Establishment of neurovascular congruency in the mouse whisker system by an independent patterning mechanism.
Journal: 2014/January - Neuron
ISSN: 1097-4199
Abstract:
Nerves and vessels often run parallel to one another, a phenomenon that reflects their functional interdependency. Previous studies have suggested that neurovascular congruency in planar tissues such as skin is established through a "one-patterns-the-other" model, in which either the nervous system or the vascular system precedes developmentally and then instructs the other system to form using its established architecture as a template. Here, we find that, in tissues with complex three-dimensional structures such as the mouse whisker system, neurovascular congruency does not follow the previous model but rather is established via a mechanism in which nerves and vessels are patterned independently. Given the diversity of neurovascular structures in different tissues, guidance signals emanating from a central organizer in the specific target tissue may act as an important mechanism to establish neurovascular congruency patterns that facilitate unique target tissue function.
Relations:
Content
Citations
(7)
References
(53)
Grants
(1)
Chemicals
(4)
Genes
(11)
Organisms
(3)
Processes
(2)
Anatomy
(5)
Affiliates
(1)
Similar articles
Articles by the same authors
Discussion board
Neuron 80(2): 458-469

Establishment of neurovascular congruency in complex tissue by a novel independent patterning mechanism

Introduction

The nervous and vascular systems are highly branched networks that are functionally and physically interdependent (Carmeliet and Tessier-Lavigne, 2005; Zlokovic, 2008). Blood vessels provide neurons with oxygen and nutrients and protect them from toxins and pathogens. Nerves, in turn, control blood vessel diameter and other hemodynamic parameters such as heart rate. The functional interdependence between nerves and vessels is reflected in their close anatomic apposition. In the periphery, nerves and vessels often run parallel to one another, a phenomenon called neurovascular congruency (Bates et al., 2003; Quaegebeur et al., 2011). The intimate association between neurons and vessels is particularly important in the brain, as neural activity and vascular dynamics are tightly coupled by a neurovascular unit (Iadecola, 2004). Moreover, recent evidence suggests that some neurodegenerative diseases once thought to be caused by intrinsic neuronal defects are initiated and perpetuated by vascular abnormalities (Quaegebeur et al., 2011; Zlokovic, 2011). Despite these important connections between the nervous and vascular systems, a key unsolved question is how nerves and vessels become physically aligned during development in order to facilitate their functional properties.

The similar branching pattern of nerves and blood vessels was first noted in the scientific literature over a hundred years ago (Lewis, 1902). Since then, tightly associated nerves and blood vessels have been termed ‘neurovascular bundles’ and the phenomenon itself has been named ‘neurovascular congruency’ (Martin and Lewis, 1989; Taylor et al., 1994). While the existence of neurovascular bundles is widespread, the best-studied example is the vertebrate forelimb skin where congruency has been shown to be established during embryogenesis. Arteries are aligned with peripheral nerves in embryonic mouse limb skin, and in mice with mutations that lead to disorganized nerves, blood vessels follow these misrouted axons. Therefore, in the developing mouse forelimb skin system, peripheral sensory nerves determine the differentiation and branching pattern of arteries (Mukouyama et al., 2005; Mukouyama et al., 2002), indicating that the nerve guides the vessel. This elegant study suggests that neurovascular congruency can be established by a “one patterns the other” general principle in which either the nervous or vascular system precedes in development and then instructs the second system to form using the already established architecture as a template. However, this particular example describes the relatively simple planar structure of skin, and given the diversity of neurovascular networks in different tissues, the “one patterns the other” model is perhaps not precise enough to serve as a general mechanism governing the establishment of neurovascular congruency in all tissues, especially where more complex neurovascular associations and spatially restricted three-dimensional structures exist.

The whisker pad system provides an excellent model to study how neurovascular congruency arises in target tissues with complex, three-dimensional anatomy. Whiskers are patterned in discrete arrays with invariant numbers in each row of follicles. The whisker pad that underlies each individual whisker receives primary somatosensory innervation from axons in the maxillary branch of the trigeminal nerve, and then faithfully relays the sensory information to the brain (Erzurumlu et al., 2010). In the mature whisker pad, each whisker is innervated by multiple types of trigeminal neurons that form morphologically distinct endings inside the follicular sinus complex (FSC) (Ebara et al., 2002). Similarly, each whisker is inhabited by a branch of the infraorbital artery, which establishes the intricate capillary network called the blood sinus surrounding the follicle and forms the FSC (Ebara et al., 2002; Katsume et al., 1984); (Fundin et al., 1997). Therefore, in the adult, each follicle is surrounded by a well-organized layer of nerve endings and an adjacent vascular network. The vascular component affiliated with each whisker affects the movement of the whisker (Wineski, 1985) and modulates the sensitivity of the sensory nerve endings (Fundin et al., 1997; Wineski, 1985). Thus, neurovascular organization in each follicle is critical for precise whisker function.

In this study, we found that the close association between sensory nerves and blood vessels in the whisker pad is established during development. We observed a dynamic nerve/vessel interaction that ultimately results in the stereotypic organization of a “double ring” structure around each follicle. Surprisingly, nerve and vessel rings form independently rather than through a “one patterns the other” model. We further demonstrate through the use of mouse genetics, that secreted Semaphorin 3E and its receptor Plexin-D1 signaling is required to establish a stereotypic “double ring” neurovascular structure around each whisker follicle. Sema3E has the potential to repel both nerves and vessels through Plexin-D1. Sema3E originated from the follicle controls the organization of the outer vessel ring while the selective down-regulation of Plexin-D1 permits the nerves to maintain their inner ring position. Given the broad diversity of neurovascular structures and the need for individualized function, the co-patterning/ independent-patterning mechanism by guidance signals emanating from a central organizer in the specific target tissue is likely to be a common mechanism that is used to establish neurovascular congruency patterns in complex tissues.

Results

Nerves and vessels form a stereotypic “double ring” structure around each whisker follicle, with the nerve ring inside and the vessel ring outside

To characterize how follicular neurovascular organization arises during development, we performed double immunostaining to detect axons (anti-neurofilament; green) and blood vessels (anti-PECAM; red) during different developmental stages. We found that trigeminal axons arrive at the whisker primordium as early as E10.5, as was previously reported (Stainier and Gilbert, 1990). The primary vascular plexus is already formed at this time and is intermingled with the incoming axons until E11.5 (Figure 1B). Around E12.5, a ring–like structure of axonal innervation forms around each whisker primordium, while vessels remain disorganized and partially intermingled with the axons (Figures 1C, 1G, 1H and S1A, Movie S1). By E14.5, vessels are recruited to the nerve ring, leading to close apposition of vessels and axons (Figures 1D and S1B, Movie S2). By E16.5, a prominent “double ring” structure is formed, with a ring of nerves on the inside, a ring of vessels on the outside, and a defined space between them (Figures 1E and S1C, Movie S3). At E18.5, both nerve and vessel ring patterning is further refined (Figure 1F). The nerve-inside and vessel-outside structure is maintained during adulthood, and is known to be important for whisker function (Ebara et al., 2002). This relatively simple system with clear stereotypic developmental characteristics (Figure 1I) allows us to dissect the molecular and cellular interactions between nerves and vessels in a systematic, stepwise manner.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f1.jpg
Nerve and blood vessels are organized into a “double ring” structure in the whisker follicle during development

(A–F) Developmental profile of trigeminal axon innervation and blood vessel patterning in the whisker follicles. Trigeminal axons and blood vessels are visualized by neurofilament (green) and PECAM (red), respectively, in the snout area (boxed region in Fig. 1A). Co-immunostained tangential sections of whisker follicles at E11.5 (B), E12.5 (C), E14.5 (D), E16.5 (E), E18.5 (F) show the dynamic processes that lead to the stereotypic organization of a double ring structure. The trigeminal axons form a ring-like structure as early as E12.5 (arrowheads in C), and are then surrounded by a blood vessel ring that results in the formation of a double ring structure around each follicle at E16.5, with the nerve ring located inside and the vessel ring outside (arrow in E). (G–H) One whisker follicle in C (dotted box) is overlaid by DAPI staining to show follicle primordium (dotted circle in G). Nerve terminals surround the outside of the primordium (arrow in H). (I) Schematic illustration of nerve and vessel ring organization in the whisker follicle. Scale bar: 100 μm (B–F), 50μm (G–H). See also Figure S1 and Movie S1–S3.

Nerve ring and vessel ring are patterned independent of each other in the whisker follicles

Based on the developmental profile of nerve and vessel organization in the whisker follicle, the “one patterns the other” model of neurovascular congruency would predict that the trigeminal axons attract surrounding blood vessels to establish the double ring structure. To examine this possibility, we analyzed the nerve/vessel organization in the whisker follicle of neurogenin-1 (Ngn1) knockout embryos, in which trigeminal ganglia (TG) and thus the trigeminal nerves are completely absent (Ma et al., 1998). To our surprise, vessel ring organization is completely normal in Ngn1 knockouts at E16.5 (Figure 2B), despite the lack of nerve ring formation (absence of green signal in Figure 2B). Both the size and position of the vessel ring in Ngn1 mutant and wild-type littermate controls has no detectable difference (Figures 2C and 2D). This result clearly demonstrates that unlike limb skin (Mukouyama et al., 2002), where arterial patterning is disrupted in Ngn1/2 knockouts, indicating that sensory axons determine the patterning of arteries, here the formation of the vascular ring structure is independent of the nerves.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f2.jpg
Nerve ring and vessel ring are patterned independent of each other in the whisker follicle

(A–B) Nerve/vessel immunostaining of wild-type littermate control exhibits normal double ring patterning (A). Nerve/vessel immunostaining of the Ngn1 knockout shows no trigeminal innervation in the whisker follicles, but blood vessel rings are normally organized at each whisker follicle (arrows in B, C–D). (C–D) For the quantification, 10 pairs of whisker follicles were analyzed and the mean distance from the center (C) and vasculature area surrounding the follicle (D) ± SEM is shown. (E–F) Endothelial specific-Npn1 deletion shows less vasculature formation around whisker follicle (white arrowheads in F and G) compared to the vasculature of control (E), but nerve rings are organized normally (H). (G–H) Vasculature area surrounding follicle (G) and mean distance from center to nerve ring (H) ± SEM is shown (n=14). Paired student t-test; ***, P<0.001; n.s, not significant. Scale bar: 200 μm (A–B and E–F)

Next, we examined whether blood vessel ring patterning has any effect on nerve ring formation. We analyzed the “double ring” structure in endothelial-specific neuropilin-1 (Npn-1) null mice (Tie-2 Cre; Npn1 flox/flox) where the vasculature surrounding the whisker follicles is poorly developed (Figures 2E and 2F) (Gu et al., 2003). However, nerve ring formation shows no detectable difference between mutants and control littermates (Figures 2G and 2H). This result demonstrates that nerve ring formation is also independent of vessel ring formation. Therefore, the neurovascular congruency of the double ring pattern in the whisker follicles is not established by the “one patterns the other” mechanism, but rather patterned independently.

Sema3e and Plxnd1 are complementarily expressed in the trigeminal-whisker follicle system

To reveal the novel mechanism governing the neurovascular congruency in the whisker pad system, we first considered the contributions from the target tissues on their coordinated patterning. Emerging evidence suggests that the neural and vascular systems share many similar mechanisms and molecular cues to regulate their development (Adams and Eichmann, 2010; Carmeliet and Tessier-Lavigne, 2005; Gelfand et al., 2009). Therefore, we next asked whether guidance cues could be found in the whisker follicle, and whether these cues could function as an organizing signal to coordinately pattern the nerves and vessels into the congruent “double ring” structure. We first performed expression screening of known guidance molecules in the trigeminal ganglion and whisker targets, and found a striking complementary expression pattern of Sema3e and its receptor Plxnd1 in these areas (Figures 3A–3H). Moreover, the spatiotemporal expression of these molecules coincides with the developmental profile of whisker follicles. Plxnd1 mRNA was expressed at very low levels in the trigeminal ganglion neurons at an early stage (E12.5), when axons arrive at the whisker target (Figure 3A, (van der Zwaag et al., 2002)). Its expression level is significantly increased by E14.5, when trigeminal axons and blood vessels begin to organize the double ring structure in the whisker follicles (Figure 3B), and continues through E16.5 and E18.5, when target innervation is refined and a clear double ring structure is apparent (Figure 3C and data not shown). Similarly to the temporal profile of Plxnd1 expression in the trigeminal ganglion, Sema3e is not expressed at E12.5 in the whisker follicles (Figure 3E). Sema3e mRNA is found in the mesenchymal sheath surrounding the hair follicles starting at E14.5 and continuing to E16.5 and E18.5 (Figures 3F3H). To characterize the precise location of Sema3e within the follicle, we performed immunostaining of axons after Sema3e in situ hybridization (ISH) on the same tissue sections and demonstrated that Sema3e was expressed inside the nerve ring (Figures 3I and 3J). Finally, to examine whether Plxnd1 is also expressed in the outer blood vessel ring, we performed double fluorescence ISH of Plxnd1 and the endothelial cell marker Flk-1. As shown in figure 3L, mRNA expression of Plxnd1 and Flk-1 completely overlaps within the vessel ring. Together these expression data suggest that in the whisker pad Sema3E expression surrounds the hair follicle (Figures 3K and 3M), while Plexin-D1 is expressed in the blood vessels and may also expressed in the innervating trigeminal axons based on its mRNA expression in the trigeminal ganglion (Figure 3M). This complementary expression pattern in the target area suggests that Sema3E-Plexin-D1 signaling plays a role in coordinating trigeminal target innervation and the formation of blood vessels into the double ring structure.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f3.jpg
Complementary expression pattern of Plexin-D1 in the TG and blood vessels and Sema3E in the whisker follicle

(A–D) Plxnd1 in situ hybridization (ISH) on sagittal sections of wild-type embryos at E12.5 (A), E14.5 (B), and E16.5 (C) shows that Plxnd1 mRNA is detected at very low level as early as E12.5 and is significantly increased at E14.5 and continues to E16.5 as the double ring structure develops. Plxnd1 mRNA is not detectable in sections of the trigeminal ganglion from plexin-D1 knockout animals at E14.5 (TG outlined by dashed line in D). (E–H) Sema3e mRNA ISH on coronal sections parallel to the whisker follicles at E12.5 (E), E14.5 (F), E16.5 (G) and E18.5 (H) shows that Sema3e mRNA is expressed in the mesenchymal sheath surrounding the hair follicles starting at E14.5 and continues at E16.5 and E18.5 (black arrows). Scale bar: 100 μm in A applies to G, and 200 μm (H) (I–J) Neurofilament immunostaining (white arrows) after Sema3e ISH on the same sections shows that Sema3e (black arrows) is expressed inside of the nerve rings (coronal image in I and tangential image in J). (K–L) Double fluorescence ISH with Plxnd1 and Sema3e (K) or Flk-1 (L) shows that Plxnd1 mRNA is expressed in the endothelial cells that form the vessel ring (white arrowhead in K and L) and Sema3e is expressed inside of both nerve and vessel rings. (M) Summary of Plxnd1 and Sema3e mRNA expression pattern in the whisker follicle. Sema3e is expressed in mesenchymal tissue (yellow) and Plxnd1 is expressed in blood vessel (red) and in TG therefore Plexin-D1 protein probably in nerve (green). Scale bar: 100 μm in A applies to G, 200 μm (H), and 100 μm in I applies to L.

Semaphorin 3E and its receptor Plexin-D1 facilitate repulsive signaling in both nerves and vessels

Does Sema3E function as a common cue to coordinately pattern nerves and vessels during the formation of the stereotypic double ring structure? Because trigeminal axon terminals in the whisker target are adjacent to Sema3e-expressing cells (Figures 3I and 3J), whereas blood vessels are located outside of the nerve ring (Figures 3K and 3L), we first tested the hypothesis that Sema3E may act as an attractive cue to trigeminal axons and a repulsive cue to blood vessels. Such an attractive guidance role for axons is consistent with prior observation that in the presence of neuropilin-1, a co-receptor for Plexin-D1, Sema3E can serve as an attractant (Chauvet et al., 2007) and our observation that Npn-1 is highly expressed in the TG at this stage (Figure S2B). However, to our surprise, application of alkaline phosphatase (AP)-tagged Sema3E to E14.5 TG explants induced significant growth cone collapse (Figure 4C) compared to the control AP-treated group (Figure 4A). Moreover, Sema3E-induced growth cone collapse was absent in trigeminal neurons isolated from Plxnd1 null mice (Figure 4D), indicating that Sema3E serves as a repulsive cue to trigeminal neurons and Plexin-D1 is required for its effects. To further examine whether Npn-1 plays any role in Sema3E-induced trigeminal growth cone collapse, we performed the same growth cone collapse assay using TG explants from Nestin-Cre driven-Npn1 conditional knockout embryos in which Npn-1 was ablated in all neuronal populations. Trigeminal ganglia isolated from these mice exhibit the same level of growth cone collapse as their wild-type littermate controls when treated with AP-Sema3E (Figures S2E and S2F), indicating that Sema3E-Plexin-D1 signaling induces trigeminal growth cone collapse independent of Npn-1. As a positive control, we also treated TG explants from these mice with AP-Sema3A. Sema3A-Npn-1 signaling causes trigeminal growth cone collapse in vitro and is required for the initial axon projection from the trigeminal ganglia to their peripheral targets in vivo (Gu et al., 2003; Kobayashi et al., 1997)(Figures 4E and 4F). As expected, axons from trigeminal ganglia isolated from Nestin-Cre; Npn1flox/flox mice were completely unresponsive to the Sema3A-induced growth cone collapse (Figure S2H). Finally, although vascular endothelial growth factor receptor 2 (VEGFR2/Flk-1) has been suggested to promote axonal growth in association with Plexin-D1/Neuroplin-1 receptor complex (Bellon et al., 2010), VEGFR2 is not expressed in the TG neurons during double ring formation (Figure S2K). Therefore, Sema3E acts on trigeminal axons as a repellant rather than an attractant.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f4.jpg
Sema3E-Plexin-D1 signaling provides a repulsive guidance cue to both TG axons and blood vessels in vitro

(A–G) Sema3E-Plexin-D1 signaling serves as repulsive cue for cultured TG neurons. Trigeminal ganglia were isolated from wild type (A, C, E) or Plexin-D1 null mice (B, D, F) at E14.5 and a growth cone collapse assay was performed. Axons (green) and growth cones (red) were visualized by anti-neurofilament and phalloidin staining, respectively. After incubation with 2 nM of AP (A, B), AP-Sema3E (C, D), or AP-Sema3A (E, F) for 30 min, >200 growth cones were scored for each experimental condition. Sema3E treatment induced significant growth cone collapse (white arrowheads in C), but growth cone collapse was absent in Plexin-D1 mutant TG neurons (D). (G) Quantification of growth cone collapse assay (n=3, data shown as mean ± SD, ANOVA; *, P<0.001). (H–L) Sema3E inhibits endothelial cell migration. VEGF-induced HUVEC transwell migration assay was performed in the presence (K) or absence (J) of Sema3E (AP-Sema3E, 0.5 nM). After a 5 hr incubation, migrated cells were fixed and visualized by Nissl staining. Sema3E prevents HUVEC migration induced by VEGF as well as the basal level of migration (I). (L) Quantification of migrated HUVECs (data shown as mean ± SEM, n=3, ANOVA; *, P<0.05; **, P<0.001). Scale bar: 50 μm (A–F, H–K). See also Figure S2.

To examine the effect of Sema3E on endothelial cells, we performed an in vitro transwell migration assay using human umbilical vein endothelial cells (HUVECs), which endogenously express Plexin-D1. As predicted, vascular endothelial growth factor (VEGF) caused the majority of endothelial cells to migrate to the bottom chamber (Figures 4J and 4L), reflecting the known function of VEGF as an attractant for these cells. However, when Sema3E and VEGF were added together, almost no migration was observed (Figures 4K and 4L), indicating that Sema3E inhibits VEGF-induced migration. Moreover, Sema3E also blocked the basal level of migration (Figure 4I). Together, these results support the idea that Sema3E can dominantly block the attractive effects exerted upon Plexin-D1-expressing endothelial cells.

Differential expression of Plexin-D1 protein allows for distinctive responses of nerves and vessels to Sema3E

The appealing hypothesis that emerged, based on these in vitro results, is that Sema3E acts as a repulsive guidance cue for both trigeminal axons and blood vessels and serves to organize the double ring neurovascular structure surrounding the follicle. If this hypothesis is true, what accounts for the relative position of the nerves and vessels, with the nerve ring consistently positioned inside of the vessel ring? In particular, it is puzzling how trigeminal axons expressing Plexin-D1 are able to innervate an area that is so close to a secreted repulsive cue. One potential possibility is that the abundance of Plxnd1 mRNA transcripts in the TG may not reflect protein levels at the nerve terminal. To test this idea, we performed AP-Sema3E binding on tissue sections to detect Plexin-D1 protein along the trigeminal nerve (Chauvet et al., 2007; Gu et al., 2005). As shown in Figure 5A, although Plexin-D1 protein is highly expressed in the trigeminal nerves projecting to the whisker follicle, surprisingly, Plexin-D1 protein is very low in the nerve terminals (white arrowheads in Figures 5B and 5C). This difference is not due to our inability to detect protein binding at the nerve terminal because Npn-1 protein is equally well detected in both the projecting axons and the nerve terminal by AP-Sema3A binding (arrowheads in Figures 5D and 5E). The absence of AP-Sema3E binding on the vessel ring (red arrow in Figure S3C) of Plxnd1 null mice further confirmed that Plexin-D1 is the only receptor for Sema3E in the vessel ring (Figures S3A–S3D). The absence of AP-Sema3E binding on the nerve ring in Sema3e null mice also ruled out the possibility that the lack of AP-Sema3E binding may be due to the sequestration of Plexin-D1 by endogenous Sema3E (Figures S3E-S3H). To further confirm the selective down-regulation of Plexin-D1 protein in the TG nerve terminals, we also performed anti-Plexin-D1 immunohistochemistry (Chauvet et al., 2007). Consistent with AP-Sema3E binding result, TG nerve terminal exhibited extremely low Plexin-D1 immunoreactivity (Figures 5F and 5G). Therefore, the Plexin-D1 protein is selectively down-regulated in the TG nerve terminals, which enables the nerves to innervate areas close to the Sema3E-expressing region and form the inner ring. In stark contrast to the low expression of Plexin-D1 in the nerve ring, Plexin-D1 protein level visualized by both AP-Sema3E and anti-Plexin-D1 antibodies in the blood vessels is remarkably high (white arrows in Figures 5B, 5C, 5H, and 5I). Therefore, the differential expression of localized Plexin-D1 protein may lead to a differential response between the nerves and vessels to Sema3E, resulting in the nerve-inside and vessel-outside double ring structure.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f5.jpg
Plexin-D1 protein expression is selectively down-regulated in the target terminal of TG axons

(A–C) Plexin-D1 protein expression is visualized by AP-Sema3E binding. Tangential sections of tissue were cut at E14.5 and then incubated with 1 nM AP-Sema3E. Binding was then detected by reaction with an AP substrate. Plexin-D1 (black arrows in A) is highly expressed within the TG (dotted red outline) and the trigeminal nerves that project both centrally toward the brainstem and peripherally towards the snout (right side from dotted green outline). But Plexin-D1 protein is extremely low in nerves when they reach the whisker follicle (white arrowhead in B and C). In contrast, Plexin-D1 protein is strongly detected in the blood vessel ring (white arrows in B and C). Yellow arrowhead (in B and C) indicates non-specific signals in membrane structure of the whisker follicle. (C) To identify the relative location of nerve and blood vessel rings, adjacent sections were immunostained with NF (green) and PECAM (red) and overlaid onto the AP-stained sections. (D–E) Neuropilin-1 protein expression is detected by AP-Sema3A (2 nM) binding in the terminal nerve ring (arrowheads in D-E), which completely overlaps with NF-positive nerve ring on the same sections after AP-staining (E). (F–I) Plexin-D1 protein expression is visualized by immunohistochemistry with anti-Plexin-D1. In the TG nerve terminal surrounding the whisker follicles, Plexin-D1 protein expression is extremely low (arrowheads in F and G). In contrast, Plexin-D1 protein is strongly detected in the blood vessel ring (arrows in H and I). Anti-Plexin-D1 displays non-specific signals in the hair sheath and glassy membrane of the whisker follicle (asterisks in F and H). Scale bar: 1 mm (A), 50 μm (BE, F-I). See also Figure S3.

Sema3E-Plexin-D1 signaling is required for proper “double ring” formation in vivo

To directly test the in vivo requirement of Sema3E-Plexin-D1 signaling in the formation of the neurovascular double ring structure in each whisker follicle, we analyzed nerve and vessel patterning in the developing whisker follicle in Plxnd1 null and Sema3e null embryos. In wild-type embryos at E16.5, the outer ring of blood vessels is distinctly separated from the inner ring of nerves (Figures 6A and 6D). However, in the mutant embryos, the two rings are intermingled (arrows in Figures 6B and 6E). To quantitatively analyze this phenotype, we measured the ratio between the inner ring radius (r) and outer ring radius (R) in the same follicle and then compared the averaged ratios from wild-type embryos to that in mutant mice (Figure 6I). We observed a significant increase in the r/R ratio in the mutant animals as compared to the wild-type littermate controls, reflecting a collapsed double ring structure in the mutant mice (Figures 6C and 6F). The average distance from the follicle center to nerve /vessel ring in mutants and controls were also measured, respectively (Figures S4A and S4B). The vessel ring moved inward significantly whereas nerve ring position relative to center showed no noticeable changes, consistent with the selective down-regulation of Plexin-D1 at the nerve terminal. To see clear structure of the nerve and vessel alignment in the whisker follicle, whole-mount follicle staining with anti-neurofilament and VE-cadherin antibodies was performed. The blood vessel layer is separated from trigeminal nerve layer along with whole follicle shaft (Figure 6G). However, the nerve and vessel alignment is overlapped in multiple regions in the whole follicle shaft (Figure 6H). Therefore, Sema3E-Plexin-D1 signaling is required in vivo for proper nerve and vessel organization in the whisker follicle.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f6.jpg
Plexin-D1 and Sema3E is required for the stereotypic double ring nerve/vessel patterning in the whisker pad in vivo

(A–C) Plexin-D1 null mice exhibit severe nerve/vessel organization defects. The whisker follicle region of Plexin-D1 mutants (B) and their littermate controls (A) at E16.5 were sectioned and immunostained with NF and PECAM. In the Plexin-D1 mutant, nerve and vessel rings were abnormally close, with some areas showing complete vessel/nerve intermingling (arrow in B). Radii of the vessel ring (R) and nerve ring (r) depicted in I were measured and their ratio was calculated for each follicle (C). (D–F) Sema3E null mice also show the same nerve/vessel patterning phenotype (white arrow in E) compared to their littermate controls (D) and r/R ratio measurement is shown in F. For the quantification, 6–8 whisker follicles per animal from three different pairs of embryos in each experimental condition were measured and the mean ratios ± SEM are shown (C, F). Paired student t-test; *, P<0.05 (G–I) The detail patterning phenotype was analyzed in coronal view of whisker follicles using confocal imaging process after neurofilament and VE-cadherin whole mount staining. The blood vessel layer (white arrows in G) is clearly separated from the TG nerve layer in wild-type (G), but in the Plxnd1 mutant, vessels (yellow arrows in H) are intermingled with nerve layers (H). White arrowhead (in G and H) indicates non-specific signals from the whisker sheath marked in white asterisks. Scale bar: 100 μm (A, B, D, and E) and 50 μm (G and H). See also Figure S4.

Attractive signals for the initial recruitment of nerve and vessel are expressed in the whisker follicles

Our data demonstrated that in the developing whisker pad, the nerve and vessel are initially intermingled, then the nerve ring forms followed by the vessel ring to eventually establish the “double ring” structure with the nerve ring inside and vessel ring outside. Moreover, the nerve and vessel rings are formed independently of each other, rather than “one patterns the other”. Sema3E originating from the mesenchymal sheath surrounding the hair follicles controls their relative position in the “double ring structure” through its receptor Plexin-D1. What are the signals that initially recruit nerves and vessels in the whisker follicle? Two major chemoattractants for nerve and vessel recruitment, NGF and VEGF, respectively, are expressed around the whisker follicle when the double ring structure is forming (Bandtlow et al., 1987; Genc et al., 2005)(Figures S5). At E12.5 when nerve ring begins to form, we detected TrkA (NGF receptor) expression in the trigeminal nerves and Ngf expression around each whisker primordium (Figures S5A-S5H). Moreover, TG and maxillary explant co-culture assay demonstrated that NGF is necessary to attract TG axons (O’Connor and Tessier-Lavigne, 1999), suggesting that NGF-TrkA signaling contributes to the initial nerve ring formation. To directly test this hypothesis, we analyzed the initial nerve ring formation in mice lacking Ngf at E12.5-E13.5. We observed a normal size of TG and normal peripheral projections of trigeminal nerves to the snout area in Ngf mutant mice, consistent with the previous finding that mouse trigeminal ganglion survival become NGF-dependent after E14.5 (Figures S6A and S6B) (Pinon et al., 1996). However, despite of normal arrival at the whisker pad, TG axons failed to form the initial “ring like” structure in the Ngf mutant embryos. At E12.5 when a ring-like structure of axonal innervation forms around each whisker primordium in the wild-type littermates (Figures 1C, S6C-D, and S6G), but the axonal innervation in Ngf mutant embryos aggregates and fails to form the fasciculated ring-like structure (Figures S6E-G). At E13.5 when a complete nerve ring structure forms in the wild-type littermates (Figure S6H), the axonal innervation still “wanders” and fails to form a ring structure around the whisker primordium (Figure S6I). In addition, this initial failure of nerve ring formation defect seems to be continued postnatally, because whisker innervation defects have also been observed in new born Bax; trkA mice at P0 (Patel et al., 2000). Together, these data provides in vivo evidence demonstrating that NGF signaling is critical for the initial nerve ring formation. At E13.5, right before the blood vessel ring forms, we detected strong VEGF expression around each whisker primordium in an area close to the nerve ring (Figures S5I-S5K). Moreover, endothelial-specific knockout of Npn1, a VEGF receptor, results in aberrant vessel ring formation (Figures 2E and 2F). These data demonstrate that VEGF signaling contributes to the initial vessel ring formation. Interestingly, both NGF and VEGF-expressing cells are arrayed along the circumference of the primordium, thereby enabling the recruitment of both nerve and vessel in a ring like structure during development. At E14.5, the repulsive signaling of Sema3E-Plexin-D1 emerges to set the two rings apart and achieve the “double ring” structure. Thus, these results suggest that balanced attractive and repulsive signals from the target tissue control the ontogenetic patterning of the neurovascular double ring structure (Figure 7).

An external file that holds a picture, illustration, etc.
Object name is nihms571974f7.jpg
Model of how stereotypic double ring neurovascular structure is established during development

(Left) During early development (E11–E12), NGF is expressed in tissue surrounding the primitive whisker follicle and attracts trigeminal axons to the whisker target area. At this stage, the blood vessel plexus is dispersed in the whisker follicle and is not organized. (Middle) As development progress, around E13–E14 when trigeminal nerve forms a nerve ring structure, VEGF expression in the whisker follicle region begins to recruit blood vessel close to the nerve ring. (Right) Around E14–E16, Sema3E starts to be expressed in the mesenchymal sheath of the whisker follicle and repels incoming vessels through its interaction with Plexin-D1, which then settle outside of nerve ring. Although Plexin-D1 is also expressed in the trigeminal nerve at this stage, its expression in the axon terminals is selectively down-regulated to a level that is not sufficient to respond to Sema3E, thereby maintaining the location of the nerve ring close to the whisker follicle. See also Figures S5 and S6.

Taken together, this study demonstrates that the stereotypic neurovascular congruency in complex tissue is established during development by a novel independent patterning mechanism by cues emanated from the target tissue. Using the mouse whisker pad as a model system, we found that nerves and vessels form a stereotypic “double ring” neurovascular structure around each of the whisker follicles during development. Surprisingly, this “double ring” neurovascular congruency is not established by a “one patterns the other” mechanism, but rather is established via a novel mechanism in which nerves and vessels are patterned independently through differential responses to a common guidance cue. Sema3E secreted from the center of the follicle has the potential to repel both nerves and vessels through Plexin-D1. However, differential Plexin-D1 expression in nerves and vessels during whisker follicle development is translated into a distinctive repulsive strength that defines the relative location of nerve and vessel rings. Finally, formation of this double ring structure is disrupted in mice lacking Plxnd1 and Sema3e. Unlike the previously described “one patterns the other” mechanism (Mukouyama et al., 2002), our study demonstrates that local signals act as a central organizer in complex tissues to establish neurovascular congruency via a novel independent patterning mechanism.

Nerves and vessels form a stereotypic “double ring” structure around each whisker follicle, with the nerve ring inside and the vessel ring outside

To characterize how follicular neurovascular organization arises during development, we performed double immunostaining to detect axons (anti-neurofilament; green) and blood vessels (anti-PECAM; red) during different developmental stages. We found that trigeminal axons arrive at the whisker primordium as early as E10.5, as was previously reported (Stainier and Gilbert, 1990). The primary vascular plexus is already formed at this time and is intermingled with the incoming axons until E11.5 (Figure 1B). Around E12.5, a ring–like structure of axonal innervation forms around each whisker primordium, while vessels remain disorganized and partially intermingled with the axons (Figures 1C, 1G, 1H and S1A, Movie S1). By E14.5, vessels are recruited to the nerve ring, leading to close apposition of vessels and axons (Figures 1D and S1B, Movie S2). By E16.5, a prominent “double ring” structure is formed, with a ring of nerves on the inside, a ring of vessels on the outside, and a defined space between them (Figures 1E and S1C, Movie S3). At E18.5, both nerve and vessel ring patterning is further refined (Figure 1F). The nerve-inside and vessel-outside structure is maintained during adulthood, and is known to be important for whisker function (Ebara et al., 2002). This relatively simple system with clear stereotypic developmental characteristics (Figure 1I) allows us to dissect the molecular and cellular interactions between nerves and vessels in a systematic, stepwise manner.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f1.jpg
Nerve and blood vessels are organized into a “double ring” structure in the whisker follicle during development

(A–F) Developmental profile of trigeminal axon innervation and blood vessel patterning in the whisker follicles. Trigeminal axons and blood vessels are visualized by neurofilament (green) and PECAM (red), respectively, in the snout area (boxed region in Fig. 1A). Co-immunostained tangential sections of whisker follicles at E11.5 (B), E12.5 (C), E14.5 (D), E16.5 (E), E18.5 (F) show the dynamic processes that lead to the stereotypic organization of a double ring structure. The trigeminal axons form a ring-like structure as early as E12.5 (arrowheads in C), and are then surrounded by a blood vessel ring that results in the formation of a double ring structure around each follicle at E16.5, with the nerve ring located inside and the vessel ring outside (arrow in E). (G–H) One whisker follicle in C (dotted box) is overlaid by DAPI staining to show follicle primordium (dotted circle in G). Nerve terminals surround the outside of the primordium (arrow in H). (I) Schematic illustration of nerve and vessel ring organization in the whisker follicle. Scale bar: 100 μm (B–F), 50μm (G–H). See also Figure S1 and Movie S1–S3.

Nerve ring and vessel ring are patterned independent of each other in the whisker follicles

Based on the developmental profile of nerve and vessel organization in the whisker follicle, the “one patterns the other” model of neurovascular congruency would predict that the trigeminal axons attract surrounding blood vessels to establish the double ring structure. To examine this possibility, we analyzed the nerve/vessel organization in the whisker follicle of neurogenin-1 (Ngn1) knockout embryos, in which trigeminal ganglia (TG) and thus the trigeminal nerves are completely absent (Ma et al., 1998). To our surprise, vessel ring organization is completely normal in Ngn1 knockouts at E16.5 (Figure 2B), despite the lack of nerve ring formation (absence of green signal in Figure 2B). Both the size and position of the vessel ring in Ngn1 mutant and wild-type littermate controls has no detectable difference (Figures 2C and 2D). This result clearly demonstrates that unlike limb skin (Mukouyama et al., 2002), where arterial patterning is disrupted in Ngn1/2 knockouts, indicating that sensory axons determine the patterning of arteries, here the formation of the vascular ring structure is independent of the nerves.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f2.jpg
Nerve ring and vessel ring are patterned independent of each other in the whisker follicle

(A–B) Nerve/vessel immunostaining of wild-type littermate control exhibits normal double ring patterning (A). Nerve/vessel immunostaining of the Ngn1 knockout shows no trigeminal innervation in the whisker follicles, but blood vessel rings are normally organized at each whisker follicle (arrows in B, C–D). (C–D) For the quantification, 10 pairs of whisker follicles were analyzed and the mean distance from the center (C) and vasculature area surrounding the follicle (D) ± SEM is shown. (E–F) Endothelial specific-Npn1 deletion shows less vasculature formation around whisker follicle (white arrowheads in F and G) compared to the vasculature of control (E), but nerve rings are organized normally (H). (G–H) Vasculature area surrounding follicle (G) and mean distance from center to nerve ring (H) ± SEM is shown (n=14). Paired student t-test; ***, P<0.001; n.s, not significant. Scale bar: 200 μm (A–B and E–F)

Next, we examined whether blood vessel ring patterning has any effect on nerve ring formation. We analyzed the “double ring” structure in endothelial-specific neuropilin-1 (Npn-1) null mice (Tie-2 Cre; Npn1 flox/flox) where the vasculature surrounding the whisker follicles is poorly developed (Figures 2E and 2F) (Gu et al., 2003). However, nerve ring formation shows no detectable difference between mutants and control littermates (Figures 2G and 2H). This result demonstrates that nerve ring formation is also independent of vessel ring formation. Therefore, the neurovascular congruency of the double ring pattern in the whisker follicles is not established by the “one patterns the other” mechanism, but rather patterned independently.

Sema3e and Plxnd1 are complementarily expressed in the trigeminal-whisker follicle system

To reveal the novel mechanism governing the neurovascular congruency in the whisker pad system, we first considered the contributions from the target tissues on their coordinated patterning. Emerging evidence suggests that the neural and vascular systems share many similar mechanisms and molecular cues to regulate their development (Adams and Eichmann, 2010; Carmeliet and Tessier-Lavigne, 2005; Gelfand et al., 2009). Therefore, we next asked whether guidance cues could be found in the whisker follicle, and whether these cues could function as an organizing signal to coordinately pattern the nerves and vessels into the congruent “double ring” structure. We first performed expression screening of known guidance molecules in the trigeminal ganglion and whisker targets, and found a striking complementary expression pattern of Sema3e and its receptor Plxnd1 in these areas (Figures 3A–3H). Moreover, the spatiotemporal expression of these molecules coincides with the developmental profile of whisker follicles. Plxnd1 mRNA was expressed at very low levels in the trigeminal ganglion neurons at an early stage (E12.5), when axons arrive at the whisker target (Figure 3A, (van der Zwaag et al., 2002)). Its expression level is significantly increased by E14.5, when trigeminal axons and blood vessels begin to organize the double ring structure in the whisker follicles (Figure 3B), and continues through E16.5 and E18.5, when target innervation is refined and a clear double ring structure is apparent (Figure 3C and data not shown). Similarly to the temporal profile of Plxnd1 expression in the trigeminal ganglion, Sema3e is not expressed at E12.5 in the whisker follicles (Figure 3E). Sema3e mRNA is found in the mesenchymal sheath surrounding the hair follicles starting at E14.5 and continuing to E16.5 and E18.5 (Figures 3F3H). To characterize the precise location of Sema3e within the follicle, we performed immunostaining of axons after Sema3e in situ hybridization (ISH) on the same tissue sections and demonstrated that Sema3e was expressed inside the nerve ring (Figures 3I and 3J). Finally, to examine whether Plxnd1 is also expressed in the outer blood vessel ring, we performed double fluorescence ISH of Plxnd1 and the endothelial cell marker Flk-1. As shown in figure 3L, mRNA expression of Plxnd1 and Flk-1 completely overlaps within the vessel ring. Together these expression data suggest that in the whisker pad Sema3E expression surrounds the hair follicle (Figures 3K and 3M), while Plexin-D1 is expressed in the blood vessels and may also expressed in the innervating trigeminal axons based on its mRNA expression in the trigeminal ganglion (Figure 3M). This complementary expression pattern in the target area suggests that Sema3E-Plexin-D1 signaling plays a role in coordinating trigeminal target innervation and the formation of blood vessels into the double ring structure.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f3.jpg
Complementary expression pattern of Plexin-D1 in the TG and blood vessels and Sema3E in the whisker follicle

(A–D) Plxnd1 in situ hybridization (ISH) on sagittal sections of wild-type embryos at E12.5 (A), E14.5 (B), and E16.5 (C) shows that Plxnd1 mRNA is detected at very low level as early as E12.5 and is significantly increased at E14.5 and continues to E16.5 as the double ring structure develops. Plxnd1 mRNA is not detectable in sections of the trigeminal ganglion from plexin-D1 knockout animals at E14.5 (TG outlined by dashed line in D). (E–H) Sema3e mRNA ISH on coronal sections parallel to the whisker follicles at E12.5 (E), E14.5 (F), E16.5 (G) and E18.5 (H) shows that Sema3e mRNA is expressed in the mesenchymal sheath surrounding the hair follicles starting at E14.5 and continues at E16.5 and E18.5 (black arrows). Scale bar: 100 μm in A applies to G, and 200 μm (H) (I–J) Neurofilament immunostaining (white arrows) after Sema3e ISH on the same sections shows that Sema3e (black arrows) is expressed inside of the nerve rings (coronal image in I and tangential image in J). (K–L) Double fluorescence ISH with Plxnd1 and Sema3e (K) or Flk-1 (L) shows that Plxnd1 mRNA is expressed in the endothelial cells that form the vessel ring (white arrowhead in K and L) and Sema3e is expressed inside of both nerve and vessel rings. (M) Summary of Plxnd1 and Sema3e mRNA expression pattern in the whisker follicle. Sema3e is expressed in mesenchymal tissue (yellow) and Plxnd1 is expressed in blood vessel (red) and in TG therefore Plexin-D1 protein probably in nerve (green). Scale bar: 100 μm in A applies to G, 200 μm (H), and 100 μm in I applies to L.

Semaphorin 3E and its receptor Plexin-D1 facilitate repulsive signaling in both nerves and vessels

Does Sema3E function as a common cue to coordinately pattern nerves and vessels during the formation of the stereotypic double ring structure? Because trigeminal axon terminals in the whisker target are adjacent to Sema3e-expressing cells (Figures 3I and 3J), whereas blood vessels are located outside of the nerve ring (Figures 3K and 3L), we first tested the hypothesis that Sema3E may act as an attractive cue to trigeminal axons and a repulsive cue to blood vessels. Such an attractive guidance role for axons is consistent with prior observation that in the presence of neuropilin-1, a co-receptor for Plexin-D1, Sema3E can serve as an attractant (Chauvet et al., 2007) and our observation that Npn-1 is highly expressed in the TG at this stage (Figure S2B). However, to our surprise, application of alkaline phosphatase (AP)-tagged Sema3E to E14.5 TG explants induced significant growth cone collapse (Figure 4C) compared to the control AP-treated group (Figure 4A). Moreover, Sema3E-induced growth cone collapse was absent in trigeminal neurons isolated from Plxnd1 null mice (Figure 4D), indicating that Sema3E serves as a repulsive cue to trigeminal neurons and Plexin-D1 is required for its effects. To further examine whether Npn-1 plays any role in Sema3E-induced trigeminal growth cone collapse, we performed the same growth cone collapse assay using TG explants from Nestin-Cre driven-Npn1 conditional knockout embryos in which Npn-1 was ablated in all neuronal populations. Trigeminal ganglia isolated from these mice exhibit the same level of growth cone collapse as their wild-type littermate controls when treated with AP-Sema3E (Figures S2E and S2F), indicating that Sema3E-Plexin-D1 signaling induces trigeminal growth cone collapse independent of Npn-1. As a positive control, we also treated TG explants from these mice with AP-Sema3A. Sema3A-Npn-1 signaling causes trigeminal growth cone collapse in vitro and is required for the initial axon projection from the trigeminal ganglia to their peripheral targets in vivo (Gu et al., 2003; Kobayashi et al., 1997)(Figures 4E and 4F). As expected, axons from trigeminal ganglia isolated from Nestin-Cre; Npn1flox/flox mice were completely unresponsive to the Sema3A-induced growth cone collapse (Figure S2H). Finally, although vascular endothelial growth factor receptor 2 (VEGFR2/Flk-1) has been suggested to promote axonal growth in association with Plexin-D1/Neuroplin-1 receptor complex (Bellon et al., 2010), VEGFR2 is not expressed in the TG neurons during double ring formation (Figure S2K). Therefore, Sema3E acts on trigeminal axons as a repellant rather than an attractant.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f4.jpg
Sema3E-Plexin-D1 signaling provides a repulsive guidance cue to both TG axons and blood vessels in vitro

(A–G) Sema3E-Plexin-D1 signaling serves as repulsive cue for cultured TG neurons. Trigeminal ganglia were isolated from wild type (A, C, E) or Plexin-D1 null mice (B, D, F) at E14.5 and a growth cone collapse assay was performed. Axons (green) and growth cones (red) were visualized by anti-neurofilament and phalloidin staining, respectively. After incubation with 2 nM of AP (A, B), AP-Sema3E (C, D), or AP-Sema3A (E, F) for 30 min, >200 growth cones were scored for each experimental condition. Sema3E treatment induced significant growth cone collapse (white arrowheads in C), but growth cone collapse was absent in Plexin-D1 mutant TG neurons (D). (G) Quantification of growth cone collapse assay (n=3, data shown as mean ± SD, ANOVA; *, P<0.001). (H–L) Sema3E inhibits endothelial cell migration. VEGF-induced HUVEC transwell migration assay was performed in the presence (K) or absence (J) of Sema3E (AP-Sema3E, 0.5 nM). After a 5 hr incubation, migrated cells were fixed and visualized by Nissl staining. Sema3E prevents HUVEC migration induced by VEGF as well as the basal level of migration (I). (L) Quantification of migrated HUVECs (data shown as mean ± SEM, n=3, ANOVA; *, P<0.05; **, P<0.001). Scale bar: 50 μm (A–F, H–K). See also Figure S2.

To examine the effect of Sema3E on endothelial cells, we performed an in vitro transwell migration assay using human umbilical vein endothelial cells (HUVECs), which endogenously express Plexin-D1. As predicted, vascular endothelial growth factor (VEGF) caused the majority of endothelial cells to migrate to the bottom chamber (Figures 4J and 4L), reflecting the known function of VEGF as an attractant for these cells. However, when Sema3E and VEGF were added together, almost no migration was observed (Figures 4K and 4L), indicating that Sema3E inhibits VEGF-induced migration. Moreover, Sema3E also blocked the basal level of migration (Figure 4I). Together, these results support the idea that Sema3E can dominantly block the attractive effects exerted upon Plexin-D1-expressing endothelial cells.

Differential expression of Plexin-D1 protein allows for distinctive responses of nerves and vessels to Sema3E

The appealing hypothesis that emerged, based on these in vitro results, is that Sema3E acts as a repulsive guidance cue for both trigeminal axons and blood vessels and serves to organize the double ring neurovascular structure surrounding the follicle. If this hypothesis is true, what accounts for the relative position of the nerves and vessels, with the nerve ring consistently positioned inside of the vessel ring? In particular, it is puzzling how trigeminal axons expressing Plexin-D1 are able to innervate an area that is so close to a secreted repulsive cue. One potential possibility is that the abundance of Plxnd1 mRNA transcripts in the TG may not reflect protein levels at the nerve terminal. To test this idea, we performed AP-Sema3E binding on tissue sections to detect Plexin-D1 protein along the trigeminal nerve (Chauvet et al., 2007; Gu et al., 2005). As shown in Figure 5A, although Plexin-D1 protein is highly expressed in the trigeminal nerves projecting to the whisker follicle, surprisingly, Plexin-D1 protein is very low in the nerve terminals (white arrowheads in Figures 5B and 5C). This difference is not due to our inability to detect protein binding at the nerve terminal because Npn-1 protein is equally well detected in both the projecting axons and the nerve terminal by AP-Sema3A binding (arrowheads in Figures 5D and 5E). The absence of AP-Sema3E binding on the vessel ring (red arrow in Figure S3C) of Plxnd1 null mice further confirmed that Plexin-D1 is the only receptor for Sema3E in the vessel ring (Figures S3A–S3D). The absence of AP-Sema3E binding on the nerve ring in Sema3e null mice also ruled out the possibility that the lack of AP-Sema3E binding may be due to the sequestration of Plexin-D1 by endogenous Sema3E (Figures S3E-S3H). To further confirm the selective down-regulation of Plexin-D1 protein in the TG nerve terminals, we also performed anti-Plexin-D1 immunohistochemistry (Chauvet et al., 2007). Consistent with AP-Sema3E binding result, TG nerve terminal exhibited extremely low Plexin-D1 immunoreactivity (Figures 5F and 5G). Therefore, the Plexin-D1 protein is selectively down-regulated in the TG nerve terminals, which enables the nerves to innervate areas close to the Sema3E-expressing region and form the inner ring. In stark contrast to the low expression of Plexin-D1 in the nerve ring, Plexin-D1 protein level visualized by both AP-Sema3E and anti-Plexin-D1 antibodies in the blood vessels is remarkably high (white arrows in Figures 5B, 5C, 5H, and 5I). Therefore, the differential expression of localized Plexin-D1 protein may lead to a differential response between the nerves and vessels to Sema3E, resulting in the nerve-inside and vessel-outside double ring structure.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f5.jpg
Plexin-D1 protein expression is selectively down-regulated in the target terminal of TG axons

(A–C) Plexin-D1 protein expression is visualized by AP-Sema3E binding. Tangential sections of tissue were cut at E14.5 and then incubated with 1 nM AP-Sema3E. Binding was then detected by reaction with an AP substrate. Plexin-D1 (black arrows in A) is highly expressed within the TG (dotted red outline) and the trigeminal nerves that project both centrally toward the brainstem and peripherally towards the snout (right side from dotted green outline). But Plexin-D1 protein is extremely low in nerves when they reach the whisker follicle (white arrowhead in B and C). In contrast, Plexin-D1 protein is strongly detected in the blood vessel ring (white arrows in B and C). Yellow arrowhead (in B and C) indicates non-specific signals in membrane structure of the whisker follicle. (C) To identify the relative location of nerve and blood vessel rings, adjacent sections were immunostained with NF (green) and PECAM (red) and overlaid onto the AP-stained sections. (D–E) Neuropilin-1 protein expression is detected by AP-Sema3A (2 nM) binding in the terminal nerve ring (arrowheads in D-E), which completely overlaps with NF-positive nerve ring on the same sections after AP-staining (E). (F–I) Plexin-D1 protein expression is visualized by immunohistochemistry with anti-Plexin-D1. In the TG nerve terminal surrounding the whisker follicles, Plexin-D1 protein expression is extremely low (arrowheads in F and G). In contrast, Plexin-D1 protein is strongly detected in the blood vessel ring (arrows in H and I). Anti-Plexin-D1 displays non-specific signals in the hair sheath and glassy membrane of the whisker follicle (asterisks in F and H). Scale bar: 1 mm (A), 50 μm (BE, F-I). See also Figure S3.

Sema3E-Plexin-D1 signaling is required for proper “double ring” formation in vivo

To directly test the in vivo requirement of Sema3E-Plexin-D1 signaling in the formation of the neurovascular double ring structure in each whisker follicle, we analyzed nerve and vessel patterning in the developing whisker follicle in Plxnd1 null and Sema3e null embryos. In wild-type embryos at E16.5, the outer ring of blood vessels is distinctly separated from the inner ring of nerves (Figures 6A and 6D). However, in the mutant embryos, the two rings are intermingled (arrows in Figures 6B and 6E). To quantitatively analyze this phenotype, we measured the ratio between the inner ring radius (r) and outer ring radius (R) in the same follicle and then compared the averaged ratios from wild-type embryos to that in mutant mice (Figure 6I). We observed a significant increase in the r/R ratio in the mutant animals as compared to the wild-type littermate controls, reflecting a collapsed double ring structure in the mutant mice (Figures 6C and 6F). The average distance from the follicle center to nerve /vessel ring in mutants and controls were also measured, respectively (Figures S4A and S4B). The vessel ring moved inward significantly whereas nerve ring position relative to center showed no noticeable changes, consistent with the selective down-regulation of Plexin-D1 at the nerve terminal. To see clear structure of the nerve and vessel alignment in the whisker follicle, whole-mount follicle staining with anti-neurofilament and VE-cadherin antibodies was performed. The blood vessel layer is separated from trigeminal nerve layer along with whole follicle shaft (Figure 6G). However, the nerve and vessel alignment is overlapped in multiple regions in the whole follicle shaft (Figure 6H). Therefore, Sema3E-Plexin-D1 signaling is required in vivo for proper nerve and vessel organization in the whisker follicle.

An external file that holds a picture, illustration, etc.
Object name is nihms571974f6.jpg
Plexin-D1 and Sema3E is required for the stereotypic double ring nerve/vessel patterning in the whisker pad in vivo

(A–C) Plexin-D1 null mice exhibit severe nerve/vessel organization defects. The whisker follicle region of Plexin-D1 mutants (B) and their littermate controls (A) at E16.5 were sectioned and immunostained with NF and PECAM. In the Plexin-D1 mutant, nerve and vessel rings were abnormally close, with some areas showing complete vessel/nerve intermingling (arrow in B). Radii of the vessel ring (R) and nerve ring (r) depicted in I were measured and their ratio was calculated for each follicle (C). (D–F) Sema3E null mice also show the same nerve/vessel patterning phenotype (white arrow in E) compared to their littermate controls (D) and r/R ratio measurement is shown in F. For the quantification, 6–8 whisker follicles per animal from three different pairs of embryos in each experimental condition were measured and the mean ratios ± SEM are shown (C, F). Paired student t-test; *, P<0.05 (G–I) The detail patterning phenotype was analyzed in coronal view of whisker follicles using confocal imaging process after neurofilament and VE-cadherin whole mount staining. The blood vessel layer (white arrows in G) is clearly separated from the TG nerve layer in wild-type (G), but in the Plxnd1 mutant, vessels (yellow arrows in H) are intermingled with nerve layers (H). White arrowhead (in G and H) indicates non-specific signals from the whisker sheath marked in white asterisks. Scale bar: 100 μm (A, B, D, and E) and 50 μm (G and H). See also Figure S4.

Attractive signals for the initial recruitment of nerve and vessel are expressed in the whisker follicles

Our data demonstrated that in the developing whisker pad, the nerve and vessel are initially intermingled, then the nerve ring forms followed by the vessel ring to eventually establish the “double ring” structure with the nerve ring inside and vessel ring outside. Moreover, the nerve and vessel rings are formed independently of each other, rather than “one patterns the other”. Sema3E originating from the mesenchymal sheath surrounding the hair follicles controls their relative position in the “double ring structure” through its receptor Plexin-D1. What are the signals that initially recruit nerves and vessels in the whisker follicle? Two major chemoattractants for nerve and vessel recruitment, NGF and VEGF, respectively, are expressed around the whisker follicle when the double ring structure is forming (Bandtlow et al., 1987; Genc et al., 2005)(Figures S5). At E12.5 when nerve ring begins to form, we detected TrkA (NGF receptor) expression in the trigeminal nerves and Ngf expression around each whisker primordium (Figures S5A-S5H). Moreover, TG and maxillary explant co-culture assay demonstrated that NGF is necessary to attract TG axons (O’Connor and Tessier-Lavigne, 1999), suggesting that NGF-TrkA signaling contributes to the initial nerve ring formation. To directly test this hypothesis, we analyzed the initial nerve ring formation in mice lacking Ngf at E12.5-E13.5. We observed a normal size of TG and normal peripheral projections of trigeminal nerves to the snout area in Ngf mutant mice, consistent with the previous finding that mouse trigeminal ganglion survival become NGF-dependent after E14.5 (Figures S6A and S6B) (Pinon et al., 1996). However, despite of normal arrival at the whisker pad, TG axons failed to form the initial “ring like” structure in the Ngf mutant embryos. At E12.5 when a ring-like structure of axonal innervation forms around each whisker primordium in the wild-type littermates (Figures 1C, S6C-D, and S6G), but the axonal innervation in Ngf mutant embryos aggregates and fails to form the fasciculated ring-like structure (Figures S6E-G). At E13.5 when a complete nerve ring structure forms in the wild-type littermates (Figure S6H), the axonal innervation still “wanders” and fails to form a ring structure around the whisker primordium (Figure S6I). In addition, this initial failure of nerve ring formation defect seems to be continued postnatally, because whisker innervation defects have also been observed in new born Bax; trkA mice at P0 (Patel et al., 2000). Together, these data provides in vivo evidence demonstrating that NGF signaling is critical for the initial nerve ring formation. At E13.5, right before the blood vessel ring forms, we detected strong VEGF expression around each whisker primordium in an area close to the nerve ring (Figures S5I-S5K). Moreover, endothelial-specific knockout of Npn1, a VEGF receptor, results in aberrant vessel ring formation (Figures 2E and 2F). These data demonstrate that VEGF signaling contributes to the initial vessel ring formation. Interestingly, both NGF and VEGF-expressing cells are arrayed along the circumference of the primordium, thereby enabling the recruitment of both nerve and vessel in a ring like structure during development. At E14.5, the repulsive signaling of Sema3E-Plexin-D1 emerges to set the two rings apart and achieve the “double ring” structure. Thus, these results suggest that balanced attractive and repulsive signals from the target tissue control the ontogenetic patterning of the neurovascular double ring structure (Figure 7).

An external file that holds a picture, illustration, etc.
Object name is nihms571974f7.jpg
Model of how stereotypic double ring neurovascular structure is established during development

(Left) During early development (E11–E12), NGF is expressed in tissue surrounding the primitive whisker follicle and attracts trigeminal axons to the whisker target area. At this stage, the blood vessel plexus is dispersed in the whisker follicle and is not organized. (Middle) As development progress, around E13–E14 when trigeminal nerve forms a nerve ring structure, VEGF expression in the whisker follicle region begins to recruit blood vessel close to the nerve ring. (Right) Around E14–E16, Sema3E starts to be expressed in the mesenchymal sheath of the whisker follicle and repels incoming vessels through its interaction with Plexin-D1, which then settle outside of nerve ring. Although Plexin-D1 is also expressed in the trigeminal nerve at this stage, its expression in the axon terminals is selectively down-regulated to a level that is not sufficient to respond to Sema3E, thereby maintaining the location of the nerve ring close to the whisker follicle. See also Figures S5 and S6.

Taken together, this study demonstrates that the stereotypic neurovascular congruency in complex tissue is established during development by a novel independent patterning mechanism by cues emanated from the target tissue. Using the mouse whisker pad as a model system, we found that nerves and vessels form a stereotypic “double ring” neurovascular structure around each of the whisker follicles during development. Surprisingly, this “double ring” neurovascular congruency is not established by a “one patterns the other” mechanism, but rather is established via a novel mechanism in which nerves and vessels are patterned independently through differential responses to a common guidance cue. Sema3E secreted from the center of the follicle has the potential to repel both nerves and vessels through Plexin-D1. However, differential Plexin-D1 expression in nerves and vessels during whisker follicle development is translated into a distinctive repulsive strength that defines the relative location of nerve and vessel rings. Finally, formation of this double ring structure is disrupted in mice lacking Plxnd1 and Sema3e. Unlike the previously described “one patterns the other” mechanism (Mukouyama et al., 2002), our study demonstrates that local signals act as a central organizer in complex tissues to establish neurovascular congruency via a novel independent patterning mechanism.

Discussion

In this study, we address a critical developmental question about which little is known -how nerves and blood vessels form congruent patterns to facilitate their interdependent functions. We demonstrate that neurovascular congruency in the whisker follicle is not established by a “one patterns the other” mechanism, but rather patterned independently through a balance of attractive and repulsive cues originating from the surrounding environment, to achieve the final congruent “double ring” structure.

While NGF and VEGF can serve as the initial attractive cues to organize nerves and vessels respectively to the center around the hair follicle, Sema3E is responsible for positioning the two rings to the final double ring structure with the nerve ring inside and vessel ring outside. Even though Sema3E-Plexin-D1 signaling exerts repulsive signals in both neurons and endothelial cells (Figure 4) (Oh and Gu, 2013) (Ding et al., 2012) the nerve ring and vessel ring respond differently to this repulsive signal because of different levels of Plexin-D1 in the nerve and vessel rings. In the case of vessels, Plexin-D1 level is high and therefore vessels actively respond to Sema3E’s repulsive signal, resulting in the outer ring position. In contrast, the selective down-regulation of Plexin-D1 protein levels in the nerve terminal removes the sensitivity of the axons to the repulsive ligand, resulting in the inner nerve ring position and thereby allowing the whole final double ring structure to form. What is the potential mechanism underlying the selective Plexin-D1 protein down-regulation at the nerve terminal? Recently, a study in the mouse spinal cord demonstrated that calpain-mediated proteolytic processing of the Plexin A1 receptor in pre-crossing commissural axons prevents responsiveness to the repulsive cue Sema3B, until the axons have crossed the midline (Nawabi et al., 2010). Indeed protease-mediated regulation of receptor levels plays a critical role in controlling spatial and temporal aspects of ligand responsiveness during neural development and disease (Bai and Pfaff, 2011). We postulate that a similar mechanism of protease activity might be responsible for the selective down-regulation of Plexin-D1 at the nerve terminal to silence the nerve ring responsiveness to Sema3E. Another possibility is that the selective down-regulation of Plexin-D1 happens at the local translation level. Recent studies demonstrated that the local synthesis of axon guidance receptors can be controlled at translational level (Colak et al., 2013; Tcherkezian et al., 2010). Testing these hypotheses will be of interest in future studies.

The use of an independent patterning mechanism in establishing neurovascular congruency provides an intriguing contrast with the “one patterns the other” model shown in previous studies of the limb skin and sympathetic system. In the developing mouse forelimb skin, peripheral sensory nerves determine the differentiation and branching pattern of arteries (Mukouyama et al., 2005; Mukouyama et al., 2002) (Li et al., 2013), indicating that the nerve guides the vessel. Conversely, there are also cases where vessels can express signals that then attract axons. For example, artemin is expressed in the smooth muscle cells of the vessels and attracts sympathetic fibers to follow these blood vessels (Honma et al., 2002). Similarly, blood vessel-expressed endothelins direct the extension of sympathetic axons from the superior cervical ganglion to the external carotid artery (Makita et al., 2008). This “one patterns the other” mechanism was thought to represent a general rule governing the establishment of neurovascular congruency. However, in these examples, there is a relatively simple organization of the aligned nerves and vessels, and neurovascular networks in different tissues are very diverse. Here in the whisker pad, the double ring neurovascular congruency is not established by one system patterning the other, but rather by a novel independent patterning mechanism.

Why should two distinct mechanisms be used to establish congruency? In the case of a target tissue with a planar structure or during pathfinding before reaching a target, the “one patterns the other” model allows for the parallel trajectories of nerves and vessels, independent of their position relative to their surroundings. However, in target tissues with complex three-dimensional structures, the precise architecture of the trio of nerves, vessels and target tissues becomes functionally relevant. This functional organization is clearly the case in the whisker follicles, where the nerve ring must be located closer to, and the vessel ring farther from, the whisker pad to enable proper neurovascular regulation of the whisker itself. The independent/or coordinate patterning model enables the target tissue to act as a central organizer to control the coordinated development of multiple tissue subcomponents. Given the diversity of neurovascular structures in various tissues, and the variable timing of neural innervation and vascular ingression in various tissues, local signals provided by a central organizer within the specific target tissue may be a common mechanism used to establish neurovascular congruency patterns that facilitate target tissue function.

The close anatomic apposition between nerves and vessels is critical for their functional interdependence and revealing the signals that mediate the tight neurovascular association provides new insights into our understanding of nervous system function. This is more evident in the brain where neural cells (neurons and glia) and vascular cells (endothelia, pericytes) form a functionally integrated network that is collectively termed the neurovascular unit (Iadecola, 2004). Neurovascular unit integrity has recently been shown to regulate important physiological functions, and is linked to the onset and progression of various neurodegenerative diseases. For example, a deficient endothelial-secreted PDGF-BB-PDGFβ signaling on pericytes within the vascular system leads to defects in cerebrovascular integrity and subsequently results in neuronal dysfunction (Armulik et al., 2010; Bell et al., 2010) (Daneman et al., 2010). Similarly, a recent study shows that deficient signaling between astrocyte-secreted apoE, a major risk factor for Alzhemimer’s dementia, and its binding protein LRP1 on pericytes can also lead to cerebrovascular defects followed by neuronal degeneration (Bell et al., 2012). The functional consequences of periphery neurovascular congruency defects on nervous system function are less understood. In the whisker system, earlier studies suggest that the vascular component affiliated with each whisker affects both the movement of the whisker and modulates the sensitivity of the sensory nerve endings (Fundin et al., 1997; Wineski, 1985). It will be an important future direction to use the animal model we have generated here to investigate whether developmental deficits in establishing neurovascular patterning in the whisker pad target areas will manifest as functional and behavioral deficits in the mature animals. Both the peripheral trigeminal ganglion and higher level barrel cortical electrophysiological properties in response to whiskering can be used as neural functional readouts. It will also be interesting to examine functional consequence of neurovascular congruency defects in other peripheral examples such as forelimb skin sensory axon / arterial branching system (Mukouyama et al., 2005; Mukouyama et al., 2002) (Li et al., 2013) and the mouse sympathetic axon/ blood vessel (Honma et al., 2002) (Makita et al., 2008). Understanding how highly stereotyped neurovascular structure is formed to facilitate organ-specific functions will provide insights into the homeostasis and pathogenesis of medical disorders that involve both nerves and vessels.

Experimental Procedures

Animals

Plxnd1flox/flox mice (Kim et al., 2011), Nestin-Cre mice (Tronche et al., 1999), and Tie2-Cre mice (Kisanuki et al., 2001) were maintained on a C57Bl/6 background. Plxnd1+/− mice (Gu et al., 2005), Npn1 mice (Gu et al., 2003) and VegflacZ mice (Miquerol et al., 1999) were maintained on an outbred Swiss Webster background. Sema3e+/− mice were maintained on an 129SVE background. Ngn1 (Ma et al., 1998) and Ngf (Crowley et al., 1994) knockout embryos were obtained from Dr. Quifu Ma (Harvard Medical School) and Dr. Rejji Kuruvilla (Johns Hopkins University), respectively. Swiss Webster, C57Bl/6, 129SVE wild-type mice were obtained from Taconic Farms, Inc. All animals were treated according to institutional and NIH guidelines approved by IACUC at Harvard Medical School.

Immunohistochemistry

Embryos were removed, immediately frozen in liquid nitrogen, and sectioned at 16 μm using a cryostat (Leica). Sections were fixed in 4% paraformaldehyde (PFA) for 5 min and briefly washed in 1X phosphate buffered saline (PBS) several times at room temperature. After washing, sections were blocked in 1X PBS containing 0.1% Triton X-100 (PBST) and 10% normal goat serum for 1 hr, and then incubated with primary antibodies in blocking buffer at 4°C overnight. For Plexin-D1 immunohistochemisty, embryos were fixed in 4% PFA in 0.1M phosphate buffer (pH 7.5) for 2 hr and equilibrated in 30% sucrose at 4°C overnight. The antibodies used are: α-neurofilament (1:100; 2H3, Developmental Studies Hybridoma Bank), α-PECAM (1:500; 553370, BD Pharmingen™), α-Plexin-D1 (1:6000, a gift from Dr. Yutaka Yoshida, Cincinnati Children’s Hospital), and α-TrkA (1:500; 06-574, Millipore). After washing in 1X PBST for 5 min three times, sections were incubated in AlexaFluor conjugated secondary antibodies (1:1,000, Invitrogen) for 1 hr, then washed several times in PBST and mounted with Fluoromount G (Electron Microscopy Sciences). Immunostained sections were analyzed by fluorescence microscopy using a Nikon Eclipse 80i microscope equipped with a Nikon DS-2 digital camera. Images were processed using Adobe Photoshop and ImageJ (National Institutes of Health).

Wholemount immunostaining

For wholemount whisker follicle staining, embryos were fixed in 4% PFA for 6 hr and equilibrated in 30% sucrose at 4°C overnight. Embryo snout areas were sectioned at 100 μm using a cryostat (Leica) and blocked in 1X PBS containing 1% Triton X-100 (PBST), 10% dimethylsulfoxide and 5% normal goat serum for 1 hr, and then incubated with primary antibodies in blocking buffer at room temperature for 2 days. The antibodies used are: α-neurofilament (1:50; 2H3) and α-VE-cadherin (1:100; ab33168, Abcam). After washing in 1X PBST for 1 hr five times, sections were incubated in AlexaFluor conjugated secondary antibodies (1:500, Invitrogen) diluted in the blocking buffer for 1 day, then washed several times in PBST and dehydrated with 100% methanol. Before mounting, sections were cleared with benzyl alcohol and benzyl benzoate mixture. Sections were analyzed by confocal laser-scanning microscopy using a Zeiss LSM 510 META.

In situ hybridization

Non-radioactive in situ hybridization (ISH) for tissue sections and whole mount ISH were performed according to standard methods (Gu et al., 2005). Briefly, 16-μm-thick- cryosections were fixed in 4% PFA, acetylated in 1% triethanolamine and 0.25% acetic anhydride, prehybridized, and hybridized with indicated probes at 60°C. After hybridization, sections were washed and incubated with sheep anti-Digoxigenin-alkaline phosphatase (AP) antibody (1:3,000, Roche) for 90 min at room temperature. After several washes, sections were incubated in BM Purple (Roche) until positive signal was detected. To perform immunostaining on the same section after ISH, sections were washed in 1X PBS several times and post-fixed in 4% PFA for 5 min. After fixation, all procedures were followed as described in the immunohistochemistry section above. Double fluorescence in situ hybridization was performed using the tyramide signal amplification method according to the manufacturer’s instructions (NEL753001KT, PerkinElmer, Inc.). Two FITC-or DIG-labeled anti-sense probes were simultaneously hybridized and stained by fluorescein or Cy3 chromogens, respectively. The following anti-sense probes were used: Plexin-D1 (Kim et al., 2011), Sema3E (Gu et al., 2005), Npn-1 (He and Tessier-Lavigne, 1997), and Flk-1 ({"type":"entrez-nucleotide","attrs":{"text":"NM_010612","term_id":"1388591291","term_text":"NM_010612"}}NM_010612, nt1277-2249).

Ligand preparation and AP-ligand binding assay

Alkaline phosphatase (AP)-tagged ligands were produced in HEK 293T cells as previously described (Gu et al., 2002). Expression construct was transfected into cells by LipofectAMINE 2000 (Invitrogen) and conditioned medium was harvested at 48 hr post-transfection. For AP-ligand binding assay (Gu et al., 2003), 20-μm-thick-cryosections were fixed in cold methanol and pre-incubated in 1X PBS containing 4 mM MgCl2 and 10% fetal bovine serum for 1 hr. Sections were incubated in the binding solution (1X PBS-MgCl2 plus 20 mM HEPES, pH 7.0) containing 1–2 nM ligands for 2 hr at room temperature. After several washes in 1X PBS-MgCl2, sections were fixed in acetone and formaldehyde fixative, then heat inactivated at 65°C for 2 hr. For AP chromogenic development, sections were incubated in AP buffer (100 mM Tris-HCl pH 9.5, 100 mM NaCl, 5 mM MgCl2) containing 4-Nitro blue tetrazolium chloride and 5-Bromo-4-chloro-3-indolyl-phosphate (Roche).

Growth cone collapse assay

TG explant growth cone collapse assay was performed as previously described (Behar et al., 1999). Briefly, TG explants isolated from E14.5 embryos are placed on growth factor reduced Matrigel (356230, BD)-coated cover glasses and cultured overnight at 37°C in DMEM/F12 containing 0.6 mg/ml cellulose and 20 ng/ml NGF (a gift from Dr. Rejji Kuruvilla, Johns Hopkins University). The next day, without removing culture media, 2x concentrated prewarmed ligands (Kim et al., 2011) were applied to explants for 30 min, and immediately fixed in 4% PFA for 30 min. After three washes in 1X PBS and blocking, explants were incubated with anti-neurofilament antibody (1:100), followed by anti-mouse Alexa-488 conjugated secondary antibody (1:1,000) and Alexa-568 conjugated phalloidin (1:50) mixture incubation. Then explants were washed in 1X PBS several times and mounted blindly. To quantify collapsed growth cone, randomly selected fields of TG neurons were imaged and collapsed versus intact growth cones were scored as done in previous reports (Cox et al., 1990; Ughrin et al., 2003). Briefly, growth cones with broad lamellipodia were defined as intact whereas growth cones lacking lamellipodia and having only a few sharp filopodia were counted as collapsed.

HUVEC migration assay

Human umbilical vein endothelial cells (HUVECs, cc-2517, Lonza) were maintained in endothelial cell basal medium-2 (EBM-2, CC-3156, Lonza) supplemented with EGM-2 growth factor mixture (CC-4176, Lonza). HUVECs (5 X 10) were seeded on upper chamber of fibronectin (Calbiochem)-coated Transwell insert (Falcon #3097, 8 mm pore size) with 0.5 nM ligands with or without 50 ng/ml of VEGF in the lower chamber. After 5 hr incubation, filters were fixed in 4% PFA and stained with 0.5% crystal violet for 10 min. Migrated HUVECs were imaged and random fields from each image were counted to calculate the migrated cell number per area.

Statistical analysis

Statistical analyses were performed using Prism4 (GraphPad Software Inc.). Summary data are reported as mean ± SD or mean ± SEM. Multiple samples were analyzed with One-way ANOVA and two samples were analyzed with nonparametric student t-test. P<0.05 was considered as statistically significant.

Animals

Plxnd1flox/flox mice (Kim et al., 2011), Nestin-Cre mice (Tronche et al., 1999), and Tie2-Cre mice (Kisanuki et al., 2001) were maintained on a C57Bl/6 background. Plxnd1+/− mice (Gu et al., 2005), Npn1 mice (Gu et al., 2003) and VegflacZ mice (Miquerol et al., 1999) were maintained on an outbred Swiss Webster background. Sema3e+/− mice were maintained on an 129SVE background. Ngn1 (Ma et al., 1998) and Ngf (Crowley et al., 1994) knockout embryos were obtained from Dr. Quifu Ma (Harvard Medical School) and Dr. Rejji Kuruvilla (Johns Hopkins University), respectively. Swiss Webster, C57Bl/6, 129SVE wild-type mice were obtained from Taconic Farms, Inc. All animals were treated according to institutional and NIH guidelines approved by IACUC at Harvard Medical School.

Immunohistochemistry

Embryos were removed, immediately frozen in liquid nitrogen, and sectioned at 16 μm using a cryostat (Leica). Sections were fixed in 4% paraformaldehyde (PFA) for 5 min and briefly washed in 1X phosphate buffered saline (PBS) several times at room temperature. After washing, sections were blocked in 1X PBS containing 0.1% Triton X-100 (PBST) and 10% normal goat serum for 1 hr, and then incubated with primary antibodies in blocking buffer at 4°C overnight. For Plexin-D1 immunohistochemisty, embryos were fixed in 4% PFA in 0.1M phosphate buffer (pH 7.5) for 2 hr and equilibrated in 30% sucrose at 4°C overnight. The antibodies used are: α-neurofilament (1:100; 2H3, Developmental Studies Hybridoma Bank), α-PECAM (1:500; 553370, BD Pharmingen™), α-Plexin-D1 (1:6000, a gift from Dr. Yutaka Yoshida, Cincinnati Children’s Hospital), and α-TrkA (1:500; 06-574, Millipore). After washing in 1X PBST for 5 min three times, sections were incubated in AlexaFluor conjugated secondary antibodies (1:1,000, Invitrogen) for 1 hr, then washed several times in PBST and mounted with Fluoromount G (Electron Microscopy Sciences). Immunostained sections were analyzed by fluorescence microscopy using a Nikon Eclipse 80i microscope equipped with a Nikon DS-2 digital camera. Images were processed using Adobe Photoshop and ImageJ (National Institutes of Health).

Wholemount immunostaining

For wholemount whisker follicle staining, embryos were fixed in 4% PFA for 6 hr and equilibrated in 30% sucrose at 4°C overnight. Embryo snout areas were sectioned at 100 μm using a cryostat (Leica) and blocked in 1X PBS containing 1% Triton X-100 (PBST), 10% dimethylsulfoxide and 5% normal goat serum for 1 hr, and then incubated with primary antibodies in blocking buffer at room temperature for 2 days. The antibodies used are: α-neurofilament (1:50; 2H3) and α-VE-cadherin (1:100; ab33168, Abcam). After washing in 1X PBST for 1 hr five times, sections were incubated in AlexaFluor conjugated secondary antibodies (1:500, Invitrogen) diluted in the blocking buffer for 1 day, then washed several times in PBST and dehydrated with 100% methanol. Before mounting, sections were cleared with benzyl alcohol and benzyl benzoate mixture. Sections were analyzed by confocal laser-scanning microscopy using a Zeiss LSM 510 META.

In situ hybridization

Non-radioactive in situ hybridization (ISH) for tissue sections and whole mount ISH were performed according to standard methods (Gu et al., 2005). Briefly, 16-μm-thick- cryosections were fixed in 4% PFA, acetylated in 1% triethanolamine and 0.25% acetic anhydride, prehybridized, and hybridized with indicated probes at 60°C. After hybridization, sections were washed and incubated with sheep anti-Digoxigenin-alkaline phosphatase (AP) antibody (1:3,000, Roche) for 90 min at room temperature. After several washes, sections were incubated in BM Purple (Roche) until positive signal was detected. To perform immunostaining on the same section after ISH, sections were washed in 1X PBS several times and post-fixed in 4% PFA for 5 min. After fixation, all procedures were followed as described in the immunohistochemistry section above. Double fluorescence in situ hybridization was performed using the tyramide signal amplification method according to the manufacturer’s instructions (NEL753001KT, PerkinElmer, Inc.). Two FITC-or DIG-labeled anti-sense probes were simultaneously hybridized and stained by fluorescein or Cy3 chromogens, respectively. The following anti-sense probes were used: Plexin-D1 (Kim et al., 2011), Sema3E (Gu et al., 2005), Npn-1 (He and Tessier-Lavigne, 1997), and Flk-1 ({"type":"entrez-nucleotide","attrs":{"text":"NM_010612","term_id":"1388591291","term_text":"NM_010612"}}NM_010612, nt1277-2249).

Ligand preparation and AP-ligand binding assay

Alkaline phosphatase (AP)-tagged ligands were produced in HEK 293T cells as previously described (Gu et al., 2002). Expression construct was transfected into cells by LipofectAMINE 2000 (Invitrogen) and conditioned medium was harvested at 48 hr post-transfection. For AP-ligand binding assay (Gu et al., 2003), 20-μm-thick-cryosections were fixed in cold methanol and pre-incubated in 1X PBS containing 4 mM MgCl2 and 10% fetal bovine serum for 1 hr. Sections were incubated in the binding solution (1X PBS-MgCl2 plus 20 mM HEPES, pH 7.0) containing 1–2 nM ligands for 2 hr at room temperature. After several washes in 1X PBS-MgCl2, sections were fixed in acetone and formaldehyde fixative, then heat inactivated at 65°C for 2 hr. For AP chromogenic development, sections were incubated in AP buffer (100 mM Tris-HCl pH 9.5, 100 mM NaCl, 5 mM MgCl2) containing 4-Nitro blue tetrazolium chloride and 5-Bromo-4-chloro-3-indolyl-phosphate (Roche).

Growth cone collapse assay

TG explant growth cone collapse assay was performed as previously described (Behar et al., 1999). Briefly, TG explants isolated from E14.5 embryos are placed on growth factor reduced Matrigel (356230, BD)-coated cover glasses and cultured overnight at 37°C in DMEM/F12 containing 0.6 mg/ml cellulose and 20 ng/ml NGF (a gift from Dr. Rejji Kuruvilla, Johns Hopkins University). The next day, without removing culture media, 2x concentrated prewarmed ligands (Kim et al., 2011) were applied to explants for 30 min, and immediately fixed in 4% PFA for 30 min. After three washes in 1X PBS and blocking, explants were incubated with anti-neurofilament antibody (1:100), followed by anti-mouse Alexa-488 conjugated secondary antibody (1:1,000) and Alexa-568 conjugated phalloidin (1:50) mixture incubation. Then explants were washed in 1X PBS several times and mounted blindly. To quantify collapsed growth cone, randomly selected fields of TG neurons were imaged and collapsed versus intact growth cones were scored as done in previous reports (Cox et al., 1990; Ughrin et al., 2003). Briefly, growth cones with broad lamellipodia were defined as intact whereas growth cones lacking lamellipodia and having only a few sharp filopodia were counted as collapsed.

HUVEC migration assay

Human umbilical vein endothelial cells (HUVECs, cc-2517, Lonza) were maintained in endothelial cell basal medium-2 (EBM-2, CC-3156, Lonza) supplemented with EGM-2 growth factor mixture (CC-4176, Lonza). HUVECs (5 X 10) were seeded on upper chamber of fibronectin (Calbiochem)-coated Transwell insert (Falcon #3097, 8 mm pore size) with 0.5 nM ligands with or without 50 ng/ml of VEGF in the lower chamber. After 5 hr incubation, filters were fixed in 4% PFA and stained with 0.5% crystal violet for 10 min. Migrated HUVECs were imaged and random fields from each image were counted to calculate the migrated cell number per area.

Statistical analysis

Statistical analyses were performed using Prism4 (GraphPad Software Inc.). Summary data are reported as mean ± SD or mean ± SEM. Multiple samples were analyzed with One-way ANOVA and two samples were analyzed with nonparametric student t-test. P<0.05 was considered as statistically significant.

Supplementary Material

supp data

supp data

Click here to view.(1.2M, pdf)

Acknowledgments

We thank Drs. Bob Datta, Michael Greenberg, Rejji Kuruvilla, Qiufu Ma, and members of the Gu laboratory for helpful comments on the manuscript; Dr. Qiufu Ma for providing Ngn1 knockout embryos; Drs. Rejji Kuruvilla and Rajshri Joshi for providing NGF and Ngf knockout embryos; Dr. David Ginty and Siyi Huang for providing Ngf knockout embryos; Dr. Yutaka Yoshida for providing Plxnd1flox/flox mice and anti-Plexin-D1 antibody; Drs. Christopher Henderson and Fanny Mann for providing Sema3E null mice; Dr. Susan Dymecki for providing Nestin-Cre mice and Vegf-lacZ mice; Dr. Reha Erzurumlu for technical advice; National Cancer Institute-Frederick for providing VEGF; Optical Imaging Program at Harvard NeuroDiscovery Center for helping with confocal images. This work was supported by the Lefler postdoctoral fellowship (W.O.), Alice and Joseph Brooks Fund Postdoctoral Fellowship (W.O.), and the following grants to C.G.: Sloan research fellowship, March of Dimes Basil O’Connor award, Armenise junior faculty award, and NIH grant R01NS064583.

Department of Neurobiology, Harvard Medical School, 220 Longwood Ave, Boston, MA 02115, U.S.A
Correspondence and requests for materials should be addressed to C.G., Chenghua Gu (ude.dravrah.smh@ug_auhgnehc), Department of Neurobiology, Harvard Medical School, Boston, MA 02115, U.S.A., (Ph) 617-432-6364 (Fax) 617-432-1639

Summary

Nerves and vessels often run parallel to one another, a phenomenon that reflects their functional interdependency. Previous studies have suggested that neurovascular congruency in planar tissues such as skin is established through a “one patterns the other” model in which either the nervous or vascular system precedes developmentally, and then instructs the other system to form using its established architecture as a template. Here we find that in tissues with complex three-dimensional structures such as the mouse whisker system, neurovascular congruency does not follow the previous model, but rather is established via a novel mechanism in which nerves and vessels are patterned independently. Given the diversity of neurovascular structures in different tissues, guidance signals emanating from a central organizer in the specific target tissue may act as an important mechanism to establish neurovascular congruency patterns that facilitate unique target tissue function.

Summary
Collaboration tool especially designed for Life Science professionals.Drag-and-drop any entity to your messages.