Crystal Structures of the Human Doublecortin C- and N-terminal Domains in Complex with Specific Antibodies

Abstract

Doublecortin is a microtubule-associated protein produced during neurogenesis. The protein stabilizes microtubules and stimulates their polymerization, which allows migration of immature neurons to their designated location in the brain. Mutations in the gene that impair doublecortin function and cause severe brain formation disorders are located on a tandem repeat of two doublecortin domains. The molecular mechanism of action of doublecortin is only incompletely understood. Anti-doublecortin antibodies, such as the rabbit polyclonal Abcam 18732, are widely used as neurogenesis markers. Here, we report the generation and characterization of antibodies that bind to single doublecortin domains. The antibodies were used as tools to obtain structures of both domains. Four independent crystal structures of the N-terminal domain reveal several distinct open and closed conformations of the peptide linking N- and C-terminal domains, which can be related to doublecortin function. An NMR assignment and a crystal structure in complex with a camelid antibody fragment show that the doublecortin C-terminal domain adopts the same well defined ubiquitin-like fold as the N-terminal domain, despite its reported aggregation and molten globule-like properties. The antibodies' unique domain specificity also renders them ideal research tools to better understand the role of individual domains in doublecortin function. A single chain camelid antibody fragment specific for the C-terminal doublecortin domain affected microtubule binding, whereas a monoclonal mouse antibody specific for the N-terminal domain did not. Together with steric considerations, this suggests that the microtubule-interacting doublecortin domain observed in cryo-electron micrographs is the C-terminal domain rather than the N-terminal one.

Keywords: cell migration; crystal structure; microtubule; microtubule-associated protein (MAP); neurogenesis; neuroscience; protein conformation; protein stability; protein-protein interaction; structural biology.

FIGURE 1.

Effect of doublecortin-specific antibodies on doublecortin binding to microtubules and on tubulin polymerization. a, the doublecortin variant with the complete domain tandem T-DCX co-pelleted with taxol-stabilized microtubules in a microtubule pelleting assay, whereas isolated N-DCX and C-DCX domains did not (M, reaction mix; S, supernatant; P, pellet; MT, microtubule). b, the controls show that the faint bands of N-DCX and C-DCX observed in the pellet are independent of microtubules. c, the N-DCX-specific antibody mAb 1/108 co-pelleted with T-DCX and microtubules while showing no effect on binding of DCX to microtubule. mAb 1/108 did not co-pellet with microtubule alone. d, the addition of the C-DCX-specific Nanobody XA4551 inhibited co-pelleting of T-DCX with microtubules. A control (Ctr) Xaperone not binding doublecortin had no effect on microtubule binding of T-DCX. e, overview of doublecortin domain arrangement. Construct boundaries for N-DCX, C-DCX, and T-DCX are given in both the UNIPROT numbering scheme used in this study and in the numbering used in some other doublecortin publications.

FIGURE 2.

The high resolution structures of N-DCX illustrate the different conformations occupied by its mobile C-terminal region. a, a superposition of three N-DCX structures from different crystals (chain A of 5IO9 colored light gray; chain E of 5IOI colored pink; chain A of 5IN7 colored magenta) onto PDB entry 2BQQ (7) (colored blue) shows new open and closed N-DCX conformations. b, the superposition shown in a but rotated by 90° around the vertical axis. c, despite the big differences in open and closed conformations of the C-terminal region, Trp²²⁷ always contacts a shallow pocket created by residues Lys¹³⁵, Val¹³⁶, Arg¹³⁷, Gly²⁰³, Glu²⁰⁴, and Ser²⁰⁵ on the DCX core. In the open conformation observed in 5IOI, this contact is intermolecular, Trp²²⁷ binding the pocket on a crystal neighbor. d, stereo image showing the 2F_o − F_c electron density at 1σ and 5IOI chain E Trp²⁷⁷ (magenta) interacting with Lys¹³⁵, Val¹³⁶, Arg¹³⁷, Gly²⁰³, Glu²⁰⁴, and Ser²⁰⁵ of chain A (yellow).

FIGURE 3.

The complex of wild type N-DCX with Fab1/108. a, schematic of the complex, N-DCX colored orange, Fab light chain colored blue, and heavy chain colored cyan. N-DCX Ile¹⁷⁶ at the center of the antibody epitope is highlighted in a ball representation. b, superposition of wild type N-DCX (orange) onto N-DCX_DD (gray) shows that the double mutation K215D/K216D introduced for crystallization did not disturb the DCX core structure in N-DCX_DD. The side chains of other residues interacting with the Fab1/108 are also shown. c, stereo image of the binding interface shows how Fab1/108 wraps around Ile¹⁷⁶ with the hydrophobic atoms of Lys¹⁰⁰ and Arg⁵⁰ from heavy chain and Tyr⁹⁴ and His⁹¹ from light chain. Additionally, N-DCX Asp¹⁴³, Arg¹⁴⁴, Tyr¹⁴⁵, Asp¹⁷⁴, Asn¹⁷⁷, and Lys²¹⁵ are hydrogen-bonded to Fab1/108, and Phe²¹³ and Val²¹⁷ make hydrophobic contacts. Notably, Lys²¹⁵ of the K215D/K216D crystallization mutation in N-DCX_DD is part of the epitope recognized by the crystallization helper Fab1/108. d, the crystal packing is dense, without voids that would provide space for the C-DCX domain.

FIGURE 4.

C-DCX assumes an ordered structure. a, thermal shift curves from the thermofluor buffer screening show thermal unfolding transitions for C-DCX in the buffers 125 mm NaAc, pH 4.5, 400 mm NaCl (blue) and 180 mm CAPS, pH 10.0, 400 mm NaCl (red), whereas in 180 mm Hepes, pH 7.0, 400 mm NaCl (green), fluorescence is already present at the start and decays with protein precipitation. b, the assigned two-dimensional ¹H-¹⁵N HSQC spectrum of C-DCX in NMR buffer at 300 K and pH 4. Additional residues at the N terminus (residues 244–250) are due to a cloning artifact. c, chemical shift index (CSI) histogram shows the predicted secondary structure for C-DCX to be consistent with the DCX fold. Consensus CSI values of −1 indicate α-helix, whereas values of 1 correspond to β-strand structures. d, sedimentation coefficient distribution c(s_20,w) for C-DCX at pH 7.5 reveals a dominant monomer peak at 1.15 S, which is consistent with a folded DCX domain. The top panel with the integrated distribution shows that the monomer peak comprises 48% of the total loading signal and a series of oligomers accounts for the rest.

FIGURE 5.

The C-terminal DCX domain has the DCX fold. a, overview of the complex between C-DCX (green) and the V_HH XA4551 (cyan). The orientation is the same as in b and c. b, superposition of the C-DCX domain (green) on DCDC2 C-terminal domain NMR structure (PDB entry 2DNF) (dark gray) shows that both assume the classical DCX fold. The N and C termini as well as a few surface loops are labeled with residue numbers. c, superposition of the C-DCX domain (green) on N-DCX core (gray). d, the hydrophobic surface patch on C-DCX (green) that is recognized by XA4551. Black labels indicate the C-DCX residues contributing to the hydrophobic surface. Superimposed is the structure of N-DCX (gray), where Tyr¹⁵¹ and Phe¹⁵⁸ side chains prevent the formation of a hydrophobic surface pocket. e, stereo image showing how XA4551 (cyan) binds its epitope, which is a cluster of aliphatic side chains on the C-DCX surface on a β-strand-loop-helix motif (residues 276–298) and on an adjacent loop (residues 260–262). Ile³¹ of XA4551 CDR1 plugs a shallow pocket lined exclusively by aliphatic side chains.

FIGURE 6.

The binding of anti-DCX domain antibody complexes is incompatible with the cryo-EM-derived binding mode of a DCX domain to 13-protofilament microtubules. a, superposition of the DCX domains in the C-DCX XA4551 complex and the N-DCX Fab1/108 complex shows that the two antibodies recognize epitopes on different regions of the DCX domain fold (N-DCX orange, C-DCX green, antibodies cyan and blue). The DCX domains are in the same orientation as in b and c. b, superposition of the C-DCX·XA4551 complex on the DCX domain observed in the doublecortin microtubule complex PDB entry 2XRP shows that XA4551 clashes with tubulin α2 and β2. The location of the patient mutations R259L, P272R, T303I, and G304E that affected doublecortin binding to curved microtubules in single molecule fluorescence microscopy experiments is indicated by arrows. c, the same superposition of the N-DCX·Fab1/108 complex shows that Fab1/108 clashes with tubulin α1.