The Delta-like 1 (Dlk1) gene encodes a transmembrane protein that belongs to the Delta-Notch family of signaling molecules. This protein has been variously named DLK1, pG2, FA-1, Pref-1, SCP-1, and ZOG (Fay et al.,1988; Laborda et al.,1993; Smas and Sul,1993; Okamoto et al.,1997; GenBank, D16847). The mouse Dlk1 and human DLK1 genes are subject to genomic imprinting; Dlk1 is expressed only from the paternally inherited allele, whereas the maternally inherited allele is silent (Kobayashi et al.,2000; Schmidt et al.,2000; Takada et al.,2000). The DLK1 protein contains a signal peptide, six EGF-like repeats, a juxtamembrane region, a transmembrane domain, and a short intracellular tail (Smas et al.,1994). DLK1 lacks the DSL domain shared by all Notch ligands, however, and is not believed to bind a canonical Notch receptor. DLK1 is proteolytically cleaved within the juxtamembrane region, producing membrane-associated and secreted domains (Bachmann et al.,1996; Smas et al.,1997). The functions of these two forms of the protein are currently the subject of debate (Smas et al.,1997; Garces et al.,1999; Mei et al.,2002). The Dlk1 gene is expressed at high levels in the embryo and placenta during development (Schmidt et al.,2000). We have shown previously that Dlk1 is first expressed at e11 in the mouse, with levels increasing until late gestation (Schmidt et al.,2000). In the human embryo, DLK1 protein was detected by immunohistochemistry in numerous tissues, including the pituitary gland, pancreas, adrenal gland, and skeletal muscle (Jensen et al.,1993, 1994; Larsen et al.,1996; Floridon et al.,2000). Expression of DLK1 is down-regulated in most adult tissues, becoming restricted to the β cells of the islets of Langerhans, the bone marrow, the pituitary gland, and the adrenal gland (Jensen et al.,1993; Larsen et al.,1996; Tornehave et al.,1996).
Current evidence suggests that DLK1 functions as a growth factor, maintaining proliferating cells in an undifferentiated state. Studies have shown that Dlk1 expression maintains proliferating 3T3-L1 preadipocyte cells, and this expression must be down-regulated for differentiation to occur (Smas and Sul,1993; Smas et al.,1997). In the developing human pancreas, DLK1 was reported to colocalize with insulin, indicating that it is expressed in the β cells (Tornehave et al.,1996). Treatment of rat neonatal islets of Langerhans with growth hormone (GH) results in a mitogenic effect and an up-regulation of Dlk1 mRNA (Carlsson et al.,1997). A recent report suggested that Dlk1 is not directly mitogenic for pancreatic β cells; however, the authors examined only the secreted form of the DLK1 protein (Friedrichsen et al.,2003). In cultures isolated from mouse bone marrow, fetal liver, and thymus, Dlk1 is expressed in stromal cells that influence the development of the surrounding hematopoietic cells (Moore et al.,1997; Bauer et al.,1998; Kaneta et al.,2000). Also supporting a growth regulatory role for DLK1 is the phenotype of mice carrying a mutation in the Dlk1 gene (Moon et al.,2002). Dlk1 knockout mice display greater than 50% perinatal lethality, and surviving animals show growth retardation, skeletal malformations, eyelid defects, and a late-onset increase in adipose tissue. The specific Dlk1-expressing cell types responsible for most of these diverse effects, however, are still unknown. The Dlk1 gene was originally identified based on its expression in small cell lung cancer, and it is up-regulated in other human cancers as well (Laborda et al.,1993; Harken Jensen et al.,1999; van Limpt et al.,2000).
Most imprinted genes are expressed in the placenta, a key regulator of embryonic growth (Reik et al.,2003). In human extraembryonic tissues, DLK1 was observed in endodermal cells surrounding the blood islands of the yolk sac and in the stromal cells of the placental villi (Floridon et al.,2000). The function of Dlk1 in the placenta is of significant interest, given its role in embryonic growth regulation. Despite their similar functions, however, placental morphologies vary widely between mammalian species. This variation prevents human experimental data from being extrapolated to the mouse. The human and mouse placentae are both hemochorial, where trophoblast cells directly contact the maternal blood supply, but humans have a villous placenta with chorionic villi extending into the maternal blood filling the intervillous space, whereas mice have a labyrinthine placenta, where the highly branched structure of the labyrinth brings maternal and fetal circulation into close proximity (Pijnenborg et al.,1981; Georgiades et al.,2002). It has been proposed that the human chorionic villi and the mouse labyrinth are functionally homologous structures (Cross,2000; Rinkenberger and Werb,2000). The labyrinth of the mouse placenta contains no stromal cells, however, so the Dlk1-expressing cell type is unknown.
To further investigate the role of the Dlk1 gene as a growth factor, and to identify those tissues where it may be exerting this role, the Dlk1 mRNA expression pattern was analyzed by in situ hybridization in mouse embryo and placenta at e12.5 and e16.5. These developmental stages represent, respectively, periods of high cellular proliferation and increasing differentiation. As Dlk1 is believed to have a role in proliferation and is decreased in differentiating cells, these time points provide a view of Dlk1 expression at two functionally distinct developmental stages.
RESULTS AND DISCUSSION
In Situ Analysis of Dlk1 Expression in Embryonic Tissues
In the e12.5 mouse embryo, Dlk1 expression was found in numerous tissues and in most undifferentiated mesenchyme (Fig. 1B). By e16.5, however, Dlk1 mRNA is much less abundant (Fig. 1C). Significant expression can still be seen in the pituitary and adrenal glands and in skeletal muscle. This more restricted pattern may be an indication of reduced proliferation and increased levels of differentiation among the embryonic cell types. Dlk1 expression was seen in the endothelium of many developing blood vessels. The endothelial cells that line the blood vessels of the developing trigeminal ganglion express Dlk1 (Fig. 1D), as do the major blood vessels of the head, such as the cerebral arteries (Fig. 1E). Dlk1 expression was also found in several regions within the developing visual system. Strong Dlk1 expression was observed in the optic stalk (developing optic nerve; Fig. 1F, arrowhead), in the optic chiasm (Fig. 1F, “oc”), and in the mesodermal condensations of the developing extrinsic ocular muscles (Fig. 1F, arrow). Dlk1 expression was also seen in the endothelial capillaries of the hyaloid vascular plexus, a transient structure that feeds the developing retina and lens (Fig. 1G).
In the brain of e12.5 embryos, the most prominent site of Dlk1 expression was in the developing pituitary gland (Fig. 2A). The pituitary is built from two different tissues, Rathke's pouch, which arises from the oral ectoderm and forms the anterior and intermediate lobes of the gland, and the infundibulum, which arises from the neural ectoderm of the ventral diencephalon and forms the posterior lobe. These two regions are directly juxtaposed during pituitary development, and inductive signaling between them is required for normal pituitary morphogenesis (Takuma et al.,1998). Dlk1 is highly expressed in both the infundibulum and in the developing pouch (Fig. 2A). Unlike in many other tissues, Dlk1 expression remains quite strong in the pituitary at e16.5 (Fig. 2B).
The anterior pituitary is composed of six different types of cells, which arise in a temporally specific pattern beginning at e12.5, and each cell type produces a characteristic hormonal product (for reviews of mouse pituitary organogenesis, see Sheng and Westphal,1999; Scully and Rosenfeld,2002). It was reported previously that DLK1 expression is restricted to the growth hormone-producing somatotroph cells of the mature human pituitary (Larsen et al.,1996). Somatotrophs first arise in the mouse pituitary beginning at e16.5 and constitute up to 50% of the cells of the mature gland (Scully and Rosenfeld,2002). To determine whether Dlk1 is expressed in the somatotroph cells of the mouse, in situ hybridization was performed on sections of postnatal day 5 (P5) embryonic head. Dlk1 expression was detected in a significant fraction of the cells of the mouse anterior pituitary, with lower levels of expression in the intermediate and posterior lobes (Fig. 2C). To further identify the Dlk1-expressing cell type, adjacent sections were subjected to immunohistochemistry with an antibody to the pituitary somatotroph marker GH (Fig. 2D). The similar expression pattern of Dlk1 and GH suggested Dlk1 expression is localized to the somatotrophs of the anterior pituitary.
To identify more precisely the Dlk1-expressing cell types within the pituitary gland, immunohistochemistry was used to colocalize the DLK1 protein with GH in individual sections from e16.5, P5, and adult (4-week) pituitaries. Sections were labeled with anti-DLK1 and anti-GH antibodies and detected with fluorescein isothiocyanate (FITC) -conjugated and Texas Red–conjugated secondary antibodies, respectively. Imaging of e16.5 pituitaries showed that DLK1 is expressed at easily detectable levels in most cells (Fig. 2E), while GH is expressed at varying levels in a subset of cells, primarily in the anterior pituitary (Fig. 2F). Figure 2G shows a merged image of DLK1 and GH expression patterns. Note that the very brightly stained cells in all three images represent autofluorescence from clusters of red blood cells. At this stage, therefore, DLK1 and GH are not specifically colocalized. As Dlk1 RNA expression is decreasing at this time point (Fig. 2B), it is likely that levels of DLK1 protein are also decreasing, while GH is just beginning to be expressed. At postnatal day 5 (data not shown), immunofluorescence of DLK1 and GH still shows no strict colocalization. DLK1 levels are lower than at e16.5, and GH levels have increased, but individual cells may express one, the other, or both proteins. Taken together, the e16.5 and P5 time points mark a period of dynamic regulation of both DLK1 and GH; a transition between embryonic and adult expression patterns. By the time mice have reached 4 weeks of age, however, a strong colocalization of DLK1 and GH is seen (Fig. 2H–J). Most DLK1-positive cells also express GH, although GH levels vary and differences in subcellular localization preclude a perfect overlap of the fluorescent signals within each cell. Given the increased levels of Dlk1 seen in β cells treated with GH (Carlsson et al.,1997), it is interesting to speculate that Dlk1 may regulate growth by itself promoting the expression of GH. An important role for Dlk1 in the pituitary might also explain the growth phenotype seen in surviving Dlk1 null mice (Moon et al.,2002).
Numerous mammalian tissues are generated by a developmental process known as branching morphogenesis, in which epithelial structures bud off to form tubules that grow distally, branching repeatedly to form structures of narrowing diameter (for a recent review, see Lubarsky and Krasnow,2003). Branching morphogenesis is dependent upon interactions between the distal growing epithelium of the bud and the surrounding mesenchyme. Several pairs of signaling molecules and their receptors, such as the fibroblast growth factors and bone morphogenetic proteins, are differentially expressed in the epithelium and mesenchyme (Warburton et al.,2000). Expression of Dlk1 was detected in the developing pancreas, lung, and submandibular gland, structures that develop by branching morphogenesis. In the e12.5 pancreas, Dlk1 expression was observed in the distal regions of the branching epithelium, with strong expression in the pancreatic mesenchyme as well (Fig. 3A). In the e16.5 pancreas, Dlk1 expression was significantly down-regulated, with only a small percentage of cells still expressing Dlk1 (Fig. 3B). It is likely these cells represent clusters of differentiating β cells, because DLK1 protein has been shown to colocalize with the expression of insulin in the developing human pancreas (Jensen et al.,1994). Differentiated β cells are present in the pancreas by this stage of embryogenesis and are producing insulin. The identity of the Dlk1-expressing cells was further analyzed by immunohistochemistry using an antibody that recognizes mouse insulin. Expression of Dlk1 mRNA and that of insulin protein was compared in adjacent sections of e16.5 pancreas (Fig. 3B,C). General colocalization of expression was detected, suggesting Dlk1 marks β cells in the mouse as well.
Dual-labeling immunohistochemistry experiments were used to analyze the cell type-specific expression of the DLK1 protein, in contrast to mRNA, in the pancreas during development and adulthood. As the DLK1 and insulin antibodies are both derived from rabbit, however, double labeling of individual sections was not possible. Instead, immediately adjacent sections were used with a goal of approximating colocalization. Whereas individual cells are unlikely to appear on adjacent sections, a single islet can be assayed for both proteins. At all stages analyzed however, e16.5, P5, and adult, DLK1 protein could not be detected in the mouse pancreas (Fig. 3D, and data not shown). That Dlk1 RNA is present, albeit at low levels, suggests the protein may be present as well, but at levels below the detection limit of the antibody used in this study. The DLK1 antibody has been proven to be functional and specific, and it readily detected DLK1 expression in pituitary (Fig. 2E). Although DLK1 protein has been reported previously in both human and rat fetal islets (Tornehave et al.,1996; Carlsson et al.,1997; Floridon et al.,2000), it may be that levels in the mouse are significantly lower than in these species. The insulin antibody detected expression in e16.5, P5, and in the islets of adult pancreas (Fig. 3E, and data not shown).
In the developing lung at e12.5, Dlk1 expression was seen in epithelium of the distal (growing) tips of the segmental bronchi (Fig. 4A, arrow), and in the surrounding undifferentiated mesenchyme. No Dlk1 expression was seen in the more proximal portions of the bronchi (Fig. 4A, arrowhead). At e16.5, Dlk1 expression was decreased significantly, remaining only in the epithelium of the terminal bronchioles (Fig. 4B) and in the connective tissue surrounding the developing pulmonary artery (Fig. 4B, arrow). The submandibular salivary gland also grows by the process of branching morphogenesis. At e16.5, Dlk1 is expressed in the mesenchyme of submandibular gland, as well as in the distal tips of the branching epithelium (Fig. 4C). Since Dlk1 is known to maintain cells in the proliferative state, it is tempting to suggest that it has the same function in these developing tissues, allowing polarized growth along a proscribed branching pathway. Of interest, in all these tissues Dlk1 is expressed in both the growing epithelium as well as in the surrounding mesenchyme. This is a different pattern than other known signaling molecules that regulate branching morphogenesis, which tend to be localized to one tissue or the other. This pattern suggests that Dlk1 is unlikely to participate in inductive signaling itself but, rather, may function to maintain pools of both proliferating epithelial and mesenchymal cells needed for further growth.
The hepatocytes of the fetal liver exhibited high levels of Dlk1 at e12.5 (Fig. 4D) with significant down-regulation of expression at e16.5 (data not shown). Although hematopoietic cells occupy more than 50% of the liver at e12.5, they do not express Dlk1. Dlk1 expression was also seen in the connective tissue capsule of the liver (Fig. 4D, arrowhead). Significant levels of Dlk1 expression were found in the adrenal gland, another major neuroendocrine organ. Similar to the pituitary, the adrenal gland is composed of two distinct functional units of different embryological origin: a cortex derived from the mesoderm of the coelomic epithelium, and a medulla derived from the neural crest. Adrenal development begins with an outgrowth of the coelomic epithelium at e9 that forms the cortex, which is then is invaded by migrating chromaffin cells beginning at e12. At e12.5, Dlk1 was detected within discrete clusters of cells in the developing adrenal gland (Fig. 4E), and by e16.5, these cells were localized to the adrenal medulla, identifying them as catecholamine-producing chromaffin cells (Fig. 4F, arrow). Dlk1 is also found in the zona glomerulosa of the cortex at e16.5, a tissue responsible for maintaining water–electrolyte balance through the secretion of aldosterone (Fig. 4F, arrowhead).
Dlk1 expression was seen widely in developing skeletal muscle and cartilage at both e12.5 and e16.5. The skeletal muscle of the tongue is one of the most highly expressing tissues at e12.5 (Fig. 5A), and Dlk1 persists in skeletal muscle at e16.5 (the diaphragm is shown), even as its expression is diminishing in most other tissues (Fig. 5B). Dlk1 is highly expressed in skeletal muscle during embryogenesis in the sheep but is down-regulated postnatally. It appears to play a growth-promoting role in this tissue, as the sheep CLPG mutation shows hypertrophy of those muscles where Dlk1 remains active (Cockett et al.,1996; Murphy et al.,2005). All cartilage primordia showed high levels of Dlk1 expression at e12.5 (Fig. 5C). By e16.5, Dlk1 expression in the developing skeleton has decreased, with only the distal regions of the developing bones, particularly the epiphyses of the long bones, still reactive for Dlk1 (Fig. 5D). This timing coincides with increasing levels of ossification of the cartilaginous skeleton, suggesting Dlk1 down-regulation in chondroblasts may be necessary to allow the invasion and proliferation of osteoblasts. The expression of Dlk1 in developing bone may help to explain the skeletal defects found in mice lacking Dlk1 (Moon et al.,2002). Dlk1 function has been investigated extensively in developing adipocytes, where it promotes the proliferation of preadipocytes and must be down-regulated for these cells to differentiate to mature adipocytes (Smas and Sul,1993). Expression of Dlk1 was found in the developing brown adipose tissue of the mouse at e16.5 (Fig. 5E). Surprisingly, however, Dlk1 was confined to a small number of preadipocyte cells, with the majority of the expression found in the mesenchyme surrounding the lobules of preadipocyte cells (Fig. 5F).
In Situ Analysis of Dlk1 Expression in Extraembryonic Tissues
The mouse placenta is a complex structure that is composed of both maternal- and fetal-derived cell types: the maternal decidua and maternal vasculature, and the fetal trophoblast and fetal vasculature. The bulk of the mouse placenta is composed of trophoblast derived from the trophectoderm of the blastocyst. It is in the tortuous labyrinth that embryonic and maternal blood comes into a close contact for physiological exchange (for a review of placental development, see Rossant and Cross,2001). Most imprinted genes are expressed in the placenta, and many are involved in regulating the production of growth factors and other hormones, as well as in maternal–fetal nutrient transfer (Georgiades et al.,2002; Reik et al.,2003; Sibley et al.,2004). In fact, it has been proposed that regulation of fetal growth was the driving force in the evolution of imprinting (Moore and Haig,1991).
At both e12.5 (Fig. 6A–D) and e16.5 (not shown), Dlk1 was detected in the endothelial cells lining the embryonic vasculature of the placental labyrinth. All other placental cells, including all trophoblast derivatives, were Dlk1-negative. Fetal and maternal vessels in the placenta can be differentiated by the presence of nucleated or enucleated red blood cells, respectively, and at high magnification Dlk1 expression was found only in those vessels carrying nucleated fetal red cells (Fig. 6D). To further confirm the identity of the Dlk1-expressing cells, immunohistochemistry was used to compare the pattern of Dlk1 expression with that of the CD31 (platelet endothelial cell adhesion molecule-1, or PECAM-1) protein. PECAM-1 is a 130-kDa integral membrane protein found in platelets and endothelial cells and is involved in cell–cell adhesion (Newman et al.,1990). PECAM-1 is commonly used as a marker for embryonic and adult endothelial cells (Vecchi et al.,1994). PECAM-1 protein and Dlk1 mRNA displayed similar patterns of expression in the endothelial cells of the labyrinth (Fig. 6E,F). The maternal blood sinuses are not an endothelial structure but are lined with trophoblast cells and, therefore, are negative for PECAM-1. The large maternal blood vessels of the decidua are PECAM-1–positive but are Dlk1–negative. The development of the placental labyrinth vasculature is often referred to as an example of branching morphogenesis. The placenta is, therefore, a tissue where two emerging roles for Dlk1—endothelial development and the regulation of branching morphogenesis—are involved in a single morphological event. The fetal vasculature of the placenta is derived from the mesoderm of the allantois, and following chorioallantoic fusion, the allantois itself develops into the umbilical cord. Dlk1 expression is detectable by whole-mount in situ hybridization within the umbilical cord (Fig. 6G). Dlk1 is also detected in the endothelial cells that line the vessels of the mature blood islands of the yolk sac, a cell type (like the allantois) that is derived from the extraembryonic mesoderm (Fig. 6H). This finding is in contrast to the human, where DLK1 protein was observed in the endodermally derived cells of the yolk sac (Floridon et al.,2000).
The expression of Dlk1 in the vasculature of the placental labyrinth is quite different from the pattern reported in human placenta, where Dlk1 is expressed in the stromal cells of the chorionic villi. While these two tissues are structurally distinct, both are closely juxtaposed to the site of maternal–fetal exchange. In both species, the Dlk1-expressing cell type is located within the few cell layers that separate the fetal and maternal blood supplies. The endothelial cell expression of Dlk1 places it in a location from which to contribute to the regulation of maternal–fetal communication. In addition, it has been shown that human DLK1 protein is secreted at high levels into amniotic fluid (Jensen et al.,1993; Bachmann et al.,1996). The finding of Dlk1 mRNA in the fetal vasculature gives a potential source for this secreted DLK1 protein.
Dlk1 expression in the placental labyrinth may also help to explain one phenotype of mice carrying uniparental disomies for chromosome 12. The term uniparental disomy describes a situation in which both copies of a given chromosome are derived from either the mother (mUPD12) or the father (pUPD12). As dosage of biallelically expressed genes is not altered, UPD phenotypes are typically due to the loss or overexpression of imprinted genes. Both mUPD12 and pUPD12 animals die during gestation, and they show under- and overgrowth of the placenta, respectively (Georgiades et al.,2001). Most significantly, pUPD12 mice show defects in the fetal vasculature of the placenta, suggesting that the increased levels of Dlk1 found in these animals may interfere with morphogenesis of the labyrinth.
The data presented here confirm that many of the known DLK1-expressing tissues in humans are also positive for Dlk1 in the mouse, yet the data reveal significant differences as well. Most striking are the previously unreported expression of Dlk1 in developing endothelial cells in the vasculature of the mouse embryo and placenta and the identification of a role for Dlk1 in branching morphogenesis. Taken together, these expression data support a role for Dlk1 in maintaining proliferating cells in the undifferentiated state and suggest that Dlk1 may also play a role in the regulation of physiological exchange that occurs within the placenta.
Mouse Breeding and Tissue Preparation
Matings were set up between FVB/N mice, and females were checked for the presence of copulation plugs each morning. Noon on the day of the presence of a plug was considered to be embryonic day 0.5 (e0.5). Embryos and placentae were isolated at different stages of development and fixed in 4% paraformaldehyde in phosphate buffered saline (PBS; 2 hr overnight). Tissues were embedded in paraffin, and 6-micron sections were cut.
In Situ Hybridization
A 737-bp fragment that corresponds to exons 4 and 5 of the Dlk1 cDNA was amplified by reverse transcriptase-polymerase chain reaction using the primers Dlk1F3, 5′-ATGGCGTCTGCACCGACATC-3′, and Dlk1R3, 5′-CACACAATAGAGCAAACTCCACCAC-3′. The fragment was cloned into the pCRII-TOPO vector (Invitrogen), excised with EcoRI, and cloned into EcoRI site of pBKSII in both orientations. These plasmids were linearized with XhoI, phenol:chloroform extracted, and ethanol precipitated. Sense and antisense digoxigenin-labeled cRNA probes were generated by in vitro transcription using 1 μg of template, digoxigenin-11-UTP (Roche), and T7 RNA polymerase (Promega).
Tissue sections were cleared in xylene, rehydrated through a graded ethanol series, and treated with 20 μg/ml proteinase K for 7.5 min at 37°C. Hybridization was carried out overnight at 68°C in formamide hybridization buffer (50% formamide, 0.75 M NaCl, 10 mM PIPES pH 6.8, 1 mM ethylenediaminetetraacetic acid [EDTA], 100 μg/ml tRNA, 0.05% heparin, 0.1% bovine serum albumin, 1% sodium dodecyl sulfate [SDS]) containing 0.5 μg/ml RNA probe. After hybridization, slides were washed twice with Wash I (300 mM NaCl, 10 mM PIPES pH 6.8, 1 mM EDTA, 1% SDS) and twice with Wash II (50 mM NaCl, 10 mM PIPES, 1 mM EDTA, 0.1% SDS), then treated with RNaseA (10 μg/ml) for 1 hr at 37°C. Slides were washed once with Wash III (50% formamide, 300 mM NaCl, 10 mM PIPES, 1 mM EDTA, 1% SDS) and once with Wash IV (50% formamide, 150 mM NaCl, 10 mM PIPES, 1 mM EDTA, 0.1% Tween 20), then incubated with anti–digoxigenin-alkaline phosphatase conjugated antibody at 4°C overnight. The slides were washed with several changes of Wash V (140 mM NaCl, 2.7 mM KCl, 25 mM Tris pH 7.6, 0.1% Tween 20 in TBST), developed with nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) and counterstained with 0.2% Nuclear Fast Red.
The primary antibodies used for immunohistochemistry were as follows: rabbit anti-human Dlk1 (Santa Cruz Biotechnology), rat anti-mouse CD31 (PECAM-1; BD Pharmingen), rabbit anti-human insulin (Santa Cruz Biotechnology), rabbit anti-mouse growth hormone for enzymatic detection, and guinea pig anti-rat growth hormone for fluorescent detection (both from National Hormone and Peptide Program). Tissue sections were dewaxed in xylene (three changes for 10 min each) and rehydrated through a graded ethanol series. For the growth hormone and insulin antibodies, microwave antigen retrieval was performed (boiling in 0.1 M pH 6.0 citrate buffer). For the PECAM antibody, the slides were incubated in 1% trypsin/PBS for 5 min, and trypsin inactivated with three changes of 1% serum. Endogenous peroxidase activity was quenched with 1% H2O2/methanol for 20 min. After blocking with 10% normal serum in PBS for 20 min, sections were incubated with primary antibody for 2 hr to overnight at 4°C. The primary antibodies were visualized by enzymatic detection using anti-rat and anti-rabbit ABC kits (Santa Cruz Biotechnology or Vector Laboratories) according to the manufacturer's protocols. For fluorescent detection, the endogenous peroxidase quenching step was omitted and the primary antibodies were visualized using FITC-conjugated donkey anti-rabbit and Texas Red-conjugated donkey anti-guinea pig antibodies (Jackson Immunoresearch). The sections were counterstained with 2% methyl green. Low-power images (i.e., Figs. 1A–C, 6A,B,E,G) were collected on a Leica MZFLIII microscope using a Leica DFC320 camera and the manufacturer's software. High-power images were collected on a Zeiss Axiovert 200M microscope equipped with ApoTome imaging, using a Zeiss AxioCam MRc5 and the manufacturer's software.
The authors thank Robert Costa (UIC) and Vladimir Kalinichenko (University of Chicago) for assistance with immunohistochemistry. All animals used in these experiments were maintained in compliance with the National Institutes of Health's Guide for the Care and Use of Laboratory Animals and The University of Illinois at Chicago Animal Care Committee guidelines. J.V.S. was supported by a grant from the National Institutes of Health.