A Novel Mammalian Lithium-sensitive Enzyme with a Dual Enzymatic Activity, 3′-Phosphoadenosine 5′-Phosphate Phosphatase and Inositol-polyphosphate 1-Phosphatase*

We report the molecular cloning in Rattus norvegicus of a novel mammalian enzyme (RnPIP), which shows both 3′-phosphoadenosine 5′-phosphate (PAP) phosphatase and inositol-polyphosphate 1-phosphatase activities. This enzyme is the first PAP phosphatase characterized at the molecular level in mammals, and it represents the first member of a novel family of dual specificity enzymes. The phosphatase activity is strictly dependent on Mg2+, and it is inhibited by Ca2+ and Li+ ions. Lithium chloride inhibits the hydrolysis of both PAP and inositol-1,4-bisphosphate at submillimolar concentration; therefore, it is possible that the inhibition of the human homologue of RnPIP by lithium ions is related to the pharmacological action of lithium. We propose that the PAP phosphatase activity of RnPIP is crucial for the function of enzymes sensitive to inhibition by PAP, such as sulfotransferase and RNA processing enzymes. Finally, an unexpected connection between PAP and inositol-1,4-bisphosphate metabolism emerges from this work.

The sulfation reactions in mammals affect many different physiological processes, including deactivation and bioactivation of xenobiotics, inactivation of hormones and catecholamines, structure and function of macromolecules, and elimination of end products of catabolism (1). Sulfation involves the transfer of a sulfate group from 3Ј-phosphoadenosine 5Јphosphosulfate (PAPS) 1 to an acceptor molecule in a reaction that is catalyzed by a family of sulfotransferase enzymes (2). Usually, the obligate co-substrate PAPS donates its sulfuryl group to a functional hydroxyl group: PAPS ϩ R-OH 3 PAP ϩ R-OSO3 Ϫ . The products are a sulfoconjugate (R-OSO3 Ϫ ) and 3Ј-phosphoadenosine 5Ј-phosphate (PAP). PAP is an inhibitory end product in any sulfation reaction because it acts as a competitive inhibitor with respect to PAPS (3). For instance, the M and P forms of human phenolsulfotransferases are po-tently inhibited by PAP, exhibiting a K i value of approximately 0.1 M (4).
In addition, a toxic effect of PAP on RNA processing enzymes (5Ј 3 3Ј exoribonucleases) has been recently described in yeast (5). This effect might be attributed to the fact that PAP mimics the monomers of a polyribonucleotide chain and consequently, it might prevent the attack to the phosphodiester bond of RNA processing enzymes. These 5Ј 3 3Ј exoribonucleases are highly conserved in evolution (5), therefore it is very likely that PAP accumulation in mammals has similar inhibitory effects on RNA processing, leading to aberrant changes in the pattern of gene expression. Consequently, a rapid removal of PAP is required both to maintain the activity of sulfotransferase enzymes and to prevent the accumulation of PAP to levels that could be inhibitory for RNA processing enzymes. Thus, it is reasonable to predict the existence in mammals of hydrolytic enzymes, such as PAP phosphatases, which could rapidly recycle PAP to AMP and inorganic phosphate. Indeed, a preliminary characterization of a 3Ј,5Ј-bisphosphate nucleotidase purified from guinea pig liver has been reported (6), although this enzyme showed a very poor affinity for PAP (K m 3 mM). No clear role for this enzymatic activity nor molecular data of the corresponding gene have been reported.
In this work, we present for the first time the molecular cloning of a mammalian PAP phosphatase. The enzyme was cloned by functional complementation of a Saccharomyces cerevisiae hal2 mutant. Hal2 is a 3Ј,5Ј-bisphosphate nucleotidase that specifically hydrolyzes the 3Ј-phosphate from PAP (7,8), thereby preventing the accumulation of PAP. In yeast, PAP is generated as a side product of PAPS reductase, which is a key enzyme in the sulfate assimilation pathway that leads to the synthesis of methionine. PAP accumulation has deleterious effects on the cell because it inhibits PAPS reductase (9), and hence methionine biosynthesis, and RNA processing enzymes (5). The PAP phosphatase activity of Hal2 is crucial for the function of PAPS reductase, which is reflected in the fact that hal2 cells are auxotrophic for methionine (7,10). We took advantage of this phenotype to clone the first mammalian PAP phosphatase by functional complementation of the auxotrophy for methionine of hal2 cells. Interestingly, the mammalian PAP phosphatase has a dual enzymatic activity, as it is also active as inositol-polyphosphate 1-phosphatase. Yeast Strain and Growth Media-The Saccharomyces cerevisiae strain used in this study was JRM4 (MATa leu2-3, 112 ura3-251,328,372 hal2::LEU2), kindly provided by Dr. J. R. Murguia (Universidad Politécnica, Valencia, Spain). Complementation of the auxotrophy for methionine of the hal2 mutant was assayed in minimal synthetic glucose medium (SD), 2% glucose, 0.7% yeast nitrogen base without amino acids and 50 mM succinic acid adjusted to pH 5 with Tris. When indicated, the SD medium was supplemented with 100 g/ml methionine to give SDM medium.

Chemicals-All
Cloning of RnPIP-The yeast strain JRM4 was transformed with a Rattus norvegicus heart cDNA library constructed in the pFL61 vector (11), a yeast expression vector where expression of the rat cDNA is driven by the constitutive phosphoglycerate kinase (PGK) promoter. Transformants were first selected by uracil prototrophy in SDM. 10 5 primary transformants were obtained and pooled together. To select clones with the ability to complement the auxotrophy for methionine of the hal2 mutant, the primary transformants were spread on minimal medium lacking methionine. Four clones (pFL61-PIP) were obtained that complemented the auxotrophy for methionine of the hal2 mutant. The yeast shuttling plasmid was recovered by electroporation of Escherichia coli WM1100 cells. Partial sequencing and restriction analysis of the plasmids recovered from the E. coli transformants showed that all of them represented the same cDNA.
The PCR-amplified product was subcloned into the pT7-Blue vector (Novagen) to give pT7-PIP. The PIP open reading frame was released with XhoI and subcloned into the XhoI site of pGEX-KG (12), thus generating a fusion protein between glutathione S-transferase (GST) and PIP. The pGEX-KG-PIP construct was verified by sequencing with an ABI 377 automatic sequencer, and it was introduced into E. coli DH5␣ cells. The expression of the recombinant protein was induced with 0.1 mM isopropyl-1-thio-b-D-galactopyranoside for a period of 1 h at 30°C, and the GST fusion protein was affinity purified on glutathione-Sepharose 4B (Amersham Pharmacia Biotech) as indicated by the suppliers. The pGEX-KG-PIP construct contains a thrombin cleavage site and a poly-glycine spacer between the GST and the RnPIP moieties (12). Thus, specific proteolysis of the fusion protein could be achieved by thrombin cleavage. The resulting RnPIP protein behaves identically to the GST fusion protein in the biochemical assays we have performed.
Enzyme Assay-PAP phosphatase and inositol-polyphosphate 1phosphatase activities were assayed by one of the two following methods (8). Briefly, phosphatase activity was routinely assayed (assay 1) in a 100-l mixture containing 50 mM Tris-HCl, pH 7.5, 1 mM magnesium chloride, and the indicated amount of GST fusion protein and substrate. After a 30-min incubation at 30°C, the inorganic phosphate released was quantified by the malachite green method as described previously (13). Under these conditions, the enzyme activity was linear with protein quantity (up to 5 g) and reaction time (up to 1 h). The concentration of either Ca 2ϩ or Li ϩ that decreased the activity of the enzyme by 50% relative to a reaction without these cations (IC 50 ) was estimated at a substrate concentration of 0.2 mM. Assay 2 is described next, and it involved a HPLC analysis of the reaction products. The products were detected either by ultravioletabsorption (8) or, in the case of radioactively labeled substrates, by coupling the HPLC device to a RadioFlow detector LB509 (EGG Berthold).

HPLC Analysis of the Hydrolysis of PAP, [ 35 S]PAPS, and [ 3 H]Inositol-1,4-bisphosphate-
The conditions for the HPLC analysis of the hydrolysis of PAP and PAPS have been described previously (8). To improve the resolution of the analysis of PAPS hydrolysis, the mobile phase was 3% methanol instead of 4%. Inositol-1,4-bisphosphate hydrolysis was assayed in a reaction mixture containing 0.2 M potassium Bicine (pH 8.0), 1 mM Mg 2ϩ , and 10 M [ 3 H]inositol-1,4-bisphosphate (10 Ci/mmol). At the indicated time, 10 l of reaction mixture were injected into a 4.6 ϫ 10-mm Guard Cartridge 10-m SAX column linked to a 4.6 ϫ 250-mm Partisil 10-m SAX column (pS Phase Sep), equilibrated in Milli-Q water, and maintained at 22°C with a flow rate of 1 ml/min. Inositol phosphates were eluted with a linear gradient of (NH 4 )H 2 PO 4 , pH 3.7, according to Jenkinson (14

RESULTS
Cloning of RnPIP-To explore the presence in animals of PAP hydrolytic enzymes, a screening was conducted for mammalian genes that complement the auxotrophy for methionine of the yeast hal2 mutant. A rat cDNA library was constitutively expressed in hal2 cells, and as a result, a cDNA was identified that complemented the auxotrophy for methionine. Sequencing of this cDNA reveals a 927-base pair open reading frame that putatively encodes a protein of 308-amino acid residues with a molecular mass of 33.2 kDa (Fig. 1). The cDNA has 27 base pairs of 5Ј-untranslated sequence and 961 base pairs of 3Јuntranslated region. The protein encoded by this gene, named RnPIP, complements the auxotrophy for methionine of hal2 cells to the same extent as Hal2, as demonstrated by the similar growth rates in methionine-free minimal medium of hal2 cells transformed with RnPIP or ScHAL2 (data not shown).
Southern and Northern Blot Analyses-A commercially available genomic DNA blot was used to analyze the presence of RnPIP in the rat genome as well as in other eukaryotic species ( Fig. 2A). A major EcoRI hybridization band of approximately 4.0 kilobase pairs was detected in the rat genome. Other bands of lower intensity were also detected, suggesting that additional sequences related to RnPIP are present in the asterisk. An in frame TGA stop codon is found 12 nucleotides 5Ј to the first ATG translation initiation codon. The three conserved regions (17) involved in the coordination of phosphate and metal ions and nucleophilic water activation are indicated in boldface and are underlined. The nucleotide sequence of the RnPIP gene has been deposited at the GenBank TM /EBI data base under the accession number AJ000347. rat genome. Hybridization bands were also detected in the genome of mouse, dog, cow, rabbit, and chicken. Data base searches in the human nucleotide data base reveal the presence of human genes with high sequence similarity to RnPIP (see below). A rat multiple tissue Northern blot containing poly(A) ϩ RNA was used to examine the transcript level of the RnPIP gene in different tissues (Fig. 2B). The RnPIP mRNA was detected in all the tissues examined. The transcript level was especially high in heart, brain, and kidney. The size of the mRNA fits well with the length of the cDNA sequence reported above.
Biochemical Characterization of the RnPIP Protein-The Rn-PIP protein was expressed in E. coli as a GST fusion protein.
Following induction with isopropyl-1-thio-b-D-galactopyranoside, the fusion protein was purified by affinity chromatography on glutathione-Sepharose resin. The phosphatase activity of the purified protein with respect to different substrates is summarized in Table I. PAP was the preferred substrate for RnPIP. The K m value for PAP hydrolysis was too low to be determined with the standard colorimetric assay (13) or even using the HPLC-based method where the products are detected by ultraviolet absorption (8). The detection limit of the HPLC analysis for PAP hydrolysis is approximately 2 M, therefore the K m for PAP must be below 2 M. PAPS was used as substrate with a 50% efficiency as compared with PAP. No appreciable activity was observed against AMP, 3Ј-AMP, fructose-1,6-bisphosphate, or pNPP. Interestingly, inositol-1,4bisphosphate also served as a substrate for the enzyme, although with a 45% efficiency as compared with PAP. In addition, RnPIP hydrolyzed inositol-1,3,4-trisphosphate but not inositol-1,4,5-trisphosphate or inositol monophosphate. Therefore, besides the preferred PAP, RnPIP hydrolyzes the typical substrates of inositol-polyphosphate 1-phosphatase (15).
Determination of the K m Values for PAPS and Inositol-1,4-bisphosphate Hydrolysis-The sensitivity limit of the colorimetric assay for determination of inorganic phosphate prevents the determination of K m values in the low micromolar range. Fortunately, the above described HPLC analysis, together with the availability of [ 3 H]inositol-1,4-bisphosphate and [ 35 S]PAPS, made it possible to determine the K m values for these substrates (Table II). The apparent K m for inositol-1,4bisphosphate was 0.2 M Ϯ 0.1. The K m for PAP could not be accurately determined, as radioactively labeled PAP is not available. [ 35 S]PAPS is commercially available and RnPIP hydrolyzes the 3Ј-phosphate of PAPS with a 50% efficiency as compared with PAP, generating [ 35 S]adenosine 5Ј-phosphosulfate (Table II). Therefore, although PAPS is not likely a physiological substrate for animal PAP phosphatases (16), we have also determined the K m value for PAPS. The apparent K m for [ 35 S]PAPS was 1.2 M Ϯ 0.3.
Cation Sensitivity of RnPIP-Hal2 and inositol-polyphosphate 1-phosphatase belong to a family of magnesium-dependent, lithium-sensitive phosphatases (17). Accordingly, the phosphatase activity of RnPIP was strictly Mg 2ϩ -dependent, with an optimal concentration of 1 mM (data not shown). Ca 2ϩ inhibited the phosphatase activity of RnPIP (Fig. 4A) (20 M CaCl 2 for 50% inhibition at 0.2 mM PAP), probably because of competition for Mg 2ϩ , as the inhibition was abolished by increasing the Mg 2ϩ concentration (data not shown). RnPIP was not affected by high Na ϩ concentrations, whereas it was very sensitive to Li ϩ (0.8 mM LiCl for 50% inhibition at 0.2 mM PAP) (Fig. 4B). The effect of Li ϩ on this enzyme is unique among the group of monovalent cations (data not shown).

DISCUSSION
Biochemical characterization of the RnPIP protein shows that it is a dual enzyme that hydrolyzes both PAP and inositol-1,4-bisphosphate. The V max value for PAP hydrolysis was approximately 2-fold higher than for inositol-1,4-bisphosphate hydrolysis. Although the K m for PAP could not be accurately determined (Ͻ2 M), we could determine the K m for the closely related substrate PAPS (1.2 M) and for inositol-1,4-bisphosphate (0.2 M), indicating a high affinity for these substrates. The enzyme is very sensitive to Li ϩ and, therefore, RnPIP is  the fourth animal enzyme that is inhibited by Li ϩ ions in the therapeutic range, the other three being inositol monophosphate phosphatase (18), inositol-polyphosphate 1-phosphatase (15), and fructose-1,6-bisphosphatase (19). RnPIP is the first PAP phosphatase characterized at the molecular level in mammals, and a human homologue of RnPIP can be identified in the human data base of expressed sequence tags (ESTs) (Fig. 5).
We predict that the human homologue of RnPIP will show a similar sensitivity to Li ϩ ions. Consequently, it might represent another target in lithium therapy, and the inhibition by lithium of the human enzyme could also contribute to the therapeutic or toxic effects of Li ϩ treatment. In addition to PAP phosphatase, RnPIP also qualifies as inositol-polyphosphate 1-phosphatase. A precedent for this sort of dual enzymatic activity is found in plants (20). The SAL1 enzyme of Arabidopsis thaliana exhibits both PAP phosphatase and inositol-polyphosphate 1-phosphatase activities. However, the hydrolysis of inositol-1,4-bisphosphate proceeds with a 34% efficiency as compared with PAP. Furthermore, the K m value for inositol-1,4-bisphosphate was much higher (90 M) than the K m value for PAP (2 to 10 M). In contrast, RnPIP hydrolyzes both PAP and inositol-1,4-bisphosphate with very high affinities. However, some features clearly distinguish the RnPIP enzyme from the canonical inositol-polyphosphate 1-phosphatase (15). First, RnPIP hydrolyzes both PAP and inositol-1,4-bisphosphate, whereas PAP hydrolysis by inositolpolyphosphate 1-phosphatase has not been reported. Second, bovine inositol-polyphosphate 1-phosphatase fails to complement the yeast hal2 mutant (data not shown). Finally, amino acid sequence comparison also reflects a clear difference between RnPIP and inositol-polyphosphate 1-phosphatase (Fig.  6). RnPIP is only 27% identical to bovine inositol-polyphosphate 1-phosphatase, whereas the mouse inositol-polyphosphate 1-phosphatase is 85% identical to bovine inositolpolyphosphate 1-phosphatase (Fig. 6). On the other hand, the sequence identity of RnPIP to non-animal PAP phosphatases is also low (25-30%) (Fig. 6). Instead, ESTs of proteins with unidentified function, which show high identity (60 -90%) to RnPIP, are present in mouse, pig, fruit fly, and man (Fig. 6). For instance, comparison of human ESTs corresponding either to the N or the C terminus of RnPIP reveals 85-90% amino acid sequence identity to the respective part of RnPIP (Fig. 5). These data, taken together with the biochemical characterization of the protein, indicate that RnPIP is a novel mammalian enzyme, which probably represents the first member of an animal gene family of dual enzymes (Fig. 6).
In summary, we propose that the PAP phosphatase activity of RnPIP, by preventing accumulation of PAP, is necessary to maintain active sulfotransferase (6,21) and RNA processing enzymes (5). RNA processing is crucial in cell metabolism. The importance of sulfation in animals is illustrated by the phenotype of brachymorphic mice, which have abnormal hepatic detoxification, bleeding times, and postnatal growth (22). This phenotype is because of a missense mutation in SK2 (sulfurylase kinase 2) (22), a member of the gene family encoding the bifunctional enzymes that synthesize the universal sulfate donor, PAPS. In addition, the inositol-polyphosphate 1-phosphatase activity of RnPIP could play a role in the phosphoinositidesignaling pathway. Finally, the unexpected connection between PAP and inositol-1,4-bisphosphate metabolism remains to be investigated.