|Home | About | Journals | Submit | Contact Us | Français|
We hypothesized that oxygen gradients and hypoxia-responsive signaling may play a role in the patterning of neural or vascular cells recruited to the developing heart. Endothelial progenitor and neural cells are recruited to and form branched structures adjacent to the relatively hypoxic outflow tract (OFT) myocardium from stages 27–32 (ED6.5–7.5) of chick development. As determined by whole mount confocal microscopy the neural and vascular structures were not anatomically associated. Adenoviral delivery of a VEGF trap dramatically affected the remodeling of the vascular plexus into a coronary tree while neuronal branching was normal. Both neuronal and vascular branching was diminished in the hearts of embryos incubated under hyperoxic conditions. Quantitative analysis of the vascular defects using our recently developed VESGEN program demonstrated reduced small vessel branching and increased vessel diameters. We propose that vascular and neural patterning in the developing heart share dependence on tissue oxygen gradients but are not interdependent.
Diverse cell types are added to the developing embryonic vertebrate heart and organized into the complex structures of the mature heart. Endothelial progenitor cells (EPCs) are recruited via the pro-epicardial organ, organized into a vascular network at the base of the outflow tract (OFT) and remodeled into a stereotypical and highly patterned coronary arterial tree (reviewed in Tomanek 2005;Reese et al. 2002;Lavine and Ornitz 2009). Neuronal precursor cells are recruited from the neural crest, also appear at the base of the cardiac OFT and similarly form stereotypical, highly branched and patterned structures within the heart (reviewed in Kirby 2007).
A number of factors that regulate the growth, survival and migration of these cell types have been identified, yet how arteries and nerves develop their stereotypical patterns in the heart is unknown. In the heart, as in many other organs, nerves and blood vessels are anatomically associated in what is termed the neurovascular bundle. Recent studies suggest that the congruence of neurovascular patterning in some tissues, for example the skin, is due to a trophic factor, vascular endothelial growth factor (VEGF), released by the nerves that pattern the forming blood vessels (Mukouyama et al. 2002). Other studies have suggested that forming vessels serve as a template and provide guidance cues to pattern the nerves (reviewed in Eichmann et al. 2005). A third model is that of shared neurovascular patterning, in which arteries and nerves are controlled by similar patterning mechanisms but are independent of one another, as proposed for the embryonic chick limb (Bates et al. 2003).
We have previously observed that the avian OFT myocardium is hypoxic relative to the remainder of the heart from Hamburger-Hamilton(Hamburger and Hamilton 1951) stages 25–32 (ED5–8) (Sugishita et al. 2004b;Sugishita et al. 2004a;Sugishita et al. 2004c;Wikenheiser et al. 2006). This period overlaps the recruitment of vascular and neural precursors into the heart. This led us to hypothesize that tissue hypoxia and hypoxia-responsive signaling may play a role in the recruitment and patterning of vascular and neural elements in the developing heart. One hypoxia-responsive gene is VEGF, a key regulator of angiogenic and vasculogenic blood vessel formation through its effects on endothelial cell proliferation and migration (reviewed in Coultas et al. 2005). By injecting a peptide (VEGFtrap) that contains the VEGF-binding portion of the VEGF receptors 1 and 2 (VEGFR1 and VEGFR2), a prior study showed that VEGF is required for coronary vasculogenesis in the quail embryo (Tomanek et al. 2006). However, in that study, it was not determined if the effect of VEGFtrap was on the recruitment of EPCs to the OFT, the formation of the vascular plexus, and/or their remodeling into a highly patterned coronary arterial tree. It was also not determined if trapping of VEGF affected neuronal patterning, and whether or not the two are interdependent.
We have used a recombinant adenovirus, AdFlk1-Fc (AdFlk1), that expresses a truncated and secreted form of VEGFR2 (Flk1) and traps VEGF (Kuo et al. 2001), to test the role of VEGF in heart development. We previously showed that AdFlk1 disrupts the apoptosis-dependent remodeling of the avian cardiac OFT required for transition to a dual in-series circulation (Liu and Fisher 2008). The goals of the current study were to use AdFlk1 to define the role of VEGF in the recruitment and patterning of vascular and neural elements in the developing heart. To more directly test the role of oxygen gradients on neurovascular patterning, the effect of incubation under increased oxygen concentrations was examined. Critical to these analyses of patterning was the implementation of a computational program developed by one of us, termed VESGEN, to provide mapping and quantification of the vascular patterned structures (Parsons-Wingerter et al. 2000a;Parsons-Wingerter et al. 2000b;Parsons-Wingerter et al. 2006a;Parsons-Wingerter et al. 2006b;McKay et al. 2008).
Cells of the endothelial lineage were identified by the QH-1 antibody (Pardanaud et al. 1987). As previously reported (Sugishita et al. 2004b;Kattan et al. 2004), EPCs are observed on the surface of the embryonic quail heart at stage 25, organize to form a vascular network at the base of the OFT at stage 30 (ED7), and are remodeled into a tree-like structure from stages 34–36 (ED8–9, Fig. 1A–C). The first coronary vascular smooth muscle cells (SMCs), identified by SMA staining, are evident at the base of the OFT at ED8, and define the epicardial coronary arteries at ED9 (Fig. 1D–F).
The TuJ antibody recognizes β-tubulin III specific to neuronal cells. Neuronal extensions first appear on the anterior surface of the heart at stages 28–29 (ED6–6.5) and form a highly branched structure over the cardiac OFT myocardium that subsequently extends over the surface of the ventricles from ED7–9 (Fig. 2A–D), consistent with prior studies (Kirby and Stewart 1983;Kuo et al. 2001;Hood and Rosenquist 1992;Verberne et al. 1998;Kirby 2007). By laser scanning confocal microscopy of ED8 and ED9 hearts, it is evident that the vascular and neural cell patterning is not matched in the x-y plane (Fig. 2E–F). In the z-plane the vascular and neural staining are present in separate stacks (not shown), indicating that these cells are separated by a distance of at least 28 microns. At each developmental time point, the vascular structure preceded the neural structure, organizing in a proximal-distal pattern from its origin at the base of the OFT. At ED9 ganglia are visible at the branch points at the base of the OFT (Fig. 2F, arrows).
To test the role of VEGF in the recruitment and organization of the neurovascular elements, we delivered to the embryonic quail heart recombinant adenovirus that expresses a truncated secreted form of the VEGFR2 (AdFlk1), and thereby functions as a VEGF trap. In this model system, expression of the truncated VEGF receptor is restricted to the OFT myocardium and adjacent tissues (Liu and Fisher 2008). The number of free EPCs on the surface of the heart in AdFlk1 embryos at ED5.5 (stage27), prior to the onset of vasculogenesis, was decreased (Con 260±22 vs AdFlk1 160±24, n=5 each, p<0.05), while there was no difference at ED7 (Con 147±25 vs AdFlk1 165±89 at ED7, n=6, p>0.05) and ED8 (Con 432±97 vs AdFlk1 539±146, n=5, p>0.05). In all 10 AdFlk1 embryos examined at ED7 (stage 30), a vascular network had formed at the base of the OFT. The vascular plexus was reduced in density in 5 of the 10 embryos. In 3 of the 10, vascular density appeared normal but the network was disorganized and the vessel diameters were increased (Fig. 3A–C). In the remaining two embryos, the vascular plexus appeared normal.
At ED8, 11 of 13 AdFlk1 hearts exhibited abnormal vascular patterning. In 5 of these embryos the vessels were increased in diameter and displayed reduced branching, giving the appearance of hyper-fused vessels (Fig. 3E). In 4 of these embryos there was the abnormal persistence of a vascular network at this stage of development, with either complete absence of, or markedly diminished, tree formation (Fig. 3F). In one embryo there were no discernible vascular structures at the base of the heart, while in another the tree-like structure had formed but the vascular density and vessel diameter were reduced (not shown).
At ED9, 16 of 19 AdFlk1 hearts exhibited abnormal vascular patterning. Seven embryos displayed reduced vessel branching and increased vessel diameter (Fig. 3H). In seven other embryos, the vascular network persisted at the base of the OFT, accompanied by markedly reduced or complete absence of a tree-like structure. In the remaining 2 abnormal embryos, vessel branching increased while vessel diameter was reduced (Fig. 3I). A subset of the ED8 and ED9 embryos were co-stained with the SMA antibody. The extent of coverage with SMA was markedly reduced in 3 of 5 AdFlk1 embryos at ED8 (Fig. 4B) and 4 of 7 AdFlk1 embryos at ED9 (n=7, Fig. 4D).
In contrast, AdFlk1 had no effect on neural patterning at ED7–9 (Fig. 5A–C, representative of n=6 at each stage; compare to controls in Fig. 2). Co-staining with QH-1 and TuJ in whole mount demonstrates normal neuronal branching in an ED9 AdFlk1 embryo in which vascular patterning was abnormal (Fig. 5D–F).
To directly test the role of tissue hypoxia in recruitment and patterning of the vascular and neural cells, oxygen gradients in the developing heart were diminished by hyperoxia. As described previously (Sugishita et al. 2004b), quail embryos were incubated in 95% oxygen beginning at stage 27 (ED5.5) for 48–72 hours to ED8–9 (stages 34–36). The embryos showed defects in vascular patterning that were similar to those of the AdFlk1 hearts. At ED7, the vascular network at the base of the OFT had formed in embryos incubated under hyperoxic conditions, but was reduced in density (Fig. 6A, n=9; compare to control, Fig. 1). At ED8, the hyperoxic embryos had reduced vessel branching and density (11/12), increased vessel diameter (4/12) and absence or reversal of vessel tapering (4/12) (Fig. 6B). In 3/12 embryos the vessel diameter was reduced rather than increased (not shown). Smooth muscle actin staining was also markedly reduced, indicating failure of coronary artery formation (Fig. 6E, compare to control, Fig. 1). Embryos incubated under hyperoxia also showed significant reductions in neuronal branching and density (Fig. 6G–I, n=9, compare to controls, Fig.2). Similar effects on neuronal and vascular patterning were observed at ED9 in 5 of 11 embryos (Fig. 6C, F, I, n=11). The lower proportion of neurovascular defects at ED9 may reflect selection bias as many embryos are non-viable with prolonged incubation.
We used VESGEN software to map and quantify vasculogenesis in representative control and experimental embryos (n=2 each) from the AdFlk1 and hyperoxia series described above. Both AdFlk1 and hyperoxia inhibited the development of branching complexity within the coronary vascular tree at ED8 primarily by inhibiting the angiogenesis of small vessels (Fig. 7 and Table 1). AdFlk1 and hyperoxia treated vessels therefore remained as less mature composite tree-network structures. At ED8 numerous parameters of branching complexity were decreased in the AdFlk1 embryos compared to the controls. The mean of vessel number density (Nv), vessel branch point density (Brv), and vessel length density (Lv) decreased from 3.65E-05 µm−2, 3.59E-05 µm−2 and 5.22E-03 µm/µm2 in control embryos to 2.03E-05 µm−2, 1.92E-05 µm−2 and 2.98E-03 µm/µm2, respectively, in AdFlk1 embryos. However, the mean diameter (Dv) increased from 36.2 µm in control embryos to 52.4 µm in AdFlk1 embryos. Within smaller branching generations (G>3) where G1 is the largest branch, the mean of Nv(G>3) and Lv(G>3) decreased from 2.73E-05 µm−2 and 2.66E-03 µm/µm2 to 1.54E-05 µm−2 and 1.16E-03 µm/µm2. This selective inhibition of small vessel development by AdFlk1, as well as the significantly decreased length of larger vessels, was accompanied by increased vessel diameter in all branching generations (Table 1). For G1–3 and G>3, respectively, the mean of vessel diameter increased from 59.5 µm and 24.4 µm in control specimens to 74.2 µm and 36.2 µm in AdFlk1 specimens.
Hyperoxia demonstrated more severe inhibition on the development of coronary vasculature than did AdFlk1. Compared to control hearts, the mean values of Nv, Brv, and Lv decreased to 0.79E-05 µm−2, 0.54E-05 µm−2 and 1.66E-03 µm/µm2, respectively, while Dv increased to 41.02 µm. For both G=1–3 and G>3, the vessel length density was significantly inhibited and the mean Lv decreased to 0.37E-05 µm/µm−2 and 1.08E-05 µm/µm−2, respectively, in embryos incubated in the 95%/5% O2/CO2 environment. Similarly, the mean of Nv for G=1–3 and G>3 was also significantly reduced to 0.16E-05 µm−2 and 0.63E-05 µm−2, respectively. On the contrary, the mean vessel diameter of both small (Dv(G>3)) and large (Dv(G=1–3)) branches increased in the hyperoxia group compared to noninjected control hearts (31.22µm and 77.02µm, respectively, for hyperoxia group versus 24.37µm and 59.45µm for controls) and are of similar magnitude to those in AdFlk injected hearts.
We used Metamorph software to map and quantify neurogenesis in representative control and experimental ED8 embryos (n=3 each) from the hyperoxia series described above. Total neurite outgrowth length from the main branch, total number of branches, and number of branching generations were all significantly reduced by hyperoxia (Table 2).
In this study we used hyperoxia and adenoviral gene delivery to test the role of oxygen gradients and hypoxia-responsive signaling in neurovascular patterning during heart development. The major effect of AdFlk1 was a failure in the transition from a vascular network to a coronary tree. This was qualitatively evident, and quantitative analysis with VESGEN software indicated reduced numbers and density of the small branches of the forming vascular tree and increased diameters and density of the larger proximal branches. The most likely explanation for this effect is that the VEGF trap disrupts a VEGF gradient that is required for the addition and sprouting of EPCs at the tips of vascular structures in the transition to a coronary tree, as described in other developmental models of vasculogenesis and angiogenesis (Ruhrberg et al. 2002;Gerhardt et al. 2003;Isogai et al. 2003;Bautch and Ambler 2004;Lundkvist et al. 2007;Cha and Weinstein 2007). Alternatively, VEGF may be required for the transition from the primary vascular network to the vascular tree through its effects on the formation and/or remodeling of the network. AdFlk1 had a modest effect on EPC numbers and primary vascular network formation; however it is still possible that the cumulative effects of AdFlk1 on these processes were sufficient to perturb the subsequent transition to a vascular tree. We did not observe an increase in indicators of cell death in the EPCs in the AdFlk1 hearts (data not shown), suggesting that VEGF is not functioning in EPC survival in this context. Fate mapping of the endothelial cells by real time imaging within single embryos will be required to determine the cellular mechanism for this effect on vasculogenesis. Alternatively, the role of VEGF at sequential stages of coronary vasculogenesis could be determined by timed delivery of the VEGF antagonists, a method that is not currently feasible in this model of gene delivery.
The coronary arteries ultimately develop a highly patterned structure that is conserved in birds and mammals. Arising from the right and left coronary cusp each courses through the atrioventricular grooves and give off fairly stereotypical patterns of sub-branches, with the anterior and posterior descending arteries always coursing adjacent to the inter-ventricular septum. The factors that mediate this stereotypical coronary patterning are not defined. In other specialized circulations it has been suggested that the nerves may participate in the patterning of the vessels, or that vessels may serve as a template for patterning the nerves, and that the co-dependence of patterning may reflect the effects of VEGF and semaphorins on nerves and vessels (reviewed in Carmeliet 2003;Eichmann et al. 2005). The current study provides several lines of evidence to suggest that neural and vascular patterning in the developing heart, at least in the initial phases, are not co-dependent. 1) We observed that while neural and vascular cell types migrate to and begin to pattern in the cardiac OFT with similar timing (stages 27–30, ED6–7) they do not co-localize as determined by scanning confocal microscopy. 2) The tree-like projections that form in the subsequent phases of vasculogenesis precedes the extension of neural fascicles, excluding a role for neural cells in this stage of vascular patterning. 3) Neural patterning was normal in embryos with impaired vasculogenesis due to AdFlk1, suggesting that this neural migration and patterning does not follow vascular cues and is not co-dependent on VEGF. It is still possible that neural and vascular patterning is co-dependent at later stages of development or in the micro-vasculature.
Prolonged incubation under hyperoxia (95%O2/5%CO2), a condition which substantially reduces oxygen gradients within the heart (Sugishita et al. 2004b), significantly attenuated the angiogenic phase of coronary vessel development. This reduction in the smaller vessels and enlarged vascular plexus is similar to that achieved with AdFlk1. The similarity of these effects suggests a model in which hypoxia driven gradients of VEGF within the heart drive vessel tip cell sprouting and the angiogenic phase of coronary vasculogenesis, as has been proposed in retinal vasculogenesis (Lobov et al. 2007). Consistent with this interpretation, hyperoxic incubation resulted in a 50% decrease in VEGF mRNA levels in the heart as measured by real-time PCR (data not shown). In the development of other branched structures, e.g. Drosophila trachea, tissue oxygen gradients establish morphogenic gradients of FGF that direct terminal branching (Jarecki et al. 1999;Centanin et al. 2008). FGFs are also required for coronary vasculogenesis (Pennisi and Mikawa 2009;Lavine et al. 2006).Further studies are needed to define the hypoxia-dependent program that may establish growth factor (or other) gradients required for coronary vascular patterning.
In contrast there has been less investigation of the role of tissue hypoxia in neural migration and patterning. Neural patterning in the OFT is blunted by hyperoxic incubation but not AdFlk1 suggesting that it is hypoxia-dependent but VEGF-independent. A recent study in C. elegans indicates that axonal pathfinding in the embryo is oxygen and HIF-1 sensitive through the regulation of VAB-1(Pocock and Hobert 2008), the Eph receptor homologue, a well described regulator of axon guidance in vertebrates (reviewed in Hinck 2004). Hypoxia also affects neurite outgrowth in the PC12 cell line in vitro (O'Driscoll and Gorman 2005). Semaphorin signaling through plexin and neuropilin receptors are another logical candidate for hypoxia-dependent neural patterning in the OFT given their established role in axonal patterning (Hinck 2004;Yazdani and Terman 2006), and their roles in cardiac OFT morphogenesis as defined by expression patterns and loss-of-function studies in mouse (Brown et al. 2001;Gitler et al. 2004) and chicken (Toyofuku et al. 2008). However there is currently little data to support the idea that their expression or activity is hypoxia-responsive (Compernolle et al. 2003)
This study has used hyperoxic exposure to dissipate oxygen gradients within the heart. The effect on neurovascular patterning is ascribed to the alleviation of tissue hypoxia and dissipation of oxygen gradients. We cannot exclude the possibility that the increased oxygen concentration was toxic, though we did not observe generalized toxicity. An alternative approach will be to inactivate hypoxic signaling (HIF) specifically in the hypoxic tissues in the mouse heart in the analogous developmental window. One limitation of the VESGEN analysis of vascular patterning is the exclusion of the primary vascular plexus that forms over the OFT myocardium due to the difficulty in resolving the vascular structures. Indeed the ultimate fate of this primary vascular network is not known. However, it is apparent that the branched vascular structures that ultimately will comprise the epicardial coronary arterial tree are not first apparent within this primary vascular network. A second limitation of the VESGEN analysis is the approximation of occasionally overlapping larger vessels within the coronary branching tree as joined vessels. Overcoming this limitation would require using 3D-reconstructed images obtained by confocal microscopy together with a VESGEN 3D analysis.
In conclusion, these observations support a model in which neural and vascular patterning in the heart, at least in the initial phases, are not co-dependent, but may have shared control mechanisms that are regulated by tissue oxygen concentrations and gradients. This model of shared control mechanisms for neurovascular patterning in the heart is analogous to that proposed in models of mouse and chick limb development (Bates et al. 2003;Schwartz et al. 1990;Vieira et al. 2007).
Fertile quail (Coturnix coturnix Japonica) eggs obtained from the Department of Animal Science (Michigan State University, MI) were incubated in a humidified room air incubator (Circulated Air Incubator Model 1250, G.Q.F. Manufacturing Co., Savannah, GA) at 38°C to the appropriate stages for each experiment. Under stereomicroscopy 0.5 µl of a solution containing AdFlk1-Fc (AdFlk1) at a titer of 1012 pfu/ml was injected into the pericardial space of Stage 17–18 quail embryos as previously described (Liu and Fisher 2008). AdFlk1 is a recombinant replication-defective adenovirus that expresses the murine Flk1 (VEGFR2) cDNA sequence encoding the signal peptide and the ectodomain fused to a murine IgG2α Fc fragment under the control of the cytomegalovirus (CMV) promoter/enhancer (Kuo et al. 2001). Un-injected embryos or embryos injected with an adenovirus expressing murine IgG2α Fc alone (Kuo et al. 2001) were used as controls. The adenovirus was kindly provided by Drs. Kenneth Walsh (Boston University) and Calvin Kuo (Stanford University). Eggs were returned to the incubator and embryo harvested at stages 30, 34 and 36, corresponding to days 7, 8 and 9 of incubation (ED7, 8, 9).
Stage 27 quail embryos were placed in a sealed chamber (Modular Incubator Chamber, Billups-Rothenberg, Inc., Del Mar, CA) with water, flushed with 95% O2/5% CO2 for 10 min to completely exchange the air, and the inlet and outlet tubes were clamped, as previously described (Sugishita et al. 2004b). The chamber was set in a standard humidified tissue culture incubator at 37°C. The sealed chamber was flushed with oxygen gas on a daily basis to maintain hyperoxia. Control eggs were placed in the same incubator in room air. The embryos were harvested at stages 30 (ED7), 34 (ED8) and 36–37 (ED9).
For QH-1 and TuJ staining the hearts were dissected from embryos and immediately fixed in freshly prepared methanol:DMSO (4:1 w/w) solution. For staining for smooth muscle actin embryos were fixed in freshly prepared 4% paraformaldehyde. After overnight fixation at 4°C, the hearts were extensively washed with PBS followed by blocking in PBS containing 1% BSA and 0.3% Triton X-100. The primary antibodies used in this study were polyclonal rabbit antibody against neuronal class III β-tubulin (TuJ, Covance, Emeryville, CA) at 1:1000 dilution; QH-1 antibody (Developmental Studies Hybridoma Bank, Iowa City, IA) against quail vascular endothelial cells at 1:1000 dilution; and monoclonal mouse antibody against smooth muscle α-actin conjugated with cy3 (Sigma-Aldrich, St. Louis, MO) at 1:500 dilution. TuJ and QH-1 antibodies were detected by goat anti-rabbit and mouse secondary antibodies conjugated with Alexa Fluor 488 and 594 (Invitrogen, Carlsbad, CA), respectively.
Representative embryos from ED8 control, AdFlk1 and hyperoxia groups were analyzed for vascular patterning by VESGEN software, an automated vascular analysis software program (lead developer, P. Parsons-Wingerter) that will be publicly available following development of user documentation. VESGEN analyzes three major types of vascular morphology: branching vascular trees, vascular networks, and tree-network composites (Vickerman et al. 2009). Output parameters include vessel diameter, length and branchpoint density, fractal dimension, tortuosity and avascular space/vessel relationships as well as various image mappings. For vascular trees, the measurements are provided as dependent functions of vessel branching generation.
Each stereomicroscopic image was transformed by semi-automatic image processing into two types of binary (black-white) images as described previously (Parsons-Wingerter et al. 2006a;Parsons-Wingerter et al. 2006b;McKay et al. 2008). The first binary image was the vascular pattern and the second the region of interest (ROI) occupied by the vessels. These two binary input images were used by VESGEN to generate various image mapping and quantification outputs. VESGEN maps branching vascular trees into successively smaller branching generations according to two conditions: 1) relatively symmetric vessel bifurcation and 2) decrease in vessel diameter to approximately 71% of the diameter of the parent vessel (using a 15% tolerance factor), which provides geometric conservation of blood flow. These conditions were developed using a two-dimensional (2D) model of angiogenesis in the quail chorioallantoic membrane (CAM) based on studies of three dimensional vascular branching in the dog and pig heart and lung (Kassab et al. 1994). The ROI was used to provide density measurements for parameters such as vessel number and vessel length in ED8 embryo hearts. The amorphous peri-conal vasculogenic region was uniformly excluded from all vascular pattern images and therefore from VESGEN analysis because after forming at ED7 this region becomes progressively less well defined through development (see Results).
The number of free endothelial cells stained by QH-1 on the surface of the heart was determined with the Image J program and Cell Counter plug-in as previously described (Liu and Fisher 2008).
Neural network formation and branching pattern analysis was performed using the neurite outgrowth module contained within Metamorph imaging software (Molecular Devices Corporation, Downington, PA). Briefly, the Neurite Outgrowth program took the branching points as cell bodies and automatically tracked the associated outgrowths. The length of each outgrowth and the number of branching junctions were measured by the program. The total length of outgrowth and total number of branches were obtained by adding values for each cell body. Branching generation was defined at each hierarchical bifurcation point. The maximum branching generations was determined by the number of hierarchical nerve generations for both control and hyperoxia treated hearts.
Confocal images were acquired with a Zeiss LSM 510 META laser scanning confocal microscope (Carl Zeiss MicroImaging, Jenna, Germany) using an Argon laser (excitation, 488 nm), a HeNe laser (excitation 594 nm) and a 5× Plan-Neofluar, NA 0.15, Ph1 objective. All images presented here are maximum intensity projections of Z stacks consisting of 56.3 micron optical slices collected every 28.19 microns (optimal interval settings determined by LSM 510 software).
All statistics were done with Prism software (GraphPad software Inc., San Diego, CA). A student’s t-test was used for comparison among experimental groups. P<0.05 was considered statistically significant.
Grant sponsors: NEI/NIDDK R01EY17529 to P. Parsons-Wingerter, NASA INSPIRE student internship program, and NIH R01 HL65314 to S. Fisher.
The authors would like to thank Maryanne Pendergast and the Neurosciences Imaging Center, and Dr. Scott Howell and the Visual Sciences Research Center, at Case Western Reserve University, for assistance with the confocal microscopy and neuronal network analysis, respectively.