-mm thick section with a DMi 6000 confocal microscope (Leica, Mannheim, Germany). We used a 10X objective with a working distance of 3,000 μm. The image matrix size (mediolateral axis × dorsoventral × anteroposterior) was 5518 × 8575 × 279 with a 4,177 × 6,491 × 673 μm3 field-of-view; achieving a voxel resolution of 0.76 × 0.76 × 2.41 μm3. Instead of staining vascular walls from the inside by perfusing an intravascular dye, we stained them from the outside using Tomato lectin because it provided a more complete staining of the vasculature in our preliminary examinations. However, because Tomato lectin 1) forms a concentration gradient as it penetrates deeper into the tissue from the bulb surface, 2) stains microglia (Acarin et al., 1994), and 3) may have an affinity for Caspr2 proteins concentrated in GL (Poliak et al., 1999), apparent dense staining in structures other than vessels are expected in outer bulb layers, such as ONL and GL (Fig. 4).

Figure 4. Vascular architecture of the olfactory bulb using CLARITY.

Figure 4

Open in a new tab

We measured characteristics of the bulb vasculature for comparison with fMRI studies. A. Maximum intensity projection of a coronal section of the right olfactory bulb (4,177 × 6,491 × 673 μm3, mediolateral axis × dorsoventral × anteroposterior) at approximately the same location as the fMRI studies. We prepared the tissue with CLARITY-based methods, stained the bulb vessels with Tomato lectin (red) and cellular nuclei with Hoechst 33258 (blue), and imaged them with confocal microscopy using a 10X objective and a 3,000-μm working distance. B. Cropped ROI at approximately the same location as the fMRI line profiles (expanded from the white rectangle in A, 2,184 × 500 × 200 μm3). Stained nuclei clearly identify the layers of interest and are described from left to right: The glomerular layer (GL, yellow box) has dense nuclear staining of periglomerular cells located at the boundaries of the spherical glomeruli, while the external plexiform layer (EPL, blue box) has sparse nuclear staining. The mitral cell layer (MCL) appears as the first vertical stripe of dense nuclear staining (immediately to the right of the second dashed line), immediately followed by the internal plexiform layer as the thin layer of sparse nuclear staining. The successive granule cell layer (GCL, green box) has a striated pattern of nuclear staining and the bulb core (cyan box) follows GCL with a less dense nuclear staining and absence of striations. Notice that this laminar pattern repeats in the reverse order to the right of the bulb core with the rightmost spherical patterns being lateral GL. C. Example 3D rendering of the lectin signal within the corresponding squares in B. D. Same as B, but with clearer visualization of the vasoarchitecture and segmentation of EPL (yellow vessels). E. The number of voxels containing EPL vessels by summation in the dorsoventral axis. D, dorsal; V, ventral; L, left; R, right; A, anterior; P, posterior.

Vessel analysis

After the image of the right bulb was acquired, we used IMARIS 8.2 software (Bitplane, Zurich, Switzerland) for cropping, Gaussian filtering, z-normalization, and binarization. Specifically, we cropped the image at approximately the same location as the fMRI line profiles (Fig. 4A–B, dashed line, 2,184 × 500 × 200 μm3). A 200-μm slice thickness was chosen due to the penetration depth limit of our confocal microscope and dye penetration. Therefore, to make our CLARITY image volume similar to the fMRI slab volume (i.e., ~2,000 × 330 × 500 μm3), the slice width was set to 500 μm. We used the Hoechst stain (Fig. 4B) to identify and segment GL, EPL, GCL, and core. As a preprocessing step (Fig. 5A), we applied a 3D Gaussian filter with a 2.41-μm FWHM for background blurring and homogenization of the vessel intensities. Then, to enhance the signal-to-noise ratio, we subtracted the blurred background by using the “subtract background” plugin in Fiji (ImageJ) (Schindelin et al., 2012; Schneider et al., 2012). Next, we normalized each z-slice to achieve a uniform level of brightness and contrast to correct for illumination variations and to reduce attenuation effects from the z-stacks. Specifically, we computed the mean and SD of the image intensity for each individual z-slice and that of the entire z-stack. We then adjusted the intensities of each z-slice to match the grand mean and SD of the entire z-stack using a linear transformation. We excluded background voxels with a zero intensity from this normalization process.

Figure 5. Overview of the image processing procedure.

Figure 5

Open in a new tab

A. Image processing diagram from the raw image to the quantification of the vasoarchitecture. The processing steps include intensity compensation with Gaussian blurring, background subtraction, and z-stack normalization. B. We applied various thresholds (THD) to binarize the images. Fixed thresholds of 40 and 70 (a.u.) of the image intensity were evaluated, in addition to a local variable threshold. C. Line profiles from the corresponding cross-sections in B, referenced by the lowercase letters. To best binarize vessels of many different diameters, the local contrast method was chosen for further analyses. D. Expanded view of a single ROI located in medial EPL. The pre-processed image had a better SNR compared to the raw image and, thus, vessel edges were easier to identify. The preprocessed image was rendered in 3D and binarized. Then, the image was skeletonized and branching (green/yellow dots) and termination points (blue/magenta dots) were identified using the 3D skeletonize plugin in Fiji. The vessel skeleton is overlaid on the pre-processed image to show the accuracy of this process (see also, Video 1). Vessels with termination points were removed from further analysis so that only whole segments were used in our quantifications, which will specifically affect the length calculations. Finally, we determined the segment diameters by the Euclidean distance transformation and calculated the segment volumes.

To further quantify the vascular structures, we generated a binary image using a local contrast thresholding method in the IMARIS software. As shown in Fig. 5B–C, an absolute threshold could over- or under-estimate the vessel diameter based on differences in peak intensity caused by light attenuation. Therefore, we used a Mexican Hat filter-based thresholding technique, termed “local contrast”, to determine the maximum contrast between the vessel and background intensities. Next, we rescaled the image to an isotropic voxel (1 × 1 × 1 μm3) for quantification. We applied two different algorithms to the binary images to acquire information on the vascular morphology. First, we generated a skeletonized image using the 3D skeletonize plugin in Fiji (Fig. 5D) and converted it into a network graph based on a 3 × 3 × 3-voxel neighborhood. In this calculation, we identified vessel branching points (Fig. 5D, “raw skeleton”, green/yellow points) and vessels that either leave EPL or the tissue section (termination (blue/magenta) points); and iteratively removed short vessel segments (< 10 voxels) that contained termination points to correct for falsely identified segments (Video 1). We then measured the vessel segment lengths as the distance between two vessel branching points, while excluding vessels that had termination points in order to include only whole segments in our vessel quantifications (Fig. 5D, “skeleton of whole segments”) (Kerschnitzki et al., 2013). Second, we calculated the vessel diameter by the Euclidean distance transformation of the fitted sphere that is located at the center point of the binary vessel image (Dougherty and Kunzelmann, 2007). We then calculated the number and volume of each segment based on graphical coordinates and a paired diameter map. Finally, we repeated this quantification for vessels that crossed the EPL border with GCL in order to further relate the vessel lengths that passed outside of EPL to the fMRI signal spread.

Results

To determine the spatial spread of the hemodynamic responses, we first acquired CBVw fMRI images from three coronal sections of the main olfactory bulb (Fig. 2A, 110 × 110 × 500 μm3) during LOT stimulation that evoked the thin, “ring-like” response. We then obtained line profiles of the evoked fMRI responses from 330-μm thick slabs that orthogonally transected the bulb layers (Fig. 3A, example slab (green rectangle) in Fig. 2A–B). Of the 30 peaks measured, the laminar fMRI response profiles had 25 peaks in EPL, four in the immediately adjacent MCL, and one in GCL. The mean FWHM of the fMRI response peaks was 454 ± 114 μm (n = 30 peaks from 5 rats, 3 slices each rat, 2 peaks each slice), whereas the mean anatomical thickness of EPL at these lines was 265 ± 65 μm. Thus, the fMRI spatial spread relative to the anatomical thickness of EPL was 190 ± 102 μm if we simply subtract the anatomical thickness from the corresponding peak FWHM. To better estimate the spatial spread of the fMRI signals beyond the stimulated layer, the FWHM measurements were plotted as a function of the anatomical EPL thickness, and the slope and intercept of the linear regression were determined. We found the intercept, which indicates the spread of the fMRI response after removing the effects of laminar thickness and approximates the spread of the fMRI signals beyond the stimulated layer, was 242 ± 80 μm (intercept ± SE, Fig. 3B, r2 = 0.21, p-value of linear regression = 0.011, df = 28). The slope of the regression line was 0.91 ± 0.24, which indicates a 1:1 relationship between the functional spread and EPL thickness (Fig. 3B).

Figure 2. LOT-evoked fMRI activation maps at 110-μm and 55-μm resolutions.

Figure 2

Open in a new tab

LOT was electrically stimulated to preferentially activate synapses in EPL. A–B. Three slices (anterior-most slice on the left) from a single, representative rat at (A) 110 × 110 × 500 μm3 and (B) 55 × 55 × 500 μm3 fMRI resolutions were compared to determine the spread of the fMRI response from the evoked layer EPL (p < 0.001 (voxel-wise) and minimum number of clustered active voxels equal to 12 for a family-wise error correction of p < 0.001). Activation maps were calculated in SPM8 after concatenating the 6 and 12 fMRI blocks for the 110-μm and 55-μm resolution images, respectively. A 330-μm thick slab (green rectangle in the 3rd slice) was drawn from the left to the right bulb extreme at a location in the dorsoventral axis (top-bottom axis in A–B) where a thick portion of EPL (inner and outer boundaries delineated by black or white dotted lines) runs orthogonal to the line. Residual negative CBVw responses are observed in the contralateral EPL (Poplawsky et al., 2015). Functional images are D, dorsal; V, ventral; L, left; R, right.

Figure 3. The CBV fMRI signal spreads ~100 μm from the preferentially evoked anatomical layer, EPL, at the 55-μm resolution.

Figure 3

Open in a new tab

A. Example line profiles and response full-width at half-maximum (FWHM) measurements (from the green rectangles in Fig. 2) for LOT stimulation-induced laminar fMRI response peaks in the medial and lateral portions of EPL (yellow bars) of the right bulb at each resolution. The smaller functional spread and increased peak amplitudes for the higher resolution line profiles provide evidence that the differences in FWHM between these resolutions are due to decreased in-plane partial volume effects. B–C. FWHM of LOT stimulation-induced laminar fMRI response peaks plotted against the corresponding anatomical thickness of EPL (n = 30 points for each resolution). Regression lines are solid and are bounded by the 95% confidence intervals indicated by dashed lines. Regression line intercepts, which approximate the fMRI signal spread beyond the corresponding anatomical thickness, were 242 ± 80 μm (slope = 0.80 ± 0.29, regression coefficient ± SE) and 106 ± 65 μm (slope = 0.91 ± 0.24) for the 110-μm and 55-μm resolutions, respectively. Blue arrows indicate profiles where peaks were in GL; green arrows in GCL. D. A differential analysis was performed to directly compare the relative reduction of the functional spreads between the 110-μm and 55-μm resolutions. The FWHM was subtracted from the corresponding EPL anatomical thickness on the y-axis to remove the regression slope (−0.09 ± 0.24 (slope ± SE) for the 55-μm resolution, −0.20 ± 0.29 for the 110-μm). Here an approximate twofold reduction of the fMRI spread for the high-resolution study is clearly observed, which is consistent with a twofold reduction in the partial volume effects in the direction of the line profile. E. In our 55-μm resolution line profiles and fMRI activation maps, we frequently observed the peak of fMRI activation in the inner portion of EPL. To quantify this, we subdivided EPL at approximately its middle depth (outer (o) and inner (i)EPL) and reported the location of each line profile’s maximum (n = 30 peaks). The majority of peaks (19 of 30) were located in iEPL.

Increasing the fMRI spatial resolution has the benefits of reducing partial volume effects and, more importantly, improving the spatial specificity if the finest vasculature control is less than 100 μm. However, increased resolution comes with the cost of a decreased signal-to-noise ratio and often requires increased averaging. To determine if increasing the spatial resolution has practical benefits for examining finer neurophysiological activity, we repeated these experiments at twice the in-plane resolution in the same rats (Fig. 2B, 55 × 55 × 500 μm3). Of the 30 peaks measured, the laminar fMRI response profiles had two peaks in GL, 23 peaks in EPL, four in MCL, and one in GCL. The mean FWHM of LOT-stimulation induced fMRI response peaks at the same anatomical locations as the lower in-plane resolution experiments was 347 ± 102 μm, resulting in a mean difference of 82 ± 83 μm from the corresponding EPL thicknesses. Similarly, the intercept of the regression analysis gave a functional spread of 106 ± 65 μm beyond EPL (intercept ± SE, Fig. 3C, r2 = 0.34, p-value of linear regression < 0.001, df = 28). To further determine the degree to which the improvement in the spatial spread between the images at the two resolutions can be explained by partial volume effects, the 55-μm resolution fMRI time series were downsampled to a 110-μm resolution and the regression analysis repeated. The intercept of the downsampled 110-μm resolution data was 212 ± 66 μm (r2 = 0.24, p = 0.007, df = 28), similar to the native 110-μm intercept (242 ± 80 μm). Therefore, the nearly twofold improvement in the spatial spread may be due to the twofold increase in the spatial resolution.

At the 55-μm resolution, qualitative inspection of the fMRI activation maps led us to suspect that the CBVw responses may be peaking in the inner half of EPL (iEPL) rather than the outer half (oEPL), see examples in Fig. 2A–B. Therefore, we subdivided EPL, using the anatomical images as references, at approximately its middle depth (oEPL = 129 ± 25 μm thick, iEPL = 136 ± 54 μm) and determined the layer where the peak fMRI response was found. We observed that 19 of 30 peaks were in iEPL, while only four peaks were in oEPL (Fig. 3E). If the preferential sublaminar activation in iEPL has neurophysiological origins, then the functional spread of the CBVw signal may be underestimated using the full thickness of EPL as a reference. Therefore, we repeated our differential and linear regression analyses using the thickness of iEPL as a reference. The mean differential spread and the regression intercept for the 55-μm resolution images were 211 ± 87 μm and 216 ± 45 μm (intercept ± SE, r2 = 0.26, p-value of linear regression < 0.001, df = 28), respectively. Therefore, depending on the anatomical reference, we determined an approximate 100 or 200 μm spread of the fMRI signal beyond EPL or iEPL, respectively.

Since the spread of the fMRI response was limited to 100 – 200 μm beyond the anatomical layer, we were interested in whether this spread was related to vascular constraints. In the second experiment, therefore, we examined the vascular architecture of the right olfactory bulb using CLARITY-based methods to render the brain transparent. We used confocal microscopy in a thin section of clarified bulb tissue to visualize the Tomato lectin-stained vessels and Hoechst-stained cellular nuclei, the latter to identify and segment the bulb layers (Fig. 4A–B). We chose a similar ROI as the fMRI line profiles (Fig. 4B, expanded view of the white rectangle in Fig. 4A). The main layers of the bulb can clearly be delineated from the Hoechst-stained image (blue). From left to right in Fig. 4B, GL is identified by peripheral staining of the spherical glomeruli, EPL by the successive volume with sparse staining, MCL by the first vertical stripe of dense staining, the internal plexiform layer by the successive vertical stripe of sparse staining, GCL by the striated islets of dense staining, and the bulb core by the centrally-located sparse staining. The vessel architecture is also layer-dependent, as can be clearly seen in the different views of the lectin-stained tissue with the Hoechst reference excluded (Fig. 4C–D). Specifically, the vasculature is very dense in the superficial GL and gradually diminishes with bulbar depth until the vessels become sparser in GCL, which is consistent to our previous blood volume MRI measurements (Poplawsky and Kim, 2014) and previously reported histological results (Borowsky and Collins, 1989). In addition, the direction of the vessels is qualitatively random in GL, EPL, and GCL, but collimate in the bulb core. It is also noted that the spherical glomeruli of GL appear in the vascular image, which is likely due to some unknown lectin staining that is nonspecific to the vessels (see methods).

To further relate our localized fMRI signal spread to the vascular characteristics, we evaluated the vessels of EPL for their number of segments, diameter, length, and volume using methods available in Fiji (Fig. 5A). First, we preprocessed the vascular image to achieve a uniform level of brightness and contrast between the vessels and background across the z-stack; and thresholded this image to achieve a binary image (Fig. 5B–C). Next, the preprocessed image was skeletonized and vessel branching (green/yellow) and termination (blue/magenta) points were identified (Fig. 5D). During this process, segments with less than 10 consecutive voxels (~17 μm) in length and vessels with termination points were removed from analysis to reduce falsely identified vessels and to include only whole vessel segments, respectively. Vessels with an outer diameter of <11 μm accounted for the majority of the total vascular volume (64.3%) and total number of segments (75.0%), which had a mean segment length of 55.3 ± 40.4 μm (n = 472 vessel segments) (Fig. 6). To evaluate whether our choice of the microvessel outer diameter threshold (<11 μm) greatly affected our mean vessel length, we also grouped vessels with <9 and <13 μm outer diameters. Here we determined that such vessels comprised 35.7 – 79.6% of the total vascular volume and 46.9 – 88.9% of the total number of segments, and had a mean segment length of 52 – 60 μm. The total blood volume of all vessels in EPL (Fig. 6) was 3.4% relative to the total laminar volume, and had a mean length of 50.5 ± 39.8 μm (n = 629 vessel segments). In addition, because vessels less than 10 voxels in length were removed, we evaluated how changing this threshold affected our mean length outcome for vessels with outer diameters of <11 μm. Specifically, we repeated our length quantifications using a wide range of thresholds (1 – 60 voxels), and determined that the mean length of these vessels in EPL is conservatively within the range of 40 – 70 μm. Finally, we separately quantified the vessels that crossed the boundary between EPL and MCL, while a similar quantification was not possible at the boundary between EPL and GL due to the nonspecific lectin staining in GL (see methods). Specifically, we were interested in the mean length of the vessels outside of EPL (segments outside of EPL, cyan; segments inside of EPL, blue in Fig. 7) since this vessel compartment will contribute to the fMRI spread beyond EPL. Vessels with outer diameters of <11 μm had a mean segment length of 72.6 ± 60.8 μm (n = 28 of 32 total crossing segments) outside of EPL, which is similar to the mean length of vessels within EPL. In addition, this mean length is comparable to the functional spread beyond one side of EPL or iEPL (~50 or 100 μm, respectively; representing half of the regression intercepts).

Figure 6. The length of microvessels in EPL approximates the spread of the CBVw fMRI signal.

Figure 6

Open in a new tab

Schematics of the vessel characteristics (1st column) plotted against vessel lengths (2nd column) and diameters (3rd column). A. Segment diameters and lengths (mean ± SEM). B. Volume of the different sized vessels normalized by the total volume of EPL. C. The number of vessels segments relative to the total number of segments in the layer. Microvessels (<11 μm outer diameter) make up the majority of the vascular volume (64.3% of total vascular volume) and segments (75.0%), and are likely the vascular compartment that our CBVw fMRI methods are most sensitive to. The average length of these vessels (55.3 ± 40.4 μm, mean ± SD, n = 472 vessel segments) is approximately 1 – 2 times the fMRI signal spread beyond EPL.

Figure 7. Vessels crossing the EPL border.

Figure 7

Open in a new tab

A. Binary vascular image with vessels that cross the boundary between EPL and MCL (white sheet). Vessels that cross the boundary between GL and EPL could not be included in this analysis due to nonspecific lectin staining in GL (see methods). The vessel portions within EPL are shown in blue, while the portions outside are shown in cyan. B–C. Alternate views of the lateral EPL vessels shown in A. A total of 32 vessels cross the medial and lateral EPL borders. The mean length of vessels outside of EPL that have outer diameters of <11 μm is 73 ± 61 μm (n = 28 vessels, cyan vessel segments), which is similar to the mean length of vessels that reside in EPL and to half of the fMRI signal spread beyond EPL.

Discussion

To determine the biological limits of the layer-specific CBVw fMRI spatial resolution, we compared the mean spatial spread of the LOT-evoked fMRI signal change (n = 5 rats), which is highly localized to a single layer of the rat olfactory bulb, to the vascular architecture of the same layer obtained from a different rat using CLARITY-based methods. First, we demonstrated a twofold improvement in the CBVw fMRI spatial spread when we increased the spatial resolution by the same factor. An agreement between the functional spread of the downsampled 110-μm images (212 ± 66 μm) and that of the native 110-μm resolution (242 ± 80 μm) corroborates that the majority of the FWHM improvement for the 55-μm resolution images could be due to reduced partial volume effects; although time dependent changes in the layer-specific synaptic activity and contrast agent concentrations should be considered as additional contributors. Because our fMRI signal spread seems to improve with the increased spatial resolution to a similar degree, further improvement of the hemodynamic spread with an increased fMRI resolution is possible; however severe SNR reductions could limit the feasibility of such a study. Second, we demonstrated that the CBVw fMRI signal spread approximately 50 – 100 μm beyond EPL, which was 1 – 2 times the average length of microvessels with outer diameters of <11 μm within and outside of EPL (40 – 73 μm). Although a more direct evaluation with greater rat numbers is needed, our range is similar to the mean capillary segment length range observed in other studies of the rat and cat cortex (56 – 112 μm) (Hudetz, 1997; Hudetz et al., 1993; Mironov et al., 1994; Motti et al., 1986; Pawlik et al., 1981; Weiss et al., 1982). This provides a possibility that microvessel-specific dilation is the dominating source of the neuronal specificity of CBVw fMRI, which is discussed below.

Microvessel regulation by synaptic activity dominates the specificity of CBVw fMRI

In the current study, we examined whether the anatomical architecture of vessels in EPL constrained the fMRI signal spread observed and, specifically, whether their average segment lengths were shorter than the fMRI signal spread. We are defining a vascular segment as the length of vessel between two branching points. It is conceivable that blood flow is regulated at the branching points where contractile muscles, such as arteriole smooth muscle or pericytes on capillaries, wrap around the vessel (Harrison et al., 2002). Therefore, if a point of neuronal activity is confined to one end of the vessel, then the dilation of the whole segment will extend the fMRI signal changes throughout the entire vessel length. This can be a significant contributor to nonspecific fMRI signal spread in addition to other sources, such as increased extravascular dephasing caused by dilation-induced increases in magnetic susceptibility gradients. The reported mean vessel lengths may serve as a conservative estimate for the potential source of the hemodynamic spread since the vessel pathways are not straight or perfectly orthogonal to the layers, but instead follow complex three-dimensional paths that will reduce the net distance a vessel travels in any single dimension (i.e., the left-right axis in our coronal bulb slices). Additionally, to report the fMRI signal spread, we chose to measure the conventional FWHM of the fMRI peaks because it does not assume the shape of the signal distribution (e.g., Gaussian or Lorentzian). But we assume that the distribution of the fMRI signal spread and the FWHM measurements are restrained by the vessel segment lengths, which needs to be directly evaluated by, for instance, two-photon microscopy.

To estimate how finely blood volume is regulated, we calculated the difference between the spread of the blood volume response evoked in EPL and the thickness of this layer. We found that the mean fMRI signal spread beyond EPL (50 – 100 μm) was approximately 1 – 2 times the mean length of microvessels with outer diameters of <11 μm within and outside of EPL (40 – 73 μm). The largest blood volume pool in EPL was also comprised of these microvessels (64.3%), which our contrast-enhanced blood volume fMRI measurements are most sensitive to, in addition to being less sensitive to nonspecific changes in draining veins and large pial vessels (Kim et al., 2013; Mandeville and Marota, 1999; Poplawsky et al., in press). However, it remains to be determined whether these microvessels are capillaries or arterioles. Since the microvessel lengths in EPL are within the range of mean capillary segment lengths (56 – 112 μm range) reported in the rat and cat cortex (Hudetz, 1997; Hudetz et al., 1993; Mironov et al., 1994; Motti et al., 1986; Pawlik et al., 1981; Weiss et al., 1982), these microvessels can be capillaries. However, it is debatable whether capillaries that lack smooth muscle cells can actively regulate blood flow in vivo (Hall et al., 2014; Hill et al., 2015). Specifically, Hill et al. (2015) estimated that blood flow regulation by smooth muscle cell contractility was on the order of a 75 – 150 μm radius in tissue. This is comparable to our hemodynamic spread when the effects of the anatomical thicknesses of EPL or iEPL were removed (100 – 200 μm) and the FWHM of the point spread functions of blood volume signals (103 – 175 μm) estimated by Vazquez et al. (2014) that determined a linear relationship between the spread of photo-stimulation induced neuronal activities and blood volume-weighted optical imaging of intrinsic signals in the mouse cerebral cortex. Thus, both passive and active capillary regulation of blood flow in vivo may be considered, especially by way of astrocyte or pericyte mediators for active regulation (Mishra et al., 2016), which can extend the vascular response beyond the length of a single vessel segment. Capillary regulation may also explain the localized velocity increase of red blood cells in GL capillaries during odor stimulation (Chaigneau et al., 2003; Shepherd, 2003). In conjunction with our previous observations in the bulb, where we measured peak CBVw fMRI responses localized to the same layer as the isolated synaptic activity (Poplawsky et al., 2015), our data suggest that the fMRI neuronal specificity is originating from the dilation of microvessels only within the vicinity of the evoked synaptic activities. Therefore, at the very least, vascular regulation points must be present near branching points of microvessels since diffuse activations spanning the layers were not observed. Consequently, the spatial specificity and resolution of CBVw fMRI are likely limited by the length of the microvessel segments, which are regulated by local neural activity.

Mapping sublaminar synaptic activity with CBVw fMRI

In our 55-μm resolution fMRI response profiles, 23 of 30 peaks were in EPL; of which 19 were in iEPL and four were in oEPL (Fig. 3E). The higher peak appearance frequency in iEPL may have neurophysiological sources. With our LOT stimulation, the anterior-posterior position of our stimulation electrode was near Bregma and likely stimulated mitral cells more than tufted cells (Igarashi et al., 2012). The lateral dendrites of these mitral cells branch perpendicularly from their apical dendrites and preferentially innervate iEPL, (Mori, 1987); thus, our particular location of stimulation may cause a preferential synaptic activation in this sublayer. In addition, our fMRI peaks may be mediated by a greater mitral cell synaptic density in iEPL (Matsuno et al., 2017) and other sublaminar differences in synaptic strengths (Bartel et al., 2015). It should also be noted that four peaks were in MCL, the thin layer (~50 μm thick) immediately below iEPL, which may simply be explained by partial volume effects, slight misregistrations between the functional and anatomical images, misidentification of bulb layers, or fMRI signal spread; or can have neurophysiological sources such as the retrograde activation of synapses on or near the mitral cell bodies. Alternatively, anterograde action potentials from LOT stimulation near Bregma will activate their neuronal targets in piriform cortex, which will disynaptically activate superficial granule cell synapses in GCL (Dennis and Beviss Kerr, 1968; Uva et al., 2006). Therefore, CBVw responses in GCL (see also Fig. 3B–C, one GCL peak indicated by green arrows), albeit being smaller than the more direct EPL targets from retrograde stimulation, may also have neurophysiological origins that could potentially shift the fMRI peak toward deeper layers. For the two fMRI peaks in GL (Fig. 3C, blue arrows, both observed in medial peaks of the same rat), LOT stimulation should also activate reciprocal synapses between periglomerular and mitral cells in GL due to active current propagation in dendrites (Bischofberger and Jonas, 1997; Chen et al., 1997; Urban and Castro, 2005), although retrograde LOT currents primarily activate reciprocal synapses between mitral and granule cells in EPL. However, electrophysiology and optical imaging with a voltage sensitive dye (Laaris et al., 2007; Wellis and Scott, 1990) suggest that synaptic responses evoked by LOT stimulation are negligible in GL. This is in agreement with most of our data in which LOT-stimulation did not evoke large blood volume changes in GL. Therefore, the GL peaks may be best explained by similar, non-neurophysiological, sources as described above since GL is immediately superficial to EPL. Nonetheless, with the greater frequency of sublaminar activations in iEPL compared to oEPL, it is interesting to speculate that synaptic events separated by very short distances can be differentiated with CBVw fMRI. However, more direct studies must be performed to definitively determine whether high-resolution fMRI is capable of measuring multiple nearby, but spatially distinct, neurophysiological events that are often characteristic of laminar circuits.

Technical consideration for the quantification of vessels using CLARITY-based methods

In the current report, we quantified vascular characteristics including the number of vessel segments and their diameters, lengths, and relative volumes. However, the tissue slab can expand by 10 – 30% following the process of rendering the tissue transparent (Epp et al., 2015). To mitigate this expansion, a refractive index matching medium (EasyIndex) is used to optically adjust the tissue to approximately its original size for imaging (Chung et al., 2013). In addition, the lectin that we used to visualize the blood vessels stains the vessel endothelial cells and represents the vessel outer (abluminal) diameter. Because the thickness of a typical endothelial cell is ~1 μm (Mishra et al., 2014), our measurements are likely overestimating the absolute vessel diameter by an additional ~2 μm. Therefore, the reported vessel outer diameters should not be considered as absolute values, but as estimations, and is the basis for classifying the largest vascular compartment (<11 μm) as microvessels rather capillaries or arterioles.

Supplementary Material

supplement. Video 1. Rotating view of the raw skeleton image.

To further show the accuracy of the skeletonization process at identifying vessel segments, the 3D “raw skeleton” image overlaid on the pre-processed vessel image (Fig. 5D) is rotated.

Download video file (43.8MB, mpg)

Acknowledgments

We thank Ping Wang for his experimental support. This work was supported by the National Institutes of Health (NS07391, MH18273, EB003324, and EB018903) and the Institute for Basic Science (IBS-R015-D1).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Acarin L, Vela JM, Gonzalez B, Castellano B. Demonstration of poly-N-acetyl lactosamine residues in ameboid and ramified microglial cells in rat brain by tomato lectin binding. J Histochem Cytochem. 1994;42(8):1033–1041. doi: 10.1177/42.8.8027523. [DOI] [PubMed] [Google Scholar]
  2. Bartel DL, Rela L, Hsieh L, Greer CA. Dendrodendritic synapses in the mouse olfactory bulb external plexiform layer. J Comp Neurol. 2015;523(8):1145–1161. doi: 10.1002/cne.23714. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bischofberger J, Jonas P. Action potential propagation into the presynaptic dendrites of rat mitral cells. J Physiol. 1997;504(2):359–365. doi: 10.1111/j.1469-7793.1997.359be.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Borowsky IW, Collins RC. Metabolic anatomy of brain: A comparison of regional capillary density, glucose metabolism, and enzyme activities. J Comp Neurol. 1989;288(3):401–413. doi: 10.1002/cne.902880304. [DOI] [PubMed] [Google Scholar]
  5. Chaigneau E, Oheim M, Audinat E, Charpak S. Two-photon imaging of capillary blood flow in olfactory bulb glomeruli. Proc Natl Acad Sci U S A. 2003;100(22):13081–13086. doi: 10.1073/pnas.2133652100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Chen WR, Midtgaard J, Shepherd GM. Forward and Backward Propagation of Dendritic Impulses and Their Synaptic Control in Mitral Cells. Science. 1997;278(5337):463–467. doi: 10.1126/science.278.5337.463. [DOI] [PubMed] [Google Scholar]
  7. Cheng K, Waggoner RA, Tanaka K. Human Ocular Dominance Columns as Revealed by High-Field Functional Magnetic Resonance Imaging. Neuron. 2001;32(2):359–374. doi: 10.1016/s0896-6273(01)00477-9. [DOI] [PubMed] [Google Scholar]
  8. Chung K, Deisseroth K. CLARITY for mapping the nervous system. Nat Methods. 2013;10(6):508–513. doi: 10.1038/nmeth.2481. [DOI] [PubMed] [Google Scholar]
  9. Chung K, Wallace J, Kim S-Y, Kalyanasundaram S, Andalman AS, Davidson TJ, Mirzabekov JJ, Zalocusky KA, Mattis J, Denisin AK, Pak S, Bernstein H, Ramakrishnan C, Grosenick L, Gradinaru V, Deisseroth K. Structural and molecular interrogation of intact biological systems. Nature. 2013;497(7449):332–337. doi: 10.1038/nature12107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dennis BJ, Beviss Kerr DI. An evoked potential study of centripetal and centrifugal connections of the olfactory bulb in the cat. Brain Res. 1968;11(2):373–396. doi: 10.1016/0006-8993(68)90032-2. [DOI] [PubMed] [Google Scholar]
  11. Dougherty R, Kunzelmann K-H. Computing Local Thickness of 3D Structures with ImageJ. Microsc Microanal. 2007;13:1678–1679. [Google Scholar]
  12. Epp JR, Niibori Y, Hsiang H-L, Mercaldo V, Deisseroth K, Josselyn SA, Frankland PW. Optimization of CLARITY for Clearing Whole-Brain and Other Intact Organs. eNeuro. 2015;2(3):e0022–0015. doi: 10.1523/ENEURO.0022-15.2015. 2015 0021–0015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Frostig RD, Lieke EE, Ts’o DY, Grinvald A. Cortical functional architecture and local coupling between neuronal activity and the microcirculation revealed by in vivo high-resolution optical imaging of intrinsic signals. Proc Natl Acad Sci U S A. 1990;87(16):6082–6086. doi: 10.1073/pnas.87.16.6082. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Fukuda M, Moon C-H, Wang P, Kim S-G. Mapping Iso-Orientation Columns by Contrast Agent-Enhanced Functional Magnetic Resonance Imaging: Reproducibility, Specificity, and Evaluation by Optical Imaging of Intrinsic Signal. J Neurosci. 2006;26(46):11821–11832. doi: 10.1523/JNEUROSCI.3098-06.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Goense J, Merkle H, Logothetis NK. High-Resolution fMRI Reveals Laminar Differences in Neurovascular Coupling between Positive and Negative BOLD Responses. Neuron. 2012;76(3):629–639. doi: 10.1016/j.neuron.2012.09.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hall CN, Reynell C, Gesslein B, Hamilton NB, Mishra A, Sutherland BA, O’Farrell FM, Buchan AM, Lauritzen M, Attwell D. Capillary pericytes regulate cerebral blood flow in health and disease. Nature. 2014;508(7494):55–60. doi: 10.1038/nature13165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Harrison RV, Harel N, Panesar J, Mount RJ. Blood Capillary Distribution Correlates with Hemodynamic-based Functional Imaging in Cerebral Cortex. Cereb Cortex. 2002;12(3):225–233. doi: 10.1093/cercor/12.3.225. [DOI] [PubMed] [Google Scholar]
  18. Hess A, Stiller D, Kaulisch T, Heil P, Scheich H. New Insights into the Hemodynamic Blood Oxygenation Level-Dependent Response through Combination of Functional Magnetic Resonance Imaging and Optical Recording in Gerbil Barrel Cortex. J Neurosci. 2000;20(9):3328–3338. doi: 10.1523/JNEUROSCI.20-09-03328.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hill RA, Tong L, Yuan P, Murikinati S, Gupta S, Grutzendler J. Regional Blood Flow in the Normal and Ischemic Brain Is Controlled by Arteriolar Smooth Muscle Cell Contractility and Not by Capillary Pericytes. Neuron. 2015;87(1):95–110. doi: 10.1016/j.neuron.2015.06.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hudetz AG. Blood Flow in the Cerebral Capillary Network: A Review Emphasizing Observations with Intravital Microscopy. Microcirculation. 1997;4(2):233–252. doi: 10.3109/10739689709146787. [DOI] [PubMed] [Google Scholar]
  21. Hudetz AG, Greene AS, Fehér G, Knuese DE, Cowley AW., Jr Imaging System for Three-Dimensional Mapping of Cerebrocortical Capillary Networks in Vivo. Microvascular Research. 1993;46(3):293–309. doi: 10.1006/mvre.1993.1054. [DOI] [PubMed] [Google Scholar]
  22. Igarashi KM, Ieki N, An M, Yamaguchi Y, Nagayama S, Kobayakawa K, Kobayakawa R, Tanifuji M, Sakano H, Chen WR, Mori K. Parallel Mitral and Tufted Cell Pathways Route Distinct Odor Information to Different Targets in the Olfactory Cortex. J Neurosci. 2012;32(23):7970–7985. doi: 10.1523/JNEUROSCI.0154-12.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Jung H, Sung K, Nayak KS, Kim EY, Ye JC. k-t FOCUSS: A general compressed sensing framework for high resolution dynamic MRI. Magn Reson Med. 2009;61(1):103–116. doi: 10.1002/mrm.21757. [DOI] [PubMed] [Google Scholar]
  24. Jung H, Ye JC, Kim EY. Improved k – t BLAST and k – t SENSE using FOCUSS. Phys Med Biol. 2007;52(11):3201. doi: 10.1088/0031-9155/52/11/018. [DOI] [PubMed] [Google Scholar]
  25. Kerschnitzki M, Kollmannsberger P, Burghammer M, Duda GN, Weinkamer R, Wagermaier W, Fratzl P. Architecture of the osteocyte network correlates with bone material quality. J Bone Miner Res. 2013;28(8):1837–1845. doi: 10.1002/jbmr.1927. [DOI] [PubMed] [Google Scholar]
  26. Kim S-G, Harel N, Jin T, Kim T, Lee P, Zhao F. Cerebral blood volume MRI with intravascular superparamagnetic iron oxide nanoparticles. NMR Biomed. 2013;26:949–962. doi: 10.1002/nbm.2885. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Laaris N, Puche A, Ennis M. Complementary Postsynaptic Activity Patterns Elicited in Olfactory Bulb by Stimulation of Mitral/Tufted and Centrifugal Fiber Inputs to Granule Cells. J Neurophysiol. 2007;97(1):296–306. doi: 10.1152/jn.00823.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Lustig M, Santos J, Donoho D, Pauly J. kt SPARSE: high frame rate dynamic MRI exploiting spatio-temporal sparsity. Proc Intl Soc Mag Reson Med. 2006;14:2420. [Google Scholar]
  29. Mandeville JB, Marota JJA. Vascular filters of functional MRI: Spatial localization using BOLD and CBV contrast. Magn Reson Med. 1999;42(3):591–598. doi: 10.1002/(sici)1522-2594(199909)42:3<591::aid-mrm23>3.0.co;2-8. [DOI] [PubMed] [Google Scholar]
  30. Matsuno T, Kiyokage E, Toida K. Synaptic distribution of individually labeled mitral cells in the external plexiform layer of the mouse olfactory bulb. J Comp Neurol. 2017;525(7):1633–1648. doi: 10.1002/cne.24148. [DOI] [PubMed] [Google Scholar]
  31. Mironov V, Hritz MA, LaManna JC, Hudetz AG, Harik SI. Architectural alterations in rat cerebral microvessels after hypobaric hypoxia. Brain Res. 1994;660(1):73–80. doi: 10.1016/0006-8993(94)90840-0. [DOI] [PubMed] [Google Scholar]
  32. Mishra A, O’Farrell FM, Reynell C, Hamilton NB, Hall CN, Attwell D. Imaging pericytes and capillary diameter in brain slices and isolated retinae. Nat Protoc. 2014;9(2):323–336. doi: 10.1038/nprot.2014.019. [DOI] [PubMed] [Google Scholar]
  33. Mishra A, Reynolds JP, Chen Y, Gourine AV, Rusakov DA, Attwell D. Astrocytes mediate neurovascular signaling to capillary pericytes but not to arterioles. Nat Neurosci. 2016;19(12):1619–1627. doi: 10.1038/nn.4428. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Moon C-H, Fukuda M, Park S-H, Kim S-G. Neural Interpretation of Blood Oxygenation Level-Dependent fMRI Maps at Submillimeter Columnar Resolution. J Neurosci. 2007;27(26):6892–6902. doi: 10.1523/JNEUROSCI.0445-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Mori K. Membrane and synaptic properties of identified neurons in the olfactory bulb. Prog Neurobiol. 1987;29(3):275–320. doi: 10.1016/0301-0082(87)90024-4. [DOI] [PubMed] [Google Scholar]
  36. Motti EDF, Imhof H-G, Yaşargil MG. The terminal vascular bed in the superficial cortex of the rat. Journal of Neurosurgery. 1986;65(6):834–846. doi: 10.3171/jns.1986.65.6.0834. [DOI] [PubMed] [Google Scholar]
  37. Pawlik G, Rackl A, Bing RJ. Quantitative capillary topography and blood flow in the cerebral cortex of cats: an in vivo microscopic study. Brain Res. 1981;208(1):35–58. doi: 10.1016/0006-8993(81)90619-3. [DOI] [PubMed] [Google Scholar]
  38. Poliak S, Gollan L, Martinez R, Custer A, Einheber S, Salzer JL, Trimmer JS, Shrager P, Peles E. Caspr2, a new member of the neurexin superfamily, is localized at the juxtaparanodes of myelinated axons and associates with K+ channels. Neuron. 1999;24(4):1037–1047. doi: 10.1016/s0896-6273(00)81049-1. [DOI] [PubMed] [Google Scholar]
  39. Poplawsky AJ, Fukuda M, Kim S-G. Foundations of layer-specific fMRI and investigations of neurophysiological activity in the laminarized neocortex and olfactory bulb of animal models. Neuroimage, Advance online publication. doi: 10.1016/j.neuroimage.2017.05.023. in press. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Poplawsky AJ, Fukuda M, Murphy M, Kim S-G. Layer-Specific fMRI Responses to Excitatory and Inhibitory Neuronal Activities in the Olfactory Bulb. J Neurosci. 2015;35(46):15263–15275. doi: 10.1523/JNEUROSCI.1015-15.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Poplawsky AJ, Kim S-G. Layer-dependent BOLD and CBV-weighted fMRI responses in the rat olfactory bulb. Neuroimage. 2014;91:237–251. doi: 10.1016/j.neuroimage.2013.12.067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, Rueden C, Saalfeld S, Schmid B, Tinevez J-Y, White DJ, Hartenstein V, Eliceiri K, Tomancak P, Cardona A. Fiji: an open-source platform for biological-image analysis. Nat Methods. 2012;9(7):676–682. doi: 10.1038/nmeth.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nat Methods. 2012;9(7):671–675. doi: 10.1038/nmeth.2089. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Shepherd GM. The single capillary and the active brain. Proc Natl Acad Sci U S A. 2003;100(22):12535–12536. doi: 10.1073/pnas.2336089100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Sheth SA, Nemoto M, Guiou M, Walker M, Pouratian N, Hageman N, Toga AW. Columnar Specificity of Microvascular Oxygenation and Volume Responses: Implications for Functional Brain Mapping. J Neurosci. 2004;24(3):634–641. doi: 10.1523/JNEUROSCI.4526-03.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Silva AC, Koretsky AP, Duyn JH. Functional MRI impulse response for BOLD and CBV contrast in rat somatosensory cortex. Magn Reson Med. 2007;57(6):1110–1118. doi: 10.1002/mrm.21246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Treweek JB, Chan KY, Flytzanis NC, Yang B, Deverman BE, Greenbaum A, Lignell A, Xiao C, Cai L, Ladinsky MS, Bjorkman PJ, Fowlkes CC, Gradinaru V. Whole-body tissue stabilization and selective extractions via tissue-hydrogel hybrids for high-resolution intact circuit mapping and phenotyping. Nat Protoc. 2015;10(11):1860–1896. doi: 10.1038/nprot.2015.122. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Urban NN, Castro JB. Tuft Calcium Spikes in Accessory Olfactory Bulb Mitral Cells. J Neurosci. 2005;25(20):5024–5028. doi: 10.1523/JNEUROSCI.0297-05.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Uva L, Strowbridge BW, de Curtis M. Olfactory bulb networks revealed by lateral olfactory tract stimulation in the in vitro isolated guinea-pig brain. Neuroscience. 2006;142(2):567–577. doi: 10.1016/j.neuroscience.2006.06.047. [DOI] [PubMed] [Google Scholar]
  50. Vanzetta I, Slovin H, Omer DB, Grinvald A. Columnar Resolution of Blood Volume and Oximetry Functional Maps in the Behaving Monkey: Implications for fMRI. Neuron. 2004;42(5):843–854. doi: 10.1016/j.neuron.2004.04.004. [DOI] [PubMed] [Google Scholar]
  51. Vazquez AL, Fukuda M, Crowley JC, Kim S-G. Neural and Hemodynamic Responses Elicited by Forelimb- and Photo-stimulation in Channelrhodopsin-2 Mice: Insights into the Hemodynamic Point Spread Function. Cereb Cortex. 2014;24(11):2908–2919. doi: 10.1093/cercor/bht147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Wang Z, Roe AW. Columnar Specificity of Microvascular Oxygenation and Blood Flow Response in Primary Visual Cortex: Evaluation by Local Field Potential and Spiking Activity. J Cereb Blood Flow Metab. 2012;32(1):6–16. doi: 10.1038/jcbfm.2011.152. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Weiss HR, Buchweitz E, Murtha TJ, Auletta M. Quantitative regional determination of morphometric indices of the total and perfused capillary network in the rat brain. Circulation Research. 1982;51(4):494–503. doi: 10.1161/01.res.51.4.494. [DOI] [PubMed] [Google Scholar]
  54. Wellis DP, Scott JW. Intracellular responses of identified rat olfactory bulb interneurons to electrical and odor stimulation. J Neurophysiol. 1990;64(3):932–947. doi: 10.1152/jn.1990.64.3.932. [DOI] [PubMed] [Google Scholar]
  55. Xu F, Kida I, Hyder F, Shulman RG. Assessment and discrimination of odor stimuli in rat olfactory bulb by dynamic functional MRI. Proc Natl Acad Sci U S A. 2000;97(19):10601–10606. doi: 10.1073/pnas.180321397. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Xu F, Liu N, Kida I, Rothman DL, Hyder F, Shepherd GM. Odor maps of aldehydes and esters revealed by functional MRI in the glomerular layer of the mouse olfactory bulb. Proc Natl Acad Sci U S A. 2003;100(19):11029–11034. doi: 10.1073/pnas.1832864100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Zhao F, Wang P, Hendrich K, Ugurbil K, Kim S-G. Cortical layer-dependent BOLD and CBV responses measured by spin-echo and gradient-echo fMRI: Insights into hemodynamic regulation. Neuroimage. 2006;30(4):1149–1160. doi: 10.1016/j.neuroimage.2005.11.013. [DOI] [PubMed] [Google Scholar]
  58. Zong X, Lee J, Poplawsky AJ, Kim S-G, Ye JC. Compressed sensing fMRI using gradient-recalled echo and EPI sequences. Neuroimage. 2014;92:312–321. doi: 10.1016/j.neuroimage.2014.01.045. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplement. Video 1. Rotating view of the raw skeleton image.

To further show the accuracy of the skeletonization process at identifying vessel segments, the 3D “raw skeleton” image overlaid on the pre-processed vessel image (Fig. 5D) is rotated.

Download video file (43.8MB, mpg)

RESOURCES