A Novel cGMP-dependent Protein Kinase from Paramecium*

An unusual monomeric cGMP-dependent protein kinase, enriched in cilia, was isolated from Paramecium cilia and whole cells. Cilia and whole cell extracts had relatively high ratios of cGMP-dependent to CAMP-dependent protein kinase activity (1:2). The calculated molecular weight of the native enzyme was 88,000. The enzyme was identified on sodium dodecyl sulfate- polyacrylamide gels as a 77,000 molecular weight band based on copurification of this protein with enzyme activity, S-NS-[~~P]CAMP labeling, and autophosphorylation. Based on the size of the native enzyme, it was concluded that the kinase is a monomer with cGMP-binding and catalytic activities on the same polypeptide. Dimer-sized cGMP-dependent protein kinase, like that of the well characterized mammalian enzyme, was never seen, despite stringent efforts to control proteolysis. The structure of the Paramecium cGMP-dependent protein kinase supports a model in which the dimeric vertebrate form of the enzyme evolved from an early monomeric form. The catalytic properties of the Paramecium enzyme differed in several respects from those of the mammalian enzyme: it could use GTP or ATP as the phosphoryl donor, it did not phosphorylate Kemptide effectively, and it had poor histone kinase activity with high M 8 + concentra-tions. Quercertin, 5"guanylyl imidodiphosphate, in-domethacin, and the isoquinolinesulfonamide drug H7 inhibited

An unusual monomeric cGMP-dependent protein kinase, enriched in cilia, was isolated from Paramecium cilia and whole cells. Cilia and whole cell extracts had relatively high ratios of cGMP-dependent to CAMPdependent protein kinase activity (1:2). The calculated molecular weight of the native enzyme was 88,000. The enzyme was identified on sodium dodecyl sulfatepolyacrylamide gels as a 77,000 molecular weight band based on copurification of this protein with enzyme activity, S-NS-[~~P]CAMP labeling, and autophosphorylation. Based on the size of the native enzyme, it was concluded that the kinase is a monomer with cGMP-binding and catalytic activities on the same polypeptide. Dimer-sized cGMP-dependent protein kinase, like that of the well characterized mammalian enzyme, was never seen, despite stringent efforts to control proteolysis. The structure of the Paramecium cGMP-dependent protein kinase supports a model in which the dimeric vertebrate form of the enzyme evolved from an early monomeric form. The catalytic properties of the Paramecium enzyme differed in several respects from those of the mammalian enzyme: it could use GTP or ATP as the phosphoryl donor, it did not phosphorylate Kemptide effectively, and it had poor histone kinase activity with high M 8 + concentrations. Quercertin, 5"guanylyl imidodiphosphate, indomethacin, and the isoquinolinesulfonamide drug H7 inhibited Paramecium cGMP-dependent protein kinase activity. The enzyme had fast and slow binding sites (with Iza values of 5-10 X lO-'s-l and 0.44 X s-') and showed an order of preference for cyclic nucleotides and cyclic nucleotide analogs similar to that of the mammalian enzyme.
Cyclic GMP is probably involved in regulating ciliary motility and membrane excitation in Paramecium tetraurelia, a unicellular model system for cellular excitation (for review, see Refs. 1 and 2). In Paramecium, increases in cGMP levels correspond with membrane depolarization and the intracellular rise in Ca2+ which brings about ciliary reversal (3). Studies in this laboratory (4) and others (5) using detergentpermeabilized cells indicate that cGMP antagonizes Ca2+regulated ciliary reversal. Studies by Bonini and Nelson (4) also suggest distinct roles for CAMP and cGMP in regulating * This work was supported by National Institutes of Health Grants GM 34906 and GM 32514 and a National Institutes of Health Training Grant in Cell and Molecular Biology. 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.
ciliary orientation. Some, if not all, of the actions of cGMP on Paramecium motility are likely to be brought about through stimulation of cGMP-dependent protein kinase. Paramecium cilia have high levels of cyclic nucleotide-dependent protein kinase activity relative to most mammalian tissues (6,7) and have an unusually high level of cGMP-dependent kinase activity, almost as high as that of CAMP-dependent kinase. The phosphorylation in vitro of several ciliary proteins is stimulated by the addition of cGMP (8,9).
Much is known about the mammalian cGMP-dependent protein kinase, which has been purified from heart and lung tissue, thoroughly characterized (for review, see Ref. lo), and sequenced (11). The bovine enzyme is a disulfide-linked dimer composed of identical 76,331-dalton subunits which do not dissociate when activated by cGMP. Each subunit contains two cGMP binding sites, distinguishable by their dissociation rates for cGMP, and a catalytic site. Mammalian cGMPdependent protein kinase has recently been studied with respect to its evolutionary history. Based on sequence homology studies by Takio and co-workers (ll), it has been proposed that the evolution of the mammalian cGMP-dependent protein kinase involved a gene fusion event between a primitive protein kinase and a cyclic nucleotide binding protein. However, details regarding the order of these events are not yet known. The study of cyclic nucleotide-dependent protein kinases in simple eukaryotes may elucidate how the mammalian enzymes evolved (12).
To study the possible role of cGMP-dependent protein phosphorylation in Paramecium swimming behavior we have isolated and characterized the cGMP-dependent protein kinase from this organism. Our goals were to isolate enzyme which could be used as a biochemical tool in studying the regulation of ciliary motility and membrane excitation and to compare the enzymatic properties of the Paramecium cGMPdependent protein kinase with those of the enzyme of vertebrate origin. We isolated the enzyme from cilia and whole cells and found it to be a monomer. This is strikingly different from the dimeric structure of the mammalian enzyme. We found that some of its catalytic properties, such as the ability to use GTP as a phosphoryl donor, and poor use of Kemptide as a substrate, also differed greatly from those of the vertebrate enzyme. However, the regulatory region of the Paramecium cGMP-dependent protein kinase had similar properties to the vertebrate enzyme; it appeared to contain two cGMP-binding sites, characterized by fast and slow dissociation rates. We propose that the Paramecium enzyme represents an early evolutionary form of the kinase, containing the catalytic site and two cGMP-binding sites, but lacking the dimerization site present in the vertebrate enzyme.

RESULTS
Partial Purification of the cGMP-dependent Protein Kinase-The cyclic nucleotide-dependent protein kinases were enriched in the cilia of Paramecium. Cilia contained about 20% of the total cell cyclic nucleotide-dependent protein kinases, and about 3% of the total cellular protein; the specific activity in the presence of cGMP or cAMP was 5-7 times higher in cilia than in whole cells. However, extracts of whole cells contained more total kinase than cilia, allowing more extensive purification with better recoveries (Table I, Fig. 1). In preliminary work, we established that the CAMP-dependent protein kinases of Paramecium were only slightly stimulated by 0.2 p~ cGMP (32), and the cGMP-dependent protein kinase was only slightly stimulated by 0.2 p~ CAMP, which allowed us to estimate the activity of each enzyme type in the presence of the other. Both CAMP-and cGMP-dependent protein kinases were present in the 40-60% ammonium sulfate pellet (Table I, Part A). The DEAE-cellulose column pass-through fraction contained the cGMP-dependent protein kinase as well as some CAMP-dependent protein kinase. Further purification on Sephadex G-150 and Cm-cellulose separated the cGMP-and CAMP-dependent protein kinases and resulted in a 200-300-fold increase in cGMP-dependent protein kinase specific activity (see Miniprint Section). Enzyme purified through the Cm-cellulose step was usually 5-10% pure.
The cGMP-dependent protein kinase was purified from cilia for enzyme characterization because of our interest in enzymes associated with ciliary regulation. The cGMP-dependent protein kinase was isolated from cilia as an apparently soluble enzyme, whereas the CAMP-dependent activity tended to be associated with insoluble ciliary fractions (see Miniprint Section). The combination of passing the ciliary extract through DEAE-cellulose followed by gel filtration chromatography resulted in a rapid and substantial purification of cGMP-dependent protein kinase (Fig. 1, Table I, Part B) which contained no detectable CAMP-dependent kinase activity. Cyclic GMP-dependent protein kinase purified to ' Portions of this paper (including "Experimental Procedures," part of "Results," Table 11, and Figs. 2-4, 6-9) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press. this extent was stable at 4 "C when stored with 0.02% NaN3, with less than 10% loss of specific activity after 3 months. Size of the Native Protein-The size of native cGMPdependent protein kinase was determined using samples prepared with strenuous efforts to avoid proteolysis. For the ciliary enzyme, cilia were isolated in the presence of PMSF' and TAME (0.3 mM each), and the ciliary pellet was homogenized in METP buffer containing leupeptin (10 pglrnl). Following centrifugation at 35,000 X g for 30 min, the supernatant fraction was applied immediately (without freezing and thawing) either to Sephadex G-150 to determine the Stokes radius or to sucrose gradients to determine the sedimentation coefficient.
Gel filtration columns and sucrose gradients were run in the presence of PMSF and TAME, and fractions were assayed with 0.2 p~ 8-bromo-cGMP, a selective activator of Paramecium cGMP-dependent protein kinase over CAMP-dependent protein kinases (32). Under these isolation conditions, the CAMP-dependent protein kinase ran in its large form (220 kDa on  indicating that proteolysis of the protein kinases was minimal or did not occur (32).
For the enzyme from whole cells, the extract was prepared as described except that aprotinin (0.17 trypsin inhibitor units/ml), leupeptin (10 pg/ml), and pepstatin A (1 pg/ml) were added to the cell homogenization buffer in addition to PMSF and TAME, and the 18,000 X g centrifugation was omitted. The cell washes, extraction, and 150,000 x g centrifugation were performed in less than 3 h at 4 "C. The supernatant fraction was then immediately loaded onto Sephadex G-150 or sucrose gradients, and the fractions were assayed for protein kinase activity in the presence of 8-bromo-cGMP.
The molecular weight of the native protein was calculated to be 88,000 based on the Stokes radius and the sedimentation coefficient (see Miniprint Section) (20). This molecular weight was used to estimate a frictional ratio, f/fo, of 1.4. These values are approximations, assuming the partial specific volume to be the same (0.74 cm3/g) as that for the mammalian enzyme (22).
Identification of cGMP-dependent Protein Kinuse on SDS-Polyacrylamide Gels-Several lines of evidence supported the identification of the cGMP-dependent protein kinase as a polypeptide of approximately 77 kDa on SDS-polyacrylamide gels. The density of a 77 kDa band correlated with cGMPdependent protein kinase activity when fractions from Sephadex G-150 and Cm-cellulose columns were observed by SDS-PAGE, and samples containing roughly equivalent amounts of cGMP-dependent protein kinase activity contained approximately the same levels of the 77-kDa protein ( Fig. 1). Furthermore, when partially purified cGMP-dependent protein kinase was passed over 8-(2-aminoethyl)-amino-cAMPagarose, the enzyme activity was removed, and this loss of activity was accompanied by the loss of a 77 kDa band on SDS-PAGE (data not shown). In contrast to the isolation of mammalian cGMP-dependent protein kinase (33), affinity chromatography was not a useful purification step because of low recoveries of enzyme activity.
When partially purified cGMP-dependent protein kinase from cilia or whole cells was incubated with the affinity label 8-N3-[32P]cAMP, a 77-kDa protein was labeled (Fig. 5). A 20fold excess of cGMP or cAMP competed for the binding of the labeled analog. This reactivity differed from that of CAMP-dependent protein kinase regulatory subunits, the labeling of which was completely blocked by a 20-fold concen-    5). This 77-kDa phosphoprotein was found in all partially purified cGMPdependent protein kinase samples analyzed and in the purest samples obtained. The phosphorylation of this protein was reduced about 40% in the presence of 1 PM cGMP, even in crude samples which contained several proteins that showed enhanced phosphorylation in the presence of cGMP. The phosphorylation of the 77-kDa protein also was inhibited about 30% by 1 PM CAMP. Cyclic nucleotides appeared to affect the extent of the phosphorylation, rather than the initial rate (data not shown). The phosphorylation of the 77-kDa protein was slow; maximal phosphorylation occurred after 40-60 min. When autophosphorylated cGMP-dependent protein kinase samples were subjected to SDS-PAGE in the absence versus the presence of 2-mercaptoethanol, the apparent size of the 77-kDa protein did not change (data not shown).
Enzyme Characterization-Most of the characterizations of catalytic activity were performed using cGMP-dependent protein kinase from cilia, purified through the Sephadex G-150 stage. Following partial purification, the ciliary cGMP-dependent protein kinase was considered suitable for enzyme characterization studies; DEAE-cellulose separated the cGMP-dependent protein kinase from the CAMP-dependent protein kinases, as well as from Ca2+-dependent protein kinases and CAMP-and cGMP-phosphodiesterase^.^ This sample was also free of measurable protein phosphatase activity as determined by ATP pulse-chase experiments, and by the fact that 5 mM NaF, a protein phosphatase inhibitor (8, 34), did not increase the measured enzyme activity (data not shown). For comparison, some experiments employed cGMPdependent protein kinase purified from whole cells through the Cm-cellulose step. Characterization of pH and divalent cation dependence, activation by cyclic nucleotides, effects of protein kinase inhibitors, and cGMP binding activity are presented in the Miniprint Section. Ability to use ATP or GTP as Phosphoryl Donor-Both ATP and GTP could serve as substrates for the Paramecium cGMP-dependent protein kinase. With either ATP or GTP as phosphoryl donor, the effect of cGMP was to increase the reaction velocity without substantially changing the K,. The R. E. Gundersen, personal communication. enzyme had a higher K , for GTP (30-40 PM) than for ATP (10 PM), and the V,,, when assayed with GTP was roughly 50% that of the same preparation assayed with ATP (Table   111). The cGMP-dependent protein kinase isolated from whole cells could also use GTP as a substrate whereas neither of the CAMP-dependent protein kinases from Paramecium could use GTP (data not shown). By thin layer chromatography, it was shown that GTP was not converted to ATP during the protein kinase reaction (see Miniprint Section).
Several lines of evidence indicate that the kinase using GTP was the cGMP-dependent protein kinase and not a different, contaminating protein kinase. The protein kinase activity with GTP was stimulated by cGMP and co-purified with cGMP-dependent protein kinase on both DEAE-cellulose and gel filtration columns (data not shown). GTP could also be used in the autophosphorylation reaction to phosphorylate the 77-kDa cGMP-dependent protein kinase (data not shown).
Protein and Peptide Substrates-The proteins that served as good substrates for the cGMP-dependent protein kinase were similar to those that are good substrates for protein kinases in general (Table 111). Using the standard assay mixture (pH 6.0,0.2 pM cGMP), histone 11-A, a mixed histone preparation, had the lowest K,,, (0.064 mg/ml) of all proteins tested. Partially dephosphorylated casein had the highest K , (4.0 mg/ml) and also had the highest V,,,.
To determine which amino acids were phosphorylated by the cGMP-dependent protein kinase, histone 11-A was phosphorylated by the enzyme, subjected to acid hydrolysis, and the phosphoamino acids were separated by electrophoresis on cellulose thin layer plates (data not shown). The 32P label was associated with serine (70%) and threonine (29%) residues. The counts associated with Ser:Thr:Tyr were 626:256:9 cpm after subtracting a background of 31 cpm.
The Paramecium cGMP-dependent protein kinase was essentially unable to phosphorylate Kemptide ( K , > 1 mM). As

DISCUSSION
This study provides the first extensive description of a cGMP-dependent protein kinase in a unicellular eukaryote. We found several unusual features of the enzyme, including its monomeric structure, its ability to use GTP, and its inability to phosphorylate Kemptide. Paramecium was found to contain cGMP-dependent protein kinase at a high specific activity relative to other organisms (39, 40; see also Ref. 7). Cilia were enriched for the enzyme compared to whole cells, consistent with the proposed role of cGMP-dependent protein phosphorylation in the regulation of ciliary motility (3)(4)(5).
Enzymatic purity of the Paramecium cGMP-dependent protein kinase was attained, in the sense that the cGMPdependent protein kinase was separated from the Ca2'4 and CAMP-dependent protein kinases by ion exchange chromatography. Purification of the cGMP-dependent protein kinase allows its use as a biochemical tool to explore its function in R. E. Gundersen, personal communication. ciliary motility and membrane excitation. Microinjection of the enzyme into cells (41, 42) or its addition to detergentpermeabilized paramecia (43) will help to determine whether effects of cGMP on ciliary motility (4, 5 ) and membrane excitation (3) are brought about through action of the kinase.
The kinase was purified about 60-fold from cilia and 300fold from whole cells. However, this probably underestimates the extent of purification because crude fractions contain other kinases which contribute to the apparent cGMP-dependent protein kinase activity. Based on stained SDS-polyacrylamide gels, the cGMP-dependent protein kinase isolated from whole cells was 5-10% pure and that from cilia 1-2% pure. The specific activity of the pure cGMP-dependent protein kinase was estimated to be 2-10 rmol. min" . mg".
Paramecium cGMP-dependent protein kinase in whole cell extracts and the bulk of the enzyme in ciliary extracts passed through DEAE-cellulose. Whole cell Paramecium cGMP-dependent protein kinase and the three forms of ciliary cGMPdependent protein kinase resolved on DEAE had the same size as assessed by gel filtration. Multiple peaks of cGMPdependent protein kinase on DEAE-cellulose may represent different isozymes, enzyme in its phosphorylated and nonphosphorylated states, enzyme bound to other proteins, or slightly proteolyzed forms. The ciliary cGMP-dependent protein kinase from Tetrahymena also elutes from DEAE-cellulose in three peaks (44), similarly consisting of one passthrough peak, and two peaks at the beginning of the NaCl gradient. The Tetrahymena cGMP-dependent protein kinase from all three peaks was the same size, as assessed by gel filtration chromatography. The only other report of multiple forms of cGMP-dependent protein kinase within one organism is the membrane-associated form in intestinal brush borders (45).
Several lines of evidence support the identification of the Paramecium ciliary or whole cell cGMP-dependent protein kinase as a 77 kDa band on SDS-polyacrylamide gels. A protein of this size corresponded with enzyme activity in fractions from several different column types. It was removed with cGMP-dependent protein kinase activity when partially purified enzyme preparations were passed over %(aminoethyl)-amino-CAMP-Sepharose. In partially purified samples, a 77-kDa protein was labeled by 8-N3-[32P]~AMP in a manner that was competed by excess cAMP or cGMP. Others have photoaffinity-labeled a protein of similar molecular weight in Paramecium cilia (9), the labeling of which was also competed by both cAMP and cGMP, suggesting it was the cGMPdependent protein kinase. A protein of this size was also labeled when partially purified enzyme preparations were incubated with [y3*P]ATP. This labeling was reduced in the presence of 1 p~ cGMP. The unusual effect of cGMP on the phosphorylation of the 77-kDa protein supports its identification as the autophosphorylated cGMP-dependent protein kinase, because it suggests that the protein interacted with cGMP, yet the phosphorylation was not increased by cGMP as a substrate of the cGMP-dependent protein kinase would be.
The vertebrate soluble cGMP-dependent protein kinase is a disulfide-linked dimer consisting of two identical 76,331dalton subunits (11) which do not dissociate upon activation by cGMP (22). Our results indicate that the native Paramecium cGMP-dependent protein kinase has a molecular weight of 77,000-88,000. The Paramecium enzyme was never observed to run on gel filtration columns or sucrose gradients as a dimer of two 77-kDa subunits. Ciliary or whole cell extracts freshly prepared in the presence of protease inhibitors then immediately loaded onto gel filtration columns or sucrose gradients showed only a single peak of cGMP-dependent protein kinase, corresponding to the 88,000 molecular weight enzyme. The monomeric Paramecium cGMP-dependent protein kinase was stable upon storage at 4 "C for months, unlike the fragments of the mammalian dimer described by Lincoln and co-workers (46). When SDS-polyacrylamide gels were run in the presence and absence of 2-mercaptoethanol, the size of autophosphorylated cGMP-dependent protein kinase did not change. Thus, despite strenuous efforts to observe a dimer, we found no evidence for this form of the enzyme. We conclude that the native Paramecium cGMP-dependent protein kinase is a monomer with a size similar to a single subunit of the vertebrate enzyme.
Only a few cGMP-dependent protein kinases of a size distinct from the dimeric vertebrate enzyme have been described. The monomeric cGMP-dependent protein kinase isolated from vertebrate brush border membranes (45) has a molecular weight of 96,000 and is tightly membrane-bound. Soluble, monomeric (70-80 kDa) cGMP-dependent protein kinases have been reported in insects (47, 48), although the 140-kDa enzyme isolated from silkworm pupae is dimer-sized (49). The soluble cGMP-dependent protein kinase from Tetrahymena has an apparent native molecular weight of 100,000 by gel filtration (44) and may thus resemble the Paramecium enzyme.
The dimeric vertebrate cGMP-dependent protein kinase is activated by cGMP in an allosteric manner (19), with no subunit dissociation (22). The subunits are proposed to be aligned in an antiparallel fashion such that the cGMP-binding portion of one chain interacts with the catalytic portion of the other (50). In the absence of cGMP, the catalytic site is blocked by the regulatory region of the opposite chain. When cGMP binds to the regulatory regions of the subunits, the protein structure changes to expose the catalytic region, thus activating the enzyme. The activation of a monomeric protein kinase by cGMP implies a different mechanism: either the protein is able to fold upon itself in the absence of cGMP so that the regulatory region blocks the catalytic region, or cGMP alters the conformation of the protein in some other way.
Differences in the autophosphorylation properties between the Paramecium and vertebrate enzymes may reflect their different structures. Autophosphorylation occurs slowly for both the Paramecium and vertebrate (51,52) enzymes, reaching maximal levels after 40-60 min of incubation with ATP. By densitometry of stained gels we estimated that Paramecium cGMP-dependent protein kinase incorporated one to two phosphates per kinase molecule, whereas the mammalian enzyme can incorporate up to four phosphates per subunit (53). The extent of autophosphorylation of the Paramecium cGMP-dependent protein kinase was reduced by cyclic nucleotides. This property differs from that of the vertebrate enzyme, for which cyclic nucleotides increase the rate and the extent of autophosphorylation (53). The lack of cyclic nucleotide stimulation of the Paramecium kinase autophosphorylation may reflect a different mechanism due to the monomeric structure of the enzyme. The 71-kDa proteolytic fragment from the membrane-bound cGMP-dependent protein kinase in intestinal brush borders undergoes autophosphorylation which is also decreased by cyclic nucleotides (45).
The catalytic properties of the Paramecium cGMP-dependent protein kinase differed in some respects from those of the vertebrate enzyme. We initially chose to characterize the enzyme isolated from cilia because of our interest in studying the enzymes involved in regulating ciliary motility. However, the resemblance between the structural and catalytic proper-ties of the cGMP-dependent protein kinase isolated from whole cells and cilia suggests that they are the same or very similar enzymes.
We confirmed that the Paramecium enzyme can use either ATP or GTP as the phosphoryl donor in the phosphotransferase reaction (7), which distinguishes it from other reported cyclic nucleotide-dependent protein kinases (54). The inhibition of Paramecium cGMP-dependent protein kinase by competitive inhibitors of ATP may reflect its ability to use GTP. For example, the enzyme was inhibited by the nonhydrolyzable GTP analog Gpp(NH)p more effectively than were the CAMP-dependent protein kinases, which could not use GTP. The bioflavenoid quercetin inhibits several protein kinases, including casein kinase I1 (36) and pp60"" (35), both of which can use GTP in place of ATP (55, 56). This compound also inhibited the Paramecium cGMP-dependent protein kinase.
Quercetin does not inhibit the vertebrate CAMP-dependent protein kinases (35,36), although we observed some inhibition of the Paramecium CAMP-dependent protein kinases. These observations suggest a catalytic similarity between the Paramecium cGMP-dependent protein kinase and other protein kinases that can use GTP. The use of the divalent cations M e and Co2+ by the Paramecium cGMP-dependent protein kinase resembled that of the cGMP-dependent protein kinases of vertebrate (40) and arthropod (57) sources. The substrate specificity for the cGMP-dependent protein kinases from Paramecium and mammalian sources may reflect a dissimilarity between the substrate phosphorylation site sequences for the two enzymes. In contrast to the mammalian enzyme (58), the heptapeptide protein kinase substrate (Kemptide) did not serve as a substrate for the Paramecium kinase. However, as with other cyclic nucleotide-dependent protein kinases, histones and casein were good substrates for the Paramecium enzyme.
While there appeared to be some striking differences between the catalytic properties of the Paramecium and vertebrate cGMP-dependent protein kinases, their regulatory properties were quite similar. The order of effectiveness of cyclic nucleotides and cyclic nucleotide analogs in stimulating protein kinase activity was similar for the Paramecium and mammalian enzymes, with roughly similar activation constants for each nucleotide (10,40). Also, the Paramecium enzyme, like that of vertebrate origin (18,59), appeared to have two types of cGMP-binding sites characterized by fast uersus slow dissociation rates. Slightly higher cGMP binding was observed when dilution was performed in the presence of 2.65 M ammonium sulfate, presumably because there was less loss of [3H]cGMP from the fast dissociation site (18). The two different cGMP-binding sites were probably not due to two different cGMP-binding proteins, because photoaffinity labeling indicated only one cyclic nucleotide-binding protein in these partially purified samples. By estimating the amount of cGMP-dependent protein kinase from densitometry of SDS-polyacrylamide gels we determined that maximal cGMP binding was close to stoichiometric, but these measurements were not accurate enough to distinguish between one or two cGMP molecules bound per monomer.
We found no evidence for cooperative regulation of the Paramecium cGMP-dependent protein kinase by cGMP. Hill plots of cGMP-binding saturation curves and cyclic nucleotide stimulation of enzyme activity had slopes of 0.8-1.1, indicating a lack of positive or negative cooperativity. Cyclic GMP stimulation of the vertebrate enzyme occurs with positive cooperativity with a Hill coefficient of 1.59 (19). With the mammalian enzyme, the k d for the fast site is 5-15 times lower in the presence of cGMP, which is also characteristic of positive cooperativity. However, we observed only a slight retardation of cGMP dissociation from the Paramecium enzyme in the presence of cGMP (0.1 mM) (data not shown). Proteolytic cleavage of the mammalian cGMP-dependent protein kinase yields a monomeric but cGMP-stimulatable protein kinase, which lacks the positive cooperativity of the dimeric enzyme (19). The lack of cooperativity demonstrated by the Paramecium enzyme is probably a reflection of its monomeric structure.
The characteristics of enzymes in simple organisms can yield information about the evolution of their mammalian counterparts (12, 60, 61). The enzymes in P. tetraureliu evolved from very early forms; sequencing of the small (62) and large5 rRNA subunits places the emergence of Paramecium early in evolution, prior to the divergence of plants, animals, and yeast. Sequence homology studies suggest that the cGMP-dependent protein kinase evolved as the result of a fusion between genes for prototypical cyclic nucleotidebinding and catalytic proteins (11). Our data suggest that this fusion occurred early in evolution, since the Paramecium cGMP-dependent protein kinase also has cyclic nucleotidebinding and catalytic activities on the same polypeptide chain. The proposed internal tandem gene duplication producing the two cyclic nucleotide-binding sites on each subunit of the mammalian cyclic nucleotide-dependent protein kinases (11, 63) must also have occurred very early in evolution since the Paramecium cGMP-dependent and CAMP-dependent (32) protein kinases also have two types of cyclic nucleotidebinding sites. However, the monomeric structure of the Paramecium cGMP-dependent protein kinase supports a scheme in which dimerization of the enzyme occurred later in evolution. A similar order of events has been proposed for the CAMP-dependent protein kinase (12), with the internal gene duplication being an early event, followed later by dimerization.

EXPERIMENTAL PROCEDURES
Isoletion of E G M P -a d e n t DZ orein kinase from cilia -Cells prepared as described .above were deciliaeed using the calcium shack procedure (15). lsolaced cilla were resuspended in METP buffer to a final prorein concenrrerion of 5 to 20 mg/ml, and were scored a t -20DC Por each purlficacion, fresh PnSF and TAME (0 3 mM each) were added to frozen cilia (0.2 fa 1 0 g), which were then thawed, homogenized, subjected to 5 cycles of freezing end thawing, and centrifuged at 28,000 x g for 30 ain. The pellet was dilutsd to npprorilnafely 10 mg/ml protein in METP buffer. subjeered co 5 more freezing and thawing cycles, and centrifuged again. Isoletion of E G M P -a d e n t DZ orein kinase from cilia -Cells prepared as described .above were deciliaeed using the calcium shack procedure (15). lsolaced cilla were resuspended in METP buffer to a final prorein concenrrerion of 5 to 20 mg/ml, and were scored a t -20DC Por each purlficacion, fresh PnSF and TAME (0 3 mM each) were added to frozen cilia (0.2 fa 1 0 g), which were then thawed, homogenized, subjected to 5 cycles of freezing end thawing, and centrifuged at 28,000 x g for 30 ain. The pellet was dilutsd to npprorilnafely 10 mg/ml protein in METP buffer. subjeered co 5 more freezing and thawing cycles, and centrifuged again.  (Table IA). This column also resolved a peak of eGNP-dependent protein kinase DOC show stilovlaCion by cyclic nuoleofidea (PIP. PA). The eGMP-deoendsnr rotei in kinase YSQ (usually eonraining CAMP-dependent protein'kineos as well) from B protein kinase that did separated from conlminaling CAMP-dependant protein kinases on Cn-cellulose. The CAMPdependent protein kineses passed thr'ough this column and the CCMP-dspendent prorein kinase with Cm-cellulose (Table IA Time (minutes) -6.5