Purification and properties of a brain enzyme which deiminates proteins

Biochimica et Biophysica Acta, 745 (1983) 285-291 Elsevier

285

BBA 31631

PURIFICATION AND PROPERTIES OF A BRAIN ENZYME WHICH DEIMINATES PROTEINS JOSEPH KUBILUS and HOWARD P. BADEN

Hart~ard Medical School, Department of Dermatology, Massachusetts General Hospital, Boston, MA 02114 (U.S.A.) (Received January 17th, 1982)

Key words: Protein deimination; Guanidyl group; (Brain)

T h e deimination of guanidyl groups of peptides, proteins and other arginine-containing compounds is catalyzed by enzymes found in mammalian brain and epidermis. In cow, the brain and epidermal enzymes differ kinetically and physically, but both may be quantitated by measuring the production of benzoyl citrulline ethyl ester from benzoyl-arginine ethyl ester. The brain enzyme has been purified to apparent homogeneity, as judged by the presence of only one 85000 dalton band in purified preparations when examined by SDS-polyacrylamide gel electrophoresis. An antibody raised to this band precipitates pure and partially purified brain enzyme but not partially purified epidermal enzyme, using the Ouchtedony technique. The antibody bound to an insoluble matrix removes brain enzyme activity from solution but not epidermal enzyme activity. The K m of the brain enzyme for benzoyl-arginine ethyl ester is about 0.33 raM.

Introduction The presence of citrulline-containing proteins in hair [1-3], epidermis [4,5] and a human myelin fraction [6] has been demonstrated. One mechanism by which citruUine occurs in the peptide linkage of proteins is through the action of an arginine-converting enzyme or peptidyl arglnine deiminase, which catalyzes the deimination of the guanidyl group to the ureido group with loss of ammonia. This enzyme was first identified in hair bulb [7,8], later partially purified from epidermis [9-11] and, most recently, partially purifed from skeletal muscle [12]. In a preliminary report [13] we detected this enzyme in bovine brain and described its partial purification. Differences in molecular weight and substrate specificities have r

.

Abbreviations: BAEE, a-N-bcnzoyl-arginine ethyl ester; BA, a-N-benzoyi-L-arginine; BAA, a-N-beazoyI-L-arsinine amide; BAPNA, a-N-benzoyl-L.ar~nine-p-nitroanilide; TAME, a-Np-tosyi-L-arginine methyl ester; BCEE, benzoyl-citrulline ethyl ester. 0167-4838/83/$03.00 © 1983 Elsevier Science Publishers B.V.

suggested that the epidermal and muscle enzymes are not identical [ 10-12]. This work describes the tissue distribution of peptidyl arginine deiminase activity, the purification to apparent homogeneity of the enzyme from whole brain, the preparation of an antibody to the enzyme, and a comparison of some kinetic and physical properties of the isolated brain enzyme with those of the partially purified epidermal enzyme. Materials and Methods a-N-Benzoyl-L-arginine ethyl ester (BAEE), aN-Benzoyl-L-arginine (BA), a-N-benzoyl-Larginine amide (BAA), a-N-benzoyl-L-arg~aine-pnitroanilide (BAPNA) and a-N-p-tosyl-L-ar~fmine methyl ester (TAME) were purchased from Sigma Chemical Co., St. Louis, MO. a-N-[ethyl-23H]Benzoyl-L-arginine ethyl ester hydrochloride ([3H]BAEE) was purchased from New England Nuclear, Boston, MA. Arginine-containing peptides and guanidine group-containing com-

286

pounds tested as potential enzyme substrates were acquired from Vega Biochemicals, Tucson, AZ or Research Plus Laboratories, Denville, NJ. Cyanogen bromide-activated Sepharose 4B, Sephadex G-200 and DEAE Sephadex A-50 were purchased from Pharmacia Fine Chemicals, Piscataway, NJ. Arginine agarose and lysine agarose were purchased from Bethesda Research Laboratories, Gaithersberg, MD. Hydroxyapatite-HT was purchased from BioRad Laboratories, Richmond, CA. The silica gel used was Bakerflex IB-2, purchased from J.T. Baker Chemical Co., Phillipsburg, NJ. All other chemicals were either of reagent grade or of the highest quality available.

Extraction of enzyme To determine the organ and tissue distribution of the enzyme, rat and bovine tissues were homogenized with a Waring blender in a buffer containing 2 mM EDTA/5 mM Tris, pH 7.5. After centrifugation at 30000 × g for 30 min, extracts were assayed for enzyme activity. For purification, enzyme was extracted from bovine brain by homogenization for 2 min in a Waring blender in a buffer containing 0.5 mM dithiothreitol/3 mM MgCIz/50 mM Tris, pH 7.5 (buffer 1), at a ratio of 3 ml per g of tissue. After centrifugation at 30000 × g for 30 min, a clear solution was obtained. Buffer 1 was used throughout purification since it stabilized enzyme activity. Assay of enzyme The amount of enzyme was proportional to the rate of conversion of BAEE to benzoyl-citrulline ethyl ester (BCEE). Quantitation of activity was done similarly to the method of Sugawara and Fuzisaki [9] in a volume of 1.6 ml, which was 10 mM in BAEE and CaCI 2, 2 mM in dithiothreitol, 50 mM in Tris-HCl, pH 7.5, and contained up to 1.0 ml of enzyme solution. Mixtures were sealed under nitrogen and incubated at 50°C for 20 h. The reaction was stopped by the addition of 0.4 ml 5 M perchloric acid and the mixtures were chilled on ice for 10 min and then centrifuged at 5000 x g for 10 min. Up to 1.0 ml of supernatant was analyzed for ureido groups by the method of Guthohrlein and Knapp [14], using citrulline as a standard. Ureido concentrations of 0.10 mM were

measurable by this technique. 1 unit of activity is that amount of enzyme which catalyzes the formation of 1 /~mol of BCEE per h at 50°C in the above assay. Specific activity was defined as units of activity per mg of protein as determined by Bradford [ 15].

Substrate specificity Other substrates or potential substrates were assayed under the same conditions as described above at concentrations of 1 or 10 mM using highly purified brain enzyme or partially purified epidermal enzyme as described previously [10]. When proteins were used as substrates, reaction mixtures were hydrolyzed in 6 N HCI and citrulline was measured using an amino acid analyzer (Beckman, model 116). K,, determination and microassay The microassay used to determine g m was similar to the assay described above, except that [3H]BAEE was used at a specific activity of 100- I0 nCi/nmol at varying concentrations; the volume of the reaction mixture was 0.40 ml and the incubation time was 1 h. The reaction was stopped by bringing the pH to 4 by the addition of 5 M acetic acid. 20 #1 of reaction mixture was streaked on silica gel along with 20 ~1 of a marker solution which was 5 mM in both BAEE and BCEE. Chromatograms were developed in CHCI3/MeOH / 17% NH,OH (2:2: 1). Spots were visualized by viewing in ultraviolet light or by spraying with p-dimethylaminocinnamaldehyde [3]. The BAEE and BCEE spots were both scraped from the chromatogram, incubated at 37°C for 30 min with 1.0 ml of HzO, and, after the addition of 10 ml of Aquasol II, counted in a liquid scintillation counter. The ratio of cpm in the BCEE and BAEE spots was the percent conversion from which the reaction rates were calculated. Immunological procedures Purified enzyme was run by SDS-polyacrylamide gel electrophoresis [16] in 1.5-mm slab gels which had been poured at least 24 h previously. The position of the 80000-85000 Mr band was determined by silver-staining [17] one track and the corresponding position was excised from the unstained tracks. The slices were passed through

287

teflon-coated wire mesh and then extracted with 50 mM Tris, pH 8.3, with 1% SDS overnight at room temperature. The extracts were dialyzed against 5 mM Tris, pH 8.3, and lyophilized; the SDS was removed by ion-pair extraction with the triethylamine solvent [18]. Protein was dissolved in a small amount of saline and diluted 1:i with Freund's complete adjuvant. The backs of rabbits were injected subcutaneously with 500 pg of protein and then boosted biweekly with 100 #g of protein for 6 weeks. 5 ml of antisera were dialyzed against 5 mM phosphate, pH 6.5, and then stirred gently for 1 h with 1.0 g of DEAE-cellulose which had been equilibrated in the same buffer. The suspension was centrifuged at 35 000 × g for 30 rain and the supernatant containing 35 mg of purified antibody was diluted 1 : 1 with 0.1 M NaHCO 3, pH 7.8, and stirred gently for 2.5 h with 2 g of CNBr-activated Sepharose 4B. The gel was washed with buffer and contained 70-80% of the applied antibody protein. Results

Enzyme distribution A number of organs and tissues from the rat and cow were analyzed for enzyme activity and

T A B L E I1 E N Z Y M E LEVELS IN BRAIN A R E A S Dissected areas of bovine brain were homogenized in buffer 1 and assayed with BAEE. Activity is expressed as units per g of tissue (wet weight). Brain area

Activity

(units/g) White matter Grey matter Cerebellum Hypothalamus Spinal cord Stem

3.8 4.2

0.9 4.7 2.8 2.4

these results are summarized in Table I. Highest levels occurred in brain, with lesser amounts in epidermis, kidney and lung. No activity was detected in heart, spleen, stomach, pancreas, serum, or large and small intestines. Since whole bovine brain contained by far the most activity on a per g basis we assayed various anatomical regions of this organ and these results are summarized in Table II. Low levels of activity were seen in the cerebellum, but the other regions were approximately equivalent.

TABLE 1 E N Z Y M E C O N T E N T O F O R G A N S A N D TISSUES Organs were homogenized in buffer I or in 5 m M T r i s / 2 M E D T A and were analyzed for enzyme content using BAEE as described. Activity is expressed as units per g of wet weight of tissue obtained from either rat or cow.

Organ o r tissue Bovine Epidermis Brain Rat Brain

Kidney Lung Spleen Stomach Small intestine Large intestine Pancreas Serum Heart

Activity (uniLVs) 0.53 3.30

A

Lktits/ ml ...... )

2SOem (--}

NeCI (M) (---)

t2-

e/

a

-~6

t0.6

-44

0.4

-43

-qO.3

4

2 1.90

0.]8 0.07 < 0.02 < 0.02 < 0.02 < 0.02 < 0.02 < 0.02 < 0.02

o

o

to

2O

3O

4O

5O

6O

I-Jo 7O

0

TIME OF ELUTION (Howt) Fig. 1. Chromatography on DEAE-cellulose. Crude extract in buffer I was appfied to a 5 × 50 cm column of DEAE-oellulose at 300 m l / h . "me column was washed with buffer l containin 8 35 m M NaCI and, after the flow rate was cut to 80 m l / h , sample was eluted with a linear gradient of 0-0.2 M NaCL 20-ml fractions were collected and aliquots were analyzed for enzyme activity.

288 ,4

Units/

28Ohm (--)

ml (.........)

NoCl

(M) (---)

i

I0

05

3

6

03

2

4

02

t

2

01

$0 0

IO

0 oI

s

ts

2o

2S

30

3S

4O

4S

TIME OF ELUTiON (HQws)

Fi s. 2. Chromatography on DEAE-Sq~hedex. Pooled enzyme from DEAE-ccilulos¢ was applied at 40 m l / h to a 2.5 x 40 cm column of DEAE-Scphadex. After washing with buffer i containing 90 m M NaCI, the sample was eluted with a 90-300 m M gradient. Protein was eluted in 20-ml fractions.

Purification of enzyme DEAE-cellulose. The buffer 1 extract of four fresh calf brains was applied to a DEAE-cellulose column at a high flow rate so that the large volume of material (2000 ml) could be processed quickly (Fig. 1). The linear NaCI gradient eluted the activity at 90 mM as a sharp peak, and tubes containing the enzyme were pooled. DEAE-Sephadex. DEA E-cellulose-purified material was applied directly to DEAE-Sephadex which had been equilibrated in buffer 1 containing

,,4 28Onto ( )

Unils / ml (-.--...-)

PHOSPHATE (M) (---)

100 mM NaCI. The linear NaCl gradient eluted the enzyme as a broad peak at 180 mM (Fig. 2). Neither changing the gradient slope nor working at a different pH improved the yield or specific activity.

Hydroxyapatite Pooled fraction from the DEAE-Sephadex were dialyzed against buffer 1 made 100 mM in KCI and adjusted to pH 6.9 and then applied to a hydroxyapatite column equilibrated,in the same buffer. Protein was eluted from the column by washing with 100 mM potassium phosphate and 500 mM potassium phosphate in starting buffer (Fig. 3). Eluting the enzyme with a phosphate gradient resulted in large dilutions and prohibitive losses of enzyme activity.

Arginine agarose Enzyme-containing tubes from the hydroxyapatite column were pooled and dialyzed into buffer 1 containing 100 mM NaCI and applied to an arginine agarose column. After washing with starting buffer, stepwise NaCI elution was performed and enzyme activity was recovered at 350 mM NaCI (Fig. 4).

Lysme agarose Arginine agarose-purified enzyme was dialyzed into buffer 1 and applied to a lysine agaros¢ column. Stepwise washes with buffer 1 containing

,4 280 am 6

12

4

(m)

~ l I

,~l h ~ 2

.

10

(___)

t2

10.6

0.2

t2

',4

16

t~

20

22

0

0.4

TiME OF ELUTION (Hours)

Fig. 3. Chromatography on hydroxyapatite. Enzyme in buffer I at pH 6.9 containing 100 m M KCI was applied to a 1.5 × 2 0 cm column of hydroxyapatite at 24 m l / h . The column was washed with starting buffer until baseline was reached, and then with 3 column volumes of starting buffer I containing 0.1 M potassium phosphate. Enzyme was ehited g~th buffer 1 containing 0.5 M potassium phosphate. Fractions of 12 mi were collected and aliquots from them were dialyzed against buffer I and

analyzed for activity.

jo

NoCI (M)

(.~)

0.4

2 0

0.6

Units / ml

0.2 0

2

4

6

8

~

I t2

0

TIME OF ELUTION ( Hovre}

Fig. 4. Chromatography on ar~inine asaros¢. E.nzym¢ in buffer 1 containing 0.10 NaCi (50-70 mJ) was applied to a 1.5 x 70 cm column of equilibrated as~i'nine agarose at 12 m l / h . The enzyme was eluted stepwis¢ with 0.20 and 0.35 M NaCI and 30-rain fractions were collected.

289

A

280 am

(

Units/

NoCI

mt

(U)

(........ }

)

0.8 0.6

:-:

0.4

0.2

0

,-./,'%. 0

2

4

6

O

i

to

(---)

16

10.20

t2

--40.t5

8

tO.tO

4

10.05

o

o

TIME OF ELUTION (Hours)

Fig. 5. Chromatography on lysine agarose. Sample in buffer 1

was appfied to a 0.9x 10 cm column of lysine agarose at 5 m l / h . Enzyme was eluted by washing with 0.10 M NaCI in buffer I.

TABLE IV SUBSTRATE SPECIFICITY OF BRAIN A N D EPIDERMAL ENZYMES Reaction rate is e x p r ~ as t~mol of ureido group produced per h per g of brain or epidermis (wet weight) for each substrate at 50°C. Substrate

BAEE BAEE BA BA BAA BAA BAPNA TAME Met-Arg-Phe Met-Arg-Phe

Bradykinin Met-Arg-Phe-Ala

NaCi eluted the enzyme at 100 mM salt, as seen in Fig. 5. Purification of the protein-deiminating enzyme from bovine brain is summarized in Table IIl. A purification of 1100-fold with a yield of 4% was achieved.

Enzyme purity The active peak from lysine agarose was assessed for purity by SDS-gel electrophoresis. The preparation consisted of a single band with a molecular weight of about 85 000. In some preparations an indistinctly defined component in the molecular weight range of 55 000-65 000 was also seen. When either purified or unpurified brain enzyme was chromatographed on Sephadex G-200,

Concentration (mM)

Reaction rate

Brain

Epidermis

10 1 10 1 10 I 1 10 10 1 1 10

3.30 0.88 0.05 < 0.01 2.30 0.44 < 0.01 0.35 0.22 < 0.01 0.05 0.22

0.55 0.22 1.05 0.15 0.64 0.10 < 0.01 0.12 2.10 0.18 < 0.01 0.05

the time of elution of the activity indicated an apparent molecular weight of 125 000.

Immunology When purified enzyme preparations were tested by the Ouchterlony technique against the antibody prepared to the enzyme, precipitin lines were seen when the enzyme concentration was 2 units per ml or greater and 0.014 units were placed in the wells. Lines were also seen with SDS-treated enzymes. Enzyme activity was strongly absorbed to antibody-coupled Sepharose 4B when partially purified preparations were applied in buffer 1 containing 150 mM NaC1. Activity could not be eluted with buffer I containing either NaCI or (NH4)2SO4

TABLE 1I! PURIFICATION O F BRAIN PEPTIDYL A R G I N I N E DEIMINASE Purification step

Volume (ml)

Protein (mg)

Activity (units)

Crude DEAE-Cellulose DEAE-Sephadex Hydroxyapatite Argi'nlne -~=rose

2 000 450 200 60 24 10

32 000 5 360 740 123 9 1.1

2 200 1270 650 336 128 85

Specific activity

Yield (%)

(units/rag)

Lysine agarose

0.07 0.24 0.88 2.73 14.2 77

58 30 i5 6 4

290 at 3 M. When the same experiment was done using Sepharose 4B coupled to IgG purified from preimmune serum no absorption occurred. Partially purified epidermal enzyme did not bind to the antibody-coupled gel, with over 90% of the activity passing directly through the column. Enzyme substrates

A number of arginine-containing compounds were shown to be deiminated by both brain and epidermal enzymes, as shown in Table IV. Peptides which at 10 mM were not deiminated by either the brain or epidermal enzyme include Arg-Phe-Ala, Arg-Glu, Arg-Val, Arg-Tyr, and Arg-Ala. At 1 mM, the highest level tested, Leu-Trp-Met-ArgPhe-Ala, Leu-Trp-Met-Arg-Phe and Arg-Pro-ProArg were not substrates of either enzyme. Bovine S-carboxymethylated prekeratin and myelin basic protein were good substrates of the enzyme, with 25 and 15~, respectively, of their arginines susceptible to enzymatic deimination. In an earlier paper, we reported that the semipurified epidermal enzyme was active against many proteins and large peptides, but unreactive toward all small peptides and arginine derivatives, including BAEE and BAA [10]. Subsequent to this, it was reported that these compounds were very good substrates of the newborn rat epidermal enzyme [9]. When we used the assay described by Sugawara and Fuzisaki [9], we found that many of these compounds were substrates not only for the rat epidermal enzyme, but also for the bovine epidermal and brain enzymes. The problem with our initial assay was related to the choice of reducing agent. With small substrates, neither cysteine nor fl-mercaptoethanol at concentrations of up to 10 mM resulted in consistently measurable reaction rates. If dithiothreitol between 1 and 5 mM was used, relatively high reproducible rates were obtained, although at a concentration of 10 mM reaction, rates of 30-50~ less than optimum were observed. This rate reduction was greater at higher dithiothreitol concentrations. At dithiothreitol concentrations of less than 1 raM, reduced or zero reaction rates were seen. K m determination

Enzyme activities were measured at BAEE concentrations of 1.0, 0.10, 0.045, 0.030 and 0.024

mM, using the microassay described in Materials and Methods. Under these conditions less than 5~ of the available BAEE was converted to BCEE, and at the lower substrate concentrations BCEE levels were less than 2/~M. Since the colorimetric assay was not sufficiently sensitive to detect this amount of BCEE, it was necessary to employ the microassay. At high BCEE levels there was good agreement between the two assays. The enzymes obeyed Michaelis Menton Kinetics and a K m of 0.33 mM was calculated. This g m value is consistent with observations that over 50% of the BAEE could be converted to BCEE in the standard assay without apparent reduction in the reaction rate. The same results were obtained for enzyme purified through the hydroxyapatite, arginine agarose or lysine agarose steps. Discussion

Although citrulline is not incorporated into proteins at the translational stage of biosynthesis, there is now overwhelming evidence that citruUine may be found in the peptide linkage of certain proteins of hair follicle, epidermis and nerve tissue [1-8]. The mechanism by which citruLline residues arise in proteins is the post-translational enzymatic deimination of arginyl residues. The enzyme responsible for this activity has been found in eukaryotes and is active against proteins, peptides and other arginine derivatives, but not the free amino acid [7-11]. An enzyme with a similar activity is arginine deiminase (EC 3.5.3.6), which has been found in prokaryotic sources and, in contrast to the eukaryotic enzyme, is active against free arginine but not against a-amino- or acarboxyl-substituted arginine derivatives [19]. As can be seen from Table IV, brain and epidermal enzymes differ in specificity. Arginine-derived compounds must be blocked at the a - N position to be substrates of either enzyme and the nature of both the a-N-blocking groups as well as groups substituted at the a-C position influence rates of enzymatic deimination. In addition to specificity differences, one also sees a difference in physical properties. When apparent molecular weights were determined with the same Sephadex G-200 column, M r values of 125000 for the brain enzyme and 69 000 for the epidermal enzyme were

291 obtained. The value for brain enzyme is similar, though not identical, to the molecular weight of 115000 seen for rabbit muscle enzyme [12]. Additional differences between the skin and brain enzymes were evident in their immunological responses. When the antibody raised to the polypeptide eluted from SDS-gels was linked to a solid support, it quantitatively removed brain enzyme from solution. This demonstrated that the isolated 85 000 dalton polypeptide, when undenatured, was the source of protein-deiminating activity in bovine brain. The inability of the antibody to form a precipitin line with epidermal enzyme or to remove its activity from solution was further evidence that peptidyl arginine deiminase activity may be associated with different tissue specific proteins. Table 1 shows that brain and skin are major sites of deiminating activity. Sugawara et al. [12] has also reported an enzyme content for rabbit skeletal muscle which is about 2-fold that measured for bovine brain using BAA as a substrate. The enzyme activity of bovine brain is 6-fold greater than that of bovine epidermis, which is markedly different than the 100-fold difference seen by Sugawara and co-workers [11,12] between the rabbit muscle and newborn rat epidermal enzymes. It is possible that the tissue distribution of the enzyme may differ from species to species. A n y such variation may be difficult to measure accurately owing to substrate-specificity differences between enzymes from different sources. The function of this enzyme is unclear at present. It does appear that structural proteins of specialiTed tissues, including hair, epidermis and nerve, are major in vivo substrates for these enzymes [4-8,20]. Other substrates may exist, but could be difficult to detect if present in small amounts. It should be noted that conversion of an arginine residue to a citrulline residue could radically change the biochemical behavior of a protein

or peptide. This could occur by a change in conformation or by modification of a binding site, and c o u l d b e of major physiological consequence. Acknowledgements This work was supported by N.I.H. grant No. AM06838.

References I Rogers,G.E. (1962) Nature 194, 1149-1151 2 Steinert, P.M., Harding, H.WJ. and Rogers, G.E. (1969) Biochim. Biophys.Acta 175, I-9 3 Harding, H.W.J. and Rogers, G.E. (1976) Biochim. Biophys. Acta 427, 315-324 4 Kubilus, J., Waltkus, R.W. and Baden, H.P. (1979) Biochim. Biophys.Acta 581, 114-121 5 Sugawara, K. (1979) Agric. Biol. Chem. 43, 2215-2217 6 Finch, P.R., Wood, D.D. and Moscarello, M.A. (1971) FEBS Lett. 15, 145-148 7 Rogers, G.E., Harding, H.W.J. and Llewellyn-Smith, 1.J. (1977) Biochim. Biophys. Acta 495, 159-175 8 Rogers, G.E., and Taylor, L.D. (1977) in Advances in Experimental Medicine and Biology(Friedman~ M., ed.), Vol. 86A, pp. 283-294, Plenum Publishing Corp., New York 9 Sugawara, K. and Fuzisaki, M. (1979) Agric. Biol. Chem. 43, 2407-2408 10 Kubilus, J., Waitkus, R.F. and Baden, H.P. (1980) Biochim. Biophys. Acta 615, 246-251 11 Fujisaki, M. and Sugawara, K. (1981) J. Biochem. 89, 257-263 12 Sogawara, K., Oikawa, Y. and Ochi, T., (1981) J. Biochem. 91, 1065-1071 13 Kubilus, J. and Baden, H.P. (1982) Fed. Proc. 41, 2209 14 Guthohrlein, G. and Knapp, J. (1968) Anal. Biochem. 26, 188-191 15 Bradford, M. (1976) Anal. Biochem. 72, 248-254 16 Laemmli,U.K. (1970) Nature 227, 680-685 17 Merril, C.R., Goldman, D., Sedman, S.A. and Ebert, M.H. (1981) Science 211, 1437-1438 18 Henderson, L.E., Oroszlan, S. and Konigsberg, W. (1979) Anal. Biochem. 93, 153-157 19 Smith, D.W., Ganaway, R.L. and Fahrney, D.E. (1978) J. Biol. Chem. 253, 6016-6020 20 St, inert, P.M. and Idler, W.W. (1979) Biochemistry 18, 6554-5669