Eur. J . Biochem. 207, 1035-1044 (1992) 0FEBS 1992

Purification and characterization of a thermostable proteinase isolated from Thermus sp. strain Rt41A Keith PEEK, Roy M. DANIEL, Colin MONK, Lynne PARKER and Tim COOLBEAR Thermophile and Microbial Biochemistry and Biotechnology Unit, University of Waikato, Hamilton, New Zealand (Received February 19/May 19, 1992) - EJB 920233

Thermus sp. strain Rt41A produces an extracellular thermostable alkaline proteinase. The enzyme has a high isoelectric point (10.25 - 10.5) which can be exploited in purification by using cationexchange chromatography. The proteinase was purified to homogeneity and has a molecular mass of 32.5 kDa by SDS/PAGE. It is a glycoprotein, containing 0.7% carbohydrate as glucose equivalents, and has four half-cystine residues present as two disulphide bonds. Maximum proteolytic activity was observed at pH 8.0 against azocasein and greater than 75% of this activity was retained in the pH range 7.0- 10.0. Substrate inhibition was observed with casein and azocasein. The enzyme was stable in the pH range 5.0-10.0 and maximum activity, in a 10-min assay, was observed at 90°C with 5 mM CaCI, present. No loss of activity was observed after 24 h at 70°C and the half-lives at 80°C and 90°C were 13.5 h and 20 min, respectively. Removal of Ca2+reduced the temperature for maximum proteolytic activity against azocasein to 60 "C and the half-life at 70 "C was 2.85 min. The enzyme was stable at low and high ionic strength and in the presence of denaturing reagents and organic solvents. Rt41A proteinase cleaved a number of synthetic amino acid p-nitrophenol esters, the kinetic data indicating that small aliphatic or aromatic amino acids were the preferred residue at the P1 position. The kinetic data for the hydrolysis of a number of peptide p-nitroanilide substrates are also reported. Primary cleavage of the oxidized insulin B chain occurred at sites where the P1' amino acid was aromatic. Minor cleavage sites (24 h incubation) were for amino acids with aliphatic side chains at the P1' position. The esterase and insulin cleavage data indicate the specificity is similar for both the P1 and P1' sites.

Members of the extremely thermophilic eubacterial genus Thermus are widely distributed in natural, geo heated waters throughout the world. Only three validly named species have been described but numerous isolates exist that show there is considerable variety within the genus [4]. For example, the validly named Thermus aquaticus does not show a close taxonomic relationship to any New Zealand isolates so far described [4]. Five proteinases, all of which are serine-type proteinases, have been characterized from extremely thermophilic Thermus isolates [5-91. Two of these enzymes, caldolase [8] (from strain Tok3) and aqualysin I [7] (from T. aquaticus) are chelator-resistant, whilst caldolysin [5] (from strain T351) and aqualysin I1 [9] (from T . aquaticus) are chelatorsensitive. This paper deals with the purification and characteristics Correspondence to K. Peek, Pacific Enzymes Ltd, Thermophile of the extracellular proteinase from Thermus sp. strain Rt41A, and Microbial Biochemistry and Biotechnology Unit, University of isolated from a geothermal hot pool in New Zealand [lo]. The Waikato, Hamilton, New Zealand Abbreviations. E-64, ~-trans-epoxysuccinyl-leucylamido[4-guidi-effects of growth conditions on the yield and stability of the nolbutane; Gdn/HCl, guanidine hydrochloride; Cbz, carbobenzoxy ; proteinase have been described previously [ll]. Also, because Suc, succinyl; tosyl, 4-toluenesulphonyl; -NH-Np, p-nitroanilide. of its high thermal stability, and rapid inactivation in the Enzymes. Thermus Rt41A proteinase (EC 3.4.21.-); proteinase K, presence of EGTA, Rt41A proteinase has been used for the subtilisin (EC 3.4.21.14); chymotrypsin (EC 3.4.21.1); thermolysin preparation of DNA [12] and mRNA [13] prior to amplifi(EC 3.4.24.4); lysozyme (EC 3.2.1.17). Note. The novel amino acid sequence data published here have cation by the polymerase chain reaction and can replace probeen deposited with the Swiss-Prot sequence data bank and are avail- teinase K in preparative methods for DNA isolation. A recent able under the accession number P80146. report has also shown that the enzyme can be used for the

Generally there is no firm evidence to suggest that an enzyme isolated from an extremely thermophilic organism will have a higher specific activity than its mesophilic counterpart (when assayed at a temperature appropriate to the growth temperature of the producing organism). However, mesophilic proteins tend to denature and become more susceptible to proteolysis at the temperatures at which thermostable enzymes operate optimally. Thus, proteinases from extreme thermophiles have higher specific activities against mesophilic proteins than do most microbial proteinases [l]. The inherent stability of thermophilic proteinases to high temperatures as well as to detergents, organic solvents and chaotropic agents, makes these enzymes potentially useful in a range of biotechnological applications [2, 31.

1036 cleaning of whey-fouled ultrafiltration membranes at elevated temperatures [14]. A preliminary report of the amino acid sequence of Rt41A proteinase, deduced from the nucleotide sequence of the gene 1151, shows it is 70% similar to aqualysin I [16] and has high similarity to the industrially important subtilisin-like proteinases [17].The differences in the properties of Thermus proteinases [I 81 potentially make them a commercially interesting group of enzymes, the diversity of which is not yet fully understood. MATERIALS AND METHODS Growth of Thermus sp. strain Rt41A Thermus sp. strain Rt41A was grown on a modified medium 162 [I 11, containing 2 g/1 sodium glutamate. 600 1 media was inoculated with a 5.0-1 batch culture (grown in a Chemap CF200 fermenter) and grown at 70°C with aeration and pH control. After 20 h of growth, cultures were clarified using a continuous-flow Sharples centrifuge (100 l/h, 13000 x g). Proteinase assay

Proteinase activity was determined by the release of trichloroacetic-acid-soluble peptides from azocasein at 75 "C. The reaction was started by the addition of 10 pl proteinase sample to 1 ml 0.2% (mass/vol.) azocasein in 50 mM Hepes/ NaOH pH 7.5 containing 5 mM CdC12, preincubated for 10 min at 75°C. The reaction was stopped after 10 rnin with the addition 500 p1 15% (massivol.) trichloroacetic acid. The mixture was left for 10 min at room temperature, then centrifuged at 15000 x g for 5 min and the absorbance of the supernatant determined at 420 nm. The rate of azocasein hydrolysis was linear up to an A420n m = 0.2. One unit (U) is defined as that amount of enzyme activity producing a change in absorbance at 420 nm of 1.0 in 1 h at 75°C. Where the proteinase was assayed in the presence of inhibitors, the enzyme sample was preincubated for 5 rnin at 70°C in Hepes buffer containing 0.01 % (by vol.) Triton X-100 and the inhibitor (concentration given in text) before assay. Protein assay

Protein was determined by the method of Lowry et al. [I91 using bovine serum albumin as a standard. Interfering substances [20] were removed by dialysis or dilution. Samples containing Triton X-100 were quantified by comparison to albumin standards containing Triton X-100. Purification of Rt41A proteinase

All steps in the following purification scheme were performed at ambient temperature. 160 1 cell-free supernatant was taken to 80% saturation with ammonium sulphate and left standing overnight. The precipitated protein was harvested by continuous-flow centrifugation (Sharples), then extracted with 500 ml 10 mM Hepes/NaOH pH 7.5. Insoluble material was separated by centrifugation (Sorvall SS3, 9000 x g, -30 mm)and re-extracted twice more. All the supernatant fractions were pooled, then desalted and concentrated by diafiltration with Tes buffer ( I 0 mM Tes/NaOH pH 7.5, containing 5 mM CaC12)in a 2.5-1 stirred cell using a PMIO ultrafiltration membrane (Amicon). The concentrate (final volume 1.O 1) was then applied to 300 ml Fast-Flow S Sepharose (Pharmacia)

packed into a Pharmacia K50/30 column and equilibrated with the Tes buffer. The unbound protein was discarded and the proteinase eluted with Tes buffer containing 1 M sodium chloride. The eluted proteinase was desalted and concentrated to 50 ml by diafiltration using Tes buffer. The preparation was then subjected to two successive purification steps by cationexchange chromatography using a Mono S l O / l O FPLC column (Pharmacia). In the first step the 50 ml of preparation was applied to the column (equilibrated in Tes buffer) and eluted with a 90-ml linear gradient of 0-0.5 M NaCl in Tes buffer at a flow rate of 3 ml/min. Fractions (3 ml) containing proteinase activity were pooled then desalted and concentrated by diafiltration to a final volume of 60 ml. In the second step, 20-ml aliquots were reapplied to the Mono S lOjl0 column and eluted with a 90-ml linear gradient of 0-0.4 M NaCl in Tes buffer. Fractions (3 ml) containing activity from three successive runs were pooled, diafiltered and concentrated then stored frozen at - 70°C until required for use. Electrophoresis

The molecular mass of Rt41A proteinase was determined by SDSjPAGE on 10- 15% T Phast gels using the Pharmacia Phast system and comparison with standard marker proteins (low-molecular-massmarker kit, Pharmacia). The preparation of samples for electrophoresis was as described by Pharmacia (separation technique file 1lo), except that the proteinase samples were inactivated prior to electrophoresis with 10 mM phenylmethanesulphonyl fluoride and samples were reduced, when required, with 75 mM dithiothreitol. After electrophoresis, gels were silver-stained to visualize protein bands (Pharmacia development technique file 240). A separate estimate of the molecular mass was obtained by cathodic PAGE [21] using gradient gels 8-45% T (total concentration of monomer), 4% C (cross-linking agent of T) containing 60 mM KOH and 375 mM acetic acid pH 4.3 prerun for 1 h at 150 V in cathodic electrophoresis buffer [21] (LKB vertical electrophoresis unit). Cytochrome c, lysozyme, trypsinogen and lactoferrin were used as molecular mass markers and basic fuchsin as tracking dye. Electrophoresis was at 150 V and 15"C for 4 - 5 kV h. Gels were fixed immediately with 10% (massivol.) sulphosalicylic acid and stained with Coomassie blue R-250. Double logarithmic plots of molecular mass versus relative mobility for marker proteins gave a linear relationship from which the molecular mass of the proteinase was estimated. Gel filtration

Gel filtration was performed on a TSK G3000 SW column (Toyo Soda Co., Japan) and a Superose 12 10/30 column (Pharmacia). The columns were equilibrated with 50 mM Hepes/NaOH pH 7.5 containing 5 mM CaCI2, 300 mM NaCl and 0.01% (by vol.) Triton X-100. The TSK column was run at 1 ml/min and the Superose 12 at 0.5 ml/min using a Waters M45 HPLC pump and protein was detected by absorbance at 280 nm with a Waters model 481 LC spectrophotometer. Pierce molecular mass markers (18-300 kDa) were used as standards. lsoelectric point determination

The isoelectric point of the proteinase was estimated against high-isoelectric-point markers (Pharmacia) by focusing on a 1% (massivol.) agarose gel containing 6.3% (by

1037 vol.) Pharmalyte 8.0 - 10.5 (Pharmacia) and 12% (mass/vol.) sorbitol, as described by Pharmacia in Isoelectric focusing: principles and methods. Gels were run at 600 V for 3 h at 24 "C on a Pharmacia flat-bed electrophoresis unit, flushed continuously with nitrogen. After electrophoresis, gels were fixed immediately with 10% (mass/vol.) trichloroacetic acid/ 5% (mass/vol.) sulphosalicylic acid and stained with Coomassie blue R-250 [22]. Carbohydrate content Total sugars were estimated using an orcinol/sulphuric acid method [23] and expressed as glucose equivalents. Free thiol groups and disulphide bonds Free sulphydryl groups were estimated using 5,5'-dithiobis(2-nitrobenzoic acid) according to the method of Ellman [24]. Qualitative detection of disulphide bonds was made by comparing the distance the proteinase migrated after SDS/ PAGE under reducing and non-reducing conditions. The presence of disulphide bonds is indicated by the non-reduced sample migrating further, due to the molecules having a smaller radius of gyration [25]. Free thiol groups were estimated by the method of Ellman [24] and quantitative determination of disulphide bonds made after the addition of cysteine (final concentration, 5 pM) to proteinase samples. The cysteine-spiked sample was then treated with Ellman's reagent and the free sulphydryl groups determined. In the presence of free thiol groups, disulphide bonds are cleaved and derivatized with Ellman's reagent [26]. The 3-carboxy-4-nitrothiophenolate group attached to the protein is only released after titration to pH 10.5 with NaOH. The number of disulphide bonds is determined from the difference in 3-carboxy-4nitrothiophenolate before and after titration to pH 10.5. Amino acid composition Purified Rt41A proteinase samples were inhibited with 10 mM phenylmethanesulphonyl fluoride for 1 h at 20°C. Excess phenylmethanesulphonyl fluoride and calcium were removed from proteinase samples by dialysis at 4°C with 100 mM EDTA pH 6.0, followed by Milli Q water. Samples were then hydrolysed at 150"C for 1 h and 2 h with 6 M HC1 containing 0.5% (by vol.) phenol in evacuated tubes (flushed three times with 02-free nitrogen). The resulting amino acids were derivatized with phenylisothiocyanate and separated using a Waters liquid chromatography system [27]. NH2-terminal amino acid analysis The NH2-terminal amino acid analysis of Rt41A proteinase was determined using a gas phase sequencer (Applied Biosystems, model 470A) equipped with an on-line phenylthiohydantoin amino acid analyser. Prior to analysis Rt41A proteinase was electroblotted from SDS/polyacrylamide gels to a polyvinylidene difluoride membrane (Immobilon transfer, 0.45 pm, Millipore) and the protein band sequenced by placing polyvinylidene difluoride membrane pieces in the upper cartridge block of the sequencer. The effect of temperature on enzyme activity The effect of temperature in the range 50-100°C on enzyme activity was determined with azocasein and Suc-Ala-

Ala-Pro-Phe-NH-Np (succinyl-alanyl-alanyl-prolyl-phenylalanine p-nitroanilide) as substrates. With azocasein the activity was determined in both the presence and absence of calcium. Activity against Suc-Ala-Ala-Pro-Phe-NH-Np was determined by incubating 10 p1 proteinase for 5 min with 500 p1 2 mM substrate in Hepes buffer containing 5 mM CaC1, and 0.01% (by vol.) Triton X-100. The reaction was stopped with the addition of 100 p1 glacial acetic acid and the absorbance determined at 410 nm. In both cases the enzyme was added to substrate pre-incubated at the appropriate temperature. Adjustment of pH was made at the temperature of incubation. Stability studies All experiments were performed with proteinase samples in Hepes buffer (pH 7.0 or 7.5) containing 0.01% (by vol.) Triton X-100 and CaC1, as indicated in the text. For studies in which half-lives of only a few minutes were observed, proteinase samples (50 yl) were incubated in sealed glass capillaries (100 mm) to ensure rapid temperature equilibration. Adsorption of proteinase onto the glass surface [28] was prevented by the addition of NaCl ( I = 0.1 M). Triplicate samples were removed at each of the indicated times, stored on ice, and assayed for proteinase activity using azocasein as a substrate at the completion of the time course. For studies involving longer half-lives, proteinase samples (2.0 ml) were held in sealed plastic sample tubes and triplicate 50-pl samples removed at the indicated times. The samples were frozen immediately at - 70 "C and stored until completion of the time course, then assayed for proteinase activity using azocasein as a substrate. Activity towards synthetic amino acid esters and peptides The esterase activity of Rt41A proteinase was determined with various carbobenzoxy (Cbz) amino acid p-nitrophenol esters (Sigma) at 50°C. Stock substrates were prepared with acetonitrile and assays started by the addition of 15 p1 substrate to 1.505 ml 35 mM Hepes/NaOH pH 7.0 containing 3.5 mM CaCI2, 0.007% (by vol.) Triton X-100, 30% (by vol.) acetonitrile and 12 units proteinase activity in stoppered 1.5ml quartz cuvettes. The increase in absorbance at 400 nm was monitored continuously for 3 min in a Perkin Elmer Lambda 3B spectrophotometer fitted with a thermoelectric five-cell holder. The rate of change in absorbance was calculated from the initial rate of reaction and activity was determined using an absorption coefficient forp-nitrophenol of 18300 M - cm-l. The activity against a number of peptide p-nitroanilides (Sigma Chemical Co. and Bachem, Switzerland) were similarly determined at 70°C. Stock peptide solutions (2-5 mM) were prepared in 50 mM Hepes/NaOH, pH 7.5, containing 5 mM CaC12, and diluted with buffer as required. The assay was started by the addition of 10 p1 Rt41A proteinase (38.4 or 384 Ujml) and monitored continuously for 5 min at 400 nm. The rate of change in absorbance was calculated from the initial rate of reaction and activity determined using an absorption coefficient for p-nitroaniline of 10 500 M - cm- I. HPLC analysis was used to confirm that the p-nitroanaline bond was the only site of cleavage in the peptides used. The K , and V,,, for the ester and peptide substrates were determined by linear regression analysis (least-squares method) of Lineweaver-Burk plots from at least six substrate dilutions.

'

1038 Table 1. Purification of Rt4lA proteinase from 160 1 culture supernatant.

Step

Culture supernatant (NH4)2S04 precipitation Fast-Flow S Sepharose FPLC Mono s lOjl0

Total activity

Totdl protein

Specific activity

Purification

kU

mg

U/mg

-fold

1120

8000

140

1

336

1300

260

2

307

200

1540

11

246

26

9370

61

Determination of the cleavage sites for Rt41A proteinase against oxidised insulin B chain Oxidized insulin B chain (bovine, Sigma) was purified before use by reverse-phase HPLC on a MPLC CI8 column (4.6 mm x 22 cm; Brownlee Labs, USA) with a Waters liquid chromatograph system. The oxidised insulin B chain was eluted on a 60-ml linear gradient of 0-100% (by vol.) acetonitrile containing 0.1YO(by vol.) trifluoroacetic acid, at a flow rate of 1 ml/min. Peptides were detected by absorbance at 230 nm and the fractions containing the insulin B chain pooled and lyophilized. For specificity studies 4.5 mg purified insulin B chain was dissolved in 10ml 5 mM Tes/NaOH pH 7.5 containing 5 mM CaClz and the preparation preheated to 70°C prior to the addition of 0.45 pg proteinase (1 :10000, by mass); 1-ml samples were removed at 5, 15, 30 min and 1, 3,6, and 24 h and the reaction stopped by the addition of 5 p1 6 M HCl and cooling on ice. All samples were stored at - 70°C until required for peptide analysis. Peptides resulting from the cleavage of insulin B chain were separated using reverse-phase HPLC, eluted with a 60-ml gradient of 0 - 60% (by vol.) acetonitrile containing 0.1 YO(by vol.) trifluoroacetic acid, at a flow rate of 1 ml/min. The eluted peptides were lyophilized and then hydrolysed by heating for 1 h at 150°C in 6 M HCI containing 1% (by vol.) phenol. The resulting amino acids were derivatized with phenylisothiocyanate [27] and separated using the Waters Picotag amino acid analysis system. The amino acid composition of the peptides were matched to the known sequence of oxidised insulin B chain to determine cleavage sites. RESULTS AND DISCUSSION Purification Table 1 shows the purification of Rt41A proteinase using ammonium sulphate precipitation and cation-exchange chromatography. The poor sedimentation of the ammonium sulphate precipitate resulted in low yields with 70% losses being typical at this stage. Similar losses were also observed at this stage for caldolase [8]. However, the yield of proteinase from the recovered precipitate was good, typically over 70% recovery being obtained (Table 1). Smaller-scale cultures (5.0 1) in which the cell-free supernatant was concentrated by ultrafiltration, and quantitative recovery of the enzyme obtained, showed that only one proteinase is secreted into the culture media (results not shown). Overall, the enzyme was purified by a factor of 67 and had a specific activity of 9370 U/mg. The purified proteinase was

present as a single band after both cathodic PAGE and SDS/ PAGE under non-reducing conditions (Fig. 1a). However, after SDS/PAGE under reducing conditions, additional lowermolecular-mass bands were observed (Fig. la), due to a low level of residual autolysis in the presence of phenylmethanesulphonyl fluoride (see Table 4). This can be prevented by an acid-denaturing step prior to electrophoresis (results not shown). The proteinase was also present as a single protein band after isoelectric focussing (Fig, lb) and focussed beyond the last PI marker (PI 10.25), but within the Pharmalyte pH range, indicating a PI of between 10.25 and 10.5. Molecular mass The molecular mass of Rt41A proteinase was 36250 1300 Da as determined by cathodic PAGE and 32500 300 Da by SDSjPAGE (Fig. la). Density gradient centrifugation gave a molecular mass of 36 kDa as calculated by the sedimentation ratio of Martin and Ames [29], using alcohol dehydrogenase and cytochrome c as markers. The results are in good agreement with the molecular mass of 35 750 Da derived from the gene sequence of the mature proteinase [15]. Accurate estimation of the molecular mass of Rt41A proteinase by gel-filtration chromatography was not possible since the behaviour on TSK G3000SW and Superose 12 101 30 columns was anomalous. On both columns the proteinase eluted in a volume greater than the bed volume; the K,, was 1.28 on the TSK column indicating adsorption to the column matrix. Increasing the ionic strength ( I = 0.4 M) of the eluent and addition of Triton X-100 (0.1Y0 by vol.) reduced the K,, to 1.14. We have already noted that Rt41A proteinase is hydrophobic in nature and is readily adsorbed onto polyethylene surfaces [28]. It is also adsorbed onto glass which may be a consequence of its high isoelectric point, and interaction with the anionic silanol groups of glass surfaces. Anomalous molecular mass estimations using gel-filtration chromatography have been observed for a number of extracellular microbial proteinases. While some workers have suggested that this is a consequence of a high PI value, others have shown this not to be the case [30]. Although the exact mechanism of the interaction of Rt41A proteinase with the column matrices is unknown, the result is an inaccurate estimate of the molecular mass using gel filtration. Effect of proteinase inhibitors The results presented in Table 2 indicate that the enzyme is a serine proteinase. The most effective inhibitor tested was phenylmethanesulphonyl fluoride, with diisopropylfluorophosphate, chymostatin and carbobenzoxy-phenylalanine chloromethane also causing significant inhibition. In contrast, tosylphenylalanine chloromethane reduced the activity only 5%, indicating the importance of the N-terminal blocking group to inhibitor specificity. In comparison, aqualysin I was only weakly inhibited by both these chloromethane inhibitors [7]. However, Rt41A proteinase was not inhibited by other serine proteinase inhibitors such as antipain, aprotonin. 4amidinophenylmethanesulphonyl fluoride, tosyllysine chloromethane and 3,4-dichloroisocoumarin. Rt41A proteinase is not a metalloproteinase as phosphoramidon and the zinc chelator 1,I 0-phenanthroline were not inhibitory. Although EDTA and EGTA reduced the activity, they did so by removing calcium ions that stabilize the tertiary structure (see below) and loss of activity was therefore due to enzyme denaturation rather than inhibition. 2-

1039

Fig. 1. (a) SDS/PAGE and (b) IEF of purified Rt4lA proteinase. (a) Samples were run on a 10- 15% T Phastgel (Pharmacia). Lanes 1, 2, 3, non-reduced; lanes 4, 5, 6 reduced with 75 mM dithiothreitol. Lanes 1 and 6, cell-free supernatant; lanes 2 and 5 , purified Rt41A proteinase; lanes 3 and 4, low-molecular-mass markers. (b) 1% (massivol.) agarose gel, containing Pharmalyte 8.0- 10.5 (Pharmacia). Lane 1 , high-pI IEF markers (Pharmacia); lane 2, purified Rt41A proteinase.

Table 2. Effect of proteinase inhibitors. The following inhibitors had less than 5% effect: 4-amidophenylmethanesulphonyl fluoride (185 pM), p-chloromercuribenzoate (1 mM), iodoacetate (1 mM), E64 (2.8 mM), leupeptin (1.2 pM), dithiothreitol (1 mM), 1,10phenanthroline (10 mM), phosphoramidon (607 pM), tosylphenylalanine chloromethane (100 pM), tosyllysine chloromethane (27 pM), antipain (83 pM), bestatin (130 pM), pepstatin (15 pM), aprotonin (1.5 pM). Inhibitor

Phenylmethanesulphonyl fluoride Diisopropylfluorophosphate 2-mercaptoethanol EDTA EGTA EGTA (30 min) Cbz-Phe-CH2C1 Chymostatin

Concentration

Remaining activity

mM

%

1 1 1 10 10 10 0.3 0.3

7.8 45.6 79.3 49.2 72.0 12.0 50.0 39.0

Mercaptoethanol reduced Rt41A proteinase activity but dithiothreitol had no effect, nor did the cysteine proteinase inhibitors iodoacetate, p-chloromercuribenzoate, leupeptin and 1-trans-epoxysuccinyl-leucylamido[4-guidino]butane (E64). Substrate inhibition Rt41A proteinase showed substrate inhibition with azocasein and casein above 0.02% (mass/vol.), an apparent Kiof 0.6% (mass/vol.) azocasein being obtained (results not shown) from a plot of I/v against substrate concentration [31]. The same phenomenon has been reported for caldolysin with azocasein as a substrate [32], and for both proteinase K and subtilisin BPN’ with azoalbumin [33]. The mechanism of inhibition is unknown but may involve the formation of an enzyme-substrate complex in which two substrate molecules bind to the active site and so prevent catalysis [31]. However, other mechanisms including allosteric interactions would also explain this phenomenon.

Carbohydrate content The carbohydrate content of Rt41A proteinase was 0.7% (by mass) as glucose equivalents using the orcinol/sulphuric acid method and is considerably lower than the 13% (by mass) reported for caldolysin [5] and 10% (by mass) for caldolase [S] in which the phenol/sulphuric acid method was used. The large difference may be due to interference caused by protein when using the latter method [23].

Free thiols and disulphide bonds Proteinase samples subjected to SDSjPAGE under nonreducing conditions migrated further than samples prepared under reducing conditions (Fig. la), indicating the presence of intramolecular disulphide bonds. No free thiol groups were detected with Ellman’s reagent at a proteinase concentration of 0.6 mg/ml. After spiking proteinase samples with 5 pM cysteine, free thiol was detected corresponding exactly to that derived from the added cysteine. After titration to pH 10.5 to release 3-carboxy-4-nitrothiophenolatefrom derivatized disulphide bonds, a net free thiol content of 3.5 mol/mol enzyme was detected. This indicates the presence of four half-cystine residues (two disulphide bonds) per enzyme molecule.

Amino acid composition Table 3 compares the amino acid composition of Rt41A proteinase with that of aqualysin 1. The compositions of the two proteinases are very similar except that the threonine content is much higher for Rt41A proteinase. The compositions of both enzymes are similar to members of the subtilisinlike family (see [7]), except for the content of cystine, arginine and lysine. The presence of two disulphide bonds further classifies the enzymes as belonging to a subclass of subtilisinlike proteinases that contain cystine, which includes proteinase K. The arginine/lysine ratio is high in both enzymes and low for the mesophilic members of this family. The substitution of lysine for arginine is a feature of a number of thermophilic enzymes when compared to their mesophilic counterparts and may be associated with increased thermostability [34].

1040 Table 3. The amino acid composition of Rt41A proteinase. Values are the mean of four determinations. Coefficient of variation < 5 % for all measurements. Amino acid

Number of residues/molecule aqua1y sin

Rt41 A proteinase

Asx Glx Ser GlY His Arg Thr Ala Pro TYr Val Met CYS Ile Leu Phe TrP LYS a

,ooo

determined

integral

31.3 11.0 25.9" 35.3 6.4 10.8 37.8" 39.3 16.2 17.1 26.2' 3.7 3.5" 12.3b 23.7 6.3 4.Id 4.6

31 11 26 35 6 11 38 39 16 17 26 4 4 12 24 6 5 5

28 9 28 38 5 16 24 41 12 17 27 2 4 10 19 3 5 2

Value extrapolated to t = 0 hydrolysis. Determined after 2 h hydrolysis at 150°C. Determined by method [26], see text. Determined spectrophotometrically.

NH2-terminal amino acid sequence

The NH2-terminal sequence of Rt41A proteinase was AlaVal- Gln-Ser-Pro5- Ala- Thr -Xaa- Gly -Leu" - Asp-Xaa- IleAsp-Gln"-. No phenylthiohydantoin was detected at the eighth cycle suggesting the presence of a tryptophan or modified residue. The derived amino acid sequence from the gene sequence [1 51 identified this residue as tryptophan. The amino acid derivative after the twelfth cycle could not be identified conclusively and is perhaps indicative of a modified residue. Gene sequence analysis [15] identified this residue as an arginine. The sequence shows considerable similarity to the NH2terminal sequence of aqualysin 1 [7] with only two differences: a valine for threonine at residue 2 and a threonine for proline at residue 7. Activity and stability as a function of pH

Maximum proteolytic activity against azocasein was observed at pH 8.0, with 95% or more of this activity retained over the range pH 7.5-9.3 and 50% or more in the range pH 6.0 - 10.4. In the presence of 5 mM CaCl, the enzyme was stable for at least 4 h at 70°C in the range pH 5-10 (results not shown). Effect of temperature on activity

The effect of temperature on the activity against azocasein is shown in Fig. 2 as an Arrhenius plot. In a 10-min assay, maximum activity was observed at 90°C (ZIT= 0.00275), decreasing thereafter due to thermal denaturation. If free calcium was removed from the assay by the addition of excess EDTA, maximum activity occurred at 60°C (l/T=0.003,

,

Temperature ("C)

190

9;

8.0

7,O

5,O

6,O

100

101 0.0026

.

I

0.0027

,

I

0.0028

,

I

0.0029

0.0030

.

,

0.0031

1IT (K ) Fig. 2. Arrhenius plot of the proteolytic activity of Rt4lA proteinase. Activity was determined with 0.2% (massivol.) azocasein in 50 mM Hepes/NaOH pH 7.5 with 4.5 mM CaC1, (0) and with 9.5 mM EDTA ( 0 ) .

Fig. 2), demonstrating the thermostabilizing effect of calcium on Rt41 A proteinase. Below 60°C the absence of calcium resulted in increased activity; the proteinase was 20% more active at 50°C (1/T= 0.0031) in the absence of calcium. This effect was also observed with haemoglobin and the insoluble substrate azocoll, but not with the peptide substrate Suc-Ala-Ala-Pro-Phe-NH-Np. The proteinase has presumably evolved to have optimum flexibility and stability for activity at the growth temperature optimum of the organism (70°C). The increased activity against protein substrates in the presence of a chelator at lower temperatures may be a result of the increased (restored) molecular flexibility of the proteinase as calcium salt bridges are broken. At a temperature where the tertiary structure is still maintained in the absence of calcium, the increased molecular flexibility may result in more efficient binding of macromolecular substrates. A more flexible structure may result in an increase in the binding energy between enzyme and substrate. The increased binding energy could then be utilized to increase k,,, without any necessary change in K,,, (see Fersht [35]). For the smaller peptide substrate where access to the binding subsites is less restricted, and fixed to a maximum of five subsites, the increased flexibility of the proteinase would not necessarily increase reaction rates. The mean activation energy in the presence of calcium was 65.3 kJ/mol and Q l o (SO-75°C) was 2.0. A change in the

1041 30

A A

h-.

(a) 5 rnM CaCI2+ 10 rnM EDTA

-A

A

A

25

100 3

-

L

E .

2

20

->x c

._ c $

15

0 2 4 6 81012 Time (rnin)

0 .c

-gz

10

c

?

n

5

I

\

, . , . , . , . & . , .

0 0

2

6

4

8

1

0

1

2

1

4

Time (min) ' 0

10

20

30

Time (h)

Fig. 3. Thermostability of Rt41A proteinase. Proteinase samples contained 50 mM Hepes/NaOH pH 7.0, 5 mM CaCI2, 0.01% (by vol.) Triton X-100 and 76 mM NaCl ( I = O . l M). Triplicate samples were prepared for each time interval in sealed glass capillaries for 70°C (A) and 90°C (0). At 80°C ( 0 )the samples were held in a plastic reaction vial (I .5 ml) and triplicate samples removed at the selected time interval. Residual activity was determined using azocasein. Mean values are plotted.

slope which occurred at about 75"C, resulted in a lower activation energy. Interpretation of this change is complicated because the effects of substrate unfolding, enzyme denaturation and possible changes to the rate determining steps of enzyme catalysis are not easily separated [36]. The Arrhenius plot obtained with the peptide. substrate Suc-Ala-Ala-Pro-Phe-NH-Np is curvilinear in the range 50 75°C (results not shown) and similar to that obtained with caldolase using Suc-Ala-Ala-Ala-NH-Np as a substrate (81. Above 75 "C, only a small increase in the activity was observed over a 15°C rise in temperature, followed by a decrease due to thermal denaturation. The curvilinear nature of this plot at temperatures where the enzyme is stable might suggest an effect on a rate step in substrate catalysis. However, because the substrate was used at a concentration close to the K , for the enzyme, non-linearity could be due to the heat of substrate binding [36]. Effect of temperature on Rt41A proteinase stability

Fig. 3 shows the thermostability of Rt41A proteinase in the presence of 5 mM CaCl, and 0.01% (by vol.) Triton X100. No loss of enzyme activity was observed after 24 h at 7 0 T , whilst at 80°C and 90°C the half-lives of the enzyme were 13.5 h and 20 min, respectively. At 90°C the loss of activity was a first-order rate process, indicating thermal denaturation, but at 80 "C denaturation is biphasic and the first- and second-order (indicative of autolysis) decay plots were also non-linear. When calcium is sequestered by the addition of excess EDTA (in glass capillaries), activity is lost within minutes at

(b) 10 pM CaCI, 140

b 0

1

.

I

.

I

I

I

I

I

.

I

.

I

.

Time (hours)

Fig. 4. Thermostabilityof RMlA proteinase at 7OOC. (a) In the absence of free C a 2 + :50 mM Hepes pH 7.0, 5 mM CaC12and 10 mM EDTA (inset first-order rate plot); (b) with 50 mM Hepes pH 7.0 and 10 pM CaCI, (inset second-order rate plot). Incubations contained 0.01 % (by vol.) Triton X-100 and I= 1.0 M. At 10 pM C a Z +incubations were in plastic reaction vials and, in the absence of free C a 2 + , in sealed glass capillaries (see Methods).

70°C (Fig. 4a). The loss of activity is a first order rate process (see inset Fig. 4a), indicating thermal denaturation, with a rate constant of 4.03 x s - l and a half-life of 2.85 min. Numerous studies have shown the stabilizing effect calcium has on proteinases isolated from a wide variety of sources (e.g. [37]). A number of proteinases that are stabilized by calcium exhibit two dissociation constants for the ion : high-affinity sites which are saturated at low concentrations of calcium that protect against thermal denaturation ; and low-affinity sites, saturated at high levels of calcium, which are involved in protection against autolysis. Rt41A appears to fall into this general scheme because at 70°C and zero calcium the enzyme undergoes thermal denaturation within minutes (Fig. 4a) and at 5 mM calcium the enzyme is stable for longer than 24 h (Fig. 3). However, at low levels of calcium although a signifi-

1042 Table 4. Stability towards denaturing reagents and organic solvents. Incubations were for 24 h at 4°C and 1 h a t 70°C. All organic solvents were used at 90% (by vol.).

Reagent

Remaining activity at 4°C

1% SDS 5 % SDS 6 M Urea 6 M Gdn/HCl Acetone Acetonitrile Ethanol Butanol Isopropanol Methanol

120 89 126 145 124 85 111 13 123 319

70 "C

73 16 62 5 28 19 55 30 87 4

cant stabilizing effect is observed when compared to zero calcium, the enzyme is not stable at 70°C. For example, at 10 pm CaCI, the half-life was 2.5 h at 70°C (Fig. 4a). The kinetics of activity loss at 10 pM CaCI, are complex; both the first-order and second-order rate plot are non-linear. The second-order rate plot (inset to Fig. 4b) shows that the rate of activity loss appears to increase with time and the mechanism of activity loss in polyethylene tubes is not completely understood. However, there are a number of factors that may complicate interpretation of the rate of of activity loss at 80°C with 5 mM CaClz and at 70°C with 10 pM CaC1,. We have shown that Rt41A proteinase is absorbed onto the walls of polyethylene tubes and this can be prevented at 70°C with 0.01% (by vol.) Triton X-100 in the presence of 5 mM CaC1, [28]. The exact effects of temperature and low calcium on the surface absorption is unknown, but previous observations (unpublished) have shown that absorption to polyethylene is possible in the presence of 0.01% (by vol.) Triton X-100 under a number of conditions. Similarly, although we have demonstrated that autolysis occurs at low concentrations of calcium (Wilson et al., unpublished), the effects of temperature and absorption on the rate of autolysis have yet to be investigated in full. Effect of ionic strength

Ionic strength did not affect proteolytic activity against azocasein in the range I = 0.018 -0.5 M (when adjusted with NaC1) at 75 "C. At I = 1.O M a 20% reduction of activity was observed. Ionic strength did not effect enzyme stability for 24 h at 70 "C in the range I = 0.018 - 1.O M, while at I = 5.0 M 96% of the activity was lost after 24 h. Stability towards denaturing reagents and organic solvents

Table 4 shows the effect of SDS, urea, guanidine hydrochloride (Gdn/HCl) and various organic solvents on enzyme stability at 4°C and 70°C. The results show Rt41A proteinase to be resistant to most of the denaturants tested, even at elevated temperatures. In a number of cases an increase in activity was observed. Since the substrate for the assay is itself a mesophilic protein it would be expected to be more susceptable to the denaturing effects of, for example, urea and

Table 5. Esterase activity of Rt41A Proteinase. Thep-nitrophenyl ester of Cbz-Ile was not cleaved. Although activity was detected with the esters of lysine, benzyl (Bzl) aspartate and proline, K, and V,,, were not determined due to poor substrate solubility or low rates of hydrolysis resulting in inaccurate rate determinations.

Cbz-Xaa p-nitrophenyl ester

GlY Ala TYr TrP Phe Bzl-Cys Asn Gln Val Leu

Km

k,,,

kcatIKm

mM

S-'

s-lM-'

3.4 5.O 2.2 0.3 0.9 0.5 0.7 19.4 0.9 1.1

33.0 38.0 11.4 3.0 6.1 1.0 5.0 21.5 0.3 2.6

9686 7 600 5 182 10133 7 389 2000 7 143 1108 305 2360

Table 6. Kinetic constants for the hydrolysis of peptide nitroanilides by Rt41A proteinase.

Substrate

Suc-Ala- Ala-Pro-Phe-NH-Np Suc-Ala- Ala-Pro-Leu-NH-Np Suc-Ala-A h -Ala-NH-Np

Km

kcat

kcatIKm

mM

s-l

s-l

2.5 3.6 13.4

508 69 15

203 200 19167 1119

M-1

Gdn/HCl. Therefore any carry-over of denaturant into the assay would result in disruption of the substrate's tertiary structure and make it more susceptable to enzymatic hydrolysis. However, this can only partially explain the observed increases, since if the equivalent concentration of urea and Gdn/HCl carry-over were included in the assays without prior treatment of the enzyme at 4"C, then only 16% and 26% increases were observed with urea and Gdn/HCl (cf. Table 4). Thus these denaturants apparently induce changes in the enzyme itself which result in greater catalytic efficiency. Esterase activity

The kinetic data for the esterase activity of RM1A proteinase against a number of Cbz-Xaap-nitrophenol esters is shown in Table 5. The specificity constants indicate that small aliphatic or aromatic amino acids are the preferred residue at the PI position. The specificity constants of tryptophan and glycine were the highest. The small aliphatic amino acids were turned over at the highest rates but had comparatively high K , values, whereas the longer chain aliphatic amino acids (Leu, Val, Ile, Nle) had lower K, values but were turned over rather slowly or not at all. All of the amino esters tested, except that of isoleucine, were cleaved by Rt41A proteinase. The types of ester cleaved by Rt41A proteinase are similar to those cleaved by the Thermus sp. proteinases caldolase [8] and aqualysin I [7], but no direct comparisons can be made due to the lack of kinetic data for these enzymes. Both caldolase and aqualysin I showed most activity towards the alanine ester, for which Rt41A proteinase had the highest k,,,. Caldolase was able to cleave the isoleucine ester which was not cleaved by Rt41A proteinase or aqualysin 1.

1043

u u u u

u

u u u u

U

u

uu

u

Phe-Val-Asn-Gln-His-Leu-Cya-Gly-Ser-His-Leu-Val-Glu-Ala-Leu-Tyr-Leu-Vai-Cya-Gly-Glu-Arg-Gly-Phe-Phe-Tyr-Thr-Pro-Lys-Ala 15min

I

I 1

I

-I1

I - *

I

- 1 24h{

H

-

I

I

I

1

-

I

l

-

I

w

1

H

Fig. 5. Hydrolysis of the oxidized insulin B chain (bovine) by Rt41A proteinase. The diagram shows the peptides recovered by HPLC after 15-min, 3-h and 24-h incubations with Rt41A proteinase. Major and minor cleavage sites are denoted by large and small arrowheads, respectively.

Hydrolysis of synthetic peptides The kinetic constants for the hydrolysis of several p nitroanilide substrates by Rt41A proteinase are shown in Table 6. Of the substrates tested, Suc-Ala-Ala-Pro-Phe-NHNp was most susceptible to hydrolysis, having the highest k,,, and specificity constant. The peptide was originally synthesized as a substrate for chymotrypsin [38] (which has a high specificity constant for the peptide), but it is also cleaved by a number of other proteinases including cathepsin G and subtilisin BPN’. Substitution of leucine for phenylalanine at the PI position results in a lower k,,,, although the K, was relatively unchanged. Similar results have been obtained for both cathepsin G and chymotrypsin [39] in which the same substitution resulted in a lower k,,, with little change in &,,The results are also consistent with the esterase data in which the same substitution resulted in a significant lowering of the k,,, with little change in K,,, (see Table 5). Hydrolysis of the oxidized insulin B chain The hydrolysis of the oxidized insulin B chain by Rt41A proteinase is shown in Fig. 5 for 15-min, 3-h and 24-h incubations. The first bond cleaved was Leul 5-Tyr16, and all these bonds were cleaved after a 15-min incubation at 70°C. After 3 h no Phel - Leul 5 peptide remained in the digest, the main peptide products indicating that the second preferred cleavage site was Gln4-His5. Although a significant proportion of the Tyrl6 -Ala30 peptide remained intact, cleavage at the Phe24Phe25 bond was also apparent in 3-h digests, suggesting this was the third most preferred site. After 24 h of incubation all the Phe24-Phe25 bonds were cleaved. Minor cleavage sites that emerged on 3-h incubation were Hiss-Leu6, Leu6-Cya7, Cya7-Gly8, Glu13-Ala14 and Phe25-Tyr26. After 24 h other minor sites identified were Leu1 1-Va112, Tyrl6-Leul7, Glu21Arg22, Tyr26-Thr27 and Lys29-Ala30. The results show that Rt41 A proteinase has a preference for amino acids with aromatic side groups at the P1’ side of the scissile bond. Minor cleavage sites show preference mainly for aliphatic amino acids at the P1’ site of the peptide. Exceptions were the cleavage of Leu6-Cya7 and Glu21-Arg22 bonds, indicating negatively and positively charged side chains can be accommodated at the S1‘ enzyme subsite.

The cleavage pattern that emerged is unique to Rt41A proteinase, but shows similarities to other microbial serine proteinases [7], the three major cleavage sites being common to both aqualysin I [7] and proteinase K [40]. In addition, however, Rt41A proteinase cleaved Leu6-Cya7 and Lys29Ala30 bonds, that were also cleaved by aqualysin I, but not proteinase K. The Alal4-Leu15 bond cleaved by Rt41A proteinase was not cleaved by either aqualysin I or proteinase K. The authors thank Dr. P. H. Janssen for the 5.0-1 Thermuscultures, Dr. D. L. Christie (University of Auckland) for the NH,-terminal amino acid analysis, Ms. D. Veitch for technical assistance and Mrs. T. Barea for typing the manuscript. This work was funded by Pacific Enzymes Limited, PO Box 1496, Hamilton, New Zealand.

REFERENCES 1 . Cowan, D. A,, Daniel, R. M. & Morgan, H. W. (1981) Znt. J . Biochem. 19,141 -143. 2. Cowan, D. A., Daniel, R. M. & Morgan, H. W. (1985) Trends Biotechnol. 3, 68 - 12. 3. Coolbear, T., Daniel, R. M. & Morgan, H. W. (1992) Adv. Biochem. Eng.lBiotechnol.45, 51 - 98. 4. Hudson, J . A,, Morgan, H. W. & Daniel, R. M. (1986) J. Gen. Microbiol. 132, 531 - 540. 5. Cowan, D. A. & Daniel, R. M. (1982) Biochim. Biophys. Acta 705,293 - 305. 6. Taguchi, H., Hamaoki, M., Matsuzawa, H. & Ohta, T. (1983) J . Biochem. (Tokyo) 93, 7-13. I . Matsuzawa, H., Tokugawa, K., Hamaoki, M., Mizoguchi, M., Taguchi, H., Terada, I., Kwon, S. T. & Ohta, T. (1988) Eur. J . Biochem. 171, 441 -447. 8. Saravani, G. A., Cowan, D. A., Daniel, R. M. & Morgan, H. W. (1989) Biochem. J. 262,409-416. 9. Matsuzawa, H., Hamaoki, M. & Ohta, T. (1983) Agric. Bid. Chem. 47.25-28. 10. Lim, S. H. (1983) Ph. D.thesis, University of Waikato. 11. Janssen, P. H., Morgan, H. W. & Daniel, R. M. (1991) Appl. Microbiol. & Biotechnol. 34, 189 -793. 12. McHale, R. H., Stapleton, P. M. & Bergquist, P. L. (1991) Biotechniques 10,20- 22. 13. Fung, M-C & Fung, K. Y-M., (1991) Nucleic Acids Res. 19,4300. 14. Coolbear, T., Monk, C., Peek, K., Morgan, H. W. & Daniel, R. M. (1992) J. Membr. Sci. 67, 93-102:

1044 15. McHale, R. H., Munro, G. K. L., Reeves, R . A. & Bergquist, P. L. (1 990) Proceedings of Ninth Australian Biotechnology Conference, pp. 296 - 301, University of Queensland Press, Brisbane. 16. Kwlon, S . T., Terada, I., Matsuzawa, H. & Ohta, T. (1988) Eur. J . Biochem. 173,491 -497. 17. Outtrup, H. & Boyce, C. 0. L. (1990) in Microbial enzymes and biotechnology, 2nd edn (Fogarty, W. M. & Kelly, C. T., eds) pp. 227 - 254, Elsevier Applied Science, London. 18. Cowan, D. A., Daniel, R. M. & Morgan, H. W. (1987) FEMS Microbiol. Lett. 43, 155- 159. 19. Lowry, 0. H., Rosenbrough, N. J., Farr, A. L. & Randall, R. J. (1951) J . Biol. Chem. 193,265-275. 20. Peterson, G. L. (1979) Anal. Biochem. 100,201 -220. 21. Reisfield, R. A., Lewis, U. J. & Williams, D. E. (1962) Nature 195,281-283. 22. References deleted. 23. White, C. A. & Kennedy, J. F. (1986) in Carbohydrate analysis: a practical approach (Chaplin, M. F. & Kennedy, J. F., eds) pp. 37 - 54, IRL Press, Oxford. 24. Ellman, G. L. (1959) Arch. Biochem. Biophys. 82, 70-77. 25. Pollitt, S . & Zalkin, H. (1983) J . Bacteriol. 153, 27-32. 26. Robyt, J. F., Ackerman, R. J. & Chittenden (1971) Arch. Biochem. Biophys. 147, 262. 27. Bidlingmeyer, B. A., Cohen, S. A. & Tarvin, T. L. (1984) J . Chromatogr. 336, 93 - 104. 28. Peek, K., Janssen, P. H., Morgan, H. W. & Daniel, R. M. (1990) in Fermentation technologies: industrial applications (Pak- Lam Yu, ed.) pp. 97- 102, Elsevier Applied Science, London.

29. Martin, R. G. & Ames, B. N. (1961) J . Biol. Chem. 236, 13721379. 30. Voordouw, G., Gaucher, G. M. & Roche, R. S. (1974) Biochem. Biophys. Res. Comrnun. 58, 8 - 12. 31. Dixon, M. &Webb, E. C. (1979) in Enzymes, 3rd edn, pp. 126136, Longman, London. 32. Cowan, D. A,, Daniel, R. M. & Morgan, H. W. (1987) Int. J . Biochem. 19,483-486. J J . Bajorath, J., Saenger, W. & Pal, G. P. (1988) Riochim. Biophys. Acta 954, 176- 182. 34. Mozhaev, V. V. & Martinek, K. (1984) Enzyme Microb. Technol. 6, 50 - 59. 35. Fersht, A. (1985) Enzyme structure and mechanism, 2nd edn, pp. 31 1 - 344, W. H. Freeman, New York. 36. Han, M. H. (1972) J . Theor. Bid. 35, 543 -568. 37. Voordouw, G. & Roche, R. S . (1975) Biochemistry 14, 46674673. 38. Delmar, E. G., Largman, C., Brodrick, J. W. & Geokos, M. C. (1979) Anal. Biochem. 99, 316-320. 39. Nakajima, K., Powers, J. C., Ashe, B. M. & Zimmerman, M (1979) J . Biol. Chem. 254,4027-4032. 40. Kraus, E., Kiltz, H. H. & Fempert, U. F. (1976) Hoppe-Seyler’s Z. Physiol. Chem. 357, 233 -237. QQ