Cyanobacterial PPP Family Protein Phosphatases Possess Multifunctional Capabilities and Are Resistant to Microcystin-LR*

The structural gene for a putative PPP family protein-serine/threonine phosphatase from the microcystin-producing cyanobacterium Microcystis aeruginosa PCC 7820,pp1-cyano1, was cloned. The sequence of the predicted gene product, PP1-cyano1, was 98% identical to that of the predicted product of an open reading frame, pp1-cyano2, from a cyanobacterium that does not produce microcystins, M. aeruginosa UTEX 2063. By contrast, PP1-cyano1 displayed less than 20% identity with other PPP family protein phosphatases from eukaryotic, archaeal, or other bacterial organisms. PP1-cyano1 and PP1-cyano2 were expressed in Escherichia coli and purified to homogeneity. Both enzymes exhibited divalent metal dependent phosphohydrolase activity in vitro toward phosphoserine- and phosphotyrosine-containing proteins and 3-phosphohistidine- and phospholysine-containing amino acid homopolymers. This multifunctional potential also was apparent in samples of PP1-cyano1 and PP1-cyano2 isolated from M. aeruginosa. Catalytic activity was insensitive to okadaic acid or the cyanobacterially produced cyclic heptapeptide, microcystin-LR, both potent inhibitors of mammalian PP1 and PP2A. PP1-cyano1 and PP1-cyano2 displayed diadenosine tetraphosphatase activity in vitro. Diadenosine tetraphosphatases share conserved sequence features with PPP family protein phosphatases. The diadenosine tetraphosphatase activity of PP1-cyano1 and PP1-cyano2 confirms that these enzymes share a common catalytic mechanism.

Strains of the cyanobacterium Microcystis aeruginosa sp. synthesize the cyclic heptapeptide microcystin-LR, a potent toxin toward humans and other animals (1,2). In eukaryotes, the immediate targets of microcystin-LR are the structurally homologous catalytic subunits of the protein-serine/threonine phosphatases PP2A and PP1 1 (3), to which it binds with nano-molar or near nanomolar affinity, respectively (4). Sensitivity to these and other toxins, such as okadaic acid, is so highly conserved among eukaryotic PP1 and PP2A that it serves as a criterion for the identification of these enzymes in cell extracts (5). Moderate sensitivity to microcystin-LR extends to homologs recently identified in members of the Archaea such as Methanosarcina thermophila TM-1 (6,7) and Pyrodictium abyssi TAG11 (8).
Recently, the presence of open reading frames potentially encoding PP1/2A-like, also known as the PPP family (9), protein phosphatases has been reported in two strains of M. aeruginosa, the microcystin-producing strain M. aeruginosa PCC 7820 and the non-producing strain M. aeruginosa UTEX 2063 (10). These open reading frames have been designated pp1-cyano1 and pp1-cyano2, respectively. Given the observation that microcystins accumulate within the interior of toxin-producing cyanobacteria (11), and the near absolute conservation of sensitivity to these compounds by members of the PP1/2A superfamily (5), how do microcystin-producing cyanobacteria protect themselves from the action of endogenous toxins against their presumptive PP1/ 2A-like protein phosphatases? It has been suggested that cyanobacteria produce these secondary metabolites to help ward off encroachments by other microorganisms upon their habitat, thus gaining a competitive advantage (12). Do the presumptive PP1/ 2A-like protein phosphatases in strains of cyanobacteria that do not synthesize microcystins differ in sensitivity from the equivalent enzymes in toxin-producing strains? In order to answer these questions we (a) obtained a complete clone of pp1-cyano1 and determined the DNA-derived amino acid sequence of its predicted protein product, PP1-cyano1; (b) expressed and purified PP1-cyano1 and PP1-cyano2 in order to study certain of their properties, including their sensitivity to toxins; and (c) compared the properties of the recombinant enzymes with those of PP1-cyano1 and PP1-cyano2 partially purified from their natural sources, M. aeruginosa PCC 7820 and M. aeruginosa UTEX 2063.

EXPERIMENTAL PROCEDURES
Growth of Cyanobacteria-M. aeruginosa PCC 7820 and M. aeruginosa UTEX 2063 were grown with continuous aeration and lighting in BG-11 medium (13). When the late exponential phase of growth was reached, the cells were harvested by centrifugation and washed with Buffer A and stored at Ϫ20°C.
Standard Procedures-Protein concentrations were measured according to the method of Bradford (14) using premixed reagent and a standardized solution of bovine serum albumin from Pierce (Rockford, IL). SDS-polyacrylamide gel electrophoresis was performed as described by Laemmli (15). To visualize polypeptides, gels were stained with Coomassie Brilliant Blue as described by Fairbanks et al. (16).
Preparation of Substrates for Assay of Phosphohydrolase Activity-[ 32 P]Phosphoproteins were prepared as described previously (17,18). Partially phosphorylated poly-L-histidine and poly-L-lysine were prepared as described by Wong et al. (19). Low molecular weight phosphomonoesters such as p-nitrophenyl phosphate and diadenosine tetraphosphate (Ap 4 A) were purchased from Sigma.
Assay of Protein Phosphatase Activity-Phosphohydrolase activity toward 32 P-labeled phosphoproteins was determined by a variation of the procedure described in Kennelly et al. (17). Samples of PP1-cyano1 or PP1-cyano2 were incubated at 37°C in a final volume of 30 l of Buffer D containing 2 M substrate-bound [ 32 P]phosphate. The reaction was terminated, typically after a period of 15-90 min, by the addition of 100 l of 20% (w/v) trichloroacetic acid, mixed, centrifuged for 3 min at 12,000 ϫ g in a microcentrifuge, and 50 l of the supernatant liquid removed and counted for 32 P radioactivity in 1 ml of Scintisafe Plus 50% liquid scintillation fluid (Fisher Scientific, Pittsburgh, PA). Activity toward [ 32 P]phosphorylase a was determined according to the procedure provided by the manufacturer (Life Technologies Inc., Gaithersburg, MD). Phosphatase activity toward partially phosphorylated homopolymers of L-histidine and L-lysine were performed essentially as described by Wong et al. (19) with the assay buffer altered to Buffer D and the incubation temperature increased to 37°C.
Assay of Phosphohydrolase Activity toward Low Molecular Weight Phosphomonoesters and Ap 4 A-Phosphomonoesterase activity toward p-nitrophenyl phosphate and other low molecular weight phosphomonoesters was assayed essentially as described by Howell et al. (18). However, the assay buffer was altered to Buffer D and the incubation temperature was increased to 37°C. Pyrophosphatase activity toward Ap 4 A was measured by the method of Plateau et al. (20), in which the adenine mononucleotides formed by the pyrophosphorolysis of Ap 4 A are hydrolyzed to adenine and organic phosphate using high levels of alkaline phosphatase. Briefly, PP1-cyano1 or PP1-cyano2 was incubated for varying times at a temperature of 37°C in a volume of 50 l of Buffer D containing 1 mM Ap 4 A and 3 units of alkaline phosphatase (U. S. Biochemical Corp., Cleveland, OH). Reaction was terminated by the addition of 150 l of DE52 cellulose equilibrated in 50 mM imidazole, pH 8.0. After mixing and centrifugation in a microcentrifuge at 12,000 ϫ g for 3 min, 50 l of the supernatant liquid was assayed for inorganic phosphate using malachite green (21).
The site cleaved in Ap 4 A was determined by identifying the adenine nucleotides formed following pyrophosphorolysis by the method of Mangold (22). Briefly, samples of PP1-cyano1 or PP1-cyano2 were incubated for varying times at a temperature of 37°C in 50 l of Buffer D containing 1 mM Ap 4 A. Reaction was terminated by adding 50 l of 10 mM EDTA. A 10-l aliquot of the quenched assay mixture was then analyzed by thin layer chromatography on a plate of LK2F microcrystalline cellulose (Whatman, Clifton, NJ) that was developed with a mixture of 35:25:15:15:10 1-butanol, acetone, acetic acid, 5% (v/v) ammonium hydroxide, water. Samples of Ap 4 A, ATP, ADP, and AMP were run in parallel as standards. Adenine nucleotides were visualized by virtue of their fluorescence when illuminated by UV light. Symmetric diadenosine tetraphosphatases produce ADP as their sole hydrolysis product, while asymmetric forms produce AMP and ATP.
Assay of Protein Phosphatase Activity following SDS-PAGE-Measurement of protein phosphatase activity following SDS-PAGE was performed essentially as described by Burridge et al. (23) using [ 32 P]casein as substrate.
Partial Purification of PP1-cyano1 and PP1-cyano2 from M. aeruginosa-The procedures for purifying PP1-cyano1 from M. aeruginosa PCC 7820 and PP1-cyano2 from M. aeruginosa UTEX 2063 were identical. Fifty grams of frozen cells were thawed and suspended in 5 volumes of Buffer A containing 1 g/ml DNase I, ruptured by sonication using six pulses, each of 1-min duration, of a Sonifier model 185 sonic disrupter fitted with a large probe, then centrifuged at 12,000 ϫ g for 20 min at 4°C. The supernatant liquid was collected and passed through a column of CM-Trisacryl. The flow-through was collected and loaded onto a 6.5 ϫ 40-cm column of DE52 cellulose equilibrated in Buffer B. After washing with Buffer B, the column was eluted with Buffer B containing 400 mM NaCl. The column eluate, DE-52 fraction I, was dialyzed overnight against Buffer B, then applied to a 3 ϫ 60-cm column of DE52 cellulose equilibrated in Buffer B. The column was washed with Buffer B, then developed with a linear gradient consisting of 400 ml each of Buffer B and Buffer B containing 400 mM NaCl. Fractions, 10 ml each, were collected and assayed for protein concentration and protein phosphatase activity. Active fractions were pooled, reduced in volume to 20 ml using a Centriprep-10 centrifugal concentrator (Amicon, Beverly, MA), and applied, in 10-ml portions, to a 5 ϫ 100 cm column of Sephacryl S-200 that had been equilibrated in Buffer C. The column was developed with Buffer C and fractions, 5 ml each, collected and assayed for protein phosphatase activity. Fractions exhibiting protein phosphatase activity were pooled, concentrated using a Centriprep-10 centrifugal concentrator, and stored at Ϫ20°C.
Cloning and Sequencing of pp1-cyano1-Genomic DNA, 2 g, isolated from M. aeruginosa PCC 7820 as described previously (10), was digested with HindIII according to the manufacturer's protocols (Life Technologies Inc.). Restriction fragments were separated on a 0.8% (w/v) agarose gel, then transferred to a Magna nylon membrane (MSI, Boston, MA). The membrane was probed with a partial clone of PP1-cyano1 (10) that had been radiolabeled by PCR amplification from M. aeruginosa PCC 7820 genomic DNA in the presence [␣-32 P]dATP (24). A single species roughly 2.7 kilobase pairs in length hybridized with the probe. Starting with 20 g of DNA, HindIII restriction fragments corresponding in size to that identified above were extracted from an agarose gel, and ligated into plasmid vector pZERO (Invitrogen, Portland, OR) that had been cut previously with HindIII. Transformation of Escherichia coli with the ligation mixture and identification of positive clones by colony lifts followed standard procedures (25). DNA inserts from positive clones were digested with restriction enzyme XbaI following the manufacturer's protocols (Life Technologies Inc.), subcloned, and positive subclones completely sequenced on both strands using Sequenase version 2.0 (U. S. Biochemical Corp.) and oligonucleotide primers (Life Technologies Inc.).
Expression of Recombinant PP1-cyano1 and PP1-cyano2 in E. coli-The complete structural genes for PP1-cyano1 and PP1-cyano2 were amplified by PCR using Pfu DNA polymerase (Stratagene, La Jolla, CA). The primers were designed to introduce annealing sequences for BamHI and PstI restriction sites at each end. A common reverse primer was used for each: 5Ј-CCAGCTGCAGTATTAATCAGATTATCAACTA-3Ј. The forward primers were 5Ј-CGATGGATCCGTATGTTGTTTAGA-AAAATAG-3Ј for PP1-cyano1 and 5Ј-CGATGGATCCGTATGTTTTTT-AGAAACATAG-3Ј for PP1-cyano2. The resulting PCR products were ligated into the expression vector pRSET C (Invitrogen, Portland, OR), transformed into competent E. coli DH5␣ (Life Technologies Inc.), and plasmid isolated therefrom. The plasmid-encoded protein phosphatase genes were sequenced to verify the fidelity of PCR amplification, etc. Competent E. coli BL21(DE3) pLyS (Promega, Madison, WI) were transformed with the plasmids, grown until they reached an OD 600 of 0.6 -1.0, then expression of PP1-cyano1 or PP1-cyano2 induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.4 mM. Cells were grown in the presence of isopropyl-1-thio-␤-D-galactopyranoside overnight at a temperature of 30°C, harvested by centrifugation, then resuspended in 50 ml of 20 mM Tris-HCl containing 0.5 M NaCl, 1 mM imidazole, 1 mM phenylmethylsulfonyl fluoride, 2 mg/ml lysozyme, and 100 g/ml DNase I and placed on ice for 30 min. The cells were then lysed by sonic disruption and the resulting lysate clarified by centrifugation at 17,000 ϫ g for 30 min. The supernatant liquid was directly applied to a 1.5 ϫ 20-cm column of chelating Sepharose Fast Flow (Pharmacia-LKB, Uppsala, Sweden) that had been charged with ZnSO 4 and equilibrated in Buffer E. The column was extensively washed with Buffer E, and adhering proteins eluted with Buffer E in which the imidazole concentration had been increased to 250 mM. The high imidazole eluate was collected and dialyzed versus Buffer F. The dialyzed material was then concentrated by centrifugal ultrafiltration (Centriprep 10) to a volume of 1 ml and portions, 0.2 ml, applied to a 1 ϫ 60-cm column of Sephacryl S-200 that had been equilibrated in Buffer F. The column was developed with Buffer F. Fractions, 0.5 ml each, were collected and assayed for protein phosphatase activity. Active fractions were pooled and stored at Ϫ20°C until needed.
Nucleotide Sequence Accession Number-The nucleotide sequence for PP1-cyano1 has been submitted to GenBank, which has assigned it accession number U80886. The nucleotide sequence and accession number of PP1-cyano2 can be found in Shi and Carmichael (10).

Cloning of the Complete Gene for and DNA-derived Amino
Acid Sequence of PP1-cyano1-Using the Ϸ113-base pair PCR fragment originally isolated by Shi and Carmichael (10) as probe, Southern analysis indicated that only a single gene potentially encoding a PPP-like protein phosphatase was present in the genome of M. aeruginosa PCC 7820 (data not shown).
A full-length clone of the pp1-cyano1 gene was isolated from the genomic DNA of M. aeruginosa PCC 7820 and sequenced using conventional methods. The complete, DNA-derived amino acid sequence of the predicted protein product, PP1-cyano1, is shown in Fig. 1. PP1-cyano1 is predicted to consist of a polypeptide 264 amino acids in length with a molecular mass of 30,426 daltons and an isoelectric pH of 5.8. The nucleotide sequence of the presumptive coding region of pp1-cyano1 is 98% identical to that of pp1-cyano2 from M. aeruginosa UTEX 2063. The predicted protein products of the two open reading frames share 98% identity at the amino acid level.
In contrast to their very high degree of sequence identity to each other, the predicted gene products of pp1-cyano1 and pp1-cyano2 exhibit relatively low, typically 17-19%, identity to their next closest homologs among the PPP family of protein phosphatases ( Fig. 1), whose most prominent members are PP1 and PP2A (9). This was to be expected in those cases where the predicted products of these cyanobacterial genes were compared with their counterparts from eukaryotes or the Archaea. However, the lack of noticeably greater identity to the only other well characterized PPPs from the Bacteria, namely PrpA and PrpB from E. coli (26), or the bacteriophage encoded PP- (27) was somewhat surprising. We postulated that the lack of an obvious "family resemblance" to other bacterial or bacterially associated PPPs might reflect the extremely deep nature of the branching between the cyanobacteria and other members of the bacterial phylogenetic domain (28). A computer-generated phylogenetic tree of representative PPP-like phosphohydrolases and related enzymes (Fig. 2) appears to confirm this supposition. It grouped pp1-cyano1 and pp1-cyano2 together with the other bacterial phosphohydrolases, including the diadenosine tetraphosphatases (29), while the archaeal and eukaryotic PPPs each clustered together in a manner consistent with current three domain models for phylogeny (28). Significantly, within the bacterial cluster, PP1-cyano1 and PP1-cyano2 were grouped with the other known protein phosphatases from or associated with bacteria, PrpA and PrpB from E. coli (26) and PP-(30), on a branch separate from that containing the bacterial diadenosine tetraphosphatases.
Expression of PP1-cyano1 and PP1-cyano2 in E. coli and Purification to Homogeneity-The presumed structural genes for pp1-cyano1 and pp1-cyano2 were cloned into vector pR-SET-C for expression in E. coli and subsequent evaluation of the catalytic capabilities of the resulting polypeptide products. The pRSET-C vector introduced an amino-terminal His 6 sequence into the recombinant proteins, which facilitated their purification by metal-chelate affinity chromatography (31). Recombinant PP1-cyano1 and PP1-cyano2, which henceforth will be referred to as rPP1-cyano1 and rPP1-cyano2, respectively, were purified to electrophoretic homogeneity (Fig. 3) as described under "Experimental Procedures" by Zn 2ϩ -affinity chromatography followed by gel filtration chromatography on Sephacryl S-200. An "in gel" protein phosphatase assay using [ 32 P]phosphoseryl casein revealed that each preparation con- tained a single source of protein phosphatase activity whose apparent M r corresponded to that calculated for the recombinant gene product (data not shown). The position of the band of activity also corresponded with that of the single polypeptide species visible following staining with Coomassie Blue.
Properties of Recombinant PP1-cyano1 and PP1-cyano2-rPP1-cyano1 and rPP1-cyano2 displayed divalent metal ion-dependent protein phosphatase activity toward [ 32 P]phosphoseryl casein (Table I). Mn 2ϩ produced by far and away the greatest stimulation of protein phosphatase activity, while Mg 2ϩ , Co 2ϩ , and Ni 2ϩ proved weakly stimulatory. Cd 2ϩ , Cu 2ϩ , Fe 2ϩ , and Zn 2ϩ substantially inhibited protein phosphatase activity when added along with the activating metal Mn 2ϩ . Half-maximal activation of rPP1-cyano1 occurred at a Mn 2ϩ concentration of approximately 0.2 mM (data not shown). rPP1-cyano1 was active toward phosphoseryl casein over a pH range that spanned from 6.5 to 9.5 (Fig. 4). Within this range, optimal activity was exhibited from pH 7.5 to 9.0. The enzyme was essentially inactive at pH values less than or equal to 5 or greater than or equal to 10.
Next, the effects of various compounds known to inhibit protein phosphatases and other phosphomonoesterases, as well as metabolites known to exert allosteric effects in bacteria, on the catalytic activity of rPP1-cyano1 and rPP1-cyano2 were determined (Table II). The various metabolites, such as adenine nucleotides, cyclic AMP, sugar phosphates, glutamine, and ␣-ketoglutarate, were without effect. The same proved to be the case for tartrate, an inhibitor of many acid phosphatases, and tetramisole, the classic inhibitor of alkaline phosphatase. The general PPP family protein phosphatase inhibitors fluoride and pyrophosphate proved inhibitory toward PP1-cyano1 and PP1-cyano2 when present at millimolar concentrations, as did orthovanadate. The latter is a general phosphohydrolase inhibitor that displays particularly high potency toward protein-tyrosine phosphatases, which are sensitive to submillimolar concentrations of this phosphate mimetic.
Microcystin-LR, okadaic acid, and calyculin A, the potent inhibitors of homologous protein phosphatases from eukaryotes, PP1, and PP2A (4, 5), proved innocuous even when present at the micromolar concentrations known to inhibit the mildly sensitive eukaryotic PPP, calcineurin (PP-2B). Ap 4 A also proved somewhat inhibitory when present at millimolar concentrations. This compound is the substrate for a set of bacterial phosphohydrolases, the diadenosine tetraphosphatases (ApaH), that exhibit significant homology to the PPP family of protein phosphatases (3,29).
Given the homology between PPP family protein phosphatases and diadenosine tetraphosphatases (3,29,(32)(33)(34), Ap 4 A was tested as a potential substrate for rPP1-cyano1 and rPP1-cyano2. Both enzymes hydrolyzed Ap 4 A at rates comparable to macromolecular substrates. Once again, it is important to note that the concentration of Ap 4 A employed, 1 mM, far exceeded that used or even achievable with radiolabeled phosphoprotein substrates (2 M). Analysis of the reaction products by thin layer chromatography revealed that both cyanobacterial protein phosphatases cleaved Ap 4 A in a symmetric manner to yield two molecules of ADP (Table IV).
Properties of PP1-cyano1 and PP1-cyano2 Isolated from Native Organisms-Bacterial expression of the catalytic subunit of mammalian PP1 results in the production of a recombinant enzyme that exhibits a number of aberrant functional properties. These include a dependence on the presence of exogenous divalent metal ions for catalytic activity (35,36) as well as a marked elevation in activity toward phosphotyrosyl proteins (37). Restoration of the recombinant mammalian enzyme to its native conformation can be accomplished by incubation with its regulatory subunit, inhibitor-2, which apparently functions as a molecular chaperone (37,38). In this study, however, incubation with mammalian inhibitor-2 had no effect on the catalytic activity, substrate specificity, or other functional properties of rPP1-cyano1 or rPP1-cyano2 (Data not shown).
To definitively establish whether the multifunctional capabilities and other functional properties exhibited by rPP1-cyano1 and rPP1-cyano2 accurately reflected the properties of the native enzymes, we partially purified these enzymes from their natural sources, the cyanobacteria M. aeruginosa PCC 7820 and M. aeruginosa UTEX 2063, respectively. The major protein-serine/threonine phosphatase activities from both of these cyanobacteria were partially purified by a combination of ion-exchange and gel-filtration chromatography as described under "Experimental Procedures." Fig. 5 displays the results obtained with extracts from M. aeruginosa PCC 7820. Similar results, not shown, were obtained with extracts from M. aeruginosa UTEX 2063. In each case, a major peak of phosphohydrolase activity was observed that eluted at a position similar, allowing for the His 6 -containing NH 2 -terminal sequence introduced by the pRSET-C vector system, to that of rPP1-cyano1 and rPP1-cyano2. Analysis of the peak fractions using an in gel protein phosphatase assay employing [ 32 P]phosphoseryl casein as substrate revealed the presence of a single catalytic species in each with an M r matching that estimated for PP1-cyano1 or PP1-cyano2 (data not shown). On each column, the phospho-TABLE I Activation of cyanobacterial protein phosphatases by divalent metal ions The activity of purified recombinant rPP1-cyano1 and rPP1-cyano2, 10 ng of each, along with that of partially purified PP1-cyanol (50 ng) and PP1-cyano2 (75 ng) from M. aeruginosa PCC 7820 and M. aeruginosa UTEX 2063, respectively, was assayed under standard conditions using [ 32 P]phosphoseryl casein with the exception that, where indicated, the compounds listed were substituted for the activating divalent metal ion, Mn 2ϩ . All compounds were present at a final concentration of 3 mM. Iron was maintained in the ferric state by the addition of ascorbic acid to a final concentration of 3 mM. All results are reported as the percentage of activity relative to that observed with the most activator, Mn 2ϩ .  4. Influence of pH on the catalytic efficiency of rPP1-cyano1 and PP1-cyano1. rPP1-cyano1 (E), 10 ng, and PP1-cyano1 (q), 50 ng, were assayed for protein phosphatase activity toward [ 32 P]phosphoseryl casein under standard conditions with the exception that the buffer composition and pH were varied as indicated. Shown is the relative protein phosphatase activity detected as a function of pH. For rPP1-cyano1, 100% activity was equal to 0.2 pmol of 32 P i released per minute, while the corresponding value for PP1-cyano1 was 0.4 pmol/min. Buffer salts used, all at a final concentration of 50 mM included sodium acetate, pH 5.0; imidazole, pH 5.0 -8.0; Tris-HCl, pH 8.0 -9.0; and glycine, pH 9.0 -11.0. hydrolase activities toward phosphoseryl casein, phosphotyrosyl casein, 3-phospho-polyhistidine, phospho-polylysine, and Ap 4 A coeluted, indicating that the native forms of PP1-cyano1 and PP1-cyano2 possess multifunctional potential (Fig. 5).
The estimated molecular mass of PP1-cyano1, as determined by comparing its elution on a Sephacryl S-200 column with that of protein standards, was 29 kDa. This indicates that the native enzyme, at least in its catalytically active form, is a monomer. A detailed comparison of the divalent metal ion dependence (Table I), inhibitory spectrum (Table II), pH activity profile (Fig. 4), and substrate specificity (Table III, Fig. 5) of the partially purified enzymes from M. aeruginosa PCC 7820 or M. aeruginosa UTEX 2063 with that of their recombinant counterparts indicated that rPP1-cyano1 and rPP1-cyano2 accurately embodied the properties of the native enzymes. DISCUSSION With the determination of the complete gene sequences for a number of bacterial organisms, it has become apparent that open reading frames potentially encoding homologs of the major family protein-serine/threonine phosphatases in eukaryotes, the PPP family, are widespread throughout the Bacteria (39,40). At present, however, little is known concerning the physical or functional properties of these potential bacterial protein phosphatases. It was particularly intriguing to observe that the cyanobacteria contain open reading frames whose  PP1-cyano1 and PP1-cyano2 The activity of purified recombinant rPP1-cyano1 and rPP1-cyano2, 10 ng of each, along with that of partially purified PP1-cyanol (50 ng) and PP1-cyano2 (75 ng) from M. aeruginosa PCC 7820 and M. aeruginosa UTEX 2063, respectively, was assayed under standard conditions, using [ 32 P]phosphoseryl casein as substrate, with the exception that the compounds listed were present at the indicated final concentrations. All results are reported as the percent of activity measured in the absence of added compounds.

TABLE III
Relative activity of PP1-cyanol toward macromolecular and low molecular weight organophosphates The listed phosphoprotein, phosphopolypeptide, and low molecular weight organophosphate compounds were tested as substrates for purified, recombinant PP1-cyano1 (rPP1-cyano1) and PP1-cyano1 partially purified from M. aeruginosa PCC 7820 (PP1-cyano1). All assays were carried out as described under "Experimental Procedures." For protein and polymer substrates, the type of phosphoamino acid(s) present is given in parentheses. Phosphoprotein substrates were assayed at a final concentration of 2 M protein-bound [ 32 P]phosphate. Partially phosphorylated polylysine and polyhistidine, p-nitrophenyl phosphate, and Ap 4 A were assayed at a final concentration of 1 mM. All assays were performed in triplicate. The quantities of rPP1-cyano1 and PP1-cyano1 used for the assay of activity toward phosphorylated forms of casein, RCM-lysozyme, and myelin basic protein were 10 and 50 ng, respectively. For assay of activity toward glycogen phosphorylase a, polylysine, polyhistidine, and Ap 4 A the quantities were increased to 20 and 100 ng while for pNPP these levels were 40

TABLE IV Analysis of adenine nucleotides produced when Ap 4 A is incubated
with cyanobacterial PPPs Purified recombinant rPP1-cyano1 and rPP1-cyano2, 20 ng each, and partially purified PP1-cyano1, (100 ng) and PP1-cyano2 (150 ng) from M. aeruginosa PCC 7820 and M. aeruginosa UTEX 2063, respectively, were incubated with Ap 4 A and the hydrolysis products analyzed by thin layer chromatography as described under "Experimental Procedures." Listed below are the mobilities (R f ), relative to the solvent front, of both adenine di-and mononucleotide standards and the products formed following incubation with the listed phosphohydrolase. Standards were run in quadruplicate, and their mobilities are therefore reported plus or minus standard error. predicted products resembled the catalytic subunits of PP1 and PP2A, since these organisms produce peptide toxins such as microcystin-LR that potently inhibit these enzymes in eukaryotes (1,2). The characterization of PP1-cyano1 from the microcystin-LR producing cyanobacterium M. aeruginosa PCC 7820 and of PP1-cyano2 from the nonproducing strain M. aeruginosa UTEX 2063 represents an important step in the exploration of potential signal transduction enzymes in the Bacteria (41). The nucleotide sequences of pp1-cyano1 and pp1-cyano2, and the resulting DNA-derived amino acid sequences of their polypeptide products, were found to be 98% identical. Thus, it was not surprising that their functional properties were quite similar as well. Characterization of recombinant forms of the proteins produced in E. coli permitted us to work with homogeneous preparations whose origins could be unambiguously traced to a defined gene product. Parallel experiments using partially purified preparations of the enzymes from their natural sources, M. aeruginosa PCC 7820 and M. aeruginosa UTEX 2063, respectively, ensured that the behaviors displayed by the recombinant gene products faithfully reflected those exhibited by the native proteins. This represented a very significant potential pitfall, as early attempts to express mammalian PP1 in E. coli had led to the production of a catalytically active protein product that displayed a number of aberrant functional properties (35)(36)(37)(38). In the case of PP1-cyano1 and PP1-cyano2, however, the behavior of the recombinantly produced proteins mirrored that of their naturally produced counterparts, and no attempt will be made to differentiate between the two forms in the remarks that follow.
PP1-cyano1 and PP1-cyano2 are divalent metal ion-dependent phosphohydrolases. The most effective activator tested was Mn 2ϩ . In this property PP1-cyano1 and PP1-cyano2 closely resemble the bacterially associated PPP, PP-from the bacteriophage gt11 (30,33,42), and the PPP homologs recently characterized in the archaeons: i.e. PP1-arch1 from Sulfolobus solfataricus (17,43), PP1-arch2 from M. thermophila TM-1 (6,7), and Py-PP1 from P. abyssi TAG11 (8). The inhibitory spectrum of PP1-cyano1 and PP1-cyano2 appears to be fairly typical for members of the PPP superfamily of protein phosphatases, with the notable exception that both enzymes were resistant to potent inhibitory toxins such as okadaic acid and microcystin-LR, even when these compounds are present at very high levels. The toxin insensitivity of PP1-cyano1 renders the microcystin-LR producing cyanobacterium M. aeruginosa PCC 7820 immune to the effects of this endogenous toxin, which accumulates at various locations, including the thylakoid and nucleoid, inside the cell (11). Interestingly, PP1-cyano1 from the non-toxin producing strain M. aeruginosa UTEX 2063 also proved insensitive to these compounds, suggesting that if these secondary metabolites serve as a means for the molecular defense of environmental habitat (12), they are not directed against the encroachments of other cyanobacteria. The bacteriophage encoded PP-also is insensitive to these compounds (30). However, toxin insensitivity is not a general property of prokaryotic (as opposed to bacterial) PPPs, since two of the three archaeal PPP family protein phosphatases characterized to date, PP1-arch2 (6, 7) and Py-PP1 (8), are toxin inhibitable.
When challenged with a variety of phosphate-containing substrates, PP1-cyano1 and PP1-cyano2 displayed a wide range of hydrolytic capabilities. They hydrolyzed phosphoseryl and phosphothreonyl bonds in a variety of protein substrates, with the noteworthy exception of glycogen phosphorylase a. PP1-cyano1 and PP1-cyano2 also displayed significant activity toward macromolecular substrates containing phosphotyrosine, 3-phosphohistidine, and phospholysine. Interestingly, PP-had previously been reported to possess both phosphotyrosine and phosphohistidine phosphatase activity (33), while PrpA and PrpB from E. coli, two partially characterized PPPs of bacterial origin, also exhibited protein-serine and proteintyrosine phosphatase activity in vitro (26). PP1, PP2A, and PP2C from eukaryotes also have been reported to exhibit protein-histidine phosphatase in vitro (44). However, data establishing physiological relevance for this activity remains lacking several years after this observation was reported, and these enzymes are still classified as phosphoserine-and phosphothreonine-specific.
The prevalence of histidine-phosphorylated proteins in bacte- FIG. 5. Coelution of protein phosphomonoesterase, phosphoramidase, and diadenosine tetraphosphatases activity during ion-exchange and gel filtration chromatography. Top, DE52 fraction I was isolated from M. aeruginosa PCC 7820 and dialyzed as described under "Experimental Procedures." DE52 fraction I, 1 liter containing approximately 1 g of protein, was applied to a 3 ϫ 60-cm column of DE52 cellulose that was washed and developed with a linear salt gradient of 50 -400 mM NaCl. Fractions, 10 ml, were collected and portions, 15 l each, assayed for phosphohydrolase activity toward the following substrates using standard procedures: E, phosphoseryl casein; q, phosphotyrosyl casein; ‚, phospho-polylysine; OE, 3-phosphopolyhistidine; and Ⅺ, Ap 4 A. Shown is the relative phosphohydrolase activity in each fraction in either counts/min of [ 32 P]phosphate released for phosphoseryl casein and phosphotyrosyl casein, or the OD 660 of the phosphate complex detected using malachite green for all other substrates. The relative protein concentration (f) is reported as the OD 595 of the protein-Coomassie Blue complex formed in the protein assay of Bradford (9). The shape of the salt gradient is indicated by the conductivity of each fraction (ϩ). Bottom, a portion, 10 ml, containing Ϸ10 mg of protein, of the active fractions from the DE52 column, described above, protein were applied to a 5 ϫ 100-cm column of Sephacryl S-200. The column was equilibrated and developed as described under "Experimental Procedures." Fractions, 5 ml, were collected and portions, 15 l each, assayed for phosphohydrolase activity toward the following substrates using standard procedures: E, phosphoseryl casein; q, phosphotyrosyl casein; ‚, phospho-polylysine; OE, 3-phospho-polyhistidine; and Ⅺ, Ap 4 A. Shown is the relative phosphohydrolase activity in each fraction as well as the relative protein concentration (f), as indicated for Fig. 5, top. ria (45,46), including the phosphohistidyl proteins of the twocomponent regulatory system and phosphoenolpyruvate:sugar phosphotransferase system, raises the possibility that the multifunctional potential of PP1-cyano1, PP1-cyano2, and/or other bacterial members of the PPP family protein phosphatases may be realized in bacterial organisms, however. This supposition is reinforced by the observation that genetic manipulations of the genes for PrpA and PrpB in E. coli perturbed both phosphohistidine levels and two-component signaling events in this bacterium (26). The heterogeneous distribution of the various protein phosphatase archetypes, i.e. PPP, PPM, PTP, and low MW PTP, among the prokaryotes also implies that a greater degree of catalytic versatility is required of these enzymes than of their highly specialized eukaryotic counterparts (39). It is tempting to extrapolate that PP1-cyano1 and PP1-cyano2, along with other bacterial PPPs, may act on the phosphoaspartyl residues present on the response regulator modules of the two-component system. However, biochemical data indicating phosphoaspartyl phosphatase activity is lacking, and the identification of distinct sources of phosphoaspartyl phosphatase activity has been reported in the literature (47)(48)(49).
The PPP family of phosphohydrolases is not limited to protein-specific phosphomonoesterases. The PPP family also shares conserved sequence features with the diadenosine tetraphosphatase family of bacterial pyrophosphatases that act on Ap 4 A (3, 29, 32-34). Both PP1-cyano1 and PP1-cyano2 hydrolyzed Ap 4 A at a significant rate in vitro relative to phosphomonoester and phosphoramide substrates. Two lines of evidence suggest that this represents a latent or vestigial activity of protein phosphatases rather than their primary natural function. First and foremost, Ap 4 A only weakly, Ϸ30 -50%, inhibited the dephosphorylation of [ 32 P]phosphoseryl casein, despite the fact that (a) it was present at a 500-fold higher concentration than the protein-bound phosphoryl groups and (b) phosphoseryl casein is a physiologically irrelevant substrate that is not found in cyanobacteria. One would expect a dedicated diadenosine tetraphosphatase to exhibit a strong preference for its natural substrate over an exogenous mammalian phosphoprotein, particularly when the natural substrate is present in gross excess. Second, sequence comparisons group PP1-cyano1 and PP1-cyano2 with known protein phosphatases such as PP-, PrpA, and PrpB on a branch distinct from that encompassing the bacterial adenosine tetraphosphatases (Fig. 2).
While perhaps physiologically irrelevant, the ability of PP1-cyano1 and PP1-cyano2 to act as both phosphomonoesterases and pyrophosphatases in vitro possesses significant evolutionary implications. Many investigators have proposed an ancestral link between the adenosine tetraphosphatases and the PPP family of protein phosphatases based on comparisons of primary sequence data (3,29,(32)(33)(34). The results of the assays reported in this study provide the first functional evidence supporting both the proposed ancestral linkage of these enzymes and the conserved nature of the catalytic mechanism which they employ for the hydrolysis of phosphomonoester or pyrophosphate bonds.