Characteristics of an Inhibitor of the Na'/K' Pump in Human Cerebrospinal Fluid*

Human cerebrospinal fluid (CSF) inhibits the Na+/K+ pump in human red cells and the activity of purified Na+/K+-ATPase (Halperin, J. A., Shaeffer, R., Galvez, L., and Malave, S. (1985) Proc. Natl. Acad. Sci. U. S. A. 80, 6102-6104, 1983; Halperin, J. A., Martin, A. M., and Malave, S. (1985) Life Sci. 37, 561-566. We describe here some properties of the CSF inhibitor of the Na+/K+ pump. Active material was extracted from human CSF with 50% methanol and then concentrated and desalted by ultrafiltration. This extract inhibited, in a dose-dependent manner, the ouabain-sensitive influx of K+ into human red cells and the activity of purified Na+/K+-ATPase. Partial separation of the inhibitory activity was achieved by gel filtration and reverse-phase high performance liquid chromatography. Inhibition of both pump and enzyme was specific in that other red cell membrane transport systems or enzymes examined were not influenced by CSF ex- tracts. Dialysis and ultrafiltration experiments indicate that the molecular weight of the inhibitor is ~ 6 0 0 . The inhibitory activity is sensitive to proteolytic enzymes indicating that the inhibitor might be a small peptide. In the presence of CSF extract the for external K+

8 To whom correspondence should be addressed. pump as well as the activity of purified Na+/K'-ATPase. However, the identity of these inhibitors still remains unknown.
It was reported previously that crude human CSF' inhibits the activity of the Na'/K' pump in human red blood cells and of a purified Na+/K+-ATPase (9). Moreover, this inhibitory effect of crude human CSF increases upon acute expansion of the extracellular fluid volume (10). Additionally, Lichtstein and colleagues (11) showed that human CSF inhibits the binding of [3H]ouabain to rat brain synaptosomes.
The work reported in this paper further characterizes the inhibitor of the Na'/K' pump present in human CSF. It has a molecular weight of about 600 and is sensitive to proteolytic digestion, suggesting that it is a small peptide. Furthermore, it is a specific inhibitor of the Na'/K' pump and of the Na+/ K'-ATPase and behaves as a partial competitive inhibitor of the activation of the Na'/K' pump by external K' .

Extraction of Na+/K+ Pump Inhibitor from Human CSF
Human CSF was obtained as discarded material collected in the course of either lumbar myelograms (9) or continuous intracerebroventricular perfusion procedures. Samples of human CSF were lyophilized and then reconstituted to their original volume in 50% (v/v) methanol in water. After 30 min a t 4 "C the insoluble material was removed by centrifugation for 45 min a t 60,000 X g in an ultracentrifuge (Beckman Instruments). The supernatant was concentrated by rotary evaporation, and the resultant aqueous phase was either lyophilized or desalted by ultrafiltration using an Amicon model 2000 Diaflo ultrafiltration cell and a Spectra/Por type C membrane of 500 M, cut-off (Spectrum Medical Industries, Los Angeles, CA).

Gel Filtration HPLC
An aliquot of the methanol extract was injected onto a Superose 12 column (Pharmacia LKB Biotechnology Inc.) connected to a Waters Associates high performance liquid chromatography (HPLC) system. The column was equilibrated and eluted with buffer containing: 130 mM NaCl, 2.8 mM KCl, 1 mM MgCl,, 10 mM glucose, and 10 mM MOPS adjusted to pH 7.4 with Tris base. The flow rate was 0.6 ml/min. The absorbance was monitored at 214 nm. Fractions of 1.2 ml were collected and assayed for biological activity.

High Performance Liquid Chromatography
Aliquots of the methanol extract of human CSF were injected to an octadecylsilane (C-18) column (5 X 100 mm; 3-pm particle size, IBM Instruments Inc., Danbury, CT) equilibrated with 0.1% trifluoroacetic acid. The column was eluted with a linear concentration gradient of 70% acetonitrile in 0.1% trifluoroacetic acid at a flow rate The abbreviations used are: CSF, cerebrospinal fluid; HPLC, high performance liquid chromatography; EGTA, [ethylenebis-(oxyethylenenitrilo)]tetraacetic acid; MOPS, 3-(N-morpholino) propanesulfonic acid; concentration of external K+ required to half-maximally activate the Na+/K+pump; K'", apparent dissociation constant of the external K+ pump site complex; V,,,.,, influx a t nonlimiting external K' concentration. of 1 ml/min. Absorbance was monitored a t 214 nm. Fractions of 2 ml

RESULTS
were collected and lyophilized. The dry material was reconstituted with the corresponding assay buffer and assayed for biological activity.

Protease Sensitivity
One-ml aliquots of the methanol extract concentrated and desalted by ultrafiltration were incubated for 2 h at 37 "C with one of the following peptidases: carboxypeptidase Y (100 pg/ml; Cooper Biomedical, Malvern, PA), Pronase (1 mg/ml; Sigma), and thermolysin (10 pg/ml; Sigma). The reaction was stopped by boiling for 3 min. After centrifugation for 45 min at 35,000 X g, the supernatants were assayed for biological activity.

Na'/K+ Pump and Enzyme Inhibition Assays
The capacity to inhibit the Na'/K' pump in human red cells was used as the principal assay to follow the activity through the isolation procedures. Some samples were also tested for their ability to inhibit the activity of a purified Na'/K' ATPase.

Na+/K+ Pump in Human Red Cells
The activity of the Na'/K' pump in human red cells was estimated by measuring ouabain-sensitive K' influx, using ffiRb as a tracer, by a slight modification of the method previously described (9). In order to minimize interassay variability, red cells used in all experiments described in this work were obtained from a single donor. Washed packed red cells were suspended to a final hematocrit of -5% in prewarmed medium (37 "C) containing 130 mM NaC1, 2.8 mM KC1, 1 mM MgCI,, 10 mM glucose, 10 mM Tris-MOPS, pH 7.4, at 37 "c, and -3 pCi/ml =Rb (specific activity, 5 mCi/mg; Amersham Corp.), with and without ouabain (0.1 mM). The radioactivity in three aliquots of influx medium was measured for determination of the initial specific activity. After 30 min at 37 "C the cells were washed 3 times with cold isotonic choline chloride solution, and the cell pellet was lysed with 1 ml of 0.02% Acationox (American Scientific Products, McGraw Park, IL). The radioactivity incorporated into the cells was measured by counting aliquots of the lysate in a y counter (Auto-Gamma 500, Packard Instruments, Lynn, MA), and the amount of cells present in each sample was calculated from the concentration of hemoglobin (A611) in the lysate and the hematocrit and hemoglobin concentration of the original cell suspension. The influx in pmollliter red blood cells/h was calculated as described previously (12) ATPase Activity Coupled Enzymatic Assay-The activity of a very highly purified Na'/K+-ATPase extracted from the salt gland of the dogfish (provided by Dr. J. C. Skou, University of Aarhus, Denmark) was measured using the coupled enzymatic assay, exactly as described (13). The enzyme was preincubated for 20 min a t a concentration of 25 pg/ml in either 140 mM NaCl, 1 mM MgCI,, 1 mM EGTA, 10 mM Tris-MOPS (pH 7.4 a t 37 "C), or in CSF extract desalted by ultrafiltration and then reconstituted in the same buffer. The assay was initiated by the rapid addition of 10 pl of the ATPase suspension to 1 ml of reaction mixture. Oxidation of NADH, which in this assay is equimolar with the hydrolysis of ATP, was continuously monitored spectrophotometrically a t 340 nm.

Na',K' and Protein Determination
Na' and K' concentrations were determined by atomic absorption spectrometry, as described (12). Protein concentration was determined by the colorimetric method of Bradford (15).
Partial Purification of the CSF Inhibitor-In order to purify the CSF inhibitor, a methanol extract was prepared from 1 liter of lyophilized human CSF, as described under "Experimental Procedures." After evaporation of the organic solvent, the aqueous residue was desalted and concentrated ~50-fold by ultrafiltration through a 500 M , cut-off membrane. Ultrafiltration was allowed to proceed until the Na' and K' content of the concentrated preparation was reduced to less than 100 FM, as measured by atomic absorption spectrometry. Aliquots of the concentrated extract were then adjusted to the ionic and pH conditions of the assays for inhibitory activity. The activity of both the Na'/K' pump in human red cells (Fig. 1) and purified Na'/K' ATPase ( Fig. 2) was inhibited in a dosedependent manner by the methanol extract of human CSF. When the extract was concentrated 50-fold the inhibition reached ~9 5 % .
Inhibition of the purified Na'/K' ATPase was time-dependent and reached maximal inhibitory effect after 20 min of preincubation of the enzyme with the CSF extract ( Fig. 2, lower inset).
Estimation of the Apparent Molecular Weight of the Znhibitor-The inhibitory activity passes through a 1000 M , cutoff dialysis membrane and is lost when the extract is filtered through a 1000, but not a 500, M , cut-off ultrafiltration membrane. The apparent molecular weight of the inhibitory substance was also estimated by gel filtration on a Superose 12 column as described under "Experimental Procedures." The CSF inhibitory activity elutes a t =34 min. Relative to several other molecules of known molecular weight, this retention time corresponds to an apparent molecular weight of -600, a result consistent with the dialysis and ultrafiltration findings.
Separation by Reverse Phase HPLC-Aliquots of the methanol extract of human CSF were injected onto a C-18 reverse phase HPLC column equilibrated with 0.1% trifluoroacetic 10001 -.

FIG. 1. Effect of human CSF extract on the ouabain-sensitive K+ influx into human red cells.
A methanol extract of human CSF was prepared and desalted by ultrafiltration as described under "Experimental Procedures." The influx of K' into fresh human red trated CSF extract. ffiRb' was used as the tracer. Final assay condi-cells was measured in the presence of increasing amounts of concentions were: 130 mM NaCl, 2.8 mM KC1, 1 mM MgCl,, 10 mM glucose, 10 mM Tris-MOPS, pH 7.4, a t 37 "C, and -3 pCi/ml ffiRb'. The hematocrit of the cell suspension was -5%. The incubation time was 30 min. The ouabain-sensitive influx of K' at each amount of CSF concentrate was calculated by the difference of the influx measured in the absence and presence of 0.1 mM ouabain. The amount of CSF concentrate is expressed as the logarithm of the volume (pl) of original CSF present in 0.5 ml of assay medium. Inset, the inset shows the percent inhibition of the ouabain-sensitive K' influx as a function of the amount of CSF concentrate in the medium. Effect of human CSF extract on the activity of a highly purified Na+/K+-ATPase. The activity of a purified Na'/ K+-ATPase extracted from the salt gland of the dogfish (cf. "Experimental Procedures" ) was measured by a coupled enzymatic assay. The enzyme was preincubated at 37 "C for 20 min at a concentration of 25 eg/ml either in buffer or in different concentrations (2.5, 6, and 50 times) of CSF extract desalted and concentrated by ultrafiltration (cf. "Experimental Procedures" ); enzyme activity was then determined spectrophotometrically. The specific activity of the purified enzyme preincubated in buffer without CSF was -1300 pmol/mg protein/h. The enzyme was 100% inhibitable by ouabain. Upper inset, percent inhibition of the Na+/K+-ATPase as a function of the volume of original CSF present in 0.1 ml of preincubation medium. Preincubation time was 20 min. Lower inset, fraction of enzyme inhibited after preincubation at 37 "C in 50-fold concentrated CSF extract for different time intervals. Fraction of inactivated enzyme was calculated as 1 -(specific activity in CSF/specific activity in control buffer). acid, as described under "Experimental Procedures." When the pH of the extract was =8, the inhibitory activity was not retained by the column and was recovered in the effluent breakthrough volume of the column between 3 and 5 min after injection. Active material obtained from this first passage through the C-18 column was concentrated by lyophilization, acidified to pH 2.8, and rechromatographed in the same system. Under these conditions, the inhibitory activity did not elute in the breakthrough volume and, presumably, was bound to the C-18 column. The column was then eluted as described under "Experimental Procedures," and fractions were collected every 2 min. After lyophilization and reconstitution in the appropriate buffer all fractions were tested for biological activity. As shown in Fig. 3, only one fraction, which eluted at 42-46 min, inhibited Na'/K' pump activity in red cells. Depending on the volume and concentration of the extract injected to the column, the inhibition of the Na'/ K' pump ranged from 22 to 65% ( n = 4), but the yield of active material recovered was =1% of the amount injected, a value too small to allow for further purification. Table  I summarizes the purification scheme described. Chemical Nature of the Inhibitor-An aliquot of the methanol extract was incubated for 2 h in the presence of either thermolysin, pronase, or carboxypeptidase Y, as described under "Experimental Procedures." Carboxypeptidase Y and Pronase significantly reduce and thermolysin completely suppresses the inhibition of the Na+/K' pump produced by the CSF methanol extract (Fig. 4). Thus, the inhibitory activity appears to be associated with a peptide of low molecular weight.
Specificity of the CSF Inhibitor-In order to determine the specificity of the CSF inhibitory activity, two types of experiments were performed. First, the specificity of the inhibition of the Na'/K' pump in the transport assay was determined by measuring the total and ouabain-resistant components of Retention Time ( m i d  FIG. 3. Separation of the CSF inhibitor by reverse phase HPLC. An aliquot of the methanol extract of human CSF was injected at pH 8 into a reverse phase HPLC column as described under "Experimental Procedures." The breakthrough volume was collected, lyophilized, and reconstituted at pH 2.8. An aliquot of the acidified materia1 was reinjected into the same column and eluted with a linear gradient of 70% acetonitrile in 0.1% trifluoroacetic acid. The flow rate was 1 mI/min. Fractions were collected every 2 min, lyophilized, and reconstituted in K' influx buffer to measure the K+ influx into human red cells as described under "Experimental Procedures."

TABLE I Purification of the CSF inhibitor of the Na+/K+ pump
The inhibitory activity of the Na+/K+ pump was determined by the uDtake of =Rb as described under "ExDerimental Procedures." "Protein was estimated by the dye binding method of Bradford 1 unit is the amount of active material in 0.5 ml of assay medium that induces 50% inhibition of the Na+/K+ pump. BTV, the breakthrough volume of the reverse phase C-18 column (cf. "Results").
HPLC fraction 23, the active fraction obtained by reverse phase HPLC using a C-18 column as described under "Experimental Procedures''; retention time, 44-46 min. e Protein as estimated by amino acid analysis.
K' influx in the presence or absence of CSF extract. Then, the influx specifically mediated through the Na'/K' pump, which is the ouabain-sensitive component of the influx, was calculated by subtracting the ouabain-resistant from the total influx. The extract of human CSF inhibited only the ouabainsensitive fraction of the influx, indicating specific inhibition of the Na'/K' pump (Table 11). Second, the specificity of the inhibition for the Na'/K+-ATPase was examined using both Na'/K'-and Ca2'-dependent ATPases from human red cell membranes (cf. "Experimental Procedures" ). A 50-fold concentrated extract of human CSF completely inhibited the Na'/K'-ATPase activity, as determined by the ouabain-sensitive component of the ATPase activity measured in the presence of Na+, K', and Mg". In contrast, the CSF extract had no influence on the Ca*'-ATPase activity present in the same preparation, measured by the hydrolysis of ATP in the presence of ouabain and either CaCl, or EGTA (Table 11). Hence, the CSF inhibitor specifically suppresses the activity of the Na'/K' pump and its enzymatic machinery, the Na'/ K'-ATPase.  FIG. 4. Effect of different proteolytic enzymes on the CSF inhibitory activity of the human red cell Na+/K+ pump. The activity of the pump was determined by the ouabain-sensitive (OS) K' influx as described under "Experimental Procedures." A methanol extract of human CSF was prepared, and aliquots were incubated at 37 "C for 2 h with carboxypeptidase Y (100 pg/ml), Pronase (1 mg/ ml), and thermolysin (10 pg/ml). The reaction was stopped by boiling for 3 min and by centrifugation at 35,000 X g. The percent inhibition of influx was calculated as 100 -(ouabain-sensitive influx in CSF/ ouabain-sensitive influx in control buffer) X 100. Influence of the proteolytic enzymes on K' influx was controlled by parallel incubations of control buffer with the proteolytic enzymes in the absence of CSF.

TABLE I1
SDecificity of the CSF inhibition  of digitalis glycosides with the Na'/K'-ATPase is antagonized by K' ions (16). In order to evaluate if external K' also antagonizes the effect of t h e CSF inhibitor, the activity of the Na'/K' pump was estimated by the ouabain-sensitive uptake of K' measured in the presence and absence of CSF extract at different concentrations of extracellular K' . In the absence of CSF, external K' normally stimulated the ouabain-sensitive K' influx into fresh human red cells (Fig. 5, upper inset) (17) with a = 1.3 mM (K'" = 0.6 mM) a n d a v,,, = 1.13 mmol/liter cell/h (Hanes plot in the lower inset, Fig. 5). In the presence of t h e CSF extract the for external K' increased to 3.1 mM (K'" = 1.3 mM), and the V,,, decreased slightly to 1.03 mmol/liter cells/h. These results suggest that the inhibitor of t h e Na'/K' pump present in human CSF partially competes with the stimulation of the pump by extracellular K' .

DISCUSSION
Previous studies have shown that crude human CSF inhibits the Na+/K' pump in human red cells and the activity of a purified Na'/K'-ATPase (9,lO). These results were in accord with earlier observations implicating the brain as a possible source of an endogenous inhibitor of the Na+/K' pump (8,18). This report describes subsequent efforts to characterize a n d purify the inhibitory substance present in human CSF.
Since the ouabain-sensitive uptake of K' by erythrocytes, determined using "jRb' as tracer, is a sensitive and well characterized indicator of the activity of the Na'/K' pump, this transport assay has been used to follow the inhibitory activity throughout the steps of purification. It is worth noting that red cells from the same donor were used for all pump assays, thereby reducing the interassay variability of the ouabain-sensitive K' influx to less than 1%.
Inhibition of the Na'/K' pump in human red cells by extracts of human CSF is dose-dependent ( Fig. 1) and becomes total when the CSF extract is desalted and concentrated 50-fold. Since the transport process mediated by the Na'/K+ pump is catalyzed by the enzyme Na'/K'-ATPase, it was important to know whether this inhibition also affected the Na'/K' ATPase. Two highly purified Na+/K+-ATPases extracted from different sources (from canine kidney: Na'/ K' ATPase from Sigma, cf. Ref. 9; and from the salt gland of the dogfish: cf. "Experimental Procedures" ) were both inhibited by the methanol extract of human CSF. Inhibition of t h e purified enzyme was also dose-dependent (Fig. 2), a n d t h e 50fold concentrated extract produced 100% inhibition of t h e enzymatic activity. Metal ions such as calcium and vanadate can influence the activity of the Na'/K'-ATPase (19), but these ions do not seem to be responsible for the inhibition seen here. The concentrated extract used in these studies had been previously desalted by ultrafiltration, and EGTA up to 1 mM did not affect the inhibitory activity. In an earlier report it was noted that vanadate is not detectable in human CSF samples (9). The inhibition of the Na'/K'-ATPase by CSF extract is time-dependent (Fig. 2, lower inset) and follows a time course similar to that for inhibition of the enzyme by ouabain (20), suggesting that in both cases the inhibition is "pseudoirreversible" (20). Furthermore, the inhibitor in human CSF is specific for the Na'/K' pump, again similar to ouabain. This is indicated by the data listed in Table I1 (see also Fig. 5) in which pump activity is measured as the ouabain-sensitive component of K' influx. Since human CSF exclusively suppresses this component, only K' transport mediated through the Na'/K' pump is affected by the inhibitor. Moreover, the CSF inhibitor specifically suppresses the activity of the Na'/ K'-ATPase extracted from red cell membranes and does not affect the Ca2'-ATPase activity present in the same preparation.
The results of dialysis, ultrafiltration, and size exclusion HPLC experiments indicate that the inhibitor has an apparent molecular weight of about 600. It is heat-stable but sensitive to proteolytic digestion (Fig. 4), suggesting that it is a small peptide, conceivably a neuropeptide. A similar low molecular weight protease-sensitive Na'/K'-ATPase inhibitor has been partially purified from cultured rat hypothalamic cells by Morgan et al. (21).
Previous attempts to isolate inhibitors of the Na'/K' pump and/or Na'/K'-ATPase from different sources have proved to be unreliable because materials generated during the extraction procedures could interfere with the different assays used to detect activity (22,23). Recent reports showed that fractions extracted from human and hog plasma exhibit Na'/ K+-ATPase inhibitory activity as well as [3H]ouabain and antidigoxin antibody-displacing activities (24,25). These fractions contained unsaturated fatty acids and lysophospholipids which were previously known to have Na+/K'-ATPase inhibitory and ouabain-displacing activities (26,27). The high concentration of these compounds required to inhibit Na+/ K'-ATPase (25,27) along with the lack of evidence supporting the specificity of their inhibitory action cast doubt on the significance of these lipids as putative regulators of the Na+/ K' pump. Since the concentration of lipids in CSF is extremely low (28), it is unlikely that lipids are involved in the inhibition of the Na'/K' pump produced by CSF.
The concentration of K' in CSF, 2.8 mM, is extremely stable and significantly lower than in plasma which is about 4.5 mM (28). It has been suggested that the Na'/K' pump is involved in maintaining a constant K' concentration in CSF (29,30). K' efflux from the ventriculocisternal system occurs by a concentration-dependent transport mechanism, inhibitable by ouabain (29), and with a sigmoidal relationship between K' concentration and K' removal (30). However, the apparent K0.5 for extracellular K' stimulation of K' efflux from the CSF compartment is =4 mM (30, 31), a value much higher than the for external K+ usually found in the pumps of vertebrate cells (31). This would he explained if a competitive inhibitor of the pump is present in CSF. The experiments presented here suggest that a CSF factor increases the of the Na'/K' pump for external K' , at least in human red cells (Fig. 5). If this factor has a similar effect in cells surrounding the ventriculocisternal system, it would increase the Kn.5 for external K' stimulation of their pumps to a value closer to the concentration of K' in CSF.
The volume flow of CSF is controlled in part by the activity of the Na'jK' pump of the choroidal epithelial cells and is inhibited by ouabain (29). It is tempting to speculate that an endogenous regulator of the Na'/K' pump present in human CSF might be one factor controlling the secretion of CSF by the choroidal plexus. Furthermore, the production of CSF is reduced by increasing the osmolarity of the serum (32). This would be in accord with previous observations that acute expansion of the extracellular cell volume with sodium increases the Na'/K' pump inhibitory activity of human CSF (10).
In summary, it is possible that a small peptide present in human CSF might regulate the Na'/K' pump in cells of the central nervous system and thereby contribute to control both production and composition of the cerebrospinal fluid. From the amount of CSF required to inhibit 50% the Na'/K' pump in human red cells (Fig. 1, inset), the number of pumps/cell (33), the estimated molecular weight of the inhibitor (~6 0 0 , cf. "Results" ), and an assumed 1:l stoichiometry of inhibitor binding to the pump, the concentration of the putative regulator of the pump can be calculated to be =50 pg/ml. This value is within the concentration range of most peptides with modulator-neurotransmitter function already detected in CSF (reviewed in Ref. 34).
It has been proposed that an endogenous inhibitor of active Na' transport might be a natriuretic hormone, different than the atrial natriuretic peptides recently purified and synthesized (reviewed in Ref. 35). It has also been proposed that such an inhibitor of active Na' transport might be involved in the pathogenesis of essential hypertension (5). Once the structure of the inhibitor of the Na'/K+ pump present in human CSF has been established, it will be of great interest to determine whether it is also present in the circulation.