Identification of protein phosphatase 1 in synaptic junctions: dephosphorylation of endogenous calmodulin-dependent kinase II and synapse-enriched phosphoproteins

A calcium/calmodulin-dependent protein kinase termed CaM-kinase II is a major component of synaptic junctions from forebrain and constitutes approximately 12% of total synaptic junction protein. CaM-kinase II phosphorylates at least seven polypeptides that are enriched in synaptic junctions, of which two represent the 50- and 60-kilodalton subunits of the protein kinase. In this report the nature of endogenous protein phosphatases which dephosphorylate each of the seven synaptic junction phosphoproteins was examined. Assays of synaptic junctions and other subcellular fractions from rat forebrain for type-1 and type-2 protein phosphatases revealed that protein phosphatase 1 (PrP-1) was specifically enriched in synaptic junctions with respect to cytosolic fractions. The activity of type-2 protein phosphatases was very low in synaptic junctions. Homogeneous PrP-1 from rabbit skeletal muscle was found to dephosphorylate each of the seven phosphoproteins in synaptic junctions. Inhibitors-1 and -2 were found to inhibit endogenous protein phosphatase activity by 70 to 80%. Since inhibitors-1 and -2 are specific inhibitors of PrP-1, these results indicate that this enzyme accounts for the majority of endogenous protein phosphatase activity in synaptic junctions. Approximately 15% of the protein phosphatase activity in synaptic junctions was type 2A, whereas PrP-2B and PrP-2C accounted for little, if any, of the activity toward endogenous or exogenous phosphoproteins. These results indicate that PrP-1 may be important in controlling the state of phosphorylation of synaptic junction proteins.

. A CaM-dependent protein kinase (designated CaM-kinase II) in brain cytosol and particulate fractions that phosphorylates synapsin I (Ueda et al., 1979;De Camilli et al., 1983) has been purified to near homogeneity and is composed of a major 50-kilodalton (kd) polypeptide and a minor 60.kd polypeptide (Bennett et al., 1983;Groswald et al., 1983;Kennedy et al., 1983a, b;Lai et al., 1983;Kelly et al., 1984). Both polypeptides are phosphorylated in a Ca'+/ CaM-dependent manner and bind CaM. CaM-stimulated protein kinase activity is present in postsynaptic densities (PSDs) from canine cerebral cortex (Grab et al., 1981) and has been shown to be concentrated at asymmetric synaptic junctions (Kelly et al., 1983(Kelly et al., , 1985. At least seven prominent phosphoproteins which serve as in vitro substrates for this CaM-dependent protein kinase are also concentrated in synaptic junctions. Two of these phosphoproteins constitute the 50. and 60.kd components of CaM-kinase II (Bennett et al., 1983;Kennedy et al., 198313;Mc-Guinness et al., 1983;Kelly et al., 1984Kelly et al., , 1985. The other five include synapsin I and phosphoproteins of 240, 207, 170, and 140 kd (Kelly et al., 1985); the putative postsynaptic functions of these phosphoproteins remain unknown.
The most abundant protein in PSDs and synaptic junctions (SJs) is a 50.kd polypeptide which constitutes approximately 50% of the total protein in isolated PSDs. This polypeptide, designated the major PSD protein (mPSDp), is thought to make up in large part the submembranous cytoskeleton that underlies the postsynaptic membrane at asymmetric synapses (Banker et al., 1974;Kelly and Cotman, 1978;Cotman and Kelly, 1980;Kelly and Montgomery, 1982). Recent work from this laboratory has demonstrated that the major PSD protein is virtually identical to the 50-kd polypeptide of CaM-kinase II (Groswald et al., 1983;McGuinness et al., 1983;Kelly et al., 1984). Similarly, a 60.kd phosphoprotein in SJs is highly related to the 60.kd subunit of CaM-kinase II. Some of these molecular similarities have recently been reported by Kennedy et al. (198313) and Goldenring et al. (1984). The enrichment of CaM-kinase II/ mPSDp at SJs, together with the presence of a number of endogenous substrate proteins, suggests that CaM-dependent protein phosphorylation may play important functional roles at CNS synapses The subcellular distribution and properties of protein kinases in brain are consistent with the notion that intracellular signals due to calcium influx into neurons and increases in cyclic nucleotides result in the phosphorylation of specific proteins (Krueger et al., 1977;Forn and Greengard, 1978). In all phosphorylation systems studied to date, the steady-state level of protein phosphorylation is determined by the balance of the protein kinase and protein phosphatase (PrP) activities.
Moreover, specific mechanisms have been identified that regulate PrP activity in response to CAMP and calcium (Ingebritsen   3414  and Cohen, 1983a, b;Huang and Glinsmann, 1976). If 'protein phosphorylation is to play important roles in physiological events underlying synaptic communication, one would expect to find PrP activity co-localized with the enzymatic machinery responsible for the phosphorylation of the same substrate proteins. Furthermore, if protein phosphorylation constitutes an activation or "on" event associated with neuronal functions, then for reversible function a given phosphoprotein must be dephosphorylated. Four enzymes, termed protein phosphatases 1, 2A, 2B, and 2C, account for virtually all of the observed protein phosphatase activity toward a large number of phosphoproteins involved in the regulation of intermediary metabolism (Ingebritsen and Cohen, 1983a, b;Ingebritsen et al., 1983a, b, c;Pato et al., 1983;Stewart et al., 1983). These four PrPs have been divided into two classes (Ingebritsen and Cohen, 1983a, b). Type-l protein phosphatase (protein phosphatase 1 or PrP-1) selectively dephosphorylates the @-subunit of phosphorylase kinase and is potently inhibited by nanomolar concentrations of two heat-stable regulatory proteins, termed inhibitors-1 and -2. Conversely, the three type-2 protein phosphatases (PrP-2A, -28, and -2C) selectively dephosphorylate the a-subunit of phosphorylase kinase and are refractory to inhibitors-l and -2. Interestingly, inhibitor-l is an active inhibitor of PrP-1 only after being phosphorylated on a specific threonine residue by CAMP-dependent protein kinase (Huang and Glinsmann, 1976).
The results presented herein demonstrate that PrP-I is the major phosphatase activity in SJ and synaptic plasma membrane (SPM) fractions from rat forebrain. PrP-1 accounts for virtually all of the protein phosphatase activity toward the major phosphoproteins present in SJ and SPM fractions that are substrates for CaM-kinase II. Lesser amounts of PrP-2A were observed in SJ and SPM fractions. The possible significance of these findings to the regulation of protein phosphorylation in nerve terminals is discussed.

Subcellular
fractionation. SPM and SJ fractions were prepared from rat forebralns (60 to 100 days of age) by the lodonitrotetrazolium violet/Triton X-100 method as previously described (Kelly and Montgomery 1982); protein concentrations were determined as outlined elsewhere (Kelly and Montgom ery, 1982). SJs were prepared from SPMs by extracting the latter with Tnton X-100 (TX); approximately 85% of the SPM protein was recovered in detergent extracts.
Other subcellular fractions from brain were prepared as previously described (Groswald and Kelly, 1984 (Kelly et al., 1984) with minor modifications. Phosphorylation reactlons were .terminated after 1 min at 30°C (ATP) or 30 min at 24°C (thioATP) by the addition of 850 ~1 of 2 mM HEPES (pH 7.2), 1 mM EGTA, and centrifuging at 10,000 x g for 20 set or 5 min for SJ and SPM fractions, respectively.
Supernatants were decanted and SJ or SPM pellets were washed two additional times by resuspension and centrifugation in the same buffer. SJ and SPM pellets were resuspended (1 to 2 @g of protein/PI) in 2 mM HEPES (pH 7.2) and used in dephosphorylation reactions (see below). Endogenous dephosphorylation of SJ and SPM proteins. 32P-or =Slabeled SJ or SPM proteins were dephosphorylated at 37°C in a standard dephosphorylation buffer that contains 50 mM Tris-HCI, 0.5 mM dithiothreitol (DTT) (pH 7.0), 25 pg/ml of a-2-macroglobulin, 10 pg/ml of leupeptin, 10 fig/ml of pepstatin A, and 1.5 mM phenylmethylsulfonyl fluoride (PMSF). Additional dephosphorylation reactions contained CaCI* and CaM (2 mM and 0.02 pg/pl, respectively), MgCIZ (2 mM), or EGTA (5 mM). In some experiments Triton X-100.soluble SPM fractions (TX-sol SPMs) or purified PrP-1 and/or PrP-2A were added to reactions at 4°C prior to Initiating protein dephospho-rylation by shifting reactions to 37°C In other experiments purified phosphatase inhibitors-l and/or -2 were added to reactions. Dephosphorylation reactions were carried out for various durations (10 to 120 min) to ensure linearity and were terminated by the addition of 4~ SDS-stop buffer and immediately heated at 70°C for 5 min. Reaction mixtures were analyzed on 7 to 16% exponential gradient slab gels. Gels were stained with Coomassle blue as described previously (Kelly et al., 1979). Autoradiography of gels was performed at -75°C with Curlx RP-1 x-ray film and Cronex intensifying screens. Autoradiographic patterns were quantltated using a scanning densitometer with Integrator (E-C Apparatus). Multiple autoradiographic exposures of varying durations were made of each gel to ensure that densitometric measurements were within the linear range of silver grain development. Following autoradiographic exposures, 32P-or 35S-labeled bands were excised and subjected to liquid scintillation counting. Results from densitometry and beta counting were in close agreement.
Dephosphorylation of endogenous and exogenous phosphoproteins by synapbc and purified PrPs. Synaptic fractions or purified PrPs were assayed for their ability to dephosph&ylate exogenous "Psynapsin I or "P-CaMklnase II. Reactions contained 1 .O UCI of ?svnaosin I or 3"P-CaM-kinase II and the desired synaptic fraction (5s pg of protein) or purified PrP in 45 11 of the standard dephosphorylation buffer. Following Incubations at 37°C or 4'C. dephosphorylation reactions were terminated and analyzed as described above.
Control reactions using different substrate dilutions and highly purified PrPs of known specific activity were performed to ensure that assays were carried out in the linear range of protein dephosphorylatlon as a function of time and the amount of added phosphatase.
Alkaline phosphatases from Escherichia co/i, human placenta, and bovine Intestine, and acid phosphatases from human semen and potato (Sigma Chemical Co.) were assayed for their ability to dephosphorylate purified "P-CaM-kinase II or 32P-SJ phosphoproteins. "P-CaM-kinase II (1.8 Fg) was incubated with 0.8 units of each phosphatase using the appropriate alkaline or acid conditions in 45 ~1 of buffer (50 mM Tris-HCI, 0.5 mM DTT) for 90 min at 4°C or 37°C. Certain reactions contained "P-SJs that were solubilized with SDS (1%) prior to the addition of different purified phosphatases. Reactions were terminated as described above and analyzed by autoradiography and liquid scintillation counting.
Assay of 3ZP-inorganic phosphate release from phosphoproteins. The release of 3ZP-inorganic phosphate from 32P-SJ phosphoproteins was conducted by the phase-separation technique described by Antoniw and Cohen (1976). Briefly, dephosphorylation reactions were terminated with 4x SDSstop buffer and split into three equal aliquots. One aliquot was analyzed by SDS-PAGE and autoradrography.
The remaining duplicates received 50 ~1 of 2% bovine serum albumin followed by 100 ~1 of ice-cold 17.5% (w/v) trichloroacetic acid (TCA). Protein precipitates were allowed to form at 4°C for 20 min. Precipitates were collected by centrifugation; supernatants were removed and mixed with 200 ~1 of 1.25 mM KHzP04 in 1 .O N HZS04, and 500 ~1 of isobutanol:benzene (1 :l v/v). Ammonium molybdate (125 ~1 of 5%, w/ v) was added and mixed, and each sample was centrifuged to separate the mixture into organic and aqueous phases. Aliquots (300 ~1) of the organic phase containing 3'P,-molybate complexes were added to liquid scintillation cocktail (4 ml) and counted; 32P-peptides which could arise from proteolytic degradation during dephosphorylation reactions are excluded from the organic phase in this assay. inactivation of endogenous phosphafases. SJ and SPM fractions (1 to 2 rg/pI), before or after phosphorylation by endogenous CaM-kinase II, were incubated in 50 mM NaF, 0.5 mM EDTA at 4°C for 30 min to 2 hr. Incubations were terminated by centrifugation (10,000 x g for 30 set), supernatants were decanted, and pellets were washed three times by resuspension/ centnfugation in 10 mM HEPES (pH 7.2). SPM or SJ fractions were then resuspended at 1 to 2 pg/pl in 50 mM Tns-HCI (pH 7.0), 0.5 mM DTT Twoto 4-hr treatments with NaF/EDTA irreversibly inhibited endogenous phosphatase activity in synaptic fractions without significantly affecting CaMkinase II activity (see "Results").
Cal\ji-kinase II was puriied as described elsewhere (Kelly et al., 1984). Synapsin I and the catalytic subunit of CAMP-dependent protein kinase were generous gifts from Teresa Mc-Guinness and Paul Greengard (Rockefeller University), and Jim Maller (University of Colorado Health Sciences Center), respectively.
Assay of type 7 and type 2 protem phosphatase actrvit/es m subcellular fract/ons. Protern phosphatase assays were carried out as in lngebntsen et al (1983c), and the release of 32P, from %abeled substrate protetns was measured as previously described (Foulkes and Cohen, 1980). PrP-1 and PrP-2A are the only protein phosphatases having srgnrfrcant phosphorylase a phosphatase actrvity (Ingebntsen and Cohen, 1983b;lngebrrtsen et al., 1983b) They were therefore assayed using 32P-phosphorylase a (1 mg/ml) as a substrate in the presence and absence of rnhrbrtor-2 (100 units/assay; see Foulkes and Cohen, 1980, for defrnrtron of unrts) as prevrously described (Ingebntsen et al , 1983c). Assays were carried out in the presence of 1 mM free Mn2+ for reasons described elsewhere (Ingebntsen et al., 1983b). Fractions were assayed at the highest possible dilution (1:150 to 1:300) to prevent possrble interference by high and low molecular weight inhibitors of protein phosphatases (Stewart et al , 1983). PrP-2A activity was taken as the phosphorylase a phosphatase actrvrty In the presence of rnhrbrtor-2. PrP-1 activity IS the difference between the phosphorylase a phosphatase actrvrtres observed in the absence and presence of rnhrbrtor-2. One unit of PrP-1 or PrP-2A was the amount of actrvtty that released 1 nmol of P, from phosphorylase a in 1 min at 30°C.
PrP-2B and PrP-2C were assayed using 0.4 PM 32P-phosphorylase ktnase (containing approximately 0.9 mol of phosphate/m01 of each (Y-and psubunit) as substrate The activity of PrP-2B was measured in the presence of 1 mM free Mn2+ and 1 PM CaM. Actrvrty was taken as the difference between the activities observed in the absence and presence of 100 PM tnfluoperazrne (Ingebntsen et al., 198313). Mn2+ can substitute for Ca2+ rn the actrvatron of PrP-2B. Stewart et al (1983) found that during the latter stages of purification PrP-2B was converted to a form that had low specrfrc acttvrty rn the presence of Ca". The enzyme's actrvrty was restored when assays were carried out in the presence of Mn*+. In pilot studies we found that PrP-2B activity in cytosol~c fractions was 5-fold lower rn the presence of Ca*+ compared to Mn2+. Consequently, Mn*+ was used in assays to increase their sensitivity. The actrvrty of PrP-2C was assessed after treating fractions with NaF (50 mM) and EDTA (10 mM) to rnactrvate PrP-2A (Ingebntsen et al.. 1983b, c). PrP-2C activity was taken as the difference between phosphoryase krnase phosphatase activities in the presence of 10 mM Mg'+ plus 1 .O mM EGTA or In the presence of 1 .O mM EGTA alone. Assays for PrP-2B and PrP-2C contained inhrbrtor-2 (100 units) to block background activity due to PrP-1. One unit of PrP-2B or PrP-2C activity IS that amount that releases 1 nmol of P, from phosphorylase krnase In 1 mm at 30°C. The activity of PrP-28 was extrapolated to V,, using a K,,, value of 5.9 PM (Stewart et al., 1983). Western blots. Electrophoretrc transfer and rmmunostarning of nrtrocellulose sheets were carried out as previously described (Kelly et al., 1984).

Results
Dephosphorylation of SJ and SPM phosphoproteins by endogenous PrPs. As a first step toward investigating phosphatase-like actrvities In synaptic fractions, SPM and SJ proteins phosphorylated by endogenous CaM-kinase II were used as substrates. Endogenous phosphatase activity was assessed by measuring the removal of 32P-phosphate from 3"P-phosphoproteins in SJs. The phosphatase activity in SJs dephosphorylated both 50. and 60-kd CaM-kinase II polypeptides as well as four additional SJ phosphoproteins of 240, 207, 170, and 140 kd (Fig. 1, lanes c and d). In general, the rate and extent of dephosphorylation of these phosphoproteins were similar. SPM fractions contained the same prominent phosphoproteins observed in SJ fractions and displayed very similar properties of endogenous protein dephosphorylation (Fig. 1, lanes a and b), despite the observation that the specific activity of CaM-kinase II in SPMs toward endogenous proteins was approximately 20-fold lower than that in SJs (Kelly et al., 1985).
The phosphatase(s) in SJ fractions effectively dephosphorylated exogenous 32P-synapsin I; 80% of the radioactivity in 32P-synapsin I with and dephosphorylated soluble 32P-synapsin I and CaM-kinase II.
To confirm that the removal of 32P-phosphatefrom phosphorylated proteins was due to protein phosphatase activity and not proteolytic degradation, 32P-SJs were incubated at either 4°C or 37°C for varying lengths of time before terminating reactions by the addition of SDS followed by precipitation of, proteins with TCA. '32P-inorganic phosphate liberated in these reactions was quantjtatively separated from [32P]ATP and 32P-peptides by its association with molybdate and subsequent partitioning in a two-phase system (see "Materials and Methods"). TCA-insoluble "P-proteins from the same reactions were solubilized in SDS 'sample 'buffer, separated by SDS-PAGE, and analyzed by autoradiography.
Densitometric scanning of autoradiograms was used to quantitate the extent of dephosphorylation for individual SJ proteins. This analysis demonstrated removal of 32Pphosphate from all SJ, phogphopioteins and a'.concomitant appearance of 32P-phosphate in the supernatant; these two indices of protein dephosphorylation were inversely related but of equivalent magnitude (Fig. 2). Figure 2 shows the very similar rates of endogenous dephosphorylation for the .50--and 60.kd phosphoproteins of CaM-kinase II. The other major, phosphoproteins in ,SJs that are substrates for CaM-kinase II tiere dephosphorylated at rates similar to those observed for the 50-and 60-kd proteins (results not shown).
TX-sol SPM fractions displayed very similar temperature-dependent phosphatase activity when compared to SJs; the former activity was measured using 32P-SJs as substrate (results not shown). Control experiments demonstrated that the endogenous dephosphorylation of 32P-proteins was not affected by the concentration of TX present in the different assays.
Table I (top) shows the results from experiments using 32P-SJs (intact) to examine the action of endogenous protein phosphatases on the dephosphorylation of 60. and 50.kd phosphoproteins of CaMkinase II in the presence of a battery of protease inhibitors. When the inhibitors PMSF, leupeptin, pepstatin, and a-2-macroglobulin were included in reaction buffers, no effect on the rate or extent of endogenous dephosphorylation of 60-and 50-kd SJ proteins was observed. Likewise, the addition of TX- contained therein to accelerate the rate of dephosphotylation of 32P-SJ proteins was unaffected by the addition of protease inhibitors (Table I, bottom). In agreement with these findings, no changes in Coomassie blue or silver staining patterns of SJ proteins were observed following dephosphorylation reactions, except for the 50and 60.kd phosphoproteins of CaM-kinase II (results not shown). These bands became narrower and were shifted to slightly lower molecular weights upon dephosphorylation.
Previous studies have shown that the apparent molecular weights of the subunits of CaMkinase II phosphoproteins increase as a result of being phosphoryated in a Ca2+/CaM-dependent manner (Kelly et al., 1984;Shields et al., 1984).
When 35S-thiophosphorylated SJ proteins were used as substrates in experiments examining endogenous and/or exogeno'us PrP activities, a dramatic decrease in the extent of dephosphorylation was observed ( Fig. 1, lanes e and f). 35S-thiophosphorylated SJ proteins were refractory to phosphatase activity and retained greater than 96% of their radioactivity following 90.min dephosphorylation reactions at 37'C. This finding is in agreement with the reported resistance ot thiophosphorylated proteins to the action of protein phosphatases (Sherry et al., 1978;Hoar et al., 1979;Cassel and Glaser, 1982).
Endogenous phosphatase activities in synaptic fractions were irreversibly inactivated by incubating SJs in 50 mM NaF, 0.5 mM EDTA for 180 min at 4°C (Fig. 3, lanes b and c). This treatment is known to inactivate PrP-1 and PrP-2A (Ingebritsen et al., 1983a, b) and resulted in greater than 98% inactivation of phosphatase activity in synaptic fractions. Control experiments demonstrated that this treatment did not inactivate CaM-kinase II activity in SJs if care was taken to remove NaF and EDTA prior to phosphorylating SJ proteins by endogenous CaM-kinase II (Fig. 3, lane d). SJs prephosphory lated with [Y-~'P]ATP and then treated with NaF/EDTA to inactivate phosphatases contained phosphoproteins that were good substrates for phosphatases present in TX-sol SPM fractions (Fig. 3, lane h), or purified PrP-I and PrP-2A (see below).
Identity of endogenous PrPs. The distribution of type-l and type-2 PrPs in subcellular fractions from rat forebrain is presented in Table  II. The specific   activity of PrP-1, -2A, and -2B was 1.5. to 5-fold higher in synaptosol compared to cytosol fractions, whereas that of PrP-2C was similar in these two fractions. The specific activity of PrP-I was slightly lower (1.5-to 2-fold) in SPM and SJ *fractions compared to synaptosol. In contrast, the specific activity of the type-2 phosphatases was dramatically lower in SPM and SJ fractions compared to cytosol. Thus, the ratio of PrP-I to PrP-2A activities was approximately IO-fold higher in the SPM and SJ fractions compared to cytosoiic and synaptosolic fractions. Type-l and type-2 phosphatases were detected in TX-sol SPM fractions, the specific activity of the former being the highest. following incubations at 37°C under various conditions. Ten micrograms of %P-labeled SJs were used in each r&action. Reactions contained buffer alone, or the following additions: CaM (2 pg/lOO PI), CaC12 (2 mM), MgC12 (2 mM), or EGTA (5 mM). Dephosphorylation reactions were in 45 ~1 of buffer containing 50 mM Trls-HCI, 0.5 mM DTr (pH 7.0); some reactions also contalned the protease inhibitors PMSF (1.5 mM), leupeptin (10 pg/ml), pepstatin A (10 pg/ml), and cu2macroglobulin (25 pg/ml). TX-sol SPMs (10 pg of protein) were added to ,certain reactions as indicated.  32P-SJs (Fig. 3, lanes e and f;  were observed with either inhibitor-l or -2. Measurements of the extent of dephosphorylation at longer reaction times (greater than 45 min) and in the presence of inhibitors were less accurate than values at shorter times because non-PrP-I phosphatases remained active. These findings indicate that, of the total protein phosphatase activity in SPM or SJ fractions, PrP-1 constitutes the major activity. The remaining activity is probably due to PrP-2A (see below).
Experiments were carried out to determine whether PrP-2B or PrP-2C contributed to the endogenous protein phosphatase activity toward SJ proteins phosphorylated by endogenous CaM-kinase II. PrP-2B is completely dependent on Ca"+ or Mn'+ (see "Materials and Methods") for activity toward all known substrates, and its activity is stimulated 1 O-fold by CaM (Stewart et al., 1983). As shown in Table I, Ca'+ plus CaM did not stimulate PrP activity in SJ fractions; moreover, endogenous activity was not inhibited by EGTA. Mn2+ was found to stimulate slightly (5 to 10%) the dephosphorylation of 32P-proteins in SJs by endogenous PrP activity (Table I). However, the activity of PrP-I and PrP-2A using phosphorylase a as substrate was also stimulated by Mn2+ (Table II; see "Materials and Methods"). These results suggest that PrP-2B does not make a significant contribution to the endogenous dephosphorylation of SJ phosphoproteins. Further experiments (see Table I) indicated that PrP activity in SJs was not stimulated by Mg'*. Since PrP-2C is completely dependent on Mg'+ for activity (Pato et al., 1983), this enzyme does not contribute to the endogenous phosphatase activity in SJ fractions.
Additional experiments examined cytosolic fractions (100,000 X g supernatants) from brain homogenates and hypotonic lysates of synaptosomal fractions for protein phosphatase activity. Dephosphorylation reactions contained cytosolic proteins and 32P-SJ fractions phosphorylated by endogenous CaM-kinase II (32P-SJs were pretreated with NaF/EDTA prior to dephosphorylation reactions). 32Plabeled SJs incubated with cytosolic or synaptoplasmic fractions lost 19% and 44% of their 32P during 30 min at 37"C, respectively (results not shown). Synaptoplasmic fractions routinely contained 2to 4-fold more phosphatase activity per microgram of protein than did conventional cytosolic fractions.
Commercially available alkaline phosphatases from E. co/i, bovine intestine, and human placenta, as well as acid phosphatases from potato and human semen, were examined for their ability to dephosphorylate purified 32P-CaM-kinase II (see "Materials and Methods"). None of these phosphatases demonstrated detectable activity in removing protein-bound 32P-phosphate from CaM-kinase II. Likewise, these phosphatases were incapable of dephosphorylating 32P-phosphoproteins in intact SJs (results not shown). In contrast, when 32P-SJs were solubilized in 1% SDS and then incubated with these different phosphatases (after diluting SDS to 0.05%), only the alkaline phosphatase from bovine intestine (10 units/ml) displayed moderate activity in dephosphorylating 32P-SJ proteins; the rate of dephosphorylation in these reactions was half of that observed for endogenous phosphatases acting on 32P-proteins in "intact" SJs. These results show that SJ phosphoproteins are very poor substrates for a number of alkaline and acid phosphatases from non-neuronal tissues. Independent measurement of PrP-2B (calcineurin) in synaptic fractions. The very low specific activity of calcineurin in SJ fractions (Table II) was surprising in light of previous findings which demonstrated by immunohistochemical methods that calcineurin was present in postsynaptic densities in fixed neural tissues (Wood et al., 1980). Since it was possible that the fractionation procedures used to isolate SJs may have inactivated any calcineurin present in synaptic structures in situ, a Western blot analysis of synaptic fractions was undertaken using an affinity-purified antibody against calcineurin (kindly provided by Dr. Claude Klee, National Institutes of Health, Bethesda, MD). The amount of calcineurin in SJ fractions was 11 -fold lower than that in SPMs; the former contained 0.21 pg of calcineurin per milligram of protein (Fig. 4). Approximately 95% of the calcineurin in SPM fractions was recovered in the Tritonsoluble extract of SPMs. Calcineurin was detected in all subcellular fractions and was present in synaptosol-enriched fractions (L-S3) in the highest relative concentration (9.36 pg of calcineurin/mg of protein). The calcineurin content of forebrain homogenate and SPM fractions was equivalent (2.3 pg of calcineurin/mg of protein). In general, values for the relative concentration of calcineurin in subcellular fractions from Western blot analysis were similar to those from PrP-2B activity measurements (see Table II), with the exception that calcineurin activity values using SPM fractions were consistenUy lower than soluble fractions when compared to ratios based on antibody binding (Fig. 4). Results from '251-peptide mapping experiments have confirmed the near absence of calcineurin in SJ fractions (P. T. Kelly, unpublished observations).
Experiments using 32P-labeled SJs demonstrated that purified and active calcineurin (kindly provided by Dr. Claude Klee) displayed very little, if any, Ca'+/Mn'+-dependent and CaMstimulated activity in dephosphoryating the 240-, 207., 170., and 140.kd proteins phosphorylated by endogenous CaM-kinase II when added at concentrations similar to those used for PrP-1 (results not shown). Calcineurin did not dephosphorylate the autophosphorylated subunits of CaM-kinase II to any appreciable extent. These results agree with the observation that purified CaM-kinase II is dephosphotylated 50.fold less effectively by calcineurin when compared to equivalent amounts of PrP-2A and PrP-1 (A. Nairn, personal communication).

Discussion
Our results demonstrate the presence of protein phosphatase activity in synaptic fractions from rat forebrain.
The endogenous PrP activity in SJ and SPM fractions was shown to dephosphorylate most, if not all, endogenous proteins that are phosphorylated by endogenous CaM-kinase II. Removal of 3'P-phosphate from endogenously phosphorylated SJ proteins by proteolysis was ruled out by a TCA-molybdate phase-partitioning assay that specifically measures 32P, release, and the activity of SJ phosphatases was unaffected by a broad battery of protease inhibitors. Moreover, observations that the protein-staining patterns of SJs remained virtually unchanged following a variety of dephosphorylation conditions indicated that proteolytic release of 32P, was most unlikely. The endogenous phosphatase activity in SJ fractions displayed irreversible inactivation by NaF/EDTA, a property shared by PrP-1 and PrP-2A (Ingebritsen et al., 198313, c). Finally, 35S-thiophosphorylated proteins in SJs were virtually resistant to dephosphorylation by endogenous phosphatases, a characteristic typical of other known thiophosphorylated ABCDEFGHIJ Figure 4. Western blot analysis of subcellular fractions from rat brain and purified calcineurin. Fractions were separated by SDS-PAGE and transferred to nitrocellulose. The 50. to 70.kd regions of nitrocellulose sheets were then incubated in affinity-purified anti-calcineuiii;'(iatjbii'I~G, tipproitimately 0.8 pg/ri71) followed by "?-protein A (2 x 1 O6 cpm/ml, 2 &i/pg). lmmunoreactive bands were visualized by autoradiography and quantitated by scanning densitometry: substrate-phosphatase interactions (Sherry et al., 1978;Hoar et al., 1979;Cassel and Glaser, 1982). j The studies herein examined whether the endogenous PrP activity in SJ and SPM fractions resulted from the presence of one or more of the type-l or type-2 protein phosphatases. Using 32P-phosphoryase a as a substrate, PrP-1 and PrP-2A were found to be present In all synaptic fractions. Moreover, addition of homogeneous preparations of these phosphatases to 32P-SJ fractions revealed that PrP-1 and PrP-2A could completely dephosphorylate all phosphoproteins that are in vitro substrates for CaM-kinase II. The rates of dephosphorylation (in terms of percentage of release of "P) for the 50-, 60., and 140-kd proteins in SJs by PrP-1 and PrP-2A were 60% and 50%, respectively, of that observed using phosphorylase a as a substrate. When inhibitor-l and/or inhibitor-2 were added to dephosphorylatlon reactions at concentrations sufficient to totally inhibit PrP-1, the endogenous phosphatase activities in SJ and SPM fractions were greatly Inhibited. The inhibition observed in each synaptic fraction was consistent with the relative amounts of PrP-1 and PrP-2A activities present in each, based on measurements using 32Pphosphorylase a as a substrate.
Further experiments demonstrated that the specific activity of PrP-28 and PrP-2C was very low in SPM and SJ fractions and that these enzymes did not contribute significantly toward the dephosphorylation of SJ proteins (see "Results"). Taken together, these results show that the endogenous phosphatase activity in synaptic fractions is due primarily to PrP-1 with a small contribution by PrP-2A. Our findings indicated that, of the total phosphatase activity in SJs, approximately four-fifths of it was PrP-1 and one-fifth was PrP-2A, whereas the ratio for SPMs was 4:l. Whether or not PrP-1 preferentially regulates in viva dephosphorylation processes at synaptic junctions remains to be answered.
The near-absence of immunoreactive staining, on the one hand, and low specific activity of PrP-2B (calcineurin) in SJ fractions, on the other, is surprising In light of immunohistochemical findings which demonstrated that calcineurin is concentrated at postsynaptic densities in situ (Wood et al., 1980). Our results indicate that the fractionation procedures used to isolate SJs removes most of the calcineurin that appears to be present in postsynaptic structures in situ. Western blot analyses performed herein demonstrated that calcineurin in rat brain IS recovered primarily in cytosolic fractions. The highest concentration of calcineurin was observed in synaptosolenriched fractions which suggests that it may be an abundant component of presynaptic terminals. Alternatively, calcineurin is a major component of postsynaptic densities in situ and, due to its dissociation properties, redistributes into soluble extracts during tissue homogenization and subcellular fractionation. Our results indicate that purified calcineurin is virtually incapable of dephosphorylating the 240., 207-, 170-, and 140-kd proteins in SJ fractions that are phosphorylated by endogenous CaM-kinase II, or the auto-phosphorylated subunits of this kinase. King et al. (1984) have shown that purified calcineurin displayed high catalytic efficiency in dephosphorylating purified G-substrate, DARPP-32 and protein K.-F.; however, synapsin I was a much poorer substrate for this ph0sphataF.e.
The endpgenbus PrP activity in SJs was clearly distinct from that of a number of commercially available acid and alkaline phosphatases in that the latter were virtually incapable of dephosphorylating nondenatured SJ phosphoproteins that are substrates for CaMkinase II.
These studies demonstrated PrP activity in cytosolic extracts from brain that was capable of dephosphorylating SJ phosphoproteins. Previous studies have demonstrated the presence of PrP activities in brain extracts and crude particulate fractions (Forn and Greengard, 1978;Yang et al., 1982;Foulkes et al., 1983). Our sttidies demonstrated that both cytosolic and synaptoplasmic fractions contain PrP activities, with the activity in the latter being greatest. -This obs@vation suggests that phosphatase activity is enriched in synaptosolic compartments of neurons compared to cytosolic fractions from brain tissue (i.e., neuronal plus non-neuronal cytosol). Approximately 35% of the measured activity in cytosolic fractions displayed PrP-1 properties.
PrP-1 is highly enriched in synaptic particulate fractions. The ratio of PrP-1 to PrP-2A in the SJ and SPM fractions is much higher (IOfold) than in soluble fractions. The specific association of PrP-1 with SPM and SJ fractions suggests that it may play a functional role in regulating the state of phosphorylation of nerve terminal proteins. It is tempting to speculate about this role within the scheme of neuronal transmission and the regulation thereof. The in vitro activity of the SJ-associated CaM-kinase II is modulated by autophosphorylation at low and possibly physiological CaM concentrations (15 to 50 nM) (Shields et al., 1984). We have demonstrated that the autophosphorylation of CaM-kinase II increases its affinity for CaM and, therefore, phosphorylated CaM-kinase II displays greater activity at low CaM concentrations than does its unphosphorylated counterpart. PrP-I is preferentially enriched at synaptic junctions and dephosphorylates the autophosphorylated subunits of CaM-kinase II. Thus, the role of PrP-1 in dephosphorylating SJ proteins, especially CaM-kinase II, represents a logical modulatory mechanism to restore SJ phosphoproteins to their unphosphorylated states. Levine et al. (1985) have recently shown that the autophosphbrylation of CaM-kinase II in cytoskeletal preparations from br;ain increases the affinity of the kinase for '251-CaM when analyzed by the gel overlay method. However, the affinity of nondenatured cytoskeletal preparations for CaM at low Ca2+ concentrations (0.5 pM) decreases following autophosphorylation.
Several mechanisms for [egulating PrP-1 activity have been identified. One involves the phosphorylation of inhibitor-l on a specific threonine residue by CAMP-dependent protein kinase. The phospho-