MECHANISMS OF DEVELOPMENT
1 3 3 ( 2 0 1 4 ) 1 –1 0
Available at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/modo
An organizing function of basement membranes in the developing nervous system Willi Halfter *, Joseph Yip Department of Neurobiology, University of Pittsburgh, Pittsburgh, USA
A R T I C L E I N F O
A B S T R A C T
Article history:
The basement membranes (BMs) of the nervous system include (a) the pial BM that sur-
Received 25 May 2014
rounds the entire CNS, (b) the BMs that outline the vascular system of the CNS and PNS
Received in revised form
and (c) the BMs that are associated with Schwann cells. We previously found that isolated
30 June 2014
BMs are bi-functionally organized, whereby the two surfaces have different compositional,
Accepted 15 July 2014
biomechanical and cell adhesion properties. To find out whether the bi-functional nature of
Available online 22 July 2014
BMs has an instructive function in organizing the tissue architecture of the developing nervous system, segments of human BMs were inserted into (a) the parasomitic mesoderm of
Keywords: Basement membrane Laminin Collagen IV Extracellular matrix Extracellular matrix proteins
chick embryos, intersecting with the pathways of axons and neural crest cells, or (b) into the midline of the embryonic chick spinal cord. The implanted BMs integrated into the embryonic tissues within 24 h and were impenetrable to growing axons and migrating neural crests cells. Host axons and neural crest cells contacted the epithelial side but avoided the stromal side of the implanted BM. When the BMs were inserted into the spinal cord, neurons, glia cells and axons assembled at the epithelial side of the implanted BMs, while a connective tissue layer formed at the stromal side, resembling the tissue architecture of the spinal cord at the pial surface. Since the spinal cord is a-vascular at the time of BM implantation, we propose that the bi-functional nature of BMs has the function of segregating epithelial and connective cells into two adjacent compartments and participates in establishing the tissue architecture at the pial surface of the CNS. 2014 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
1.
Introduction
Basement membranes (BMs) are thin sheets of extracellular matrix (ECM) that are located at the basal side of every epithelium, that outline muscle fibers and that are present at the basal surface of the vascular endothelial cells (Yurchenco and Patton, 2009; Halfter et al., 2013a). BMs are comprised of multi-domain extracellular matrix (ECM) proteins that either polymerize or bind to other ECM proteins (Timpl and Brown, 1996; Erickson and Couchman, 2000).
Essential in the assembly of BMs are cellular receptors, such as integrin family members (Stephens et al., 1995; Fassler and Meyer, 1995) and dystroglycan (Henry and Campbell, 1998). BMs are evolutionary conserved in all metazoans, and mutations of BMs are either embryonic lethal or lead to muscular dystrophy, vasculature ruptures, skin blistering or CNS malformations (Gautam et al., 1996; Arikawa-Hirasawa et al., 1999; Smyth et al., 1999; Costell et al., 1999; Willem et al., 2001; Fukai et al., 2002; Halfter et al., 2002; Po¨schl et al., 2004; Bader et al., 2005; Gould et al., 2005; Lee and Gross, 2007).
* Corresponding author. Address: Department of Neurobiology, University of Pittsburgh, 1415W Biological Science Tower, Pittsburgh, PA 15261, USA. Tel.: +1 (412) 648 9424; fax: +1 (412) 648 1441. E-mail address: [email protected] (W. Halfter). http://dx.doi.org/10.1016/j.mod.2014.07.003 0925-4773/ 2014 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
2
MECHANISMS OF DEVELOPMENT
BMs are very thin, and direct experimental testing of isolated BMs has been complicated by the difficulty of handling the delicate and barely visible ECM sheets. The functions of BMs have, therefore, mainly been indirectly deduced from the phenotype analysis of humans and mice with mutations of BMs proteins. The current data indicate that the underlying cause for BM-related pathologies, such as muscular dystrophy, vascular instability, skin blistering diseases and CNS dysplasias, are due to (a) the rupture of BMs, (b) the failure to provide a stable border for epithelial/endothelial tissues or muscle fibers or (c) the separation of epithelia from the underlying connective tissue layers. The problem of direct testing for BM properties has been recently overcome by using adult human BMs for experimentation: adult human BMs are thicker than BMs from shortlived laboratory animals and can be isolated, handled, mounted on glass slides and tested for their composition, their biomechanical properties and their cell adhesion activities (Halfter et al., 2013b). These studies have shown that human as well as non-human BMs are bi-functionally organized with an epithelial side that is characterized by a high abundance of laminin, a high stiffness and a strong cell adhesion-promoting activity. The connective tissue side of BMs is softer, has a different protein composition and is anti-adhesive for epithelial cells and neurons. In the present study, we examined whether the bi-functional nature of BMs has organizing functions in embryogenesis by isolating human BMs and transplanting them into developing chick embryos.
1 3 3 ( 2 0 1 4 ) 1 –1 0
in PBS and inserted into embryonic day (E) 2.2/HH stage 14–15 chick embryos. The pre-staining of the BMs was necessary to render the otherwise transparent and very thin BMs visible and facilitate the positioning and insertion of the BMs. BMs were transplanted into the parasomitic mesoderm (n = 28) or into the midline of the neural tube (n = 32). The implants were inserted lateral to the neural tube, at the level of the segmental plate below the last formed somite (somite 27). As shown previously, neural crest cell outgrowth has not occurred at that level at that stage (Yip, 1987). For controls, segments of saran wrap (PGC Scientific #12901173; n = 5) or thin sheets of nitrocellulose (n = 15) were also transplanted. The nitrocellulose sheets were prepared by drying a drop of nitrocellulose dissolved the methanol (Lagenaur and Lemmon, 1987) onto a glass slide. The nitrocellulose sheets were lifted off the glass slides after submerging in PBS.
2.3.
Immunocytochemistry
Human cadaver eyes (n = 6) ranging from 60 to 72 years of age were obtained from CORE, the Center of Organ Recovery and Education. A list of the donor characteristics has been provided previously (Halfter et al., 2013b). The eyes were harvested within 4–24 h of death and delivered to the laboratory in less than two days. The use of the human eyes for this project was approved by the internal review board of the University of Pittsburgh under the IRB protocol number #0312072. The inner limiting membrane (ILM) was obtained by incubating segments of retinas in 2% Triton-X-100 in distilled water overnight, and transferring the ILMs into new detergent that also included 2% deoxycholate (Candiello et al., 2010). The Descemet’s membrane (DM) was obtained from corneas that were incubated in 2% Triton-X-100 for 3 h followed by micro-dissection of the DM from the corneas. Lens capsules were obtained by micro-dissection. Flat mounting of the BMs was achieved by suspending segments of DM or ILM in a droplet of PBS on poly-lysine-coated slides, draining the liquid and firmly immobilizing the BMs by centrifugation at 1200 rpm for 5 min (Halfter et al., 2013b).
Flat mounted BMs were incubated with primary antibody for 3 h to overnight, washed, and incubated with Cy3 or Cy2labeled secondary antibodies (Jackson ImmunoResearch, West Grove, PA) for 2 h. For staining of spinal cord cross sections, embryos were fixed in 4% paraformaldehyde, cryoprotected in 25% sucrose, embedded in OCT compound and sectioned at 25 lm. The sections were mounted on poly-lysine-coated Superfrost Plus slides (Fisher Scientific, Pittsburgh, PA). Images were taken with an Olympus FlowView confocal microscope. A polyclonal antibody to mouse laminin-111 and a monoclonal antibody to human laminin a5 (Sigma, St. Louis, MO; Invitrogen) were used to detect laminin in the human BMs. A mouse monoclonal antibody to the 7S domain of human collagen IV a3 (Mab J3-2; provided by Dr. Nirmal SundarRaj (SundarRaj and Wilson, 1982) was used to detect the human BM implants in the host chick embryos. This antibody also served as a marker for the stromal side of implanted BMs. An IgG monoclonal antibody to NCAM 180 (Mab 9H2; Halfter et al., 1997) and an IgM monoclonal to neurofilament (Mab EC/8; Developmental studies Hybridoma Bank) were used to trace axons in the chick embryos. Migrating neural crest cells were detected with a monoclonal antibody to the HNK-1 epitope (Mab 1C10; Developmental Studies Hybridoma Bank). Chick-specific monoclonal antibodies were used to detect host ECM proteins. These included antibodies to collagen II (3F1; Halfter et al., 2006), laminin (Mab 3H11; Halfter, 1993; Dev Studies Hyb Bank), perlecan (Mab 5C9; Balasubramani et al., 2004; Dev Studies Hyb Bank), fibronectin (Mab 5F3), tenascin-C (Mab M1; Dev Studies Hyb Bank) and CSPG (Mab 9BA12; Ring et al., 1995; Dev Studies Hyb Bank). A monoclonal antibody (Mab 1A3) was used to detect vascular endothelial cells. For TEM, the chick embryos were fixed in 2.5% glutaraldehyde and 0.05% tannic acid overnight. The fixed samples were osmicated and embedded in EPON according to standard protocols.
2.2.
2.4.
2.
Methods
2.1.
Basement membranes isolation and flat mounting
Implantation of human BMs into chick embryos
Segments of lens capsule (LC) and Descemet’s membrane (DM) from 60 to 72-year old human eyes were stained for 1 min in 0.5% a-naphthol blue (Sigma), washed multiple times
Cell adhesion and neurite outgrowth assays
Embryonic chick retinal cells were suspended at 200,000 cells/ml in DMEM/2% ovalbumin and plated onto the flat mounted BMs. After incubation for 15 min, the
MECHANISMS OF DEVELOPMENT
non-adherent cells were washed off with PBS and the preparations were fixed in 4% paraformaldehyde, 0.05% glutaraldehyde for 15 min. The adherent cells were detected by nuclear dye staining using Sytox Green or Sytox Orange (Molecular Probes/Invitrogen). Adherent cells were counted per unit area of 300 · 200 lm on the epithelial and (or) stromal side of the same BM with three counts per preparation and with an ‘‘n’’ of at least 3 biological repeats. The average was calculated with standard deviation, and Students T-tests confirmed the statistical significance. Retinal and dorsal root ganglion explants from E6 chick embryos were prepared as described (Halfter et al., 1983). Axon outgrowth was detected by staining the cultures with a monoclonal antibody to tubulin (mAb 6G7; Halfter et al., 2005a; Dev Studies Hyb Bank) after 24 h of incubation.
3.
Results
3.1.
Human BMs have side specific properties
Three types of basement membranes (BMs) were isolated from adult human donor eyes: the inner limiting membrane (ILM): a thin BM that separates the retina from the vitreous. The Descemet’s membrane (DM), a BM of intermediate thickness located between the corneal endothelium and the corneal stroma, and the lens capsule (LC), the BM that outlines the lens fibers. The LC is one of the thickest BM of the human body. All three BMs have an epithelial side that in vivo is apposed to the retinal neuroepithelium (ILM), the corneal endothelial cell layer (DM) or the lens epithelium (LC). The opposite, stromal side of these BMs borders the vitreous (ILM and LC) or the collagen-rich corneal stroma (DM). All of these BMs, when isolated, curl up, whereby their epithelial sides are consistently facing outward and the stromal sides are facing inward (Halfter et al., 2013b). The consistent rolling allows predicting the orientation of the isolated BMs for reliable side-specific flat mounting in vitro or side-specific implantation in vivo. For in vitro assays, the BMs can also be folded to expose both surfaces side-by-side. Immunocytochemistry of such folded and flat mounted BMs showed that laminin is prominently localized at the epithelial sides of all three BMs (Suppl. Fig. 1A, B). The vitreal/stromal surface of human BMs is defined by an abundance of the 7S domain of collagen IV a3 (Suppl. Fig. 1B). When dissociated chick retinal cells were plated on top of the flat-mounted BMs, the cells adhered to the retinal/epithelial side and not to the vitreal/stromal side (Suppl. Fig. 1C). When retinal explant stripes or dorsal root ganglia were placed across folded BMs, axons only grew out from the explants on the epithelial surface of the DM (Suppl. Fig. 1D). In the case of dorsal root ganglion explants, axon outgrowth on the epithelia/retinal surface of the ILM was greater as compared to outgrowth on the stromal/vitreal surface of the ILM, and the axons were longer and much less fasciculated (Suppl. Fig. 1E). The in-vitro experiments confirmed previous data showing that adult human BMs have a sidedness that is detectable by immunocytochemistry, by cell adhesion and by neurite outgrowth assays. In the following experiments, the
1 3 3 ( 2 0 1 4 ) 1 –1 0
3
side-specific properties of human BMs were tested in vivo by transplanting human BMs into chick embryos.
3.2. Human BMs maintain their side-specific properties after transplantation into chick embryos In a first set of experiments, segments of DM were implanted into the para-somitic connective tissue of E2.2/ HH stage 14 chick embryos parallel to the neural tube (Fig. 1A; n = 25). The axial position was at or below the level of the forelimb to intersect the pathway of motor axons and neural crest cells. For the transplantation, only DM and LC were usable for the experiments, since ILM samples were too thin and delicate for their reliable grafting as a single sheet. After 1–6 days of survival, the embryos were processed for cryostat sectioning and immunostaining. In the majority of cases (n = 14), the experiments were terminated 2 or 3 days after BM implantation. A monoclonal antibody specific to the 7S domain of human collagen IV a3 was used to locate the implanted BMs (Fig. 1B, C). The antibody also served as
Fig. 1 – Implantation of human DM into the mesenchymal tissue of a 2.2 day-old (HH stage 14) chick embryo. A top view of the embryo one day after BM implantation is shown in panel (A). The BM had been is implanted into the parasomitic mesenchyme (S: somites) lateral and parallel to the neural tube (NT). The BM had been labeled with alphanaphthol blue for better visualization (white star). The concave curvature shows that the epithelial side of the implanted BM is facing the neural tube (A). One day after transplantation, the implanted BMs (arrow) were localized in crosssections through the trunk by staining for the 7S domain of human collagen IV a3 (B; red; NC: notochord). The section was counter-stained with SytoxGreen, a generic nuclear dye. The section shows the tight integration of the human DM. A high power micrograph shows the prominent staining of the stromal side of the implanted DM with the 7S-specific antibody to collagen IV a3 (green; C). The section had also been stained for CD44, a label for macrophages (red). The micrograph shows that the implantation did not cause an inflammation-induced accumulation of macrophages around the human BM. Bars: B: 250 lm; C: 25 lm.
4
MECHANISMS OF DEVELOPMENT
a marker for the orientation of the implanted BMs, since the staining was most prominent at the stromal side of all human BMs (Fig. 1C). Experiments showed that the excessive scrolling of the LCs made it impossible to obtain un-curled and single-layered BM implants in the host tissue (not shown; n = 7). The implantation experiments were, therefore, continued with human DMs. Experiments showed that the DM sheets were firmly integrated into the connective tissue after less than 24 h (Fig. 1B; n = 5). Staining for macrophages using an antibody to CD44 showed that the implanted BMs did not cause an inflammatory response (Fig. 1C). To find out whether the implanted human BMs were modified in-vivo by the deposition of host ECM proteins, we employed a series of chick-specific anti-ECM antibodies. The sidedness of the implanted BMs was confirmed by co-staining for human collagen IV a3 (Fig. 2B, D, F, H, J, L). Immunocytochemistry showed that chick collagen II, tenascin-C, fibronectin and CSPG were present in the trunk mesenchyme of the embryos. Of these, collagen II and tenascin-C were detected as a defined, linear staining along the implanted BMs indicating a binding of these proteins onto the BMs. The collagen II and tenascin-C staining was detected exclusively at the stromal surface of the DMs (Fig. 2A–D). Chick laminin was present in all chick embryo BMs, but it was not detected along the implanted human BMs (Fig. 2E, F). Fibronectin and CSPG, though abundant in the chick embryo mesenchyme, were not detected as a defined layer along the implanted BMs (CSPG: Fig. 2G, H; fibronectin: Fig. 2I, J). Chick perlecan, which is present in all chick BMs bound preferentially to the epithelial side of the implanted human DM (Fig. 2K, L). The deposition of ECM proteins onto the implanted BMs was not dependent on their positioning in the host, as the BMs were implanted in random orientation, and host ECM proteins were detected on both sides of the grafted BMs (Fig. 2A, C, G, I, K). The selective deposition of ECM proteins to the epithelial or the stromal sides of the implanted human BMs indicated that the transplanted human BMs have side-selective binding activity for subset of host ECM proteins.
1 3 3 ( 2 0 1 4 ) 1 –1 0
3.3. Host axons and neural crest cells contact epithelial side but avoid stromal side of transplanted human BMs To determine whether the bi-functionality of BMs seen in cell adhesion and neurite outgrowth assays in vitro is also apparent in vivo, we transplanted human BMs into the parasomitic mesoderm and the neural tube of stage 14 chick embryos to examine the response of the host tissue to the implanted BM. Experimental embryos were sectioned in the transverse plane and stained with the human 7S domain of collagen IV a3 to detect the implanted DMs (Figs. 1–4). Monoclonal antibodies to chick NCAM 180 (Fig. 3A, C, D, F) or neurofilament (Figs. 4A, 5, Suppl. Fig. 2D, E) were used to detect host axons. For the detection of host neural crest cells, sections were stained with a monoclonal antibody to the HNK1 epitope (Fig. 3B, E; Suppl. Fig. 2F). Data showed that by the second (n = 7) and third day post-transplantation (n = 7), axons and neural crest cells had passed the BM implants on their way to more peripheral targets (Fig. 3A–F). There was not a single case where axons or neural crest cells permeated through the DMs, except in a few cases where the DM was physically disrupted. The axons and neural crest cells were found to have migrated around the implanted BM either ventrally (Fig. 3A, B, E) or dorsally (Fig. 3C, D, F). The human BMs appeared to be barriers to growing axons and migrating neural crest cells, consistent with earlier results (Martins-Green and Erickson, 1987). High power micrographs showed that the closeness of the encounter of axons and neural crest cells with the human BMs depended on the orientation of the BMs: when the epithelial side of the DM was facing the approaching axons/neural crest cells, there was direct contact of axons and neural crest cells with the BM surface (Fig. 3D, E). When the stromal side of the DM was facing the advancing axons and neural crest cells, the approaching axons and neural crest cell kept a distance of approximately 3 lm from the implanted BMs (Fig. 3F) suggesting a repulsive function of the stromal side of BMs to neural crest cells and axons.
Fig. 2 – Side-specific deposition of chick ECM proteins onto human DMs three days after implantation. Crosssections through the trunk were double-stained for the host/chick ECM proteins (red channel in A, C, E, G, I, K) and a monoclonal antibody specific to the 7S domain of human collagen IV a3 (green channel in B, D, F, H, J). A combined red and green staining is shown in (L). The strong linear staining along the grafted human BMs showed that chick collagen II (A, B), and chick tenascin-C (C, D) bound to the stromal side of the human DMs. No binding to the implanted human BMs was detected for chick laminin (E, F), CSPG (G, H) and fibronectin (I, J). Chick perlecan bound preferentially to the epithelial side of the implanted human DM (K, L). Bar: 50 lm.
MECHANISMS OF DEVELOPMENT
1 3 3 ( 2 0 1 4 ) 1 –1 0
5
Fig. 3 – Encounter of axons and neural crest cells with the implanted human BMs. To record the encounter of spinal cord axons and migrating neural crest cells with the implanted DM, sections were stained for NCAM 180 (A, C, D, F) or the HNK-1 epitope (B, E). The stromal side of the implanted DMs were stained with an antibody to the 7S domain of collagen IV a3 (green). Outgrowing axons and neural crest cells went around the implanted DM either in ventral direction (A, B, E) or in dorsal direction (C, D, F). When axons and neural crest cells encountered the human BMs they directly contacted the BMs when their epithelial side (Ep) was facing their advance (D, E). In case of an encounter with the stromal side (St) of the implanted BMs, axons and neural crest cells never contacted the BMs directly but stayed at a distance (F). Bars: A–C: 500 lm; D–F: 50 lm.
To further investigate the interaction of axons and neurons with BMs we implanted human DM into the midline of neural tubes of E2.2/HH stage 14 chick embryos. The embryos were processed for histology 2 to 3 days after grafting. A total number of 13 implanted embryos were processed for cryostat sectioning and immunostaining, whereas 14 embryos were EPON embedded and processed for TEM. Immunocytochemistry showed that the implanted DM led to a split and a duplication of the spinal cord; each of the duplicated spinal cords were outlined by a complete vascular layer, and each had a separate central canal (Fig. 4A, B). High-power imaging of the sections after staining for NCAM 180 showed that spinal cord axons and spinal cord neurons had directly contacted the epithelial side of the DM but kept a distance to the stromal surface (Fig. 4D). Staining of the sections for vascular endothelial cells showed that a continuous vascular layer had assembled at the stromal side of the BMs (Fig. 4E). To find out whether other thin barrier resulted in a similar duplication, sheets of saran wrap (n = 5) or nitrocellulose (n = 10) were implanted as well. Results showed that saran wrap and nitrocellulose sheet-implants were not integrated into the spinal cord tissue and regularly ended outside the spinal cord (Fig. 4C). In a second set of control experiments (n = 5), nitrocellulose films were coated with laminin and implanted into the midline of the neural tube. In difference to the un-coated nitrocellulose sheets, the laminin-coated sheets were firmly integrated into the midline of the spinal cord and remained in place. Two to three days after implantation, the laminincoated nitrocellulose implants did not lead to a split of the spinal cored, and blood vessels assembled symmetrically on both sides of the implant (Fig. 5A). In contrast, the implantation of the DM led to the assembly of axons, glia cells and
neurons to the epithelial side and an assembly of vasculature to the stromal side of the implanted BMs (Fig. 5B) resulting into a rather symmetrical split of the spinal cord (Fig. 5B). A total of 14 embryos with DM implants into the spinal cord were processed for high resolution TEM imaging. Semi-thin sections of plastic embedded spinal cords confirmed the splitting of the spinal cord into two, each containing a separate central canal (Fig. 6A). High power views confirmed that the stromal side of the implanted BMs was lined by a connective tissue layer that ran parallel to the entire length of the stromal DM surface (Fig. 6B). The sidedness of the implanted BMs was recognizable by their concave curvature (Fig. 6A, B) and subsequent TEM that showed a granular sub-structure of the BM matrix that at the stromal side of the DM (Fig. 6E) that allows a clear distinction from the smooth and uniform ECM structure of the epithelial side of the DM (Fig. 6D). TEM views of the grafted BMs showed cells and axons on the epithelial side (Fig. 6C, D) and a 2 lm thick connective tissue layer with extracellular space on the stromal side of the BMs (Fig. 6C, E). This cellular distribution was detected in all 14 DM-spinal cord implants. The cellular arrangement along the implanted BMs faithfully reproduces the cellar arrangement along the pial surface of the brain and spinal cord with neurons, glia cells and axons on the epithelial side of the pial BM and a vascularized connective tissue layer that is typical for the stromal side of the pial BM (Fig. 6F).
3.4. Bi-functional nature of basement membrane has denovo organizing functions in developing nervous system The segregation of neurons and glia on the epithelial side and blood vessels and connective tissues on the stromal side of the transplanted BMs suggest that BMs plays a role in
6
MECHANISMS OF DEVELOPMENT
1 3 3 ( 2 0 1 4 ) 1 –1 0
Fig. 4 – Implantation of human DMs into the neural tube of chick embryos. Human DM segments were implanted into the neural tube midline of stage 14 chick embryos, and the embryos were allowed to survive between 2 and 3 days. Crosssections through the spinal cord double-stained for endothelial cells (A, Endo, red) and neurofilament (A, NF, green), and for NCAM 180 alone (B, NCAM, red) showed that 3 days after BM implantation, the neural tube had duplicated at the level of the transplant (SC1 and SC2). Each spinal cord showed a meningeal layer (M) on the outside and a central canal on the inside. Grafting of thin nitrocellulose sheets were not integrated (C). High power views showed that axons directly contacted the epithelial side (Ep) on the implanted DMs, whereas axons stayed away from the stromal side (St) of the BMs (B, D). Staining of sections for endothelial cells showed a vascular layer at the stromal side of the implanted DM (E). The implanted DMs were visualized by staining for human collagen IV a3. The antibody preferentially labels the stromal side of the BM. The nitrocellulose had a green autofluorescence (C). The Bars: A–C: 500 lm; D, E: 50 lm.
Fig. 5 – Implantation of a laminin-coated nitrocellulose sheet (A) into the neural tube of chick embryos. Laminin-coated nitrocellulose sheets were implanted into the middle of the neural tubes at stage 14, and the embryos allowed surviving for three days. Crosssections through the spinal cord were double-stained for endothelial cells (Endo, red) to label blood vessels (BV) and neurofilament (NF, green) to label axons (Ax). With laminin-coated nitrocellulose sheets (arrows in A), the spinal cord remained undivided and had only one central canal (CC). Blood vessels were located on both sides of the graft (A). The nitrocellulose was visible due to its green autofluorescence. For comparison, a crosssection of a spinal cord three days after implantation of a human DM is shown in panel (B). With the DM implant, the neural tube had duplicated at the location of the BM transplant (arrows in B), and there were two independent central canals (CC). A thin, continuous layer of blood vessels, as indicated by the red-labeled endothelial cells, outlined the stromal side of the BM graft (white stars). The blood vessel layer along the stromal side of the BM graft was continuous with the meningeal blood vessel layer of the spinal cord. Several axonal bundles (Ax) were detected at the epithelial side of the grafted BM (arrows in B), as seen by the green labeled axons. Bar: 250 lm. organizing the tissue arrangements of the nervous system. In order to distinguish whether the implanted BMs led to the redirection of already existing axons, migrating neural crest
cells or vasculature or whether the implanted BMs participated in organizing the distribution of these tissue structures de novo, we examined the state of axon outgrowth, neural
MECHANISMS OF DEVELOPMENT
1 3 3 ( 2 0 1 4 ) 1 –1 0
7
Fig. 6 – High resolution imaging of human DM implanted into chick neural tubes. The cell and tissue distribution at the pial surface of embryonic chick spinal cord is shown for comparison in panel (F). Semithin crosssection of a spinal cord 3 days after DM implantation (A). Note the perfect integration of the implanted DM (arrow) and the duplication of the spinal cord with two central canals (stars). A higher power image of another implanted DM is shown in panel (B). Note the high density of cells on the epithelial side of the grafted DM (Ep) and the connective tissue-like cell layer on the stromal side of the graft (St). A low-power TEM of a grafted DM is shown in (C). The white bar outlines the connective tissue layer along the stromal side (St) of the DM. High-power micrographs show a high density of cells and processes, summarily referred to as neuropil, that closely abut the epithelial side of the implanted DM (Ep). An ECM-rich, fibrous and vascular layer, referred to as stroma, outlines on the stromal side of the DM (E). The stromal side of the DM (St, E) is recognizable by the granular appearance of the BM matrix (white arrow) as compared to the smooth ECM matrix on the epithelial side (D). Panel (F) shows the pial surface of a chick neural tube (NT) with cells and axons (neuropil) on the epithelial side of the pial BM (Ep) and connective tissue (CT; stroma) at the stromal/meningeal side of the pial BMs (BM). The tissue arrangement along the pial surface of the spinal cord is analogous to the tissue arrangement along the two surfaces of the implanted DM. Bars: A: 100 lm, B: 10 lm, C: 2 lm; D–F: 1 lm.
crest cell migration and blood vessels development at the time BM implantation and thereafter. Results showed that the spinal cord was a-vascular at the time of BM grafting (E2.2; HH stage 14; Suppl. Fig. 2A). Even one day after BM grafting, the neural tube remains avascular. At that stage, the dorsal aorta is the dominant vascular supply for the dorsal part of the embryo. However, first vascular sprouts from the dorsal aorta curved around the notochord towards the base of the spinal cord (Suppl. Fig. 2B, D). A complete vascularized layer around the entire spinal cord was present only 2 days after grafting (Suppl. Fig. 2C, E, F). Data also showed that that few axons and neural crest cells had exited the neural tube at the time of BM implantation (Suppl. Fig. 2A). Larger number of axons and neural crest cells grew out of the spinal cord at stage 20, approximately 1 day after the BM grafting. At that stage, axons and neural crest cells first encountered the BM implants (Suppl. Fig. 2D; n = 5). Two days after BM implantation, there was extensive axonal outgrowth from the spinal cord (Suppl. Fig. 2E) and a
large population of neural crest cells in the trunk of the embryos (Suppl. Fig. 2F). The current data showed that BM implantation led to a side-specific organization of tissues around the grafted BMs and indicate the BMs have an instructive function in the organization at the pial surface of the central nervous system.
4.
Discussion
4.1.
BMs have side-specific properties in vitro and in vivo
A previous report (Halfter et al., 2013b) showed that adult human BMs have a sidedness that is detectable in vitro by (a) an asymmetric protein/peptide epitope distribution, by (b) differences in the biomechanical quality of both BM sides and (c) by side-selective cell adhesion (Suppl. Fig. 1). In this project, we found that the asymmetrical organization of BMs can also be detected in vivo and that this BM asymmetry has a functional role in organizing tissues during
8
MECHANISMS OF DEVELOPMENT
development. Asymmetric distributions of proteins within BMs have been independently reported for human LC, DM and glomerular BMs (Danysh and Duncan, 2008; Kobosova et al., 2007, and biochemical studies recently confirmed that the epidermal BM has a layered structure, with an abundance of laminin on the epithelial side (Behrens et al., 2012). While an asymmetric distribution of proteins has not been detected for mouse or rat BMs, a functional asymmetry with respect to a selective cell adhesion applied to all BM so far tested, including BMs from neonatal and adult mice and chick embryos (Halfter et al., 2013b). In difference to adult human BMs, BMs from laboratory animals are much thinner, and the detection of a side-selective protein distribution is a much greater challenge. However, adult human BMs also start out as very thin ECM sheets and, at embryonic and neonatal stages, share the textbook BM morphology from mice and rats (Candiello et al., 2010; Murphy et al., 1984; Danysh and Duncan, 2008). The increase in BM thickness in humans is a natural process that gradually proceeds with advancing age. A major increase in BM thickness does not occur in mice and rats because of their much shorter life span. We propose that all BMs are asymmetrically organized from the outset, but that BM asymmetry is more easily detectable when the BMs become thicker and more easily manageable for experimental testing. The origin and structural basis for the BM asymmetry is currently unknown. An explanation could be the origin of a selected group of BM constituents from the epithelial tissues and others from the adjacent stromal tissues (Sarthy and Fu, 1990; Dong and Chung, 1991; Sarthy, 1993; Dong et al., 2002; Halfter et al., 2008). The asymmetric structure of the BMs may also be explained by the way BMs are assembled, namely that laminin assembles first followed by collagen IV later in development (Yurchenco and Patton, 2009). Best evidence for the latter hypothesis comes by a detailed developmental study of the human DM that showed a two-phased growth: a fast fetal growth, and a slow postnatal growth resulting in a two-layered BM with a striated and an amorphous layer (Murphy et al., 1984). We propose that the functional sidedness of BMs is universal for all metazoans, and we hypothesize that the bi-functional nature of BMs participates in organizing multicellular organisms in alternating layers of epithelial and connective tissues. By transplanting human BMs into developing chick embryos we found that the bi-functional organization of human BMs as detected in vitro can also be detected in vivo and that this asymmetry contributes to the establishment of the tissue architecture along the pial surface of the CNS. The asymmetry of BMs in vivo was readily detectable by the side-selective deposition of host ECM proteins onto the implanted human BMs: interstitial ECM proteins, such as collagen II and tenascin-C bound preferentially to the stromal side of the implanted DMs indicating that BM protein epitopes exposed at this side are capable of binding proteins that anchor the BMs to the adjacent interstitial connective tissue. The side-selective binding of protein to the implanted human BMs was restricted to a subset of host ECM proteins. The preference of cell adhesion to the opposite, epithelial side of BMs indicate that BMs are structured such that the laminin-rich epithelial side promotes cell adhesion, whereas the stromal side provides the binding sites that allow BMs
1 3 3 ( 2 0 1 4 ) 1 –1 0
to firmly connect the adjacent stromal ECM. The antiadhesive prosperities of the stromal side of BMs prevent the establishment of two directly adjacent epithelial cell layers.
4.2. BMs as borders for axon growth and neural crest cell migration The current data showed that neither growing axons not migrating neural crest cell were capable of penetrating into or pass through the implanted human BMs. In all cases, axons and neural crest cells detoured the BM implants on their way to more distal targets. Since the implanted human BMs did not cause an inflammatory response, we did not observe diapedesis of macrophages or neutrophil through the BMs, but we predict that migration of neutrophils and macrophages through BMs is possible as it occurs naturally during inflammation. During embryogenesis, neural crest migrate from the neural tube into the adjacent connective tissues, but at the stage of migration, the pial BM of the dorsal neural tube is at best very thin and fragile or may even be incomplete (Martins-Green and Erickson, 1987). Interestingly, neural crest cells and axons did not sense the implanted BMs as a stop signal, rather progressed with their distal migration by growing around the BM obstacle. The interaction of axons and neural crest cells with the implanted BMs differed depending on the sidedness of the BMs: in cases when the axons and neural crest cell approached the epithelial side, they got in direct contact with the BMs surface. In cases when axons and neural crest cell approached the stromal side of the BMs, they did not contact the BMs, rather stayed at a distance of few microns from the inserted BMs. This is consistent with in vitro data showing that axons outgrowth and cell adhesion are restricted to the epithelial side of BMs and confirm the functional sidedness of BMs in vivo. We propose that diapedesis of cells through BMs is easier and faster when cells approach from the epithelial side of the BMs. The transmigration would also be greatly facilitated for cells that don’t show an adhesion preference to only the epithelial side of BMs as it was found in cell adhesion experiments using invasive tumor cell lines (Halfter, unpublished).
4.3. An organizing function of BMs in the central nervous system When implanted into the developing neural tube, the BM led to a split into two spinal cords with two separate central canals and a dividing meningeal layer. It is of note that the neural tube has no vasculature and no axonal layers at the time of BM implantation. Thus, the arrangement of cells following the implantation was orchestrated by the de-novo interaction of cells with the inserted BMs. We found that in all cases axons, neural and glia cells were in direct contact with the epithelial side of the implanted BMs and assembled as a dense neuropil at this side. Axons, glia cells and neurons stayed away from the stromal side of the implanted BMs and there was always a several micrometer-wide connective tissue/vascular layer, indicating that the stromal surface environment of BMs is anti-adhesive for neurons, glia cells and axons but promotes the assembly of a connective tissue layer
MECHANISMS OF DEVELOPMENT
that is highly vascularized. Based on the current in vivo data and consistent with the in vitro assays, we propose that BMs have an organizing function in vivo that plays a major role in the assembly of the tissue architecture at the pial surface of the spinal cord and brain: the epithelial side of BMs promotes the adhesion of neurons, glia cells and axons, whereas the stromal side promotes the assembly of a highly vascularized connective tissue layer. The bi-functional nature of BMs may also play a role in other developmental processes that require the separation or connection of epithelial and connective tissues. We also propose that the bi-functional nature of the human BMs promoted the splitting of the neural tube into two rather symmetric spinal cords. In difference to laminincoated nitrocellulose implants (Fig. 5A), the human BMs promoted the formation of a continuous vascular layer though the midline of the spinal cord. The ectopic vascular layer through the spinal cord midline was continuous with the meningeal vascular layer that outlines the spinal cord. A caveat of these experiments is that the BMs used in these experiments were isolated from adult human eyes, thus are not the thin BMs that neurons, axons and neural crest cells encounter in embryos. However, the major goal of the current experiments was not to mimic the situation of axons and neural crest cell encountering BMs in vivo, rather to find out whether in vivo-derived BMs show side-specific properties in vivo and that these properties have the potential to organize tissues and cells in a life embryo.
Acknowledgements We would like to thank Drs. Nirmala SundarRaj, and Katharine Davoli for generously providing us with collagen IV antibodies. The current project was funded by an in-house grant from the University of Pittsburgh.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.mod.2014.07.003. R E F E R E N C E S
Arikawa-Hirasawa, E., Watanabe, H., Takami, H., Hassell, J.R., Yamada, Y., 1999. Perlecan is essential for cartilage and cephalic development. Nat. Genet. 23, 354–358. Bader, B.L., Smyth, N., Nedbal, S., Miosge, N., Baranowsky, A., et al, 2005. Compound genetic ablation of nidogen 1 and 2 causes basement membrane defects and perinatal lethality in mice. Mol. Cell. Biol. 25, 846–6856. Balasubramani, M., Bier, M.E., Hummel, S., Schneider, W.J., Halfter, W., 2004. Perlecan and its immunoglobulin like domain IV are abundant in vitreous and serum of the chick embryo. Matrix Biol. 23, 143–152. Behrens, D.T., Villone, D., Koch, M., Brunner, G., Sorokin, L., et al, 2012. The epidermal basement membrane is a composite of separate laminin- or collagen IV-containing networks connected by aggregated perlecan, but not by nidogens. J. Biol. Chem. 287, 18700–18709.
1 3 3 ( 2 0 1 4 ) 1 –1 0
9
Candiello, J., Cole, G.J., Halfter, W., 2010. Age-dependent changes in the structure, composition and biophysical properties of a human basement membrane. Matrix Biol. 29, 402–410. Costell, M., Gustafsson, E., Aszodi, A., Moergelin, M., Bloch, E., et al, 1999. Perlecan maintains the integrity of cartilage and some basement membranes. J. Cell Biol. 147, 1109–1122. Danysh, B.P., Duncan, M.K., 2008. The lens capsule. Exp. Eye Res. 88, 151–164. Dong, L.J., Chung, A.E., 1991. The expression of the genes for entactin, laminin A, laminin B1 and laminin B2 in murine lens morphogenesis and eye development. Differentiation 48, 157– 172. Dong, S., Landfair, J., Balasubramani, M., Bier, M.E., Cole, G., et al, 2002. Expression of basal lamina protein mRNA in the early embryonic chick eye. J. Comp. Neurol. 447, 261–273. Erickson, A.C., Couchman, J.R., 2000. Still more complexity in mammalian basement membranes. J. Histochem. Cytochem. 48, 1291–1306. Fassler, R., Meyer, M., 1995. Consequences of lack of b1 integrin gene expression in mice. Genes Dev. 9, 1896–1908. Fukai, N., Eklund, L., Marneros, A.G., Oh, S.P., Keene, D.R., et al, 2002. Lack of collagen XVIII/endostatin results in eye abnormalities. EMBO J. 21, 1535–1544. Gautam, M., Noakes, P.G., Moscovo, L., Rupp, F., Scheller, R.H., et al, 1996. Defective neuromuscular synaptogenesis in agrindeficient mutant mice. Cell 85, 525–535. Gould, D.B., Phalan, F.C., Breedveld, G.J., van Mil, S.E., Smith, R.S., et al, 2005. Mutations in Col4a1 cause perinatal cerebral hemorrhage and porencephaly. Science 308, 1167– 1171. Halfter, W., Newgreen, D.F., Sauter, J., Schwarz, U., 1983. Oriented axon outgrowth from avian embryonic retinae in culture. Dev. Biol. 95, 56–64. Halfter, W., 1993. A heparan sulfate proteoglycan in developing avian axonal tracts. J. Neurosci. 13, 2863–2873. Halfter, W., Schurer, B., Yip, J., Yip, Y.P., Lee, J.A., Cole, G.J., 1997. Distribution and substrate properties of agrin, a heparan sulfate proteoglycan of developing axonal pathways. J. Comp. Neurol. 383, 1–17. Halfter, W., Willem, M., Mayer, U., 2002. A critical function of the pial basement membrane in cortical histogenesis. J. Neurosci. 22, 6029–6040. Halfter, W., Willem, M., Mayer, U., 2005. Basement membranedependent survival of retinal ganglion cells. Invest. Ophthalmol. Vis. Sci. 46, 1000–1009. Halfter, W., Winzen, U., Bishop, P.N., Eller, A., 2006. Regulation of eye size by the retinal basement membrane and vitreous body. Invest. Ophthalmol. Vis. Sci. 47, 3586–3594. Halfter, W., Dong, S., Dong, A., Eller, A.W., Nischt, R., 2008. Origin and turnover of ECM proteins from the inner limiting membrane and vitreous body. Eye 22, 1207–1213. Halfter, W., Candiello, J., Hu, H., Zhang, P., Schreiber, E., Balasubramani, M., 2013a. Protein composition and biomechanical properties of in vivo-derived basement membranes. Cell Adh Migr 7, 64–71. Halfter, W., Monnier, C., Mu¨ller, D., Oertle, P., Uechi, G., Balasubramani, M., Safi, F., Lim, R., Loparic, M., Henrich, P.B., 2013b. The bi-functional organization of human basement membranes. PLoS One 8, e67660. Henry, M.D., Campbell, K.P., 1998. A role of dystroglycan in basement membrane assembly. Cell 95, 859–870. Kobosova, A., Azar, D.T., Bannikov, G.A., Campbell, K.P., Durbeej, M., et al, 2007. Compositional differences between infant and adult human corneal basement membranes. Invest. Ophthalmol. Vis. Sci. 48, 4989–4999. Lagenaur, C., Lemmon, V., 1987. An L1-like molecule, the 8D9 antigen, is a potent substrate for neurite extension. Proc. Natl. Acad. Sci. U.S.A. 84, 7753–7757.
10
MECHANISMS OF DEVELOPMENT
Lee, J., Gross, J.M., 2007. Laminin beta1 and gamma1 containing laminins are essential for basement membrane integrity in zebra fish eye. Invest. Ophthalmol. Vis. Sci. 48, 2483–2490. Martins-Green, M., Erickson, C.A., 1987. Basal lamina is not a barrier to neural crest cell emigration: documentation by TEM and by immunofluorescent and immunogold labelling. Development 10, 517–533. Murphy, C., Alvarado, J., Juster, R., 1984. Prenatal and postnatal growth of the human Descemet’s membrane. Invest. Ophthalmol. Vis. Sci. 25, 1402–1415. Po¨schl, E., Schlotzer-Schrehardt, U., Brachvogel, B., Saito, K., Ninomiya, Y., et al, 2004. Collagen IV is essential for basement membrane stability but dispensable for initiation of its assembly during early development. Development 131, 1619– 1628. Ring, C., Lemmon, V., Halfter, W., 1995. Two chondroitin sulfate proteoglycans differentially expressed in the developing chick visual system. Dev. Biol. 168, 11–27. Sarthy, P.V., Fu, M., 1990. Localization of laminin B1 mRNA in retinal ganglion cells by in situ hybridization. J. Cell Biol. 110, 2099–2108. Sarthy, V., 1993. Collagen IV mRNA expression during development of the mouse retina: an in situ hybridization study. Invest. Ophthalmol. Vis. Sci. 34, 145–152.
1 3 3 ( 2 0 1 4 ) 1 –1 0
Smyth, N., Vatansever, H.S., Murray, P., Meyer, M., Frie, C., et al, 1999. Absence of basement membranes after targeting the LANC1 gene results in embryonic lethality due to failure of endoderm differentiation. J. Cell Biol. 144, 151–160. Stephens, L.E., Sutherland, A.E., Klimanskaya, I.V., Andrieux, A., Meneses, J., et al, 1995. Deletion of beta 1 integrins in mice results in inner cell mass failure and peri-implantation lethality. Genes Dev. 9, 1896–1908. SundarRaj, N., Wilson, J., 1982. Monoclonal antibody to human basement membrane collagen IV. Immunology 47, 133–140. Timpl, R., Brown, J.C., 1996. Supramolecular assembly of basement membranes. Bioassays 18, 123–132. Willem, M., Miosge, N., Halfter, W., Smyth, N., Janetti, I., et al, 2001. Specific ablation of the nidogen binding site in the laminin c1 chain interferes with kidney and lung development. Development 129, 2711–2722. Yip, J.W., 1987. Target cues are not required for the guidance of sympathetic preganglionic axons. Brain Res. 429, 155–159. Yurchenco, P.D., Patton, B.L., 2009. Developmental and pathogenic mechanisms of basement membrane assembly. Curr. Pharm. Des. 15, 1277–1294.
1 3 3 ( 2 0 1 4 ) 1 –1 0
Available at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/modo
An organizing function of basement membranes in the developing nervous system Willi Halfter *, Joseph Yip Department of Neurobiology, University of Pittsburgh, Pittsburgh, USA
A R T I C L E I N F O
A B S T R A C T
Article history:
The basement membranes (BMs) of the nervous system include (a) the pial BM that sur-
Received 25 May 2014
rounds the entire CNS, (b) the BMs that outline the vascular system of the CNS and PNS
Received in revised form
and (c) the BMs that are associated with Schwann cells. We previously found that isolated
30 June 2014
BMs are bi-functionally organized, whereby the two surfaces have different compositional,
Accepted 15 July 2014
biomechanical and cell adhesion properties. To find out whether the bi-functional nature of
Available online 22 July 2014
BMs has an instructive function in organizing the tissue architecture of the developing nervous system, segments of human BMs were inserted into (a) the parasomitic mesoderm of
Keywords: Basement membrane Laminin Collagen IV Extracellular matrix Extracellular matrix proteins
chick embryos, intersecting with the pathways of axons and neural crest cells, or (b) into the midline of the embryonic chick spinal cord. The implanted BMs integrated into the embryonic tissues within 24 h and were impenetrable to growing axons and migrating neural crests cells. Host axons and neural crest cells contacted the epithelial side but avoided the stromal side of the implanted BM. When the BMs were inserted into the spinal cord, neurons, glia cells and axons assembled at the epithelial side of the implanted BMs, while a connective tissue layer formed at the stromal side, resembling the tissue architecture of the spinal cord at the pial surface. Since the spinal cord is a-vascular at the time of BM implantation, we propose that the bi-functional nature of BMs has the function of segregating epithelial and connective cells into two adjacent compartments and participates in establishing the tissue architecture at the pial surface of the CNS. 2014 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
1.
Introduction
Basement membranes (BMs) are thin sheets of extracellular matrix (ECM) that are located at the basal side of every epithelium, that outline muscle fibers and that are present at the basal surface of the vascular endothelial cells (Yurchenco and Patton, 2009; Halfter et al., 2013a). BMs are comprised of multi-domain extracellular matrix (ECM) proteins that either polymerize or bind to other ECM proteins (Timpl and Brown, 1996; Erickson and Couchman, 2000).
Essential in the assembly of BMs are cellular receptors, such as integrin family members (Stephens et al., 1995; Fassler and Meyer, 1995) and dystroglycan (Henry and Campbell, 1998). BMs are evolutionary conserved in all metazoans, and mutations of BMs are either embryonic lethal or lead to muscular dystrophy, vasculature ruptures, skin blistering or CNS malformations (Gautam et al., 1996; Arikawa-Hirasawa et al., 1999; Smyth et al., 1999; Costell et al., 1999; Willem et al., 2001; Fukai et al., 2002; Halfter et al., 2002; Po¨schl et al., 2004; Bader et al., 2005; Gould et al., 2005; Lee and Gross, 2007).
* Corresponding author. Address: Department of Neurobiology, University of Pittsburgh, 1415W Biological Science Tower, Pittsburgh, PA 15261, USA. Tel.: +1 (412) 648 9424; fax: +1 (412) 648 1441. E-mail address: [email protected] (W. Halfter). http://dx.doi.org/10.1016/j.mod.2014.07.003 0925-4773/ 2014 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
2
MECHANISMS OF DEVELOPMENT
BMs are very thin, and direct experimental testing of isolated BMs has been complicated by the difficulty of handling the delicate and barely visible ECM sheets. The functions of BMs have, therefore, mainly been indirectly deduced from the phenotype analysis of humans and mice with mutations of BMs proteins. The current data indicate that the underlying cause for BM-related pathologies, such as muscular dystrophy, vascular instability, skin blistering diseases and CNS dysplasias, are due to (a) the rupture of BMs, (b) the failure to provide a stable border for epithelial/endothelial tissues or muscle fibers or (c) the separation of epithelia from the underlying connective tissue layers. The problem of direct testing for BM properties has been recently overcome by using adult human BMs for experimentation: adult human BMs are thicker than BMs from shortlived laboratory animals and can be isolated, handled, mounted on glass slides and tested for their composition, their biomechanical properties and their cell adhesion activities (Halfter et al., 2013b). These studies have shown that human as well as non-human BMs are bi-functionally organized with an epithelial side that is characterized by a high abundance of laminin, a high stiffness and a strong cell adhesion-promoting activity. The connective tissue side of BMs is softer, has a different protein composition and is anti-adhesive for epithelial cells and neurons. In the present study, we examined whether the bi-functional nature of BMs has organizing functions in embryogenesis by isolating human BMs and transplanting them into developing chick embryos.
1 3 3 ( 2 0 1 4 ) 1 –1 0
in PBS and inserted into embryonic day (E) 2.2/HH stage 14–15 chick embryos. The pre-staining of the BMs was necessary to render the otherwise transparent and very thin BMs visible and facilitate the positioning and insertion of the BMs. BMs were transplanted into the parasomitic mesoderm (n = 28) or into the midline of the neural tube (n = 32). The implants were inserted lateral to the neural tube, at the level of the segmental plate below the last formed somite (somite 27). As shown previously, neural crest cell outgrowth has not occurred at that level at that stage (Yip, 1987). For controls, segments of saran wrap (PGC Scientific #12901173; n = 5) or thin sheets of nitrocellulose (n = 15) were also transplanted. The nitrocellulose sheets were prepared by drying a drop of nitrocellulose dissolved the methanol (Lagenaur and Lemmon, 1987) onto a glass slide. The nitrocellulose sheets were lifted off the glass slides after submerging in PBS.
2.3.
Immunocytochemistry
Human cadaver eyes (n = 6) ranging from 60 to 72 years of age were obtained from CORE, the Center of Organ Recovery and Education. A list of the donor characteristics has been provided previously (Halfter et al., 2013b). The eyes were harvested within 4–24 h of death and delivered to the laboratory in less than two days. The use of the human eyes for this project was approved by the internal review board of the University of Pittsburgh under the IRB protocol number #0312072. The inner limiting membrane (ILM) was obtained by incubating segments of retinas in 2% Triton-X-100 in distilled water overnight, and transferring the ILMs into new detergent that also included 2% deoxycholate (Candiello et al., 2010). The Descemet’s membrane (DM) was obtained from corneas that were incubated in 2% Triton-X-100 for 3 h followed by micro-dissection of the DM from the corneas. Lens capsules were obtained by micro-dissection. Flat mounting of the BMs was achieved by suspending segments of DM or ILM in a droplet of PBS on poly-lysine-coated slides, draining the liquid and firmly immobilizing the BMs by centrifugation at 1200 rpm for 5 min (Halfter et al., 2013b).
Flat mounted BMs were incubated with primary antibody for 3 h to overnight, washed, and incubated with Cy3 or Cy2labeled secondary antibodies (Jackson ImmunoResearch, West Grove, PA) for 2 h. For staining of spinal cord cross sections, embryos were fixed in 4% paraformaldehyde, cryoprotected in 25% sucrose, embedded in OCT compound and sectioned at 25 lm. The sections were mounted on poly-lysine-coated Superfrost Plus slides (Fisher Scientific, Pittsburgh, PA). Images were taken with an Olympus FlowView confocal microscope. A polyclonal antibody to mouse laminin-111 and a monoclonal antibody to human laminin a5 (Sigma, St. Louis, MO; Invitrogen) were used to detect laminin in the human BMs. A mouse monoclonal antibody to the 7S domain of human collagen IV a3 (Mab J3-2; provided by Dr. Nirmal SundarRaj (SundarRaj and Wilson, 1982) was used to detect the human BM implants in the host chick embryos. This antibody also served as a marker for the stromal side of implanted BMs. An IgG monoclonal antibody to NCAM 180 (Mab 9H2; Halfter et al., 1997) and an IgM monoclonal to neurofilament (Mab EC/8; Developmental studies Hybridoma Bank) were used to trace axons in the chick embryos. Migrating neural crest cells were detected with a monoclonal antibody to the HNK-1 epitope (Mab 1C10; Developmental Studies Hybridoma Bank). Chick-specific monoclonal antibodies were used to detect host ECM proteins. These included antibodies to collagen II (3F1; Halfter et al., 2006), laminin (Mab 3H11; Halfter, 1993; Dev Studies Hyb Bank), perlecan (Mab 5C9; Balasubramani et al., 2004; Dev Studies Hyb Bank), fibronectin (Mab 5F3), tenascin-C (Mab M1; Dev Studies Hyb Bank) and CSPG (Mab 9BA12; Ring et al., 1995; Dev Studies Hyb Bank). A monoclonal antibody (Mab 1A3) was used to detect vascular endothelial cells. For TEM, the chick embryos were fixed in 2.5% glutaraldehyde and 0.05% tannic acid overnight. The fixed samples were osmicated and embedded in EPON according to standard protocols.
2.2.
2.4.
2.
Methods
2.1.
Basement membranes isolation and flat mounting
Implantation of human BMs into chick embryos
Segments of lens capsule (LC) and Descemet’s membrane (DM) from 60 to 72-year old human eyes were stained for 1 min in 0.5% a-naphthol blue (Sigma), washed multiple times
Cell adhesion and neurite outgrowth assays
Embryonic chick retinal cells were suspended at 200,000 cells/ml in DMEM/2% ovalbumin and plated onto the flat mounted BMs. After incubation for 15 min, the
MECHANISMS OF DEVELOPMENT
non-adherent cells were washed off with PBS and the preparations were fixed in 4% paraformaldehyde, 0.05% glutaraldehyde for 15 min. The adherent cells were detected by nuclear dye staining using Sytox Green or Sytox Orange (Molecular Probes/Invitrogen). Adherent cells were counted per unit area of 300 · 200 lm on the epithelial and (or) stromal side of the same BM with three counts per preparation and with an ‘‘n’’ of at least 3 biological repeats. The average was calculated with standard deviation, and Students T-tests confirmed the statistical significance. Retinal and dorsal root ganglion explants from E6 chick embryos were prepared as described (Halfter et al., 1983). Axon outgrowth was detected by staining the cultures with a monoclonal antibody to tubulin (mAb 6G7; Halfter et al., 2005a; Dev Studies Hyb Bank) after 24 h of incubation.
3.
Results
3.1.
Human BMs have side specific properties
Three types of basement membranes (BMs) were isolated from adult human donor eyes: the inner limiting membrane (ILM): a thin BM that separates the retina from the vitreous. The Descemet’s membrane (DM), a BM of intermediate thickness located between the corneal endothelium and the corneal stroma, and the lens capsule (LC), the BM that outlines the lens fibers. The LC is one of the thickest BM of the human body. All three BMs have an epithelial side that in vivo is apposed to the retinal neuroepithelium (ILM), the corneal endothelial cell layer (DM) or the lens epithelium (LC). The opposite, stromal side of these BMs borders the vitreous (ILM and LC) or the collagen-rich corneal stroma (DM). All of these BMs, when isolated, curl up, whereby their epithelial sides are consistently facing outward and the stromal sides are facing inward (Halfter et al., 2013b). The consistent rolling allows predicting the orientation of the isolated BMs for reliable side-specific flat mounting in vitro or side-specific implantation in vivo. For in vitro assays, the BMs can also be folded to expose both surfaces side-by-side. Immunocytochemistry of such folded and flat mounted BMs showed that laminin is prominently localized at the epithelial sides of all three BMs (Suppl. Fig. 1A, B). The vitreal/stromal surface of human BMs is defined by an abundance of the 7S domain of collagen IV a3 (Suppl. Fig. 1B). When dissociated chick retinal cells were plated on top of the flat-mounted BMs, the cells adhered to the retinal/epithelial side and not to the vitreal/stromal side (Suppl. Fig. 1C). When retinal explant stripes or dorsal root ganglia were placed across folded BMs, axons only grew out from the explants on the epithelial surface of the DM (Suppl. Fig. 1D). In the case of dorsal root ganglion explants, axon outgrowth on the epithelia/retinal surface of the ILM was greater as compared to outgrowth on the stromal/vitreal surface of the ILM, and the axons were longer and much less fasciculated (Suppl. Fig. 1E). The in-vitro experiments confirmed previous data showing that adult human BMs have a sidedness that is detectable by immunocytochemistry, by cell adhesion and by neurite outgrowth assays. In the following experiments, the
1 3 3 ( 2 0 1 4 ) 1 –1 0
3
side-specific properties of human BMs were tested in vivo by transplanting human BMs into chick embryos.
3.2. Human BMs maintain their side-specific properties after transplantation into chick embryos In a first set of experiments, segments of DM were implanted into the para-somitic connective tissue of E2.2/ HH stage 14 chick embryos parallel to the neural tube (Fig. 1A; n = 25). The axial position was at or below the level of the forelimb to intersect the pathway of motor axons and neural crest cells. For the transplantation, only DM and LC were usable for the experiments, since ILM samples were too thin and delicate for their reliable grafting as a single sheet. After 1–6 days of survival, the embryos were processed for cryostat sectioning and immunostaining. In the majority of cases (n = 14), the experiments were terminated 2 or 3 days after BM implantation. A monoclonal antibody specific to the 7S domain of human collagen IV a3 was used to locate the implanted BMs (Fig. 1B, C). The antibody also served as
Fig. 1 – Implantation of human DM into the mesenchymal tissue of a 2.2 day-old (HH stage 14) chick embryo. A top view of the embryo one day after BM implantation is shown in panel (A). The BM had been is implanted into the parasomitic mesenchyme (S: somites) lateral and parallel to the neural tube (NT). The BM had been labeled with alphanaphthol blue for better visualization (white star). The concave curvature shows that the epithelial side of the implanted BM is facing the neural tube (A). One day after transplantation, the implanted BMs (arrow) were localized in crosssections through the trunk by staining for the 7S domain of human collagen IV a3 (B; red; NC: notochord). The section was counter-stained with SytoxGreen, a generic nuclear dye. The section shows the tight integration of the human DM. A high power micrograph shows the prominent staining of the stromal side of the implanted DM with the 7S-specific antibody to collagen IV a3 (green; C). The section had also been stained for CD44, a label for macrophages (red). The micrograph shows that the implantation did not cause an inflammation-induced accumulation of macrophages around the human BM. Bars: B: 250 lm; C: 25 lm.
4
MECHANISMS OF DEVELOPMENT
a marker for the orientation of the implanted BMs, since the staining was most prominent at the stromal side of all human BMs (Fig. 1C). Experiments showed that the excessive scrolling of the LCs made it impossible to obtain un-curled and single-layered BM implants in the host tissue (not shown; n = 7). The implantation experiments were, therefore, continued with human DMs. Experiments showed that the DM sheets were firmly integrated into the connective tissue after less than 24 h (Fig. 1B; n = 5). Staining for macrophages using an antibody to CD44 showed that the implanted BMs did not cause an inflammatory response (Fig. 1C). To find out whether the implanted human BMs were modified in-vivo by the deposition of host ECM proteins, we employed a series of chick-specific anti-ECM antibodies. The sidedness of the implanted BMs was confirmed by co-staining for human collagen IV a3 (Fig. 2B, D, F, H, J, L). Immunocytochemistry showed that chick collagen II, tenascin-C, fibronectin and CSPG were present in the trunk mesenchyme of the embryos. Of these, collagen II and tenascin-C were detected as a defined, linear staining along the implanted BMs indicating a binding of these proteins onto the BMs. The collagen II and tenascin-C staining was detected exclusively at the stromal surface of the DMs (Fig. 2A–D). Chick laminin was present in all chick embryo BMs, but it was not detected along the implanted human BMs (Fig. 2E, F). Fibronectin and CSPG, though abundant in the chick embryo mesenchyme, were not detected as a defined layer along the implanted BMs (CSPG: Fig. 2G, H; fibronectin: Fig. 2I, J). Chick perlecan, which is present in all chick BMs bound preferentially to the epithelial side of the implanted human DM (Fig. 2K, L). The deposition of ECM proteins onto the implanted BMs was not dependent on their positioning in the host, as the BMs were implanted in random orientation, and host ECM proteins were detected on both sides of the grafted BMs (Fig. 2A, C, G, I, K). The selective deposition of ECM proteins to the epithelial or the stromal sides of the implanted human BMs indicated that the transplanted human BMs have side-selective binding activity for subset of host ECM proteins.
1 3 3 ( 2 0 1 4 ) 1 –1 0
3.3. Host axons and neural crest cells contact epithelial side but avoid stromal side of transplanted human BMs To determine whether the bi-functionality of BMs seen in cell adhesion and neurite outgrowth assays in vitro is also apparent in vivo, we transplanted human BMs into the parasomitic mesoderm and the neural tube of stage 14 chick embryos to examine the response of the host tissue to the implanted BM. Experimental embryos were sectioned in the transverse plane and stained with the human 7S domain of collagen IV a3 to detect the implanted DMs (Figs. 1–4). Monoclonal antibodies to chick NCAM 180 (Fig. 3A, C, D, F) or neurofilament (Figs. 4A, 5, Suppl. Fig. 2D, E) were used to detect host axons. For the detection of host neural crest cells, sections were stained with a monoclonal antibody to the HNK1 epitope (Fig. 3B, E; Suppl. Fig. 2F). Data showed that by the second (n = 7) and third day post-transplantation (n = 7), axons and neural crest cells had passed the BM implants on their way to more peripheral targets (Fig. 3A–F). There was not a single case where axons or neural crest cells permeated through the DMs, except in a few cases where the DM was physically disrupted. The axons and neural crest cells were found to have migrated around the implanted BM either ventrally (Fig. 3A, B, E) or dorsally (Fig. 3C, D, F). The human BMs appeared to be barriers to growing axons and migrating neural crest cells, consistent with earlier results (Martins-Green and Erickson, 1987). High power micrographs showed that the closeness of the encounter of axons and neural crest cells with the human BMs depended on the orientation of the BMs: when the epithelial side of the DM was facing the approaching axons/neural crest cells, there was direct contact of axons and neural crest cells with the BM surface (Fig. 3D, E). When the stromal side of the DM was facing the advancing axons and neural crest cells, the approaching axons and neural crest cell kept a distance of approximately 3 lm from the implanted BMs (Fig. 3F) suggesting a repulsive function of the stromal side of BMs to neural crest cells and axons.
Fig. 2 – Side-specific deposition of chick ECM proteins onto human DMs three days after implantation. Crosssections through the trunk were double-stained for the host/chick ECM proteins (red channel in A, C, E, G, I, K) and a monoclonal antibody specific to the 7S domain of human collagen IV a3 (green channel in B, D, F, H, J). A combined red and green staining is shown in (L). The strong linear staining along the grafted human BMs showed that chick collagen II (A, B), and chick tenascin-C (C, D) bound to the stromal side of the human DMs. No binding to the implanted human BMs was detected for chick laminin (E, F), CSPG (G, H) and fibronectin (I, J). Chick perlecan bound preferentially to the epithelial side of the implanted human DM (K, L). Bar: 50 lm.
MECHANISMS OF DEVELOPMENT
1 3 3 ( 2 0 1 4 ) 1 –1 0
5
Fig. 3 – Encounter of axons and neural crest cells with the implanted human BMs. To record the encounter of spinal cord axons and migrating neural crest cells with the implanted DM, sections were stained for NCAM 180 (A, C, D, F) or the HNK-1 epitope (B, E). The stromal side of the implanted DMs were stained with an antibody to the 7S domain of collagen IV a3 (green). Outgrowing axons and neural crest cells went around the implanted DM either in ventral direction (A, B, E) or in dorsal direction (C, D, F). When axons and neural crest cells encountered the human BMs they directly contacted the BMs when their epithelial side (Ep) was facing their advance (D, E). In case of an encounter with the stromal side (St) of the implanted BMs, axons and neural crest cells never contacted the BMs directly but stayed at a distance (F). Bars: A–C: 500 lm; D–F: 50 lm.
To further investigate the interaction of axons and neurons with BMs we implanted human DM into the midline of neural tubes of E2.2/HH stage 14 chick embryos. The embryos were processed for histology 2 to 3 days after grafting. A total number of 13 implanted embryos were processed for cryostat sectioning and immunostaining, whereas 14 embryos were EPON embedded and processed for TEM. Immunocytochemistry showed that the implanted DM led to a split and a duplication of the spinal cord; each of the duplicated spinal cords were outlined by a complete vascular layer, and each had a separate central canal (Fig. 4A, B). High-power imaging of the sections after staining for NCAM 180 showed that spinal cord axons and spinal cord neurons had directly contacted the epithelial side of the DM but kept a distance to the stromal surface (Fig. 4D). Staining of the sections for vascular endothelial cells showed that a continuous vascular layer had assembled at the stromal side of the BMs (Fig. 4E). To find out whether other thin barrier resulted in a similar duplication, sheets of saran wrap (n = 5) or nitrocellulose (n = 10) were implanted as well. Results showed that saran wrap and nitrocellulose sheet-implants were not integrated into the spinal cord tissue and regularly ended outside the spinal cord (Fig. 4C). In a second set of control experiments (n = 5), nitrocellulose films were coated with laminin and implanted into the midline of the neural tube. In difference to the un-coated nitrocellulose sheets, the laminin-coated sheets were firmly integrated into the midline of the spinal cord and remained in place. Two to three days after implantation, the laminincoated nitrocellulose implants did not lead to a split of the spinal cored, and blood vessels assembled symmetrically on both sides of the implant (Fig. 5A). In contrast, the implantation of the DM led to the assembly of axons, glia cells and
neurons to the epithelial side and an assembly of vasculature to the stromal side of the implanted BMs (Fig. 5B) resulting into a rather symmetrical split of the spinal cord (Fig. 5B). A total of 14 embryos with DM implants into the spinal cord were processed for high resolution TEM imaging. Semi-thin sections of plastic embedded spinal cords confirmed the splitting of the spinal cord into two, each containing a separate central canal (Fig. 6A). High power views confirmed that the stromal side of the implanted BMs was lined by a connective tissue layer that ran parallel to the entire length of the stromal DM surface (Fig. 6B). The sidedness of the implanted BMs was recognizable by their concave curvature (Fig. 6A, B) and subsequent TEM that showed a granular sub-structure of the BM matrix that at the stromal side of the DM (Fig. 6E) that allows a clear distinction from the smooth and uniform ECM structure of the epithelial side of the DM (Fig. 6D). TEM views of the grafted BMs showed cells and axons on the epithelial side (Fig. 6C, D) and a 2 lm thick connective tissue layer with extracellular space on the stromal side of the BMs (Fig. 6C, E). This cellular distribution was detected in all 14 DM-spinal cord implants. The cellular arrangement along the implanted BMs faithfully reproduces the cellar arrangement along the pial surface of the brain and spinal cord with neurons, glia cells and axons on the epithelial side of the pial BM and a vascularized connective tissue layer that is typical for the stromal side of the pial BM (Fig. 6F).
3.4. Bi-functional nature of basement membrane has denovo organizing functions in developing nervous system The segregation of neurons and glia on the epithelial side and blood vessels and connective tissues on the stromal side of the transplanted BMs suggest that BMs plays a role in
6
MECHANISMS OF DEVELOPMENT
1 3 3 ( 2 0 1 4 ) 1 –1 0
Fig. 4 – Implantation of human DMs into the neural tube of chick embryos. Human DM segments were implanted into the neural tube midline of stage 14 chick embryos, and the embryos were allowed to survive between 2 and 3 days. Crosssections through the spinal cord double-stained for endothelial cells (A, Endo, red) and neurofilament (A, NF, green), and for NCAM 180 alone (B, NCAM, red) showed that 3 days after BM implantation, the neural tube had duplicated at the level of the transplant (SC1 and SC2). Each spinal cord showed a meningeal layer (M) on the outside and a central canal on the inside. Grafting of thin nitrocellulose sheets were not integrated (C). High power views showed that axons directly contacted the epithelial side (Ep) on the implanted DMs, whereas axons stayed away from the stromal side (St) of the BMs (B, D). Staining of sections for endothelial cells showed a vascular layer at the stromal side of the implanted DM (E). The implanted DMs were visualized by staining for human collagen IV a3. The antibody preferentially labels the stromal side of the BM. The nitrocellulose had a green autofluorescence (C). The Bars: A–C: 500 lm; D, E: 50 lm.
Fig. 5 – Implantation of a laminin-coated nitrocellulose sheet (A) into the neural tube of chick embryos. Laminin-coated nitrocellulose sheets were implanted into the middle of the neural tubes at stage 14, and the embryos allowed surviving for three days. Crosssections through the spinal cord were double-stained for endothelial cells (Endo, red) to label blood vessels (BV) and neurofilament (NF, green) to label axons (Ax). With laminin-coated nitrocellulose sheets (arrows in A), the spinal cord remained undivided and had only one central canal (CC). Blood vessels were located on both sides of the graft (A). The nitrocellulose was visible due to its green autofluorescence. For comparison, a crosssection of a spinal cord three days after implantation of a human DM is shown in panel (B). With the DM implant, the neural tube had duplicated at the location of the BM transplant (arrows in B), and there were two independent central canals (CC). A thin, continuous layer of blood vessels, as indicated by the red-labeled endothelial cells, outlined the stromal side of the BM graft (white stars). The blood vessel layer along the stromal side of the BM graft was continuous with the meningeal blood vessel layer of the spinal cord. Several axonal bundles (Ax) were detected at the epithelial side of the grafted BM (arrows in B), as seen by the green labeled axons. Bar: 250 lm. organizing the tissue arrangements of the nervous system. In order to distinguish whether the implanted BMs led to the redirection of already existing axons, migrating neural crest
cells or vasculature or whether the implanted BMs participated in organizing the distribution of these tissue structures de novo, we examined the state of axon outgrowth, neural
MECHANISMS OF DEVELOPMENT
1 3 3 ( 2 0 1 4 ) 1 –1 0
7
Fig. 6 – High resolution imaging of human DM implanted into chick neural tubes. The cell and tissue distribution at the pial surface of embryonic chick spinal cord is shown for comparison in panel (F). Semithin crosssection of a spinal cord 3 days after DM implantation (A). Note the perfect integration of the implanted DM (arrow) and the duplication of the spinal cord with two central canals (stars). A higher power image of another implanted DM is shown in panel (B). Note the high density of cells on the epithelial side of the grafted DM (Ep) and the connective tissue-like cell layer on the stromal side of the graft (St). A low-power TEM of a grafted DM is shown in (C). The white bar outlines the connective tissue layer along the stromal side (St) of the DM. High-power micrographs show a high density of cells and processes, summarily referred to as neuropil, that closely abut the epithelial side of the implanted DM (Ep). An ECM-rich, fibrous and vascular layer, referred to as stroma, outlines on the stromal side of the DM (E). The stromal side of the DM (St, E) is recognizable by the granular appearance of the BM matrix (white arrow) as compared to the smooth ECM matrix on the epithelial side (D). Panel (F) shows the pial surface of a chick neural tube (NT) with cells and axons (neuropil) on the epithelial side of the pial BM (Ep) and connective tissue (CT; stroma) at the stromal/meningeal side of the pial BMs (BM). The tissue arrangement along the pial surface of the spinal cord is analogous to the tissue arrangement along the two surfaces of the implanted DM. Bars: A: 100 lm, B: 10 lm, C: 2 lm; D–F: 1 lm.
crest cell migration and blood vessels development at the time BM implantation and thereafter. Results showed that the spinal cord was a-vascular at the time of BM grafting (E2.2; HH stage 14; Suppl. Fig. 2A). Even one day after BM grafting, the neural tube remains avascular. At that stage, the dorsal aorta is the dominant vascular supply for the dorsal part of the embryo. However, first vascular sprouts from the dorsal aorta curved around the notochord towards the base of the spinal cord (Suppl. Fig. 2B, D). A complete vascularized layer around the entire spinal cord was present only 2 days after grafting (Suppl. Fig. 2C, E, F). Data also showed that that few axons and neural crest cells had exited the neural tube at the time of BM implantation (Suppl. Fig. 2A). Larger number of axons and neural crest cells grew out of the spinal cord at stage 20, approximately 1 day after the BM grafting. At that stage, axons and neural crest cells first encountered the BM implants (Suppl. Fig. 2D; n = 5). Two days after BM implantation, there was extensive axonal outgrowth from the spinal cord (Suppl. Fig. 2E) and a
large population of neural crest cells in the trunk of the embryos (Suppl. Fig. 2F). The current data showed that BM implantation led to a side-specific organization of tissues around the grafted BMs and indicate the BMs have an instructive function in the organization at the pial surface of the central nervous system.
4.
Discussion
4.1.
BMs have side-specific properties in vitro and in vivo
A previous report (Halfter et al., 2013b) showed that adult human BMs have a sidedness that is detectable in vitro by (a) an asymmetric protein/peptide epitope distribution, by (b) differences in the biomechanical quality of both BM sides and (c) by side-selective cell adhesion (Suppl. Fig. 1). In this project, we found that the asymmetrical organization of BMs can also be detected in vivo and that this BM asymmetry has a functional role in organizing tissues during
8
MECHANISMS OF DEVELOPMENT
development. Asymmetric distributions of proteins within BMs have been independently reported for human LC, DM and glomerular BMs (Danysh and Duncan, 2008; Kobosova et al., 2007, and biochemical studies recently confirmed that the epidermal BM has a layered structure, with an abundance of laminin on the epithelial side (Behrens et al., 2012). While an asymmetric distribution of proteins has not been detected for mouse or rat BMs, a functional asymmetry with respect to a selective cell adhesion applied to all BM so far tested, including BMs from neonatal and adult mice and chick embryos (Halfter et al., 2013b). In difference to adult human BMs, BMs from laboratory animals are much thinner, and the detection of a side-selective protein distribution is a much greater challenge. However, adult human BMs also start out as very thin ECM sheets and, at embryonic and neonatal stages, share the textbook BM morphology from mice and rats (Candiello et al., 2010; Murphy et al., 1984; Danysh and Duncan, 2008). The increase in BM thickness in humans is a natural process that gradually proceeds with advancing age. A major increase in BM thickness does not occur in mice and rats because of their much shorter life span. We propose that all BMs are asymmetrically organized from the outset, but that BM asymmetry is more easily detectable when the BMs become thicker and more easily manageable for experimental testing. The origin and structural basis for the BM asymmetry is currently unknown. An explanation could be the origin of a selected group of BM constituents from the epithelial tissues and others from the adjacent stromal tissues (Sarthy and Fu, 1990; Dong and Chung, 1991; Sarthy, 1993; Dong et al., 2002; Halfter et al., 2008). The asymmetric structure of the BMs may also be explained by the way BMs are assembled, namely that laminin assembles first followed by collagen IV later in development (Yurchenco and Patton, 2009). Best evidence for the latter hypothesis comes by a detailed developmental study of the human DM that showed a two-phased growth: a fast fetal growth, and a slow postnatal growth resulting in a two-layered BM with a striated and an amorphous layer (Murphy et al., 1984). We propose that the functional sidedness of BMs is universal for all metazoans, and we hypothesize that the bi-functional nature of BMs participates in organizing multicellular organisms in alternating layers of epithelial and connective tissues. By transplanting human BMs into developing chick embryos we found that the bi-functional organization of human BMs as detected in vitro can also be detected in vivo and that this asymmetry contributes to the establishment of the tissue architecture along the pial surface of the CNS. The asymmetry of BMs in vivo was readily detectable by the side-selective deposition of host ECM proteins onto the implanted human BMs: interstitial ECM proteins, such as collagen II and tenascin-C bound preferentially to the stromal side of the implanted DMs indicating that BM protein epitopes exposed at this side are capable of binding proteins that anchor the BMs to the adjacent interstitial connective tissue. The side-selective binding of protein to the implanted human BMs was restricted to a subset of host ECM proteins. The preference of cell adhesion to the opposite, epithelial side of BMs indicate that BMs are structured such that the laminin-rich epithelial side promotes cell adhesion, whereas the stromal side provides the binding sites that allow BMs
1 3 3 ( 2 0 1 4 ) 1 –1 0
to firmly connect the adjacent stromal ECM. The antiadhesive prosperities of the stromal side of BMs prevent the establishment of two directly adjacent epithelial cell layers.
4.2. BMs as borders for axon growth and neural crest cell migration The current data showed that neither growing axons not migrating neural crest cell were capable of penetrating into or pass through the implanted human BMs. In all cases, axons and neural crest cells detoured the BM implants on their way to more distal targets. Since the implanted human BMs did not cause an inflammatory response, we did not observe diapedesis of macrophages or neutrophil through the BMs, but we predict that migration of neutrophils and macrophages through BMs is possible as it occurs naturally during inflammation. During embryogenesis, neural crest migrate from the neural tube into the adjacent connective tissues, but at the stage of migration, the pial BM of the dorsal neural tube is at best very thin and fragile or may even be incomplete (Martins-Green and Erickson, 1987). Interestingly, neural crest cells and axons did not sense the implanted BMs as a stop signal, rather progressed with their distal migration by growing around the BM obstacle. The interaction of axons and neural crest cells with the implanted BMs differed depending on the sidedness of the BMs: in cases when the axons and neural crest cell approached the epithelial side, they got in direct contact with the BMs surface. In cases when axons and neural crest cell approached the stromal side of the BMs, they did not contact the BMs, rather stayed at a distance of few microns from the inserted BMs. This is consistent with in vitro data showing that axons outgrowth and cell adhesion are restricted to the epithelial side of BMs and confirm the functional sidedness of BMs in vivo. We propose that diapedesis of cells through BMs is easier and faster when cells approach from the epithelial side of the BMs. The transmigration would also be greatly facilitated for cells that don’t show an adhesion preference to only the epithelial side of BMs as it was found in cell adhesion experiments using invasive tumor cell lines (Halfter, unpublished).
4.3. An organizing function of BMs in the central nervous system When implanted into the developing neural tube, the BM led to a split into two spinal cords with two separate central canals and a dividing meningeal layer. It is of note that the neural tube has no vasculature and no axonal layers at the time of BM implantation. Thus, the arrangement of cells following the implantation was orchestrated by the de-novo interaction of cells with the inserted BMs. We found that in all cases axons, neural and glia cells were in direct contact with the epithelial side of the implanted BMs and assembled as a dense neuropil at this side. Axons, glia cells and neurons stayed away from the stromal side of the implanted BMs and there was always a several micrometer-wide connective tissue/vascular layer, indicating that the stromal surface environment of BMs is anti-adhesive for neurons, glia cells and axons but promotes the assembly of a connective tissue layer
MECHANISMS OF DEVELOPMENT
that is highly vascularized. Based on the current in vivo data and consistent with the in vitro assays, we propose that BMs have an organizing function in vivo that plays a major role in the assembly of the tissue architecture at the pial surface of the spinal cord and brain: the epithelial side of BMs promotes the adhesion of neurons, glia cells and axons, whereas the stromal side promotes the assembly of a highly vascularized connective tissue layer. The bi-functional nature of BMs may also play a role in other developmental processes that require the separation or connection of epithelial and connective tissues. We also propose that the bi-functional nature of the human BMs promoted the splitting of the neural tube into two rather symmetric spinal cords. In difference to laminincoated nitrocellulose implants (Fig. 5A), the human BMs promoted the formation of a continuous vascular layer though the midline of the spinal cord. The ectopic vascular layer through the spinal cord midline was continuous with the meningeal vascular layer that outlines the spinal cord. A caveat of these experiments is that the BMs used in these experiments were isolated from adult human eyes, thus are not the thin BMs that neurons, axons and neural crest cells encounter in embryos. However, the major goal of the current experiments was not to mimic the situation of axons and neural crest cell encountering BMs in vivo, rather to find out whether in vivo-derived BMs show side-specific properties in vivo and that these properties have the potential to organize tissues and cells in a life embryo.
Acknowledgements We would like to thank Drs. Nirmala SundarRaj, and Katharine Davoli for generously providing us with collagen IV antibodies. The current project was funded by an in-house grant from the University of Pittsburgh.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.mod.2014.07.003. R E F E R E N C E S
Arikawa-Hirasawa, E., Watanabe, H., Takami, H., Hassell, J.R., Yamada, Y., 1999. Perlecan is essential for cartilage and cephalic development. Nat. Genet. 23, 354–358. Bader, B.L., Smyth, N., Nedbal, S., Miosge, N., Baranowsky, A., et al, 2005. Compound genetic ablation of nidogen 1 and 2 causes basement membrane defects and perinatal lethality in mice. Mol. Cell. Biol. 25, 846–6856. Balasubramani, M., Bier, M.E., Hummel, S., Schneider, W.J., Halfter, W., 2004. Perlecan and its immunoglobulin like domain IV are abundant in vitreous and serum of the chick embryo. Matrix Biol. 23, 143–152. Behrens, D.T., Villone, D., Koch, M., Brunner, G., Sorokin, L., et al, 2012. The epidermal basement membrane is a composite of separate laminin- or collagen IV-containing networks connected by aggregated perlecan, but not by nidogens. J. Biol. Chem. 287, 18700–18709.
1 3 3 ( 2 0 1 4 ) 1 –1 0
9
Candiello, J., Cole, G.J., Halfter, W., 2010. Age-dependent changes in the structure, composition and biophysical properties of a human basement membrane. Matrix Biol. 29, 402–410. Costell, M., Gustafsson, E., Aszodi, A., Moergelin, M., Bloch, E., et al, 1999. Perlecan maintains the integrity of cartilage and some basement membranes. J. Cell Biol. 147, 1109–1122. Danysh, B.P., Duncan, M.K., 2008. The lens capsule. Exp. Eye Res. 88, 151–164. Dong, L.J., Chung, A.E., 1991. The expression of the genes for entactin, laminin A, laminin B1 and laminin B2 in murine lens morphogenesis and eye development. Differentiation 48, 157– 172. Dong, S., Landfair, J., Balasubramani, M., Bier, M.E., Cole, G., et al, 2002. Expression of basal lamina protein mRNA in the early embryonic chick eye. J. Comp. Neurol. 447, 261–273. Erickson, A.C., Couchman, J.R., 2000. Still more complexity in mammalian basement membranes. J. Histochem. Cytochem. 48, 1291–1306. Fassler, R., Meyer, M., 1995. Consequences of lack of b1 integrin gene expression in mice. Genes Dev. 9, 1896–1908. Fukai, N., Eklund, L., Marneros, A.G., Oh, S.P., Keene, D.R., et al, 2002. Lack of collagen XVIII/endostatin results in eye abnormalities. EMBO J. 21, 1535–1544. Gautam, M., Noakes, P.G., Moscovo, L., Rupp, F., Scheller, R.H., et al, 1996. Defective neuromuscular synaptogenesis in agrindeficient mutant mice. Cell 85, 525–535. Gould, D.B., Phalan, F.C., Breedveld, G.J., van Mil, S.E., Smith, R.S., et al, 2005. Mutations in Col4a1 cause perinatal cerebral hemorrhage and porencephaly. Science 308, 1167– 1171. Halfter, W., Newgreen, D.F., Sauter, J., Schwarz, U., 1983. Oriented axon outgrowth from avian embryonic retinae in culture. Dev. Biol. 95, 56–64. Halfter, W., 1993. A heparan sulfate proteoglycan in developing avian axonal tracts. J. Neurosci. 13, 2863–2873. Halfter, W., Schurer, B., Yip, J., Yip, Y.P., Lee, J.A., Cole, G.J., 1997. Distribution and substrate properties of agrin, a heparan sulfate proteoglycan of developing axonal pathways. J. Comp. Neurol. 383, 1–17. Halfter, W., Willem, M., Mayer, U., 2002. A critical function of the pial basement membrane in cortical histogenesis. J. Neurosci. 22, 6029–6040. Halfter, W., Willem, M., Mayer, U., 2005. Basement membranedependent survival of retinal ganglion cells. Invest. Ophthalmol. Vis. Sci. 46, 1000–1009. Halfter, W., Winzen, U., Bishop, P.N., Eller, A., 2006. Regulation of eye size by the retinal basement membrane and vitreous body. Invest. Ophthalmol. Vis. Sci. 47, 3586–3594. Halfter, W., Dong, S., Dong, A., Eller, A.W., Nischt, R., 2008. Origin and turnover of ECM proteins from the inner limiting membrane and vitreous body. Eye 22, 1207–1213. Halfter, W., Candiello, J., Hu, H., Zhang, P., Schreiber, E., Balasubramani, M., 2013a. Protein composition and biomechanical properties of in vivo-derived basement membranes. Cell Adh Migr 7, 64–71. Halfter, W., Monnier, C., Mu¨ller, D., Oertle, P., Uechi, G., Balasubramani, M., Safi, F., Lim, R., Loparic, M., Henrich, P.B., 2013b. The bi-functional organization of human basement membranes. PLoS One 8, e67660. Henry, M.D., Campbell, K.P., 1998. A role of dystroglycan in basement membrane assembly. Cell 95, 859–870. Kobosova, A., Azar, D.T., Bannikov, G.A., Campbell, K.P., Durbeej, M., et al, 2007. Compositional differences between infant and adult human corneal basement membranes. Invest. Ophthalmol. Vis. Sci. 48, 4989–4999. Lagenaur, C., Lemmon, V., 1987. An L1-like molecule, the 8D9 antigen, is a potent substrate for neurite extension. Proc. Natl. Acad. Sci. U.S.A. 84, 7753–7757.
10
MECHANISMS OF DEVELOPMENT
Lee, J., Gross, J.M., 2007. Laminin beta1 and gamma1 containing laminins are essential for basement membrane integrity in zebra fish eye. Invest. Ophthalmol. Vis. Sci. 48, 2483–2490. Martins-Green, M., Erickson, C.A., 1987. Basal lamina is not a barrier to neural crest cell emigration: documentation by TEM and by immunofluorescent and immunogold labelling. Development 10, 517–533. Murphy, C., Alvarado, J., Juster, R., 1984. Prenatal and postnatal growth of the human Descemet’s membrane. Invest. Ophthalmol. Vis. Sci. 25, 1402–1415. Po¨schl, E., Schlotzer-Schrehardt, U., Brachvogel, B., Saito, K., Ninomiya, Y., et al, 2004. Collagen IV is essential for basement membrane stability but dispensable for initiation of its assembly during early development. Development 131, 1619– 1628. Ring, C., Lemmon, V., Halfter, W., 1995. Two chondroitin sulfate proteoglycans differentially expressed in the developing chick visual system. Dev. Biol. 168, 11–27. Sarthy, P.V., Fu, M., 1990. Localization of laminin B1 mRNA in retinal ganglion cells by in situ hybridization. J. Cell Biol. 110, 2099–2108. Sarthy, V., 1993. Collagen IV mRNA expression during development of the mouse retina: an in situ hybridization study. Invest. Ophthalmol. Vis. Sci. 34, 145–152.
1 3 3 ( 2 0 1 4 ) 1 –1 0
Smyth, N., Vatansever, H.S., Murray, P., Meyer, M., Frie, C., et al, 1999. Absence of basement membranes after targeting the LANC1 gene results in embryonic lethality due to failure of endoderm differentiation. J. Cell Biol. 144, 151–160. Stephens, L.E., Sutherland, A.E., Klimanskaya, I.V., Andrieux, A., Meneses, J., et al, 1995. Deletion of beta 1 integrins in mice results in inner cell mass failure and peri-implantation lethality. Genes Dev. 9, 1896–1908. SundarRaj, N., Wilson, J., 1982. Monoclonal antibody to human basement membrane collagen IV. Immunology 47, 133–140. Timpl, R., Brown, J.C., 1996. Supramolecular assembly of basement membranes. Bioassays 18, 123–132. Willem, M., Miosge, N., Halfter, W., Smyth, N., Janetti, I., et al, 2001. Specific ablation of the nidogen binding site in the laminin c1 chain interferes with kidney and lung development. Development 129, 2711–2722. Yip, J.W., 1987. Target cues are not required for the guidance of sympathetic preganglionic axons. Brain Res. 429, 155–159. Yurchenco, P.D., Patton, B.L., 2009. Developmental and pathogenic mechanisms of basement membrane assembly. Curr. Pharm. Des. 15, 1277–1294.
