Cerebral Autosomal Dominant Arteriopathy with Subcortical infarcts and Leukoencephalopathy (CADASIL) is the most prominent known cause of inherited stroke and vascular dementia in human adult. The disease gene, NOTCH3, encodes a transmembrane receptor primarily expressed in arterial smooth muscle cells (SMC). Pathogenic mutations lead to an odd number of cysteine residues within the NOTCH3 extracellular domain (NOTCH3ECD), and are associated with progressive accumulation of NOTCH3ECD at the SMC plasma membrane. The murine homolog, Notch3, is dispensable for viability but required post-natally for the elaboration and maintenance of arteries. How CADASIL-associated mutations impact NOTCH3 function remains a fundamental, yet unresolved issue. Particularly, whether NOTCH3ECD accumulation may titrate the ligand and inhibit the normal pathway is unknown. Herein, using genetic analyses in the mouse, we assessed the functional significance of an archetypal CADASIL-associated mutation (R90C), in vivo, in brain arteries. We show that transgenic mouse lines expressing either the wild-type human NOTCH3 or the mutant R90C human NOTCH3, at comparable and physiological levels, can rescue the arterial defects of Notch3−/− mice to similar degrees. In vivo assessment of NOTCH3/RBP-Jk activity provides evidence that the mutant NOTCH3 protein exhibits normal level of activity in brain arteries. Remarkably, the mutant NOTCH3 protein remains functional and does not exhibit dominant negative interfering activity, even when NOTCH3ECD accumulates. Collectively, these data suggest a model that invokes novel pathogenic roles for the mutant NOTCH3 protein rather than compromised NOTCH3 function as the primary determinant of the CADASIL arteriopathy.
Stroke is a major determinant of disability and dementia in adults, and, vascular dementia is the second leading cause of irreversible cognitive impairment in the elderly after Alzheimer's disease (1). Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy (CADASIL) (MIM 125310) is the most common genetic form of stroke and vascular dementia. The disease affects young and older individuals irrespective of vascular risk factors and manifests clinically with recurrent ischaemic strokes, cognitive deficits, migraine with aura, psychiatric disorders and epilepsy (2–5). The underlying vascular lesion is a unique systemic arteriopathy, although involving primarily brain arteries, characterized by prominent alterations of vascular smooth muscle cells (SMC) and pathognomonic granular osmiophilic (GOM) deposits (6,7). Virtually, no therapies are available to modulate the progression of the disease. In 1996, mutations in the gene NOTCH3, which encodes a 2321-amino acid-long single pass transmembrane protein belonging to the evolutionary conserved Notch receptors family, were shown to cause CADASIL (8).
Notch receptor that functions at the cell surface as a heterodimer is composed of a large extracellular domain (NOTCHECD), non-covalently attached to the membrane-tethered intracellular domain (NOTCHTMIC) (9). Upon binding to ligands of the Delta/Jagged family expressed on juxtaposing cells, Notch family proteins undergo sequential proteolytic cleavages producing the active Notch intracellular domain that forms a transcriptional activation complex with a DNA-binding protein termed CSL/RBP-Jκ, which regulates the transcription of target genes (10–14). The Notch signaling pathway plays a central role in the development and maturation of most vertebrate organs, with pleiotropic effects depending on dose and context (15). Recent studies documented a crucial role for Notch3 in the elaboration and maintenance of arteries. In late embryos and adults, NOTCH3 is expressed predominantly in the SMC of blood vessels with a preferential expression in arteries (16,17). Mice completely lacking Notch3, although viable and fertile, exhibit soon after birth prominent defects of arteries, particularly of brain arteries, due to a requirement for Notch3 in arterial differentiation and post-natal maturation of SMC (18,19).
Over the past decade, more than 80 NOTCH3 mutations have been characterized in more than 400 CADASIL families. Mutations are distributed throughout the 34 epidermal growth factor repeats (EGFR) that compose the extracellular domain of NOTCH3, with a strong clustering within EGFR2–5. The vast majority are mis-sense mutations, a very few are splice-site mutations or small inframe deletions. Remarkably, all the mutations lead to an odd number of cyteine residues (20–24). Notably, immunohistochemical and biochemical studies have shown that mutations resulted in an abnormal accumulation of aggregated NOTCH3ECD at the SMC plasma membrane in patients' blood vessels, due to an impaired clearance of the receptor from the cell surface. Of interest, NOTCH3ECD accumulates in close vicinity to the GOM deposits, however, whether it is part of the GOM remains controversial (16,25–27).
How CADASIL-associated mutations impact NOTCH3 function remains largely unknown. Contrasting with the highly stereotyped nature of the mutations, in vitro studies failed to reveal an underlying common mechanism. A few mutations, located within or near the ligand-binding domain, showed strongly reduced signaling activity (28–32). In contrast, the majority of mutations, including those located in the mutational hotspot region, retained the ability to activate RBP-Jκ in response to Delta/Jagged-binding, although some mutations exhibited impaired maturation and decreased cell surface expression of the receptor. Notably, interpretation of these data is complicated by the fact that these experiments do not mimic the effect of NOTCH3 mutations in human arteries since mutant NOTCH3 was vastly overexpressed, and, importantly, NOTCH3ECD accumulation at the SMC plasma membrane, which could titrate the ligand and inhibit the normal pathway, was not recapitulated in these assays. In addition, wild-type NOTCH3 expressed in cultured cells may have confounded the interpretation of the data. Moreover, only response to membrane-bound ligands of the Delta/Jagged families was investigated while additional ligands have now been identified (33–36).
Herein, we assessed the functional significance of an archetypal CADASIL mutation (R90C), in a physiological context. Using genetic analyses in the mouse, we investigated whether the mutant protein, in vivo, in brain arteries, exhibits reduced activity or acts in a dominant negative manner to interfere with the function of wild-type NOTCH3 or alternatively exhibits increased activity. Importantly, we carried out the analyses before and also upon appearance of NOTCH3ECD accumulation in view of the evidence that NOTCH3ECD accumulation may impact Notch3 activity. Collectively, our data provide compelling evidence that the R90C mutation does not compromise the canonical NOTCH3 function in brain arteries in vivo.
A functional assay to assess the activity of mutant NOTCH3, in vivo, in brain arteries
To test the hypothesis that mutant NOTCH3 may cause CADASIL because of a reduced functional activity in artery, we used transgenic mice that express either wild-type or mutant NOTCH3 and examined their respective ability to rescue the arterial defects of Notch3−/− mice. Using the arterial SMC-specific SM22α promoter, we previously established two transgenic lines that express human NOTCH3 with an arginine substituted in a cysteine (R90C) in the EGFR2 termed TghN3(R90C) (lines ma and ve). This particular mutation is a recurrent mutation located within the mutational hotspot region and is associated with a classical CADASIL phenotype in human patients. Moreover, prior histological analyses showed that aged TghN3(R90C) mice recapitulated the arterial defects as typically seen in CADASIL patients, including GOM deposits and NOTCH3ECD aggregation (23,37). Using a similar approach, we generated and established three independent transgenic lines expressing wild-type human NOTCH3 termed TghN3(WT).
The rescue of the Notch3−/− phenotype is contingent upon appropriate spatio-temporal expression of the hNOTCH3 transgene in arterial SMC. The SM22α promoter is active in arterial SMC but our previous analyses showed that its activity was greatly diminished in Notch3−/− mice because of an impaired arterial identity of SMC (18). As a first step to assess the feasibility of the rescuing experiment, we examined in detail the spatial and temporal activity of the SM22α promoter on Notch3+/+ and Notch3−/− backgrounds. Visualization of SM22α promoter activity with a SM22α-lacZ transgene showed that on a Notch3+/+ background, SM22α promoter was active in brain arteries from mid-late gestation onwards, although, in the medium-sized and smaller brain arteries, expression was detected only post-natally and gradually increases beyond (Supplementary Material, Fig. S1 upper-left and lower panels). On the Notch3−/− background, the SM22α promoter was silent in the small brain arteries at all stages investigated, indicating a requirement for Notch3 in the initial activation of the promoter in these vessels. Thus, the SM22α promoter-driven hNOTCH3 transgene should not rescue the Notch3−/− arterial defects in small brain arteries. In contrast, in the medium-sized and larger brain arteries, the SM22α promoter remained active until about post-natal day 6 and was down-regulated thereafter, although more modestly in the largest arteries (Supplementary Material, Fig. S1 upper-right and lower panels). Thus, in the medium-sized and larger brain arteries, Notch3 is required for the maintenance but not for the initial activation of the promoter. Therefore, SM22α promoter-driven hNOTCH3 transgene should rescue the Notch3−/− arterial defects in these vessels.
We also controlled for transgenes expression level. This is particularly important since the function of NOTCH3 is potentially influenced by its level of expression. First, NOTCH3 immunohistochemical staining indicated that human NOTCH3 transgene was expressed in brain arteries at low levels in wild-type and mutant lines (data not shown). We further developed a real-time RT–PCR strategy to precisely quantify human NOTCH3 (hNOTCH3) transcript levels in isolated medium-sized and larger brain arteries. We found that hNOTCH3 mRNA expression was almost comparable in these vessels in both mutant lines ma and ve and in wild-type line 46 and lower in wild-type lines 38 and 23 (Supplementary Material, Fig. S2). Notably, hNOTCH3 transgenes were expressed from about 19 to 86% to that of the endogenous mNotch3 in wild-type and mutant transgenic lines (Fig. 1).
Rescue of Notch3−/− arterial defects by wild-type and mutant hNOTCH3 transgenes
The three transgenic lines expressing human wild-type NOTCH3 were tested for their rescuing activity. The SM22α-lacZ transgene was introduced in addition to the hNOTCH3 transgene in Notch3−/− mice to first assess SMC arterial identity. We performed X-gal staining of 1 month-old brain and examined at least three mice from each line and genotype. The three wild-type transgenic lines yielded a different efficiency in their ability to restore SMC arterial identity in the medium-sized and larger brain arteries. In WT line 46, which expresses the higher level of hNOTCH3 (approximately 73% of mNotch3), β-galactosidase expression was indistinguishable from that of Notch3+/+ and Notch3+/− littermate mice (Fig. 2A–D). WT line 38 (approximately 50% of mNotch3) showed a nearly complete restoration of β-galactosidase expression, although arteries showed a patchier staining (Fig. 2E). In contrast, the lower expressor WT line 23 (approximately 19% of mNotch3) exhibited almost no restoration (Fig. 2F). As anticipated from the temporo-spatial pattern of the SM22α promoter activity, β-gal expression was not corrected in the small brain arteries, even in the rescued lines. We next analyzed the morphology of the brain arteries and the fine structure of SMC by high resolution optic microscopy analysis on semi-thin sections and electron microscopy analysis on ultra-thin sections. We choose the middle cerebral artery as a reference artery because it was easy to select analogous anatomical artery for analysis and we focused on its distal segments. In Notch3−/− mice, arteries become enlarged with a flattened elastica lamina, and abnormally thin SMC when compared with Notch3+/+ or Notch3+/− littermate mice (Fig. 3A–C and J–K). Figure 3D–F and L shows that the arterial defects were completely rescued in WT line 46 or nearly completely rescued in WT line 38, whereas rescuing was nearly absent in WT line 23. Thus, these results establish that wild-type hNOTCH3 transgene can rescue the Notch3−/− arterial defects. Notably, the data indicate that biological activity of wild-type hNOTCH3 transgenes is correlated to their expression levels in this assay.
The two mutant transgenic lines (ma and ve) were next tested for rescue activity. Mutant transgenic mice were backcrossed to Notch3−/− mice and investigated at 1 month of age as described above. At this age, brain arteries from TghN3(R90C) mice do not yet exhibit the pathological hallmarks of the CADASIL arteriopathy, including GOM deposits and NOTCH3ECD accumulation. Figure 2H–I shows that X-gal staining of N3−/−, TghN3(R90C), SM22α-lacZ mice from lines ma (n = 6) and ve (n = 4) was indistinguishable from that of WT line 46 rescued mice (n = 4) with full restoration of expression in the medium-sized brain arteries and no correction in the smaller arteries. Toluidine blue staining of semi-thin sections and electron micrographs of artery from N3−/−, TghN3(R90C) mice (line ma, n = 3 mice; line ve, n = 3 mice) showed complete rescue of the arterial morphology and SMC ultrastructure (Fig. 3H, I and M). Thus, the results strongly support the view that the CADASIL-linked R90C NOTCH3 mutant retains sufficient level of activity to support normal SMC arterial differentiation and maturation. Furthermore, analysis of TghN3(R90C) mice on a Notch3+/− background provides evidence that the R90C mutant protein does not exert a dominant negative effect on the endogenous murine Notch3 (Figs 2G and 3G and data not shown).
Wild-type and mutant NOTCH3 receptors exhibit comparable dose-dependent RBP-Jκ activity in brain arteries
We then sought to test the alternative hypothesis that the mutant receptor may increase rather than lower receptor activity. To address this issue, we assessed RBP-Jκ activity of wild-type and mutant NOTCH3, in vivo, in brain arteries. A fairly limited number of Notch/RBP-Jκ target genes have been identified and among these are the Hes and Hrt gene families that comprise in the mouse six and four members, respectively. Our previous and recent analysis showed that expression level of these genes was unaltered in the brain arteries of Notch3−/− mice strongly suggesting that Notch3 effects in arteries are not mediated by these genes (18) and data not shown. To overcome our lack of knowledge of Notch3 transcriptional targets in brain arteries, we took advantage of the NAS reporter line, which has a transgene, here referred to TP1-nlacZ, composed of a lacZ reporter gene linked to a well-characterized promoter, which consists of 12 RBP-Jκ binding motifs upstream from a minimal promoter (38,39). This transgenic mouse reporter has been shown to require the presence of functional RBP-Jκ with LacZ expression being suppressed in a RBP-Jκ null background, and to allow the readout of the general status of Notch-mediated RBP-Jκ activity in a number of sites during embryonic development (40). Figure 4A and J shows that in 1-month-old wild-type mice heterozygous for the TP1-nLacZ transgene (n > 5) robust β-galactosidase activity was detectable in the brain arteries. Histological analysis clearly showed that, in arteries, at the post-natal stages investigated, SMC were the sites of TP1-nLacZ expression, a finding further supported by the observation of co-localization of β-galactosidase activity and α-SM actin immunostaining (Supplementary Material, Fig. S3). Notably, the transgene was completely silent in the venous system as was expected from the preferential arterial expression of Notch3 and its involvement in arterial differentiation of SMC (Fig. 4A and Supplementary Material, Fig. S3). Most importantly, in Notch3−/− mice (n > 5), TP1-nLacZ expression was abrogated in arterial SMC, whereas non-vascular staining persisted, indicating that NOTCH3 was not active or had a redundant role with other NOTCH receptors in these non-vascular tissues (Fig. 4C, L and data not shown). Notch3+/− mice exhibit an intermediate reduction in Notch3 mRNA and protein levels in isolated arteries, as determined by real-time RT–PCR and western blot analysis, respectively (Supplementary Material, Fig. S4A and data not shown). Remarkably, a 50% reduction in wild-type murine Notch3 in Notch3+/− mice resulted in significantly reduced TP1-nLacZ expression in arterial SMC (n > 5) (Fig. 4B and K), although arterial structure and SMC arterial identity are preserved in these mice (Figs 2B and 3B). To try and quantify the observed reduction in X-gal staining in muscular arteries of these mice, we measured LacZ mRNA levels in isolated arteries. We found a 68 ± 13% reduction in LacZ mRNA level in Notch3+/− mice in comparison with Notch3+/+ mice (Supplementary Material, Fig. S4B). Together, these data validate the TP1-nLacZ transgene as a reliable qualitative and quantitative reporter of NOTCH3-mediated RBP-Jκ activity in brain arteries.
We next assessed RBP-Jκ activity of wild-type and mutant NOTCH3 receptors in 1 month-old mice. The TP1-nLacZ transgene was introduced in addition to the human NOTCH3 transgene in Notch3+/− and Notch3−/− mice. As expected, TP1-nLacZ expression in N3−/−, TghN3 (WT) mice was detected in the large- and medium-sized brain arteries but not in the smaller arteries (Fig. 4D–F). Also, histological analysis showed that TP1-nLacZ transgene expression was derived from SMC (Fig. 4M and N). Notably, visualization of TP1-nLacZ expression in the brain arteries of N3−/−, TghN3(WT) mice provided evidence for a progressive increase in RBP-Jκ activity across WT lines 23 (n = 6 mice), 38 (n = 4 mice) to 46 (n = 4 mice), confirming that these three lines express graded increase in hNOTCH3 transcript levels. Levels of RBP-Jκ activity in the large- and medium-sized arteries were almost comparable with those of Notch3−/− mice in WT line 23, intermediate to Notch3−/− and Notch3+/− in WT line 38, and, intermediate to Notch3+/− and Notch3+/+ in WT line 46 (Fig. 4A–F). Importantly, N3−/−, TghN3(R90C) mice from both lines ma and ve (n = 6) exhibited comparable TP1-nLacZ transgene expression levels and patterns to those observed in N3−/−, TghN3(WT) mice from line 46 (Fig. 4D, H and I). Also, the extent of RBP-Jκ activity in N3+/−, TghN3(R90C) mice (n > 5) was substantially higher to that seen in Notch3+/+ mice (Fig. 4A and G). Taken together, these results provide compelling evidence that the R90C CADASIL–NOTCH3 mutation retains normal NOTCH3/RBP-Jκ activity, in vivo, in brain arteries.
Mutant NOTCH3 retains normal function in aged mice despite NOTCH3ECD accumulation and GOM deposits
In view of the possibility that NOTCH3ECD accumulation may impact NOTCH3 activity, we generated a small cohort of TghN3(R90C) mice on the different Notch3 backgrounds to assess the function of mutant NOTCH3 in aged mice when arteries exhibit NOTCH3ECD accumulation and GOM deposits. Our previous analysis of the tail arteries of N3+/+, TghN3(R90C) mice had showed that NOTCH3ECD accumulation and GOM deposits were apparent from 14 to 16 months of age (37). Our current analysis of the medium-sized and larger brain arteries of N3+/+, TghN3(R90C) mice revealed significant NOTCH3ECD aggregation and GOM deposits at 12 months of age (Fig. 5B and E), whereas control animals, including N3+/+, TghN3(WT) or non-transgenic mice on either Notch3 wild-type, heterozygous or null background did not show these changes (Fig. 5A, D, F and data not shown). The earlier detection of NOTCH3 and GOM deposits in the brain arteries is likely explained by the higher levels of TghN3 transcript in brain arteries than in tail arteries (data not shown), in addition to the use of more robust Notch3 immunostaining and electron microscopy protocols. We verified that TghN3 mRNA levels were unchanged in aged TghN3(R90C) mice when compared with 1-month-old mice, further supporting the notion that NOTCH3ECD accumulation arises from an impaired clearance of the receptor (Supplementary Material, Fig. S5). Remarkably, TghN3(R90C) mice on a Notch3−/− background (n = 3) exhibited at the same age comparable NOTCH3ECD accumulation and GOM deposits (Fig. 5C and G). Since, the only NOTCH3 protein present in N3−/−, TghN3(R90C) mice is the mutant one, GOM deposits and NOTCH3ECD aggregation can be definitely attributed to the CADASIL mutation.
The 12-month-old mice were examined for SMC arterial identity, arterial morphology and RBP-Jκ activity. In N3−/−, SM22α-lacZ mice, X-gal staining was dramatically reduced in the middle-size and larger arteries (n > 5) (Fig. 6B). In contrast, TghN3(R90C), SM22α-lacZ mice on either Notch3−/− (n = 3) or Notch3+/− (n > 5) background retained strong arterial staining (Fig. 6C and D) comparable with that seen in non-transgenic N3+/+, SM22α-lacZ control mice (n = 4) (Fig. 6A). Also, arteries of Notch3−/− mice (n = 3) exhibited a prominent ‘butterfly-like’ enlargement with a completely flattened elastica lamina and extremely thin or defective SMC (Fig. 6F). In contrast, arterial structure of TghN3(R90C) mice on either a Notch3−/− or Notch3+/− background (n = 5 and 4, respectively) appeared identical to that of control littermate mice (n = 4) (Figs 5G and 6G–H). At 12 months of age, TP1-nLacZ transgene expression was down-regulated in both non-transgenic and transgenic animals when compared with 1-month-old mice (Fig. 6I–L). Importantly, the extent of down-regulation was similar in non-transgenic and TghN3(R90C) mice. Together, these results strongly support the view that the mutant NOTCH3 protein remains fully functional and does not exhibit dominant negative interfering activity, in aged mice, even when NOTCH3ECD accumulates.
Although the CADASIL–NOTCH3 mutations are highly stereotyped, their functional consequences remain a fundamental yet unresolved issue. The fact that all mutations result in an odd number of cysteine residues within a given EGFR has suggested that CADASIL disease is associated with a gain of function activity of mutant protein. Gain of function mutations, in principle, include: (i) hypermorphic mutations which elevate gene activity; (ii) dominant negative mutations which reduce wild-type gene activity; and (iii) neomorphic mutations in which mutant protein has additional activity. However, it remains possible that CADASIL mutations are hypomorphic mutations that lower activity of the receptor, since there is ample evidence that Notch function is sensitive to gene dosage. Moreover, mutations located within or close to the Delta/Jagged ligand-binding domain of NOTCH3 that strongly reduce signaling activity in vitro have been identified in human patients (28,29). In addition, loss of Notch3 in the mouse results in prominent SMC alterations, although they do not appear to phenocopy those seen in human CADASIL patients (18).
In this study, we assessed in vivo the functional activity of the R90C mutation, which we believe is an archetypal CADASIL mutation. This particular mutation is located within the mutational hotspot EGFR 2–5 and leads to the addition of a seventh cysteine residue, like two-thirds of CADASIL mutations. This mutation appears to retain normal RBP-Jκ activity in response to Delta/Jagged binding when overexpressed in cultured mammalian cells, like the majority of mutations investigated so far (29,31). Using genetic analyses in the mouse, we examined whether this mutation may abnormally lower or elevate NOTCH3 function in brain arteries (hypomorphic and hypermorphic hypotheses). We also assessed whether the NOTCH3ECD when abnormally aggregated at the plasma membrane of SMC may dominantly inactivate wild-type NOTCH3 function, through, for example, titration of the ligand (dominant-negative hypothesis). First, we established the validity of expressing SM22α promoter-driven human NOTCH3 transgenes in Notch3 null mice to assess functional activity of human NOTCH3 molecules in the absence of endogenous murine Notch3. Secondly, we provided evidence that mutant NOTCH3 displayed comparable ability to rescue the post-natal arterial defects of Notch3−/− mice relative to wild-type NOTCH3. Thirdly, we established the validity of using the NAS transgenic reporter line to monitor NOTCH3-mediated RBP-Jκ activity in arterial SMC in a qualitative and, importantly, in a quantitative manner. Fourth, we obtained further evidence that neither did mutant NOTCH3 lowered RBP-Jκ activity nor did it increase activity in brain arteries. Finally, we documented that mutant NOTCH3 remained functional in aged mice, despite NOTCH3ECD accumulation and GOM deposits. Taken together, our current data strongly suggest that the R90C mutant NOTCH3 protein is fully functional, does not exhibit dominant-negative interfering activity and does not alter the normal level of RBP-Jκ-mediated signaling activity, in brain arteries, in vivo. Studies in both invertebrate and vertebrate species have pointed towards CSL/RBP-Jκ-independent Notch signaling activities in some specific cell contexts (41–44). Thus, it remains formally possible that mutant NOTCH3 may cause CADASIL as a result of alteration of an RBP-Jκ-independent NOTCH3 signaling activity, although, so far, we do not have evidence for such an activity in adult brain arteries.
Potential limitations of using transgenic lines to evaluate the functionality of NOTCH3 molecules, are first that wild-type and mutant transgenes are inserted at random genomic locations that may lead to distinct expression levels, and, secondly, that the transgenes may be expressed above physiological levels. The use of mice with a point mutant Notch3 allele in the endogenous murine Notch3 locus would have been probably a more straightforward approach, although it is unclear at the present time whether such mice may develop the CADASIL phenotype. Indeed, a recent study reported that mice carrying an R142C NOTCH3 knock-in mutation do not show any phenotype (45). In this study, we carefully controlled the potential confounding factors of using transgenic lines. We examined several independent wild-type and mutant lines. Quantitative analysis of hNOTCH3 transcript levels in the brain arteries demonstrated that hNOTCH3 transcript levels in the two mutant lines were almost comparable with those seen in one wild-type line, allowing valid comparison between the wild-type and mutant transgenic lines in their functional activities. More importantly, we documented that, in mutant lines, the hNOTCH3 transgene was expressed to physiological levels, with transcript levels intermediate to 0.5 and 1 unit of endogenous mNotch3. The validity of our quantitative analysis was supported by the findings of a strong correlation between hNOTCH3 transcript expression levels and biological activities across the three wild-type lines investigated. Specifically, WT line 23 (19 ± 4% of mNotch3) exhibited almost no ability to rescue the Notch3−/− phenotype while WT line 38 (50 ± 9% of mNotch3) and 46 (73 ± 19% of mNotch3) exhibited a nearly complete or full ability to rescue the Notch3−/− defects, respectively, as anticipated from the observation that 50% of mNotch3 is sufficient to support normal SMC arterial differentiation and maturation. Moreover, as anticipated from the observed Notch3 gene dosage dependency of RBP-Jκ activity, levels of activity were, in WT line 38, intermediate to Notch3−/− and Notch3+/− and, in WT line 46, intermediate to Notch3+/− and Notch3+/+.
Our current findings support the hypothesis that CADASIL mutation exerts its effect through the gain of novel function(s) by mutant NOTCH3. How mutant NOTCH3 may mediate this effect at the molecular level is so far unknown. Also, whether NOTCH3ECD accumulation and GOM deposits, whose nature remains so far largely unknown, participate in the CADASIL pathogenic process or, alternatively, are an outcome of this novel function remains unclear. Previous studies showed that ultrastructural alterations of SMC could be detected prior to GOM deposits or NOTCH3ECD accumulation in CADASIL patients or mice (37,46). Although we acknowledge that immunohistochemistry and electron microscopy analyses may miss the deposits at the very beginning, we believe that the deposits are rather an outcome than a primary causative event in the CADASIL pathogenic process. Considering that all CADASIL mutations result in an odd number of cysteine residues, we speculate that mutant NOTCH3 may be involved, through its unpaired cysteine residue, in novel protein–protein interactions, unrelated to the canonical Notch pathway elements, such as with cytokines or structural matrix elements of the arterial wall that could be sequestered within the GOM deposits. In this scenario, mutant NOTCH3 would exert its pathogenic effect through the titration of this (or these) yet unknown protein(s). As such, characterization of component(s) of the GOM deposits may lead to the identification of the crucial biochemical pathways involved in the pathogenic process. However, the lack of histological or biochemical marker of the GOM deposits makes their characterization tricky.
In summary, our results strongly support the notion that CADASIL mutations do not alter the normal level of NOTCH3 activity in brain arteries. These findings raise the likely possibility that the mutations exert their effect through the gain of a novel function not seen with the wild-type protein. Conclusive demonstration awaits the identification and characterization of this(these) novel function(s). However, our results should encourage this further investigation that will be needed to develop treatments for CADASIL. Although the highly stereotyped nature of CADASIL mutations suggest that all mutations share at a primary level a similar disease mechanism, it is important to avoid generalization of this conclusion to all CADASIL-causing mutations in the absence of experimental analysis, particularly to those mutations exhibiting a significant reduction in Delta/Jagged-induced signaling activity, in cultured cells (28,29). As such, the combination of NAS reporter line, Notch3 null mice and mutant hNOTCH3 transgenic mice, whose generation is easier and quicker than that of mice with a point mutant Notch3 allele in the endogenous murine Notch3 locus, offer a unique system to assess in vivo the functionality of these particular mutations. In vivo analysis of these mutations should allow determining whether they share a similar disease mechanism with the archetypal R90C mutation or alternatively represent a distinct class of mutations that cause CADASIL because of a reduced NOTCH3 activity.
MATERIALS AND METHODS
Experimental mice and genotyping
Transgenic mice expressing human NOTCH3 with the R90C mutation, under the control of murine SM22α promoter [TghN3(R90C)] have been described previously (37). Transgenic mice expressing human wild-type NOTCH3 [TghN3(WT)] were similarly generated. Eight transgenic founders had germline transmission and three transgenic lines expressing human wild-type NOTCH3 were established upon quantification of transgene mRNA expression. Notch3 null mice (18,19) and Sm22α-lacZ mice (47) have been reported elsewhere. Notch pathway Activity Sensor (NAS) mice have been described recently. Briefly, the transgene is composed of a lacZ reporter gene coding for a nuclear β-galactosidase linked to a minimal β-globin promoter and six copies of the 50 bp EBNA2 response element of the TP-1 promoter (40). All mice were bred on a C57BL/6J background. The genotyping of mice was performed by PCR using proteinase K-digested tail or toe biopsies as DNA templates and specific primers as follows: human NOTCH3 transgene (sense) 5′-cgatggaatgggtttccact-3′ and (antisense) 5′-aggcaggagcaggaaaagga-3′, Notch3 null allele (sense) 5′-tcgcc ttcttgacgagttct-3′ and (antisense) 5′-gcgatgcaatttcct cattt-3′; wild-type Notch3 allele (sense) 5′-ccatgaggatgctatctgtgac-3′ and (antisense) 5′-cacattggcacaagaatgagcc-3′, SM22α-lacZ (sense) 5′-gg atcgatctcgccatacagcgcg-3′ and (antisense) 5′-ccagacaccgaagc tactctcctt-3′ and TP1-nLacZ (sense) 5′-gatgcgcccatctacacc-3′ and (antisense) 5′-gctctggccttcctgta-3′. All mouse protocols were approved by the Animal Care and Use Committees of Paris, Ile de France.
Real-time RT–PCR analysis
Brain arteries, including arteries of the circle of Willis and the medium-sized branches, were dissected under scope, snap-frozen in liquid nitrogen and stored at −80°C. Arteries were homogenized on ice in SV RNA lysis buffer (Promega) using a polytron homogenizer, and RNA was extracted following the manufacturer's protocol (SV total RNA isolation system, Promega). Total RNA was converted to cDNA using the MMLV reverse transcriptase (Invitrogen). Each RNA was derived from two to three mice and checked for the absence of genomic DNA by PCR. Quantitative PCR was performed using the SYBR Green PCR master mix (Biorad) on a MyiQ™ Single-Color Real-Time PCR detection system (Biorad). cDNAs were amplified in triplicate using gene-specific primers showing efficient amplification (Supplementary Material, Table S1). To accurately determine the hNOTCH3 to mNotch3 mRNA ratios in brain arteries, we constructed a pBluescript recombinant plasmid containing one copy of the human NOTCH3 transgene and one copy of murine Notch3 cDNA transcript. The same series of dilutions of this plasmid were used to generate two different standard curves for each set of human NOTCH3 transgene and murine Notch3-specific primers. By using this plasmid containing the human and murine transcripts in a 1:1 ratio as template, equal quantities could be assigned to each dilution point of the two standard curves; brain artery cDNAs were amplified in the same assay and the expression level of both human and murine transcripts was calculated from these standard curves.
β-Galactosidase staining, histology and immunohistochemistry
Whole-mount staining for β-galactosidase activity was performed as described (18). For detailed localization of β-galactosidase activity, tissues were dehydrated through graded alcohols, embedded in paraffin, sectioned at 7 µm and counterstained with hematoxylin and eosin. Immunostaining with α-SM actin antibody (clone 1A4, Dako) or NOTCH3 [clone 1E4; (16)] was performed on paraffin sections using the avidin–biotin–peroxidase system (Vector Laboratories) as described (18). Detection of NOTCH3ECD accumulation was performed on paraffin sections pretreated with trypsin, with a concentrated 1E4 hybridoma supernatant (dilution 1:500) using a peroxidase-conjugated goat anti-mouse IgG1 (Jackson Immunoresearch).
High-resolution and transmission electron microscopy
Artery and surrounding brain tissue were dissected under scope, fixed in CARSON solution and embedded in Epon E812 resin. We observed semi-thin sections stained with toluidine blue under a DMR microscope (Leica) and ultrathin sections, stained with lead citrate and uranyl acetate, in a CM100 electron microscope (Philips) as described (18).
Supplementary Material is available at HMG Online.
We thank J.P. Rio for assistance with electron microscopy. This work was funded by grants from GIS-ANR Maladies Rares and National Institutes of Health (R01 NS054122) to A.J. and by grants from ANR JC05–41835 to M.C.T. M.M. is a recipient of a fellowship from the French Ministry of Education and Research and C.S. from CNRS (Bourse de Doctorat pour les Ingénieurs).
Conflict of Interest statement. No conflicts declared.