Crystal structure of human CRMP-4: correction of intensities for lattice-translocation disorder

Abstract

Collapsin response mediator proteins (CRMPs) are cytosolic phosphoproteins that are mainly involved in neuronal cell development. In humans, the CRMP family comprises five members. Here, crystal structures of human CRMP-4 in a truncated and a full-length version are presented. The latter was determined from two types of crystals, which were either twinned or partially disordered. The crystal disorder was coupled with translational NCS in ordered domains and manifested itself with a rather sophisticated modulation of intensities. The data were demodulated using either the two-lattice treatment of lattice-translocation effects or a novel method in which demodulation was achieved by independent scaling of several groups of intensities. This iterative protocol does not rely on any particular parameterization of the modulation coefficients, but uses the current refined structure as a reference. The best results in terms of R factors and map correlation coefficients were obtained using this new method. The determined structures of CRMP-4 are similar to those of other CRMPs. Structural comparison allowed the confirmation of known residues, as well as the identification of new residues, that are important for the homo- and hetero-oligomerization of these proteins, which are critical to nerve-cell development. The structures provide further insight into the effects of medically relevant mutations of the DPYSL-3 gene encoding CRMP-4 and the putative enzymatic activities of CRMPs.

Keywords: CRMP-4; lattice-translocation disorder.

Figure 1

(a) Typical pseudo-precession diffraction pattern from CRMP-4 crystals (form B). The h0l plane is shown. Sharp reflections are observed along h for l = 5n. The other reflections are diffuse, with the streaks being most prominent at l = 5n + 2 and l = 5n + 3. LABELIT (Sauter et al., 2004 ▶) was used to calculate the diffraction image from the experimental data and to produce the figure. (b) Layer-averaged intensities before and after different corrections. A similar curve for calculated intensities is shown for comparison. Observed intensities show clear modulation, with every 5n layer being strong (corresponding to the l layers with sharp Bragg spots). The different corrections remove the modulation and improve the fit to the calculated data, with DIGS being more efficient.

Figure 2

The crystal packing, tNCS and lattice-translation disorder in crystal form B. The two monomers related by tNCS, which form the asymmetric unit of an ordered P2₁2₁2 crystal domain, are shown in red and blue. The tNCS vector corresponds to the largest non-origin Patterson peak. There are two distinct packing possibilities for two adjacent vertical rows of tetramers. The tops of the tetramers from one row can pack against the bottoms of the tetramers from another row, as shown for the magenta and cyan rows, respectively. Alternatively, the tetramers from these rows could pack bottoms against tops, with the magenta row shifted down along z by 24 Å. The shift vector t _d is shown by the red arrow and corresponds to the second largest non-origin Patterson peak. In the first approximation, the contacts between rows are geometrically identical in the two packing modes although made by different pairs of tetramers. This packing ambiguity is responsible for translocation defects with translocation vector t _d.

Figure 3

Model and electron-density maps (blue, 2F _o − F _c, 0.25 e⁻ Å⁻³, 1.15 r.m.s.d.; green and red, F _o − F _c, ±0.16 e⁻ Å⁻³, 2.8 r.m.s.d.) using no or different corrections of lattice-translocation disorder. The final model is shown in atomic stick representation. A C^α trace shows this model translated by t _d, the translocation vector in (1), which also represents the vector between origin and non-origin Patterson peaks. Dotted lines connecting two helices indicates this translation vector. The second trace is shown to illustrate that the ghost density is a translated copy of the actual density. (a) Map originating from refinement without using any correction. Clear ghost density is visible. (b) Maps from simple correction, i.e. demodulation using the cosine function. (c) Maps resulting from correction using the Patterson map flattening protocol. (d) Maps resulting from DIGS (§3.3).

Figure 4

Correction of intensities reveals new traces. The final model of CRMP-4 (crystal form B) and OMIT electron-density maps are shown. The C-­terminal residues from 489 onwards of chain B (magenta C atoms) were excluded from the map calculations. (a) The electron density calculated using nonmodified data. Unambiguous tracing of the C-terminal residues is not possible owing to spurious ghost density. (b) Electron density from DIGS-corrected data. The C-terminal residues can be traced. The 2F _o − F _c maps contoured at 0.24 e⁻ Å⁻³ (1.1 r.m.s.d.) and F _o − F _c maps contoured at ±0.16 e⁻ Å⁻³ (±2.8 r.m.s.d.) are shown in blue and green/red, respectively.

Figure 5

A model of the lattice-translocation disorder. Tetramers are shown as blue shapes with white lines indicating the approximate contours of monomers. As in Fig. 2 ▶, two adjacent columns of tetramers can be related by one of the two alternative translation vectors, v ₁ or v ₂. Thin black rectangles indicate ordered domains with local P2₁2₁2 symmetry, in which v ₁ and v ₂ alternate. Sequences of repeated vectors (here, v ₁ v ₁ v ₁) occur at lattice-translocation defects. The translocation vector t _d defines the relative shift of neighbouring domains from the position in which they would have formed a continuous single crystal.

Figure 6

Location of the I141V mutation of CRMP-4 in ALS patients. (a) Ribbon presentation of the CRMP-4 tetramer. The arm–lever interface is formed by, for example, chains A and B (red and blue) and the arm–arm interface by, for example, chains A and D (red and gold). The position of the I141V mutation is indicated by grey spheres. (b) Close-up stereoview of the I141V mutation. The mutated Ile is shown in yellow stick representation, with the surrounding interacting residues in green. Mutation of Ile to Val would result in a small void in the shown hydrophobic cluster.

Figure 7

Residues that are potentially important in determining hetero-oligomerization and homo-oligomerization in CRMPs. The crystal structures of CRMP-1 (cyan), CRMP-2 (yellow) and CRMP-5 (green) are superimposed onto CRMP-4 (orange). Both interfaces, (a) arm–arm and (b) arm–lever, contain conserved and nonconserved residues. Secondary-structure elements and interacting residues are labelled for CRMP-4, with the corresponding residues of CRMP-1, CRMP-2 and CRMP-5 given in parentheses.