ANALYTICAL

207, 129-133 (19%)

BIOCHEMISTRY

The Determination of C-Amino Groups in Soluble and Poorly Soluble Proteinaceous Materials by a Spectrophotometric Method Using Trinitrobenzenesulfonic Acid William

A. Bubnis

Philadelphia 600 South

College Forty-Third

Received

March

and Clyde of Pharmacy Street,

M. Ofner

III

and Science, Department Philadelphia, Pennsylvania

of Pharmaceutics, 19104

30, 1992

earliest (7-9) and some later procedures (10,ll) did not differentiate between CY-and e-amino groups. Some procedures (11-14) measured a yellow absorbance at 420 nm but were required to account for an interfering complex formed between the trinitrophenyl (TNP) derivative and the sulfite ion by-product (see Fig. 1). Kakade and Liener (15) introduced an extraction step which removed both excess unreacted TNBS and TNP-a-amino derivatives. This step reduces potential complications from TNBS present during measurement and allowed the specific determination of E-amino groups. These authors also measured uv absorbance at 346 nm in an acid pH to avoid interferences from the sulfite-TNP derivative complex. A variation of this procedure has been used to monitor chemical modifications of side chain c-amino groups in collagen (3,4,16). The chemical modifications in these studies, however, substantially reduce the solubility of samples and may result in erroneously Press, Inc. low values because insufficient hydration may not allow TNBS to react stoichiometrically with the amino groups. Preliminary experiments on crosslinked gelatin Proteinaceous materials such as albumin (1,2), colla- in this laboratory indicate that poor solubility can lower gen (3,4), and gelatin (5,6) have been investigated as assay values. The procedure reported here utilizes an biodegradable carriers for implantable drug delivery. A increased TNBS reaction time to allow a complete reacconvenient assay of e-amino groups could be used to tion. In addition, an autoclave hydrolysis step that was determine the number of lysine residues during subse- originally introduced by Kakade and Liener (15) is also quent stages of chemical modification. One important employed to assure dissolution of the samples for specmodification in drug delivery is crosslinking between C- trophotometric measurements. amino groups. The reagent 2,4,6-trinitrobenzenesulfonic acid (TNBS)’ has been used as a uv chromophore AND METHODS in various procedures to determine primary amino MATERIALS groups in peptides, proteins, and foodstuffs (Fig. 1). The Materials

A procedure using 2,4,6-trinitrobenzenesulfonic acid (TNBS) for the determination of t-amino groups in soluble and poorly soluble proteinaceous materials is presented. The major modification from previous procedures is an extended TNBS reaction time to allow a stoichiometric reaction with amino groups. In addition, autoclave hydrolysis is used to assure sample dissolution for spectrophotometric measurements. The assay accuracy was evaluated by determining t-amino groups of insulin and bovine albumin. The determinations differed from literature values by ~3.3%. The c-amino group content of Type B gelatin was found to be 33.0 mobgelatin molecule of 1000 residues and is in agreement with similar source gelatins and collagen. The coefficient of variation for determinations on all three materials was ~5.3%. The assay should be applicable to a broad range of proteinaceous materials. 8 1992 Academic

’ Abbreviations TNP, trinitrophenyl. 0003-x97/92

used:

TNBS,

2,4,6-trinitrobenzenesulfonic

$5.00

Copyright 8 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.

acid,

Type B gelatin was prepared by alkaline treatment of bone by Rosselot (Paris, France; Lot 10287). The USP grade granules were used without further purification. 129

130

FIG. 1. derivative

BUBNIS

Reaction of 2,4,6-trinitrobenzenesulfonic and sulfite ion.

acid with

AND

primary

The moisture content of the granules during storage was approximately 11% (w/w). The pH of a 1% solution was 5.8. The isoionic point (pl) of a sample deionized by a mixed-bed ion-exchange column was 5.2 (6). Poorly soluble gelatin was produced by exposure to 105°C and 10 pm Hg for 5 days. These conditions have been believed to induce covalent crosslinking (17). Humulin R, human insulin for injection (100 units/ml of recombinant DNA origin) was from Eli Lilly and Co. (Indianapolis, IN). Bovine albumin (high purity) was obtained from Fisher Biotech (Fair Lawn, NJ). The moisture content of the bovine albumin during storage was approximately 6.5%. TNBS (298% pure) was purchased from Pierce (Rockford, IL). The water used was purified by reverse osmosis. All other chemicals were of ACS reagent grade. TNBS

Assay

Procedure

Approximately 11 mg of solid sample or l-2 ml of dissolved protein (3.85-0.200 mg/ml) was placed in a 50-ml screw cap test tube. One milliliter of 4% NaHCO, (pH 8.5) and 1.00 ml of 0.50% TNBS were added to the tubes. For the albumin studies (0.200 and 0.400 mg/ml), the addition of the 4% NaHCO, was unnecessary because the albumin was initially dissolved in the buffer solution. The reaction mixture was heated at 40°C for 4 h with mild shaking. This reaction time is double the time used in earlier assays (3,4,15,16). Three milliliters of 6 N HCl was added and the mixture was autoclaved at 120°C and 15-17 psi for 1 h to hydrolyze and dissolve any insoluble material. For the 0.400 mg/ml bovine albumin study, 12 N HCl was required to hydrolyze and completely dissolve the denatured protein. The TNP-tamino derivative is resistant to hydrolysis and stable under these conditions (9). The hydrolysate was then diluted with 5.0 ml of H,O. The complete hydrolysis of the proteinaceous material is probably unnecessary because small TNP-derivative peptides have approximately the same solubility (9) and spectral properties (7) as free TNP-amino acids. Any a-amino groups produced during hydrolysis do not react with TNBS since the reaction requires an alkaline pH. For insulin solu-

OFNER

amino

groups

of protein

at an alkaline

pH

to form

a trinitrophenyl

tion determinations, the hydrolysate was diluted with 3.0 ml of H,O because 2 ml of sample was used initially. The hydrolysate dilution was extracted with three 20-ml portions of anhydrous ethyl ether to remove excess unreacted TNBS and TNP-a-amino groups (15). A 5.0ml aliquot of the aqueous phase was removed and heated for 15 min in a hot water bath to evaporate residual ether. The aliquot was diluted with 15.0 ml of H,O and the absorbance was measured at 346 nm in a Perkin-Elmer Model 552A double beam spectrophotometer. All samples were read against a reagent blank. The blank was prepared by the same procedure as the samples; however, the HCl was added before the addition of TNBS to prohibit any reaction of TNBS with the protein. Blanks were run in triplicate; the blank closest to the mean was used for sample measurements. Each assay trial contained four or five replicates. The trials were repeated three or four times. For the evaluation of TNBS reaction time, the autoclave hydrolysis step was replaced with mild shaking at 60°C for 2 h. Potential interferences with the assay were ruled out by conducting three control experiments. The first control experiment checked the efficiency of the extraction of TNP-a-amino derivatives with ethyl ether. The protein sample was replaced with 1 ml of a 0.272 mg/ml stock solution of glycine and reacted with TNBS as described in the assay procedure. This stock concentration is equivalent to the number of available r-amino groups determined in the gelatin sample and approximately represents a IO-fold excess of terminal a-amino groups. A value representing 2.7% of the gelatin value was measured and is within experimental error. Therefore, the potential influence of TNP-c-w-amino derivatives is negligible. The second control experiment checked that a low pH prevents further amino group reaction with TNBS by spiking the hydrolysate with a-amino groups. The hydrolysate of a gelatin determination was spiked with 1 ml of the above stock glycine solution. A value representing 1.2% of a value from gelatin alone was determined, which also is within experimental error, and demonstrates that no further reaction has taken place. The third control experiment checked

SPECTROPHOTOMETRIC

DETERMINATION

OF

c-AMINO

GROUPS

IN

131

PROTEINS

that the potential hydrolysis of TNBS to picric acid does not interfere with the determination. One milliliter of a 0.5% solution of picric acid was added to a gelatin determination before autoclaving. This concentration is 1.5 times the molar amount of TNBS present. A value representing 0.5% of a value from gelatin alone was determined. This difference is negligible and demonstrates that the hydrolysis of TNBS has no effect upon the determination. Results for albumin and insulin were obtained by moles lys moles protein

so ’ 0

liters)(MW) zz B(Absorbance)(O.OBO (1.46 X lo4 liters/mole mcm)(b)(x)

AND

2 TNBS

’

“I

where MW is the protein molecular weight with the units of g/mole, 1.46 X lo4 liters/mole. cm is the molar absorptivity of TNP-lys (7), b is the cell path length in cm, and x is the sample weight in g. The molecular weights used for bovine albumin and insulin were 66,500 (18) and 6000 (19), respectively. Gelatin results were calculated by modifying Eq. [l] to express results as moles lys/g gelatin due to the uncertainty of the molecular weight of gelatin.

RESULTS

1

DISCUSSION

TNBS reacts predominantly with primary amines. Satake et al. (8) have shown TNBS to be unreactive with the nitrogen of proline, the imidazole nitrogens of histidine, and the hydroxyl groups of tyrosine, serine, and threonine. While TNBS has some reactivity with sulfhydryl groups, the resulting TNP-S-derivative is very labile (9). At an alkaline pH, TNBS reacts with terminal a-amino groups as well as side chain t-amino groups. During the ether extraction at an acid pH, TNP-aamino groups are un-ionized and removed from the aqueous phase. Their removal allows the specific determination of t-amino groups. Lysine as the terminal amino acid will also be removed during extraction but this should be rare and only have a negligible effect on results. The duration of the TNBS reaction time was evaluated to determine the time required for complete reaction in a poorly soluble proteinaceous material. The absorbance measurements of crosslinked and therefore poorly soluble gelatin were compared after TNBS reaction times of 1 to 4 h. These results are shown in Fig. 2 where absorbance is expressed per gram sample weight and indicates that 3-4 h is required for completion of the TNBS reaction. Autoclave times ranging from 0.25 to 1.0 h duration were examined with insulin but the

REACTION

3 TIME

4 (h)

FIG. 2. Determination of time for the complete reaction of 2,4,6trinitrobenzenesulfonic acid (TNBS) in Fig. 1 using poorly soluble crosslinked gelatin. Absorbance is expressed per gram of sample weight. After reaction with TNBS, samples were hydrolyzed at 60°C for 2 h with mild shaking and then extracted with ethyl ether. Absorbance was measured at 346 nm against a reagent blank. Each data point is the average of five replicates; bars represent SD.

optimum time of 1.0 h was the same as reported previously (15). The assay accuracy was evaluated by analysis of insulin and bovine albumin. Both proteins have known amino acid sequences. For each molecule of insulin there is one lysine residue (19) containing an t-amino group, and for each molecule of bovine albumin there are 59 lysine residues containing an t-amino group (20). The results of the insulin analysis and the bovine albumin analysis are shown in Table 1. Four insulin trials yielded a range of 0.94-1.04 mol lys/mol insulin. The overall coefficient of variation and accuracy were 4.1 and 3.0%, respectively. The bovine albumin results for 0.400 mg/ml samples were less precise and accurate than the results from samples of 0.200 mg/ml. These results of 5.3% coefficient of variation and 3.3% accuracy, however, are acceptable. The t-amino determination was conducted on uncrosslinked gelatin. Gelatin is the denatured and partially hydrolyzed, soluble product of collagen. It has a theoretical molecular weight of 100,000 and its amino acid sequence is that of its parent collagen molecule. In gelatin, lysine and hydroxylysine residues contain tamino groups. Table 2 shows the results of the t-amino group determination for uncrosslinked gelatin. Three trials yielded an average of 33.0 X 1O-5 mol/g gelatin. This value corresponds to 33.0 e-amino groups on one gelatin molecule of 1000 amino acid residues. The precise amino acid content, and consequently the t-amino group content of gelatin, will depend on the source of the parent collagen. These values are not well established. The experimentally determined value for gelatin

132

BUBNIS

AND

OFNER

TABLE

Determination Protein Insulin

Trial”

Moles Iysine per mole proteinb

SD

Coefficient variation

0.95 1.04 0.96 0.94 0.97

0.01 0.03 0.01 0.02 0.04

1.1 2.9 1.0 2.1 4.1

1’

5.0 4.0 4.0 6.0 3.0

1 2 3 4

Averaged Bovine albumin (0.400 mg/ml)

Literature value

Percentage difference

57 58 54 58 57

1 1 5 2 3

1.6 1.7 9.3 3.4 5.3

59’

3.3 1.7 8.5 1.7 3.3

1 2 3 4

61 61 60 59 60

1 2 2 1 1

1.9 2.5 3.7 1.6 1.7

59e

3.4 3.4 1.7 0.0 1.7

Averaged

a Trials contained four or five replicates. b Based on a molar absorptivity value of 1.46 X 10’ liters/m01 ’ Ref. (19). d Calculated from all individual values. e Ref. (20).

. cm (7) for TNP-c-amino

is in agreement with values of 32.8 (21) and 32.3 (22) for similar source gelatins, and 32.9 for a similar source collagen (21). The assay provides a sensitive and reliable measure of the f-amino groups in a range of sample weights of these proteinaceous materials. A sample weight range of 0.200-11.0 mg was used. However, the analysis is dependent upon the number of t-amino groups present in the sample which is dependent on the number of tamino groups per molecule of protein. In the above sample weight range, 1.76 X 10e4-3.68 X 10e3 mmol of tamino groups was analyzed. A useful equation to determine the minimum sample weight to correspond

TABLE

Determination

of

1 2 3 4

Average” Bovine albumin 0.200 mg/ml)

1

of t-Amino Groups in Insulin and Bovine Albumin

2

of t-Amino Groups in

Type

B Gelatin

Trial

e-Amino groups moles/g” (X106)

SD

Coefficient variation

1 2 3 Average

33.3 32.5 33.1 33.0

0.4 0.6 0.7 0.6

1.2 1.8 2.1 1.9

of

D Based on a molar absorptivity value of 1.46 X 10’ liters/m01 +cm (7) for TNP-c-amino lysine. Value corresponds to the number of c-amino groups on one gelatin moelcule of 1000 amino acid residues.

Iysine.

with the least amount is mg protein

=

MW

assayed by the current

procedure

protein z

X (1.76 X 10e4 mmol e-amino groups),

[2]

where MW is the molecular weight in units of mg/mmol of the protein, and z is the number of mmoles of t-amino groups per mmole of protein. This assay is currently being used for the quantitative determination of crosslinking extent in proteinaceous materials crosslinked via amino groups. CONCLUSIONS A modified TNBS assay was developed on soluble and poorly soluble proteinaceous materials and was found to be accurate and reproducible. Experimentally determined values for bovine albumin, insulin, and gelatin were in agreement with literature values. It is likely that the assay can be used to determine t-amino groups in a broad range of proteinaceous materials. Further uses could include the determination of lysine residues during stages of chemical modification. ACKNOWLEDGMENTS The authors thank SmithKline Beecham tin and Dr. Edwin T. Sugita for his helpful

for their donation of gelareview of the manuscript.

SPECTROPHOTOMETRIC

DETERMINATION

OF

REFERENCES 1. Longo, W. E., Iwata, H., Lindheimer, T. A., and Goldberg, E. P., (1982) J. Pharm. Sci. 71, 1323-1328. 2. Wilmott, N., and Harrison, P. J., (1988) Znt. J. Pharm. 43, 161166. 3. Gilbert, D. L., Okano, T., Miyata, T., and Kim, S. W. (1988) Znt. J. Pharm. 47, 79-88. 4. Gilbert, D. L., and Kim, S. W. (1990) J. Biomed. Mater. Res. 24, 1221-1239.

11. Snyder, 284-288. 12. Goldfarb,

S. L., and

Sobocinski,

A. R. (1966)

Biochemistry

P. Z. (1975)

Anal.

5, 2570-2574.

Biochem.

64,

IN

133

PROTEINS

14. Hazra, A. K., Chock, S. P., and Albers, them. 137,437-443. 15. Kakade, M. L., and Liener, I. E. (1969) 280.

R. W. (1984) Anal.

Anal.

Bio-

27,

273-

Biochem.

16. Wang, C., Miyata, T., Weksler, B., Rubin, A. L., and K. H. (1978) Biochim. Biophys. Acta 544,555-567. 18. Hirayma, Biochem.

J.

GROUPS

13. Fields, R. (1972) in Methods of Enzymology (Him, C. H. W., and Timasheff, S. N., Eds.), Vol. 25, pp. 464-468, Academic Press, New York.

17. Yanas,

5. Tabata, Y., and Ikada, Y. (1989) Pharm. Res. 6,422-427. 6. Welz, M. M., and Ofner III, C. M. (1992) J. Pharm. Sci. 81,85-90. 7. Okuyama, T., and Satake, K. (1960) J. Eiochem. 47,454-466. 8. Satake, K., Okuyama, T., Ohashi, M., and Shinoda, T. (1960) Biochem. (Japan) 47,654-660. 9. Kotaki, A., and Satake, K. (1964) J. Biochem. 56,299-307. 10. Habeeb, A. F. (1966) Anal. Biochem. 14, 328-336.

t-AMINO

I. V., and Tobolsky,

A. V. (1967)

K., Akashi, S., Furuya, Biophys. Res. Commun.

19. Sober, H. A., (Ed.) Data for Molecular Cleveland. 20. Brown, J. R. (1975)

M.,

Nature and

Stenzel,

215,509-510.

Fukuhara,

K. (1990)

173, 639-646.

(1968) Handbook of Biochemistry, Selected Biology, pp. C184, Chemical Rubber Co., Fed. Proc.

Fed. Am.

21. Veis, A. (1964) The Macromolecular demic Press, New York. 22. Lewis, M. S., and Piez, K. A. (1964)

Sot. Exp. Biol.

Chemistry Biochemistry

34,

of Gelatin,

591. Aca-

3, 1126-1131.