The tight junction is a cell–cell contact that defines the apical and basal surfaces of epithelial and endothelial cells (van der Wouden et al., 2003) and bars the paracellular passage of small molecules between these cells (Stevenson and Keon, 1998). The formation of the trophoectoderm epithelium of the preimplantation embryo is coincident with the formation of intercellular junctions, including tight junctions (Fleming et al., 2000, 2001). The reorganization and further development of epithelia during morphogenesis and organogenesis requires the regulated control of tight junction formation, cell adhesion, and cell migration (Denker and Nigam, 1998; Mlodzik, 2002).
Junctional adhesion molecule-A (JAM-A or JAM-1; UniGene designation F11r) is a 32-kDa transmembrane glycoprotein consisting of two immunoglobulin (Ig) loops on the extracellular domain and a short intracellular domain (Bazzoni, 2003; Naik and Eckfeld, 2003; Ebnet et al., 2004). JAM-A is localized to tight junctions of epithelial and vascular endothelial cells, and its cytoplasmic domain interacts with several tight junction–associated proteins, including ZO-1 and PAR-3 (Ebnet et al., 2000, 2001). JAM-A can also participate in homophilic interactions with JAM-A proteins on adjacent cells and heterophilic interactions with integrins, including LFA-1 and αvβ3 (Bazzoni, 2003). JAM-A controls leukocyte transmigration between and platelet adhesion to vascular endothelial cells (Bazzoni and Dejana, 2004) and is required for basic fibroblast growth factor (bFGF)-induced angiogenesis (Naik et al., 2003a). Whereas JAM-A mRNA and protein have been detected in preimplantation human and mouse embryos (Ghassemifar et al., 2003; Thomas et al., 2004) and JAM-A protein is detected in several epithelial tissues (Martin-Padura et al., 1998), JAM-A mRNA has been reported to be absent from mouse embryos from 7 to 17 days post coitum (dpc; Aurrand-Lions et al., 2001). Here, we investigated the developmental expression pattern of JAM-A during organogenesis using both mice harboring a lacZ-tagged gene trap insertion in the JAM-A gene and immunofluorescence staining. Our results demonstrate that JAM-A expression initiates early in the development of most epithelia and is found in the early embryonic vasculature.
RESULTS AND DISCUSSION
JAM-A Gene Trap Insertion
Previously, a mutant in the JAM-A gene was identified in a screen of gene trap insertions enriched for genes that encode secreted or transmembrane proteins (Mitchell et al., 2001). Transcription of this locus results in the production of the first Ig domain of JAM-A fused to TMβ-geo, an artificial protein created from the transmembrane domain of rat CD4 joined to the catalytically active domains of β-galactosidase and neomycin transferase containing an internal transmembrane domain (Skarnes et al., 1995). Because the expression of this fusion protein is directed by the endogenous JAM-A promoter, mice were created from these embryonic stem (ES) cells and X-gal staining was used to determine the expression pattern of the JAM-A gene during early embryonic development.
Early Embryonic Expression of JAM-A
Tight junctions and other intercellular membrane junctions first form as an eight-cell embryo develops into the morula, the first point in embryogenesis where vertebrate cells develop different cell fates (Fleming et al., 2000). No JAM-A gene activity was detected in morula heterozygous for the JAM-A gene trap after 7 hours of staining (Fig. 1A); however, weak staining was observed after 20 hr (data not shown). In blastocysts, appreciable lacZ activity localized to the inner cell mass and trophectoderm (stained for 7 hr; Fig. 1B), indicating the expression of JAM-A. This finding correlates well to the previous observation that JAM-A mRNA and protein are present in both mouse and human preimplantation embryos (Ghassemifar et al., 2003; Thomas et al., 2004). Notably, the tight junction protein ZO-1 is also expressed at low levels in cleavage stage embryos and up-regulates as morula develop into blastocysts (Fleming et al., 1989). The formation of tight junctions in the trophectoderm are necessary to control the paracellular movement of fluid during the formation of the blastocoele (Watson et al., 2004), and JAM-A appears to participate in this process (Thomas et al., 2004).
At 8.5 dpc, the chorion (data not shown) and the ventral side of the embryo (Fig. 1C) express JAM-A. Sectioning through this embryo revealed that JAM-A gene activity was present in the endoderm, surface ectoderm, and the foregut diverticulum (Fig. 1D–F). In contrast, little JAM-A expression was detected in the mesoderm, neural plate/fold, heart, or allantois. By 9.5 dpc, JAM-A gene activity generally has decreased on the ventral side of the embryo (Fig. 1G), and by 12.5 dpc, JAM-A expression is highly tissue-specific (Figs. 1H, 2, 3B–F, 4, 5), while a wild-type embryo at the same age, containing no knock-in allele, did not stain for β-galactosidase activity (Fig. 1F). These localizations are described in greater detail below.
JAM-A Expression in Embryonic Blood Vessels
Blood vessel precursors first form during embryonic development when clusters of mesoderm form blood islands consisting of angioblasts surrounding a core of hemopoietic cells. These blood islands then fuse as the angioblasts differentiate into endothelial cells and a vessel lumen forms (Risau and Flamme, 1995). The JAM-A gene does not appear to be expressed in primary angioblasts or the early primary vessels, because the allantois of 8.5 dpc embryos did not stain with X-gal (Fig. 1C), although Tal1+/Flk1+ angioblasts and early platelet endothelial cell adhesion molecule-1 (PECAM-1/CD31)-positive vessels have been detected in this structure (Drake and Fleming, 2000). In contrast, JAM-A gene activity is detected by X-gal staining in vascular networks throughout 9.5–12.5 dpc embryos (Figs. 1G,H, 2A–C). Specifically, temporal JAM-A gene expression is evident at sites of intersomitic vasculature beginning at 9.5 dpc. Localization of JAM-A to cranial vasculature was confirmed in 11.5 dpc embryos by double staining for JAM-A (X-gal) and PECAM-1 (immunohistochemical; Fig. 2B,C), a known marker of blood vessel endothelial cells (Albelda et al., 1991; Ilan and Madri, 2003). Furthermore, we demonstrated by immunofluorescent staining that the endogenous JAM-A protein is also colocalized with PECAM-1 in vascular networks of 11.5 dpc embryos, validating the use of the gene trap to determine the developmental expression pattern of JAM-A (Fig. 2D–F). This finding is consistent with a recent study reporting the presence of JAM-A protein at tight junctions in adult human microvessels responsible for the formation of the blood–brain barrier (Vorbrodt and Dobrogowska, 2004).
The expression pattern of JAM-A suggests that it is a relatively late marker of vascular endothelial cells. It has been shown that JAM-A is localized to the tight junctions of vascular endothelial cells and its homotypic interactions are implicated in the regulation of tight junction integrity (Martin-Padura et al., 1998; Bazzoni et al., 2000). JAM-A redistributes from the junction to the cell periphery upon treatment with TNF-α, interferon-γ, and bFGF, which appear to alter leukocyte transmigration (Ozaki et al., 1999; Ostermann et al., 2002; Naik et al., 2003b). We previously demonstrated that function blocking JAM-A antibodies prevent bFGF-induced tube formation by human umbilical vein endothelial cells (HUVECs) and inhibit bFGF-induced angiogenesis in chick chorioallantoic membranes (Naik et al., 2003b). Furthermore, overexpression of JAM-A in HUVECs is sufficient to induce tube formation and knock-down of JAM-A expression using RNAi inhibited bFGF-induced cell migration, suggesting a central role for JAM-A in blood vessel development and function (Naik et al., 2003a).
JAM-A Is Expressed in the Developing Sensory Organs
While JAM-A is prominently expressed in blood vessels, intense JAM-A gene activity as detected by X-gal staining was also noted in several other locations, especially in the inner ear. The inner ear develops from a region of head ectoderm lateral to the neural tube denoted the otic placode (Riley and Phillips, 2003). While JAM-A gene activity was not seen in the otic placode at 8.5 dpc, it was prominently expressed in the epithelium of otic pit/vesicle at 9.5 dpc (Fig. 3A). As the otic vesicle elaborates into its mature structure, JAM-A gene activity is still detectable in the entire otocyst that will form the inner ear of 11.5–12.5 dpc embryos, including the endolymphatic duct, utricle, and cochlea (Fig. 3B,C). It is likely that JAM-A is participating in tight junction biology in the developing ear, because tight junctions are observed in the otic epithelium as early as the otocyst stage (Anniko and Bagger-Sjoback, 1982). Correct tight junction formation appears to be essential for normal hearing because mutations in the tight junction protein claudin 14 result in deafness in humans (Wilcox et al., 2001). It is believed that tight junctions function in the mature inner ear to compartmentalize the perilymph and endolymph and may be selective ion barriers (Kitajiri et al., 2004).
Similarly, the olfactory system develops from surface ectoderm overlying the developing forebrain denoted the nasal placode. JAM-A gene expression is seen during the formation of the nasal pits and olfactory epithelium at 11.5 dpc (Fig. 3D,E). This expression persists as the nasopharynx forms (Fig. 3F) and becomes continuous with JAM-A gene expression in the oral ectoderm, pharynx, and trachea at 12.5 dpc. Notably, the tight junction protein ZO-1 is also expressed in the olfactory epithelium at 11.5 dpc, although its expression appears to initiate in the nasal placodes by 9 dpc (Miragall et al., 1994), when JAM-A gene activity is minimal. JAM-A is not found in all sensory organs during their induction, because JAM-A gene activity is notably absent in the lens placode and developing eye through 12.5 dpc, although the blood vessels of the tunica vasculosa do express JAM-A (Kang and Duncan; unpublished data).
JAM-A Is Expressed During Branching Morphogenesis
The epithelial component of the lung bronchi initially forms as a bud from the endodermal component of the pharynx called the laryngotracheal groove. This groove bifurcates into the lung buds and then the final bronchial tree in response to the surrounding mesenchyme (Shannon and Hyatt, 2004). Whole-mount and section X-gal staining of 11.5 dpc mouse embryos reveals the presence of JAM-A gene activity in the lung buds and trachea (Fig. 4A,B). This expression is maintained in the bronchi as they continue to branch to form the primary bronchial tree at 13.5 dpc (Fig. 4C). Tight junctions are normally found in the bronchial epithelium, and abnormalities in the regulation of paracellular permeability have been implicated in human airway disease (Montefort et al., 1993; Godfrey et al., 1994). The presence of JAM-A in the bronchial epithelium suggests that studies on the modulation of JAM-A function in airway disease may be a fertile topic of future study.
The mesonephric kidney is a transient structure that forms when the pronephric duct enters the mesonephric mesenchyme and begins to branch (Dressler, 2002; Vainio and Lin, 2002). JAM-A gene activity is obvious in these branches at 11.5 dpc in both whole-mount embryos (Fig. 4D) and sections through this structure (Fig. 4E). As the mesonephric duct grows caudally and the uteric bud branches off to induce the metanephric kidney (Sakurai, 2003), JAM-A expression is still seen and is maintained as it enters the metanephric mesenchyme and begins to branch, forming the renal pelvis and connecting duct system (not shown). Furthermore, JAM-A gene activity is prominent in the developing glomeruli (Fig. 4F), which are formed from mesenchymal to epithelial transitions induced by the tips of the branching collecting system. It is notable that the further differentiation of the glomeruli will require the migration of endothelial cell precursors and the interdependent development of the podocytes and capillaries (Nikolova and Lammert, 2003). Because JAM-A is present on both the epithelial and endothelial components of the adult kidney (data not shown), it is tempting to speculate that the cross-talk required to develop a glomerulus is partially mediated by homotypic JAM-A interactions.
JAM-A Is Expressed Broadly in Developing Epithelia
Further sites of epithelial X-gal staining, in addition to the lungs and kidneys, include the developing skin and hair follicles, choroid plexuses, and gut (Fig. 5A,C,E). JAM-A gene activity is patchy in the surface ectoderm of developing mouse embryos at 8.5 dpc and is observed at 9.5 dpc in the head ectoderm (Fig. 3A) as the single-layered epithelium splits into the periderm and basal layer. Notably, this finding is coincident with the first morphologically visible tight junctions in the developing skin (M'Boneko and Merker, 1988). The expression of JAM-A in the epithelial component of the skin remains detectable as the skin further develops and is maintained in derivatives of the skin, including the vibrissae at 13.5 dpc as visualized by both X-gal staining (Fig. 5A) and anti–JAM-A immunofluorescent staining (Fig. 5B).
In the developing brain, the only notable sites of JAM-A expression besides the blood vessels are in the choroid plexuses (Fig. 5C,D shown by X-gal and immunofluorescent staining, respectively). These structures are composed of epithelia connected by tight junctions lined by blood vessels and are responsible for the regulated secretion of cerebrospinal fluid (CSF; Sturrock, 1979). JAM-A gene activity is first seen by β-gal staining at 12.5 dpc in the epithelial component of the newly forming choroid plexus (data not shown). JAM-A expression continues at 13.5 dpc as detected by X-gal staining (Fig. 5C) and anti–JAM-A staining (Fig. 5D). Tight junctions are observable both morphologically and functionally in the mammalian choroid plexus from its earliest development, correlating with the distribution of JAM-A expression, although the regulation of CSF composition is quite different in the embryonic and adult brain (Dziegielewska et al., 2001).
The presence of JAM-A as detected by X-gal staining on the ventral surface of the unturned embryo (Fig. 1C) and in the foregut diverticulum (Fig. 1E) suggests that it is present in the endodermal precursors to the developing mid-gut (Tam et al., 2003). By 9.5 dpc, JAM-A is expressed in the epithelial lining of both the ectodermal component of the oral cavity, including Rathke's pouch (data not shown) as well as the endodermally derived foregut (Kaufman and Bard, 1999), including the nasopharynx, esophagus, and trachea (Figs. 1G, 2B, 3A,F, 4A), and to a lesser extent the liver (Fig. 4F) and pancreas (data not shown). JAM-A gene activity is maintained in the endodermally derived gut epithelium at 12.5 (Fig. 1H) and 13.5 dpc (Fig. 5E,F, as shown by X-gal and immunofluorescent staining, respectively) again showing a relationship between JAM-A expression and the presence of tight junctions in epithelia (Madara et al., 1981; Hirano and Kataoka, 1986).
In summary, the JAM-A gene is active in the blood vessel endothelium and the epithelia of the embryo derived from all three primary germ layers. The expression of JAM-A in these tissues correlates well with the presence of tight junctions, supporting further investigations into the function of this protein.
All experiments using animals were approved by the University of Delaware Institutional Animal Care Committee. Heterozygous mice harboring the JAM-A β-geo-tagged allele were generated from the KST235 ES cell line created by an insertion of the pGT1pfs gene trap vector (Mitchell et al., 2001). The KST235 sequence tag and ES cell line are available from BayGenomics (http://baygenomics.ucsf.edu). Embryonic mice were staged by designating noon of the day on which a semen plug was observed in the dam as 0.5 dpc. All mice were maintained in a 12-hr light/dark cycle at 21–24°C and were given food and water ad libitum.
Histochemical Staining of Preimplantation Mouse Embryos for β-Galactosidase Activity
Wild-type female mice were injected with 5 IU of pregnant mare serum gonadotropin (National Hormone and Peptide Program, Torrance, CA) followed 46 hr later by 5 IU of human chorionic gonadotropin (Sigma-Aldrich Co. St. Louis, MO) then mated with males homozygous for the JAM-A gene trap allele. Seventy-two hours later, embryos at the eight-cell to early morula stage were collected and cultured in vitro for 24–48 hr under standard conditions (Hogan et al., 1994).
Preimplantation embryos were washed three times for 7 min each at room temperature in phosphate-buffered saline (PBS) for tissue culture (Invitrogen). Embryos were fixed for 30 min at 4°C in PBS containing 1% paraformaldehyde, 0.02% glutaraldehyde, and 5 mM ethyleneglycoltetraacetic acid (EGTA) and washed again in PBS three times for 7 min each. The embryos were transferred to staining solution (PBS containing 0.1% X-gal (5-bromo-4-chloro-3-indoyl-β-D-galactopyranoside), 5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6, 2 mM MgCl2, 0.02% Nonidet P-40, and 5 mM EGTA) and incubated at 30°C in the dark for the times noted in the text. Preimplantation embryos were then washed three times for 7 min each at room temperature in 3% dimethylformamide (DMF) in 1× PBS before being photographed on a light microscope.
Histochemical Staining of Postimplantation Embryos for β-Galactosidase Activity
Embryos from 8.5 until 11.5 dpc were isolated and whole embryos incubated at 4°C for 1 hr in fixation solution (PBS containing 1% paraformaldehyde, 0.2% glutaraldehyde, 0.02% NP-40, and 0.01% sodium deoxycholate). Specimens were then washed three times (7–10 min each) in PBS, and stained for β-galactosidase activity at 30°C overnight in an X-gal solution (50 mM K3Fe(CN)6, 50 mM K4Fe(CN)6, 0.02% NP-40, 0.01% sodium deoxycholate, 2 mM MgCl2, and 1 mg/ml X-gal [from 40 mg/ml stock in DMF] in PBS). Specimens were washed three times (7–10 min each) with PBS containing 3% DMF and then post-fixed in 4% paraformaldehyde in PBS for 30 min at 4°C. Finally, specimens were rinsed once in PBS with 3% DMF and stored in the dark at 4°C in the same solution. Staining was visualized either by clearing the embryos as described below or by embedding them in OCT (Tissue Tek), sectioning at 20 μm on a cryostat, mounting the sections on slides and cover-slipping with Gel mount (Biomedia, Foster City, CA) mounting media.
Methyl salicylate was used to clear X-gal stained whole embryos for photodocumentation as described at (http://www.paperglyphs.com/wmc/docs/lacZ_bible.html). Briefly, embryos were washed with distilled water twice for 30 min at room temperature and subsequently dehydrated with agitation at room temperature as follows: 30 min in 70% ethanol, 30 min in 95% ethanol, twice in 100% ethanol for 30 min each. Embryos were then transferred to 100% methyl salicylate and agitated at room temperature until the tissues cleared (approximately 15 min). Cleared embryos were photographed through a dissecting microscope within 1 hr of clearing.
Embryos at 12.5 and 13.5 dpc were dissected, embedded immediately in OCT, sectioned at 20 μm on a cryostat, incubated at 4°C for 1 hr in fixation solution (PBS containing 1% paraformaldehyde, 0.2% glutaraldehyde, 0.02% NP-40, and 0.01% sodium deoxycholate), stained for β-galactosidase activity, and cover-slipped as above.
Enzyme immunohistochemical staining for PECAM-1 was performed on X-gal–stained sections as described (Wakayama et al., 2003). Briefly, sections generated from whole-mount X-gal staining of 11.5 dpc embryos as described above, were treated with Triton X-100 in PBS for 1 hr, washed with PBS, and then immersed in 3% H2O2 in methanol for 10 min. Sections were then incubated with 3% bovine serum albumin (BSA) in PBS for 30 min and then incubated overnight at 4°C with a PECAM antibody (rat anti-mouse PECAM-1/CD31 antibody, catalog no. 550274, BD-Pharmingen) diluted to 1 μg/ml in PBS. This step was followed by incubation with biotinylated anti-rat IgG antibody for 1 hr at room temperature. Sections were then treated with horseradish peroxidase–labeled streptavidin for 1 hr at room temperature and incubated with 3′,3′-diaminobenzidine tetrahydrochloride until a light brown color was observed.
The immunofluorescence procedures were taken from Reed et al. (2001). Briefly, unfixed cryosections were fixed at −20°C for 20 min in 1:1 acetone:methanol. The sections were then blocked in 1% BSA in PBS at 4°C for 1 hr followed by incubation with goat polyclonal anti-mouse JAM-A antibody (1:100 dilution, catalog no. AF1077, R&D systems) at room temperature for 1 hr. Sections were then washed 2–3 times in 1× PBS for 5–10 min each and incubated in the dark for 1 hr at room temperature with Alexa Fluor 568 donkey anti-goat IgG antibody (1:50 dilution, Molecular Probes, Eugene, OR) and the nuclear counterstain TO-PRO-3 (1:3,000, Molecular Probes). Double labeling for JAM-A and PECAM-1 expression was performed similarly, except that rat anti-mouse PECAM-1/CD31 antibody (1:15.6 dilution, BD-Pharmingen) was added to the primary antibody solution and Alexa Fluor 488–conjugated anti-rat IgG (Molecular Probes) was added to the secondary antibody solution. Slides were stored at −20°C until photographed on a Zeiss LSM 510 confocal microscope configured with an argon/krypton laser (488-nm and 568-nm excitation lines) and helium laser (633-nm excitation line; Carl Zeiss, Inc., Göttingen, Germany).
We thank Richard Focht of the University of Delaware Transgenic and Knockout Facility for assistance with superovulation and blastocyst manipulation, Dr. Kirk Czymmek of the University of Delaware Core Microscopy Facility for confocal microscopy support, and Dr. Kevin Mitchell for characterization of the KST235 cell line. PMS Gonadotropin for superovulation was obtained through the National Hormone and Peptide Program funded by NIDDKD, Dr. A.F. Parlow, Director. M.K.D. and U.P.N. were funded by the National Institutes of Health and the University of Delaware Core Imaging facility was supported by an INBRE program grant. J.P.P. was supported by the Howard Hughes Medical Institute.