Cell Biology Program, Sloan-Kettering Institute, New York, New York
Arthritis and Tissue Degeneration Program and Cell Biology, Program, Caspary Research Building, Room 426, Hospital for Special Surgery, Weill Medical College of Cornell University, 535 E. 70th St., New York, NY 10021
ADAMs (adisintegrin and metalloprotease) are a family of membrane-anchored glycoproteins that have critical roles in a variety of different processes, including sperm–egg interactions, neurogenesis, neovascularization, and heart development (Black and White, 1998; Schlöndorff and Blobel, 1999; Primakoff and Myles, 2000; Horiuchi et al., 2003; Jackson et al., 2003; Seals and Courtneidge, 2003; Kurohara et al., 2004; Zhou et al., 2004). A typical ADAM is composed of several conserved and characteristic protein modules, including a prodomain, metalloprotease domain, disintegrin domain, cysteine-rich region, epidermal growth factor (EGF) repeat, transmembrane domain, and cytoplasmic tail. Only approximately half of the known ADAMs contain the catalytic site consensus sequence for metalloproteases (HEXXHXXGXXHD; Stocker et al., 1995; Sahin et al., 2004), and ADAMs lacking this site most likely do not possess metalloprotease activity. Several ADAMs that contain a catalytic site have been shown to participate in the release of cytokines, growth factors and receptors from the cell membrane, a process that is referred to as protein ectodomain shedding (Hooper et al., 1997; Schlöndorff and Blobel, 1999; Hooper and Turner, 2000). In addition, the disintegrin domain and cysteine-rich region of certain ADAMs has been implicated in cell–cell and cell–matrix interactions (White, 2003). Finally, ADAMs frequently contain potential signaling motifs such as SH3-ligand domains in their cytoplasmic domain. These are predicted to play a role in intracellular signaling and/or regulation of ADAM activity (Schlöndorff and Blobel, 1999; Seals and Courtneidge, 2003).
To date, the biological roles of only a few catalytically active ADAMs have been elucidated. ADAM10 (also referred to as kuzbanian or KUZ), first identified in Drosophila, has a role in axonal guidance and Notch signaling (Fambrough et al., 1996; Rooke et al., 1996; Pan and Rubin, 1997; Sotillos et al., 1997; Hartmann et al., 2002; Lieber et al., 2002). ADAM17/TACE (tumor necrosis factor-α [TNF-α] converting enzyme) was identified by its ability to cleave TNF-α (Black et al., 1997; Moss et al., 1997) and has since been shown to have a critical role in shedding a variety of other substrates from cells, including several ligands of the EGF receptor (Peschon et al., 1998; Merlos-Suarez et al., 2001; Sunnarborg et al., 2002; Sahin et al., 2004), fractalkine (Garton et al., 2001), amyloid precursor protein (Buxbaum et al., 1998), p75 neurotrophin-receptor (Weskamp et al., 2004), and MUC1 (Thathiah et al., 2003). ADAM15 has a role in pathological neovascularization in mice (Horiuchi et al., 2003), whereas ADAM19 is essential for cardiovascular morphogenesis (Kurohara et al., 2004; Zhou et al., 2004). Finally, mutations in the ADAM33 gene have been linked to asthma and bronchial hyperresponsiveness in humans (Van Eerdewegh et al., 2002).
ADAM8 (also referred to as MS2 or CD156a) is a catalytic site-containing ADAM that was initially identified in mouse macrophages and macrophage cell lines by using a differential display approach and that is up-regulated in response to macrophage stimulators (Yoshida et al., 1990). In adult mice, ADAM8 expression has also been observed in the central nervous system, in neurons and oligodendrocytes (Schlomann et al., 2000). In addition, ADAM8 is expressed in human immune cells with the exception of T-cells (Yoshiyama et al., 1997). Promoter studies have identified LPS (lipopolysaccharide), interferon-γ, interleukin (IL) -6, and TNF-α response elements in the 5′ region of the ADAM8 gene (Kataoka et al., 1997; Schlomann et al., 2000). A biochemical characterization of recombinantly expressed soluble ADAM8 confirmed that it possesses catalytic activity and demonstrated that ADAM8 is activated through autocatalytic removal of its inhibitory pro-domain (Amour et al., 2002; Schlomann et al., 2002). By comparison, most other catalytically active ADAMs require pro-protein convertases such as furin to remove their prodomain (Clarke et al., 1998; Loechel et al., 1998; Lum et al., 1998; Roghani et al., 1999; Schlöndorff et al., 2000). Recombinantly expressed and purified soluble ADAM8 can cleave myelin basic protein (MBP), a peptide representing the membrane-proximal region of IL-1 receptor type II (IL1-RII), and CD23, a low affinity IgE receptor (Amour et al., 2002; Schlomann et al., 2002; Fourie et al., 2003). The tissue inhibitor of matrix metalloproteinases (TIMP) -inhibitor profile of ADAM8 is like that of ADAM9 in that it is not inhibited by any of the known TIMP proteins (Amour et al., 2002; Schlomann et al., 2002). Finally, the cytoplasmic domain of ADAM8 also contains putative proline rich SH3 binding sequences, suggesting it may interact with cytoplasmic proteins (Yoshida et al., 1990; Yoshiyama et al., 1997).
After the identification of ADAM8 in macrophages, a transgenic mouse overexpressing its soluble ectodomain was generated (Higuchi et al., 2002). These animals had attenuated leukocyte infiltration and down-regulated expression of L-selectin. Experiments in Wobbler mice, which have an accelerated course of neurodegeneration, showed an increase in ADAM8 expression in activated glial cells (astrocytes and activated microglia), suggesting that ADAM8 has a role in pathological neuron–glia interactions (Schlomann et al., 2000). ADAM8 expression was enhanced during differentiation of an immortalized murine osteoclast cell line, and reduction of ADAM8 expression by antisense oligonucleotides inhibited osteoclast formation in a vitamin D treated mouse bone marrow culture (Choi et al., 2001). Expression of ADAM8 is increased after activation of the peroxisome proliferator-activated receptor γ in a macrophage cell line, raising the possibility that ADAM8 might have a role in the pathology of atherosclerosis (Hodgkinson and Ye, 2003). ADAM8 is also up-regulated in tissue surrounding loosened hip prosthesis (Mandelin et al., 2003). Finally, ADAM8 reportedly can function as an adjuvant in the administration of vaccines against autoimmune diseases, although the mechanism underlying this effect remains to be established (Schluesener, 1998). The goal of this study was to learn more about the role of ADAM8 in mice by determining its expression pattern during mouse embryogenesis and by evaluating how a targeted deletion of ADAM8 affects mouse development and adult animals.
RESULTS AND DISCUSSION
Analysis of ADAM8 Expression During Mouse Development
To identify the tissues and cells in which ADAM8 may function during development, we evaluated its expression by mRNA in situ hybridization at different embryonic stages. At E8.5, ADAM8 expression was the most abundant in the giant trophoblast layer surrounding the embryo (Fig. 1A–C) and in the decidua basalis (mesometrial decidua, Fig. 1D–F). In the decidua, numerous ADAM8-positive cells were found between the diploid decidual cells around the uterine lumen, and strong expression was also detected in scattered cells around blood lacunae (Fig. 1F).
The expression of ADAM8 was continuous toward and within the developing placenta. In addition to the prominent signal in the invading trophoblast giant cells (Fig. 1G,H, arrows), a weaker signal was detected in the chorionic plate of the fetal placenta (Fig. 1G–I, arrowheads), whereas the allantois showed little or no expression of ADAM8 (Fig. 1I, asterisk). Similarly, the tissues of the embryo proper did not exhibit detectable expression of ADAM8. Therefore, at this stage of development, the expression of ADAM8 mRNA was confined predominantly to maternal cells in the decidua basalis. The only embryo-derived cells expressing ADAM8 were of trophectodermal origin (giant trophoblast cells and the fetal placenta).
At E11.5, ADAM8 expression was prominent in the developing gonadal ridge (Fig. 2A,B), whereas lower levels of ADAM8 expression were seen in the spinal cord (Fig. 2C,D). However, the expression in the central nervous system and gonads was transient and was no longer evident at later stages of development (after E14.5, data not shown).
At E14.5, the expression pattern of ADAM8 was considerably different from that described above. Expression of ADAM8 was observed adjacent to the cartilage primordia of various structures, including the mandible (Fig. 3A,B), the basi-occipital bone, (Fig. 3C,D), the ribs (Fig. 3E,F), the upper palate (Fig. 3G,H), as well as the femur, tibia, and fibula (data not shown). Pronounced expression of ADAM8 was found in the epithelium of the nasal sinuses (Fig. 3I,J). The expression of ADAM8 was also seen in the developing thymus (Fig. 4A,B) and in an area of the liver, where major venous vessels such as hepatic veins, ductus venosus, and vitelline veins connect to the inferior vena cava (Fig. 4C,D, arrowheads). In addition, patchy expression was found in single cells scattered throughout the liver (Fig. 4C,D), and along tissue surrounding the major umbilical vessels (Fig. 4C,D, arrows). A structure with nest-like expression of ADAM close to the thymus may correspond to a developing lymph node (Fig. 4E–G, arrowheads). In some sections, expression of ADAM8 was seen in tissues adjacent to major vessels (Fig. 4H–M, arrows). Of interest, where this staining pattern was observed, expression of ADAM8 did not completely surround the vessels but was instead restricted to one side of a vessel, or to an area between two vessels (Fig. 4H–M). In transverse sections, ADAM8 expression was most prominent in an area between the internal jugular vein and the jugular lymph sac (Fig. 5A,B, arrows), as well as surrounding one side of the subclavian vein (Fig. 5E,F, arrows). The expression of ADAM8 in these structures is adjacent to that of Prox1, a marker for early lymphatic endothelial cell specification (Fig. 5C,G, arrowheads). Expression of Prox1 begins at E9.5 and is restricted to the endothelial cells on one side of the vessels (Oliver, 2004). Prox1 has been shown to have a critical role in committing a subpopulation of competent venous endothelial cells to becoming lymphatic endothelial cells. It remains to be determined whether the asymmetric expression of Prox1 is established in response to an intrinsic signaling mechanism of endothelial cells or through a localized inductive signal coming from mesenchymal cells. Therefore, the expression pattern of ADAM8 adjacent to Prox1-expressing cells raises the possibility that ADAM8 has a role in the differentiation of lymphatic endothelial cells.
Generation of Mice Null for the ADAM8 Allele
To address the role of ADAM8 in development and adult homeostasis, we generated mice with a targeted mutation in the ADAM8 gene (see Experimental Procedures section for details). Briefly, a gene cassette containing an internal ribosomal entry site (IRES) sequence upstream of a β-galactosidase reporter gene was used to replace the catalytic and disintegrin domains of the ADAM8 gene (Fig. 6A). The targeted deletion of ADAM8 in genomic DNA was detected by Southern blot analysis (Fig. 6B). Matings between heterozygous Adam8+/− parents produced offspring with a Mendelian distribution of the targeted ADAM8 allele (Fig. 6C). Western blot analysis using concanavalin A–enriched glycoprotein extracts confirmed that expression of ADAM8 protein was indeed abolished in Adam8−/− mice, whereas the ADAM8 protein could be readily detected in extracts from wild-type and Adam8+/− animals (Fig. 6D). During routine handling, adult Adam8−/− mice were indistinguishable in appearance and behavior from their wild-type and heterozygous littermates. Male and female Adam8−/− mice were viable and fertile, producing litters at a frequency and size comparable to that of wild-type mating pairs, and the postnatal growth curve of Adam8−/− mice was identical to that of wild-type controls (data not shown). In addition, a histopathological examination of age-matched 3-month-old ADAM8 null and wild-type males and females did not uncover any apparent morphological or pathological abnormalities in Adam8−/−mice.
A histopathological analysis of Adam8−/− embryos at stages embryonic day (E) 8.5, E11.5, and E14.5 revealed no obvious abnormalities during development compared with wild-type littermate controls (data not shown). A close examination of structures where ADAM8 expression is high during development, specifically the trophoblast and decidua at E8.5, the gonadal ridge and neural tube at E11.5, and the thymus and developing cartilage at E14.5 also did not bring to light any evident histopathological defects (data not shown). To assess potential defects in development of lymphatic vessels, sections of an E14.5 Adam8−/− embryos were stained with antibodies against Prox1 to examine the junction between the jugular vein and the lymphatic duct. However, the pattern of Prox1 staining in Adam8−/− mice (Fig. 5D,H, arrowheads) was comparable to that seen in wild-type embryos (Fig. 5C,G, arrowheads), suggesting that ADAM8 is not required for Prox1 expression or for lymphatic vessel formation. Future studies will be necessary to compare the temporal and spatial relationship of Prox1 and ADAM8 expression during lymphatic development and to further explore any potential contribution of ADAM8 to this process.
ADAM8 has been implicated previously in osteoclast differentiation (Choi et al., 2001). Therefore, we compared bone sections of young (2-week-old) and adult (10-week-old) Adam8−/− mice with age-matched wild-type animals (Fig. 7A–D). Impairments in osteoclastogenesis can hamper bone turnover and, therefore, could cause defects in bone structure. However, no apparent defects were observed in bone sections from Adam8−/− mice compared with wild-type controls upon light microscopic examination (Fig. 7A–D). Specifically, there were no evident deviations in the morphology and proportion of the different regions of the growth plate or in the histological appearance of trabecular bone and thickness of cortical bone in Adam8−/− mice. Furthermore, a comparison of whole-mount skeletal preparations of newborn Adam8−/− mice and wild-type littermate controls stained with alizarin red and Alcian blue did not reveal any macroscopic defects in bone development in Adam8−/− mice at this stage (Fig. 7E). Thus, the lack of ADAM8 does not result in major defects in bone development. Further studies will be necessary to evaluate a possible involvement of ADAM8 in osteoclast function under pathological conditions, such as menopausal osteoporosis or bone metastasis. Finally, there were no significant changes in the differential blood counts between wild-type and Adam8−/− mice, indicating that there was no severe deficiency in immune cell numbers.
The targeted allele of ADAM8 contains a lacZ gene with an internal ribosomal entry site, providing an opportunity to extend the expression analysis of ADAM8 through X-gal staining of Adam8+/− and Adam8−/− mice (Mountford et al., 1994). Strong X-gal staining was seen in the decidua basalis in sections of E8.5 Adam8+/− embryos (data not shown), which is consistent with the results of ADAM8 mRNA in situ hybridizations shown in Figure 1D,G. However, no prominent X-gal staining was observed in other locations where ADAM8 expression was detected by mRNA in situ hybridization at E8.5 and E14.5, as shown in Figures 1–5 (results of X-gal staining not shown). A potential explanation for this finding is that there might be differences in the stability of mRNA derived from the targeted allele that contains the lacZ sequence compared with wild-type ADAM8 mRNA in some cells and tissues, or that the stability of the lacZ protein might be different from that of the wild-type ADAM8 mRNA. Nevertheless, X-gal staining was also used to determine where ADAM8 might be expressed in adult Adam8+/− mice. The strongest X-gal staining was seen in epithelial cells in the lung (Fig. 8A,B), salivary gland (Fig. 8C–E), and kidney (Fig. 8F–H). However, when the X-gal staining patterns of Adam8−/− and Adam8+/− mice were compared, no differences were detected at E8.5, E14.5, or in adults, arguing against essential functions of ADAM8 in the development and survival of the X-gal stained cells in these tissues. Thus, despite the quite restricted and specific expression pattern of ADAM8 during development and in adults, a targeted inactivation of ADAM8 in mice did not lead to any apparent defects in development or adult survival or fertility of Adam8−/− mice and did not cause any evident pathological phenotypes.
In summary, analysis of ADAM8 expression by in situ hybridization revealed high expression in the extraembryonic and decidual tissues at early stages and expression in the developing gonads, thymus, cartilage primordia, and in cells adjacent to developing bone and cartilage as well as lymphatic vessels at later stages of mouse development. Furthermore, a X-gal–staining provided evidence for expression of ADAM8 in bronchial epithelium, salivary gland, and renal epithelial cells in adult mice. Nevertheless, a targeted deletion of ADAM8 showed that this protein is not essential for mouse development or adult survival. Further studies, including crosses of Adam8−/− mice with other Adam−/− mice, and more detailed functional studies of the respiratory and immune system as well as bone homeostasis in Adam8−/− mice may help uncover physiologically relevant functions of ADAM8.
Chemicals and reagents were purchased from Sigma (St. Louis, MO). Restriction enzymes and reagents for molecular biology were purchased from Roche Diagnostics (Mannheim, Germany) and New England Biolabs (Beverly, MA).
Generation of Adam8−/− Mice
To generate a mutation in the mouse ADAM8 gene, the coding sequence for the catalytic and disintegrin domains was replaced with a reporter selection cassette (Nehls et al., 1996) containing the β-galactosidase coding sequence linked to an upstream IRES and a downstream neomycin selection marker expressed from its own MC1 promoter (Thomas et al., 1986). The targeting vector consisted of a 5.1-kb DNA fragment containing the reporter selection cassette flanked on one side by 3.6 kb of genomic sequence and on the other by 3.8 kb of genomic sequence to create the homology arms. An HSV thymidine kinase gene cassette was incorporated into the end of the 3.8-kb homology arm for use as a negative selection marker against random integration (Mansour et al., 1988). The linearized construct was electroporated into E14Tg2aIV embryonic stem (ES) cells, and cells were selected in G418 and ganciclovir (Mansour et al., 1988). ES cells carrying a targeted allele were identified after digestion of their DNA by KpnI, SacI, and EcoRI and Southern blot analysis by using DNA probes specific for regions flanking the ends of both targeting vector homology arms, as well as for the neomycin selection marker. Correctly targeted ES cells were injected into C57BL/6J blastocysts to create chimeric offspring. Male mice with a high degree of chimerism (as evidenced by a high percentage of the agouti coat color) were mated with C57/BL/6 females to generate mice carrying the targeted ADAM8 allele in their germline. Tail biopsy DNAs were analyzed by Southern blotting as described above, and heterozygous pups were identified from this analysis. Heterozygous adults of mixed genetic background (129/Ola;C57/BL/6) were intercrossed to generate homozygous Adam8−/− mice.
Western Blot Analysis
Freshly prepared lungs from wild-type mice, or Adam8+/− or Adam8−/− mice were homogenized in RIPA buffer (phosphate buffered saline [PBS], 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate [SDS]) containing CompleteTM ethylenediaminetetraacetic acid–free inhibitor mixture (Roche Molecular Biochemicals, Mannheim, Germany) and 10 mM 1,10-ortho phenanthroline. After lysis on ice for 1 hr, samples were spun for 5 min at 13,000 × g at 4°C. The supernatants were incubated with ConA-Sepharose beads (Amersham Biosciences, Braunschweig, Germany) for 1 hr. The ConA-Sepharose beads were then washed twice with 20 mM Tris-HCl (pH 7.4), 0.5 M NaCl, 1 mM MnCl2, 1 mM CaCl2, and bound glycoproteins were eluted with 1× SDS-polyacrylamide gel electrophoresis (PAGE) loading buffer (for reducing conditions: 60 mM Tris-HCl, pH 6.8, 2% SDS, 9% glycerol, 0.0025% bromphenol blue, and 0.36 M mercaptoethanol). Samples were run on 5% SDS-PAGE gels and blotted onto Nylon membranes (Biodyne B, PALL, Portsmouth, UK) by electroblotting. Membranes were blocked overnight with 5% reconstituted milk from powder, then incubated with antibodies against the cytoplasmic domain of ADAM8 (Schlomann et al., 2000), washed, and then incubated with horse radish peroxidase (HRP) -tagged anti-rabbit IgG (1:4,000). The blots was then incubated with LumiLight Plus (Roche) as chemiluminescent substrate for the bound HRP-tagged antibody, and used to expose Kodak XAR autoradiography film.
In Situ Hybridization
To analyze the expression of ADAM8 during mouse development by mRNA in situ hybridization, embryos of specific gestation stages (E8.5, E11.5, and E14.5) were generated by timed matings. Embryos were fixed overnight in 4% paraformaldehyde at 4°C, then washed twice with PBS, dehydrated through a graded series of ethanol washes, cleared with Histoclear, and then embedded in paraffin. Sagittal sections (prepared at a thickness of 8 μm), and transverse sections (prepared at a thickness of 10 μm) were mounted on Fisher Superfrost Plus slides. Single-stranded cRNA probes were labeled with 33P using T3/T7 RNA polymerases and a ribonucleoside triphosphate mixture containing 12 μM cold UTP and 4 μM [33P]UTP. The antisense and sense probes were both prepared from the first 700 base pairs of the 5′ ADAM8 cDNA sequence (Schlomann et al., 2000). The mRNA in situ hybridization procedure was performed as previously described (Manova et al., 1990; Weskamp et al., 2002). Slides were exposed for 10–14 days and developed in Kodak developer D-19, fixed in Kodak fixer, and counterstained with hematoxylin and eosin.
In Situ lacZ Staining
Cryosections were fixed in 4% glutaraldehyde, and staining for β-galactosidase was performed overnight in a solution of 4% (v/v) 5-bromo-chloro-indolyl-β-D-galactopyranoside (X-Gal, B 4252, Sigma), 1 mM MgCl2, 150 mM sodium chloride, 3.3 mM potassium ferrocyanide (K4 Fe(CN)6 3H2O), 3.3 mM potassium ferricyanide (K3 Fe(CN)6) in 10 mM phosphate buffer (pH 7.0) at 37°C. Subsequently, sections were washed in 50 mM phosphate buffer (pH 7.0), counterstained with eosin (0.5%), and embedded in Entellan (Merck, Darmstadt, Germany).
Histopathology and Clinical Pathology
Timed matings were set up to obtain embryos at E8.5, E11.5, and E14.5 days of development for histopathological analysis of potential defects in developing Adam8−/− mice. Embryos were fixed, washed, dehydrated, and embedded in paraffin as described above; cut into sections of 8μm; and mounted on Fisher Superfrost Plus slides. The slides were kept at 42°C overnight and then left at 56°C for 1 hr to heat the paraffin, dewaxed by using histoclear, and then rehydrated through a graded series of ethanol dilutions. The rehydrated sections were stained with hematoxylin and eosin, then dehydrated and mounted by using 50% Permount in Histoclear. A histopathological and hematological analysis of 3-month-old mice (two female and two male Adam8−/− mice and two female and two male wild-type littermate controls) was performed by the Transgenic Mouse Facility at Weill Cornell Hospital. No major histopathological defects or differences in the differential blood count were observed in Adam8−/− mice compared with wild-type controls.
Two-week-old and 10-week-old Adam8−/− male mice and the corresponding wild-type control were used for bone morphological analysis. The dissected lower limbs were fixed with 4% paraformaldehyde in PBS and decalcified using Cal-Ex Decalcifier (Fisher Scientific) following the manufacturer's instructions. The specimens were embedded in paraffin wax, and the sections were cut at 6μm thickness. After deparaffinization, the sections were subjected to consecutive staining with hematoxylin, 0.02% Fast green and 0.2% Safranin-O. To prepare a whole-mount skeletal preparation, eviscerated newborn mice were fixed in 95% ethanol overnight, and then stained overnight with 0.05% Alcian blue 8GX in 95% ethanol and 5% acetic acid. The embryos were then rinsed and incubated with 95% ethanol overnight, and subsequently placed in 2% KOH until the bones were clearly visible. The whole-mount preparation was then stained with 0.1% alizarin red in 1% KOH overnight and rinsed in 1% KOH, 20% glycerol for 2 days. Specimens were transferred into a 1:1 mixture of 95% ethanol and glycerol for storage.
All images were acquired with a Zeiss Axiocam HRC camera mounted on a Zeiss Axioplan2 microscope, using Zeiss axiovision software. Images were processed with Adobe Photoshop 7.0 software.
We thank Katia Bojilova and Thadeous Kacmarczyk for excellent technical assistance, Sarah Lane for animal care, and Mike McAndrew for undertaking the PCR analysis at Celltech. C.P.B. was funded by the National Institutes of Health National, and J.W.B. was supported by DFG. The Gene Targeting Laboratory at the Institute for Stem Cell Research was supported by the UK Biotechnology and Biological Sciences Research Council.