Structure-based mutagenesis of the substrate-recognition domain of Nrdp1/FLRF identifies the binding site for the receptor tyrosine kinase ErbB3


  • Samuel Bouyain,

    1. Department of Biophysics and Biophysical Chemistry, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, USA
    Current affiliation:
    1. Division of Molecular Biology and Biochemistry, School of Biological Sciences, University of Missouri-Kansas City, 5100 Rockhill Road, Kansas City, MO 64110, USA.
    Search for more papers by this author
  • Daniel J. Leahy

    Corresponding author
    1. Department of Biophysics and Biophysical Chemistry, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, USA
    • Department of Biophysics and Biophysical Chemistry, The Johns Hopkins University School of Medicine, 725 N. Wolfe Street, Baltimore, MD 21205, USA; fax: 410 614-8839.
    Search for more papers by this author


The E3 ubiquitin ligase neuregulin receptor degrading protein 1 (Nrdp1) mediates the ligand-independent degradation of the epidermal growth factor receptor family member ErbB3/HER3. By regulating cellular levels of ErbB3, Nrdp1 influences ErbB3-mediated signaling, which is essential for normal vertebrate development. Nrdp1 belongs to the tripartite or RBCC (RING, B-box, coiled-coil) family of ubiquitin ligases in which the RING domain is responsible for ubiquitin ligation and a variable C-terminal region mediates substrate recognition. We report here the 1.95 Å crystal structure of the C-terminal domain of Nrdp1 and show that this domain is sufficient to mediate ErbB3 binding. Furthermore, we have used site-directed mutagenesis to map regions of the Nrdp1 surface that are important for interacting with ErbB3 and mediating its degradation in transfected cells. The ErbB3-binding site localizes to a region of Nrdp1 that is conserved from invertebrates to vertebrates, in contrast to ErbB3, which is only found in vertebrates. This observation suggests that Nrdp1 uses a common binding site to recognize its targets in different species.

Abbreviations: RTK, receptor tyrosine kinase; Nrdp1, neuregulin receptor degrading protein 1; TRIM/RBCC, tripartite motif/RING B-box coiled-coil; Nrdp1C, C-terminal region of Nrdp1; SAD, single-wavelength anomalous diffraction; USP8, ubiquitin-specific protease 8; TEV, tobacco etch virus; hGH, human growth hormone.

The ubiquitin-proteasome proteolytic pathway is a major mediator of cytosolic protein degradation in eukaryotic cells and is essential for the regulation of many processes including signal transduction, apoptosis, and cell cycle progression (Hershko and Ciechanover 1998). Proteins are targeted for degradation in the proteasome by covalent attachment of ubiquitin, a 76-amino acid protein that may itself be modified by ubiquitin to form poly-ubiquitin chains. Ubiquitination of target proteins is coordinated by three proteins and involves transfer of ubiquitin from an E1 activating enzyme to an E2 ubiquitin-carrier protein. An E3 ubiquitin ligase then catalyzes transfer of ubiquitin to the substrate protein either directly or indirectly via a covalent E3 ubiquitin intermediate (Passmore and Barford 2004).

Cellular levels and distribution of receptor tyrosine kinases (RTKs2) such as the nerve growth factor receptor (TrkA), the hepatocyte growth factor (Met), and the epidermal growth factor receptor (EGFR/ErbB1) are regulated by ubiquitin modification subsequent to their ligand-dependent activation in a process known as “down-regulation” (Abella et al. 2005; Arevalo et al. 2006; Huang et al. 2006). Ubiquitination of these RTKs modulates their signaling activity by regulating their recycling to the cell surface or their lysosomal degradation following endocytosis (Wiley and Burke 2001), and disruption of normal ubiquitin-mediated regulation of RTK signaling can contribute to uncontrolled cell growth in human cancers (Bache et al. 2004). In contrast, the EGFR-homolog ErbB3 is degraded by the ubiquitin-proteasome pathway in a ligand-independent manner (Diamonti et al. 2002; Qiu and Goldberg 2002). Ubiquitination of ErbB3 is mediated by the E3 ubiquitin ligase neuregulin receptor degrading protein 1 (Nrdp1), also known as fetal liver ring finger protein (FLRF) (Abdullah et al. 2001), which regulates the steady-state level of ErbB3 receptors at the cell surface, thereby influencing the potency of ErbB3-mediated signaling. ErbB3 expression levels are commonly increased concomitantly with the EGFR-homolog ErbB2/HER2 in mammary tumors, and this increased expression occurs at the protein and not at the mRNA level, suggesting that alterations in ErbB3 turnover may figure in these tumors (Siegel et al. 1999). Indeed, a recent study has shown that loss of Nrdp1 expression is associated with increased levels of ErbB3 in a mouse model of ErbB mammary tumor and in primary human breast tumors (Yen et al. 2006).

Nrdp1 is a 317-amino acid protein composed of a “tripartite” motif that consists of an N-terminal RING domain, two zinc-finger domains called the B-Box, and a coiled-coil segment (Abdullah et al. 2001). This tripartite motif is characteristic of a family of E3 ubiquitin ligases, designated TRIM/RBCC (Tripartite Motif/Ring B-Box Coiled-coil) proteins, which are involved in a wide variety of cellular processes including apoptosis, regulation of the cell cycle, and control of transcription (Meroni and Diez-Roux 2005). This functional diversity is reflected in the sequence diversity found in the C-terminal portions of those E3 ligases, which are thought to be involved in selectively recognizing target proteins, perhaps in concert with the coiled-coil regions. Of particular interest, the C-terminal region of Nrdp1 (amino acids 179–317) has no homology with other proteins of known structure (Diamonti et al. 2002; Qiu and Goldberg 2002).

Several features of Nrdp1-mediated degradation of ErbB3 have been characterized. Mutation of key cysteine, histidine, and aspartate residues in the RING domain of Nrdp1 inhibits degradation of ErbB3, and the RING and B-box motifs are not required for binding to ErbB3 (Qiu and Goldberg 2002). Interestingly, the coiled-coil and C-terminal regions of Nrdp1 act as a dominant negative form of Nrdp1 and cause accumulation of ErbB3 in cells cotransfected with ErbB3 and Nrdp1 (Diamonti et al. 2002; Qiu and Goldberg 2002). This observation is reminiscent of results obtained for EGFR/ErbB1, ubiquitination of which by the E3 ligase c-Cbl following ligand-dependent phosphorylation is inhibited by overexpression of an oncogenic form of the E3 ligase (v-Cbl) that encompasses its ErbB1-recognition region but not the ubiquitin-transfer region (Levkowitz et al. 1999). The role of Nrdp1 in the degradation of ErbB3 stimulated our interest in analyzing the interaction between ErbB3 and Nrdp1 at a molecular level. We present here the 1.95 Å crystal structure of the C-terminal region of murine Nrdp1 (Nrdp1C, amino acids 193–317) and show that it adopts a fold thus far unique to Nrdp1 molecules. Using this structure as a guide for site-directed mutagenesis, we have also mapped regions on the Nrdp1C surface important for both ErbB3 binding and Nrdp1-mediated degradation of ErbB3 in transfected HEK 293 cells. The Nrdp1C structure presented here is very similar to the recently reported structure of a human Nrdp1C domain (Avvakumov et al. 2006).

Results and Discussion

Nrdp1C structure

The crystal structure of Nrdp1C was determined using single-wavelength anomalous diffraction (SAD) with L-selenomethionyl-labeled protein, and the final atomic model refined using data to 1.95 Å Bragg spacings (Rcryst/Rfree = 0.212/0.234). Data collection and refinement statistics are presented in Table 1. Nrdp1C adopts a globular structure in which a central three-strand antiparallel β-sheet formed exclusively by C-terminal residues is encircled by a ring of loops and α-helices (Fig. 1A). This structure superimposes with the recently reported structure of human Nrdp1C with a root mean square deviation of 0.62 Å for 123 Cα positions (Avvakumov et al. 2006). An interesting feature of Nrdp1C is a negatively charged surface that is formed by the side chains of Glu-195, Glu-245, Glu-248, Asp-298, Asp-299, and Glu-303 and encompasses the majority of one face of Nrdp1C (Fig. 1B). The conservation of acidic amino acids at these positions in Nrdp1 homologs from different species indicates that this negatively charged surface is characteristic of this domain (Fig. 2).

Table Table 1.. Crystallographic data
original image
Figure Figure 1..

Crystal structure of Nrdp1C. (A) Ribbon diagrams of Nrdp1C. Nrdp1C is colored in a rainbow progression from blue at the N-terminus (N) to red at the C-terminus (C). Elements of secondary structure are indicated by α (α-helices) and β (β-strands). The right view is obtained by a 90° rotation along a vertical axis. (B) Electrostatic surface representation of Nrdp1C. The view on the left is in the same orientation as the left view in A. Regions with negative electrostatic potential are colored red and regions with positive electrostatic potential are colored blue (scale ± 10 e/kT). The acidic side chains that make up the negatively charged surface on the left view are indicated in white.

Figure Figure 2..

Alignment of Nrdp1C sequences. The positions of α-helices and β-strands in the Nrdp1C structure are indicated by the letters α and β, respectively. Strictly conserved residues are shaded in black and similar residues are colored gray. The numbering refers to the mouse sequence. The position of residues changed to alanine in the Nrdp1C mutant sets are indicated above the sequence by boxed letter As that are color-coded according to the different mutation sets. Sets 1–5, in which three to six residues were changed to alanine are colored cyan, green, orange, pink, and red, respectively. Sets 6–10, in which pairs of residues were substituted to alanine, are colored yellow, brown, lime, blue, and purple, respectively. This figure was prepared with ESpript (Gouet et al. 1999).

Binding of Nrdp1C to ErbB3

Previous studies have shown that the interaction between ErbB3 and Nrdp1 is mediated by a region of Nrdp1 that includes the coiled-coil region and Nrdp1C (Diamonti et al. 2002; Qiu and Goldberg 2002). A pull-down binding assay was designed to test whether Nrdp1C alone is sufficient to interact with ErbB3. Lysates of MDA-MB-453 breast cancer cells, which express high levels of ErbB2 and ErbB3 but negligible levels of ErbB1 and ErbB4, were prepared, and Nrdp1C-coupled resin was indeed capable of pulling ErbB3 down from the lysate as compared to uncoupled resin (Fig. 3; Moasser et al. 2001). The interaction between ErbB3 and Nrdp1C-coupled resin was inhibited by the presence of comparable amounts of soluble Nrdp1C. These results indicate that Nrdp1C binds specifically to ErbB3 and that the coiled-coil region is not required for ErbB3 binding.

Figure Figure 3..

Binding of ErbB3 to Nrdp1C-coupled resin. (A) ErbB3 was precipitated from the lysate of MDA-MB-453 cells using the anti-ErbB3 monoclonal antibody SGP1 (Anti-ErbB3), Nrdp1C-coupled resin (Nrdp1C) or Nrdp1C-coupled resin in the presence of 50 μg of soluble Nrdp1C (Competition). A control reaction using uncoupled resin is shown in the second lane (Control). Bound ErbB3 was visualized using a rabbit polyclonal antibody directed against the 17 C-terminal residues of ErbB3. (B) ErbB3 was precipitated from the lysate of Chinese Hamster Ovary cells overexpressing a fusion protein of hGH and ErbB3 that lacks the C-terminal regulatory region of ErbB3. The experiment was conducted as described in A except that bound ErbB3 was visualized using a rabbit polyclonal antibody against hGH.

The cytoplasmic region of ErbB3 consists of a ∼25-amino acid juxtamembrane region followed by a ∼360-amino acid kinase domain and a ∼280-amino acid C-terminal regulatory region. Previous studies have mapped the Nrdp1-binding site in ErbB3 to its C-terminal regulatory region (Diamonti et al. 2002). We took advantage of a Chinese Hamster Ovary cell line expressing a form of ErbB3 lacking the C-terminal regulatory region to test this result. As shown in Figure 3B, Nrdp1C-coupled resin pulls down this truncated form of ErbB3 from cell lysates. Addition of soluble Nrdp1C inhibits this interaction. These results suggest that the Nrdp1-binding site on ErbB3 resides within the juxtamembrane domain or tyrosine kinase domain, but not in the C-terminal tail as previously proposed. This discrepancy may originate from the fact that previous results were obtained using a yeast two-hybrid assay, but not a pull-down binding assay, which is more direct.

Identification of an ErbB3-interacting surface on Nrdp1C

The Nrdp1C crystal structure was used to guide selection of clusters of residues that map to discrete regions of the Nrdp1C surface. Five nonoverlapping groups of three to six residues on the surface of Nrdp1C were selected and residues in each group were changed to alanine to evaluate their importance to ErbB3 binding (Figs. 2, 4A). Only solvent-exposed residues were selected to minimize global effects on Nrdp1C structure when changed to alanine, and all alanine-substituted Nrdp1C proteins expressed and behaved comparably to wild-type Nrdp1C. In particular, all proteins are monomeric as determined by size-exclusion chromatography, indicating that the alanine substitutions do not disrupt the global structure of the protein or its monomeric state.

Figure Figure 4..

Identification of an ErbB3-interacting surface on Nrdp1C. (A) Clusters of residues substituted to alanine are indicated in color (cyan, set 1; green, set 2; orange, set 3; pink, set 4; red, set 5). The view on the left is in the same orientation as the left view in Fig. 1A. Other views are related by 120° rotations around a vertical axis. The boxed view shows the positions of pairs of residues substituted with alanine in mutant sets 6–10 (yellow, set 6; brown, set 7; lime, set 8; blue, set 9; purple, set 10). The color-coding of the different sets is identical to that used in Fig. 2. (B) ErbB3 was precipitated from the lysate of MDA-MB-453 cells using the anti-ErbB3 monoclonal antibody SGP1 (anti-ErbB3) or with resins coupled to wild-type or the indicated alanine-substituted Nrdp1C proteins. The mutant set number and color code are indicated above each lane. A control reaction using uncoupled resin is shown in the second lane (Control). Bound ErbB3 was visualized using a rabbit polyclonal antibody against the 17 C-terminal residues of ErbB3. (C) The amount of ErbB3 in the lysate of HEK 293 cells that were cotransfected with a plasmid encoding ErbB3 (1 μg) along with wild-type or mutated Nrdp1 constructs (1 μg) was visualized in a Western blot using a rabbit polyclonal antibody against ErbB3. ErbB3 appears as a doublet in ErbB3-transfected HEK 293 cells, which presumably reflects the presence of phosphorylated and non-phosphorylated species as observed by Qiu and Goldberg (2002). The ErbB3 lane includes 1 μg of ErbB3 along with 1 μg of empty vector. The amount of Nrdp1 present in these lysates was visualized using a rabbit polyclonal antibody against the C-terminal portion of Nrdp1. Immunoblotting against actin is shown to control for the amount of protein loaded in each lane.

Each of the five Nrdp1C molecules bearing alanine substitutions was purified and coupled to resin for use in pull-down assays. Binding of ErbB3 to Nrdp1C was impaired by alanine substitutions in clusters 1, 2, and 4 (colored cyan, green, and pink, respectively, in Fig. 4) but not by substitutions in clusters 3 and 5 (Fig. 4B). Interestingly, normal binding was observed for set 5 (colored red in Fig. 4), which includes the side chains of Glu-245, Glu-248, and Glu-303 and makes up a significant portion of the acidic region found on one face of Nrdp1C (Fig. 1B). This result indicates that this acidic region is not required to mediate interactions with ErbB3 and raises the question of what role this conserved region has.

The effects of the alanine substitutions were also tested in an ErbB3-degradation assay, which offers the advantage of correlating the effects of Nrdp1 mutations with the cellular stability of ErbB3. For this assay HEK 293 cells were transfected with ErbB3 and either wild-type or mutated Nrdp1 cDNAs (Fig. 4C). Nrdp1 promotes the degradation of ErbB3, and cotransfection of ErbB3 with wild-type Nrdp1 led to decreased levels of ErbB3 compared to transfection of ErbB3 alone (Fig. 4C). HEK 293 cells were chosen for these experiments because they express low levels of endogenous Nrdp1, which may indicate that Nrdp1 is marginally stable in these cells (Wu et al. 2004); Nrdp1 mediates its own ubiquitination and consequent degradation by the ubiquitin-proteasome pathway (Qiu and Goldberg 2002; Wu et al. 2004). In our hands, transfection of full-length Nrdp1 in HEK 293 cells does not lead to increased Nrdp1 protein levels compared to those observed with cells transfected with ErbB3 or empty vector only, which most likely reflects its inherent instability (Fig. 4C).

In line with the binding assay shown in Figure 4B, cotransfection of ErbB3 and Nrdp1 mutant sets 3 and 5, which retain ErbB3 binding, leads to a decrease in the levels of ErbB3. In contrast, cotransfection of ErbB3 with mutant set 4 leads to an increase in the amount of ErbB3 comparable to the levels observed with transfection of ErbB3 and vector only. Thus, introduction of this cluster of substitutions, which impair Nrdp1 binding to ErbB3, also impairs the ErbB3-degrading function of Nrdp1. Curiously, cotransfection of ErbB3 with Nrdp1 bearing mutations in sets 1 and 2, which prevent Nrdp1C from pulling down ErbB3 from cell lysates, does not lead to increased ErbB3 levels relative to cotransfection with wild-type Nrdp1. It is difficult to imagine how loss of ErbB3 binding could be compatible with relatively normal Nrdp1-mediated ErbB3 degradation, and these results likely indicate that mutant sets 1 and 2 retain some ErbB3-binding activity that is not fully detected in qualitative pull-down assays.

To localize the ErbB3-interaction site more accurately, five additional pairs of residues in and around the mutant set 4 surface were changed to alanine (Figs. 2,4A, boxed view). The activities of these Nrdp1 variants were evaluated in the context of the ErbB3-degradation assay because of the better sensitivity of these experiments compared to pull-down binding assays. Changing residues Gln-284/His-312 (set 6) and Gln-266/Arg-269 (set 10) to alanine impairs the ability of Nrdp1 to mediate ErbB3 degradation, but alanine substitutions at the other residue pairs do not (Fig. 4C). This result further maps the region of the Nrdp1 surface important for mediating degradation of ErbB3 to the central-top region of mutant set 4 as depicted in Figure 4A. This region is made up of the α4 helix and strand β5 and occurs on the opposite end of Nrdp1 relative to the N terminus (Figs.1, 2), reinforcing the observation that the coiled-coil region is not required for binding to ErbB3.

The residues composing the ErbB3 binding surface are mostly hydrophilic and are well conserved among different species (Fig. 2). Of the residues found at this surface, Arg-213, Thr-220, Gln-266, Arg-269, Gln-284, and His-312 are conserved in all sequences except in a second Nrdp1 homolog in Drosophila, and Leu-262 is strictly conserved in all available Nrdp1 sequences. Thr-261 is either conserved or substituted with a serine residue. Interestingly, this surface is conserved in invertebrates even though ErbB3 is only found in vertebrates, which suggests that Nrdp1 utilizes this same region to interact with other substrates in different species.

Like other E3 ubiquitin ligases, Nrdp1 has multiple substrates (Hershko and Ciechanover 1998), which include parkin and ubiquitin-specific protease 8 (USP8) (Wu et al. 2004; Zhong et al. 2005), and the crystal structure of a complex between human USP8 and the C-terminal domain of Nrdp1 was reported during preparation of this manuscript (Avvakumov et al. 2006). The USP8-binding site on Nrdp1 includes the side chains of residues Ser-219, Thr-220, Asp-222, Leu-262, Arg-265, Gln-266, and Arg-269. Mutation of Thr-220 and Leu-262 in mutant set 4 and Gln-266 and Arg-269 in mutant set 10 impair ErbB3 degradation, indicating that the binding sites for USP8 and ErbB3 on Nrdp1 overlap (Fig. 5). No USP8 residue contacts either Gln-284 or His-312 as the closest neighbor residues are 10 Å and 7 Å, respectively. This observation suggests that the two binding modes are not identical. On the other hand, the sequence HXXDD appears to be a consensus binding motif in USP8 for Nrdp1, but this motif is not found in the amino acid sequence of ErbB3, which further suggests that the determinants of the Nrdp1•ErbB3 interaction are not identical to those of Nrdp1•USP8.

Figure Figure 5..

Comparison of the USP8- and ErbB3-binding sites on Nrdp1C. Residues in the ErbB3-binding site identified by site-directed mutagenesis are shown on the left panel and colored red. Residues involved in USP8 binding in the crystal structure of human Nrdp1 in complex with USP8 (Avvakumov et al. 2006) were identified using CNS (Brunger et al. 1998) and are shown in blue on the right panel. The two views are identical to the boxed view in Fig. 4A. Thr-261 is not visible in this view.


The results presented here demonstrate that the C-terminal domain of Nrdp1 binds specifically to its substrate ErbB3 via residues located around helix α4 and strand β5. Conservation of this binding region in diverse species suggests that it has an important role in the ubiquitin-proteasome pathway that extends beyond mediating interactions with ErbB3, which is only found in vertebrates. Nrdp1 may have common and specific substrates in different species although the determinants of Nrdp1 substrate specificity have yet to be discovered.

Materials and Methods


The following materials or reagents were purchased from the indicated source: mouse E13.5 cDNA library, the plasmid pCDNA6/V5-His and cell culture media (Invitrogen), restriction enzymes (New England Biolabs), oligonucleotide primers (Invitrogen and Integrated DNA Technologies), the expression vector pET41 (Novagen), chelating sepharose, chromatography columns and CNBr-activated sepharose fast flow (GE Healthcare), MDA-MB-453 and HEK 293 cells (ATCC), Superfect transfection reagent (Qiagen), and EDTA-free protease inhibitors (Roche). The anti-ErbB3 monoclonal antibody SGP1 was kindly provided by William Gullick (University of Kent at Canterbury, United Kingdom, Rajkumar and Gullick 1994). The rabbit polyclonal antibodies against ErbB3 (C-17), against actin (H-300), and protein G-agarose were purchased from Santa Cruz Biotechnologies. The rabbit polyclonal antibody against human growth hormone (hGH) was purchased from Research Diagnostics. The rabbit polyclonal antibody against Nrdp1 was obtained from Bethyl; this antibody recognizes an epitope encompassed by amino acids 250–300 of human Nrdp1.

Sequence alignments

Nrdp1 homologs were identified by BLAST searching the protein database with the amino acid sequence of the C-terminal region of mouse Nrdp1. GenBank accession numbers for these sequences are NP_005776.1 (Homo sapiens), NP_080535.2 (Mus musculus), NP_998681.1 (Danio rerio), AAI08759.1 (Xenopus laevis), CAG03957.1 (Tetraodon nigroviridis), XP_321185.2 (Anopheles gambiae), XP_623219.1 (Apis mellifera), NP_648816.1 (Drosophila melanogaster 1), and NP_609668.2 (Drosophila melanogaster 2). Sequence alignments were performed using CLUSTALW (

Cloning and mutagenesis of Nrdp1

A cDNA sequence encoding full-length Nrdp1 was amplified from a mouse E13.5 cDNA library using 5′ and 3′ oligonucleotides encoding flanking BamHI and EcoRI sites, respectively. The resulting cDNA fragment was subcloned into the plasmid pCDNA6/V5-His and sequenced. A cDNA encoding the C-terminal region of Nrdp1 (residues Ala-179 to Ile-317) was amplified from the full-length Nrdp1 gene, subcloned into a modified plasmid pET41 that includes a tobacco etch virus (TEV) protease-recognition site, and sequenced. This vector directs expression of glutathione S-transferase followed by a hexahistidine tag, a TEV protease-recognition site and the C-terminal domain of Nrdp1. This protein was purified (see below) and subjected to limited proteolysis with subtilisin, which led to identification of a protease-stable domain corresponding to residues 193–317 (designated Nrdp1C) via SDS-PAGE and N-terminal sequencing. A cDNA encoding Nrdp1C was amplified by PCR and subcloned into the modified pET41 expression vector. Mutagenesis of the Nrdp1C cDNA subcloned in pET41 to introduce alanine substitutions for binding and degradation assays was carried out by a multistep megaprimer method (Geisbrecht et al. 2003). cDNAs encoding full-length Nrdp1 mutants were obtained by amplifying cDNA fragments encoding the N-terminal region of Nrdp1 and the first 30 base pairs of the Nrdp1C sequences and cDNA fragments encoding Nrdp1C mutants in separate reactions. These fragments were combined and amplified into full-length cDNAs using two oligonucleotides that bind to the 5′ end of the fragment encoding the N-terminal domain of Nrdp1 and to the 3′ end of Nrdp1C, respectively. The resulting cDNA sequences were subcloned into the plasmid pCDNA6/V5-His as described above. All DNA sequences were confirmed by sequencing.

Protein purification

cDNAs encoding Nrdp1C and its mutants were transformed into Escherichia coli strain BL21(DE3). For wild-type and mutant Nrdp1C, an aliquot of an overnight culture of transformed bacteria (10 mL) was used to inoculate 1 L of terrific broth media containing 50 μg/mL of kanamycin, and the culture was grown at 30°C and 225 rpm. Expression of targeted proteins was induced at OD550 of 1.5 by adding isopropyl-β-D-thiogalactopyranoside to a final concentration of 1 mM. The temperature was decreased to 23°C and cells were grown for a further 16–18 h. Harvested cells were suspended in 50 mL of loading buffer (500 mM NaCl, 10 mM imidazole, and 20 mM Tris-HCl, pH 8.0) and lysed by microfluidization. The lysate was centrifuged 30 min at 30,000g, and the supernatant was applied to a 2-mL column of chelating sepharose previously charged with nickel sulfate. Bound proteins were eluted with 250 mM imidazole subsequent to washing the column. After cleavage with TEV protease, Nrdp1C was purified by ion exchange using a 6-mL Resource Q column followed by size-exclusion chromatography using a Superdex 75 26/60 column equilibrated in 200 mM NaCl and 20 mM Tris-HCl, pH 8.0.

For L-selenomethionine labeling, the cDNA encoding Nrdp1C was transformed into E. coli strain B834(DE3), and the transformed cells were grown in M9 minimal medium containing 50 mg/L of L-selenomethionine as previously described (Beneken et al. 2000). The labeled protein was purified as described above except that all buffers contained 0.01% (v/v) β-mercaptoethanol. Purified selenomethionine-labeled Nrdp1C was dialyzed against 2 mM Tris-HCl pH 8.0, 0.01% (v/v) β-mercaptoethanol, concentrated to 8 mg/mL and stored frozen at −80°C in 200-μL aliquots.

Crystallization and structure determination

Crystals were grown at 20°C by the method of hanging-drop vapor diffusion. Selenomethionine-labeled Nrdp1C (1 μL) was mixed with 1 μL of a 3:1 mixture of water:reservoir buffer (25% [w/v] polyethylene glycol 8000, 180 mM potassium thiocyanate, and 100 mM Tris-HCl pH 8.4). Crystals grew as large “snowflake”-like plates within 5 d. Single plates were isolated and transferred to a solution of 25% (w/v) polyethylene glycol 8000, 10% (v/v) glycerol, 50 mM potassium thiocyanate, and 100 mM Tris-HCl pH 8.4 and flash-frozen in liquid nitrogen prior to data collection.

Diffraction data were collected from a single crystal of selenomethionyl-substituted Nrdp1C at the selenium peak wavelength (0.9792 Å) to 1.95 Å Bragg spacings at National Synchrotron Light Source beamline X4A and processed with DENZO and SCALEPACK (Table 1; Otwinowski and Minor 1997). These data were used both for de novo phasing by SAD methods using the signal from the selenium atoms and for refinement of the structure. The crystals belong to space group C2 and contain two molecules per asymmetric unit. The positions of twelve selenium atoms were found using SOLVE (Terwilliger 2003), and an initial model obtained using RESOLVE was completed after manual building using O (Jones et al. 1991). The final model was obtained after several rounds of model building using O and refinement using CNS (initial stages) and REFMAC (latter stages) (Brunger et al. 1998; Winn et al. 2001).

The final model includes residues 195–317 in molecule A, 193–316 in molecule B, 122 water molecules, two glycerol molecules, and one thiocyanate ion. An additional serine residue derived from the expression vector is found at the N terminus of the B molecule. Analysis with PROCHECK shows that 92.1% of the residues occupy the most favored regions of the Ramachandran plot with the remainder occupying allowed regions. The two Nrdp1C molecules in the asymmetric unit superimpose with a root mean square deviation of 0.41 Å for 118 Cα positions; the following structural descriptions refer to molecule B because it is slightly more complete. Electrostatic potentials were calculated with DELPHI and structural images made with PYMOL ( (Rocchia et al. 2001, 2002).

Pull-down binding assays

Wild-type and variant Nrdp1C proteins were coupled to CNBr-activated sepharose according to the manufacturer's instructions. The resins were stored as 50% (v/v) slurry in 500 mM NaCl, 100 mM Tris-HCl pH 8.0, 0.5 mM EDTA at 4°C. MDA-MB-453 cells were maintained in Leibovitz's media supplemented with 10% (v/v) fetal bovine serum and 1% (v/v) penicillin/streptomycin in non-vented flasks. For pull-down binding assays, cells at 90%–100% confluency were detached from the plastic plate in phosphate-buffered saline containing 5 mM EDTA, harvested by centrifugation, and suspended in lysis buffer (100 mM NaCl, 10% [v/v] glycerol, 1% [v/v] Triton X-100, 1% [w/v] sodium deoxycholate, 1 mM EDTA, 50 mM Tris-HCl pH 8.0, protease inhibitors) at a density of 4–10 M cells for each mL of buffer. After lysis on ice, cellular debris was removed by centrifugation and lysates were cleared with either protein G-agarose along with normal mouse IgG (in the case of immunoprecipitations) or quenched CNBr-activated sepharose (in the case of resins of Nrdp1C).

For immunoprecipitation of ErbB3, cleared lysate (1 mL) was incubated with protein G-agarose and 1 μg of the anti-ErbB3 monoclonal antibody SGP1 directed against the extracellular domain of ErbB3 (Rajkumar and Gullick 1994). For binding assays with Nrdp1C, 1 μL of cleared lysate was incubated with 30 μL of a 50% (v/v) slurry of Nrdp1C resin coupled at a density of 5 mg of protein per mL of gel. Samples were incubated overnight at 4°C and resins were washed with 100 mM NaCl, 25 mM Tris-HCl pH 8.0. In competition pull-down binding assays, the Nrdp1C-coupled resin and the cell lysate were incubated in the presence of 50 μg of soluble Nrdp1C. Samples were resolved by SDS-PAGE and affinity-purified ErbB3 was detected by immunoblotting with a rabbit polyclonal antibody against the C-terminal tail of human ErbB3.

Pull-down binding assays were also performed with Chinese Hamster Ovary cells expressing a fusion protein of hGH and a truncated form of ErbB3 (amino acids 20–1019, starting from the initiator methionine) that lacks its N-terminal signal sequence as well as the C-terminal regulatory region (P.A. Longo and D.J. Leahy, unpubl.). The cDNA construct directing the expression of the hGH–ErbB3 fusion was designed by subcloning the appropriate fragment of ErbB3 into the XbaI and NotI restriction sites of the plasmid pSGHV0 (Leahy et al. 2000). The density of cells at lysis was about 2 M cells per mL of lysis buffer. Affinity-purified ErbB3 was detected by resolving each sample by SDS-PAGE followed by transfer to a PVDF membrane and immunoblotting with a rabbit polyclonal antibody against hGH. The binding of ErbB3 from the lysate of MDA-MB-453 cells to resins coupled to wild-type or variant Nrdp1C proteins was assayed as described above, except that the densities of coupled proteins on the resin were 2 mg/mL.

ErbB3-degradation assays

HEK 293 cells were maintained in Dulbecco's modified eagle medium with high glucose medium supplemented with 10% (v/v) fetal bovine serum and 1% (v/v) penicillin/streptomycin. Cells in six-well plates were transfected with Superfect transfection reagent using 2 μg of total DNA according to the manufacturer's suggestions. Cells were transfected with a cDNA construct encoding full-length human ErbB3 subcloned into the plasmid pCDNA6/V5-His and either empty pCDNA6 (control) or wild-type/variant pCDNA6-Nrdp1 plasmids. Cells were detached with phosphate-buffered saline containing 5 mM EDTA and lysed with lysis buffer 16–18 h after transfection. Lysates were resolved by SDS-PAGE, and proteins detected after immunoblotting with antibodies against ErbB3, Nrdp1, or actin for control of protein load.

PDB deposition

The atomic coordinates and structure factors have been deposited in the Protein Data Bank ( as PDB ID code 2OGB.


This work was supported by grants from the National Institutes of Health (R01-CA90466 to D.J.L.) and The Wellcome Trust (066611/Z/01/Z to S.B.). We thank the High-Throughput Crystallization Laboratory at the Hauptman-Woodward Institute (Buffalo, NY) for performing initial crystallization screens with Nrdp1C, Randy Abramovitz and John Schwanof for assistance at beamline X4A at Brookhaven National Laboratory, Chen Qiu for assistance with data collection, and William Gullick (University of Kent at Canterbury, United Kingdom) for providing the monoclonal anti-ErbB3 antibody SGP1. We also thank Brian V. Geisbrecht for comments on the manuscript.