|Home | About | Journals | Submit | Contact Us | Français|
The metalloproteinase and major amyloid precursor protein (APP) α-secretase candidate ADAM10 is responsible for the shedding of proteins important for brain development, such as cadherins, ephrins, and Notch receptors. Adam10 −/− mice die at embryonic day 9.5, due to major defects in development of somites and vasculogenesis. To investigate the function of ADAM10 in brain, we generated Adam10 conditional knock-out (cKO) mice using a Nestin-Cre promotor, limiting ADAM10 inactivation to neural progenitor cells (NPCs) and NPC-derived neurons and glial cells. The cKO mice die perinatally with a disrupted neocortex and a severely reduced ganglionic eminence, due to precocious neuronal differentiation resulting in an early depletion of progenitor cells. Premature neuronal differentiation is associated with aberrant neuronal migration and a disorganized laminar architecture in the neocortex. Neurospheres derived from Adam10 cKO mice have a disrupted sphere organization and segregated more neurons at the expense of astrocytes. We found that Notch-1 processing was affected, leading to downregulation of several Notch-regulated genes in Adam10 cKO brains, in accordance with the central role of ADAM10 in this signaling pathway and explaining the neurogenic phenotype. Finally, we found that α-secretase-mediated processing of APP was largely reduced in these neurons, demonstrating that ADAM10 represents the most important APP α-secretase in brain. Our study reveals that ADAM10 plays a central role in the developing brain by controlling mainly Notch-dependent pathways but likely also by reducing surface shedding of other neuronal membrane proteins including APP.
ADAM10 is a member of the large family of ADAMs (a disintegrin and metalloproteases) and is expressed in oligodendrocytes and a subset of developing neurons and brain nuclei and in differentiating gray matter (Lin et al., 2008). It is thought to be responsible for the shedding of several cell surface proteins in the brain, such as ephrins (Hattori et al., 2000; Janes et al., 2005), amyloid precursor protein (APP) (Hooper and Turner, 2002; Asai et al., 2003; Postina et al., 2004), prion protein (Vincent, 2004), and axon guidance molecules such as neuronal adhesion molecule (Hinkle et al., 2006), L1 adhesion molecule (Gutwein et al., 2003), and N-cadherin (Reiss et al., 2005). It also controls the proteolytic processing of the Notch receptor (Pan and Rubin, 1997; Wen et al., 1997; Hartmann et al., 2002; Lieber et al., 2002), which mediates lateral inhibition during neurogenesis. Adam10 knock-out mice die at embryonic day 9.5 (E9.5) due to major developmental defects, including disturbed somitogenesis, a shortened forebrain primordium, and severe vascular defects (Hartmann et al., 2002). This early lethality has precluded further investigation of ADAM10 function in the brain in vivo.
The signals transduced by the Notch receptors are crucial in cell fate decisions, both during embryonic development and in adult tissue homeostasis (Gordon et al., 2008). Signaling of the Notch receptor is dependent on three types of proteolytic events (Mumm and Kopan, 2000). After the first cleavage the heterodimeric receptor is able to interact with Notch ligands allowing a second (S2) cleavage within the Notch extracellular juxtamembrane region by ADAM proteases. The S2 cleavage is necessary for subsequent γ-secretase-mediated proteolysis within the transmembrane region. This allows the Notch intracellular domain (NICD) to translocate into the nucleus, thereby directly activating gene expression. Both ADAM10 and ADAM17 have been implicated in this S2 cleavage. Recent in vitro studies suggest that ADAM10 is absolutely required for ligand-induced Notch-1 signaling (van Tetering et al., 2009), while ligand-independent Notch-1 signaling requires ADAM17 (Bozkulak and Weinmaster, 2009).
APP ectodomain shedding is followed by further cleavage of the transmembrane fragment by the γ-secretase complex releasing an APP intracellular domain (AICD). The nature of the metalloprotease responsible for the physiologic processing of APP is not completely clear. Overexpression studies and in vitro cleavage assays indicate that ADAM9, ADAM10, and ADAM17 are able to act as α-secretases (Buxbaum et al., 1998; Postina, 2008). Overexpression of ADAM10 in an Alzheimer's mouse model resulted in a decrease in β-amyloid deposition and cognitive deficits (Postina et al., 2004) but since transgenic overexpression of proteins creates a highly unphysiological situation, formal proof that ADAM10 is the crucial neuronal α-secretase needs an additional loss-of-function approach.
We decided to specifically disrupt the protease in neurons using newly generated Adam10 floxed mice crossed with Nestin-Cre transgenic mice. The neural progenitor-specific disruption of ADAM10 led to late embryonic lethality, a reduction of the ganglionic eminence (GE), and a disrupted organization of the cortical region. We analyzed the phenotype further using primary neurospheres and primary cell cultures to investigate Notch signaling and APP processing.
A mouse cosmid clone containing the 5′ region of the Adam10 gene with exon 2 was isolated from a 129/ola cosmid library (RZPD clone ID F11344Q2). For construction of the targeting vector, an 8.2 kb KpnI-PstI DNA restriction fragment of Adam10 covering exon 2, 7.3 kb 5′ sequence, and 5.8 kb 3′ downstream sequence was subcloned into the plasmid vector pUC-18. The hygromycin B resistance gene, driven by the PGK (phosphoglycerate kinase) promoter flanked with two flippase recognition target sequences, one loxP sequence downstream of the hygromycin B resistance gene, was inserted in the NotI-ApaI site 5′ upstream of exon2 in intron1. A second loxP sequence was inserted into the SphI-SpeI site in intron 2. The targeting vector was linearized with KpnI and introduced into the embryonic stem (ES) cell line E14 129 SvEv by electroporation (ITL). Hygromycin B-resistant (100 μg/ml) colonies were screened by Southern blot analysis. Genomic DNA of ADAM10 ES cells was digested with PstI and BanI and hybridized either with a 5′ external gDNA probe (KpnI-BglII 767 bp fragment), internal hygromycin probe and a 3′ external gDNA probe (5′ TCT TGG CAG AGC TTG AGC A 3′-5′ ACA CAG GGG CTG GAA TAC TTC TA 3′) as indicated in Figure 1 A and displayed in Figure 1 B. Mutated ES cell lines were microinjected into blastocysts of C57BL/6J mice. Chimeric males were obtained and mated with C57BL/6J females to transmit the modified Adam10 alleles to the germ line. Animals carrying a null allele were obtained after breeding with transgenic females expressing a Nestin-driven Cre-recombinase. Determinations of the genotypes of the floxed mice or yolk sac of embryos were done by PCR analysis using the primers as indicated in Figure 1 A and displayed in Figure 1 C. To generate Adam10 cKO mice, homozygous floxed mice (Adam10 Fl/Fl) were crossed with Nestin-Cre transgenic mice (Tronche et al., 1999) to obtain mice heterozygous for the floxed Adam10 allele and hemizygous for the Nestin-Cre allele, which were then crossed with Adam10 Fl/Fl mice to obtain Adam10 cKO mice. Adam10 Fl/Fl, Adam10 Fl/+, or Nestin-Cre; Adam10 Fl/+ mice were used as a control.
Whole brains were taken from E12.5 embryos or postnatal day 1 (P1) mice and were lysed using cell lysis buffer (5 mm Tris base, 1 mm EGTA, 250 mm sucrose, 1% Triton X-100). Postnuclear fractions were taken, proteins were quantified using standard BCA assay (Pierce), and 30 μg protein/lane was loaded on SDS-PAGE gels and transferred to polyvinylidene difluoride membranes (Roth) to perform Western blot analysis. ADAM10 was detected using a polyclonal antiserum (B42.1) generated against the 17 C-terminal amino acid residues of ADAM10. Notch-1 fragments were detected with anti-Notch-1 mN1A antibody (Sigma) raised against the cdc10-NCR region of Notch 1 and anti-activated Notch-1 8925 (Abcam) against the NICD. N-cadherin was detected using a mouse monoclonal antibody from BD Transduction Laboratories (#610920) directed against the C terminus of N-cadherin. Blots were developed using the ECL detection system (GE Healthcare).
Embryos were fixed by immersion in 4% buffered formalin. After overnight postfixation, embryos were dehydrated in ascending ethanol concentrations and embedded in low-melting-point paraffin following standard laboratory procedures. From all paraffin blocks, 4 μm sections were stained with hematoxylin and eosin (H&E) following standard laboratory procedures and were submitted to immunostaining following standard immunohistochemistry procedures using the Ventana Benchmark XT machine (Ventana). Briefly, deparaffinated sections were boiled for 30–60 min in 10 mm citrate buffer, pH 6.0, for antigen retrieval. All solutions provided by Ventana. Sections were then incubated with primary antibody in 5% goat serum (Dianova), 45% Tris buffered saline pH 7.6 (TBS), 0,1% Triton X-100 in antibody diluent solution (Zytomed) for 1 h. Followed by detection with anti-rabbit or anti-goat histofine Simple Stain MAX PO Universal immunoperoxidase polymer, or mouse stain kit (for detection of mouse antibodies on mouse sections). All secondary antibody-polymers were purchased from Nichirei Biosciences. Detection Of Antibodies was performed with an ultraview universal DAB detection kit from Ventana according to the standard settings of the machine. The counterstaining was also performed by the machine. Vasculature and connective tissue served as internal negative controls, additional negative controls included sections treated with secondary antibody only. The following primary antibodies were used: monoclonal anti-NeuN (1:50; Millipore), polyclonal anti-KI67 (1:100; Abcam), and polyclonal activated caspase-3 (1:1000; R&D Biosystems).
Pregnant females were injected intraperitoneally with BrdU (20 mg/ml) at 100 mg/kg body mass at gestational stage E13.5 and labeled embryos were removed at E17.5, processed, and serially sectioned. Serial sections (6 μm) were used for immunohistological screening using monoclonal anti-BrdU antibody (1:100; Roche). For this, the brain sections were deparaffinized in Clear-Rite (Prosan), rehydrated, and incubated in 2N HCl for 20 min at 37°C, neutralized in 0.1 m borate buffer, pH 8.5, for 10 min at room temperature, rinsed twice in PBS sequentially quenched for endogenous peroxidase with 3% hydrogen peroxide for 10 min and blocked 30 min in a solution of TNB plus 5% of normal goat serum (DAKO) and incubated ON at 4°C with 0.1 mg/ml unconjugated fab fragment goat-anti-mouse IgG(Jackson ImmunoResearch) in TNB. Next day the primary antibody was applied for 1 h at room temperature in blocking solution, and after washing, was labeled with biotin-conjugated affinipure fab fragment goat anti-mouse IgG (1:250; Jackson ImmunoResearch) for 20 min and detected by streptavidin –HRP (1:200; PerkinElmer) for 20 min and a tyramide-based signal amplification technique (TSA, NEN-DuPont).
Neurospheres were generated as described previously (Reynolds et al., 1992; Tropepe et al., 1999) with slight modifications. Lateral and medial ganglionic eminences were removed from embryos, mechanically dissociated using a pipette and resulting neural stem/progenitor cells (NSC) were grown in a defined medium composed of a 1:1 mixture of DMEM and F-12 supplemented with glucose (0.6%) (Merck), sodium bicarbonate (3 mm) (Invitrogen), B27 (2%; Invitrogen), glutamine (2 mm) (Invitrogen), HEPES buffer (5 mm) (Invitrogen). The medium was supplemented with epidermal growth factor (EGF; 20 ng/ml; TEBU) and fibroblast growth factor-2 (FGF-2; 20 ng/ml; TEBU). The cells grew in neurospheres that were passaged by mechanical dissociation after pretreatment with accutase (PAA) for 15 min approximately every fifth day and reseeded with a density of 100,000 cell/ml. The number of vital cells was determined by 0.5% Trypan blue (Invitrogen) exclusion staining. From the first passage onward, neurospheres were passaged every fifth day and stem cell self-renewal was assessed.
For immunohistochemistry of proliferating neurospheres, neurospheres cultured for 5 d in vitro were transferred and washed in PBS. Neurospheres were then fixed in 4% buffered formalin overnight. The following day, neurospheres were pelleted and put in a drop 2% low melting agarose in PBS in a 1:1 ratio and further processed as above for paraffin embedding and immunostaining with the proper antibodies: KI67 (1:100; Abcam), Nestin (1:50; Millipore), GFAP (1:1000; Dako), activated caspase-3 (1:1000; R&D Biosystems), doublecortin (1:400; Millipore). H&E and Bielschowsky staining was performed according to standard procedures.
For differentiation experiments, neurospheres as a whole or dissociated into single cells with accutase (PAA) were plated at a density 500,000 cells/ml onto 15 mm glass coverslips coated with poly-l-ornithin (100 μg/ml PBS, pH 7.3) (Sigma) in serum-free culture medium without growth factors plus 1% of FBS. Five days after plating, coverslips were fixed in 4% paraformaldehyde in PBS for 30 min at room temperature followed a wash in PBS. Cells were then permeabilized for 5 min in PBS, pH 7.3, containing 0.1% Triton X-100, rinsed in PBS and blocked for 1 h in PBS containing 5% donkey serum (Sigma). After blocking, coverslips were incubated with anti-MAP-2c mouse monoclonal (1:400; Sigma) and anti-GFAP rabbit polyclonal (1:600; DAKO) antibodies diluted in PBS containing 5% serum for 1 h. Coverslips were then rinsed in PBS, pH 7.3, containing 0.1% Triton X-100 and subsequently incubated in the appropriate Alexa (Invitrogen) secondary antibodies for 1 h. Coverslips were rinsed three times in PBS. Separate coverslips (from similar culture conditions) were used for oligodendrocyte staining. Coverslips were incubated in anti-CNPase mouse monoclonal antibody (1:1000; Sigma) in PBS containing 5% of donkey serum 1 h. Then, the coverslips were rinsed and subsequently incubated in secondary Alexa (Invitrogen) antibody. Cultures were counterstained with 4′,6-diamidino-2-phenylindole (10 μg/ml in H2O; Roche).
RNA was extracted out of whole brains taken from P1 control (n = 5) and Adam10 cKO (n = 3) animals using the NucleoSpin RNAII kit (Macherey Nagel). Two micrograms of RNA was reverse transcribed using the RevertAid first-strand cDNA synthesis kit (Fermentas). Expression levels of Hes1, Hes5, Hey1, and Hey2 were determined by real-time PCR analysis of 0.5 μl of cDNA on a 7900HT fast real-time PCR system using the TaqMan gene expression assays-on-demand system and TaqMan universal PCR master mix (all Applied Biosystems; assay numbers Mm00468601_m1, Mm00439311_g1, Mm00468865_m1, and Mm00469280_m1, respectively) in 10 μl reaction volumes. The expression levels were normalized against glyceraldehyde 3-phosphate dehydrogenase using the ΔΔCt method, and mean expression of the wild-type (WT) animals was set to 100%.
Primary neuronal cultures were generated from trypsinized brain obtained from 14-d-old embryos. Neurons were resuspended and plated on cell culture dishes (Nunc) precoated with 1 mg/ml of poly-l-lysine (Sigma) and incubated in neurobasal medium (Invitrogen) as described previously (Annaert et al., 1999). Proliferation of non-neuronal cells was prevented by adding 5 μm cytosine arabinoside. Three- to five-day-old mouse primary neuronal cultures in neurobasal medium were incubated with 10-fold-diluted Semliki Forest Virus (SFV) expressing the pSFV-1 plasmid bearing APP-WT. Cultures were incubated for 1 h at 37°C to allow entry of the virus, followed by incubation in conditioned medium in the absence of virus (for 2 h). Medium was collected for 5 h. Cells were rinsed twice with ice-cold PBS and lysed in 1% Triton, and postnuclear fractions were isolated by centrifugation at 10,000 × g for 15 min at 4°C. Proteins were quantified by using a standard Bradford assay (Pierce) and 30 μg of proteins per lane was loaded on Bis-Tris SDS-PAGE gels (Invitrogen) and transferred to nitrocellulose membranes for Western blot detection for the indicated proteins. Polyclonal antibody B63 was used to detect APP-FL and C-terminal fragments (CTFs) (De Strooper et al., 1998; Annaert et al., 1999) in the cell extracts and 6E10 (Covance) and neo-epitope sAPPβ antibody (Covance) for sAPPα, Aβ, and sAPPβ, respectively, in the media. Levels of α- and β-cleaved carboxy-terminal stubs and secreted sAPP and total Aβ were normalized to the level of expression of APP holoprotein from the cell extracts.
Aβ40 and Aβ42 were measured via ELISA1-40 and ELISA1-42. Ninety-six-well plates were coated with monoclonal antibodies either JRF/cAβ40/28 or JRF/cAβ42/26 (Mathews et al., 2002) at 1.5 μg/ml in coating buffer containing the following (in mm): 10 Tris, 10 NaCl, and 10 NaN3, pH 8.5, overnight at 4 °C. The next day, plates were blocked in 0.1% casein in PBS for 4 h at RT to reduce nonspecific binding. Standards and conditioned medium (undiluted for Aβ42 and 1/5 diluted for Aβ40) together with the detection antibody huAβ25-HRPO (Vandermeeren et al., 2001) were loaded into triplicate wells for ELISA and incubated overnight at 4°C. ELISA plates were developed with TMB/H2O2 substrate (Pierce) according to the manufacturer's specifications. The reaction was stopped with 2N H2SO4 and the plates were read on Envision Alphascreen (PerkinElmer) at 450 nm. The different Aβ species were analyzed by urea-based Aβ SDS-PAGE and immune blot (Wiltfang et al., 2002) using 82E1 (Demeditec Diagnostics), a monoclonal antibody specific for the N terminus of Aβ. Signals were detected using ECL chemiluminescence. The individual Aβ species were normalized to the level of expression of APP holoprotein from the cell extracts.
Statistical significance was calculated by Student's t test using Microsoft Excel software. Error bars indicate the mean ± SD of the mean. p values are as follows: *p < 0.05, **p < 0.01, ***p < 0.001.
To circumvent the early embryonic lethality of Adam10 −/− mice, we used the Cre/loxP recombination system to restrict ADAM10 inactivation to the CNS (Fig. 1 A). The Adam10 gene was targeted conditionally with loxP sequences flanking exon 2 (Adam10Fl). Successful targeting was confirmed by Southern blot (Fig. 1 B). Homozygous floxed Adam10Fl/Fl mice and the Adam10 Fl/+ hemizygous mice were viable and fertile and genotypes of tail-clips were determined by PCR (Fig. 1 C). We then crossed Adam10Fl mice with Nestin-Cre transgenic mice, in which Cre is expressed in neural precursor cells under the control of the Nestin promoter (Tronche et al., 1999), to generate a neural progenitor cell (NPC)-specific Adam10 cKO mouse. Expression of Cre transcripts in Nestin-cre transgenic mice begins at E9.0 in the spinal cord and the forebrain–midbrain junction and then extends to the entire CNS by E10.5 (Yang et al., 2004). The expression of Cre in the neural progenitor cells localized in the ventricular zone (VZ) was demonstrated previously (Graus-Porta et al., 2001). Western blotting established strongly decreased levels of ADAM10 protein in total brain extracts of Adam10 cKO mice (Fig. 1 D), with a very low residual expression level likely caused by non-neuronal cells.
We intercrossed Nestin-Cre; Adam10 Fl/+ with Adam10Fl/Fl animals. We analyzed 72 pups from birth until the time of weaning (Table 1). Only six (8.3%) homozygous Adam10 floxed null mutants that survived >12 h were observed, indicating late embryonic and perinatal lethality of the mice. Macroscopically these cKO embryos did not display gross abnormalities except for intracranial hemorrhages (Fig. 2 A).
We examined serial coronal brain sections of the Adam10 cKO and littermate control brains at different developmental stages between E12.5 and E17.5 by hematoxylin–eosin costaining (Fig. 2 B). The cerebral cortex was disorganized from E15.5 onward (Fig. 2 Bi,Bj,Bo,Bp) with the exception of the upper cortical layer. The boundary between the intermediate zone (IZ), a region in which cells have undergone final mitosis and initiate migration, and the VZ, in which proliferating neuroepithelial cells reside, had disappeared. The GE in the cKO mutants also appeared to be smaller at E12.5–E15.5 (Fig. 2 Be,Bf,Bk,Bl), a phenotype that became more apparent by E17.5 (Fig. 2 Bq,Br). Furthermore, the cortical region and GE displayed severe cell loss (a diminished cell number) at E17.5 (Fig. 2 Bp,Br). These findings suggest that loss of ADAM10 disrupts the cortical organization and impairs the normal development of the GE.
The development of neurons in the embryonic brain is a complex cellular and molecular mechanism. It starts at approximately E11 with the birth of new neurons from stem cell precursors, a process called neurogenesis, and continues through E17 compromising 11 cell cycles (Caviness et al., 1995; McConnell, 1995). During early neurogenesis, very few progenitor cells in the neuroepithelium exit the cell cycle to become postmitotic neurons, while the vast majority of progenitor cells re-enter the cell cycle after mitosis, resulting in an expansion of the neural progenitor population in the ventricular zone. As neurogenesis progresses, a fraction of neural progenitor cells complete their final mitosis, differentiate into postmitotic neurons and migrate from the ventricular zone into the cortical plate, forming an “inside-out” gradient of maturation. This process is guided by radial glial fibers (Angevine and Sidman, 1961; Rakic, 1988; Hatten, 1990; Mission et al., 1991; Hatten, 1993). A massive neuronal loss in the brain and a disturbance of the lamellar composition of the neocortex can occur via several possible mechanisms, such as the following: (1) premature differentiation of progenitor cells, (2) decreased proliferation, (3) increased apoptotic cell death, or (4) cortical migration defects.
To assess neuronal differentiation in the Adam10 cKO and control brains, we performed immunostaining for NeuN, a marker for postmitotic neurons, at E12.5, E15.5, and E17.5 (Fig. 3 A). At E12.5 there was no difference in the pattern of the NeuN immune-reactive neurons of the Adam10 cKO mice compared with controls (Fig. 3 Aa–Af). At E15.5 more postmitotic neurons labeled by NeuN were found in the GE and the neocortex of the Adam10 cKO (Fig. 3 Ag–Al). The ventricular zone, which consists of NeuN-negative progenitor cells in the control, contained also postmitotic neurons in the Adam10 cKO (Fig. 3 Ai,Aj). However, at E17.5 the number of NeuN-positive cells of the cortex and the GE had decreased in Adam10 cKO (Fig. 3 Am–Ar). Together, these observations point toward a premature differentiation of neural progenitor cells into postmitotic neurons in the absence of ADAM10 resulting in an early depletion of the neural progenitor population.
Bromodeoxyuridine (BrdU) birth dating was used to examine whether the abnormalities in cortical layering reflected abnormal neuronal migration. Neurons were labeled at E13.5 and appeared more brightly stained at E17.5 in the Adam10 cKO than in the control brains (Fig. 3 Bb). This indicated that neurons had exited the cell cycle more early following the pulse in the cKO brains than in the wild-type brains, thereby retaining all of the incorporated BrdU. (Fig. 3, compare Bb to Ba). This observation was in agreement with the NeuN immunostaining in Figure 3 A. The labeled cells in the control brain were localized mostly in the cortical plate, whereas in the cKO more cells were scattered across the intermediate zone (Fig. 3 Ba,Bb). These results indicated a role for ADAM10 in cortical layering and neuronal migration.
The early neuronal differentiation could explain the cell loss observed in the Adam10 cKO brains, but alternatively changes in proliferation or apoptosis had to be considered as well. We first immunolabeled sections from control and Adam10 cKO embryos with the cell cycle marker KI67 (Fig. 4). No difference in the number of KI67-positive proliferating cells in the Adam10 cKO and littermate control brains at E12.5 (Fig. 4 a–f) was observed, showing that the lack of ADAM10 does not lead to a reduction in the proliferation rate of neural progenitor cells at early embryonic age. However, at E15.5 the Adam10 cKO cortex displayed a minor and the GE a drastic decrease in the number of KI67-positive cells (Fig. 4 g–l), which corresponded to the increased number of NeuN-positive cells in Figure 3 A at the same embryonic age. By E17.5, considerably fewer KI67-positive cells remained in the ventricular zone of the cKO brains (Fig. 4 m–r). We also performed staining for activated caspase-3 on serial coronal brain sections at E12.5, E15.5, and E17.5 in Adam10 cKO and control brains (data not shown). A minor number of positive cells were observed in cKO and the control brains and although close observation suggested a slight increase in the number of apoptotic cells in the cortex and GE from cKO brains, this difference was not significant (Table S1, available at www.jneurosci.org as supplemental material), indicating that apoptosis is not the primary cause of cell loss in the E17.5 Adam10 cKO brains.
We examined the number of neural stem cells in the Adam10 cKO embryonic brain using the neurosphere assay (Reynolds et al., 1992; Tropepe et al., 1999). The ganglionic eminence from E13.5 Adam10 cKO embryos and their littermate controls were dissociated into single cells to generate neurospheres (Fig. 5 A). The number of neural stem cells isolated from the E13.5 Adam10 cKO brains was decreased by 50% (p < 0.01). In addition, an increased number of differentiated cells were attached to the flask compared with littermate controls (Fig. 5 A). Single primary neurosphere colonies were capable of producing new secondary neurospheres after 5 d in vitro. The number of descendant secondary neurospheres generated from subcloning the single primary neurosphere can be used to estimate the extent to which the initial primary neurosphere from stem cells undergoes symmetric expansionary divisions (Reynolds and Weiss, 1996). The numbers of secondary neurosphere colonies from E13.5 Adam10 cKO and control were the same (data not shown). There was also no difference in the amount of KI67 proliferating cells or apoptotic cells within the neurospheres (Fig. 5 Bc–Bf,Bm,Bn,C), suggesting that the remaining proliferating cells in the Adam10 cKO brains retained a normal proliferation capacity.
Each sphere represents a mixture of stem and early committed precursor cells (Campos et al., 2004). To examine the cellular composition and distribution inside the neurospheres, we cut paraffin sections of intact proliferating neurospheres and immunostained them with different antibodies (Fig. 5 B). We then compared the distribution of stem cell markers (KI67, Nestin) (Fig. 5 Bc,Be) with markers of early committed precursor cells for neurons (doublecortin) (Fig. 5 Bi) and astrocytes (GFAP) (Fig. 5 Bg). KI67+ and Nestin+ cells were present around the edge of the sphere, whereas the differentiation markers were mainly present inside the sphere, demonstrating the separation of stem cell and early committed precursor populations in the neurospheres. This might reflect the situation in the developing nervous system, in which the proliferating stem cells in the ventricular zone form a distinct layer from the newly differentiated postmitotic neurons present in more superficial regions of the cortex. This was further illustrated by a separation of KI67 and NeuN expression in the control embryonic cortex as shown in Figures 3 and and4.4. The Adam10 cKO spheres displayed a disturbed organization, similar to the situation in vivo, suggesting that the neurosphere model is indeed a good model to investigate the function of ADAM10 in neuronal differentiation (Fig. 5 Ba–Bn). Together, there were fewer neural stem cells in the Adam10 cKO brains, but the remaining proliferating cells were perfectly capable of proliferating and generating neurospheres in vitro, demonstrating their neuronal stem cell capacity.
To assess neural stem cell multipotentiality, secondary neurospheres as a whole or in single-cell suspensions were cultured on poly-l-ornithin-coated coverslips in the presence of 1% serum for 5 d. The cells were immunostained for MAP2 (neurons), GFAP (astrocytes), and CNPase (oligodendrocytes) (Fig. 6 A). The neurospheres derived from E13.5 Adam10 cKO neural stem cells gave rise to more neuronal progeny than those from control neural stem cells as demonstrated by the strong increase in MAP2-expressing cells. However, the increase in neurons occurred at the expense of the glial lineages as seen by the decreased number of GFAP+ cells (Fig. 6 Aa–Ad; schematized in Fig. 6 C). No difference was observed in the number of oligodendrocytes, but it should be noted that they represented only a minor fraction of the total cell number (Fig. 6 Ae,Af,B). In addition, we saw a fourfold increase in the number of apoptotic cells in the Adam10 cKO when compared with control spheres (Fig. 6 Ag,Ah) (p < 0.05).
Notch activity has been shown to be important in cell fate decision during neurogenesis (Ishibashi et al., 1995; Schuurmans and Guillemot, 2002; Yoon et al., 2004; Alexson et al., 2006). To understand the molecular base of the premature neuronal differentiation in Adam10 cKO embryos, we analyzed extracts from P1 brains from Adam10 cKO and controls for the presence of Notch-1 fragments (Fig. 6 D). Using an antibody against the mouse Notch-1 Cdc10/ankyrin repeat cytokine response region, an accumulation of the furin-generated S1 Notch-1 cleavage product in the absence of ADAM10 was evident (Fig. 6 D). In line with these observations, immunoblotting of Adam10 cKO brain extracts using an antibody directed against the cleaved intracellular (activated) form of the NICD also revealed a strong reduction in the generation of NICD compared with Adam10Fl/Fl brain extracts not expressing the Cre transgene (Fig. 6 D).
To address the question whether the observed alteration in Notch-1 processing affects Notch-1-dependent signaling, we also investigated Hes1, Hes5, Hey1, and Hey2 transcripts, which encode basic helix-loop-helix transcription factors, major downstream effectors of the Notch signaling pathway. Quantitative RT-PCR (qRT-PCR) of whole-brain extracts from newborns demonstrated significant reductions of 47, 90, 40, and 49%, respectively (Fig. 6 E), indicating a downregulation of Notch signaling in the absence of ADAM10.
The generation of ADAM10-deficient neurons provided a unique opportunity to address the importance of ADAM10 in the physiologic constitutive processing of APP. To address this question, we analyzed the generation of APP cleavage products in primary neuronal cultures from E14.5 Adam10 cKO embryos compared with controls. The APP ectodomain (sAPPα, sAPPβ), full length (APP-FL), membrane-bound CTF (APP-CTFα, APP-CTFβ), and total Aβ were analyzed from supernatants and cell extracts of dissociated neurons (Fig. 7). Together, the diminished appearance of the membrane-bound α-secretase-cleaved APP CTF (CTFα), and the observed 90% reduction in the levels of secreted sAPPα indicate that ADAM10 is the most important APP α-secretase in these neurons. Surprisingly, we also observed a paradoxical reduction in CTFβ, sAPPβ, and total Aβ peptide production and in the amount of Aβ1-40 and Aβ1-42 in the Adam10 cKO (Fig. 7 A,B). Comparison of the Aβ spectra between control and Adam10 cKO showed no shift toward longer Aβ peptide species (Aβ1-42) or shorter Aβ peptides (Aβ1-37, Aβ1-38, Aβ1-40) but confirmed the general decrease in all the Aβ species in the cKO (Fig. 7 C,D).
In this manuscript, we demonstrate the crucial role of ADAM10 in neuronal differentiation and the establishment of the cerebral cortex. Furthermore, we provide evidence for its essential role in Notch signaling and in the physiological α-secretase processing of APP in neurons.
Using a CNS-specific Nestin-Cre mouse, which expresses Cre-recombinase in the NPC (Zimmerman et al., 1994), we were able to circumvent the previously documented early embryonic lethality of Adam10 KO mice (Fig. 1 and Table 1). Nevertheless, CNS organization was severely affected in these embryos, leading to massive intracranial bleedings and complete disorganization of the cortical layering of the brain (Fig. 2). Detailed histological analyses of the Adam10 cKO brain showed an increased number of postmitotic neurons revealed by NeuN staining (Fig. 3 A) and a reduced number of KI67-positive progenitor cells (Fig. 4). Thus, lack of ADAM10 causes a premature differentiation of neural progenitor cells into postmitotic neurons, resulting in a partial depletion of the neural progenitor population, which also explains the small GE. These studies were further confirmed with the neurosphere assay, which indicates that more mature neurons are already obtained when plating cells from ADAM10-deficient brains (Fig. 5). The relative depletion of neurons was not associated with increased activated caspase-3 staining, showing that loss of ADAM10 does not result in massive apoptotic cell death at early stages (Table S1, available at www.jneurosci.org as supplemental material), ruling out the possibility of apoptosis as the primary cause of neuronal cell loss.
The effect of ADAM10, as would be predicted for a premature differentiation phenotype, is not observed in the neuronal progenitor cells per se. Although fewer progenitors are present in the plating assay (Fig. 5), they remain fully capable of generating neurons and glia cells (Fig. 6). This was also further confirmed in vivo as KI67-positive, proliferating cells were equally present in Adam10 cKO and control brains at an early developmental age, E12.5 (Fig. 4). Thus, ADAM10 is not involved in the regulation of the early neural progenitor proliferation. Our H&E stainings (Fig. 2) revealed a disruption of the cortical layering and BrdU cell birth dating analyses showed an apparent defect in neuronal migration in the Adam10 cKO, with neurons arriving in ectopic sites (Fig. 3). This suggests a role for ADAM10 in the regulation of the neuronal migration during cortical development.
The Notch signaling pathway is known to play a fundamental role in cell fate decisions in the development of the nervous system at many different levels. (Artavanis-Tsakonas et al., 1999). Notch signals govern the decision between neuronal (in an inhibitory fashion) and glial lineages (promoting this fate), modulate the fate of terminally differentiated neurons, and affect neuronal migration and even learning and memory (Yu et al., 2001; Costa et al., 2003; Saura et al., 2004; Wang et al., 2004; Louvi and Artavanis-Tsakonas, 2006; Corbin et al., 2008; Hashimoto-Torii et al., 2008). In addition, Notch pathway components, including ligands, receptors, modulators, and targets, are expressed in the telencephalic ventricular zone (Lindsell et al., 1996; Zhong et al., 1997; Irvin et al., 2001; Yun et al., 2002). Mouse mutants deficient in genes of the Notch signaling pathway support the view that this pathway plays a critical role in regulating the neural progenitor pool during neocortical development (Artavanis-Tsakonas et al., 1999; Hitoshi et al., 2002; Yoon et al., 2004; Mizutani and Saito, 2005; Yoon and Gaiano, 2005; Gaiano, 2008). Conditional deletion of Notch-1 in the neural progenitor pool, similar to the strategy used for the ADAM10 deletion reported here, indeed resulted in precocious neuronal differentiation (Yang et al., 2004). A similar phenotype was also found in Nestin-Cre transgene-driven inactivation of presenilin (PS) (Kim and Shen, 2008), which is known to mediate the Notch-1 S3 cleavage and NICD production (De Strooper et al., 1998, 1999). Both the Adam10 conditional knock-out presented here and the PS inactivation in NPCs and NPC-derived neurons and glia show that both Notch signaling proteases are required for neural progenitor cells to re-enter cell cycle and for the expansion of neural progenitor population during embryonic development. The depletion of progenitor cells may indeed also lead to a reduced number of astrocytes, which are involved in the formation of the blood–brain barrier, thereby contributing to the hemorrhages observed in presenilin (PS) cKO mice (Kim and Shen, 2008) and Adam10 conditional knock-out mice (this study).
To understand the molecular mechanism underlying the premature neuronal differentiation observed in Adam10 cKO brain, we examined the proteolytic processing of Notch-1 as well as the expression of Notch-1 target genes Hes1 and Hes5 and Hey1 and Hey2, which are basic helix-loop-helix transcriptional regulators that antagonize proneural genes like Mash1 and Neurogenins. This antagonism blocks early neuronal gene expression and is central to the inhibition of neuronal differentiation by Notch-1 signaling. By qRT-PCR, we found a clear reduction in the mRNA expression of Hes1 and Hes5 and Hey1 and Hey2 target genes, further proving the important role of ADAM10 in the regulation of the Notch signaling (Fig. 6 E).
Whereas in Drosophila, the ADAM10 ortholog Kuzbanian has been amply demonstrated to act as Notch sheddase, in the mammalian system, cell-based assays have in contrast implicated mainly ADAM17 in the S2 Notch-1 cleavage (Brou et al., 2000; Mumm et al., 2000). The Notch-like phenotype of classical Adam10-deficient mice (Hartmann et al., 2002), the novel data in the current work showing a direct effect of ADAM10 on the ability to generate signal-relevant Notch intracellular domain (Fig. 6 D), and the lack of Notch-related alterations in Adam17-deficient mice (Peschon et al., 1998), clearly underline the essential role of ADAM10 as the Notch sheddase in vivo also in mammalian species. Although the phenotype of our cKO mice clearly highlights the role of the ADAM10 in Notch-1 signaling, we cannot completely exclude the possibility that part of the abnormalities may be explained by a reduced shedding of additional neuronal surface molecules. In this regard, it is interesting to note that in ADAM10-deficient neurons the CTF1 generation of N-cadherin was severely reduced (Fig. S1, available at www.jneurosci.org as supplemental material), confirming our earlier data (Reiss et al., 2005) in fibroblast cells. We also cannot rule out the possibility that the decreased level of sAPPα contributes to the alteration in neural progenitor cell proliferation/differentiation in our Adam10 cKO phenotype, as previous studies have shown that sAPPα is a neurotrophic factor that affects the development of embryonic brain cells, including progenitor cells. sAPPα has also been suggested to play a role in developmental and synaptic plasticity in the nervous system (Mattson, 1994; Furukawa et al., 1996; Caillé et al., 2004; Rohe et al., 2008).
We finally focused on the processing of one of the most medically important putative ADAM10 substrates, i.e., APP. We provide here convincing evidence that loss of ADAM10 disturbs the α-secretase-mediated cleavage of APP to a large extent, indicating that ADAM10 is the major α-secretase in the cells that are most relevant to Alzheimer's disease (Fig. 7). Our work agrees with previous overexpression experiments in mice and in cell-based systems (Postina et al., 2004; Postina, 2008). Although it is possible that the other α-secretase candidates, including ADAM9 and ADAM17, also contribute to a certain extent to APP processing in neurons, their contribution appears to be quantitatively much less significant. Surprisingly, we found in the ADAM10 KO neurons a paradoxical significant reduction in the β-secretase cleavage, which is in contradiction to the general dogma that BACE1 and α-secretase are in competition for ectodomain cleavage of APP (Skovronsky et al., 2000). This concept is supported by a number of observations; for instance, several pharmacological agents known to stimulate α-secretase cleavage of APP also reduce β-cleavage, thereby reducing Aβ peptide generation (Allinson et al., 2003). Likewise, overexpression of ADAM10 in an Alzheimer's disease mouse model decreases Aβ plaque load and cognitive deficits (Postina et al., 2004), while BACE1 reduction results in an elevation of α- and a decrease of β-cleavage (Vassar et al., 1999). The molecular basis of this competition between the two secretase pathways is not very well understood and could involve as well direct competition between the two proteases for the substrate, as effects on subcellular trafficking of substrate or protease. In our experiments, in which ADAM10 is taken away in primary neurons, we see an opposite effect, i.e., a decrease in β-secretase activity [although β-secretase levels are not decreased (results not shown)]. We speculate that ADAM10 deficiency and the effects on APP processing result in a change in localization or subcellular trafficking of either BACE1 or APP, but further research is clearly needed to explain this intriguing phenomenon.
In conclusion, we present the first characterization of neuron-specific Adam10-deficient mice and provide substantial evidence for the key function of ADAM10 in neural Notch signaling, in Notch-dependent CNS development, and in neuronal α-secretase processing of APP, with implications for Alzheimer's disease.
This work was supported by Deutsche Forschungsgemeinschaft Sonderforschungsbereich 415, Interuniversity Attraction Poles Program IUAP P6/58 of the Belgian Federal Science Policy Office, DeZnit (EU-FP VI), the Center of Excellence “Inflammation at Interfaces,” the Fund for Scientific Research, Flanders; the Vlaams Instituut voor Biotechnologie (VIB); and a Methusalem grant from Katholieke Universiteit (KU) Leuven and the Flemish government. M.G. is supported by Deutsche Forschungsgemeinschaft Forschergruppe 885. E.J. was supported by Innovatie door Wetenschap en Technologie and a short-term fellowship from the European Molecular Biology Organization. C.B. was supported by National Institutes of Health Grants GM64750 and EY015719. We thank Dr. Hermans-Borgmeyer for assistance in generating the knock-out mice, Dieter Hartmann for initial help with the study, Wim Annaert for providing the anti-ADAM10 antibody, and Marlies Rusch and Véronique Hendrickx for excellent technical assistance.