Crystal structures of holo and Cu‐deficient Cu/Zn‐SOD from the silkworm Bombyx mori and the implications in amyotrophic lateral sclerosis

Most of the Cu/Zn-SODs are homodimeric in solution with one copper and one zinc ion per subunit. Each subunit folds as an eight-strand Greek-key β-barrel stabilized by an intrasubunit disulfide bond near the active site.3 Two major structural elements protruding from the β-barrel are termed electrostatic loop and zinc loop. The electrostatic loop connecting β6 and β7 has positively charged residues finely arranged around the active site to facilitate the diffusion of the negatively charged superoxide anion.4 The zinc loop contains two distinct substructures, one of which is anchored to the β-barrel via the disulfide bond between Cys57 and Cys146 and is thus called "disulfide loop." The other substructure includes all four zinc ligands and forms the zinc binding site. The electrostatic loop and zinc loop are bridged via hydrogen bonds between the nonligand nitrogen of the copper ligand His46 and the side chain of Asp124. These two structural elements form the substrate accessing channel leading to the active site.3 Mutations in human Cu/Zn-SOD (hSOD1) are associated with about 20% of the familial cases of amyotrophic lateral sclerosis (fALS), which is known as one of the neurodegenerative disorders characterized by the progressive loss of motor neurons in brains and spinal cord leading to paralysis and eventual death.5, 6 The mutations in fALS-linked hSOD1 (fALS-hSOD1) have been mapped to nearly all regions, including the dimeric interface, loop regions, β-barrel, disulfide bond cysteines, and the Cu/Zn ligands (a full list can be found at http://www.alsod.org). It was first believed that the fALS-hSOD1 had impaired enzymatic activity leading to the increased oxidative damage to neurons.7 However, with the identification of more fALS associated human SOD1 mutants, the majority of these mutant proteins were biochemically characterized to have enzymatic activity comparable to the wild-type, suggesting that these mutations are most likely to confer a dominant toxicity on the protein, rather than a loss-of-function.8, 9 This hypothesis is strongly supported by the mouse model of ALS disease in which the transgenic mice expressing fALS associated mutants of hSOD1 develop a late-onset progressive motor neuron disease that mimics the human disease,10 in contrast with the SOD1 knock-out mice that do not show any of the ALS symptoms.11 Although the mechanism of the toxicity is still unknown, the aberrant aggregation of SOD1 mutant proteins is strongly suggested to play important roles in the etiology of the disease.12 Several fALS associated hSOD1 mutant proteins have been confirmed in vitro to form soluble oligomers or even linear/helical fibrils via nonnative interfaces.13-16 The Cu/Zn-SOD from the silkworm Bombyx mori (BmSod1) shares 63% sequence identity with hSOD1, with 56 different residues out of the total 153 residues and a number of these "substituted" sites can be mapped to the ALS-linked hSOD1 mutation sites (Fig. 1). Here, we report the crystal structures of BmSod1 in both holo and Cu-deficient forms. Structural analyses reveal that both forms are arranged as helical fibrils and further into water-filled nanotubes via nonnative interfaces in the crystalline environment. Combined with sequence analysis, it was suggested that amino acid substitution in those regions responsible for forming nonnative interfaces may impair the hSOD1-like native conformation, which originally opposes aggregation by negative design.17 Pairwire alignment of BmSod1 and hSOD1. The alignment was performed using MultAlin18 and ESPript.19 The secondary structural elements of hSOD1 are displayed at the top of the alignment. The α-helices, 310 helices, β-sheets, and strict β-turns are denoted α, η, β, and TT, correspondingly. The mutation sites related to ALS were annotated and T96 of BmSod1 was marked by a filled circle. Total RNA was extracted from the silkglands of 5 instar larvae of the silkworm P50 with Trizol reagent. The first strand of cDNA was synthesized from 20 ng of the total RNA using Super Script II reverse transcriptase and random primer. The target gene was then amplified using sense (5′CGTCTCATATGCCCGCCAAAGCAGTT3′) and antisense (5′CTTGCGGCCGCTTAAATCTTGGCCAA GCC3′) primers. The amplified fragments encoding BmSod1 were then cloned to a modified pET28a vector (Novagen) with an additional 6×His coding sequence at the 5′ end of the genes. The recombinant plasmid was transformed into E. coli Rosetta (DE3) competent cells, which were cultured in 400 mL 2×YT with supplement of 0.01 mg/ml kanamycin. The culture was grown at 37°C to an A600 nm of 0.6 and induced with 0.2 mM IPTG for 4 h. After harvesting, cells were resuspended in 50 mL buffer of 100 mM NaCl, 20 mM Tris-HCl, pH 8.0. After three cycles of freeze-thaw followed by 3 min sonication, the lysed cells were centrifuged at 16,000g for 20 min. The supernatant was loaded onto either a Ni2+-NTA or a Cu2+-NTA affinity column (GE Healthcare) followed by gel filtration using Superdex 75 column (Amersham Biosciences) equilibrated with 100 mM NaCl, 20 mM Tris-HCl, pH 8.0, 14 mM β-mercaptoethanol. The purity of the fractions was checked on the SDS-PAGE. The protein sample was crystallized using the hanging-drop vapor-diffusion method at 289 K. BmSod1 purified with Ni2+-NTA column is Cu-deficient and the crystals of Cu-deficient BmSod1 were grown from a mixture of equal volumes of 20 mg/ml protein in 100 mM NaCl, 20 mM Tris-HCl pH 8.0, 14 mM β-mercaptoethanol, 10 mM DTT and a 400 mL reservoir solution consisting of 0.2M calcium acetate, 0.1M sodium cacodylate pH 6.5, 17% PEG 4K, 15% isopropanol. When purified with Cu2+-NTA, the copper content of BmSod1 can be restored and the crystals of holo BmSod1 were grown from a mixture of equal volumes of 10 mg/ml protein in 100 mM NaCl, 20 mM Tris-HCl pH 8.0, and a 400 μl reservoir solution consisting of 2.0 M (NH4)2SO4, 0.1M Tris-HCl, pH 8.5, 10% glycerol, 1 mM CuCl2. Diffraction data of Cu-deficient BmSod1 were collected in a stream of nitrogen gas at 100K using a Rigaku MM007 X-ray generator (λ = 1.54178 Å) with a MarRearch 345 image-plate detector at School of Life Sciences, University of Science and Technology of China (USTC, Hefei, China). The data were processed with the program MOSFLM 7.0.420 and scaled with SCALA.21 Diffraction data of holo BmSod1 were collected at a wavelength of 0.9794 Å at Shanghai Synchrotron Radiation Facility (SSRF) using beamline 17U at 100 K with a MX225 CCD (MARresearch, Germany). The diffraction data was indexed, integrated, and scaled with HKL2000.22 The structure of Cu-deficient BmSod1 was solved by the molecular replacement method with the program MOLREP23 using the wild-type hSOD1 (PDB code 1AZV) as the search model. Two dimers were located in the crystallographic asymmetric unit. The initial model was refined by using the maximum likelihood method implemented in REFMAC524 as part of CCP425 program suite and rebuilt interactively by using the σA-weighted electron density maps with coefficients 2mFo-DFc and mFo-DFc in the program COOT.26 Five percent of reflections were set side to calculate an R-free factor. Each monomer was refined independently, without application of noncrystallographic symmetry restraints. Bulk solvent correction was applied and solvent molecules were added using COOT. Refinement finally converged to a R-factor of 20.1% and R-free of 24.8% at a resolution of 2.05 Å. The holo BmSod1 structure was also determined by the molecular replacement method with MOLREP using the coordinates of a monomer of the Cu-deficient BmSod1 as the search model. One dimer was located in the asymmetric unit. The initial model was refined using the program REFMAC5 and manually rebuilt in the program COOT. No NCS restraints were applied during the refinement. Refinement finally converged to a R-factor of 22.3 and R-free of 25.8% at a resolution of 1.80 Å. To reduce geometry bias of the metal ligands, no stereo-chemical restraints on metal ligand bonds or angles were applied. Both models were validated by using the program MOLPROBITY27 to correct the obvious clashes and bad rotamers in the later stages of refinement. The final models were evaluated with the programs MOLPROBITY and PROCHECK.28 The final coordinates and structure factors were deposited in the Protein Data Bank (http://www.rcsb.org/pdb) under the accession codes of 3L9E and 3L9Y for Cu-deficient and holo BmSod1, respectively. The data collection and structure refinement statistics were listed in Table I. The buried surface area was calculated with AREAIMOL as part of the CCP4i program suite and the solvation free energy gain ΔiG was calculated with the EMBL-EBI PISA server29 (http://www.ebi.ac.uk/msd-srv/prot_int/cgi-bin/piserver). All structure figures were prepared with the program PyMOL.30 The recombinant BmSod1 produced by E. coli has extremely low occupancy of copper, as indicated by atomic absorption spectroscopy, and we thus refer to it as Cu-deficient BmSod. We restored the copper content by chromatography in a Cu-NTA column, and refer to that protein as holo BmSod. The crystal of Cu-deficient BmSod1 crystal diffracts to 2.05 Å resolution and belongs to the P65 space group, with four molecules in an asymmetric unit, whereas that of holo BmSod1 diffracts to 1.80 Å resolution and belongs to the P64 space group, with two molecules in an asymmetric unit. The overall structures of the holo as well as the Cu-deficient BmSod1 preserve the typical Cu/ZnSOD tertiary structure, which adopts a flattened Greek-key β-barrel motif consisting of eight antiparallel β-strands connected by loops, turns or short α-helices [Fig. 2(A)]. The copper binding site of Cu-deficient BmSod1 was not perturbed by copper deletion and structures of both holo and Cu-deficient BmSod1 dimers are quite similar, with an overall root mean square deviation (RMSD) of 0.4 Å over 289 Cα atoms (Fig. 2B, 2C). Two monomers assemble into a tight dimer through mutual interactions between loop IV and strand β8, with a buried surface area of ∼750 Á2. The two strictly conserved residues Cys56 and Cys146 form an intrasubunit disulfide bond that anchors loop IV to the β-barrel scaffold. These two cysteines are essential for the enzyme's folding. All these structural feathers are very similar to those of hSOD1 as well as Cu/Zn SODs from other species. A: Cartoon representation of the holo BmSod1 monomer. The bound zinc and copper ions were shown in sphere. B: Superposition of the dimers of holo (cyan) and Cu-deficient BmSod1 (pink). C: Superposition of the zinc and copper binding sites of holo (cyan) and Cu-deficient BmSod1 (pink). D: Fibril assembly interfaces of BmSod1, the nonnative interfaces were boxed with black rectangles. E and F: Top view and front view of the water-filled nanotube of BmSod1. ALS-linked hSOD1 mutants, such as S134N, H46R, and Zn-H46R (zinc bound H46R mutant), have been reported to form either linear or helical filaments in the crystalline environment via newly gained "nonnative" interfaces. The electrostatic loop or zinc loop in these mutant proteins is partially disordered and adopts a nonnative conformation that packs onto edges of strands β5 and β6.14 Strikingly, crystal packing of both holo and Cu-deficient BmSod1 showed helical fibril arrangement quite similar to the Zn-H46R mutant of hSOD1. In the case of hSOD1 Zn-H46R, zinc loop residues 78–81 adopt a nonnative conformation and hydrogen bond to the deprotected beta-edge of the β6, resulting in a nonnative interface of 550 Á2. Repetition of this nonnative interface yields hollow nanotubes with an overall diameter of 95 Á and an inner diameter of 30 Á. For the two structures of BmSod1, the newly emerged nonnative interface is also formed by the zinc loop (involving residues P73, S74, S75, and A76) from one dimer packing onto the exposed edge of strand S6 (involving S98, I99, Q100, and S102) from the neighboring subunit, leading to helical fibrils with a diameter of ∼120 Á and a translation of ∼190 Á per turn (i.e., six dimers). The buried interface is up to ∼650 Á2, with a solvation free energy gain (ΔiG) of −5.3 kcal/mol, which is comparable to the typical tight dimeric interface of ∼750 Á2, with a ΔiG of −8.9 kcal/mol [Fig. 2(D)]. The helical fibrils further stick together to form a water-filled nanotube with an outer diameter of ∼120 Á and a water-filled inner cavity of ∼40 Á in diameter (Fig. 2E, 2F). The interface between helical fibrils buried up to ∼450 Á2 (ΔiG −4.2 kcal/mol) on average [Fig. 2(D)], involving interactions between the electrostatic loop residues E132-L133 and zinc loop residues E66-K67. The fibril arrangement of holo and Cu-deficient BmSod1 in the crystalline environment suggests an ALS-linked mutant-like nature of the wild-type BmSod1. The nonnative interface involved in fibril assembly is located in the electrostatic loop, zinc loop, and strand β6, all of which are suggested as major candidates for forming aggregation-prone interfaces.14 For ALS-linked hSOD1 mutants, structural perturbation because of metal depletion is proved to be responsible for the emergence of nonnative interfaces.14, 31 However, for either holo or Cu-deficient BmSod1, both the zinc loop and electrostatic loop remain intact, prompting us to speculate that the amino acid substitution in those regions may impair the hSOD1-like native conformation which originally opposes aggregation by negative design. Indeed, the "SSAV" tetrad (residues 74–77) from the zinc loop of BmSod1 accounting for the helical fibril assembly were replaced by a tetrad of ionizable amino acid "KDEE" in hSOD1 (Fig. 1), which may prevent the nonnative contacts through electrostatic and/or steric repulsion of the bulky charged side-chains. Also, the residues participating in the fibril-fibril stacking of BmSod1, such as E66, K67, and L133, are replaced by different types of residues in hSOD1 so that nonnative contacts could possibly be avoided. Being aggregation prone as hSOD1 mutants, why the wild-type BmSod1 does not have apparent toxic effects on Bombyx mori neurons and cause ALS-like symptoms? One possible reason is that ALS is a progressive and long term disease. Unlike humans, the life span of Bombyx mori may not be long enough for the wild-type BmSod1 to form insoluble aggregates or cause the loss of the motor neurons. It has been reported that the exogenously expressed wild-type hSOD1 as well ALS mutants in Drosophila motor neurons did not form high molecular weight aggregates from 1–49 days.32 Collectively, we speculate that Cu/Zn-SOD might be an intrinsically aggregation-prone protein during earlier stages of phylogenesis, but undergo substantial site mutations opposing self-aggregation during evolution because endogenous protein aggregates are highly toxic and even lethal to mammalian/human neurons.