Insights into Autoregulation of Notch3 from Structural and Functional Studies of Its Negative Regulatory Region.

Article

Insights into Autoregulation of Notch3 from Structural and Functional Studies of Its Negative Regulatory Region Graphical Abstract

Authors Xiang Xu, Sung Hee Choi, ..., Christy Fryer, Stephen C. Blacklow

Correspondence [email protected]

In Brief Xu et al. report an X-ray structure of the negative regulatory region of human Notch3 in its autoinhibited conformation. The structure and accompanying functional studies demonstrate increased ligand-independent signaling for diseaseassociated mutations, uncover distinguishing structural features associated with increased basal activity of Notch3, and identify a conserved dimerization interface present in multiple Notch receptors.

Highlights d

Structure of the human Notch3 negative regulatory region in its autoinhibited state

d

Low intrinsic stability of autoinhibited Notch3 results in detectable basal activity

d

Structure reveals a conserved dimerization interface among Notch NRRs

Xu et al., 2015, Structure 23, 1–9 July 7, 2015 ª2015 Elsevier Ltd All rights reserved http://dx.doi.org/10.1016/j.str.2015.05.001

Accession Numbers 4ZLP

Please cite this article in press as: Xu et al., Insights into Autoregulation of Notch3 from Structural and Functional Studies of Its Negative Regulatory Region, Structure (2015), http://dx.doi.org/10.1016/j.str.2015.05.001

Structure

Article Insights into Autoregulation of Notch3 from Structural and Functional Studies of Its Negative Regulatory Region Xiang Xu,1,2 Sung Hee Choi,1,2 Tiancen Hu,3 Kittichoat Tiyanont,1,2 Roger Habets,4 Arjan J. Groot,4 Marc Vooijs,4 Jon C. Aster,5 Rajiv Chopra,3 Christy Fryer,3 and Stephen C. Blacklow1,2,5,* 1Department

of Cancer Biology, Dana Farber Cancer Institute, Boston, MA 02215, USA of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA 02115, USA 3Novartis Institutes for Biomedical Research, Cambridge, MA 02139, USA 4Department of Radiotherapy (MAASTRO)/GROW – School for Developmental Biology & Oncology, Maastricht University, Maastricht 6229, the Netherlands 5Department of Pathology, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA 02115, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.str.2015.05.001 2Department

SUMMARY

Notch receptors are transmembrane proteins that undergo activating proteolysis in response to ligand stimulation. A negative regulatory region (NRR) maintains receptor quiescence by preventing protease cleavage prior to ligand binding. We report here the X-ray structure of the NRR of autoinhibited human Notch3, and compare it with the Notch1 and Notch2 NRRs. The overall architecture of the autoinhibited conformation, in which three LIN12-Notch repeat (LNR) modules wrap around a heterodimerization domain, is preserved in Notch3, but the autoinhibited conformation of the Notch3 NRR is less stable. The Notch3 NRR uses a highly conserved surface on the third LNR module to form a dimer in the crystal. Similar homotypic interfaces exist in Notch1 and Notch2. Together, these studies reveal distinguishing structural features associated with increased basal activity of Notch3, demonstrate increased ligand-independent signaling for disease-associated mutations that map to the Notch3 NRR, and identify a conserved dimerization interface present in multiple Notch receptors.

INTRODUCTION The Notch signaling pathway influences numerous cell fate decisions during development and maintains tissue homeostasis in adults. Mammals have four Notch receptors and five canonical ligands, three homologous to Drosophila Delta (DLL1, DLL3, and DLL4), and two homologous to Drosophila Serrate (Jag1, Jag2). Whereas mammalian Notch1 and Notch2 are expressed in a wide variety of tissues and are essential for mammalian development, Notch3 expression is largely restricted to vascular smooth muscle. Thus, mice lacking Notch3 exhibit developmental defects limited primarily to vascular smooth muscle maturation and arterial specification (Domenga

et al., 2004). Aberrant Notch3 expression has also been linked to a variety of different disease states, including the hereditary stroke syndrome CADASIL and ovarian cancer (Cancer Genome Atlas Research Network, 2011; Joutel et al., 1996; Park et al., 2006). Normally, Notch receptors transmit signals by undergoing regulated proteolysis in response to transmembrane ligands presented on the surface of adjacent cells. The intrinsic resistance of Notch receptors to activating proteolysis is dependent on the integrity of a negative regulatory region (NRR), which encompasses a series of three LIN12-Notch repeats (LNRs) and a juxtamembrane heterodimerization domain (HD). The HD is cleaved during normal receptor maturation by a furin-like protease at a site called S1 (Logeat et al., 1998), but the NRR is resistant to further proteolysis in the absence of ligand (Gordon et al., 2007; Sanchez-Irizarry et al., 2004). Ligand stimulation induces receptor sensitivity to metalloprotease cleavage at a site called S2 (Brou et al., 2000; Groot et al., 2014; Mumm et al., 2000), which lies near the C-terminal end of the HD (Figure 1A). After metalloprotease cleavage, the truncated receptor, called NEXT, is primed for intramembrane cleavage at site S3 and additional sites by gamma-secretase, which releases the intracellular part of Notch (NICD) from the membrane. NICD migrates to the nucleus, where it assembles a transcriptional activation complex that turns on the expression of Notch-responsive genes (Kopan and Ilagan, 2009). The X-ray structures of the Notch1 and Notch2 NRRs show that autoinhibition results from the masking of the metalloprotease cleavage site in the HD by the first two LNR modules and the linker between them (Gordon et al., 2009, 2007). Moreover, mutations of the Notch1 NRR that are found in nearly half of human T-cell acute lymphoblastic leukemia/lymphomas lead to increased ligand-independent signaling (Weng et al., 2004), highlighting the importance of the NRR in stabilizing the quiescent, ‘‘off’’ state of the receptor. Here, we report the structure of the Notch3 NRR in its autoinhibited conformation. The structure reveals differences in packing interactions among the three NRRs that may account for the increased basal proteolytic sensitivity of Notch3. We also observe an increase in ligand-independent signaling associated with various point mutations that have been identified in different

Structure 23, 1–9, July 7, 2015 ª2015 Elsevier Ltd All rights reserved 1


Figure 1. Structure of the Notch3 NRR and Comparison with NRRs from Other Notch Receptors (A) Top: Domain organization of the Notch3 NRR. Bottom: Ribbon representation of the X-ray structure of the Notch3 NRR. The three LNR modules are colored in different shades of pink as in the schematic, and the HD domain is shaded cyan up to the furin cleavage site, and in darker blue C-terminal to the furin site. The three calcium ions are in green, disulfide bonds in yellow, and the sugar residues at the glycosylation site on the HD domain have a yellow carbon backbone with heteroatoms in CPK colors. The zoomed-in views shown in (B) and (C) are boxed in red. (B–D) Comparison among Notch1 (gray; PDB: 3ETO), Notch2 (gold; PDB: 2OO4), and Notch3 (multicolored) NRRs. (B) Zoomed-in view of the LNR-A interface with LNR-B, highlighting the divergence of Notch3 at P1408, and its impact on the conserved cluster of tryptophan residues. The sugars have been removed for clarity. (C) View focusing on the interface between the LNR-C module and the HD domain, illustrating the different packing arrangement in Notch3 compared to Notch1 and Notch2. (D) Close-up view of the N-acetylglucosamine residue attached to N1438 interacting with residues at the C-terminal end of helix 3 of the HD domain. See also Figure S1.

disease states, and identify a dimerization interface conserved among the Notch1, Notch2, and Notch3 receptors. RESULTS Structure Overview and Unique Features of the Notch3 NRR The structure of the Notch3 NRR was solved to 2.4 A˚ resolution by X-ray crystallography, using the Notch1 NRR (PDB: 3ETO) as a search model for molecular replacement (Table 1; see Experimental Procedures). The Notch3 NRR adopts a compact structure, with the three LNR modules enveloping the HD domain in a closed, autoinhibited conformation (Figure 1A). The domain architecture is similar to the arrangement seen in the structures of the human Notch1 and Notch2 NRRs (Gordon et al., 2009, 2007; Wu et al., 2010), and confirms that the structural basis for Notch3 autoinhibition results from the packing of the LNRs against the HD domain, precluding access of metalloprotease to the S2 cleavage site. Detailed comparison of the Notch3 NRR with the Notch1 and Notch2 NRRs reveals substantial differences among the structures at the interdomain interfaces (Figures 1B and 1C). The backbone root-mean-square differences (RMSDs) for superposition of the Notch3 NRR on that of Notch1 or Notch2 are 2.25 and 2.29 A˚, respectively, compared with a 1.65 A˚ RMSD for the Notch1/Notch2 NRR superposition (Figure S1).

One important distinguishing feature of the Notch3 structure occurs at the interface between LNR-A and LNR-B (Figure 1B). Three absolutely conserved tryptophan residues (W1412, W1425, and W1434 in Notch3) play a critical role in stabilizing this interface. In Notch1 and Notch2, this aromatic cluster is further reinforced by an LNR-A histidine residue (H1471 in Notch1 and H1446 in Notch2), which engages the third tryptophan of the cluster in an aromatic p-stacking interaction. In Notch3, a proline residue (P1408) substitutes for histidine and makes only limited van der Waals contact with W1434, suggesting that this interface is less stable in the Notch3 NRR. Another difference that distinguishes the Notch3 NRR is the nature of the interface between LNR-C and the first helix of the N-terminal part of the HD domain (HD-N) (Figure 1C). In both Notch1 and Notch2, there is a salt bridge formed between an aspartate in LNRC and an arginine from the external face of the helix (D1534 to R1595 in Notch1, and D1506 to R1567 in Notch2). This linkage defines the interdomain orientation of the two modules and establishes the distance of closest approach between the helix and LNR-C. In Notch3, the analogous interaction becomes a hydrogen bond between a glutamate (E1472) on LNR-C and a glutamine (Q1533) on the helix. The shorter glutamine side chain at that position allows the helix to pack more tightly against LNR-C. This close-packed arrangement is further stabilized by an additional interaction between the same E1472 residue of LNR-C and R1534, an arginine residue at the next position of the helix. An additional striking feature of the NRR from Notch3 is the position of the N-linked sugars attached to N1438 of LNR-B

2 Structure 23, 1–9, July 7, 2015 ª2015 Elsevier Ltd All rights reserved


was prepared in bacteria, and is thus unglycosylated, the Notch1 NRR solved in complex with an inhibitory Fab (PDB: 3L95) was expressed in insect cells, and is glycosylated at this site (Wu et al., 2010). The sugar residues that are modeled in the structure also come within hydrogen-bonding distance of helix 3, and may serve to increase the stability of the ‘‘off’’ state of these proteins as well. The other main site of structural divergence among the three NRRs lies at the convergence point of the LNR-B and LNR-C modules with the HD domain (Figure S1D). In the Notch3 NRR, the connector between strand b4 and helix a3 containing the disulfide knuckle of the HD domain is not helical as in Notch1 and Notch2, but is instead an extended loop that projects farther from the core of the protein. The structural diversity in this region may also contribute to differential responsiveness of various receptors to different canonical ligands.

Table 1. X-Ray Data Collection and Refinement Statistics Notch3 NRR X-Ray Data Collection X-Ray source Wavelength (A˚)

24-ID-E (NECAT) 0.9792

Resolution (A˚)

49.18–2.48 (2.57–2.48)b

Space group

C2

Cell dimensions a, b, c (A˚)

122.5, 64.3, 83.1

a, b, g ( )

90.0, 105.2, 90.0

Rmergea (%)

13.2 (78.6)

CC1/2

99.2 (68.2)

I/s (I)

13.27 (2.23)

Completeness (%)

99.99 (99.87)

Redundancy

7.5 (7.5)

Mosaicity

0.15

Model Refinement Unique reflections

22,317 (2,231)

Subunits/asymmetric unit

2

Rwork/Rfree

0.1829/0.2069 (0.2595/0.3157)

No. of non-hydrogen atoms

3,701

Macromolecules

3,450

Ligand/ion (Ca2+)

96/6

Water

155

Protein residues

459

Average B factor (A˚2)

53.2

Macromolecules

53.0

Ligand/ion

78.3

Water

41.3

RMSD Bond lengths (A˚)

0.004

Bond angles ( )

0.83

Ramachandran plot Favored (%)

97

Allowed (%)

3

Outliers (%) 0.22 P P P P a Rmerge = hkl ijIi(hkl)i j/ hkl iIi(hkl). b Numbers in parentheses refer to the highest-resolution shell.

(Figure 1D). Both the first and second sugars of the chain exhibit well-defined electron density, assigned to two N-acetylglucosamine (GlcNAc) residues. Additional density is present for two more sugars, which are modeled as mannose (Figure S1C). The first GlcNAc forms several hydrogen bonds with nearby residues, including the side chain of R1619 and the backbone carbonyl oxygen of V1617. These two residues are located at the C-terminal end of helix 3 of the HD domain, and it is possible that these hydrogen-bonding interactions between LNR-B and helix 3 help in stabilizing the closed conformation of the protein. Interestingly, both Notch1 and Notch2 have a consensus sugar modification site at the asparagine residue immediately preceding the first cysteine of LNR-B, which is positioned at the N-terminal end of helix 3. Although the Notch2 protein used to solve the X-ray structure

Structural Basis for Increased Basal Activity of Notch3 Our Notch3 NRR structure has two molecules in the asymmetric unit (ASU), with an RMSD of less than 1 A˚ between the two copies (Figure 2A). The most striking difference between the two copies resides in the LNR-A domain, which is partially disordered in one molecule (named mol-B), but is well defined in the other (named mol-A). The position of LNR-A is constrained by packing interactions with a symmetry-related molecule in mol-A but not in mol-B, suggesting that LNR-A is more dynamic in the Notch3 NRR than anticipated. We revisited the question of LNR-A positional flexibility for the other Notch NRRs by calculating normalized B-factor values for the LNR-A domain relative to the rest of the structure for each Notch NRR in the PDB (Table S1). Although the absolute value of the B factor is affected by multiple influences including the quality of the data set, the resolution, and crystal-packing interactions, the normalized B-factor ratio for the LNR-A domain of the Notch3 NRR is substantially higher than in other NRR structures (Table S1), consistent with the idea that the LNR-A module of Notch3 is more dynamic than its counterparts in Notch1 and Notch2. We tested whether Notch3 exhibits a higher level of basal (ligand-independent) signaling activity than either Notch1 or Notch2 in U2OS cells using a reporter gene assay. These experiments were performed using isogenic stable cell lines with chimeric receptors (Gordon et al., 2009) that contain an intracellular Gal4-responsive element to eliminate potential confounding effects from endogenous receptors or differences in expression levels (Figure S2A, bottom schematic). We examined the responsiveness of the different receptors to the DLL1 ligand immobilized on the surface of the culture dish or on OP9 cells (Figures 2B and 2C), as well as to the Jag2 ligand expressed on NIH3T3cells (Figure S2B). In assays performed without ligand-expressing cells, with control cells that do not express the ligand, or with immobilized immunoglobulin G (IgG) in place of DLL1, the basal reporter activity of the Notch3-expressing cells is substantially higher than that seen in Notch1- or Notch2-expressing cells, even though the strength of the signal in the presence of either DLL1 or Jag2 is comparable for all three receptors (Figures 2B and 2C; Figure S2B). The amount of the mature Notch3 protein is comparable with that of Notch1 and less than that of Notch2 in the compared isogenic lines, and thus the increased basal activity of Notch3 cannot be attributed to an increased abundance of the protein (Figure S2C). Together, the increased level of



Figure 2. Notch3 Shows Elevated Basal Activity (A) Superposition of the two copies of the Notch3 NRR in the asymmetric unit of the crystal. The copy labeled Mol-B shows disorder in the LNR-A module, shown in the close-up view (boxed). (B and C) Notch3 shows higher amounts of reporter gene activity in the absence of ligand than either Notch1 or Notch2. Isogenic stable cell lines expressing Notch1, Notch2, or Notch3 chimeric receptors that report on a Gal4-response element (Malecki et al., 2006) were cultured either on plates coated with IgG control, or the DLL1 ligand (B), or in co-culture assays with OP9 stromal cells alone or stably expressing DLL1 (C). The firefly luciferase activity, normalized to a renilla luciferase control (see Experimental Procedures) was measured in cell lysates after 24 hr. In both (B) and (C), assays were also performed in the presence of the gamma-secretase inhibitor compound E (GSI) as a control. Luciferase activity is normalized to the Notch1 signal with IgG control (B), or with OP9 control cells (C). Error bars represent the SE of three replicates. Using a one-way ANOVA, statistically significant differences are indicated with asterisks: ***p < 103; ****p < 104. (D) Urea dissociation assay. Conditioned media from 293T cells expressing either the Notch1 or Notch3 NRR were immunoprecipitated with antiHA (a-HA)-coupled beads, followed by 30 min of incubation at room temperature in buffer with different concentrations of urea (0–4 M). Subunit dissociation was evaluated by SDS-PAGE followed by western blot analysis with anti-FLAG and anti-HA antibodies, which are specific for the N- and C-terminal subunits, respectively. See also Figure S2.

ligand-independent Notch3 proteolysis and the increased positional disorder of the LNR-A domain in the Notch3 NRR crystal suggest that the autoinhibited conformation is more dynamic and/or less stable in Notch3 than in Notch1 or Notch2. To determine experimentally whether the Notch3 NRR was less stable than the Notch1 NRR, we performed a subunit dissociation assay in which the integrity of the furin-processed NRR heterodimer was monitored as a function of added urea (Malecki et al., 2006). Whereas the subunits of the Notch1 NRR heterodimer remain non-covalently associated at concentrations up to 4 M urea as reported previously (Malecki et al., 2006), the Notch3 NRR is substantially more sensitive to urea-induced dissociation, indicating that the Notch3 NRR is less stable than the Notch1 NRR (Figure 2D). Analysis of Disease-Associated Mutations Mapping to the Notch3 NRR Mutations of Notch3 are responsible for the hereditary stroke syndrome known as CADASIL. Although the vast majority of the CADASIL mutations lie in the epidermal growth factor

(EGF)-like repeats, there is one CADASIL mutation in the Notch3 HD, L1515P (Fouillade et al., 2008), which lies on the central b strand of the HD domain (Figure 3A). In addition, a L1519P mutation has been reported to segregate with multiple affected individuals in a pedigree with familial autosomal-dominant infantile myofibromatosis (Martignetti et al., 2013), and an acquired S1580L mutation of Notch3 has been identified in the T-cell acute lymphoblastic leukemia/lymphoma cell line TALL1, (Cancer Genome UK). L1519P is located in the linker region between b1 and b2 of the HD (Figure 3A), while S1580 is located in b-strand 4 of the hydrophobic core, with its side-chain oxygen atom within hydrogen-bonding distance of the backbone carbonyl oxygen atoms of L1519 and P1520. To test the effects of these mutations on ligand-independent and ligand-dependent signaling, we introduced each of these three mutations in the context of the full-length Notch3 protein (Figure S2A, top schematic), and tested the expression of a luciferase reporter gene under control of the Notchresponsive TP1 promoter in transient transfection assays (Kurooka et al., 1998). The reporter gene assays show that the S1580L mutation results in a large increase in ligand-independent signaling. The L1515P mutation, as previously reported, also produces a large increase in ligand-independent signaling



Figure 3. Analysis of Notch3 DiseaseAssociated Mutations (A) Cartoon close-up highlighting disease-associated sites of mutations and their interactions with nearby residues. (B) Reporter gene assay examining the ligand-independent activity and Jag2-stimulated activity of full-length wild-type Notch3, and the S1580L, L1515P, and L1519P mutated forms of Notch3. The firefly luciferase gene was under control of the TP1 promoter (Kurooka et al., 1998) and the Jag2 protein was expressed in stably transfected 3T3 cells (Luo et al., 1997). The ratio of firefly to control renilla luciferase activity is plotted both in the absence and presence of the gamma-secretase inhibitor compound E (GSI). Luciferase activity is normalized to the signal observed with vector alone co-cultured with 3T3 cells. Error bars represent the SE of three replicates. Using a oneway ANOVA, statistically significant differences are indicated with asterisks: *p < 0.05; ****p < 104.

(Fouillade et al., 2008), whereas L1519P results in a modest, but statistically significant increase in ligand-independent signaling (Figure 3B).

respond to ligand comparably with the wild-type receptor (Figure 5). DISCUSSION

Dimerization of the Notch3 NRR The Notch3 NRR crystal contains two copies of the protein in the ASU. The interface in the crystallographic dimer is entirely derived from surface residues on the LNR-C module (Figure 4A), which bury 1,088 A˚2 of surface area at the interface. The dimer interface present in the Notch3 NRR structure closely resembles the packing interface seen in all Notch NRR structures deposited in the PDB (Figures 4A and S3A), including two different X-ray structures of the isolated Notch1 NRR, a structure of a Notch1 NRR-Fab complex, and the structure of the isolated Notch2 NRR (PDB: 3ETO, 3I08, 3L95, and 2OO4, respectively). Multiple sequence alignment shows that this surface patch on LNR-C is highly conserved among all Notch NRRs (Figure 4B). Two key residues in this interface of the Notch3 NRR dimer are Y1474 and H1478, which form symmetric sandwich-like p-p interactions (Figure 4A, right panels). These reciprocal interactions are buttressed on each molecule by Y1471. Other important residues at the interface include R1483, D1485, and Q1486 from both molecules. Equilibrium sedimentation studies show that the Notch3 NRR does not conform to a single-species model, and is best fit by a monomer-dimer equilibrium model with an estimated dimerization Kd of 30 mM (Figure S3B). To examine whether the LNR-C dimerization interface contributes functionally to Notch3 signaling, we created four mutated forms of the full-length Notch3 receptor designed to disrupt the dimer interface, Y1474E, Y1474R, H1478E, and H1478R, and tested the effect of these mutations on ligand-independent and ligand-dependent signaling in both plated ligand and cell co-culture assays. Both mutations of H1478 lead to significant increases in ligand-independent reporter gene activity compared with the unmutated receptor (Figures 5 and S4), suggesting that the residues at the dimerization interface play a role in maintaining the quiescent state of the receptor. Importantly, cells expressing the mutated receptors appear to reach the cell surface normally (Figure S4), and

Notch receptor quiescence is mediated by the NRR, which encompasses both the maturation cleavage site processed by a furin-like proprotein convertase and the metalloprotease cleavage site that undergoes the key activating cleavage in response to ligand stimulation. The structural and functional studies of the Notch3 NRR reported here deepen the understanding of the architectural features shared among Notch NRRs that encode receptor autoinhibition, and highlight differences among the Notch1, Notch2, and Notch3 NRRs that distinguish their sensitivity to receptor activation. Several lines of evidence support the conclusion that autoinhibition of the Notch3 NRR is less stringent than that of the NRR regions of Notch1 or Notch2. First, in the X-ray structure of the Notch3 NRR, the LNR-A domain has a higher average B factor than the other domains of the protein (Table S1), and is more disordered than the analogous domain from Notch1 or Notch2 in the other NRR structures. In addition, the purified Notch3 NRR undergoes subunit dissociation at a much lower urea concentration, indicating that the Notch3 NRR is less stable than the Notch1 or Notch2 NRRs. Likewise, the reporter gene assay testing ligand-independent and ligand-dependent signaling shows that the basal, ligand-independent signal of Notch3 in the reporter assay using the Gal4 fusions with identical intracellular regions is significantly stronger than the basal signal from Notch1 or Notch2, even though the ligand-dependent signal from all three receptors is comparable in the reporter assay. It remains unclear, however, why the basal activity of full-length Notch3 receptors in transient transfection assays is not as consistent, nor why gamma-secretase inhibition (e.g. Figure 3B) does not consistently suppress basal activity of full-length receptors. One possibility is that the use of the Gal4 reporter system improves assay signal-to-noise ratio by eliminating the confounding effects of endogenous receptor signaling. Another possibility is that the Notch3 intracellular domain, which is not part of the molecules used to examine basal activity (Figure 2),



Figure 4. A Conserved Dimerization Interface Is Observed in Different NRR Structures (A) Comparison of dimerization interfaces seen in the Notch3 NRR, the Notch1 NRR (PDB: 3ETO), and the Notch2 NRR (PDB: 2OO4). The left panels illustrate the mode of interaction between the two molecules, and the right panels, which represent the boxed regions, highlight conserved features among the three interfaces. (B) Multiple sequence alignment showing conservation of the residues at the interface (left), and a surface representation of the Notch3 NRR structure (right) colored on a sliding scale from maroon to teal blue according the extent of sequence conservation using the program ConSurf (Ashkenazy et al., 2010). See also Figure S3.

may contribute to restraint of basal signaling in the full-length protein, compensating in part for the reduced stability of the Notch3 NRR. Increased basal Notch3 signaling activity might have particular relevance in pathogenic conditions where an increased amount of Notch3 is expressed. For example, NOTCH3 lies within a locus that is amplified in high-grade ovarian serous adenocarcinomas (Cancer Genome Atlas Research Network, 2011; Park et al., 2006), and Notch3 has also been implicated

in the pathogenesis of certain lung cancers (Konishi et al., 2007). Both loss-of-function and gain-offunction mutations of Notch3 receptors are also associated with human developmental disorders and cancer (Aster et al., 2011; Gridley, 2003; Joutel et al., 1996). The three mutations known to map to the HD of the Notch3 NRR include the CADASIL mutation L1515P, an infantile myofibromatosis mutation L1519P, and the S1580L mutation found in the TALL1 leukemia cell line. All three mutations tested here are associated with ligand-independent gain of function, suggesting that inhibitory antibodies capable of blocking ligand-independent signaling by binding the NRR might have therapeutic utility in these diseases. Finally, the presence of a conserved dimerization interface among Notch1, Notch2, and Notch3, combined with the analytical ultracentrifugation data that fit best to a monomer-dimer equilibrium model, suggest that it possesses functional importance. The restricted distribution of molecules at the plasma membrane, the observation of receptor clustering in signaling complexes (Ahimou et al., 2004; Bardot et al., 2005; Nichols et al., 2007), and the presence of an additional weak dimerization interface within the intracellular ankyrin repeat domain of Notch (Allgood and Barrick, 2011; Nam et al., 2006, 2007) also point toward a potential role for dimerization in modulating signal transduction. Consistent with this idea, reporter gene assays suggest that autoinhibition is partially disrupted by mutations that destabilize the dimerization interface. However, a more detailed understanding of the role of cell-surface Notch dimers will require future study.



Figure 5. Effect of Mutations of Conserved Residues in the Dimer Interface upon Signaling (A) Reporter assay using plated ligand. Wild-type or mutated Notch3 receptors were cultured either on plates coated with IgG control or the DLL1 ligand, and treated with either DMSO carrier or a gamma-secretase inhibitor. (B) Reporter assay using co-cultured cells expressing the DLL1 ligand. Wildtype or mutated Notch3 receptors were co-cultured either with OP9 stromal cells alone or with OP9 cells stably expressing the DLL1 ligand, and treated with either DMSO carrier or a gamma-secretase inhibitor. In both (A) and (B), firefly luciferase activity, normalized to a renilla luciferase control, was measured in cell lysates after 24 hr. Luciferase activity is normalized to the signal observed for cells transfected with full-length Notch3 cultured on the plated IgG control (A), or with OP9 cells alone (B). Error bars represent the SE of three replicates. Using a one-way ANOVA, statistically significant differences are indicated with asterisks: *p < 0.05; **p < 0.01. See also Figure S4.

EXPERIMENTAL PROCEDURES Protein Expression and Purification For crystallization, the Notch3 NRR was subcloned into a modified pcDNAseries plasmid encoding the IgG kappa signal sequence, followed by the Notch3 NRR (1,378–1,640), and an in-frame gly-ser-his6 tag at the C terminus. The encoded protein was expressed and secreted into the conditioned media by HEK293S GnTi cells (Reeves et al., 2002). The medium was collected after 4 days, and the NRR was bound to Ni-NTA beads (Qiagen) over a 3-hr incubation period at 4 C. After a wash with 50 mM HEPES buffer containing 500 mM NaCl and 20 mM imidazole, bound protein was eluted in 50 mM HEPES buffer containing 500 mM NaCl and 250 mM imidazole. After buffer exchange into 50 mM Tris (pH 8.0) containing 50 mM NaCl and 10 mM CaCl2, the Notch3 NRR protein was passed over a Hi-trap Q HP column and eluted using an NaCl gradient. The recovered Notch3 NRR was incompletely processed into two subunits by furin, as judged by SDS-PAGE. Thus, the protein was treated with furin (30 U/mg) at room temperature for 6 hr, and the fully processed protein was further purified by size-exclusion chromatography on a Superdex 75 HR 10/30 column after buffer exchange into 20 mM HEPES buffer (pH 7.4) containing 150 mM NaCl and 10 mM CaCl2. The protein was concentrated to 45 mg/ml for initial crystallization screening.

Crystallization, Data Collection, and Structural Refinement Crystallization trials were carried out using the sitting drop vapor-diffusion method at both 22 C and 4 C. Initial crystals were obtained in two conditions: one at 4 C containing 0.2 M ammonium sulfate, 15% PEG4000, and 0.1 M trisodium citrate (pH 5.6), and the other at 22 C containing 20% isopropanol, 20% PEG4000, and 0.1 M trisodium citrate (pH 5.6). After iterative refinement, final crystals were grown at 22 C by mixing equal volumes of Notch3 NRR at 5 mg/ml with a precipitant solution containing 0.2 M sodium acetate, 20% PEG4000, and 0.1 M trisodium citrate at pH 5.4. Plate-like crystals typically grew within 2 weeks reaching a maximal size of 0.05 3 0.5 3 0.5 mm3. For data collection, a single crystal was flash-frozen from a precipitant solution containing 20% (v/v) glycerol. Diffraction data were collected from a flash-cooled crystal at 100 K using the 24ID-E beamline at Argonne National Laboratory and were processed, integrated, and scaled together with HKL2000 (Otwinowski and Minor, 1997). The criterion used for data inclusion was based on the value of CC1/2 (Karplus and Diederichs, 2012). The Notch3 NRR structure was solved by molecular replacement as implemented in Phenix (Adams et al., 2010) using a Notch1 NRR structure (PDB: 3ETO) as the search model. Structure refinement was carried out using conjugate-gradient energy minimization, torsion-constrained molecular dynamics-simulated annealing, group B-factor refinement, individual B-factor refinement, and TLS (Translation/Libration/Screw) protocols in Phenix with 5% of the reflections omitted for free R-factor calculation. Electron density peaks in difference Fourier maps at a height of above 3s were assigned as water molecules in later refinement stages if they had reasonable geometry in relation to hydrogen bond donors or acceptors and their B factors did not rise above 50 A˚2 during subsequent refinement. Model building was performed in Coot (Emsley and Cowtan, 2004) guided by sA-weighted 2Fo Fc maps, Fo Fc maps, and composite omit maps. All structure figures were generated using PyMOL v1.7 (Schrodinger, 2010). Luciferase Reporter Assay Luciferase reporter gene assays were carried out using methods previously described (Gordon et al., 2009). For the studies in Figures 2 and S2B, experiments were performed using isogenic stable cell lines with chimeric Notch1-, Notch2-, or Notch3-Gal4 fusions, and the reporter was a Gal4responsive element. For studies testing disease-associated or dimer interface mutations, the full-length Notch3 protein was used in transient transfections of U2OS cells, and the reporter was the TP1 Notch-responsive element. In plated ligand assays, the various Notch-expressing cells were cultured on tissue culture plates treated with either soluble DLL1 ligand (spanning from the N terminus through EGF-like repeat 5 [Andrawes et al., 2013]) or an IgG control protein in the absence or presence of a gamma-secretase inhibitor, compound E (1 mM). In the co-culture assays, OP9 cells alone or stably expressing DLL1, or 3T3 cells alone or stably expressing Jag2, were used as signal-sending cells, again in the absence or presence of a gamma-secretase inhibitor, compound E (1 mM). Urea Dissociation Assay Urea dissociation assays were performed essentially as described previously (Malecki et al., 2006). The Notch1 (Malecki et al., 2006) or Notch3 NRR (S1382S1640, flanked by N-terminal FLAG and C-terminal hemagglutinin [HA] tags) was recovered from conditioned media using anti-HA antibody-conjugated beads in 50 mM Tris buffer (pH 7.5) containing 150 mM NaCl and 0.1% NP40, and subjected to varying concentrations of urea for a period of 30 min before recovery of the beads by centrifugation. After three rapid washes in the incubation buffer, the proteins retained on the beads were analyzed by western blotting. Analytical Ultracentrifugation Analytical ultracentrifugation was performed in six-sectored cells using a Beckman Coulter Optima XL-A analytical ultracentrifuge equipped with an An-60 Ti analytical rotor. Before centrifugation, wild-type Notch3 NRR protein was extensively dialyzed into 25 mM HEPES buffer (pH 7.5) containing 150 mM NaCl and 5mM CaCl2. Protein concentrations of 3, 9, and 27 mM were prepared by dilution of a stock solution with the dialysate, and the dialysate was used in the three reference channels. Samples were centrifuged at 14,000 and 18,000 rpm at 4 C until equilibrium was reached, as judged by the stability



of absorption scans taken at wavelengths of 230, 280, and 290 nm over two consecutive 2-hr intervals. Data for final analysis were acquired by averaging ten scans at each wavelength using 1-nm steps, and only data in the linear range of the instrument (between 0.01 and 1 absorbance unit) were used for subsequent analysis. Global fitting was carried out using the Origin 6.0 software package comparing fits to distinguish between a single-species model and a monomer-dimer equilibrium, and to estimate a value for the dimerization Kd. Plots comparing fits to a monomer-dimer equilibrium with a 30-mM value of Kd (left panel) or a single-species monomer (right panel) are shown for 280-nm scan of the 9-mM sample spun at 18,000 rpm in Figure S3B.

involved in Notch signaling: the role of the disintegrin-metalloprotease TACE. Mol. Cell 5, 207–216.

Immunofluorescence and Flow Cytometry Detection of various Notch proteins on the surface of U2OS cells was carried out using a mouse IgG anti-Notch3 antibody A13 (Li et al., 2008), followed by secondary detection with goat anti-mouse IgG coupled to Alexa546. Flow cytometry was carried out using the same antibodies, using a BD Accuri C6 instrument. The data were analyzed using BD Accuri C6 software.

Fouillade, C., Chabriat, H., Riant, F., Mine, M., Arnoud, M., Magy, L., Bousser, M.G., Tournier-Lasserve, E., and Joutel, A. (2008). Activating NOTCH3 mutation in a patient with small-vessel-disease of the brain. Hum. Mutat. 29, 452.

ACCESSION NUMBERS Coordinates of the Notch3 NRR have been deposited in the PDB (PDB: 4ZLP). SUPPLEMENTAL INFORMATION Supplemental Information includes four figures and one table and can be found with this article online at http://dx.doi.org/10.1016/j.str.2015.05.001. ACKNOWLEDGMENTS This work was supported by NIH grant R01 CA092433 (to S.C.B.), and P01 CA119070 (to J.C.A. and S.C.B.). M.V., A.J., and R.H. are supported by European Research Council ERC grants 208259 and 334987. T.H., R.C., and C.F. are employees of Novartis, and this work was supported in part by the Dana Farber – Novartis Drug Discovery Program. Received: January 7, 2015 Revised: May 1, 2015 Accepted: May 4, 2015 Published: June 4, 2015 REFERENCES Adams, P.D., Afonine, P.V., Bunkoczi, G., Chen, V.B., Davis, I.W., Echols, N., Headd, J.J., Hung, L.W., Kapral, G.J., Grosse-Kunstleve, R.W., et al. (2010). PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66, 213–221. Ahimou, F., Mok, L.P., Bardot, B., and Wesley, C. (2004). The adhesion force of Notch with Delta and the rate of Notch signaling. J. Cell Biol. 167, 1217–1229. Allgood, A.G., and Barrick, D. (2011). Mapping the Deltex-binding surface on the notch ankyrin domain using analytical ultracentrifugation. J. Mol. Biol. 414, 243–259. Andrawes, M.B., Xu, X., Liu, H., Ficarro, S.B., Marto, J.A., Aster, J.C., and Blacklow, S.C. (2013). Intrinsic selectivity of Notch 1 for Delta-like 4 over Delta-like 1. J. Biol. Chem. 288, 25477–25489. Ashkenazy, H., Erez, E., Martz, E., Pupko, T., and Ben-Tal, N. (2010). ConSurf 2010: calculating evolutionary conservation in sequence and structure of proteins and nucleic acids. Nucleic Acids Res. 38, W529–W533. Aster, J.C., Blacklow, S.C., and Pear, W.S. (2011). Notch signalling in T-cell lymphoblastic leukaemia/lymphoma and other haematological malignancies. J. Pathol. 223, 262–273. Bardot, B., Mok, L.P., Thayer, T., Ahimou, F., and Wesley, C. (2005). The Notch amino terminus regulates protein levels and Delta-induced clustering of Drosophila Notch receptors. Exp. Cell Res. 304, 202–223. Brou, C., Logeat, F., Gupta, N., Bessia, C., LeBail, O., Doedens, J.R., Cumano, A., Roux, P., Black, R.A., and Israel, A. (2000). A novel proteolytic cleavage

Cancer Genome Atlas Research Network. (2011). Integrated genomic analyses of ovarian carcinoma. Nature 474, 609–615. Domenga, V., Fardoux, P., Lacombe, P., Monet, M., Maciazek, J., Krebs, L.T., Klonjkowski, B., Berrou, E., Mericskay, M., Li, Z., et al. (2004). Notch3 is required for arterial identity and maturation of vascular smooth muscle cells. Genes Dev. 18, 2730–2735. Emsley, P., and Cowtan, K. (2004). Coot: model-building tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132.

Gordon, W.R., Vardar-Ulu, D., Histen, G., Sanchez-Irizarry, C., Aster, J.C., and Blacklow, S.C. (2007). Structural basis for autoinhibition of Notch. Nat. Struct. Mol. Biol. 14, 295–300. Gordon, W.R., Roy, M., Vardar-Ulu, D., Garfinkel, M., Mansour, M.R., Aster, J.C., and Blacklow, S.C. (2009). Structure of the Notch1-negative regulatory region: implications for normal activation and pathogenic signaling in T-ALL. Blood 113, 4381–4390. Gridley, T. (2003). Notch signaling and inherited disease syndromes. Hum. Mol. Genet. 12, R9–R13. Groot, A.J., Habets, R., Yahyanejad, S., Hodin, C.M., Reiss, K., Saftig, P., Theys, J., and Vooijs, M. (2014). Regulated proteolysis of NOTCH2 and NOTCH3 receptors by ADAM10 and presenilins. Mol. Cell. Biol. 34, 2822– 2832. Joutel, A., Corpechot, C., Ducros, A., Vahedi, K., Chabriat, H., Mouton, P., Alamowitch, S., Domenga, V., Cecillion, M., Marechal, E., et al. (1996). Notch3 mutations in CADASIL, a hereditary adult-onset condition causing stroke and dementia. Nature 383, 707–710. Karplus, P.A., and Diederichs, K. (2012). Linking crystallographic model and data quality. Science 336, 1030–1033. Konishi, J., Kawaguchi, K.S., Vo, H., Haruki, N., Gonzalez, A., Carbone, D.P., and Dang, T.P. (2007). Gamma-secretase inhibitor prevents Notch3 activation and reduces proliferation in human lung cancers. Cancer Res. 67, 8051–8057. Kopan, R., and Ilagan, M.X. (2009). The canonical Notch signaling pathway: unfolding the activation mechanism. Cell 137, 216–233. Kurooka, H., Kuroda, K., and Honjo, T. (1998). Roles of the ankyrin repeats and C-terminal region of the mouse notch1 intracellular region. Nucleic Acids Res. 26, 5448–5455. Li, K., Li, Y., Wu, W., Gordon, W.R., Chang, D.W., Lu, M., Scoggin, S., Fu, T., Vien, L., Histen, G., et al. (2008). Modulation of Notch signaling by antibodies specific for the extracellular negative regulatory region of NOTCH3. J. Biol. Chem. 283, 8046–8054. Logeat, F., Bessia, C., Brou, C., LeBail, O., Jarriault, S., Seidah, N.G., and Israel, A. (1998). The Notch1 receptor is cleaved constitutively by a furin-like convertase. Proc. Natl. Acad. Sci. USA 95, 8108–8112. Luo, B., Aster, J.C., Hasserjian, R.P., Kuo, F., and Sklar, J. (1997). Isolation and functional analysis of a cDNA for human Jagged2, a gene encoding a ligand for the Notch1 receptor. Mol. Cell. Biol. 17, 6057–6067. Malecki, M.J., Sanchez-Irizarry, C., Mitchell, J.L., Histen, G., Xu, M.L., Aster, J.C., and Blacklow, S.C. (2006). Leukemia-associated mutations within the NOTCH1 heterodimerization domain fall into at least two distinct mechanistic classes. Mol. Cell. Biol. 26, 4642–4651. Martignetti, J.A., Tian, L., Li, D., Ramirez, M.C., Camacho-Vanegas, O., Camacho, S.C., Guo, Y., Zand, D.J., Bernstein, A.M., Masur, S.K., et al. (2013). Mutations in PDGFRB cause autosomal-dominant infantile myofibromatosis. Am. J. Hum. Genet. 92, 1001–1007. Mumm, J.S., Schroeter, E.H., Saxena, M.T., Griesemer, A., Tian, X., Pan, D.J., Ray, W.J., and Kopan, R. (2000). A ligand-induced extracellular cleavage regulates gamma-secretase-like proteolytic activation of Notch1. Mol. Cell 5, 197–206.



Nam, Y., Sliz, P., Song, L., Aster, J.C., and Blacklow, S.C. (2006). Structural basis for cooperativity in recruitment of MAML coactivators to Notch transcription complexes. Cell 124, 973–983.

restricted and homogeneous N-glycosylation by a tetracycline-inducible N-acetylglucosaminyltransferase I-negative HEK293S stable mammalian cell line. Proc. Natl. Acad. Sci. USA 99, 13419–13424.

Nam, Y., Sliz, P., Pear, W.S., Aster, J.C., and Blacklow, S.C. (2007). Cooperative assembly of higher-order Notch complexes functions as a switch to induce transcription. Proc. Natl. Acad. Sci. USA 104, 2103–2108.

Sanchez-Irizarry, C., Carpenter, A.C., Weng, A.P., Pear, W.S., Aster, J.C., and Blacklow, S.C. (2004). Notch subunit heterodimerization and prevention of ligand-independent proteolytic activation depend, respectively, on a novel domain and the LNR repeats. Mol. Cell. Biol. 24, 9265–9273.

Nichols, J.T., Miyamoto, A., Olsen, S.L., D’Souza, B., Yao, C., and Weinmaster, G. (2007). DSL ligand endocytosis physically dissociates Notch1 heterodimers before activating proteolysis can occur. J. Cell Biol. 176, 445–458. Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326. Park, J.T., Li, M., Nakayama, K., Mao, T.L., Davidson, B., Zhang, Z., Kurman, R.J., Eberhart, C.G., Shih Ie, M., and Wang, T.L. (2006). Notch3 gene amplification in ovarian cancer. Cancer Res. 66, 6312–6318. Reeves, P.J., Callewaert, N., Contreras, R., and Khorana, H.G. (2002). Structure and function in rhodopsin: high-level expression of rhodopsin with

Schrodinger, L.L.C.. (2010). The PyMOL Molecular Graphics System, Version 1.3r1. Weng, A.P., Ferrando, A.A., Lee, W., Morris, J.P.t., Silverman, L.B., SanchezIrizarry, C., Blacklow, S.C., Look, A.T., and Aster, J.C. (2004). Activating mutations of NOTCH1 in human T cell acute lymphoblastic leukemia. Science 306, 269–271. Wu, Y., Cain-Hom, C., Choy, L., Hagenbeek, T.J., de Leon, G.P., Chen, Y., Finkle, D., Venook, R., Wu, X., Ridgway, J., et al. (2010). Therapeutic antibody targeting of individual Notch receptors. Nature 464, 1052–1057.


Structural insights into regulatory mechanisms of MO25-mediated kinase activation.

The structural and functional characterization of human RecQ4 reveals insights into its helicase mechanism.

Biophysical studies suggest a new structural arrangement of crotoxin and provide insights into its toxic mechanism.

Structural insights into the autoregulation and cooperativity of the human transcription factor Ets-2.

Structural and Functional Studies of Pavine N-Methyltransferase from Thalictrum flavum Reveal Novel Insights into Substrate Recognition and Catalytic Mechanism.

Functional Genomic Insights into Regulatory Mechanisms of High-Altitude Adaptation.

Structural and functional insights into the E3 ligase, RNF126.

Structural and functional insights into the lipopolysaccharide ABC transporter LptB2FG.

Functional and structural insights into the piRNA factor Maelstrom.

TFPI2: Structural insights from molecular modeling studies.

Structural and Functional Studies of H. seropedicae RecA Protein - Insights into the Polymerization of RecA Protein as Nucleoprotein Filament.

Vestibular migraine pathophysiology: insights from structural and functional neuroimaging.

Structural and functional insights into the mode of action of a universally conserved Obg GTPase.

Structural insights into ribosome translocation.

Structural and functional studies of phosphoenolpyruvate carboxykinase from Mycobacterium tuberculosis.

Structural and Functional Studies of NirC from Salmonella typhimurium.

Structural insights into the regulatory mechanism of the Pseudomonas aeruginosa YfiBNR system.

Structural and functional insights into the N-terminus of Schizosaccharomyces pombe Cdc5.

Structural and Functional insights into the catalytic mechanism of the Type II NADH:quinone oxidoreductase family.

Structural insights into the negative regulation of BRI1 signaling by BRI1-interacting protein BKI1.

Structural and functional insights into enzymes of the vitamin K cycle.

Insights into RNA structure and function from genome-wide studies.

Structural and Functional Insights into the Cryoprotection of Membranes by the Intrinsically Disordered Dehydrins.

Recent structural and functional insights into the family of sodium calcium exchangers.