Reversible modification of arginine residues. Application to sequence studies by restriction of tryptic hydrolysis to lysine residues.

1, 2-Cyclohexanedione reacts specifically with the guanidino group of arginine or arginine residues at pH 8 to 9 in sodium borate buffer in the temperature range of 25-40 degrees. The single product, N-7, N-8-(1,2-dihydroxycyclohex-1,2-ylene)-L-arginine (DHCH-arginine) is stable in acidic solutions and in borate buffers (pH 8 to 9). DHCH-Arginine is converted to N-7-adipyl-L-arginine by periodate oxidation. The structures of the two compounds were elucidated by chemical and physicochemical means. Arginine or arginyl residues can be regenerated quantitatively from DHCH-arginine by incubation at 37 degrees in hydroxylamine buffer at pH 7.0 FOR 7 TO 8 hours. Analysis of native egg white lysozyme and native as well as oxidized bovine pancreatic RNase, which were treated with cyclohexanedione, showed that only arginine residues were modified. The utility of the method in sequence studies was shown on oxidized bovine pancreatic ribonuclease A. Arginine modification was complete in 2 hours at 35 degrees in borate buffer at pH 9.0 with a 15-fold molar excess of the reagent. The derived peptides showed that tryptic hydrolysis was entirely limited to peptide bonds involving lysine residues, as shown both by two-dimensional peptide patterns and by isolation of the resulting peptides. The stability of DHCH-arginyl residues permits isolation of labeled peptides.

SUMMARY 1,2-Cyclohexanedione reacts specifically with the guanidino group of arginine or arginine residues at pH 8 to 9 in sodium borate buffer in the temperature range of 25-40". The single product, W, N8-(1,2-dihydroxycyclohex-1,2ylene)-L-arginine (DHCH-arginine) is stable in acidic solutions and in borate buffers (pH 8 to 9). DHCH-Arginine is converted to N7-adipyl-L-arginine by periodate oxidation. The structures of the two compounds were elucidated by chemical and physicochemical means. Arginine or arginyl residues can be regenerated quantitatively from DHCHarginine by incubation at 3 '7" in hydroxylamine buffer at pH 7.0 for 7 to 8 hours.
Analysis of native egg white lysozyme and native as well as oxidized bovine pancreatic RNase, which were treated with cyclohexanedione, showed that only arginine residues were modified.
The utility of the method in sequence studies was shown on oxidized bovine pancreatic ribonuclease A. Arginine modification was complete in 2 hours at 35" in borate buffer at pH 9.0 with a S-fold molar excess of the reagent. The derived peptides showed that tryptic hydrolysis was entirely limited to peptide bonds involving lysine residues, as shown both by two-dimensional peptide patterns and by isolation of the resulting peptides. The stability of DHCH-arginyl residues permits isolation of labeled peptides.
In the last decade a number of bifunctional aldehydes and ketones have been tested for their ability to modify selectively arginine residues in proteins either for the purpose of blocking the hydrolytic action of trypsin at arginyl residues or for mod&ation of arginyl residues in native proteins.
Treatment of proteins with benzil (1)) cyclohexanedione (2)) nitromalondialde-* This investigation was aided by Grant GM 11061 from the National Institute of General Medical Sciences, United States Public Health Service.
$ Present address, Institute of Biochemistry, Biological Research Center, Hungarian Academy of Sciences, Szeged, Hungary. 8 To whom inquiries and requests for reprints should be sent.
Unfortunately, all of these procedures may result in significant side reactions with a-or t-amino groups.
To prevent these side reactions the amino groups have to be blocked prior to arginine modification (5,9,(12)(13)(14). An additional problem is caused by the multiple products arising when butancdione (14,15) or cyclohexanedione (5) is used for arginine modification. In most cases arginine modification is irreversible, a major disadvantage in protein-sequencing studies.
In other instances, such as the reaction of phenylglyoxal with arginine (11) , rcgeneration to arginine occurs spontaneously in neutral and alkaline solutions.
This instability of the product places limitations on its use in studies of sequences and for specific labeling.
2,3-Butanedione may also form a product with arginine which regenerates spontaneously to free arginine in the absence of excess reagent (15).
The various problems that have been encountered with the above methods impose limitations on their use in protein st,ructural studies.
In this paper we describe the reaction of 1,2-cyclohexancdione with arginine and arginyl residues of proteins at pH 8 to 9 in borate.
Apart from the mild conditions used, the method has several additional advantages. Under the conditions specified, cyclohexanedione reacts only with arginine residues. Arginine is converted quantitatively to a single product, DHCH-arginine,' which is stable in acidic solutions and in borate buffers, allowing isolation and sequence determination of peptides by conventional methods.
Under controlled conditions (neutral hydroxylamine or hydrazinc solutions), arginine or arginyl residues can be regenerated quantitatively.
The experiments described herein were designed to elucidate the chemical nature of the modified argininc and to est.ablish the optimal conditions for its formation and regeneration.
The method has been tested in protein sequence work and in enzymological studies, as described herein and in the following paper (16  Aliquots were removed at intervals and diluted into cold 30y0 acetic acid for analysis by paper chromatography and paper electrophoresis. The rates of reaction in 0.2 M sodium borate buffer at pH 9.0 were determined at 25", 30", 35", and 40' with 0.15 M 1,2-cyclohexanedionc and 0.025 M L-arginine. Aliquots were removed into cold 30% acetic acid, lyophilized, and analyzed on the short column of the automatic analyzer. Preparation of ilT7,A;8-(1 ,2-DihUdroxycyclohex-1,2-ylene)-L-arginine Hydrochloride (DHCH-Arginine)-L-Arginine-HCl (1.05 g, 5 mmol) was dissolved in 100 ml of freshly prepared 0.2 M sodium borate buffer at pII9.0 and 0.67 g (6 mmol) of 1,2-cyclohexanedione were added; the reaction mixture was kept under Nz at room temperature.
After 24 hours the solution was concentrated to about 20 ml by rotary evaporation and adjusted to pH 3 with HCl. The preparation was desalted and purified by gel filtration on Sephadex (i-10 columns (2 X 150 cm) in 0.01 M IICI at a flow rate of 40 ml per horn.
Five-milliliter fractions were collected and 5-~1 aliquots were removed for analysis by high voltage electrophoresis at pI1 4.7 and paper chromatography. The product was bound to an SP-Sephadex column (2 X 50 cm) in 0.01 s I-ICI, washed with 0.01 N HCl to remove the salts, then eluted with 0.5 M sodium chloride in 0.01 N HCl.

Fractions
of 5 ml were collected and analyzed as in the case of DHCH-arginine.
Samples containing the amino acid derivative were pooled and desalted by gel filtration on Sephadex G-10 columns (2 X 150 cm) in 0.01 h% HCl at a flow rate of 40 ml per hour. Fractions containing pure adipylarginine were pooled and lyophilized.
The The co&se of reaction -was followed, and the products analyzed as described above.
The l.%cvclohexanedione dioxime formed during the reaction wit,11 hydroxylamine was detected by spraying the electrophoretograms with 0.01 M NiCl, solution.
The dioxime forms a red complex with nickel salts (19).
The rate of decomposition of modified arginine (0.1 M) in 0.5 M hydroxylamine at pII 7.0 was followed by the quantitative determination of arginine and DHCH-arginine on the short column of the amino acid analyzer.
Oxidation of DHCH-arginine was performed with 0.02 M sodium metaperiodate in 0.2 M sodium acetate at pII 4.0. lleaction was allowed to proceed for 16 hours at 0" in the dark.

Reaction
of Cyclohezunetlione with Proteins-Bovine pancreatic ribonuclease A was oxidized by performic acid by the method of Hirs (20) prior to arginine modification. The oxidized enzyme (2 PmoI per ml) was treated with 0.15 M 1,2-cyclohexanedione in 0.25 1\1 sodium borate buffer at pH 9.0 in a sealed vial at 35" for 2 hours.
The final concentrations are given; thus the reagent is present at 75 times the concentration of enzyme. An equal volume of 3055 acetic acid w-as then added and the protein was dialyzed against 15C/G, 7.5y,, and lcjG solutions of acetic acid in the cold.
After freeze-drying, the sample was dissolved in 0.1 M SOdium borate at pI1 8.0 at a concentration of 1 pmol per ml and hydrolyzed with trypsin at 37" for 4 hours. The digestion was arrested by adding 3Oo/0 acetic acid; samples were withdrawn for peptide analysis and carboxypeptidase B treatment. The tryptic hydrolysate of 7 pmol of oxidized RNase was applied to a Sephadex (i-50 column (3.8 x 150 cm) equilibrated with 30% acetic acid and eluted with t,he same solvent at a flow rate of 60 ml per hour.
Fractions were further purified by electrophoresis at pH 1.9 or paper chromatography.
Modification of native lysozyme was performed as in the case of RNase, but the enzyme and reagent were incubated at 25" for 20 hours.

Reaction of Cyclohexanedione with Arginine
The pI1 of the react,ion mixture of 1,2-cyclohexancdione and argininc has a dramatic effect on the nature and number of products formed, as shown in Fig. 1. ,4s described previously (a), in strong alkali a single major product, CIlD-arginine, is formed, whereas at slightly less alkaline pII additional products appear, e.g. at pH 10 to 11 (triethylamine buffers) as many as five different products are present in substantial amounts. At pH 7.5 to 9.0, however, only a single product is formed. Since homogeneity of product is a desirable attribute of modification reactions, ne investigated in considerable detail the conditions required for the quantitative formation of this product. Tris buffers cannot be used inasmuch as Tris reacts with 1,2-cyclohexanedione, thereby preventing arginine modification. Initial experiments showed that the rate of reaction was highest at pH 9.0 in sodium borate buffer.
The rate of reaction in this buffer was studied at different temperatures.
As shown in Fig.  2, the reaction is first order with respect to disappearance of arginine.
Furthermore, there is almost a IO-fold decrease in the half-time of the reaction on going from 25-40" (Fig. 2, inset) with no change in the nature of the reaction product, as demonstrated by various chromatographic studies. The inset shows the change of half-time of reaction with temperature.

Description of DHCH-Arginine
Unlike CHD-arginine, the product is a strong base; its relative electrophoretic mobility (76% of that of arginine) is the same from pH 1.9 to pH 8.0 as shown in Fig. 3. Titration of the modified group was not feasible, since at higher pH values the product undergoes rapid decomposition and secondary reactions (see below).
DHCH-Arginine is eluted from the short column of the amino acid analyzer between ammonia and arginine, further indicating its basicity.
Unlike CHD-arginine (2), the determination of DHCH-arginine does not require special conditions for separation from other amino acids.
Based on modification and regeneration experiments (hydroxylamine), the ninhydrin color value of DHCH-arginine is identical with that of arginine. In paper chromatography (1-butanol-pyridine-acetic acid-water) the RF value of DHCH-arginine is 0.34 as compared to 0.23 for arginine and 0.46 for CHD-arginine (Fig. 1). Upon hydrolysis in 6 N HCl at 110" for 24 hours the product is destroyed.
There is an 18 to 20% regeneration of arginine; the remainder is converted to unknown basic products that are not eluted from the amino acid analyzer under normal conditions. If acid hydrolysis is carried out in the presence of excess mercaptoacetic acid (20 pl), DHCH-arginine is converted to a neutral product, which is eluted after the aromatic amino acids on the short column.
Since under these conditions no arginine regeneration occurs, the amount of arginine in modified peptides and proteins was always determined after hydrolysis in the presence of mercaptoacetic acid. Alkaline hydrolysis in 2.5 M NaOH at 110" for 24 hours results in the formation of ornithine, citrulline, and l-amino-l-cyclopentanecarboxylic acid as identified by paper chromatography, paper electrophoresis, or amino acid analysis. Formation of l-amino-l-cyclopentanecarboxylic acid is due to the fact that in strong alkali, DHCH-argininc forms CHD-arginine (see below), which has been shown to yield I-amino-l-cyclopentanecarboxylic acid (2).
Studies on the stability of DHCH-arginine in neutral and weakly alkaline solutions indicate that it decomposes to regenerate cyclohexanedione and arginine (Scheme I). The secondary reactions yield the same product (or products) as those observed when the arginine modification is carried out under the same conditions. Thus in triethylamine buffer at pH 11.0 the five characteristic products appear (see Amino acids were determined on the amino acid analyzer.
arginine is formed.
In the latter case the progress curve clearly shows this secondary reaction (Fig. 4).
As expected, at pH 8.0 to 9.0, the liberated arginine does not yield new products.
The rate of decomposition to free arginine increases with pH and temperature and is also strongly dependent on the chemical nature of the buffer. The decomposition is more rapid in buffers that react with 1,2-cyclohexanedione (Tris, hydrazine, hydroxylamine) and buffers that are strong nucleophiles (hydroxylamine, hydrazine, Tris, ammonium). This is probably due to nucleophilic catalysis of decomposition and trapping of liberated cyclohexanedione. Trapping of cyclohexanedione explains why no secondary products appear even at pH 10 to 11, when hydroxylamine or hydrazine solutions are used. Since the pH of these solutions does not seem to affect greatly the rate of decomposition, we used hydroxylamine buffer at pH 7.0 for regeneration of arginine. On customary semilogarithmic plots, the decomposition of DHCH-arginine is first order.
DHCH-Arginine is remarkably stable in sodium borate buffers at pH 8.0 to 9.0, since no arginine is liberated over a 24.hour period at room temperature or 37", whereas in sodium phosphate buffer at pH 8.0 there is 10% (room temperature) and 40% (37") regeneration in 24 hours. The reason for this unusual stability is that borate forms a complex with the product, thereby stabilizing it (see below). The exceptional stability of DHCHarginine in borate buffer also explains the high rate of formation of this product in this buffer.
DHCH-Arginine is also stable in acidic solution, e.g. no detectable regeneration occurs in 30% acetic acid at 25-37" over a 24.hour period.

Characlerization of DHCH-Arginine
Complex formation of DHCH-arginine with borate was shown by a change in its RF value (0.27 instead of 0.35) when boric acidimpregnated papers (21) were used for paper chromatography in I-butanol-pyridine-acetic acid-water; the RF values of other amino acids were unaffected.
In buffers containing 0.1 M boric acid the relative electrophoretic mobility of DHCH-arginine, in agreement with the introduction of a negative charge, decreases significantly (Fig. 3, and Scheme I).
The effect of complex formation on the stability and rate of formation of DHCHarginine has been discussed above.
DHCH-Arginine reacts with 1 molecule of periodate, as determined by sodium thiosulfate titration of unreacted periodate, to yield A"-adipylarginine (Scheme II). The latter is eluted from the short column of the analyzer at the position of lysine; thus special conditions are required for its resolution. The ninhydrin color value of adipylarginine is 115% of that of arginine. Its Rp value is 0.39 on paper chromatography with l-butanolpyridine-acetic acid-water. In accord with the presence of an amide bond between adipic acid and one of the guanidino nitrogen atoms of arginine, on hydrolysis in 6 N HCl at 110" for 24 hours, adipylarginine is hydrolyzed quantitatively to yield arginine and adipic acid. Adipic acid was identified by paper chromatography in l-butanolacetic acid-water and I-butanol-pyridine-acetic acid-water, as well as by electrophoresis at pH 3.6,4.7, and 6.5. The mobilities were in all instances identical with those of an authentic sample of adipic acid.
Bromphenol blue was used to detect the acid. Adipylarginine is stable in acidic solutions (pH lower than 5) at room temperature, but is hydrolyzed to arginine and adipic acid in alkaline solutions (pH above 7).
Structure of DHCII-Arginine-DHCH-Arginine is the addition product of arginine and 1,2-cyclohexanedione, as shown in Scheme Lack of reaction with phenylhydrazine and borohydride are also in agreement with this structure. Dissociation of DHCH-arginine to arginine and cyclohexanedione on removal of borate also shows that the cyclohexane ring is intact in the molecule.
Dihydroxyl addition products analogous to DHCHarginine have been described for t,he reaction of butanedione with benzamidine (23,24) and glyoxal or benzil with urea (25).
A Kl 0 ADIPIC ACID ARGININE SCHEME II While our work was in progress a similar structure was suggested for the reaction of monomeric 2,3-butanedione with arginine (26), but proof of the structure was not presented.
As predicted from the structure of DHCH-arginine shown in Scheme I, periodate oxidation yields adipylarginine (Scheme II). NMR and infrared spectra (see below) indicate that the product formed on periodate oxidation is N7-adipyl-n-arginine (Structure  II) rather than N7, N*-adipyl-L-arginine (Structure I). The cyclic intermediate (I) is presumably rapidly hydrolyzed because of the strain inherent in this structure, to yield Structure ZZ. The change in electrophoretic mobility of adipylarginine at acidic pH values is in good agreement with the pK1 = 4.43 of adipic acid (Fig. 3). Furthermore, on standing in alkaline solutions, arginine was released from adipylarginine without any evidence for the formation of an intermediate, as expected from the postulated structure with one amide bond.
Comparison of NMR spectra of DHCH-arginine and adipylarginine shows that on periodate oxidation an asymmetrical structure is formed (Fig. 6). The chemical shifts for the peaks arising after periodate oxidation are in agreement with published values for related structures: Sadtler Standard Spectra, No. 4406 (adipic acid) ; Nos. 5126, 5137, and 5184 (N-acylguanidino compounds (27)).
The infrared spectrum of DHCH-arginine shows a characteristic peak at 1055 cm-i, probably corresponding to vicinal hydroxyl groups.
This peak is absent from the spectrum of adipylarginine which also possesses peaks at 1680 and 1720 cm-i; these are compatible with reported values for adipic acid (SSS No. 281) and N-acylguanidino compounds (SSS Nos. 23182, 23185). showed no reaction of lysine residues. Oxidized bovine pancreatic RNase was modified with cyclohexanedione and digested with trypsin as described under experimental procedures.
Analysis of tryptic hydrolysates of RNase by two-dimensional peptide mapping showed that only peptides containing or adjacent to arginine residues were affected by cyclohexanedione treatment (Fig. 7). Treatment with carboxypeptidase B yielded, as expected, nearly theoretical amounts of lysine and arginine from tryptic digests of untreated, oxidized RNase, whereas only lysine was released from the tryptic digest of modified, oxidized RNase.
These results indicate that tryptic

Modi$cation of Arginine Residues in Proteins
Amino acid analysis of proteins treated with 1,2cyclohexanedione and peptides obtained from modified RNase indicate that only arginine residues are modified under the conditions employed (Tables I and II) (16). As described previously (5), lysine residues may react with 1,2cyclohexanedione to form a yellow product (440 nm) when longer reaction times are employed. This side reaction has a pH optimum of 11 and does not occur at neutral pH or in strong alkali.
Under the conditions specified in our work, this side reaction was not observed.
Treatment of    The various peptides were isolated from the pools as shown.
cleavage was limited to bonds involving lysine residues by the arginine modification.
In a preparative experiment the tryptic peptides obtained from modified RNase were fractionated by gel filtration and all the peptides were purified by paper chromatography or paper electrophoresis (Fig. 8). The peptides derived from the cyclohexanedione-treated RNase are designated by the prefix T(c) and numbered consecutively from the NH&erminal end. Analysis of the peptides present in the tryptic digest confirms that cleavage had taken place only at lysine residues (Table II). Partial cleavage at lysine residues was observed in both control and cyclohexanedione-treated RNase and in the isolation of T(c)3 +4 and T(c)6 +7.
In the present study the peptides were purified without prior regeneration of arginine to show that this does not occur during purification procedures at acidic pH values. In subsequent studies of protein sequences, it may be desirable to regenerate arginine residues after the tryptic hydrolysis, depending on the solubilities of the peptides with and without the blocking group. Under the conditions used for regeneration of arginine (pH 7.0, 37" for 6 to 7 hours), hydroxylamine does not produce other reactions with proteins (28), in the absence of Cu2+ or 02 (29).* DISCUSSION 1,2-Cyclohexanedione reacts with free arginine in neut,ral and slightly alkaline solutions to form a single new amino acid derivative, fl ,N8-(1 ,2-dihydroxycyclohex-1 ,2-ylene)-I,-arginine, which is stable in acidic solutions and in borate buffers at pH 8 to 9. The blocking group can be removed by hydroxylamine at pH 7.0, with quantitative recovery of arginine. The structure of DHCH-arginine was elucidated by chemical and physicochemical methods. Consistent with the presence of vicinal cis-hydroxyl groups in the molecule, it forms a complex with borate, and the carbon-carbon bond between the hydroxyl z All of the reaction mixtures were kept under Nt in sealed tubes.
Solutions were made with deionized water.
Alternatively, EDTA can be used to bind Cut+. groups is cleaved by periodate oxidation, yielding iV-adipylarginine.
Adipylarginine yields quantitatively adipic acid and arginine on hydrolysis. NMR and infrared spectra of both DHCH-arginine and adipylargininc are in agreement with the predicted structures.
Evidence for an intact cyclohexane ring comes from the fact that DHCH-arginine is slowly hydrolyzed to yield cyclohexanedione and arginine on removal of borate. Analogous reactions have been described between dicarbonyl compounds and benzamidine or urea (23-25). 1,2Xyclohexanedione treatment of proteins results in specific modification of arginyl residues, restricting the action of trypsin to hydrolysis at lysine residues.
The homogeneity of the product and its stability in different solutions simplify subsequent analytical procedures.
The tryptic peptides thus obtained can be subjected to secondary digestion with trypsin after removal of the blocking groups by hydroxylamine at pH 7.0. In this manner the reversible blocking of arginine residues provides a convenient method for sequence studies, supplementing the various methods for selective, reversible modification of lysine residues. In the absence of methods for the specific modification of arginine residues under mild conditions, very little has been known of the role of arginine residues in protein structure and funct'ion. The mild conditions used in the present procedure, the homogeneit'y of product, specificity of reaction, reversibility under mild conditions, etc., suggest its possible usefulness as a probe for ascertaining the role of arginine residues in specific proteins. Some first experiments with native lysozyme and RNase are reported in the following paper (16).