Atrioactivase , a Specific Peptidase in Bovine Atria for the Processing of Pro-atrial Natriuretic Factor PURIFICATION

A seryl protease which catalyzes conversion of proatrial natriuretic factor (ANF) to the active circulating form, ANF(ss-lzs,, was purified from a particulate fraction of bovine atria. The enzyme was solubilized with 1.6 M KCl. The molecular mass of the purified enzyme was 580 kDa on gel filtration, whereas by sodium dodecyl sulfate-polyacrylamide gel electrophoresis a cluster of six bands with molecular masses around 30 kDa was observed. The purified enzyme produced ANF~ss-lzs) from partially purified bovine pro-ANF by the selective cleavage of the arginyl peptide bond in the -Pros7-Argss-Serss-sequence in pro-ANF. The enzyme was localized mainly in the microsomal fraction rather than the granule fraction. It is likely that the enzyme selectively cleaves the  Argss-Serss  peptide bond in pro-ANF during the process of secretion.

avoiding the problems that arise when prohormone and active hormone are mixed in storage granules and in blood, the distinct localization of pro-ANF and active ANF in different spaces facilitates identification of the activating enzyme.
Proteases (or a protease) in serum were found to produce ANF(w-126) from pro-ANF (6,7). However, further work indicated that it is unlikely that serum enzymes are involved because plasma has no processing activity (8). It was further shown that conversion does not take place in blood after secretion because addition of protease inhibitors to the perfusing buffer had no effect on the appearance of ANF(as-1~6) from a rat heart Langendorf perfusion (9).
Recently we obtained evidence for the presence of a pro-ANF processing enzyme in rat atria (10). To facilitate detection of the enzyme we used Boc-Ala-Gly-Pro-Arg-MCA (AGPR-MCA) which contains the sequence on the aminoterminal side of the peptide bond cleaved by the processing enzyme. The enzyme was found to be bound to the microsomal membrane fraction of rat atrial extract and was solubilized by 1.6 M KC1 solution. The enzyme possesses general properties compatible with a seryl protease and produces 28residue ANF by selectively cleaving the Arg9'-Serw bond rather than other arginyl peptide bonds such as those in Arg'01-Arg'02-Ser'03. Although the complete amino acid sequence of bovine pro-ANF is not yet known, we found that the rat atrial enzyme substrate could be used for the determination of bovine enzyme activity. In the present study we purified the enzyme from bovine atria and report its properties.
Measurement of Enzymatic Activity Activity of the enzyme was measured fluorometrically using AGPR-MCA as substrate (10). To a cuvette containing 2 ml of 50 mM Tris-HC1, pH 8.0, and 0.2 ml of 1 mM substrate which had been dissolved in 1 mM HCl was added 50 pl of an enzyme fraction. The reaction was followed at room temperature by determining the increase in fluorescence emitted at 460 nm and excited at 380 nm in a Perkin-Elmer model 650-15 spectrofluorometer. One unit of the enzymatic activity was defined as the amount of enzyme which releases 1 fimol of 7-amino-4-methylcoumarin from the substrate/min. 9515 Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis Polyacrylamide gel electrophoresis with sodium dodecyl sulfate was performed using 10% acrylamide gel (11). Proteins were stained by the silver staining method according to Wray et al. (12). A molecular weight standard kit from Sigma was used for calibration.
For fluorography of [3H]DFP-labeled enzyme, the gel was soaked in an Enlightning (Du Pont-New England Nuclear) solution for 30 min before drying and exposed to Kodak XAR-5 film for 3 days at -70 "C.
Purification of the Enzyme Enzyme Extraction-Ten bovine atria (360 g) freshly obtained from a local slaughterhouse were minced and homogenized in a Polytron homogenizer for 30 s in 2 liters of 1 mM HEPES, pH 7.0, containing 0.25 M sucrose and 1 mM EDTA, followed by centrifugation for l h at 10,000 X g. The sediment was homogenized again and centrifuged under the same conditions in order to remove cytosolic proteins. The resulting sediment was homogenized in a Polytron homogenizer for 30 s in 2 liters of a high ionic concentration buffer, 10 mM HEPES, pH 7.0, containing 0.25 M sucrose, 1 mM EDTA, and 1.6 M KC1 to release the enzyme from the membrane fraction. The homogenate was centrifuged for 1 h at 10,000 x g, and the supernatant was dialyzed against 10 mM HEPES, pH 7.4, containing 0.1 M NaCl, 1 mM EDTA (buffer A).
Chromatography on Heparin-Agarose-The dialysate was applied to a heparin-agarose column (2.7 X 9 cm) equilibrated with buffer A and subsequently washed with the buffer. The enzyme was eluted with a linear gradient from 0.1 to 1.5 M NaCl in buffer A. Fractions containing activity were collected and dialyzed against buffer A containing heparin at 10 pg/ml (buffer B).
Chromatography on Arginine-Agarose-The eluate from heparinagarose was applied to an arginine-agarose column (1.3 X 4 cm) equilibrated with buffer B, and the column was eluted by a linear gradient of NaCl from 0.1 to 1.0 M. Fractions containing the enzyme (total 10 ml) were directly concentrated in a Centricon-30 cartridge (Amicon) to 1 ml. The recovery was nearly 90%.
Gel Filtration on Sephacryl S-300-The concentrate was applied to a Sephacryl S-300 column (1.8 X 100 cm) equilibrated with buffer B. The active fraction eluted just after the void fraction.
Chromatography on Aprotinin-Agarose-The eluate from Sepha-cry1 S-300 was applied to an aprotinin-agarose column (1.3 X 4 cm) equilibrated with buffer B and eluted with a linear gradient of NaCl from 0.1 to 1.0 M.
Rechromatography on Arginine-Agarose-This step was performed essentially by the same method as the previous step except that elution was with a gradient of NaCl from 0.1 M to 0.5 M.

Determination of pH Optimum
The pH optimum for the enzyme was determined by measuring the activity in buffers of different pH. The buffers were all 50 mM in concentration and were citrate-phosphate for pH 5.0-7.0, Tris for pH 7.0-9.0, and carbonate for above pH 9.0.

Radioimmunoassay
Radioimmunoassay was performed by the method for human ANF described previously (13). Since bovine ANF(99126) has the same sequence as human ANF(w-126) (14), human ANF(w-126) was used as standard. The amount of ANF was expressed as equivalents of human ANF(w-Lz6). For determination of the amount of bovine pro-ANF, a factor of 0.4 was used for immunocross-reactivity.
Subfractionation of Atrial Homogenate tially by the method of DeBold and Bencosme (15). Briefly, 60 g of Subcellular fractions of bovine atrial tissues were isolated essenbovine atrial tissue was minced and homogenized using a Polytron homogenizer for 5 s in 10 mM HEPES, pH 7.4, containing 0.25 M sucrose and 1 mM EDTA. The suspension was further homogenized by 5 strokes of a Teflon homogenizer. The resultant homogenate was subfractionated by centrifugation, and the fractions collected were: nuclear fraction (1,900 X g for 10 min), mitochondrial fraction (second centrifugation at 1,900 X g for 10 min), crude granule fraction (32,000 x g for 10 min), and microsomal fraction (145,000 X g for 90 rnin). Each fraction was resuspended in the buffer used for homogenization.
The amount of immunoreactive ANF was measured by radioimmunoassay after boiling the samples in 1 N acetic acid. Alkaline phosphatase activity was measured by the method of Ray (16). For measurement of general protease activities, fractions were mixed with 5 volumes of 2.0 M KC1, 10 mM HEPES, pH 7.4, containing 1 mM EDTA. After incubation for 15 min at 22 "C the mixtures were centrifuged at 145,000 X g for 90 min. The resultant supernatants were used for the measurements.

Cleavage of Bovine Pro-ANF by the Purified Enzyme
To a mixture of 0.4 ml of 50 mM Tris buffer, pH 8.0, containing 0.1% bovine serum albumin and 0.1 ml of pro-ANF solution (1.7 nmol eq of immunoreactive ANF(*-Ix)) was added 0.1 ml of the enzyme solution (1 pg/ml) or buffer B (control) followed by incubation for 1 h at 37 "C. After adding 0.4 ml of 1 N acetic acid, the mixture was boiled for 15 min to stop the reaction and centrifuged to obtain a clear supernatant, which was lyophilized and reconstituted in 0.5 ml of 0.1% trifluoroacetic acid in preparation for HPLC.

Characterization of ANF Generated by the Atrial Enzyme
High performance gel filtration was performed on a Bio-Rad TSK-125 column (30 X 7.5 mm) using a solvent system consisting of 30% acetonitrile, 0.1% trifluoroacetic acid, and 0.2 M NaCl (17). Fifty p1 of the atrial enzyme-treated pro-ANF or control pro-ANF were applied. A flow rate of 0.5 ml/min was used, and the eluate was collected in 0.25-ml fractions.
Reverse-phase HPLC was run on a Vydac Cl8 column (0.46 X 25 cm, Altech) eluted isocratically in 22% acetonitrile and 0.1% trifluoroacetic acid. Fifty pl of sample or control pro-ANF were applied, and the eluate was collected in 0.25-ml fractions. The fractions from both runs were evaporated in a Speed Vac concentrator (Savant) and redissolved in the radioimmunoassay buffer (13).

Purification of ANF Generated from Pro-ANF by Enzyme Treatment
Immunoreactive ANF(m-126) fractions obtained by reaction of the purified enzyme with partially purified pro-ANF by reverse-phase HPLC (on Vydac C, ) were collected and further purified using a combination of a Zorbax CN column (0.46 x 15 cm, Du Pont-New England Nuclear) and the Vydac Cla column by the method described previously (18).

Amino Acid Sequence Analysis
Amino acid sequence analysis was performed on an automated gasphase instrument (model 470A, Applied Biosystems) by the method described by the manufacturer.

Preparation of Bovine Pro-ANF
Partially purified bovine pro-ANF was prepared by the method of Trippodo et al. (19). Bovine atrium was boiled in 1 N acetic acid for 15 min followed by homogenization for 15 s using a Polytron homogenizer. After centrifugation to obtain a clear supernatant, the crude extract was applied to a Sephadex G-100 column (1.0 X 90 cm). The pro-ANF peak fractions identified by radioimmunoassay and elution position were pooled. The concentration of immunoreactive pro-ANF was estimated by radioimmunoassay.
Estimation of the Molecular Mass of the Enzyme The purified enzyme was applied to a TSK G 3000 SW column (0.75 X 60 cm) pre-equilibrated and eluted in a solution of 50 mM phosphate buffer, pH 6.

RESULTS
The purified enzyme (100 ng) was treated with N-glycosidase F Purification of the Pro-ANF Converting Enzyme Atrioactiuase- Table I summarizes the overall purification of the enzyme. Activity in the crude homogenate was readily lost in a high ionic concentration solution. The low recovery from "One unit is defined as the enzyme activity which hydrolyzes 1 pmol of Boc-Ala-Gly-Pro-Arg-MCA/min. the solubilization step and heparin-agarose chromatography may be due to instability of the enzyme. We found that heparin stabilizes the activity in a low ionic concentration buffer. Therefore, we added heparin to all the buffer systems after the heparin-agarose step. Without heparin activity was lost during purification.
Estimation of Molecular Mass- Fig. 1 shows the SDS-polyacrylamide gel electrophoresis pattern of the purified enzyme. As shown in lane A the final product consisted of three doublet bands with approximate masses of 30 kDa. The apparent masses of these pairs were estimated as 31.5 and 31.0 kDa, 29.0 and 28.5 kDa, and 26.5 and 26.0 kDa. The purified enzyme was labeled with [3H]DFP, electrophoresed in the same SDS gel, and the [3H]DFP-labeled proteins identified (Fig. 1, lane B ) . All the bands observed by silver staining were labeled by [3H]DFP. Relative intensities of the radioactivity and silver staining were correlated by visual inspection.
By contrast, gel filtration of the purified enzyme on a TSKG-3000 SW column showed a mass of 580 kDa (Fig. 2).
Enzymatic Properties-The enzyme has a single pH optimum between 8.0 and 8.5. Activity was inhibited by several seryl-protease-specific inhibitors including DFP, aprotinin, leupeptin and benzamidine. Chelating or alkylating reagents had no effect (Table 11).
Effect of N-Glycosidase F (N-GlycanaseTM) Treatment of the Enzyme-To determine whether the multiplicity of molecular forms of the enzyme is due to differences in glycosylation, the purified enzyme was treated with N-glycosidase F (Fig. 3).  The cluster of bands (Fig. 3, lane A) collapsed to two distinct bands (a major band at 28.0 kDa and a minor band at 30.0 kDa) as shown in Fig. 3 (lane B).
Effect on Pro-ANF-Partially purified pro-ANF was treated with the enyzme, and the resultant product was fractionated by gel filtration. Fig. 4a shows the elution profile. Partially purified bovine pro-ANF was eluted in fraction 35 in this system. After enzyme treatment the elution position of immunoreactive ANF was shifted to the elution position of the circulating form, ANF(99-126). The recovery was 76%.
Further examination of the product from the enzyme treatment by reverse-phase HPLC revealed that the elution position was shifted from that of pro-ANF to that of ANF(99-126). This position was distinguishable from those of ANF(105-121), ANF(102-126), ANF(103-126)r or A N F (~G~~~) , as shown in Fig. 4b. In order to obtain unequivocal identification of the ANF produced by the enzyme, the peak fractions (shown by a bar in Fig. 4b) were collected, purified by HPLC, and subjected to sequence analysis. The following amino-terminal sequence was determined Ser-Leu-Arg-Arg-Ser-Ser-. In addition the amino acid composition of the peptide was compatible with that of ANF(99-126).
Subcellular Distribution of the Enzyme-The subcellular distribution of the enzyme was determined with fractions obtained from a homogenate of bovine atria (15). In addition to the enzyme activity detected by AGPR-MCA, we also measured nonspecific protease activity with Z-Phe-Arg-MCA (ZFR-MCA), immunoreactive ANF, and alkaline phosphatase  3. Effect of N-glycosidase F (N-GlycanaseTM) treatment on the purified enzyme. The enzyme (100 ng) was treated for 24 h with N-glycosidase F by the method given in the manufacturer's instruction. Lanes A and B show control (incubation without N-glycosidase F) and sample incubation with N-glycosidase F, respectively. Standards used for calibration are described in the legend to Fig. 1.   (Table 111). The AGPR-MCA activity appeared mainly in the microsomal fraction, whereas ZFR-MCA hydrolyzing activity was predominantly localized in the cytosolic fraction (high speed supernatant). Immunoreactive ANF was found mainly in the granular fraction, as expected from previous studies (15). 97 a Immunoreactive ANF.
The present study represents the identification and purification of a protease in bovine atrial membrane fractions capable of catalyzing the specific processing of pro-ANF to the circulating form of ANF. Although the amino acid sequence of active bovine ANF was shown to be the same as that of human ANF (14), the entire sequence of bovine pro-ANF has not yet been determined. However, we found the substrate used in the rat study served as a substrate for the bovine enzyme.
The purified enzyme has the following characteristics: 1) it has a high affinity to heparin-agarose and is stabilized by heparin; 2) the mass is 580 kDa by gel filtration, whereas on SDS-polyacrylamide gel electrophoresis a cluster of bands is seen around 30 kDa. We could not exclude the possibility of nonspecific aggregation of the enzyme, but a sharp single peak of the activity in Fig. 2 suggests that the enzyme exists in a 580-kDa form.
Three doublets observed on SDS-polyacrylamide gel electrophoresis around 30 kDa indicate that the enzyme is an aggregate of smaller subunits. To examine the possibility that these multiple bands arise from proteolytic cleavage of the native subunits or are due to heterogeneous glycosylation, three experiments were performed. First, the enzyme was purified in the presence of N-ethylmaleimide. This cysteinyl protease inhibitor, which has been shown not to inhibit the present enzyme (Table 11), was added to all the buffers at 1 mM during purification to prevent proteolytic cleavage by cathepsin-like cysteinyl proteases which are often present in the cytosol. The resultant product gave an identical pattern on SDS-gel electrophoresis.
Second, we labeled the crude extract with [3H]DFP and subjected it to SDS-gel electrophoresis. The pattern of radioactive bands obtained from the crude extract was identical to that ( Fig. 1) obtained from the purified and [3H]DFP-treated enzyme (data not shown). In the third experiment, we treated the enzyme with an N-glycosidase. This treatment reduced the mass and the number of the bands, indicating that the multiplicity of the bands is mainly due to heterogeneity in glycosylation, similar to the pattern observed with tryptase isolated from the pituitary gland (22).
Recently several high molecular weight seryl proteases have been obtained from various tissues (22)(23)(24)(25)(26). The enzyme purified by us has characteristics similar to these, but does not seem identical. Tryptases isolated from human lung (23,24) and human pituitary (22) were shown to have high affinity to heparin and to be stabilized by heparin. However, their masses are 120-130 kDa, which are much smaller than that of the atrial enzyme. While high ionic concentrations stabilize tryptases, the atrial enzyme rapidly lost activity under the same conditions. An enzyme isolated from rat liver has been shown to have a mass of 600 kDa (25), which is close to that of the atrial enzyme. However, the liver enzyme exists only in the cytosol, and the substrate specificity is completely different from the atrial enzyme.
The enzyme we identified seems to be different from that reported by Baxter et al. (27). The present enzyme is a seryl protease whereas their enzyme is a cysteine protease.
The observation that the present enzyme exists mainly in the microsomal fraction distinguishes it from the protease activity detected by ZFR-MCA which is localized in the cytosol. Twenty percent of the AGPR-MCA hydrolyzing activity which appeared in the cytosolic fraction (Table 111) could be due to nonspecific hydrolysis of this substrate by cytosolic proteases. It is also noteworthy that the present enzyme is not localized in atrial granules containing pro-ANF, in agreement with the fact that ANF exists as pro-ANF in the granules rather than being processed in the granules. Recently Page et al. (28) suggested the involvement of the endoplasmic reticulum and Golgi in the processing of pro-ANF just before secretion as active ANF. This agrees with our observation of high activity in the microsomal fraction. However, the microsomal fraction obtained by the present method also contains plasma membranes, as indicated by the presence of alkaline phosphatase activity, a plasma membrane marker (as shown in Table 111). Further investigation of the distribution of the enzyme may disclose the path of processing ANF.
HPLC and sequence analysis of the peptide produced from pro-ANF by this enzyme demonstrates that ANF(99-126) is the major or exclusive product. Since the amino acid sequence of bovine pro-ANF is not known and since antibodies to the amino-terminal segment of bovine pro-ANF are not available, it is not feasible to investigate whether this enzyme hydrolyzes other arginyl peptide bonds in the amino-terminal segment of pro-ANF. However, recently Michener et al. (8) made an interesting approach to this problem by utilizing an antibody to the NH2-terminal fragment of rat pro-ANF. They detected only one (14 kDa) molecular form, which corresponds to the 98-residue NHz-terminal fragment in rat blood, and concluded that processing of pro-ANF is performed at only one site, -Glyg6-Prog7-Argg8-Sergg-.
In view of the specific functional features of the enzyme we propose to call it "atrioactivase." One of the unique features of this enzyme is the specific cleavage of the arginyl peptide bond in -Prog7-Argg8-Serg9-instead of bonds involving a double basic residue sequence such as -Arg'o'-Arg'02-Ser'03-. Cleavage of the latter peptide bonds gives rise to physiologically active peptides with 24 or 25 residues. Although such ANFs have not been found in plasma, they have been reported in rat brain (29,30). In the brain the processing of pro-ANF may be catalyzed by proteases specific for double basic residues (31,32), sequences which are commonly found in the processing of precursors of many peptide hormones such as preopiomelanocortin.