SEARCH

SEARCH BY CITATION

Keywords:

  • Neurogenesis;
  • Extracellular matrix;
  • Heparan sulfate proteoglycan;
  • Basement membrane;
  • Meninges Basic fibroblast growth factor

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

The novel extracellular matrix structures called fractones are found in the lateral ventricle walls, the principal adult brain stem cell niche. By electron microscopy, fractones were shown to contact neural stem and progenitor cells (NSPC), suggesting a role in neurogenesis. Here, we investigated spatial relationships between proliferating NSPC and fractones and identified basic components and the first function of fractones. Using bromodeoxyuridine (BrdU) for birth-dating cells in the adult mouse lateral ventricle wall, we found most mitotic cells next to fractones, although some cells emerged next to capillaries. Like capillary basement membranes, fractones were immunoreactive for laminin β1 and γ1, collagen IV, nidogen, and perlecan, but not laminin-α1, in the adult rat, mouse, and human. Intriguingly, N-sulfate heparan sulfate proteoglycan (HSPG) immunoreactivity was restricted to fractone subpopulations and infrequent subependymal capillaries. Double immunolabel for BrdU and N-sulfate HSPG revealed preferential mitosis next to N-sulfate HSPG immunoreactive fractones. To determine whether N sulfate HSPG immunoreactivity within fractones reflects a potential for binding neurogenic growth factors, we identified biotinylated fibroblast growth factor 2 (FGF-2) binding sites in situ on frozen sections, and in vivo after intracerebroventricular injection of biotinylated FGF-2 in the adult rat or mouse. Both binding assays revealed biotinylated FGF-2 on fractone subpopulations and on infrequent subependymal capillaries. The binding of biotinylated FGF-2 was specific and dependent upon HSPG, as demonstrated in vitro and in vivo by inhibition with heparatinase and by the concomitant disappearance of N-sulfate HSPG immunoreactivity. These results strongly suggest that fractones promote growth factor activity in the neural stem cell niche.

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


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

Neural stem and progenitor cells (NSPC) produce new neurons and glia throughout adulthood in restricted zones of the brain, termed niches. The most active neurogenic niche, and the richest source of NSPC, is located in the subependymal layer (SEL) of the lateral ventricle (LV) [1, [2], [3]4], but secondary neurogenic zones exist, such as the granular layer of the dentate gyrus [5]. NSPC proliferate and differentiate in the niches in response to growth factors (GF) [4, 6, [7], [8], [9]10], but extracellular matrix (ECM) molecules may also intervene in neurogenesis [5, 11, [12]13]. However, the mechanisms that capture and present GF to NSPC for orchestrating neurogenesis are unknown.

We recently characterized extracellular structures in the ventricle walls, in direct contact with NSPC in the adult neurogenic niche [14]. These structures appear in confocal microscopy as series of laminin-immunoreactive (-ir) puncta aligned along the SEL. However, transmission electron microscopy revealed that each punctum is a branched fractal structure that we termed a fractone [14, 15]. The branched structure allows each fractone to be the target of converging NSPC processes in a limited space.

Although they have a distinctive morphology, fractones ultrastructurally resemble basement membranes (basal laminae) in their electron density [14]. Immunoreactivity for laminin further suggests that fractones are affiliated with basement membranes [14, 15]. Basement membranes are dense mats of ECM molecules intervening in cell proliferation, differentiation, and morphogenesis during development [16, [17]18], but their role during adulthood is unclear [12]. Moreover, basement membranes contain heparan sulfate proteoglycans (HSPG), which capture, activate, and target activated GF to cell surface receptors [19, [20], [21]22].

We hypothesize that fractones represent the core of the NSPC niche, regulating neurogenesis via ECM/GF interactions at the surface contact between NSPC processes and fractones. Here, we investigated the anatomical relationships between proliferating cells and fractones in the neurogenic niche. To determine whether fractones are affiliated with basement membranes, we explored fractone immunoreactivity for diverse ECM components in different adult mammalian species. To identify fractone functions in the NSPC niche, we investigated the binding of fibroblast growth factor 2 (FGF-2), a powerful neurogenic GF that requires HSPG as cofactors [19, 20, 23].

Our data indicated that (a) fractones exist in adult mice and adult humans and consist of the usual basement membrane components plus unique N-sulfate HSPG; (b) cells in the NSPC niche proliferate next to fractones, and particularly next to fractones that are immunoreactive for N-sulfate HSPG; and (c) fractones and some SEL blood vessels are the exclusive brain structures that capture FGF-2 from the extracellular milieu. These results suggest that fractones promote heparin-binding GF activity and cell proliferation in the neural stem cell niche.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

Animal and Human Tissues

Human brains were collected fresh, 4 hours after death, from two adult human cadavers, one 79-year old female and one 76-year old male. Frozen sections generated from pieces of the LV and spinal cord segments were fixed with either 4% paraformaldehyde (PFA) or −20°C cooled acetone. Male and female Balb/c mice (4–8 weeks old; n = 24) and Holtzman Sprague-Dawley male rats (4–10 weeks old; n = 8) were anesthetized with ketamine/xylazine (100 mm/kg + 15 mg/kg of body weight) for in vivo studies or directly euthanized with 300 mg/kg + 45 mg/kg ketamine/xylazine mix. All animal experimental protocols followed NIH guidelines and were approved by the Institutional Animal Care and Use Committee at the University of Hawaii.

Intracerebroventricular and Intraperitoneal Injection of Bromodeoxyuridine in the Adult Mouse

Bromodeoxyuridine (BrdU) (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) diluted in artificial cerebrospinal fluid (CSF) (Harvard Apparatus, Halliston, MA, http://www.harvardapparatus.com) was i.p. (200 μl of 10 mg/ml) or intracerebroventricular (ICV) (2 μl of 20 mg/ml) injected in four adult mice to determine the distribution of mitotic cells and fractones. Animals that were i.p. and ICV injected showed similar BrdU immunoreactivity. For ICV injections, craniotomy was performed upon stereomicroscopy using 0.5-mm burrs at the stereotaxic coordinates bregma −0.2 mm, 0.6 mm lateral from the midline. Thirty-gauge needle tips mounted on Mikrolitterpritzen syringes were positioned at 2 mm from the skull surface prior to 15-minute BrdU injection assisted with an Ultramicropump II/Micro-4 (WPI, Sarasota, FL) in the LV.

Immunocytochemistry

Immunocytochemistry was processed on PFA-fixed or acetone-fixed frozen sections of adult human, rat, or mouse brains (as described in [24, 38]). Except for significant differences with the anti-laminin antibody (L9393; 1/1,000; Sigma-Aldrich), which served as a reference antibody for visualizing fractones [14, 15], immunolabels did not significantly differ with the mode of fixation. The list of other primary antibodies against ECM molecules is as follows: laminin γ1 (monoclonal antibody [MAb] 1920; 1/1,000); laminin β1 (MAb 1928; 1/500), collagen-IV, α2 chain (MAb 1910; 1/200); perlecan (MAb 1948; clone A7L6); anti-HSPG (MAb 2040; 1/400); entactin, G2 domain (MAb 1886; clone JF6; 1/500; Chemicon, Temecula, CA, http://www.chemicon.com), anti-human collagen-IV (c-1926; 1/250; Sigma-Aldrich); laminin α1 (a gift of Takako Sasaki; 1/300); and anti-N-sulfate glycosamines (10E4; 1/300; Seikagaku Co., East Falmouth, MA, http://www.acciusa.com/bio/customer_service.html). Proliferating cells were visualized with anti-BrdU antibodies (OBT0030; 1/500; Oxford Biotechnology, U.K., http://www.immunologicalsdirect.com). Ependymal cells were immunodetected with anti-vimentin (MAb IF01; 1/50; Calbiochem, Cambridge, MA, http://www.emdbiosciences.com). The secondary antibody detection systems used for immunolocalization were as follows: donkey anti-rabbit conjugated to fluorescein isothiocyanate (N1034vs; 1/100), sheep anti-mouse biotin (RPN1001; 1/250) followed by streptavidin-Texas Red (RPN1233; 1/250; Amersham Biosciences, Piscataway, NJ, http://www.amersham.com), Cy-5 goat anti-rat (62-9516; 1/200; Zymed; Invitrogen, San Francisco, http://www.invitrogen.com) and Alexa-Fluor 546 goat anti-rat (1/500; Molecular Probes, Carlsbad, CA, http://www.probes.invitrogen.com). Immunolabeling controls were performed by omitting primary antibodies in the immunocytochemistry procedure. Immunocytochemistry involving BrdU label was performed as followed: frozen sections were fixed in cold acetone (when using anti-laminin L9393) or 4% PFA (when using anti-N-sulfate HSPG 10E4), incubated with anti-laminin (L9393) or 10E4 and then anti-rabbit-fluorescein isothiocyanate (FITC) (Amersham Biosciences) or Alexa-488 goat anti-mouse (Molecular Probes; Invitrogen) prior to postfixation with 4% PFA for 10 minutes, and incubation with 2 N HCl at 37°C for 30 minutes and then 0.1 M sodium borate buffer for 10 minutes. The sections were then incubated with rat anti-BrdU (1/500) at 4°C overnight prior to incubation with Alexa-546 goat anti-rat (1/500; Molecular Probes), washing, and mounting in Vectashield mounting medium (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com).

FGF-2 Biotinylation

Biotinylation of FGF-2 was carried out by incubating 2 μg of biotin (EZ-link Micro-sulfo-NHS Biotinylation kit, 21425; Pierce, Rockford, IL, http://www.piercenet.com) with 25 μg of human recombinant FGF-2 (Chemicon, Invitrogen, or Peprotech [Rocky Hill, NJ, http://www.peprotech.com]) in 200 μl of phosphate-buffered saline, pH 7.4, for 2 hours on a rotating plate. The reaction was stopped by adding 20 μg of glycine at 4°C overnight.

In Situ Binding of Biotinylated FGF-2

Frozen sections (30 μm thick) were fixed in 4% PFA for 5 minutes prior to preincubation with 50 mM Tris-HCl, pH 7.5, for 15 minutes and incubation with 5 μg/ml biotinylated-FGF-2 and anti-laminin (L9393; Sigma-Aldrich) or perlecan for 2 hours, and they were subsequently revealed by incubation with streptavidin-Texas Red (1/250; Amersham Biosciences) and appropriate secondary antibodies for 40 minutes. Results were recorded by confocal microscopy.

ICV Injection and Revelation of Biotinylated FGF-2 in the Adult Mouse or Rat

ICV injection was performed as described above, with biotinylated FGF-2 diluted in artificial CSF. Craniotomy coordinates were bregma +0.2 or −0.2 mm and 0.6 mm lateral from the midline for mice, and bregma +0.5 mm and 1.5 mm lateral to the midline for rats. Over a period of 15 minutes, 0.5 μl (mouse) or 2 μl (rat) of 0.125 μg/μl of biotinylated FGF-2 was injected. Biotinylated FGF-2 binding in vivo was revealed on brain sections with streptavidin-Texas Red.

Digestion of Heparin Sulfate Chains by Heparanase I/Heparatinase I for FGF-2 Binding Assays

For in vitro experiments, frozen sections were fixed with 4% PFA, incubated in 50 mM Tris-HCl, pH 7.4, at 37°C for 15 minutes prior to 20 mU/ml of heparatinase one (100704; Seikagaku) or heparanase I/heparatinase I mix (Sigma-Aldrich) in Tris-HCl, pH 7.4, for 2 hours at 37°C. The frozen sections were then incubated with biotinylated FGF-2, 10E4, and perlecan (1/200; Chemicon) antibodies for 2 hours prior to incubation with secondary antibodies (Alexa-488 goat anti-mouse IgM, 1/200; and Alexa-647 goat anti-rat IgG; 1/300; Molecular Probes) and streptavidin-Texas Red. Heparatinase I or heparanase I/heparatinase I mix treatments yielded similar results. For in vivo experiments, 10 mU/μl of heparanase I/heparatinase I mix (Sigma-Aldrich) was injected ICV 3 hours before ICV injection of biotinylated FGF-2.

Epifluorescence, Confocal Laser Scanning, and Transmission Electron Microscopy

Micrographs were recorded with a Zeiss Axiocam digital camera mounted on an Axioskop microscope, a Zeiss confocal laser scanning system mounted on an Axiovert 200 microscope, or a Zeiss LSM 510 confocal laser scanning system mounted on an Axiovert 100 microscope (Carl Zeiss, Jena, Germany, http://www.zeiss.com). Confocal images were obtained at a resolution of 2,048 × 2,048 pixels using ×20 or ×63 oil objectives. The green, red, and magenta/blue colors displayed on photomicrographs indicate that labeling was revealed with FITC, Texas Red/Alexa 548, or Cy-5/Alexa 647 fluorochromes, respectively. For ultrastructural studies, adult rat brain samples were processed according to our published protocol [24]. All images were processed with Photoshop software (version 7.1; Adobe Systems, Mountain View, CA, http://www.adobe.com). Software adjustments for brightness and contrast were not made or were minimal.

Statistical Methods

Mean differences between groups were compared using a standard t test in the case of equal variances. The Behrens-Fisher method [25] was used when heteroscedasticity was detected. Degrees of freedom were computed via the Satterthwaite approximation [26]. p values were adjusted for multiplicity using the Hochberg sequential rejective procedure. The underlying normality of variables was assessed using bivariate log-normal plots [27]. When appropriate, a normalizing transformation was applied to the data. All p values were based on two-sided tests. Results were considered to be statistically significant for p < .05. Statistical analyses were performed using SAS software (Cary, NC, http://support.sas.com).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

Fractones are extracellular structures that appear as series of fragmented basement membranes in the brain ventricle walls in adulthood [14, 15]. Figure 1I shows the ultrastructure of a typical fractone unit (between arrowheads) in the SEL adjacent to the corpus callosum. High electron density, location beneath ependymocytes, branching morphology, and direct contact with numerous cell processes are typical fractone ultrastructural characteristics. After immunostaining for laminin (using the antibody directed against laminin isoforms), fractones appear in confocal microscopy as a series of puncta along the ventricle walls (Figs. 1A–1C, 2H). Fractones, blood vessels, and choroid plexus are immunoreactive for laminin, but fractones can easily be distinguished by their small size (1–5 μm), morphology (puncta vs. elongated structures), and location in the SEL (Fig. 2H). In the SEL of the LV, fractones are smaller but more numerous in the SEL adjacent to the caudate putamen, the most active NSPC niche during adulthood, than in the SEL adjacent to the lateral septal nucleus (Fig. 2H).

thumbnail image

Figure Figure 1.. Neural stem and progenitor cells (NSPC) proliferate next to fractones in the LV walls. (A–D): Bromodeoxyuridine+ (BrdU+) proliferating NSPC (red) next to fractones (visualized by laminin immunoreactivity, green) in the subependymal layer (SEL) of the adult mouse 5 hours after intracerebroventricular injection of BrdU. Image location is shown in (G). Spatial relationships between BrdU+ cells, fractones, and blood vessels is analyzed in (H). (E): BrdU+ cells clustered (arrow) around a capillary in the SEL of the CPu. (F): Cluster of BrdU+ cells around a fractone (arrow) in the SEL of the CPu. (G): Location of images (A–D). (H): BrdU+ cells were closer to fractones than to blood vessels. The diagram indicates average distance (in μm) and SD between the surface of each BrdU+ cell nucleus and the closest fractone (black) or closest blood vessel (red). (I): Ultrastructure of the adult rat CC wall showing postmitotic cells (asterisks) and a fractone (between arrowheads). Scale bars = 25 μm (A–F) and 2 μm (I). Abbreviations: BV, blood vessel; CC, corpus callosum; CPu, caudate putamen; EP, ependymocyte; LSN, lateral septal nucleus; LV, lateral ventricle.

Download figure to PowerPoint

thumbnail image

Figure Figure 2.. Expression of N-sulfate heparan sulfate proteoglycan (HSPG) is restricted to fractones and adjacent capillaries. All confocal images display perlecan-immunoreactive (perlecan-ir) (red) and N-sulfate HSPG-ir (10E4, green) in the adult mouse. (A): Perlecan-ir was observed in fractones (arrow), blood vessels (arrowhead), and CP. (B): Same field showing N-sulfate HSPG-ir in fractones (arrow) but not in blood vessels (arrowhead). (C): Merged image showing that numerous fractones express N-sulfate HSPG and perlecan (which appear yellow; asterisk). (D): Some subependymal layer (SEL) capillaries displayed N-sulfate HSPG-ir plus perlecan-ir (which appear yellow; arrow) next to fractones. Although numerous fractones were double-labeled (arrowhead), some showed N-sulfate HSPG-ir only (asterisk) or perlecan only (double arrow). (E): N-Sulfate HSPG-ir (arrow) in the LMx. (F): Arachnoid basement membrane ultrastructure (arrow). (G): Location of (A–F) images. (H): Expression of N-sulfate HSPG in fractones of the neurogenic niche (arrow) and SEL capillaries (arrowheads) but not distal capillaries, which exclusively expressed perlecan (asterisk). (I): Location of (H). Scale bars = 50 μm (A, D, E), 1 μm (F), and 100 μm (H). Abbreviations: Br, brain; CC, corpus callosum; CP, choroid plexus; CPu, caudate putamen; Hip, hippocampus; LMx, leptomeninges; LSN, lateral septal nucleus; LV, lateral ventricle.

Download figure to PowerPoint

NSPC Proliferate Next to Fractones

We previously established that fractones directly contact neuroblasts and other progenitor cells in the LV walls [14], but it is not clear whether those cells proliferate next to fractones. Six adult mice were sacrificed 5 hours after i.p. injection of BrdU to identify mitotic cells in the neurogenic niche and their anatomical relationships with fractones and blood vessels. Most BrdU-ir cells in the ventricle wall are NSPC [4]. Serial frozen sections were prepared coronally from bregma 1 to bregma −1 or sagittally 0.5–1 mm from the midline, and 60 sections originating from the six animals were double-immunolabeled with antibodies directed against BrdU and laminin. Fifteen confocal microscopy images (originating from eight brain sections showing the LV walls) were chosen at random within our confocal image collection and were statistically analyzed to assess the distance between BrdU-ir cells, fractones, and blood vessels. We measured the distances between the periphery of each BrdU-ir cell nucleus, the closest fractone (revealed as laminin-ir puncta), and the closest blood vessel (revealed as laminin-ir lines).

As shown in a section located 0.7 mm anterior to bregma, most BrdU-ir cells were located next to fractones (Fig. 1A–1F). Statistical analysis on the 91 BrdU-ir cells shown in Figure 1A–1D indicated BrdU-ir cell nuclei at an average of 4.2 μm (SD 4.4) from fractones (Fig. 1H, blue). The p value adjusted for multiplicity was <0.001 (described in the Statistical Methods section in Materials and Methods). Occasionally, BrdU-ir cells clustered around SEL capillaries, but BrdU-ir cells were located, on average, 13.4 μm (SD 9.8) from capillaries (Fig. 1H, orange). A t test for the 91 cells shown in Figure 1A–1D (3.0 × 10−13) indicated that the two groups differed, therefore showing that BrdU-ir cells were significantly closer to fractones than from capillaries in the SEL. To determine whether this section is representative, we performed similar statistical tests on nine other confocal images chosen at random among our stained frozen sections. Results on 409 BrdU-ir cells indicated that BrdU-ir cells were located at 4.9 μm (SD 7.6) from the closest fractone and 13.7 μm (SD 15.0) from the closest blood vessel. A t test for the 409 cells investigated (3.1 × 10−24) indicated that the two groups differed and that BrdU-ir cells were significantly closer to fractones than from capillaries in the SEL. The p value adjusted for multiplicity was <0.001. These results clearly indicate that NSPC incorporate BrdU (i.e., enter cell cycle) next to fractones and not (or rarely) next to blood vessels, in the ventricle walls.

Because NSPC display elongated processes that contact fractones [14], a distance of 4 or 5 μm between a NSPC nucleus and a fractone may indicate NSPC/fractone direct contacts. Indeed, our previous investigation of the LV walls ultrastructure in the adult rat revealed postmitotic cells next to fractones. Figure 1I shows postmitotic cell bodies with electron-dense cytoplasm 6 μm from a fractone in the SEL of the corpus callosum. An electron-dense cell process, which may belong to one of the postmitotic cells, directly contacted the fractone (Fig. 1I, on the left of the left arrowhead).

Fractones and Meninges Are Unique Structures Immunoreactive for N-Sulfate HSPG

To determine whether fractones contain potential molecules that bind heparin-binding GF, we immunostained adult mouse frozen sections for perlecan using an antibody directed against the core protein of this HSPG, and for N-sulfated HSPG, which may be found on glycosylated chains of perlecan or other HSPG. It is known that the antibody 10E4 directed against the N-sulfate glycosamines recognizes N-sulfate HSPG and not chondroitin sulfate proteoglycans [28]. Figure 2A shows that fractones (arrows), SEL blood vessels (arrowhead), and choroid plexus were immunoreactive for perlecan. Most brain blood vessels were not immunoreactive for N-sulfate HSPG (Fig. 2B, arrowhead, 2H). Double perlecan/N-sulfate HSPG revealed different color levels ranging from red to green (via yellow) in the merged images throughout the SEL (Fig. 2C, 2D, 2H). This existence of variable levels of N-sulfate HSPG-ir and perlecan-ir from a given fractone to the next (Fig. 2D) indicated chemical heterogeneity of fractone HSPG in the LV walls. Although these results do not preclude the possibility that perlecan is N-sulfated in fractones, the weak correlation between immunoreactivity for N-sulfate HSPG and perlecan suggests that an HSPG other than perlecan exists in fractone populations and that perlecan may not be N-sulfated but O-sulfated in other fractone populations. Interestingly, several blood vessels, particularly in the neurogenic niche, were immunoreactive for N-sulfate HSPG (Fig. 2D, 2H). Meninges (Fig. 2E) and choroid plexus stroma (Fig. 5A) also displayed significant levels of N-sulfate HSPG immunoreactivity. N-Sulfate HSPG-ir and perlecan-ir rarely overlapped, indicating that perlecan may be infrequently N-sulfated in these locations. We showed here fractone-like structures in the arachnoid surrounding arteries of the circle of Willis (Fig. 2F). These arachnoid structures likely correspond to scattered laminin-ir structures that we previously detected in the meninges [24].

Neurogenesis Is Correlated with N-Sulfate HSPG Immunoreactive Fractones

To determine whether N-sulfate HSPG immunoreactivity in the SEL correlates with mitotic activity in the neurogenic niche, we investigated the relative distributions of BrdU-incorporating cells and N-sulfate HSPG-ir in the LV wall. Brain sections of adult mice were generated 5 hours after a single pulse of BrdU injected by i.p. or ICV and were immunolabeled with anti-N-sulfate HSPG and anti-BrdU antibodies. Figure 3A–3D shows that BrdU-ir cells were spatially correlated with N-sulfate HSPG-ir fractones (arrows in Fig. 3A, 3C). Statistical analysis on the 62 BrdU-ir cells shown in Fig. 3 indicates that the nuclei of BrdU-ir cells were located, on average, 1.7 μm (SD 2.5 μm) from N-sulfate HSPG-ir structures, which were fractones for 55 of these cells and capillaries for 7 of these cells (Fig. 3F). Comparing Figures 1 and 3, BrdU-ir cells were significantly closer to fractones immunoreactive for N-sulfate HSPG than to the whole population of fractones (1.7 μm vs. 4.2 μm), suggesting that fractone-HSPG may promote cell proliferation in the NSPC niche. In addition, these experiments confirmed results presented in Figure 1, indicating that most BrdU-ir cells are found next to fractones, not blood vessels. Figure 3D shows a cluster of BrdU cells next to an N-sulfate HSPG-ir capillary (arrow). However, the associations between BrdU-ir cells and N-sulfate HSPG-ir capillaries were not systematically observed along the ventricle walls. Most BrdU-ir cells shown in Figure 3D were not associated with N-sulfate HSPG-ir capillaries.

thumbnail image

Figure Figure 3.. Neural stem and progenitor cells proliferate next to N-sulfate heparan sulfate proteoglycan (HSPG)-immunoreactive fractones. (A–D): Bromodeoxyuridine (BrdU) (red) and N-sulfate HSPG (10E4, green) immunoreactivity in the adult mouse brain 5 hours after intracerebroventricular injection of BrdU. Image locations are shown in (E) and (G). Most BrdU+ cells were located next to 10E4+ fractones (arrows in [A] and [C]). In (B), the arrow indicates BrdU+ cells next to a capillary. The CP stroma, not the CP Epi (arrowhead), showed 10E4 immunoreactivity. (E, G): Location of images in (A–D). (F): BrdU+ cells proliferated next to 10E4+ fractones or capillaries. The diagram indicates average distance between each BrdU+ cell nucleus and the closest 10E4+ structure. Scale bars = 30 μm (A, B, D) and 15 μm (C). Abbreviations: BV, blood vessel; CC, corpus callosum; CP, choroid plexus; CPu, caudate putamen; Epi, epithelium; LSN, lateral septal nucleus; LV, lateral ventricle.

Download figure to PowerPoint

Interestingly, all N-sulfate HSPG-ir zones were active zones of cell proliferation, including neurogenic niches and meningeal niches (meninges rapidly replace their fibroblasts and macrophages). SEL of the LV, choroid plexus stroma, brain surface meninges, and meningeal projections within the brain (not shown) were immunoreactive for N-sulfate HSPG and BrdU. The small portion of choroid plexus shown in Fig. 2B shows high N-sulfate HSPG immunoreactivity and no BrdU+ cells, but each section containing the choroid plexus displayed 2–5 BrdU+ cells in the choroid plexus. Therefore, N-sulfate HSPG immunoreactivity in brain was spatially correlated with, but not proportional to, cytogenic activity in the niches.

The Neurogenic Factor FGF-2 Binds Fractones In Situ

In situ binding on frozen sections is commonly used to locate GF binding sites and determine GF binding properties [29]. It has been shown in peripheral tissues that FGF-2 requires binding to basement membrane HSPG, such as perlecan or agrin, to exert its full biological activity [19, 20, 30, 31]. For example, perlecan binds and presents activated FGF-2 to FGF receptors on target cells that directly contact basement membranes [20]. It was further demonstrated in vitro that HSPG are implicated in FGF-2 presentation to NSPC FGF-2 receptors [32]. Our hypothesis is that fractones represent the niche structures that control NSPC fate via mechanisms that implicate GF/HSPG interactions and, ultimately, the presentation of activated GF to abutting NSPC. To determine whether fractones are target sites for neurogenic heparin-binding GF, we tested in situ binding of FGF-2 on rat/mouse brain serial frozen sections containing the LV. FGF-2 was biotinylated and then incubated on coronal frozen sections prior to revealing binding sites with streptavidin-Texas Red. The same sections were also immunolabeled with laminin antibodies (L9393) to locate all fractones [14] or with N-sulfate HSPG (antibody 10E4) to label a subset of fractones (Fig. 5A).

On more than 100 brain frozen sections observed, in situ binding of biotinylated FGF-2 occurred as puncta in rat (Fig. 4B, 4E, 4G) and in mouse (Fig. 5A, 5C, arrows) in the exact locations of fractones. Numerous, but not all, fractones displayed FGF-2 label, with a complex label heterogeneity along the ventricle walls. LV wall fractones were much more heavily labeled than fractones of the third ventricle walls (not shown). Interestingly, we observed numerous fractones labeled by biotinylated FGF-2 in the SEL of the lateral septal nucleus (Figs. 4G, 5C) and a moderate number of labeled fractones of the SEL of the caudate putamen (Figs. 4G, 5C) and corpus callosum (Fig. 4E). For example, at bregma +0.7 mm (rat), biotinylated FGF-2 bound more than 80% of fractones in the SEL of the lateral septal nucleus (Fig. 4G, arrow) but only rare fractones in the SEL of the caudate putamen (Fig. 4G, asterisk). A comprehensive overview of the distribution of in situ FGF-2 binding sites to fractones throughout the LV walls is beyond the scope of this report.

thumbnail image

Figure Figure 4.. Fractones bind fibroblast growth factor 2 (FGF-2) in situ on brain frozen sections (here shown in rat). (A): Location of confocal images. (B): Overlap of laminin (L9393, green) and biotinylated-FGF-2 binding (red). Fractones that bound FGF-2 appeared yellow (arrows). Most fractones of the CPu wall did not bind FGF-2 (green, arrowhead). ∗, location of (C). (C): Here, the CP weakly bound biotinylated FGF-2. (D): Overlap of laminin (green) and biotinylated-FGF-2 showing fractones (arrows) and capillaries (arrowheads) in the corpus callosum wall. (E): Biotinylated-FGF-2 only. Fractones (arrows), but not capillaries (arrowheads), bound biotinylated FGF-2. (F): Overlap of laminin (green) and biotinylated FGF-2 (red). (G): Biotinylated FGF-2 only. Most fractones (arrow) but not all (arrowhead) bound biotinylated FGF-2 in the SEL of the LSD. Few fractones bound biotinylated FGF-2 in the SEL of the CPu (asterisk). Blood vessels did not bind biotinylated-FGF-2 (double arrow). Scale bars = 50 μm. Abbreviations: CP, choroid plexus; CPu, caudate putamen; LSD, lateral septal dorsal nucleus; LSN, lateral septal nucleus; LV, lateral ventricle; SEL, subependymal layer.

Download figure to PowerPoint

thumbnail image

Figure Figure 5.. Fibroblast growth factor 2 (FGF-2) binding to fractones depends on N-sulfate heparan sulfate proteoglycan (HSPG) (here shown in mouse). (A): N-Sulfate HSPG-immunoreactive (HSPG-ir) (10E4, green) on fractones in the SEL of the LSN (arrows) and CPu (arrowheads) in the adult mouse. (B): Image location. (C): In situ binding of biotinylated FGF-2 (red) to fractones (arrows and arrowheads) and to the CP. FGF-2 binding occurred in N-sulfate HSPG-ir fractones. (D): Perlecan immunoreactivity (blue) in several fractones and blood vessels (asterisks). (E): Merged image. (F–I): N-Sulfate HSPG-ir (F), perlecan-ir (H), in situ FGF-2 binding (G), and merged image (I) after heparatinase I treatment on frozen sections. Heparatinase I induced disappearance of N-sulfate HSPG-ir (F) and of FGF-2 binding to fractones (G) but not of immunoreactivity for perlecan (H). (I): Merged image. Scale bar = 50 μm. Abbreviations: CP, choroid plexus; CPu, caudate putamen; LSN, lateral septal nucleus; LV, lateral ventricle.

Download figure to PowerPoint

Fractones were the principal forebrain structures that bound biotinylated FGF-2. Except for choroid plexus (Fig. 5C) and meninges (not shown), we did not observe binding of biotinylated FGF-2 in other forebrain structures. As shown in Figure 5C, binding of biotinylated FGF-2 was often robust, but it was irregular along the choroid plexus. Portions of the choroid plexus weakly bound FGF-2 (Fig. 4C). Biotinylated FGF-2 did not bind significantly to capillaries (Fig. 4E, arrowheads; Fig. 4G, double arrow), except rare SEL capillaries (not shown here but shown in in vivo FGF-2 binding experiments in Fig. 6A).

thumbnail image

Figure Figure 6.. Fractones capture fibroblast growth factor 2 (FGF-2) in vivo. (A): Binding of biotinylated FGF-2 (red) in the adult rat 20 minutes after intracerebroventricular (ICV) injection of biotinylated FGF-2. Note the diffuse plus distinct high-intensity punctate (arrows) pattern of bound FGF-2 in the horn. (B): Location of images (A, C–I). (C): Punctate binding of biotinylated FGF-2 (red) in the SEL (arrows) 1 day after injection of biotinylated FGF-2. Note the diffuse pattern of biotinylated-FGF-2 in the CP (asterisk). (D): Merged biotinylated-FGF-2 (red) and laminin (L9393, green) showing that biotinylated-FGF-2 has primarily bound to fractones (as indicated by yellow and arrows). (E, F): Magnification of the area indicated by an asterisk in (D) showing biotinylated FGF-2 binding to fractones (arrows). FGF-2 label was absent on BV (double arrow) and on some fractones (arrowhead). (G): Diffuse pattern of biotinylated FGF-2 in the SEL 1 day after successive ICV injections of heparanase I/heparatinase I mix and of biotinylated FGF-2. (H, I): Double label for laminin (L9393, green) and biotinylated FGF-2 (red) after ICV injection of biotinylated FGF-2 in the adult mouse showing biotinylated FGF-2 binding to fractones (arrows) and SEL capillaries (asterisk) but not to distal capillaries (double asterisk). Some fractones did not bind biotinylated FGF-2 (arrowhead). (J): Location of (H, I). Scale bars = 25 μm (A), 10 μm (C–G), and 50 μm (J, K). Abbreviations: BV, blood vessels; CC, corpus callosum; CP, choroid plexus; CPu, caudate putamen; Fi, fimbria hippocampus; LSN, lateral septal nucleus; LV, lateral ventricle; SEL, subependymal layer.

Download figure to PowerPoint

Fractones Bind FGF-2 via HSPG (In Vitro Demonstration)

FGF-2 is a heparin-binding GF that requires binding to heparan sulfate chains of HSPG to exert its activity [19, 21, 22]. Thus, fractones may require HSPG to bind FGF-2. To test this possibility, we incubated mouse frozen sections with heparatinase I prior to performing in situ incubation of biotinylated-FGF-2. Without heparatinase I treatment, numerous fractones were immunoreactive for N-sulfate HSPG (Fig. 5A) and perlecan (Fig. 5C), and biotinylated FGF-2 binding occurred on N-sulfate HSPG/perlecan-ir fractones (Fig. 5C, arrows). After heparatinase I treatment, the immunoreactivity for N-sulfate HSPG (Fig. 5F) and biotinylated-FGF-2 binding (Fig. 5G) totally disappeared. This signifies that biotinylated-FGF-2 binding depends on the presence of HSPG in fractones. As anticipated, heparatinase I treatment was not detrimental to immunoreactivity for the perlecan core protein (Fig. 5H).

Fractones Capture and Store FGF-2 In Vivo After ICV Injection

We next determined whether fractones capture FGF-2 in vivo. Biotinylated FGF-2 was injected in the LV of adult rats and mice that were sacrificed at various time points. As a negative control, we injected nonconjugated biotin into animals. Controls did not show any detectable label after streptavidin-Texas Red label on frozen sections (data not shown). Twenty minutes after ICV injection, biotinylated FGF-2 was found in the horn located at the caudate putamen/corpus callosum interface (Fig. 6A) and in all the LV walls. In these locations, the label appeared as a faint diffuse pattern through the SEL plus intense puncta with a diameter of 2–5 μm. Double labeling for biotinylated FGF-2 and laminin revealed biotinylated FGF-2 in fractones, in the SEL adjacent to the caudate putamen, lateral septal nucleus/fimbria hippocampus (Fig. 6C, 6D), and corpus callosum (not shown). The strong binding of biotinylated FGF-2 in vivo (Fig. 6C) but weak binding in vitro (Fig. 4G) in the SEL of the caudate putamen (neurogenic zone) is surprising, but it may be explained by an impaired accessibility of HSPG by biotinylated FGF-2 in vitro after PFA fixation, at specific locations.

Biotinylated FGF-2 was also bound to fractones 1 day (Fig. 6C, 6E) and 2 days (Fig. 6K) after ICV injection, strongly suggesting that fractones store FGF-2. Most blood vessels did not significantly capture FGF-2 (Fig. 6C, 6E, double arrow). As observed in FGF-2 binding in situ, subpopulations of fractones captured biotinylated FGF-2 (Fig. 6E). Some fractones did not bind biotinylated FGF-2 (Fig. 6F, arrowhead). That biotinylated FGF-2 binds on all sides of the LV walls whereas neurogenesis occurs only in the caudate putamen/corpus callosum suggests that FGF-2 could be stored in non-neurogenic zones for further release in the LV or that FGF-2 is used for a biological activity unrelated to neurogenesis in the lateral septal nucleus.

Fractones Capture FGF-2 via HSPG (In Vivo Demonstration)

To determine whether FGF-2 binding in vivo depends on HSPG, as observed in vitro, we injected a heparanase I/heparatinase I mix by ICV 3 hours prior to a second injection with biotinylated FGF-2. Figure 6G shows that biotinylated FGF-2 failed to bind fractones in vivo after heparanase I + heparatinase I treatment. We have not yet determined whether heparatinase I by itself abolishes biotinylated FGF-2 binding in vivo.

Fractones Are Immunoreactive for Laminins β1 and γ1, Collagen IV, Nidogen, and Perlecan

To identify fractone components, we double-labeled adult rat, mouse, and human brain sections with anti-laminin-isoform antibodies (L9393) and diverse antibodies directed against basement membrane components. Immunoreactivity for heparan sulfate chains (MAb2040) showed a diffuse heparan sulfate label throughout the SEL, with a more intense label in fractones (Fig. 7A, 7B). All fractones were immunoreactive for laminin β1 (Fig. 7D), laminin γ1 (Fig. 7J), and nidogen (Fig. 7H). Collagen IV-ir was also found in all fractones of the adult rat, mouse, and human (shown in the adult human in Fig. 7Q). Perlecan immunoreactivity was found in large populations of fractones (Fig. 5D, 7H, 7F). As anticipated, all these ECM components were found in basement membranes throughout the vasculature (Fig. 7C, arrowhead, 7S). Double-immunostaining for laminin α1 and perlecan showed laminin α1 in large blood vessels but not in capillaries (Fig. 7K). In all sections containing the LV, we did not detect laminin α1-ir in fractones (Fig. 7L).

thumbnail image

Figure Figure 7.. Fractone components in the adult rat (A–H), mouse (J–L), and human (M, S). (A): Heparan sulfate proteoglycan-immunoreactive (HSPG-ir) (monoclonal antibody [MAb] 2040, red) in the whole SEL and CP. Note the concentrated label within the SEL (arrow). (B): Same field showing merged immunoreactivity for HSPG (red) and laminin (L9393, green). Most fractones, identified by green label, displayed strong HSPG-ir (arrow). (C, D): Double immunolabel for laminin (L9393, green) and laminin β1 (red) showing that fractones (arrows) and BV (arrowhead) are immunoreactive for laminin β1. (E, F): Double immunolabel for laminin (L9393, green) and perlecan-ir (MAb 1948, far-red channel) showing that numerous fractones are immunoreactive for perlecan. (G, H): Double immunolabel for laminin (L9393, green) and nidogen (far-red channel) showing that fractones are immunoreactive for nidogen. (I): Location of confocal images. (J): Laminin γ1-ir in fractones (arrow) and BV. (K): Merged laminin α1-ir (green) and perlecan-ir (red) in the dorsal cortex showing laminin α1 in the vasculature of the IG and Art arising from IG. Note the transition from double perlecan/laminin α1 label in Art to perlecan-ir only in cap. (L): Merged laminin α1-ir (green) and perlecan-ir (red) in the LV walls. Fractones (arrows) and cap (arrowhead) were not immunoreactive for laminin α1, but large BV are (double arrow). (M): Human fractones in the SEL of the LV (arrows) identified by laminin immunoreactivity (L9393, green). Fractones contact vimentin-ir (red) cell processes (arrowhead) and Ep. (N): Image location. (O): Fractones (arrow) in the SEL of the spinal canal (L9393, green). (P, Q): Double staining of fractones for laminin (L9393, green) and laminin γ-1 (red). R, S: Double staining of human fractones for laminin (L9393, green) and collagen-IV (red). Scale bars = 50 μm (A–C, E–H), 25 μm (J, P, S), 100 μm (K, L), 10 μm (M), and 75 μm (O). Abbreviations: Art, arterioles; BV, blood vessels; cap, capillaries; CC, corpus callosum; CP, choroid plexus; CPu, caudate putamen; Ep, ependymocytes; IG, indusium griseum; LV, lateral ventricle; SC, spinal canal; SEL, subependymal layer.

Download figure to PowerPoint

Fractones Exist in Adult Humans

We immunostained frozen sections of adult human brain and spinal cord to determine whether fractones exist in humans. Figure 7M shows a series of laminin-ir ranging from 1 to 7 μm in diameter in the SEL of the LV. Pattern, location, and size strongly suggest that these labeled structures are fractones. Double-labeling experiments demonstrated that human fractones were immunoreactive for collagen IV and laminin-γ1 (Fig. 7Q, 7S). In addition, we detected fractones by laminin-ir in the SEL of the human spinal canal (Fig. 7O).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

The physiological mechanisms regulating NSPC proliferation, differentiation, and migration in the adult brain are not fully understood. In this study, we demonstrated that the novel extracellular structures fractones are present in the NSPC niche of different adult mammalian species, including humans. We further demonstrated that fractones, like blood vessels and meningeal basement membranes, are immunoreactive for collagen IV, laminin β1 and γ1, and nidogen. Fractones, infrequent SEL blood vessels, choroid plexus, and meninges are the sole brain structures immunoreactive for N-sulfate HSPG. We provided evidence that cell proliferation in the LV neurogenic niche is anatomically correlated with fractones and poorly correlated with blood vessels. Furthermore, we have shown that cell proliferation preferentially occurs next to fractones that are immunoreactive for N-sulfate HSPG. Finally, we demonstrated that fractones and infrequent blood vessels sequester FGF-2 by a heparan sulfate-dependent mechanism. These results give new insight into the potential mechanisms that regulate the NSPC niche and strongly suggest that fractones are responsible for ECM/GF interactions that ultimately lead FGF-2 to NSPC in the adult brain.

Fractone Niche for Adult Neurogenesis

It has previously been demonstrated that cell proliferation in the adult hippocampus occurs next to blood vessels [33]. We have shown here that cell proliferation in the LV walls predominantly occurs next to fractones, and particularly next to N-sulfate HSPG-ir fractones. In contact with fractones and blood vessels, NSPC may encounter basement membrane molecules that promote GF neurogenic activity. Among basement membrane molecules, HSPG are of a particular interest because they are known to mediate heparin-binding GF activity. It has been demonstrated with peripheral cell lines that HSPG activate FGF-2 in the extracellular space before initiating receptor-mediated intracellular protein kinase cascades that ultimately lead to cell proliferation [19, 20, 22, 34]. That HSPG target FGF-2 on neuroepithelial cell FGF receptors [32] further suggests that HSPG-GF interactions are crucial for promoting and regulating NSPC proliferation. Anatomical features of the SEL and fractones may explain why fractones, more than vascular basement membranes, may operate in the LV walls. The immense number of fractones and their three-dimensional organization allows multiple contacts between fractones and NSPC. In addition, fractone exposure to ventricular cerebrospinal fluid via interstitial clefts [35] permits fractone-HSPG to efficiently capture ventricle-borne GF. In the light of our results, it would be interesting to determine whether blood vessels that reside next to BrdU-incorporating cells in the granular layer of the dentate gyrus (GLDG) are immunoreactive for N-sulfate HSPG. The GLDG does not contain a significant number of fractones (not shown), but N-sulfate HSPG are expressed in the GLDG [36, 37]. Therefore, blood vessels may play in the dentate gyrus the role fractones play in the LV wall: capturing heparin-binding GF by N-sulfate HSPG. HSPG may be crucial for promoting heparin-binding GF activity in the neurogenic niches.

Expression of N-Sulfate HSPG in Fractones and Meninges

Our results show that brain blood vessels and fractones are both immunoreactive for HSPG, particularly perlecan, but a striking difference between blood vessels and fractones is immunoreactivity for N-sulfate HSPG. In the forebrain, only large populations of fractones, rare SEL blood vessels, and meninges (including choroid plexus stroma) were immunoreactive for N-sulfate HSPG. We have previously shown laminin-ir arachnoid structures [24] (Fig. 1A). Arachnoid basement membranes resembling fractones in their complex three-dimensional morphology were characterized by ultrastructure (Fig. 2F). However, we do not know whether these arachnoid basement membranes are the structures immunoreactive for N-sulfate HSPG. Our results show that NSPC proliferation preferentially occurs next to N-sulfate HSPG-ir fractones and rare SEL capillaries that are also immunoreactive for N-sulfate HSPG. Therefore, it is possible that cell proliferation in the neurogenic niche primarily depends on the presence of HSPG that are N-sulfated. In other words, FGF-2 and other neurogenic heparin-binding GF may interact with fractone-borne N-sulfate HSPG to exert their neurogenic activity. Interestingly, it has been shown that N- and 2-O-sulfate groups are essential for the FGF-2 binding [37, 38]. Perlecan, agrin, and collagen-XVIII may be the basement membrane HSPG that influence angiogenesis and neurogenesis [30, 39, [40], [41], [42]43]. Candidate heparin-binding GF that influence neurogenesis and may interact with fractones in FGF-4, FGF-8, [44], amphiregulin [9], heparin-binding growth-associated molecule [10], and interferons [45].

Fractones Capture and Store Ventricular FGF-2

FGF-2 is highly neurogenic in vivo and in vitro. In vitro, in the presence of heparin, FGF-2 promotes the formation of multipotent neurospheres by NSPC [46]. Heparin, a fully sulfated molecule, likely compensates for the absence of HSPG in vitro. Because our in vitro and in vivo binding studies demonstrate that fractones, rather than blood vessels, bind FGF-2, we hypothesize that fractone-HSPG bind and activate FGF-2 and other neurogenic heparin-binding GF to promote neurogenesis. That FGF-2 is predominantly bound to fractones, 20 minutes and 2 days after ICV injection, indicates that fractones have unique capabilities for trapping and storing FGF-2. Biotinylated FGF-2 is quickly available in the SEL but cannot efficiently bind blood vessels even 1 day after penetration in the SEL. Thus, a difference must exist between the affinity of most vascular basement membranes and fractones for FGF-2 in the ventricle walls. Here, we show that heparan sulfate chains of HSPG are crucial for the binding of FGF-2 to fractones. Indeed, heparatinase I digestion in vivo and in vivo demonstrated that FGF-2 binding is dependent upon the presence of HSPG (Figs. 5, 6G).

Fractones Are Specialized Basement Membranes That Contain Laminins, Collagen IV, Nidogen, and HSPG

We show here that fractones are immunoreactive for laminin β1 and γ1, collagen IV, nidogen, the HSPG perlecan, and N-sulfate HSPG. This strongly reinforces our previous assertion [14] that fractones are basement membranes [18, 47]. However, fractones are unique in morphology and location (they do not cover a surface but are three-dimensional). Basement membranes also exist in the blood vessel walls and at the interface between the meninges and glia limitans astrocytes [12, 14, 15, 24, 48, 49]. Although HSPG, and particularly their N-sulfated forms, are promising molecules for regulating GF activity in the neurogenic niche, other basement membrane components, including laminins, may play important roles in regeneration and neural tissue plasticity [16, 50, [51]52].

Neurogenesis: An HSPG-Associated Mechanism Regulated by Fractones?

In light of the present results, we suggest that ECM/GF interactions occurring within fractones are critical for adult neurogenesis. Fractones express HSPG and bind and store FGF-2 by a mechanism that is heparan binding-dependent, within minutes after FGF-2 injection and thus prior to induction of neurogenesis by FGF-2. We propose that HSPG may function as fractone-associated regulators of neurogenesis and that those HSPG-GF interactions are three-dimensionally orchestrated within fractones to ultimately present GF to NSPC. The production of neurogenic GF by the choroid plexus [53, 54] and GF infiltration in between ependymocytes also argues for implication of choroid plexus in neurogenesis. Furthermore, we believe that macrophages/microglia associated with fractones along the ventricle walls [15, 55] also participate in neurogenesis. Macrophages/microglia likely provide ECM molecules that are necessary for the turnover/degradation of fractone components and produce GF, cytokines [56], and chemokines that may influence neurogenesis [55]. Our current goal is to determine whether fractones serve neurogenesis by presenting HSPG-bound GF to NSPC. Understanding the ECM/GF interactions that occur in fractones will give new insights into the basic mechanisms that govern the NSPC niche in vivo.

Disclosure of Potential Conflicts of Interest

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

The authors indicate no potential conflicts of interest.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

We thank Katalin Csiszar (Cardiovascular Research Center, John A. Burns School of Medicine, University of Hawaii, Honolulu, HI) for support during the development of this work, as well as Vivek Nerurkar, Maureen Saint-Georges Chaumet, and Dirk Vellinga (Department of Tropical Medicine Medical Microbiology and Pharmacology, John A. Burns School of Medicine); Glenn I. Hatton (Department of Cell Biology and Neuroscience, University of California Riverside, Riverside, CA); and Tatsunori Seki (Juntendo School of Medicine, Tokyo) for helpful comments on a previous draft of this article. This work was supported by grants from the Research Centers in Minority Institutions Program (G12 RR003061), National Center for Research Resources, NIH (NS009140), and the Hawaii Community Foundation (20022105 and 20040456); a High Technology Research Center Grant; and Grants-in-Aid 17082008 and 17046019 from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References