Solution structure of the catalytic domain of RICH protein from goldfish Guennadi Kozlov, Alexey Y. Denisov, Ekaterina Pomerantseva, Michel Gravel, Peter E. Braun and Kalle Gehring Department of Biochemistry, McGill University, Montreal, Quebec, Canada Axonal injuries in the mammalian central nervous sys- tem do not cause any significant regeneration response due to inhibitory signaling suppressing axon outgrowth and low trophic response [1–5]. In contrast, the axons of teleost fish regenerate upon nerve injury and have been used as a model system to study nerve regener- ation in the central nervous system [6]. A better under- standing of molecular processes leading to axonal regeneration in teleost fish could find important appli- cations for treatment of human central nervous system injuries. Previous studies have identified numerous axonal growth-associated proteins, which are induced during nerve regeneration in teleost fish [6–9]. Regeneration- induced CNPase homologs (RICH) proteins are axonal growth-associated proteins that were originally discov- ered in the studies of regenerating optical nerve in goldfish and were termed p68 ⁄ 70 based on their appar- ent molecular weight [10]. RICH proteins are induced in the retinal ganglion cells during axonal regrowth upon the optic nerve crush, and also expressed in the germinal neuroepithelium of retina, which generates new neurons throughout the lifespan of the fish [11]. The cloning of the RICH proteins from goldfish (gRICH68 and gRICH70) and zebrafish (zRICH) revealed significant homology with mammalian brain 2¢,3¢-cyclic nucleotide 3¢-phosphodiesterases (CNPases) [12–14]. CNPases hydrolyze 2¢,3¢-cyclic nucleotides in vitro and are abundant in oligodendrocytes and Schwann cells [15]. Recent studies on CNPase-null Keywords CNPase; 2H phosphoesterase superfamily; NMR solution structure; RICH protein; tRNA splicing Correspondence K. Gehring, Department of Biochemistry, McGill University, 3655 Promenade Sir William Osler, Montreal, Quebec, Canada H3G 1Y6 Fax: +1 514 3987384 Tel: +1 514 3987287 E-mail: kalle.gehring@mcgill.ca Website: http://www.mcgill.ca/biochemistry/ department/faculty/gehring/ (Received 6 December 2006, revised 16 January 2007, accepted 17 January 2007) doi:10.1111/j.1742-4658.2007.05707.x Regeneration-induced CNPase homolog (RICH) is an axonal growth- associated protein, which is induced in teleost fish upon optical nerve injury. RICH consists of a highly acidic N-terminal domain, a catalytic domain with 2¢,3¢-cyclic nucleotide 3¢-phosphodiesterase (CNPase) activity and a C-terminal isoprenylation site. In vitro RICH and mammalian brain CNPase specifically catalyze the hydrolysis of 2¢,3¢-cyclic nucleotides to produce 2¢-nucleotides, but the physiologically relevant in vivo substrate remains unknown. Here, we report the NMR structure of the catalytic domain of goldfish RICH and describe its binding to CNPase inhibitors. The structure consists of a twisted nine-stranded antiparallel b-sheet sur- rounded by a-helices on both sides. Despite significant local differences mostly arising from a seven-residue insert in the RICH sequence, the active site region is highly similar to that of human CNPase. Likewise, refinement of the catalytic domain of rat CNPase using residual dipolar couplings gave improved agreement with the published crystal structure. NMR titra- tions of RICH with inhibitors point to a similar catalytic mechanism for RICH and CNPase. The results suggest a functional importance for the evo- lutionarily conserved phosphodiesterase activity and hint of a link with pre-tRNA splicing. Abbreviations CNPase, 2¢,3¢-cyclic nucleotide 3¢-phosphodiesterase; RICH, regeneration-induced CNPase homolog. 1600 FEBS Journal 274 (2007) 1600–1609 ª 2007 The Authors Journal compilation ª 2007 FEBS mutant mice revealed that the absence of CNPase cau- ses axonal swelling and neuronal degeneration pointing to a role for CNPase in the maintenance of the mye- lin–axonal interface [16]. While the enzymatic activity of CNPase has been well characterized, its physiologi- cal substrate remains a mystery [17]. Structurally, RICH protein consists of three regions: a glutamate- and aspartate-rich N-terminal domain, a catalytic phosphodiesterase domain, and a C-terminal isoprenylation site (Fig. 1). Recent studies showed that the catalytic domain is fully sufficient for the in vitro activity of RICH [18], as previously shown for mamma- lian CNPase [19,20]. The catalytic domains of RICH and CNPase share a pair of conserved sequence motifs H-X-(T ⁄ S)-X (Fig. 1B) with three other groups of enzymes: fungal ⁄ plant RNA ligases, bacterial RNA ligases and fungal ⁄ plant cyclic phosphodiesterases. Together, the catalytic domains of these proteins form a superfamily of so-called 2H enzymes, which occur in evolutionary kingdoms ranging from bacteria to mammals [21]. The RNA ligases are involved in tRNA splicing. In bacteria and archaea, they join tRNA half- molecules containing 2¢,3¢-cyclic phosphate and 5¢-hydroxyl termini. Plant and yeast cyclic phosphodi- esterases (CPD or CPDase) hydrolyze ADP-ribose 1¢,2¢-cyclic phosphate to yield ADP-ribose 1¢-phosphate (at least one of these latter enzymes also hydrolyzes nucleoside 2¢,3¢-cyclic phosphates). CPDases are also thought to play a role in the tRNA splicing pathways. NMR titrations with CNPase inhibitors and mutagen- esis studies of rat CNPase [20] in combination with the high-resolution crystal structure [22] of human CNPase catalytic domain have been used to propose a cata- lytic mechanism involving the catalytic H-X-(T ⁄ S)-X motifs. Here, we report the structure of the catalytic domain from goldfish RICH determined by NMR. We show that its structure and its active site are highly similar to that of the mammalian CNPase. NMR titrations with CNPase inhibitors identified the residues in RICH involved in inhibitor binding and suggest the proteins use a similar catalytic mechanism. These findings underline the importance of the evolutionarily con- served phosphodiesterase activity of 2H proteins and suggest that a not yet understood link exists between RNA metabolism and axon growth and maintenance. RICH H-T H-T plant tRNA ligase H-T H-TGIPGxAKS T4 Pnk H-T H-T GCPGSGKS mammalian CNPase H-T H-T GLPGSGKS fish CNPase H-T H-T GLPGSGKS yeast tRNA ligase H-T H-TGCGKT CNPase domain acidic domain A B isoprenylation site Fig. 1. (A) Domain organization of RICH and the related 2H proteins. Domains are repre- sented by rectangles with functional motifs added. Cyclic phosphodiesterase domains are shown in purple. RICH and CNPase contain a C-terminal isoprenylation motif shown in red, domains with experimentally confirmed polynucleotide kinase and aden- ylyltransferase activity are in cyan and green, respectively. The negatively charged low-complexity N-terminal domain of RICH is in magenta. (B) Sequence alignment of catalytic domains of goldfish RICH (gRICH68), human and rat CNPase (hCNP and rCNP, respectively) and a homologous protein from puffer fish (Gi:47207595). The secondary structure elements refer to the solution structure of goldfish RICH. The con- served catalytic H-X-(T ⁄ S)-X motifs are shown in bold. G. Kozlov et al. Solution structure of the RICH catalytic domain FEBS Journal 274 (2007) 1600–1609 ª 2007 The Authors Journal compilation ª 2007 FEBS 1601 Results Structure of the RICH catalytic domain We determined the structure of the 24 kDa catalytic fragment of goldfish RICH protein (Fig. 2). The previ- ously determined resonance assignments [23] were used to assign NOEs from 15 N- and 13 C-edited 3D NOESY experiments (Fig. S1). The 10 structures with the low- est energy and least number of restraint violations were chosen to represent the final ensemble (Fig. 2A). The structural statistics are shown in Table 1. The folded domain extends from Leu175 to Phe386 and presents its N- and C-termini together close in space. This would position the N-terminal domain of full-length RICH protein relatively close to its Fig. 2. Structure of goldfish RICH catalytic domain. (A) Stereo view of the backbone superposition of 10 lowest energy structures of the RICH catalytic domain. The superposition was carried out using regions Pro174–Glu208 and Leu233–Phe386. (B) Ribbon representation of the RICH catalytic domain. Secondary structure elements and the N- and C-termini are labeled. (C) Backbone overlay of catalytic domains of goldfish RICH (in cyan) and human CNPase (in purple) showing overall similarity of the structures. The lowest-energy structure from the RICH NMR ensemble is used for the overlay. (D) The surface of the RICH catalytic domain shows several negatively charged patches of res- idues. The catalytic site itself is not charged. Positive charges are shown in blue, negative charges are in red. Solution structure of the RICH catalytic domain G. Kozlov et al. 1602 FEBS Journal 274 (2007) 1600–1609 ª 2007 The Authors Journal compilation ª 2007 FEBS C-terminal isoprenylation site. This feature could be responsible for the smaller degree of association with the plasma membrane observed for RICH compared with CNPase [24] as the negative charge of the N-ter- minal domain of RICH should lead to electrostatic repulsion with the membrane. The catalytic domain of RICH is composed of a highly twisted antiparallel b-sheet consisting of nine b-strands (b1–b9) (Fig. 2B). Both sides of the b-sheet are covered with a-helices. The twisted nature of the b-sheet creates two extended grooves on the opposite sides of the protein, which are occupied by the longest helices a1 and a9. A number of short helical fragments group together in the vicinity of the N-terminus. This helical patch is the most basic part of the molecule and a potential interaction surface for the preceding acidic N-terminal domain. Structural comparison with CNP A structural similarity search using the Dali database [25] showed that the best hit (Z ¼ 21.0) was the cata- lytic domain of CNPase (PDB code 1WOJ) with an rmsd of 2.8 A ˚ over 197 residues (Fig. 2C). The struc- tural similarity to 2¢-5¢ RNA ligase (PDB code 1VDX) is much weaker with an rmsd of 5.3 A ˚ over 123 resi- dues. As noticed previously [22], the NMR structure [20] of the rat CNPase catalytic domain contained an erro- neously positioned helix. This was caused by the sparse number of NMR constraints in this part of the mole- cule. To address this, we measured residual dipolar couplings for the catalytic domain of rat CNPase using the C 12 E 5 ⁄ hexanol liquid crystalline medium (data not shown). Analysis of these residual dipolar couplings added invaluable information about this region of the rat CNPase structure and allowed us to identify several misassigned NOE constraints. The structure was recal- culated with the addition of residual dipolar coupling constraints and deposited to the RCSB database under the accession code 2ILX (supplementary Table S1). The corrected solution structure is in a good agree- ment with the crystal structure of human CNPase cata- lytic domain (rmsd of 2.3 A ˚ over 205 residues). Sequence alignment of the catalytic domain of RICH and related proteins (Fig. 1B) identifies the big- gest difference between RICH and CNPase as the seven-residue insert in the helical region between b4 and b5 strands of RICH. This insert results in addi- tional a-helical structure in RICH comprising helices a6 through a8 and causes this to be the most structur- ally dissimilar region when comparing the two pro- teins. The functional significance of this difference is unclear but, of note, the recently identified, CNP- related protein from the puffer fish (gi:47207595) also contains a long 34-residue insert, on this side of the molecule, between helices a4 and a5 (Fig. 1B). The catalytic domains of RICH and CNPase differ significantly in their surface charges. This changes the overall highly positive charge of the CNPase catalytic domain to a surface dominated by negatively charged patches in RICH (Fig. 2D). Interestingly, the region around the catalytic H-X-(T ⁄ S)-X motifs is relatively neutral in both RICH and CNP. Thus, it is likely that the overall charge difference between RICH and CNPase is more related to protein–partner interactions and less related to their catalytic activity on physiological substrate(s). We measured heteronuclear NOEs for the RICH cat- alytic domain to identify mobile regions of the structure. Besides the unstructured N-terminus, the most flexible part of the protein fragment is the internal loop immedi- ately following the helix a2 (Fig. 3). This is highly remi- niscent of CNP, where the corresponding region in the primary sequence, Gly208 to Lys214, produced the most Table 1. Structural statistics for RICH protein Restraints for structure calculations Total restraints used 1789 Intraresidue NOEs 664 Sequential NOEs 502 Medium and long range NOEs 222 Hydrogen bonds 79 Backbone angles 322 Final energies (kcalÆmol )1 ) E total 190.1 ± 3.0 E bond 7.1 ± 0.3 E angle 91.4 ± 3.0 E impr 9.1 ± 0.4 E repel 39.6 ± 1.4 E NOE 34.0 ± 2.0 E cdih 9.0 ± 1.7 rmsd from idealized geometry Bond (A ˚ ) 0.0014 ± 0.0001 Bond angles (°) 0.30 ± 0.01 Improper torsions (°) 0.18 ± 0.01 rmsd for experimental restraints a Distances (A ˚ ) 0.017 ± 0.001 Dihedral angles (°) 0.48 ± 0.05 Coordinate rmsd from average structure (A ˚ ) b Backbone atoms (N,C a ,C¢) 0.61 ± 0.05 All heavy atoms 1.15 ± 0.07 Ramachandran analysis (%) Residues in most favored regions 84.5 ± 1.8 Residues in additional allowed regions 12.3 ± 1.6 Residues in generously allowed regions 3.0 ± 2.6 a Calculated structures had 3–8 dihedral angle violations >2° and three distance violations >0.2 A ˚ . b For residues 174 : 208 and 233 : 386. G. Kozlov et al. Solution structure of the RICH catalytic domain FEBS Journal 274 (2007) 1600–1609 ª 2007 The Authors Journal compilation ª 2007 FEBS 1603 intense peaks in the 15 N– 1 H heteronuclear single-quan- tum correlation spectroscopy spectrum indicative of backbone flexibility [20]. This region is far from the nuc- leotide binding site (vide infra) and unlikely to play a role in catalysis. Binding to CNPase inhibitors To obtain more information about the active site of RICH, we titrated 15 N-labeled catalytic domain of RICH with orthophosphate and the CNPase inhibitor, adenosine-3¢-monophosphate (3¢ -AMP). The titrations were followed by 1 H– 15 N heteronuclear single-quan- tum correlation spectra and the shifts of amide signals as a function of ligand addition recorded. These sig- nals act as a fingerprint to identify amino acid residues affected by binding and to measure the binding affinity (Fig. 4). Titration of the catalyti c domain of RICH with 3¢-AMP resulted in chemical sh ift changes of roughly 20 backbone amides, indicating binding to the protein. The biggest chemical shift changes were o bserved for T hr322 (0.67 p.p.m.), Thr236 (0.60 p.p.m.), Asp241 (0.36 p.p.m. ), Val332 (0.35 p.p.m.), Gly335 (0.27 p.p.m.), Phe239 (0.27 p.p.m.) and Ala319 (0.27 p.p.m.) (Fig. 5A). Thr236 and Thr322 are part of the H-X-(T ⁄ S)-X motifs, which are essential for the catalytic activity of the related CNPase Fig. 3. Identification of the mobile regions in the RICH catalytic domain. (A) Plot of heteronuclear NOEs identifies the a2–a3 loop (Gly215–Val221) as the most mobile place in the RICH catalytic fragment. Secondary structure and the two catalytic motifs (*) are shown. (B) Representation of flexibility in the solution structure of the RICH catalytic domain. The width of the sausage is reversely proportional to the heteronuclear NOE values. The figure was gen- erated with MOLMOL [46]. Fig. 4. NMR titration of the RICH catalytic domain with 3¢-AMP. (A) Overlay of six heteronuclear single-quantum correlation spectra of the 15 N-labeled RICH catalytic domain at different concentrations of 3¢-AMP. The color changes from cyan (unliganded RICH) to dark blue (9.3 : 1 ratio of 3¢-AMP to RICH). The most shifted amides are labeled. (B) Determination of the dissociation constant of 4.6 ± 0.3 m M for 3¢-AMP binding from the amide chemical shift changes of Gly335. Solution structure of the RICH catalytic domain G. Kozlov et al. 1604 FEBS Journal 274 (2007) 1600–1609 ª 2007 The Authors Journal compilation ª 2007 FEBS [19,20]. Mapping of the chemical shift chan ges on the RICH catalytic domain structure (Fig. 5B) shows that all the changes are closely grouped in space, thereby unam- biguously identifying this region as the catalytic site of the protein (Fig. 6). A similar pattern of chemical shift changes was observed for the rat CNPase catalytic domain [20], which confirms the structural and catalytic relatedness of the two protein s. The binding of orthophosphate results in chemical shift changes very similar to those observed upon bind- ing of 3¢-AMP (Fig. 5C). As previously shown for CNPase [20,22], phosphate binds in the active site. Fewer residues are affected by phosphate binding, which reflects the smaller size of the phosphate group leading to a more local effect (Fig. 5D). Comparison of the 3¢-AMP and orthophosphate titrations allowed us to identify the residues of RICH affected by binding of the adenine group (Fig. 5E). Located in the loop between strand b2 and helix a2, Phe239 and Asp241 appear to be in a proximity of adenine base in the RICH ⁄ 3¢-AMP complex (Fig. 5F). A 081 022 062 003 043 0 83 0.1 0.3 0.5 0.7 Residue Number Δ p.p.m.)(tfihs.mehc T236 F239 D241 T322 A319 V332 G335 081 022 062 003 0 4 3 083 Residue Number 0.1 0.3 0.2 Δ p.p.m.)(tfihs.mehc F239 D241 081 022 062 00 3 04 3 08 3 Residue Number 0.1 0.3 0.5 0.7 Δ p.p.m.)(tfihs.mehc T236 T322 V332 G335 T236 T322 V332 T236 T322 V332 F239 D241 F E D C B Fig. 5. Chemical shift perturbation plot of the 15 N-labeled catalytic domain of RICH upon titration with 3¢-AMP (A) and ortho- phosphate (C) and mapping of the chemical shift changes upon binding of 3¢-AMP (B) and orthophosphate (D) on the RICH cata- lytic domain structure. The color representa- tion is white for no change to red for the maximum change. (E) The difference between chemical shift changes from titra- tion with 3¢-AMP and orthophosphate identi- fies orientation of 3¢-AMP when bound to the catalytic domain of RICH. (F) Mapping of the chemical shift changes due to the adenine group of 3¢-AMP on the RICH structure. Fig. 6. Catalytic site of RICH. Histidines and threonines from the catalytic motifs and other residues affected by inhibitor binding are shown as sticks and labeled. G. Kozlov et al. Solution structure of the RICH catalytic domain FEBS Journal 274 (2007) 1600–1609 ª 2007 The Authors Journal compilation ª 2007 FEBS 1605 The NMR titration experiments also allowed us to estimate binding affinities. Under conditions of weak binding and fast exchange, the shifts of the signals in heteronuclear single-quantum correlation spectra can be used to measure the amount of inhibitor bound. These shifts can be fitted using a simple equation, assuming K d  [protein], to estimate the dissociation constant (K d ). The resulting values were 4.6 ± 0.3 and 12 ± 2 mm for 3¢-AMP and orthophosphate, respect- ively, binding to RICH (Fig. 4B and Fig. S2). In com- parison, orthophosphate binding to CNPase shows an identical binding affinity (K d of 12 ± 2 mm), while 3¢-AMP binds CNPase with a much better K d of 0.57 ± 0.04 mm [20]. The physiological significance of the relatively poor affinity of catalytic domain of RICH for 3¢-AMP is unclear, since 3¢-AMP is not a substrate of CNPase activity. Discussion The role of 2H proteins in myelination and nerve growth remains mysterious. RICH shows highest struc- tural similarity with CNPase in its catalytic domain and catalytic site, which suggests that the conserved 2¢,3¢-cyclic phosphodiesterase activity is important for the in vivo function of both proteins. The best-charac- terized members of the 2H protein superfamily are involved in RNA-processing pathways, specifically tRNA splicing and ligation. This leads to speculation about possible physiological substrate(s) for RICH and CNP. One of the mechanisms of tRNA splicing involves endonuclease cleavage of an intron-containing tRNA at two exon–intron borders, yielding 2¢,3¢-cyclic phos- phates and 5¢-OH termini. Following cleavage, three reactions are required to put the ends of fragmented tRNA together: first, the 2¢,3¢-cyclic phosphate is hydrolyzed by a cyclic phosphodiesterase; secondly, the 5¢-OH terminus is phosphorylated by an NTP- dependent polynucleotide kinase; and thirdly, the modified ends are joined by an ATP-dependent RNA ligase [26–31]. RNA ligases in plants and fungi con- sist of a single polypeptide chain with three domains. Despite the low sequence similarity, the domain organization of plant and fungal ligases is very sim- ilar: an N-terminal adenylyltransferase ⁄ ligase domain, followed by a polynucleotide kinase domain and a C-terminal cyclic phosphodiesterase domain (Fig. 1). In multicellular animals, all three domains are still essential, but are not necessarily in the same polypeptide [31]. The C-terminal domains in all these ligases contain two H-X-(T ⁄ S)-X motifs, which identify them as 2H proteins. The central GTP-dependent polynucleotide kinase domain of yeast ⁄ plant tRNA ligases contains an NTP-binding P-loop consensus sequence of GxxGxGKS that is critical for function. The sequence of the putative P-loop in the N-terminal domain of CNPase ( 37 GLPGSGK 44 S) is strikingly similar to that of plant tRNA ligases (GIPGSAKS for Arabidopsis thaliana) (Fig. 1; [32]), which reinforces the connection between CNPase and tRNA maturation. While CNPase is missing a ligase domain, this activity could be performed by another protein. In T4 bacteriophage, tRNA ligation is carried out by two different enzymes. The bifunctional enzyme T4 Pnk, which contains a P-loop (GCPGSGKS) almost identical to CNP, pre- pares the 3¢ and 5¢ ends of the cleaved tRNA, and T4 Rnl1 ligase reconnects the ends [33–35]. This advocates the hypothesis that CNPase is a functional homolog of T4 Pnk and participates in tRNA splicing and maturation. Intriguingly, the fish homolog of CNPase (gi:47207595) contains an additional N-terminal domain, which could potentially possess an adenylyl- transferase activity (Fig. 1). While showing higher sequence homology to the fish CNPase (56% identity versus 47% identity to human or rat CNPase), RICH appears to lack both the adenylyltransferase and kin- ase domains; little is known about the function of its acidic N-terminal segment. The cellular localization of CNPase does not contra- dict its involvement in pre-tRNA splicing. Recent stud- ies revealed that the yeast tRNA splicing endonuclease mainly localizes on mitochondria and this localization is important for its function [36]. Interestingly, one CNPase isoform (CNP2) is specifically targeted to mitochondria [37]. More intriguingly, RICH and CNPase may be involved in other RNA splicing events. XBP1 mRNA, in humans, and HAC1 mRNA, in yeast, undergo cytoplasmic splicing as part of the unfolded protein response that regulates the endoplas- mic reticulum volume and protein composition [38]. While no 2¢,3¢-cyclic phosphate intermediates have been identified in these reactions or in regulation by micro RNAs, it is not impossible that the evolutionari- ly ancient phosphoesterase activity of 2H proteins is involved in regulating membrane biogenesis in oligo- dendrocytes or neurons via RNA. In conclusion, RICH proteins have been less char- acterized than mammalian CNPases, but the strong structural similarity of these proteins suggests a sim- ilar function. The structure of RICH and nucleotide binding studies presented here represent another step towards understanding of CNP ⁄ RICH function and suggest new avenues to study these still enigmatic proteins. Solution structure of the RICH catalytic domain G. Kozlov et al. 1606 FEBS Journal 274 (2007) 1600–1609 ª 2007 The Authors Journal compilation ª 2007 FEBS Experimental procedures Protein expression and purification The catalytic domain of goldfish gRICH68 protein (residues 172–389) was subcloned into pET15b (Novagen, Inc., Madison, WI, USA) and expressed in the Escherichia coli expression host BL21 (DE3) (Stratagene, La Jolla, CA, USA) as a His-tagged fusion protein. The protein was puri- fied by immobilized metal affinity chromatography on Ni 2+ -loaded chelating sepharose column (Amersham Phar- macia Biotech, Piscataway, NJ, USA). Isotopically labeled RICH was prepared from cells grown on minimal M9 media containing 15 N ammonium chloride and 13 C glucose (Cambridge Isotopes Laboratory, Andover, MA, USA). The N-terminal His-tag was cleaved from RICH by over- night dialysis with thrombin (Amersham Pharmacia Bio- tech, Piscataway, NJ, USA) at 1 unit per mg fusion protein at room temperature. Benzamidine sepharose and Ni 2+ -loa- ded chelating sepharose were used to remove thrombin and the His-tag peptide from RICH. The resulting 222 amino acid protein contained four N-terminal extraneous residues (GSHM) from the His tag. The sequence composition of purified RICH was confirmed by mass spectrometry. NMR spectroscopy NMR resonance assignments of the catalytic domain of RICH were determined previously [23]. All NMR experi- ments were recorded at 307 K. NMR samples were 1 mm protein in 50 mm 4-morpholineethanesulfonic acid buffer, 0.15 m NaCl, 1 mm dithiothreitol at pH 6.0. NMR spectra were processed with nmrpipe [39] and xwinnmr software version 3 (Bruker Biospin) and analyzed with xeasy [40]. For titrations, 3¢-AMP and Na 2 HPO 4 were purchased from Sigma (Saint-Louis, MO, USA) and used without any additional purification. Titrations were monitored by 15 N- 1 H heteronuclear single-quantum correlation spectra following addition of inhibitors to 15 N-labeled RICH (172–389) on a Bruker DRX 600 MHz spectrometer. The experiments were recorded with 128 increments using 4–8 scans and lasted for 10–20 min. Chemical shift changes were calculated as (DHN 2 + (0.2*DN) 2 ) 1 ⁄ 2 in p.p.m. Samples contained 50 mm 4-morpholineethanesulfonic acid, 0.15 m NaCl and 1 mm di- thiothreitol at pH 6.0 and 0.4–0.5 mm RICH at 307 K. The inhibitor concentrations ranged from 0.1 to 60 mm depend- ing on the affinity and solubility of the inhibitor. The hetero- nuclear single-quantum correlation spectra of complexes were assigned by monitoring chemical shift changes upon addition of the substrate, since the binding takes place in the fast exchange. The pH of the NMR samples was monitored during the titrations and adjusted as needed. Chemical shift changes for individual residues were fit to a one-site binding equation: d ¼ d max Æ [L] ⁄ (K d + [L]) where d is the chemical shift change, [L] is the total ligand concentration (uncor- rected for binding to RICH), and K d is the dissociation con- stant. The fitting was carried out using the computer program grafit, version 3.0 (Ericathus Software, Horley, UK) to determine d max and the K d of binding. Structure calculation NOESY constraints for the structure determination were obtained from 15 N-edited NOESY (mixing time 80 ms) and 13 C-edited NOESY (mixing time 80 ms) 3D experiments acquired on a Varian Unity Inova 800 MHz spectrometer at the Quebec-Eastern Canada High-Field NMR Facility. For the structure determination, a set of ARIA-assigned [41] and manually verified 1388 NOEs were collected from 15 N- and 13 C-edited NOESY spectra of RICH (172–389). Three hundred and twenty-two backbone angles resulted from chemical shift index using the TALOS database [42]. Hydrogen bonds were predicted from NOE analysis. The starting structure was generated with modeller 6v2 [43] using human CNPase crystal structure (PDB code 1WOJ) and was in agreement with manually assigned NOEs. One hundred and fifty structures were calculated and refined using standard protocols in cns v.1.1 [44]. procheck-nmr [45] was used to check the protein stereochemical geometry. The structural statistics for 10 structures are shown in Table 1. The coordinates have been deposited in the Research Collaboratory for Structural Bioinformatics (RCSB) under PDB accession code 2I3E and the NMR assignments under BMRB accession number 7167. Acknowledgements We acknowledge Dr. M. D. Uhler (University of Michigan, USA) for a gift of gRICH68 cDNA. We thank T. Sprules for help in running NMR experi- ments at the Quebec-Eastern Canada High-Field NMR Facility. This work was funded by operating grant MOP-43967 to KG and PB from the Canadian Institutes of Health Research. References 1 Bahr M & Bonhoeffer F (1994) Perspectives on axonal regeneration in the mammalian CNS. Trends Neurosci 17, 473–479. 2 Fawcett JW & Asher RA (1999) The glial scar and central nervous system repair. Brain Res Bull 49, 377–391. 3 Horner PJ & Gage FH (2000) Regenerating the damaged central nervous system. Nature 407, 963–970. 4 Spencer T, Domeniconi M, Cao Z & Filbin MT (2003) New roles for old proteins in adult CNS axonal regen- eration. Curr Opin Neurobiol 13, 133–139. 5 Schwab ME (2004) Nogo and axon regeneration. Curr Opin Neurobiol 14, 118–124. G. Kozlov et al. Solution structure of the RICH catalytic domain FEBS Journal 274 (2007) 1600–1609 ª 2007 The Authors Journal compilation ª 2007 FEBS 1607 6 Bernhardt RR (1999) Cellular and molecular bases of axonal regeneration in the fish central nervous system. Exp Neurol 157, 223–240. 7 Skene JH (1989) Axonal growth-associated proteins. Annu Rev Neurosci 12, 127–156. 8 Caroni P (1997) Intrinsic neuronal determinants that promote axonal sprouting and elongation. Bioessays 19 , 767–775. 9 Plunet W, Kwon BK & Tetzlaff W (2002) Promoting axonal regeneration in the central nervous system by enhancing the cell body response to axotomy. J Neu- rosci Res 68, 1–6. 10 Heacock AM & Agranoff BW (1982) Protein synthesis and transport in the regenerating goldfish visual system. Neurochem Res 7 , 771–788. 11 Wilmot GR, Raymond PA & Agranoff BW (1993) The expression of the protein p68 ⁄ 70 within the goldfish vis- ual system suggests a role in both regeneration and neurogenesis. J Neurosci 13, 387–401. 12 Ballestero RP, Wilmot GR, Leski ML, Uhler MD & Agranoff BW (1995) Isolation of cDNA clones encoding RICH: a protein induced during goldfish optic nerve regeneration with homology to mammalian 2¢,3¢-cyclic- nucleotide 3¢-phosphodiesterases. Proc Natl Acad Sci USA 92, 8621–8625. 13 Ballestero RP, Wilmot GR, Agranoff BW & Uhler MD (1997) gRICH68 and gRICH70 are 2¢,3¢-cyclic-nucleo- tide 3¢-phosphodiesterases induced during goldfish optic nerve regeneration. J Biol Chem 272, 11479–11486. 14 Ballestero RP, Dybowski JA, Levy G, Agranoff BW & Uhler MD (1999) Cloning and characterization of zRICH, a 2¢,3¢-cyclic-nucleotide 3¢-phosphodiesterase induced during zebrafish optic nerve regeneration. J Neurochem 72, 1362–1371. 15 Sprinkle TJ (1989) 2¢,3¢-Cyclic nucleotide 3¢-phosphodi- esterase, an oligodendrocyte-Schwann cell and myelin-associated enzyme of the nervous system. Crit Rev Neurobiol 4, 235–301. 16 Lappe-Siefke C, Goebbels S, Gravel M, Nicksch E, Lee J, Braun PE, Griffiths IR & Nave KA (2003) Disrup- tion of Cnp1 uncouples oligodendroglial functions in axonal support and myelination. Nat Genet 33, 366–374. 17 Braun PE, Lee J & Gravel M (2004) 2¢,3¢-Cyclic nucleo- tide 3¢-phosphodiesterase: structure, biology and func- tion. In Myelin Biology and Disorders, Vol. 1 (Lazzerini, RA, ed.), pp. 499–521. Elsevier, Academic Press, New York. 18 Challa M, Chapa GR, Govindaraju S, Gonza ´ lez-Garcı ´ a M & Ballestero RP (2006) Characterization of the domains of zRICH, a protein induced during optic nerve regeneration in zebrafish. Brain Res 1100, 42–54. 19 Lee J, Gravel M, Gao E, O’Neill RC & Braun PE (2001) Identification of essential residues in 2¢,3¢-cyclic nucleotide 3¢-phosphodiesterase. Chemical modification and site-directed mutagenesis to investigate the role of cysteine and histidine residues in enzymatic activity. J Biol Chem 276, 14804–14813. 20 Kozlov G, Lee J, Elias D, Gravel M, Gutierrez P, Ekiel I, Braun PE & Gehring K (2003) Structural evidence that brain cyclic nucleotide phosphodiesterase is a mem- ber of the 2H phosphodiesterase superfamily. J Biol Chem 278, 46021–46028. 21 Mazumder R, Iyer LM, Vasudevan S & Aravind L (2002) Detection of novel members, structure-function analysis and evolutionary classification of the 2H phosphoesterase superfamily. Nucleic Acids Res 30, 5229–5243. 22 Sakamoto Y, Tanaka N, Ichimiya T, Kurihara T & Nakamura KT (2005) Crystal structure of the catalytic fragment of human brain 2¢,3¢-cyclic-nucleotide 3¢-phos- phodiesterase. J Mol Biol 346, 789–800. 23 Denisov AY, Kozlov G, Gravel M, Sprules T, Braun PE & Gehring K (2006) 1 H, 13 C and 15 N resonance assignments of the catalytic domain of the goldfish RICH protein. J Biomol NMR 36 (Suppl. 5), 75. 24 Leski ML & Agranoff BW (1994) Purification and char- acterization of p68 ⁄ 70, regeneration-associated proteins from goldfish brain. J Neurochem 62, 1182–1191. 25 Holm L & Sander C (1995) Dali: a network tool for protein structure comparison. Trends Biochem Sci 20, 478–480. 26 Abelson J, Trotta CR & Li H (1998) tRNA splicing. J Biol Chem 273, 12685–12688. 27 Greer CL, Peebles CL, Gegenheimer P & Abelson J (1983) Mechanism of action of a yeast RNA ligase in tRNA splicing. Cell 32, 537–546. 28 Apostol BL, Westaway SK, Abelson J & Greer CL (1991) Deletion analysis of a multifunctional yeast tRNA ligase polypeptide: identification of essential and dispensable functional domains. J Biol Chem 266, 7445– 7455. 29 Phizicky EM, Consaul SA, Nehrke KW & Abelson J (1992) Yeast tRNA ligase mutants are nonviable and accumulate tRNA splicing intermediates. J Biol Chem 267, 4577–4582. 30 Westaway SK & Abelson J (1995) Splicing of tRNA precursors. In Trna: Structure, Biosynthesis and Function (Soll, D & RajBhandary, UL, eds), pp. 79–92 American. Society for Microbiology, Washington, DC. 31 Sawaya R, Schwer B & Shuman S (2003) Genetic and biochemical analysis of the functional domains of yeast tRNA ligase. J Biol Chem 278, 43928–43938. 32 Koonin EV & Gorbalenya AE (1990) Related domains in yeast tRNA ligase, bacteriophage T4 polynucleotide kinase and RNA ligase, and mammalian myelin 2¢,3¢-cyclic nucleotide phosphohydrolase revealed by amino acid sequence comparison. FEBS Lett 268, 231– 234. 33 Amitsur M, Levitz R & Kaufmann G (1987) Bacter- iophage T4 anticodon nuclease, polynucleotide kinase Solution structure of the RICH catalytic domain G. Kozlov et al. 1608 FEBS Journal 274 (2007) 1600–1609 ª 2007 The Authors Journal compilation ª 2007 FEBS and RNA ligase reprocess the host lysine tRNA. EMBO J 6, 2499–2503. 34 Cameron V & Uhlenbeck OC (1977) 3¢-Phosphatase activity in T4 polynucleotide kinase. Biochemistry 16, 5120–5126. 35 Galburt EA, Pelletier J, Wilson G & Stoddard BL (2002) Structure of a tRNA repair enzyme and molecu- lar biology workhorse: T4 polynucleotide kinase. Struc- ture (Camb) 10, 1249–1260. 36 Yoshihisa T, Yunoki-Esaki K, Ohshima C, Tanaka N & Endo T (2003) Possibility of cytoplasmic pre-tRNA splicing: the yeast tRNA splicing endonuclease mainly localizes on the mitochondria. Mol Biol Cell 14, 3266– 3279. 37 Lee J, O’Neill RC, Park MW, Gravel M & Braun PE (2006) Mitochondrial localization of CNP2 is regulated by phosphorylation of the N-terminal targeting signal by PKC: implications of a mitochondrial function for CNP2 in glial and non-glial cells. Mol Cell Neurosci 31, 446–462. 38 Bernales S, Papa FR & Walter P (2006) Intracellular signaling by the unfolded protein response. Annu Rev Cell Dev Biol 22, 487–508. 39 Delaglio F, Grzesiek S, Vuister GW, Zhu G, Pfeifer J & Bax A (1995) NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J Biomol NMR 6, 277–293. 40 Bartels C, Xia T-H, Billeter M, Gu ¨ ntert P & Wu ¨ thrich K (1996) The program XEASY for computer-supported NMR spectral analysis of biological macromolecules. J Biomol NMR 6, 1–10. 41 Nilges M, Macias MJ, O’Donoghue SI & Oschkinat H (1997) Automated NOESY interpretation with ambigu- ous distance restraints: the refined NMR solution structure of the pleckstrin homology domain from beta-spectrin. J Mol Biol 269, 408–422. 42 Cornilescu G, Delaglio F & Bax A (1999) Protein back- bone angle restraints from searching a database for che- mical shift and sequence homology. J Biomol NMR 13, 289–302. 43 Sali A & Blundell TL (1993) Comparative protein mod- elling by satisfaction of spatial restraints. J Mol Biol 234, 779–815. 44 Bru ¨ nger AT et al. (1998) Crystallography & NMR sys- tem: a new software suite for macromolecular structure determination. Acta Crystallogr D Biol Crystallogr 54, 905–921. 45 Laskowski RA, Rullmannn JA, MacArthur MW, Kaptein R & Thornton JM (1996) AQUA and PRO- CHECK-NMR: programs for checking the quality of protein structures solved by NMR. J Biomol NMR 8, 477–486. 46 Koradi R, Billeter M & Wu ¨ thrich K (1996) MOLMOL: a program for display and analysis of macromolecular structures. J Mol Graph 14, 51–55. Supplementary material The following supplementary material is available online: Fig. S1. Plot of NOE number per residue. Fig. S2. Determination of the dissociation constant of 12.0 ± 1.9 m M for orthophosphate binding from the amide chemical shift changes of Gly335. Table S1. Structural statistics for rCNP catalytic domain. This material is available as part of the online article from http://www.blackwell-synergy.com Please note: Blackwell Publishing is not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corres- ponding author for the article. G. Kozlov et al. Solution structure of the RICH catalytic domain FEBS Journal 274 (2007) 1600–1609 ª 2007 The Authors Journal compilation ª 2007 FEBS 1609 . 1601 Results Structure of the RICH catalytic domain We determined the structure of the 24 kDa catalytic fragment of goldfish RICH protein (Fig. 2). The previ- ously. similarity of the structures. The lowest-energy structure from the RICH NMR ensemble is used for the overlay. (D) The surface of the RICH catalytic domain