Glutamyl ribose 5-phosphate storage disease. A hereditary defect in the degradation of poly(ADP-ribosylated) proteins.

A patient with a lysosomal storage disease, progressive neurologic degeneration, and renal failure was found to have accumulated a low molecular weight ninhydrin and phenol-H2SO4 reactive compound. Amino acid analysis and gas chromatography-mass spectrometry identified a glutamic acid moiety. Direct insertion mass spectrometry proved the carbohydrate portion to be a sugar phosphate. NaB3H4 reduction and borate electrophoresis, paper chromatography, and enzymatic digestion indicated the presence of ribose 5-phosphate. Quantitative analysis of the intact compound indicated a 1:1:1 ratio for glutamic acid: ribose:phosphate. Brain was found to contain 0.96 mumol/g, wet weight, and kidney 0.60 mumol/g, wet weight, of glutamyl ribose 5-phosphate. This substance is the linkage region in ADP-ribosylation of histones and other proteins. It is suggested that the primary defect in this patient is a genetic abnormality of ADP-ribose protein hydrolase (Okayama, H., Honda, M., and Hayaishi, O. (1978) Proc. Natl. Acad. Sci. U. S .A. 75, 2254-2257).

The covalent attachment of carbohydrate and protein in mammalian glycoproteins and proteoglycans may be divided into N-linked and 0-linked groups. The former consist of Nacetylglucosamine-asparagine linkages which predominate in plasma proteins. The 0-linkages consist of 1) N-acetylgalactosamine-serine/threonine found in secretory proteins of epithelial tissues, 2) galactose-hydroxylysine existing only in collagen type glycoproteins, and 3) the xylose-serine linkage of proteoglycans (1). Recently, several unusual linkage groups have been isolated from normal human urine including /I-Dglucopyranosyl-(1 + 3)-cu-~-fucopyranosyl-~-threonine (2) and digalactosylcysteine (3). A new type of carbohydrateprotein linkage has been described in poly(ADP-ribosylated) proteins. This covalent modification has been proposed to function in the regulation of gene expression and DNA repair ( 4 , 5 ) . Ogata et al. (6,7 ) have demonstrated that, in histones, the linkages consist of an ester bond between both the glutamic acid y-carboxyl and the terminal carboxyl of lysine with the reducing position of ribose.
The study of oligosaccharides excreted in the urine and stored in the tissues of patients with hereditary deficiencies of lysosomal glycosidases has been helpful in elucidating the *This work was supported by Grant 1-831 from the National Foundation-March of Dimes and Grant GM-28949 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ To whom correspondence should be addressed.
wide variety of oligosaccharide structures present in glycoconjugates (8). But, only one inborn error of metabolism has been identified which is due to a deficiency of an enzyme which functions to cleave the covalent linkage region of glycoproteins. This, a deficiency of 2-N-acetamido-(4'-~-aspartyl)-2deoxy-P-D-glucosylamine amidohydrolase (EC 3.5.1.26), results in an accumulation of the N-acetylglucosamine-asparagine linkage group and the higher homologous glycoasparagines (9, 10). We report the accumulation of glutamyl ribose 5-phosphate in an 8-year-old male who succumbed after a 6year course of progressive neurologic deterioration and renal failure. The patient had a biopsy proven lysosomal storage disease and the known lysosomal enzyme deficiencies were excluded. Family history indicated an X-linked recessive inheritance. A detailed clinical report will be published elsewhere. We propose that the glutamyl ribose 5-phosphate is derived from the linkage region of poly(ADP-ribosylated) proteins and that the defect in this patient is due to a deficiency of ADP-ribose protein hydrolase.

EXPERIMENTAL PROCEDURES
Extraction and Purification of Accumulated Compounds-Frozen tissue, stored a t -70 "C, was homogenized and sonicated in 10 volumes of distilled water or 15% glacial acetic acid, and centrifuged a t 10,000 X g for 30 min. The pellet was re-extracted, centrifuged, and the supernatants combined. Protein was precipitated by the addition of 50% trichloroacetic acid to a final concentration of 10% and centrifugation at 10,000 X g for 30 min. The supernatant was extracted 3 times with 5 volumes of ether to remove trichloroacetic acid and lyophilized. Respective samples were dissolved in 1 ml of 15% glacial acetic acid and chromatographed on a Bio-Gel P-2 column (0.9 X 228 cm) (Bio-Rad < 400 mesh). Carbohydrate was monitored using the phenol-HzS04 method of Dubois et al. (11). Ion exchange chromatography was performed on Dowex 50-H+ and Dowex l-CIcolumns (0.5 X 5.0 cm) (Bio-Rad) previously washed extensively with distilled water. Following application of samples and subsequent washing, retained compounds were eluted with 1 N HCl.
Carbohydrate Analysis-Acid hydrolysis was performed in sealed glass ampules by two methods: 1) 2 N HC1 a t 100 "C for 3 h, or 2) 0.5 N HCI a t 80 "C for 1 h. HC1 was removed by repeated evaporation in uucuo at 37 "C. Quantitation was achieved by the method of Takasaki and Kobata (12). 14C-1abeled sugars were obtained from New England Nuclear and reduced with NaBH4. Unlabeled sugars were obtained from Sigma and labeled with NaB3H4 (New England Nuclear).
Descending Paper Chromatography-Samples were spotted on Whatman 1 paper and developed in ethyl acetate, pyridine, 5 mM boric acid (3:2:1) for 16 h. Following drying, chromatograms were either scanned on a Packard 7201 Radiochromatoscanner, or cut into 1.0-cm segments, transferred to scintillation vials, eluted with 0.5 ml of water followed by addition of 5. FIG. 1. Chromatography of brain extract. Bio-Gel P-2 (<400 mesh, 0.9 X 228 cm, 0.67 ml/h, 1.3 ml/ fraction) chromatography was run in 15% acetic acid and followed by phenol-H2S04. The black boxes (m) indicate patient uersus control (0). The arrows indicate the elution volumes of glucose oligomers.
Diagnostics Inc., Somerville, NJ), and radioactivity determined in a Tracor Mark-111 Liquid Scintillation Counter. Instrumental Methods-Amino acid analysis was performed on a Dionex D-400 analyzer using the physiologic fluid program. X-ray fluorescence quantitation of phosphorus was done on a Kevex Ultratrace 0600 X-ray Fluorescence Subsystem pX Analytic Spectrometer with a 6100 data processor. Before analysis by mass spectrometry, samples were dried in uucuo at room temperature, and Tmsl-derivatives were prepared using N,O-bis-(trimethy1silyt)-trifluoroacetamide (Pierce Chemical Co.) and dry pyridine (l:l, v/v) at 60 "C for 30 min. Analyses were performed using a Finnigan gas chromatographmass spectrometer (Model 3200) with a INCOS data system (glass column (6' X 2 mm) packed with 3% OV-1 on Gas-chrom Q, 100/120 mesh; injector temperature 250 "C, column temperature 80 "C to 290 "C final at 12 "C/min, 70 eV). Phosphorylated samples were directly introduced into the mass spectrometer source without prior gas chromatography.
Enzymatic Digestion-Escherichia coli alkaline phosphatase (Sigma P-4252), 5'-nucleotidase (Sigma N-4005), and 3'-nucleotidase (Sigma N-7008) were incubated with 2 X lo6 dpm of authentic [3H] ribositol5-phosphate or the alkaline NaB3H4 derivative of the isolated compound under the following conditions: (a) 5'-nucleotidase, 5 units in 30 p1 of 0.67 M glycine:NaOH, pH 8.5, containing 33 mM MgClz for 1 h at 37 "C; (b) 3'-nucleotidase, 5 units in 200 pl of 0.1 M Tris-HC1, pH 7.5, for 1 h at 37 "C, and (c) alkaline phosphatase, 5 units in 200 pl of 0.1 M Tris-HCI, pH 8, at 37 "C for 16 h. The reaction mixtures were passed over Dowex 1-CI-columns (0.5 X 5.0 cm) previously washed extensively with distilled water. Following application of samples and washing with water, columns were eluted with 1 N HCI. The degree of dephosphorylation was determined as per cent of disintegrations/min in the water wash compared to the total recovered in water + HC1 eluates. Total recovery was 97 to 105%.

RESULTS
Chromatography of an aqueous or 15% glacial acetic acid extraction of brain on a P-2 gel filtration column gave the profile of carbohydrate-containing material depicted in were observed at positions corresponding to the void volume and a tetrasaccharide. The material eluting in the void volume has not been characterized. Based upon the molecular weight of the proven structure (Fig. 6), the yield of material isolated from the gel filtration column was 0.96 pmol/g, wet weight, in brain. Similar profiles were obtained from kidney tissue, except that the yield was 0.60 pmol/g, wet weight. Early efforts to isolate the included peak were tedious due to its extreme instability, in that the P-2 column was run in 0.15 M NaHC03.
The compound was found to he very stable upon changing the purification procedure to 15% acetic acid.
Thin layer chromatography of the included peak with subsequent spraying with ninhydrin and orcinol reagents indicated a single carbohydrate component and multiple amino acid spots, one of which corresponded to the orcinol positive compound (data not shown). This compound was retained on Dowex l-Cl-and washing with water removed all of the contaminating amino acids. The unknown compound was eluted with 1 N HC1. Preliminary characterization revealed the following: binding to Dowex 50-H' and Dowex 1-Cl-, reaction with ninhydrin and orcinol, no reaction with alkaline silver nitrate without prior exposure to periodic acid, and a decrease in electrophoretic mobility after treatment with bacterial alkaline phosphatase (data not shown). These data suggested the unknown compound to consist of an amino acid covalently linked to the reducing end of a sugar phosphate.
Hydrolysis under strong acid conditions (2 N HCI a t 100 "C for 3 h) and amino acid analysis revealed a single peak whose retention time and co-elution with an authentic standard indicated glutamic acid (chromatogram not shown). After repeated lyophilization to remove HCl, the hydrolysate was silylated and analyzed by gas chromatography-mass spectrometry. The total ion chromatogram is shown in Fig. 2A. Aside from volatile materials in the silylation mixture which FIG. 2. Gas chromatography-mass spectrometry of the acid hydrolysates of the isolated compound. A is the ion chromatogram of a 2 N HCl hydrolysate with the inset illustrating the mass spectrum and structure of the glutamic acid derivative. B is the spectrum and structure of pyroglutamic acid seen in a 0.5 N HCI hydrolysate. The third minor component was present in amounts too small to unequivocally identify, but exhibited a fragmentation pattern suggestive of a saccharide. The strong reaction of the intact compound with orcinol compared to its weak ninhydrin reactivity suggested that more saccharide would have been identified unless destroyed by the hydrolysis conditions. This would be the case for a deoxyhexose or pentose.
Mild acid hydrolysis (0.5 N HCl at 80 "C for 1 h), silylation and gas chromatography-mass spectrometry yielded a single peak whose mass spectrum corresponded to the Tms derivative of pyroglutamic acid (Fig. 2B)  156 (M-C02Tms)+. The absence of any peaks corresponding to phosphate or saccharide derivatives suggested that they remained covalently attached after mild acid hydrolysis because Tms derivatives of sugar phosphates are not volatilized under these gas chromatograph-mass spectrometry conditions. When the sample was examined by a direct insertion technique, the mass spectrum indicated in Fig. 3A resulted. Based on characteristic fragmentation patterns (m/z: 103, 129, 133, 191, 211, 215, 217, 230, 243, 299, and 315), the spectrum was judged to be that of the Tms derivative of a pentose phosphate (14). The closest match was with that of Tms derivative of ribose 5-phosphate. Fig. 3B illustrates the spectrum of authentic ribose 5-phosphate derivatized and run under the same conditions. In the higher mass region, ion (m/ t) 575 ("CH3)+, in the sample spectrum was weaker than in the standard, but easily identifiable. Ions characteristic of the phosphate group (m/z: 299 and 315) were present in both spectra. The ions typical of the carbohydrate moiety (m/z: 103,129,133,169,191,211,215,217,230, and 243) were a close match. Two additional ions ( n / z : 155 and 189) may arise from contaminants in the unpurified hydrolysate.
In order to identify positively the sugar phosphate, the intact unknown was reduced with NaB3H4 under mild alkaline conditions and electrophoresed according to the method of Takasaki and Kobata (12). As shown in Fig. 4A, a single species was observed migrating ahead of the standards maltitol, xylitol, ribositol, and arabinitol. This peak was in the same position as authentic ribositol5-phosphate. Treatment of the reduction product with alkaline phosphatase or acid hydrolysis (2 N HC1 at 100 "C for 3 h) followed by electrophoresis resulted in the radioscans shown in Fig. 4, B and C, respectively. This position corresponded to that of ribositol and arabinitol which were not separated by this technique. Paper chromatography of the alkaline phosphatase-treated material yielded a species co-chromatographing with the ribositol standard (Fig. 5 ) .
The position of the phosphate moiety was confirmed by enzymatic digestion. Bacterial alkaline phosphatase quantitatively released phosphate from the reduced sugar phosphate and from a ribositol 5-phosphate standard. Incubation with 5'-nucleotidase released phosphate from both at the same rate, while 3'-nucleotidase was ineffective (Table I).
The molar ratio of phosphate, ribose, and glutamic acid moieties was assessed by amino acid analysis, spectrophotometric assay using a ribose standard and x-ray fluorescence. Table I1 indicates that a 1:1:1 molar ratio for glutamic acidrib0se:phosphate was found.  * Spectrophotometric by phenol-H,S04.

DISCUSSION
These studies clearly indicate that the compound isolated from the tissues of the patient contains covalent linkage of glutamic acid to ribose phosphate. The phosphate group is in the 5 position as determined by enzymatic digestion. The reactivity of the intact compound with alkaline silver nitrate only after treatment with periodic acid indicates that the phosphate cannot be in the 1, 2, or 3 position of ribose when one considers that the I position is labeled with NaB3H, without loss of phosphate and periodic acid reactivity requires vicinal hydroxyls. The intact compound is ninhydrin reactive which excludes ribose linkage to the a-carboxyl or a-amino group (15). The extreme instability of this compound to base and the formation of pyroglutamic acid upon mild acid hydrolysis suggest an ester linkage of ribose, a good leaving group, via the reducing position to the y-carboxyl of glutamic acid. It is unlikely that ribose is linked to glutamine, as Nlinked glycosides are stable to mild alkaline NaBH4 reduction. Thus, the suggested structure as shown in Fig. 6 is glutamyl ribose 5-phosphate. The anomeric linkage has yet to be determined. In 1963, Chambon et al. (16) found that proteins could be covalently modified by the incorporation ofpoly(ADP-ribose). chondrial, and cytoplasmic subcellular fractions (17). Mono(ADP-ribosylation) is primarily a cytoplasmic event and poly(ADP-ribosylation) a nuclear activity (18). The individual synthetases responsible for these activities have been highly purified and partially characterized (19,20). The structure of the poly(ADP-ribose) modification has been established as a repeating unit of adenosine diphosphate ribose in which branching may occur via 1'-3' glycosidic linkages uersm the usual 1'-2' glycosidic bond (21). The acceptor moiety for ADP-ribose has received much attention due to its unique structure and possible relevance to physiologic function. Studies of mono(ADP-ribosylation) in turkey erythrocytes have indicated an arginine acceptor, although detailed structural studies have not been reported (20). Poly(ADP-ribosylation) of mammalian nuclear proteins (predominantly histones), via a base sensitive ester linkage to glutamic acid, has been suggested by Riquelme et al. (22) and Burzio et al. (23). Recently, Ogata et al. has elegantly demonstrated ribose linked to glutamic acid through an ester bond, probably by means of the C-1 position of ribose and the y-carboxyl of glutamic acid in histones H1 and H2B (6). In HI, the COOHterminal lysine residue was also labeled via the carboxyl group (7). These represent the only examples known in mammalian biochemistry of a sugar amino acid ester linkage. The structural identity of our isolated compound to that found in poly(ADP-ribosylated) histones suggests that it is derived from poly(ADP-ribosylated) proteins. Our demonstration of the structure of glutamyl ribose 5-phosphate by direct physical analysis supports the earlier work on the structure of the histone linkage.
The in uiuo degradation of poly(ADP-ribose) is primarily due to poly(ADP-ribose) glycohydrolase which cleaves the 1'-2' ribose-ribose glycosidic bond (24). Endo-and exophosphodiesterases are also known which act upon the ADP-ribose homopolymer (25,26). The result of the combined action of these enzymes and proteases would, in the absence of ADPribose protein hydrolase, be the accumulation of glutamyl ribose 5-phosphate as found in our patient (Fig. 6).
Recently, Okayama et al. (27) have partially purified an enzyme which cleaves mono(ADP-ribose) from its histone acceptor. The enzyme was clearly separated from poly(ADPribose) glycohydrolase and was much less active against the poly(ADP-ribose) histone. The product of this ADP-ribosyl histone hydrolase was a mixture of ADP-ribose and a compound similar to it (28). The authors speculated that the ribose was linked to glutamic acid via an unknown, nearly neutral compound. This enzyme had a pH optimum of 6 and on crude subcellular fractionation was found distributed in nuclear, mitochondrial, and cytosolic compartments. Whether or not a separate ADP-ribose protein hydrolase for intact poly(ADP-ribose) substrates exists or prior cleavage af polymeric substrates to mono(ADP) derivatives is required, has not been determined. We propose that the hereditary defect in this patient is a deficiency of a lysosomal ADP-ribose protein hydrolase, although an abnormality in a phosphatase specific for ribose 5-phosphate cannot be excluded.
The putative roles of poly(ADP-ribosylation) in the regulation of enzyme function, DNA repair, and gene expression may be dependent on the sequential extension and hydrolysis of ADP-ribose units from the glutamyl ribose 5-phosphate core. Thus, the inability to degrade the core linkage group may have little effect on these processes. However, cells from this patient would provide a valuable tool for studying at what point in time a given protein is ADP-ribosylated and a means of measuring the rate of post-translational modification of that protein. The block in degradation would also allow the identification of proteins whose poly(ADP-ribosylation) may only be a transitory event. The primary clinical expression of this storage disease in brain and kidney cannot be explained at this time. Further studies to discern the enzymatic deficiency in the tissues and cultured fibroblasts of this patient will contribute to a better understanding of the metabolism of poly(ADP-ribosylation) and the degradation of histones in various tissues.