gRICH68 and gRICH70 are 2',3'-cyclic-nucleotide 3'-phosphodiesterases induced during goldfish optic nerve regeneration.

Biochemical characterization of changes in gene expression that accompany optic nerve regeneration has led to the identification of proteins that may play key roles in the regeneration process. In this report, a cDNA encoding gRICH70, a novel isoform of the regeneration-induced gRICH68 protein, has been identified and characterized in goldfish. Both gRICH68 and gRICH70 show significant homology (34-36%) to mammalian 2′,3′-cyclic-nucleotide 3′-phosphodiesterases (CNPases), hence the name oldfish egeneration-nduced NPase omolog (gRICH). The predicted 431-amino acid gRICH70 protein is 88% homologous to gRICH68, and the retinal mRNA for gRICH70 is coordinately induced with gRICH68 mRNA during optic nerve regeneration. Enzymatic analysis of recombinant proteins confirms that both gRICH proteins possess CNPase activity. Despite the relatively limited sequence homology, the kinetic constants obtained suggest that both gRICH proteins are at least as efficient as recombinant mouse CNP1 in catalyzing the hydrolysis of 2′,3′-cAMP. Immunoprecipitation studies indicate that gRICH proteins are responsible for the majority of the CNPase activity detected in regenerating goldfish retinas. The evidence presented demonstrates that gRICH68 and gRICH70 correspond to a previously described doublet of acidic proteins that are selectively induced in the goldfish retina during optic nerve regeneration. Thus, CNPase enzyme activity is implicated for the first time in the process of nerve regeneration.

Biochemical characterization of changes in gene expression that accompany optic nerve regeneration has led to the identification of proteins that may play key roles in the regeneration process. In this report, a cDNA encoding gRICH70, a novel isoform of the regenerationinduced gRICH68 protein, has been identified and characterized in goldfish. Both gRICH68 and gRICH70 show significant homology (34 -36%) to mammalian 2,3-cyclic-nucleotide 3-phosphodiesterases (CNPases), hence the name goldfish regeneration-induced CNPase homolog (gRICH). The predicted 431-amino acid gRICH70 protein is 88% homologous to gRICH68, and the retinal mRNA for gRICH70 is coordinately induced with gRICH68 mRNA during optic nerve regeneration. Enzymatic analysis of recombinant proteins confirms that both gRICH proteins possess CNPase activity. Despite the relatively limited sequence homology, the kinetic constants obtained suggest that both gRICH proteins are at least as efficient as recombinant mouse CNP1 in catalyzing the hydrolysis of 2,3-cAMP. Immunoprecipitation studies indicate that gRICH proteins are responsible for the majority of the CNPase activity detected in regenerating goldfish retinas. The evidence presented demonstrates that gRICH68 and gRICH70 correspond to a previously described doublet of acidic proteins that are selectively induced in the goldfish retina during optic nerve regeneration. Thus, CNPase enzyme activity is implicated for the first time in the process of nerve regeneration.
Regeneration of the axotomized optic nerve in cold blooded vertebrates has served as a model system for the characterization of cellular and molecular events underlying regeneration in the central nervous system (CNS) 1 (1,2). While warm blooded vertebrates readily regenerate peripheral nerves, the mature CNS does not support axonal regeneration. Thus, lesioned CNS axons undergo axonal degeneration and permanent loss of nervous function (3,4). The differences involved in the loss of the regenerative capacity between the CNS and peripheral nervous system of warm blooded species on the one hand and between warm blooded and cold blooded CNS on the other are not fully understood. Experimental analysis suggests that soon after embryonic development, changes in both the neurons and the glial cell populations establish a nonpermissive environment for axonal growth in the CNS (5,6). A better understanding of the biochemical basis of these differences will have direct relevance for the understanding and treatment of traumatic and degenerative human neurological disorders. The goldfish visual system has been used extensively in CNS nerve regeneration studies (2,7). After optic nerve crush, the retinal ganglion cells undergo morphological and biochemical changes (8 -10) and regrow their axons to reform connections within the tectum with a high degree of spatial specificity (11,12), resulting in recovery of visual function (13). Studies of the molecular correlates of optic nerve regeneration have implicated several proteins in the process of regrowth, including cytoskeletal components (14 -16), cell adhesion molecules (17), ion channels (18), transcription factors (19), and several proteins of as yet unknown function (15, 20 -25).
p68/70 is a doublet of acidic proteins that is markedly induced in goldfish retinal ganglion cells and transported into the optic nerve during optic nerve regeneration (15). The doublet was purified from goldfish tissues, and a polyclonal antibody was generated and used to demonstrate a marked increase of expression of p68/70 in the retinal ganglion cells during regeneration (26). More recently, peptide sequences obtained from the purified p68/70 (27) were used to isolate a corresponding cDNA from a goldfish nerve-regenerating retina cDNA library (28). The encoded protein was designated RICH (regenerationinduced CNPase homolog) to underscore the homology with mammalian CNPases, enzymes highly expressed in mammalian myelin with an as yet unknown biochemical function (29). The region of homology between RICH and CNPases was localized to the COOH terminus of the proteins (28), a region that contains the catalytic activity of mammalian CNPases (30). The RICH protein showed several characteristics that supported its relationship with the p68/70 doublet, including a similar isoelectric point (acidic) and amino acid composition. The RICH mRNA was found to be induced during regeneration as well (28).
It remained unclear which of the two peptides in the p68/70 doublet was encoded by gRICH. It had been established previously that the p68/70 protein components migrated anomalously in SDS-PAGE, probably due to effects of protein shape and low isoelectric point and of possible post-translational modifications (27 differences in apparent molecular weights of p68/70 on SDS-PAGE gels and the predicted molecular mass for gRICH68 of approximately 45 kDa (28).
In the present report, a cDNA encoding a second isoform of RICH has been isolated and sequenced. It encodes a slightly larger protein that is highly homologous to RICH. The corresponding mRNA is also induced during regeneration with a very similar time course to that for the initially characterized cDNA. The evidence reported here supports the identity of these proteins with the p68/70 doublet components, and hence, they have been renamed gRICH68 and gRICH70. Moreover, the enzymatic properties of each of the gRICH proteins have been determined, confirming that both are highly active CNPases.

MATERIALS AND METHODS
Animals-Goldfish (Carassius auratus) 6 -9 cm in body length were obtained from Grassy Forks Fisheries (Martinsville, IN) and maintained in aerated tanks at 25°C. The intraorbital, unilateral optic nerve crush procedure has been described previously (13).
Library Screenings-The isolation of the clone g-RICH-1 from a cDNA library from goldfish optic nerve-regenerating retina has been described previously (28). It contained two unrelated cDNA inserts (28). A PCR fragment containing the full ORF of gRICH68 is described below under "Bacterial Expression and Purification of gRICH Proteins and mCNP1." Approximately 20 ng of this fragment (template) and the antisense primer were used to synthesize a radiolabeled probe (5 ϫ 10 9 cpm/g) by asymmetric PCR. The probe was used to screen a goldfish genomic library (kindly provided by Dr. N. Schechter) by standard hybridization techniques (31). Clone g-RICH-G4 was purified by two further rounds of hybridization, and its insert was mapped by restriction analysis. Southern blotting using the same probe was used to determine the restriction fragments bearing sequences of interest. The full RICH-related cDNA insert from the g-RICH-1 clone and a fragment from the g-RICH-G4 genomic insert (part of exon 1) were sequenced (both strands). DNA sequencing was performed by the dideoxynucleotide chain termination method by using a Sequenase DNA polymerase kit (U.S. Biochemical Corp./Amersham Corp.). Sequence analysis was performed with the University of Wisconsin Genetics Computer Group and DNASTAR software packages. The sequence reported here has been submitted to the GenBank data base.
RNase Protection Assays-The procedure used for the RNase protection assays and the total RNA isolation from goldfish retinas in a time course experiment have been described previously (28). A 183-base pair EcoRI-BamHI fragment from the gRICH70 ORF, highly divergent from the gRICH68 corresponding sequence, was selected as template and subcloned into pBluescript-KS (Stratagene), generating plasmid pBKS-gRICH70-EB. The plasmid pBKS-gRICH70-EB was digested with EcoRI, and a 234-base-long riboprobe (10 9 cpm/g) was synthesized by run-off transcription using T7 RNA polymerase (Life Technologies, Inc.) by standard protocols (31). To generate sense RNA, pBKS-gRICH70-EB was digested with BamHI, and T3 RNA polymerase (Life Technologies, Inc.) was used in the synthesis reaction. Four g of the retina total RNAs (0, 2, 5, 9, 20, and 35 days post-optic nerve crush (PC)) were used in the sample tubes. For the sense controls, several amounts of sense RNA (0, 0.3, 1, 3, 10, and 30 pg) were mixed with 4 g of yeast tRNA. The RNAs were hybridized to 4 ϫ 10 5 cpm of antisense riboprobe. After the RNase protection assay procedure, the protected fragments (188 bases in sample tubes, 187 bases in sense controls) were separated in a 6% acrylamide sequencing gel. Radiolabeled DNA markers and 2 ϫ 10 3 cpm of the antisense riboprobe were run in the gel as references. PhosphorImager (Molecular Dynamics) quantitation was performed as described previously (28).
Bacterial Expression and Purification of gRICH Proteins and mCNP1-All of the enzymes used were from Life Technologies unless otherwise indicated. The plasmid pKK233-2 (Clontech) was modified for the purpose of expression of the gRICH proteins and mCNP1. The unique BamHI site was removed by digestion with BamHI, treatment with Klenow DNA polymerase, and blunt end religation. The polycloning site was modified to include a BglII restriction site immediately next to the ribosome binding site using a previously described PCR mutagenesis method (32). A BglII-PstI synthetic fragment, containing an ATG codon encoding the initiator methionine, followed by seven codons for histidine and by an in-frame BamHI, was inserted in the plasmid. The insert was generated by hybridization of two oligonucleo-tides obtained from the University of Michigan DNA Core Facility (5Ј-GAT CTA CCA TGG CAC ATC ATC ACC ATC ACC ACC ACG GAT CCA GAT CTG CA-3Ј and 5Ј-GAT CTG GAT CCG TGG TGG TGA TGG TGA TGA TGT GCC ATG GTA-3Ј). The plasmid generated in this way was designated pKKR2. Any BamHI-or BglII-flanked insert containing the ORF of interest in frame can be cloned in the unique BamHI site of pKKR2, allowing isopropyl-1-thio-␤-D-galactopyranoside-regulated expression of an NH 2 terminus heptahistidine-tagged protein in bacteria. Specific sense and antisense primers were used to generate PCR fragments containing the full ORF of gRICH68 and gRICH70 flanked by BglII sites, as well as the full ORF of mCNP1 flanked by BamHI sites. A subclone containing the full cDNA insert from clone g-RICH-8 and a subclone that fuses sequences from the g-RICH-G4 genomic clone and the g-RICH-1 cDNA clone were used as templates in the gRICH68 and gRICH70 PCR reactions. For the mCNP1 reaction, mouse brain total RNA was used as template in a reverse transcriptase-PCR reaction. The fragments were subcloned in pKKR2 generating the pKKR2-gRICH68, pKKR2-gRICH70, and pKKR2-mCNP1 constructs. In all cases, the subcloned PCR fragments were fully sequenced. No mutations were detected in the gRICH68 and gRICH70 ORF fragments, and only silent mutations were detected in the mCNP1 ORF fragment.
Escherichia coli (XLI-Blue strain, Stratagene) were transformed with the pKKR2-gRICH68, pKKR2-gRICH70, and pKKR2-mCNP1 plasmids and grown at 37°C in 0.5 liters of LB medium until they reached an optical density (at 600 nm) of 0.3. Then isopropyl-1-thio-␤-D-galactopyranoside was added to reach a concentration of 0.3-0.5 mM, and expression of the proteins was allowed for 3 h at room temperature. The bacteria were collected by centrifugation, resuspended in 25 ml of ST-lysis buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 g/ml pepstatin) and lysed by sonication (total of five pulses of 1-min duration at medium intensity). The lysates were cleared of debris by centrifugation at 10,000 ϫ g for 10 min. Imidazole (pH 7) was added to reach a final concentration of 20 mM, and the lysates were loaded on a nickel-nitrilotriacetic acid column (0.5 ml of resin, Quiagen). The column was washed with ST-20 buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 20 mM imidazole) and, if necessary, with ST-50 buffer (same as ST-20 but with 50 mM imidazole). The affinity-purified proteins were eluted with ST-250 buffer (same as ST-20 but with 250 mM imidazole). Glycerol was added to reach a final concentration of 20%, and the purified proteins were stored at Ϫ80°C. A working dilution of the proteins (50 ng/l) was stored at Ϫ20°C in ST buffer containing 50% glycerol with no apparent loss of enzyme activity for several months. Protein concentrations were determined (Bio-Rad) on a Beckman Biomek-1000 automated workstation. SDS-PAGE was used to monitor the purification procedure and the purified proteins (Bio-Rad minigel apparatus). The purified histidine-tagged proteins (total tag size 12 amino acids) were designated H7-gRICH68, H7-gRICH70, and H7-mCNP1.
Determinations of Kinetic Constants-CNPase activity was measured with an alkaline phosphatase (AP)-coupled assay method based on a previously published procedure (33). The reactions were performed in CNPase reaction buffer (50 mM MES, pH 6) at 30°C for 20 min in a final volume of 100 l containing the indicated concentrations of 2Ј,3Ј-cAMP as substrate (Sigma). Five ng of H7-gRICH68, 2.5 ng of H7-gRICH70, and 10 ng of H7-mCNP1 were used in the assays. The CNPase reaction was stopped by boiling for 1 min, and the coupled reaction was initiated then by the addition of 50 l of CAP-buffer (300 mM Tris-HCl, pH 9, 21 mM MgCl 2 ) containing 1 unit of calf intestine AP (Boehringer) and was incubated at 37°C for 20 min. The released inorganic phosphate was detected by a sensitive chromogenic method (34). All measurements were performed in triplicate and were corrected with control values obtained in parallel in assays containing all of the components but the specific CNPase enzyme. The measured rates were used to obtain Lineweaver-Burk plots using SigmaPlot software (Jandel).
Paper Chromatography Separation of CNPase Reaction Products-CNPase reactions were performed with 10 mM 2Ј,3Ј-cAMP in CNPase reaction buffer for 30 min at 30°C in a final volume of 50 l. One g each of H7-gRICH68, H7-gRICH70, and H7-mCNP1 were used. The products of the reactions were separated by paper chromatography as described previously (35). Five l of each of the reactions were loaded side by side with 5 l each of a 10 mM solution of 2Ј,3Ј-cAMP, 2Ј-AMP, or 3Ј-AMP, employed as standards for comparison (all from Sigma). The developed chromatogram was photographed under short wave ultraviolet light transillumination.
Transient Transfections into HEK-293 Cells-The PCR fragments containing the ORFs for mCNP1, gRICH68, and gRICH70 (described above under "Bacterial Expression and Purification of gRICH Proteins and mCNP1") were subcloned into the unique BglII site of pCMVneo (36), generating the plasmids pCMVneo-mCNP1, pCMVneo-gRICH68, and pCMVneo-gRICH70. Thirty g of each of these constructs and of the parental pCMVneo vector were transiently transfected into HEK-293 cells (100-mm Petri dishes at 50% confluency) by the calcium phosphate coprecipitation method (31). Forty-eight h after transfection, the medium was removed, and the plate was stored at Ϫ80°C for future lysate preparation.
A polyclonal antibody was generated against gRICH protein by injection of purified H7-gRICH68 into rabbits (Research Genetics). The anti-gRICH antiserum, collected 10 weeks after the initial injection of the antigen (with two additional boost immunizations), contained a high titer of anti-gRICH antibodies as detected by Western blotting and enzyme-linked immunosorbent assay with the purified gRICH proteins. Only slight cross-reactivity with mCNP1 could be detected, even with larger amounts of mCNP1. Western blotting was used to analyze the expression in transfected HEK-293 and goldfish retina extracts. Two g of total protein from the HEK-neo, HEK-mCNP1, HEK-gRICH68, and HEK-gRICH70 extracts were loaded side by side with 40 g of total protein from the gf-retina-R35 and gf-retina-C35 extracts in 10% SDS-PAGE. Fifty ng of the purified histidine-tagged proteins were also loaded as controls. A 1:1000 dilution of anti-gRICH antiserum was used as primary antibody. A 1:5000 dilution of goat anti-rabbit-AP (Life Technologies) was used as secondary antibody. The blots were developed with the nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate substrate system (Life Technologies, Inc.). Additionally, the blots were photographed with a Kodak DC40 digital camera for quantitation with Kodak 1D image analysis software.
CNPase assays were performed with the extracts by the AP method (described under "Determinations of Kinetic Constants"). The concentration of the substrate, 2Ј,3Ј-cAMP, was 4 mM. The amounts of total protein used as a source of enzyme were 0.1 g of the HEK-neo, HEK-mCNP1, HEK-gRICH68, and HEK-gRICH70 extracts and 1 g of the gf-retina-R35 and gf-retina-C35 extracts. Assays were performed in triplicate, and control values were obtained from assays with all of the components except the enzyme source and subtracted from the values obtained with the extracts.
Immunodepletion Analysis-The HEK-mCNP1, HEK-gRICH68, and HEK-gRICH70 extracts were used as positive controls. They were diluted to 0.05 g/l in ST-0.1% T buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1% Triton X-100), and 50 l of the diluted extracts were then incubated with 5 l of anti-gRICH antiserum at 4°C for 1 h in a horizontal rotator. Next, 50 l of a suspension of protein A-Sepharose beads (Pharmacia Biotech Inc.) in ST-0.1% T buffer (40% beads) was added and incubated at 4°C for 1 h in a horizontal rotator. The tubes were then spun in a microcentrifuge, and the immunodepleted supernatants were transferred to new tubes. For the gf-retina-R35 extract, the same procedure was followed except that the extract was diluted to 1 g/l and the volumes used were half of those described for the controls. In all cases, control immunodepletions were performed with the preimmune serum obtained from the same rabbit. All of the immunodepletion procedures were performed in triplicate. The immunodepleted supernatants were used in CNPase assays by the AP method (described under "Determinations of Kinetic Constants"). The determined specific activities were then normalized to the results obtained with the preimmune controls for the same extract.

RESULTS
Cloning of a Second Isoform of RICH, gRICH70 -The isolation of a cDNA encoding RICH, based on peptide sequences derived from purified p68/70 has been previously reported (28). A third clone (g-RICH-1) was obtained in the original screening, but the cDNA insert of that clone was not fully characterized at the time, since it did not contain a full ORF and was in fact the result of the fusion of two unrelated cDNAs (28).
Therefore, a goldfish genomic library was screened to obtain the sequences corresponding to the 5Ј-end of the new mRNA that were missing in the g-RICH-1 cDNA insert. Clone g-RICH-G4 was isolated in this screening, and partial sequence of the region immediately upstream of that of the g-RICH-1 cDNA insert was determined. The sequence obtained showed continuous homology to that of the g-RICH-8 cDNA up to the 5Ј-untranslated region, 25 bases upstream of the initiator ATG codon, suggesting that this sequence corresponded to a single exon. The composite sequence is represented in Fig. 1. Evidence presented below has indicated the identity of the two isoforms of RICH with the components of the p68/70 protein doublet. The first isoform (28) corresponds to the faster migrating protein of the p68/70 doublet, while the novel isoform described here corresponds to the slower migrating protein; hence, they have been renamed gRICH68 and gRICH70, respectively. The predicted gRICH70 protein is 431 amino acids in length, 20 amino acids longer than gRICH68, and has a predicted molecular mass of 46,972 Da and a calculated isoelectric point of 4.39.
An alignment of gRICH70 with gRICH68 is presented in Fig.  2. The two isoforms are highly homologous (88% amino acid identity) over their entire sequence, and both proteins contain an identical isoprenylation consensus sequence (CTIL) at the COOH terminus (37). The homology with mammalian CNPases is clearly localized to the COOH terminus (Fig. 2), corresponding to the catalytically active region of these proteins (30).
gRICH70 mRNA Is Induced in Nerve-regenerating Retinas during Goldfish Optic Nerve Regeneration-A significant induction of the gRICH68 mRNA during goldfish optic nerve regeneration has been previously demonstrated by a specific RNase protection assay (28). In similar studies of the gRICH70 mRNA, induction was detected in the regenerating retinas (R lanes) as early as 2 days PC and continued to increase up to 20 days PC, declining at later times (Fig. 3A). No significant induction was detected in the control retinas (C lanes). This time course of the induction resembled that seen previously for the gRICH68 mRNA, suggesting coordinate regulation of gRICH68 and gRICH70 mRNAs. Sense RNA controls were included in the assay for accurate quantitation in a Phosphor-Imager. The quantitation indicated a maximal induction of gRICH70 mRNA of 6-fold over control retina levels 20 days PC (Fig. 3B). It also suggested that the levels of gRICH70 mRNA are approximately 2-3-fold lower than those of gRICH68 in normal retinas. These retinal RNA samples have been previously tested by an RNase protection assay with a goldfish 18 S rRNA riboprobe and yielded a band of similar intensity in all of the lanes (28).
The induction detected with whole retinas is likely to be an underestimation of the induction in the retinal ganglion cell layer, since constitutive expression of p68/70 proteins has been demonstrated in other retinal layers (26).
Bacterial Expression and Purification of gRICH Proteins and mCNP1-The predicted gRICH proteins showed very significant homology to the mammalian myelin CNPases, particularly in the catalytic region of these proteins, including an isoprenylation motif at the COOH terminus ( Fig. 2 and Ref. 28). CNPases hydrolyze 2Ј,3Ј-cyclic-nucleotide monophosphates specifically to the 2Ј-nucleotide monophosphate product, but their in vivo role is as yet unclear (29).
To explore further the significance of the homology between the gRICH proteins and mammalian CNPases, gRICH68 and gRICH70 proteins were expressed and purified from E. coli. mCNP1 was also purified using the same system. Bacterial expression constructs were designed so that the full ORFs for the three proteins were fused in frame to a sequence encoding a heptahistidine tag. The NH 2 -terminal heptahistidine-tagged proteins (H7-gRICH68, H7-gRICH70, and H7-mCNP1) could therefore be purified from the bacteria to near homogeneity by a single column affinity purification method (described under "Materials and Methods"). Five g of each of the purified proteins were analyzed by SDS-PAGE (Fig. 4). While the three proteins (mCNP1, gRICH68, and gRICH70) are of similar length (400, 411, and 431 amino acids long, respectively), their apparent molecular weight in SDS-PAGE is significantly different. While mCNP1 seems to migrate accordingly with the predicted molecular weight, gRICH proteins migrate significantly more slowly, slightly below the 71 kDa molecular mass standard (Fig. 4). These electrophoretic mobilities support the assignment of gRICH proteins with the p68/70 doublet. The   FIG. 1. Sequence of gRICH70 cDNA and its encoded protein. The sequence represents a composite of the sequence derived from the g-RICH-1 cDNA clone and the sequence from the g-RICH-G4 genomic clone (exon 1). The sequence contains partial 5Ј-and 3Ј-untranslated regions and the full ORF for the 431-amino acid-long predicted gRICH70 protein, as represented below the DNA sequence with the one-letter amino acid codes. Both strands of the DNA were fully sequenced.

FIG. 2.
Homology of gRICH70 with gRICH68 and mCNP1. The alignment of gRICH68 and gRICH70 predicted proteins was obtained with the BESTFIT program of the Genetics Computer Group software. gRICH70 is 20 amino acids longer than gRICH68 and shows 88% amino acid identity. Positions conserved are indicated with an asterisk under the alignment. An x is substituted in those positions also conserved in mCNP1. Note that all of the homology with mCNP1 is localized in the carboxyl terminus of the gRICH proteins.
anomalous migrations may be dependent on the sequence of the proteins themselves, since the relevant post-translational modifications do not occur in bacteria.
gRICH Proteins Are Novel and Efficient CNPases-The purified H7-gRICH68 and H7-gRICH70 were tested for CNPase activity. Assays with H7-mCNP1 were performed for comparison. Initial rates of reaction were determined in assays performed in triplicate at several concentrations (0.25-8 mM) of the substrate 2Ј,3Ј-cAMP, and the values were used to obtain Lineweaver-Burk plots. A representative experiment is presented in Fig. 5. V max and K m values obtained from three independent experiments for each enzyme were then averaged ( Table I). The kinetic constants obtained for the recombinant gRICH proteins indicated that they are more efficient CNPases than the recombinant mCNP1 (higher V max and lower K m ).
The results determined with recombinant mCNP1 cannot be directly compared with the kinetic constants obtained with native proteins purified from myelin, since mammalian CNPases have been purified only from other species and have been tested under different conditions (29). However, the K m values for all three (H7-mCNP1, H7-gRICH68, and H7-gRICH70) fall well within the range determined for mammalian CNPases purified from myelin. The V max values are slightly lower than those reported with highly purified mammalian CNPases (approximately 800 -1000 units/mg) (29). In other experiments, a different bacterial expression system (New England Biolabs), which generates proteins fused to the E. coli maltose-binding protein (MBP), was used to purify MBP-mCNP1 and MBP-gRICH68. Both enzymes were active and yielded virtually identical kinetic constants to those obtained with the corresponding heptahistidine-tagged proteins (data not shown).
The CNPase activity assay employed does not distinguish the product isomer (2Ј-NMP or 3Ј-NMP) generated in the reac- FIG. 3. gRICH70 mRNA expression during regeneration. A, RNase protection assay of gRICH70. The expression of gRICH70 mRNA during optic nerve regeneration was studied using an RNase protection assay. Total RNA was isolated from both control (C) and regenerating (R) goldfish retinas 0, 2, 5, 9, 20, and 35 days postcrush. Four g of each were hybridized to a gRICH70-specific riboprobe (234 bases). After RNase digestion, the protected fragments (188 bases) were separated by denaturing gel electrophoresis in a 6% sequencing gel, and the gel was dried and then exposed to autoradiographic film. The position of the protected fragments is indicated with an arrow. Assays containing sense RNA standards were included for accurate quantification. The time course shows a detectable induction of gRICH70 mRNA 5 days postcrush, with continuing increases up to 20 days postcrush and declining levels afterward. B, PhosphorImager quantification of the assay. The expression of gRICH70 mRNA in both control and nerve-regenerating retinas is 2-3-fold lower than that previously determined for gRICH68, but the induction time course is very similar for both. Maximal induction (approximately 6-fold) occurs 20 days postcrush (8-fold was previously observed for gRICH68). tion. Mammalian CNPases hydrolyze 2Ј,3Ј-cNMP substrates exclusively to the 2Ј-NMP, while RNases most often yield the 3Ј-NMP. It was therefore of interest to analyze the product generated from the hydrolysis of 2Ј,3Ј-cAMP by gRICH proteins. H7-gRICH68 and H7-gRICH70 generated the 2Ј-AMP specifically as did H7-mCNP1, confirming that these proteins are true CNPases (Fig. 6).
gRICH Proteins Are Equivalent to the p68/70 Doublet Components-To compare directly the gRICH68 and gRICH70 proteins encoded in the cDNAs with the p68/70 doublet from goldfish, a polyclonal antibody recognizing specifically gRICH proteins was generated. Purified H7-gRICH68 was used as the antigen for rabbit immunizations (see "Materials and Methods"), and the antibody was characterized using the purified proteins. The anti-gRICH antibody recognized both H7-gRICH68 and H7-gRICH70 and showed very weak cross-reactivity with H7-mCNP1.
Constructs for the eukaryotic expression of the encoded gRICH68 and gRICH70 proteins were generated and transfected transiently into HEK-293 cells, a system that can produce isoprenylated proteins. Control transfections were performed with the parental plasmid (pCMVneo) and with a construct expressing mCNP1 as well. Protein extracts were obtained from the transfected HEK-293 cells (HEK-neo, HEK-mCNP1, HEK-gRICH68, and HEK-gRICH70 extracts) as well as from goldfish nerve-regenerating and control retinas 35 days PC (gf-retina-R35 and gf-retina-C35 extracts).
The protein extracts were analyzed by Western blot (Fig. 7). The anti-gRICH antibody detected a doublet of proteins in the 68 -70-kDa area in the goldfish extracts. Additionally, the doublet was induced in the optic nerve-regenerating retinas and comigrated perfectly with the major protein products detected in the HEK-gRICH68 and HEK-gRICH70 extracts. All these data indicate that gRICH68 and gRICH70 proteins are identical to the p68/70 doublet components. Quantitation of the Western blot signal demonstrated an induction of the gRICH68/gRICH70 proteins detected in the regenerating retinas of 4.17 Ϯ 1.40 fold, a result correlating with the induction detected previously for the p68/70 doublet itself (26). No specific signal was detected in the HEK-neo or HEK-mCNP1 extracts (Fig. 7).
Analysis of the Enzymatic Activity of Recombinant and Native gRICH Proteins-The protein extracts used for the Western blot analysis were also tested for CNPase activity by the previously described AP-coupled method. A representative experiment is presented in Fig. 8. As expected, the HEK-mCNP1, HEK-gRICH68, and HEK-gRICH70 extracts showed high specific activity compared with the control HEK-neo extract (Fig.  8A). The activity detected in the HEK-mCNP1 extract confirmed the presence of the enzyme, indicating that the absence of a band in the Western blot (Fig. 7) was not due to lack of expression but to the specificity of the anti-gRICH antibody. An estimation of the V max of the eukaryotic recombinant gRICH68 and gRICH70 proteins can be calculated based in the levels quantitated in the Western blot ( Fig. 7; assuming that anti-gRICH recognizes equally the bacterial and eukaryotic proteins) and the specific activity levels of the extracts. This calculation yields values in the range of 500 -1000 mol min Ϫ1 mg Ϫ1 , correlating fairly well with the values obtained with the purified bacterial proteins, suggesting that the post-translational modifications in eukaryotic cells do not significantly affect the catalytic properties of the enzymes.  6. Analysis of reaction products by paper chromatography. Paper chromatography was performed to separate the products of the reactions with 2Ј,3Ј-cAMP as substrate as described under "Materials and Methods." Reactions were driven to near completion by an excess of enzyme. Standards were run for comparison. The chromatogram was air-dried and photographed under short wave ultraviolet light. All three enzymes (gRICH68, gRICH70, and mCNP1) generated 2Ј-AMP specifically. The origin and direction of the chromatogram are indicated with an arrow. The activity levels in goldfish retina were much lower than those in the overexpressing extracts (Fig. 8A), correlating well with the lower levels of gRICH proteins detected in the Western blot (Fig. 7). However, an induction of CNPase activity could be clearly detected in the optic nerve-regenerating retinas over the levels of control retinas (Fig. 8B). The average induction of three independent experiments was 3.54 Ϯ 0.84fold. The induction of the CNPase activity correlates well with the induction of the gRICH doublet detected by Western blot (Fig. 7). This result further supports the identity of gRICH68 and gRICH70 with the p68/70 doublet components. Moreover, it suggested that the gRICH proteins are contributing significantly to the total CNPase activity in retinas.
The Majority of the CNPase Activity in Nerve-regenerating Retinas Is Due to gRICH Proteins-From the analysis of the retina extracts, it could be inferred that a significant fraction of the CNPase activity detected was due to the gRICH proteins. To directly explore the proportion of the CNPase activity attributable to gRICH proteins, immunodepletion experiments with the anti-gRICH antiserum were performed. The gRICH proteins present in the extracts were immunoprecipitated by binding to the anti-gRICH polyclonal antibody, followed by precipitation with protein A-Sepharose beads, and the immunodepleted supernatants were tested for CNPase activity. As expected, the CNPase activity was severely reduced in the HEK-gRICH68 and HEK-gRICH70 extracts by the immunodepletion procedure, while CNPase activity was not significantly affected in the HEK-mCNP1 extract (Fig. 9). Further controls indicated that the preimmune serum did not have a significant effect on the activity of any extract. In addition, the anti-gRICH antibody per se (without the protein A-Sepharose precipitation step) was able to block the activity of the gRICH enzymes only by approximately 30 -40% (data not shown). The immunodepletion protocol removed most of the CNPase activity from the gf-retina-R35 extract (Ͼ98%; Fig. 9). This result was confirmed by three independent experiments and provides direct evidence indicating that the majority of the CNPase activity in nerve-regenerating retinas is accounted for by the gRICH68 and gRICH70 proteins. DISCUSSION In this report, previous biochemical studies of optic nerve regeneration in goldfish (7, 26 -28) have been extended with the cloning of a cDNA encoding a novel isoform of RICH proteins. The available cDNAs have allowed the expression of the encoded proteins both in prokaryotic and eukaryotic recombinant systems, as well as the generation of a specific polyclonal antibody that recognizes the gRICH proteins efficiently. The experimental evidence obtained confirmed the identity of gRICH68 and gRICH70 with the p68/70 protein doublet components, two induced proteins previously identified in the goldfish optic nerve regeneration model system.
The studies performed here with the purified recombinant proteins have identified gRICH68 and gRICH70 as efficient CNPases. These novel CNPases are the first cloned nonmammalian members of this family of enzymes. Since they are nevertheless related to the mammalian CNPases, they should be helpful in delimiting the regions of the protein relevant for the enzymatic activity. These highly conserved motifs will probably correspond to those relevant for the in vivo function of these proteins. Of note, those conserved sequences do not include several sequence motifs that were previously thought to be involved in catalysis (28,29).
The gRICH68 and gRICH70 proteins showed very similar migrations in SDS-PAGE and enzymatic activity when expressed in E. coli or HEK-293 cells, suggesting that post-translational modification is not the cause of the anomalous migration of gRICH proteins, nor is it necessary for enzymatic FIG. 8. Analysis of the CNPase activity of the eukaryotic gRICH proteins. A, CNPase activity analysis of the extracts. The protein extracts used in the Western blot analysis were used in CNPase assays (AP method). As seen in the bar graph, the HEK cells expressing mCNP1, gRICH68, or gRICH70 showed a large increase in CNPase activity compared with HEK-neo control, indicating further that gRICH68 and gRICH70 are potent 2Ј,3Ј-cyclic-nucleotide phosphodiesterases (and confirming the expression of mCNP1 in the transfected cells). Much lower levels could be detected in the goldfish retinas, as expected from the difference in expression levels shown in the Western blot. The results represent the average Ϯ S.D. from triplicate determinations, and this experiment is representative of three independent experiments. B, induction of CNPase activity on nerve-regenerating goldfish retinas. The enlargement of the last two bars from panel A shows an increase in CNPase activity in the nerve-regenerating retinas 35 days post-crush (versus the contralateral control retinas). The induction in CNPase activity correlates well with the induction of the gRICH doublet seen in the Western blot, suggesting that the gRICH68/ gRICH70-derived CNPase activity makes a significant contribution to the total CNPase activity of the retina extracts.
FIG. 9. Immunodepletion of CNPase activity from nerve-regenerating retinas with anti-gRICH antiserum. Anti-gRICH antiserum was used to test directly if gRICH68 and gRICH70 are the major CNPases present in goldfish retina. gRICH proteins were immunoprecipitated from the extracts, and the immunodepleted supernatants were tested for CNPase activity. Controls showed that the majority of CNPase activity was depleted from HEK-gRICH68 and HEK-gRICH70 extracts, while no significant decrease was observed with HEK-mCNP1 extract. The majority of the activity was also depleted from regenerating goldfish retina extract (98.6%), further indicating that the majority of the activity detected in the extract was due to gRICH68 and gRICH70. The results represent the average Ϯ S.D. from triplicate determinations, and this experiment is representative of three independent experiments. Black bar, preimmune; gray bar, anti-gRICH. activity. Alternatively, the acid pI of both gRICH68 and gRICH70 relative to mCNP1 may be responsible for the anomalous migration in SDS-PAGE. The overexpression systems described here should be useful for the studies of post-translational modifications of gRICH proteins and to discern their role on the gRICH proteins function. The isoprenylation of mammalian CNPases have been shown to be important for their membrane localization and for the phenotypic effects of their overexpression (38,39).
The two known isoforms of mammalian CNPases, CNP1 and CNP2, differ only in the amino terminus. They arise by alternative splicing and promoter usage (40,41). The gRICH68 and gRICH70 cDNAs showed differences all along the sequence, suggesting that they are derived from two distinct genes. Both the gRICH68 and gRICH70 mRNAs showed similar time courses of induction during optic nerve regeneration (Ref. 28 and Fig. 3). The mechanisms of transcriptional induction of specific genes following nerve injury are not well understood. The time course of induction of gRICH mRNAs closely resembles that of the ␣-tubulin (18), with a progressive increase in message levels during the period of axonal regrowth. The levels of a nicotinic acetylcholine receptor subunit mRNA follow a different time course, being induced late in the regeneration process (18), corresponding to the period of synapse reformation. The study and comparison of the promoters of the induced genes should lead to a better understanding of the signaling mechanisms underlying the transcriptional control during this process.
The expression systems utilized in these studies demonstrated that gRICH68 and gRICH70 are novel members of the CNPase family of enzymes. The best studied members of this family are the mammalian CNPases (29), although enzymes with CNPase activities have been detected in organisms as diverse as crustaceans (42) and even plants (43). The mammalian CNPases are very highly expressed in oligodendrocytes and Schwann cells, and they have been widely used as a marker of these glial cells (29). Additionally, CNPases are strongly associated with myelin membranes both by a hydrophobic isoprenyl group in their COOH terminus (38,39) and by interaction with the cytoskeleton (44,45). Overexpression experiments in cell lines (39) and in transgenic mice (46) have suggested that they are involved in the generation of membranous extensions and myelin biogenesis. The conservation between gRICH proteins and mammalian CNPases of both the enzymatic activity and the COOH-terminal isoprenylation signals strongly suggests that they must have related biochemical roles in vivo. This is particularly curious inasmuch as it is clear that, in the goldfish visual system, gRICH proteins are induced in neurons (the retinal ganglion cells) and are axonally transported down the regenerating nerve fibers (26). A possible explanation for the appearance of CNPase activity in neurons may be found in a functional commonality between the glial cell that is proliferating large amounts of membrane and supporting cytoskeleton (myelin) and the regenerating cell body of neurons, which are engaged similarly in the generation of long axonal processes. In each instance, the cell supports a massive outgrowth at a great distance from the cell body and nucleus. Although CNPase-immunoreactive protein has previously been identified in neuronal tissue (47), the present studies implicate this activity in nerve regeneration for the first time. Further studies may lead to a better understanding of the molecular basis of nerve regeneration as well as to the identification of the true physiological substrate of CNPases.