THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 238, No. 1, January 1963

Printed in U.S. A.

The Action of Hydrazine on Collagen

I. CHEMICAL MODIFICATION*

ROGER DE LA BURDE,? LEONARD PECKHAM, AND ARTHUR VEIS$

From the Armour Leather Company and Northwestern University Medical School, Chicago, Illinois

(Received for publication, July 16, 1962)

The current view of the collagen fibril (1) pictures it as an ordered aggregate of collagen rods, which themselves consist of three intertwined peptide chains. The macromolecular mono- mer unit rods, tropocollagen, appear to be packed in the same way in all collagens regardless of the origin of the collagenous tissue or its age. The variations in the properties of the colla- gens have been ascribed in part to the presence of covalent cross- linkages between tropocollagen units (2-10) (inter-TC).l There is also good evidence for the existence of covalent cross-linkages between the individual peptide chains within the tropocollagen molecule (3, 6, 11, 12) (intra-TC). Some of the current im- portant questions in the chemistry of the maturation of collagen are those of the chemical nature of the various cross-linkages, their number, and their distribution.

Several investigators have proposed that the cross-linkages in collagen are ester-like in character (10). The first direct evi- dence came from the work of Grassmann et al. (8), in which it was shown that lithium borohydride reductively cleaved a small number of covalent bonds in citrate-soluble calf skin collagen. Konno and Altman (9) reported the isolation of a glycine-carbo- hydrate complex, reducible with lithium borohydride, from rat muscle collagen. Major support for the presence of ester-like linkages came from the work of Gallop et al. (7), who reported that hydroxylamine and hydrazine reacted with gelatins to produce protein-bound hydroxamate and hydrazide, respec- tively. These investigators relied on the fact that. both reagents react much more rapidly with simple esters than with simple amides or peptides in proposing that the hydrolyzable bonds were ester linkages. They found that with acid precursor gela- tin,2 or the gelatin produced by the denaturation of calf skin or

* This work was performed in the Central Research Laboratory of Armour and Company.

t That part of the work carried out by R. de la Burde was used for the preparation of a dissertation in partial fulfillment of the requirements for the degree of Dr.-Ing. at Technische Hochschule Aachen, at which he had completed other requirements before coming to the United States. He wishes to thank the Armour Leather Company for encouraging this work and permitting pub- lication and Prof. H. Zahn, of Aachen, for acting as his sponsor and assisting in the preparation of the dissertation.

$ Inquiries should be addressed to Dr. Arthur Veis. 1 The abbreviation used is: TC, tropocollagen. 2 Commercial gelatins are manufactured by either of two pro-

cedures. Pretreatment of collagen with acid, followed by neutral or acid extraction, yields the acid-precursor gelatins. Pretreat- ment with a base, usually Ca(OH)z, followed by neutral extrac- tion yields the alkali-precursor gelatins. One essential distin- guishing feature is that the acid-precursor gelatins have isoelectric points near pH 9 while the alkali-precursor gelatins have isoelectric points near pH 5. This difference is the result of

ichthyocol TC, 6 moles of hydroxamate were formed per lo5 g of protein. The gelatin was degraded to units with molecular weights near 20,000. Since there seemed to be neither a decrease in the number of amide groups in the gelatin, nor any increase in NHz-terminal amino acid residues, it was concluded that native collagen and acid-pretreated gelatins contained 6 x 10-c equivalent of weak ester-like linkages per g. Alkali-pretreated gelatins, from which approximately one-half of the amide group content was removed during pretreatment, formed only approxi- mately 3 X lo+’ equivalent per g of bound hydroxamate. How- ever, in this case, the molecular weight was still reduced to a value near 20,000.

The reduction of the molecular weight of the individual or-pep- tide chains, with the concomitant formation of bound hydroxa- mate, led Gallop et al. (7) to postulate that the ester-like link- ages were intrachain bonds, that, is, part of the gelatin back- bone. From this point of view, the ester linkages could not be involved in intra-TC or inter-TC cross-linking. Gallop et al. (13) did, however, suggest that the ester linkages occurred in pairs in the a-chains and that an isomerization and rearrange- ment of bonds could establish ester-like cross-linkages.

The hydrasine and hydroxylamine reactions were carried out by Gallop el al. (7) at pH 9 or pH 10 in the range of 1 to 3 M

reagent concentrations. Under these conditions, high concen- trations of urea or thiocyanate were required for the reaction to proceed at an appreciable rat.e at room temperature. The reac- tions could be carried out in the absence of the urea at 40”.

Bello (14) attempted to apply these same reactions to fibrous steer corium collagen as well as to an acid precursor gelatin. In his experiments, the acid precursor gelatin yielded 0.5 to 1.5 X lop5 equivalent of bound hydroxamate per g of collagen, and the insoluble corium collagen, presumably more highly cross- linked, only 2 x 10e5 equivalent of hydroxamate per g. There was no way to reconcile his data for the acid precursor gelatin with that of Gallop et al. (7). Bello found, in addition, that the insoluble collagen would not react at all with hydroxylamine or hydrazine (pH 8.6) unless the collagen was denatured before treatment with the base. Bello suggested that all the bound hydroxamate could be accounted for by the presence of carbo- hydrate in his collagen preparations. Hormann (15) also ex- amined the reaction of hydroxylamine with fibrous collagen. He confirmed that the collagen had to be denatured before it would react. Hijrmann et al. (16) found that, on the average, mature steer corium collagen yielded 11 X lop5 equivalent of bound hydroxamate per g of collagen. They also treated solu-

partial deamidation of the alkali-precursor gelatin stock during the pretreatment step.

189

by guest on Novem

ber 5, 2020http://w

ww

.jbc.org/D

ownloaded from

190 Action of Hydrazine on Collagen. I Vol. 238, No. 1

TABLE 1

Characterization of purified and limed collagen

Analysis

1. Content of solu- ble compo- nents

Neutral salt- soluble

Acid-soluble

2. Swelling pH 2.5 pH 5.3

3. Total nitrogen 4. Amide nitrogen

5. Arginine 8.43-8.55 8.47-8.52 6. Citrulline Negative Negative 7. Hydroxyproline 12.9 12.9

8. Shrinkage tem- perature

62-64” 58-59 ’

Purified Lied collagen collagen

%

0.04

0.59

415 515 310 325

18.08 17.82 0.66 0.42

Method

Weight and hy- droxyproline (Neuman and Lo- gan (22))

Solution uptake (Bowes (23))

Kjeldahl analysis NH3 in partial hy-

drolysate (Cassel and Kanagy (24))

Rosenberg et al. (25) Fearon test (26) Neuman-Logan (22)

ble TC with hydroxylamine and found 6.8 X 10e5 equivalent of bound hydroxamate per g. They therefore concluded that there were 4 moles of ester-like linkages per 10-S g of insoluble collagen which participated in inter-TC linkages. Following the hydroxamate formation, carried out with 0.75 M NHQOH in 4 M LiCl at 37” and pH 9.55 for 10 hours, the collagen was com- pletely dissolved and degraded, presumably to an average mo- lecular weight near 20,000. Both the collagen and TC con- tained a total of approximately 4 X IO-+ mole of glucose or glu- cose and galactose per g. Hormann (17) concluded that the cross-links in collagen consisted of hexoses whose two hydroxyl groups formed ester links between the different peptide chains while the hexoses in tropocollagen were bound by only a single ester group. The reducing group of the hexose was supposed to be attached to a hydroxyl side chain of one of the backbone pep- tide units by a glucoside-like oxygen bridge.

All of these data provide substantial evidence for the presence of hydroxylamine-sensitive, ester-like linkages in collagen but do not provide conclusive proof that the ester linkages partici- pate in inter-TC unit bonding. The dissolution of the mature collagen fibers after hydroxamate formation could conceivably occur merely due to the extensive degradation of the individual backbone peptide chains and not because of the cleavage of in- terchain bonds. A clarification of this question would appear to require the selective cleavage of the inter-TC bonds while leaving intact the intra-TC bonds of weak ester-like character.

I f one accepts the premise that the inter-TC bonds are ester bonds as Hormann proposes (15-17), it is difficult to see why these groups would not be readily accessible for reaction with hydroxylamine or hydrazine.

Veis and Cohen (2, 18) and Veis et al. (3, 11) showed that collagen fibers could be selectively degraded in acid or at neutral pH to yield soluble fractions with nongaussian molecular weight distributions and containing substantial numbers of molecules with both the peptide backbone chains and the intra-TC cross-

linkages intact. These molecules could be reversibly denatured and renatured. Courts (19) showed that treatment of mature corium collagen with acid in the cold, followed by prolonged soaking in Ca(OH)2 in the presence of salts that reduced swell- ing, also appeared to lead to the selective rupture of inter-TC bonds. Collagen fibers treated in this way were soluble in cold dilute acetic acid to give high viscosity solutions very similar to TC in behavior. Fibrous precipitates of this soluble collagen could be prepared by dialysis (20). These data illustrate the accessibility of the inter-TC bonds to both acid and alkaline hydrolysis but do not indicate their chemical nature.

In a search for a system in which the inter-TC bonds could be cleaved in fibrous collagen without denaturing the collagen struc- ture, we found that hydrazine, in contrast to hydroxylamine, would react with native collagen under appropriate conditions. We therefore investigated the chemical and physical changes occurring in purified native steer corium collagen, and Ca(OH)2- soaked, but otherwise undegraded steer corium collagen follow- ing reaction with hydrazine. This report describes the chemical aspects of the collagen-hydrazine reaction over an extended range of hydrazine concentration.

EXPERIMENTAL PROCEDURE

Preparation of Collagens-Purified steer corium collagen was prepared from fresh steer hide by the method of Veis, Anesey, and Cohen (3). The limed collagen was obtained from a simi- lar hide subjected to a 5-day lime treatment. This material was purified by the method of Bowes and Kenten (21). Both col- lagens were lyophilized for storage. A number of determinations were made to characterize the collagens. These data are sum- marized in Table I. The short liming process did not decrease the arginine content of the collagen, and there was no indication of citrulline formation. The principal result of the liming was to reduce the amide content by 36.4%.

Preparation of Hydra&e-Hydrazine (Olin Mathieson Chemi- cal Corporation, 95 + %) was redistilled and used as a standard for the preparation of hydrazine solutions of different concentra- tions. The concentration of the standard was determined colori- metrically by the method of Pesez and Petit (27) and was kept in the range of 96 to 98%. Dilutions of this stock solution were made in an acetone-Dry Ice bath to avoid overheating. Since aqueous hydrazine easily undergoes autoxidation, which can be accelerated by dilution, by the addition of NaOH, or by the presence of trace metals (2830), the change in concentra- tion of 10 and 70% aqueous hydrazine on standing in closed glass containers at room temperature was determined. There was no significant decrease in hydrazine concentrations within 6 days.

Reaction of Collagen with Hydra&e-Lyophilized collagen was treated with aqueous hydrazine at 25 =t 2” for 30 hours. A ratio of 1 weight of collagen to 10 volumes of hydrazine solution was used. The moisture content of the collagen (-9%) did not cause an increase in the solution temperature on mixing. These reaction mixtures were shaken in glass-stoppered Erlenmeyer flasks, When solutions above 58.8% hydrazine were used, a wax coating was applied to the stoppers.

At the end of the reaction period the collagen was washed with many changes of cold distilled water (-7”) for 3 to 4 days. The washings were monitored for the presence of hydrazine as follows. After each 20 hours of washing, 2 g of collagen were washed separately for 3 hours in 50 ml of distilled water. This

by guest on Novem

ber 5, 2020http://w

ww

.jbc.org/D

ownloaded from

January 1963 R. de la Burde, L. Peckham, and A. Veis 191

washing solution was tested qualitatively with salicylaldehyde (31) for the presence of hydrazine. After the first negative test, the protein was washed for another 30 hours and then lyophil- ized. The hydrazide content of the collagen was determined by the method described by Seifter et al. (32). Crystalline hy- drazine sulfate, the reaction product of hydrazine hydrate and excess sulfuric acid, was used for the preparation of the hydra- zine standards.

It was noticed that the amount of protein-bound hydrazide appeared to decrease with prolonged washing. To determine the extent of this loss, samples of collagen treated with 40y0 and 70 $Zo hydrazine were extracted with repeated changes of absolute ethanol for 40 hours to remove the free hydrazine. The spot test with salicylaldehyde indicated the absence of free hydrazine in the final rinse. The extracted collagens were rinsed with water to remove the ethanol, lyophilized, and then subjected to the usual aqueous wash procedure. The hydrazine content was de- termined at intervals during a 70-hour wash. These data are shown in Fig. 1 and indicate a gradual decrease in protein-bound hydraaide that could decrease the values reported subsequently by5tolO%.

Arginine Detemination-Half-gram portions of collagen were hydrolyzed under reflux for 16 hours in 10 ml of 20.5% HCl. The hydrolysates were analyzed by the method of Rosenberg et al. (25). The presence of hydrazine in the amounts expected from the hydrolysis of the bound hydrazides (up to 90 x 10m5 equivalents of NHfNHz per g) did not affect the arginine analysis.

Citrulline Analysis-The hydrazine-treated collagens were examined for the presence of citrulline, one of the possible break- down products of arginine. The method described by Fearon (26) was used, and no citrulline was found.

Amide Analysis-The amide content of the collagens was determined as the amount of ammonia distilled from partial acid hydrolysates in the presence of excess sodium and calcium hy- droxide (24). A comparable standard hydrazine solution was prepared for each of the distillates, according to the hydrazide content. The steam distillates of these solutions were titrated as in the amide analyses. In each case, the amount of steam- distilled hydrazine that could have accompanied ammonia in the collagen distillates was negligible and could not be detected by titration.

Amino Acid Analyses-Purified collagen and the residual and soluble fractions resulting from 40 and 70 7O hydrazine treatment were hydrolyzed and examined for their amino acid content by a quantitative paper chromatographic technique (33). Particular attention was given to the ornithine content. The standard amino acid mixtures contained ornithine as well as the amino acids present in collagen. The presence of ornithine increased the color yield of the lysine. Correction factors were computed for various concentrations of ornithine. These were obtained by running the standard amino acid mixture with ornithine con- centrations ranging from 0 to 6.02%.

NHz-terminal Amino Acids-Purified native collagen treated with 50% hydrazine was dinitrophenylated under the conditions described by Levy (34). The dinitrophenyl derivatives were identified qualitatively by paper chromatography.

RESULTS

The Hydrazine System The reaction systems examined ranged from 0 to 50% hydra-

zine, corresponding to 0 to 21.8 M. The basicity of the solutions

r

G s 5

z 40, I I I I I I I I I I I

8 16 24 32 40 48 56 64 72

DURATION OF WATER WASH, HOURS

FIG. 1. Decrease in protein-bound hydraaide during the wash- ing procedure. A---A, collagen treated in 7070 hydrazine; O-O, collagen treated in 40% hydrazine. The upper figures on the ordinate refer to the 707, and the lower to the 4070 hydra- zine-treated collagens.

varies continuously over this range. Deno (35) has shown that the Hammett function, H- is a linear function of the hydrazine concentration. H- increases from a value of 10.75 for an in- finitely dilute hydrazine solution to 15.93 at 18.72 M (60%). In the overlapping range up to pH 14, the H- and pH functions have identical values and, therefore, the pH is also a direct linear function of the hydrazine concentration. In the ensuing discussion, all the data could have been presented in terms of H-, or pH, but are usually presented in terms of the hydrazine weight concentration since this is a more easily appreciated unit and was the experimental variable.

Reactions during Hydrazinolysis

Formation of Protein-bound Hydrazide-Hydrasine was found to react with the undenatured fibrous collagens to yield protein- bound hydraaide over the entire hydrazine concentration range. The amount of protein-bound hydrazide in the native and limed collagen is shown in Fig. 2 as a function of the hydrazine solu- tion concentration. The two striking features of these data are first, that the native collagen reacted with substantially more hydrazine than did the limed collagen at. each hydrazine con- centration, and second, that in the 40 to 50% hydrazine concen- tration range (H-, 14 to 15) there was an abrupt rise in the amount of hydrazide bound to the native collagen, whereas this did not take place with the limed collagen.

The collagen fibers did not appear to be denatured by their treatment and reaction with hydrazine. Electron micrographs of the limed fibers, the native hydrazine-treated fibers, and the limed, hydrazine-treated fibers showed all of the cross-striations and periodicity of structure characteristic of native collagen as long as the hydrazine reaction solution concentration was < 55 %. The original limed collagen, however, showed some evidence for decreased structural integrity in having a shrinkage temperature range of 58-59” as compared with the native collagen 62-64O range (Table I). Amorphous, nonstriated, treated collagen fibers appeared at hydrazine concentration > 60%. This did not coincide with the 40 to 50% hydrazine concentration range in which the native collagen showed its enhanced formation of bound hydrazides.

by guest on Novem

ber 5, 2020http://w

ww

.jbc.org/D

ownloaded from

192 Action of Hydraxine on Collagen. I Vol. 238, No. 1

0 IO 20 30 40 50 60 70

HYDRAZINE CONCENTRATION, %

FIG. 2. The formation of bound hydrazide by collagen as a function of hydraeine concentration. O-O, native purified collagen; O-0, limed collagen.

36

32

26

24

20

1 t I I I I I

0 IO 20 30 40 50 60 70

HYDRAZINE CONCENTRATION, %

FIG. 3. The amide group content of hydrazine-treated collagen. O--O, native purified collagen; O-0, limed collagen.

Deamidation-The amide analyses (Fig. 3) showed that the amide contents of both the native and limed collagens were de- creased during hydrazinolysis and that the decreases were directly proportional to the hydrazine concentration. An un- expected result, however, was that the treatment of the native collagen with 60% hydrazine removed only an amount of amide equivalent to that removed by the 5-day liming. The residual amide in the limed collagen was not very reactive. These data clearly suggest that two groups of amides exist in collagen and that approximately 640/, of the amides are situated in such a way as to be sterically inaccessible to the hydrazine, or are held by particularly strong bonds. The correspondence between the extent of deamidation and the basicity of the reaction medium probably indicates that the amide removal is a general base- catalyzed reaction and is not governed by any specific reactivity of amide bonds with hydrazine. However, in the presence of a nucleophilic reagent such as hydrazine, every amide hydrolyzed

I I I I I I ! I

0 IO 20 30 40 50 60 70

HYDRAZINE CONCENTRATION, %

FIG. 4. The arginine content, of hydrazine-treated collagen. O-O, native purified collagen; a-0, limed collagen.

TABLE II

Arginine and ornithine content of native collagen after reaction with hydrazine

Hydrazine concentration Collagen fraction Arginine* Ornithinet

% moles/l0~ g

0 Native fibers 48 0 40 Insoluble residue 45 2.5 70 Insoluble residue 23 33

40 Soluble, nondialyz- 23 33 able

70 Soluble, nondialyz- 8 53 able

* Determined by direct analysis, method of Rosenberg et al. (25).

t Determined by paper chromatography (33). This method is subject to a larger experimental error than the arginine analysis. These data seem to be consistent.ly high by approximately 157,.

will react to form a protein-bound hydrazide except at very low hydrazine concentrations.

Deguanidination-Analyses of the arginine content of the two collagens, illustrated in Fig. 4, showed that at high hydrazine solution concentrations (hydrazine > SO%, H- > 15) the ar- ginine content was decreased. The reaction was that of con- version of the arginine to ornithine and, as shown in Table II, there was reasonably good quantitative correlation between the decrease in number of residues of arginine and the number of residues of ornithine created. There was no indication of citrul- line formation, and in accord with the work of Hamilton and Anderson (36)) it is safe to conclude that the direct deguanidina- tion of arginine was the principal reaction in the presence of hydrazine. The deguanidination reaction also resulted in the folmation of several amino-substituted guanidines and of sym-

diaminotetrazine. This interesting aspect of the chemistry of the reaction system will be described elsewhere.3

The deguanidination reaction does not involve the formation of protein-bound hydraaide, as does the cleavage of an amide bond. Therefore, although the deguanidination reaction began

3 R. de la Burde and A. Veis, in preparation.

by guest on Novem

ber 5, 2020http://w

ww

.jbc.org/D

ownloaded from

January 1963 R. de la Burde, L. Peckham, and A. Veis

I t I I t I

ZO-

0 10 20 30 40 50 60

WDRAZINE coNcENmATioN , %

193

FIG. 5. The number of peptide bonds cleaved in the residue fibers of hydrazine-treated collagen. O--O, native collagen; O--O limed collagen.

to be significant in the same hydrazine concentration range as the very rapid rise in bound hydrazide in the native collagen, these two reactions were not directly related.

Hydrazinolysis of Peptide Bonds-Bradbury (37) determined the kinetics of the hydrazinolysis of simple peptides and found that the rate of cleavage of peptide bonds was finite in anhydrous hydrazine even at 2”, the freezing point of hydrazine. Rees and Singer (38) also reported that there was a continuous decrease with time of the specific viscosity of a sample of y-globulin in anhydrous hydrazine at 25” and that this was due to peptide bond hydrazinolysis. We therefore examined the hydrazine- treated collagen for evidence of peptide bond hydrolysis and chose the 50 To hydrazine-treated native collagen for this analysis. At this hydrazine concentration, the number of bound hydra- aides was almost at its maximal value of 40 X 10m5 mole bound per g, whereas only approximately 0.3 of the arginine residues had been deguanidinated and there had been no dramatic in- crease in amide removal. Paper chromatography of hydroly- sates of the fluorodinitrophenylated, 50% hydrazine-treated collagen showed the presence of glycine, aspartic acid, and glutamic acid as new NHn-terminal residues. Semiquantita- tively, the NH2-terminal glycine was present in largest amount and glutamic acid in the smallest amount. These data are in accord with those of Bradbury (37, 39) who found that peptide bonds containing glycine were particularly susceptible to hy- drazinolysis, and that, in addition, bonds involving serine, glu- tamic acid, and aspartic acid were also especially labile in insulin, Iysozyme, and wool.

The number of peptide bonds cleaved by the hydrazinolysis can be determined by subtracting the number of moles of bound hydrazide formed by amide replacement from the total hydraaide content. Fig. 5 shows the results of such a calculation for both the native and limed collagen. At the maximum, approximately 22 x lOA mole of peptide bonds are cleaved per g of native collagen. At the same hydrazine concentration, only 3 x lop5 mole of peptide bonds per g were converted to the hydrazide in

the limed collagen. These data indicate that approximately 19 X 1O-6 mole of peptide bonds per g collagen were hydrolyzed during the liming process and that the same bonds were appar- ently involved in both liming and hydrazinolysis. Bowes (23) found evidence for an increase in both basic groups and fluoro- dinitrobenzine-reactive groups in limed collagen. She ascribed a portion of these to newly created residues from peptide bond hydrolysis and the remainder to the presence of ornithine result- ing from the deguanidination of arginine. In this study, the native and limed collagens had the same arginine content and all of the increase in reactive groups appears to be related to the hydrazinolysis of particularly labile peptide bonds. The idea that a set of such bonds exists arises because of the clear-cut maximum in hydrazide formation at 55% hydraaine (Fig. 2).

Dissolution during Hydrazinolysis

As the hydrazinolysis reaction proceeds, a number of physical changes indicate a decrease in the structural integrity of the collagen fibers. The fiber tensile strength decreases, the swell- ing increases, and part of the collagen dissolves. Each of these alterations will be discussed in detail in a subsequent report. For the present purpose, we can take the extent of dissolution as a measure of the degree of rupture of cross-linkages as well as of backbone peptide chain hydrazinolysis. The jilled circles in Fig. 6 represent the solubility of the native collagen as a function of hydrazine treatment. The limed collagen (open circles) was somewhat more readily dissolved at low hydrazine concentra- tions, but in both cases the amount of collagen dissolved in- creased most. rapidly at hydrazine concentration greater than 70%. The rapid increase in ease of dissolution, in other words, did not coincide with the abrupt increase in content of bound hydrazide at 40 to 50% hydrazine (Fig. 2) but followed the same course as the deguanidination reaction, As shown in Table II, the soluble fractions had a much lower arginine content than the insoluble fibrous residues and a correspondingly higher amount of ornithine. This observation suggests that the arginine residues

by guest on Novem

ber 5, 2020http://w

ww

.jbc.org/D

ownloaded from

Action of Hydrazine on Collagen. I Vol. 238, No. 1

0 to 20 30 40 50 60 70 00 so 100

HYDRAZINE CONCENTRATION. %

FIG. G. The dissolution of collagen during reaction with hy- drazine. O-O, native purified collagen; O----O, limed col- lagen.

52

12

6

4

0 IO 20 30 40 50 60 70

HYDRAZlNE CONCENTRATION, %

FIG. 7. A summary of the modification of native purified colla- gen. All points except the open circles refer to the Zeft-hand ordi- nate. l ----0, moles of bound hydrazide; O-U, moles of amide removed; V-V, moles of arginine destroyed; O--O, dissolution of the collagen.

may play an important role in maintaining the insolubility of the intact collagen structure.

DISCUSSIOli AND COiTCLUSIONS

The results reported in the foregoing are fairly clear-cut and unequivocal. The objective of this discussion is 2-fold: 17-e wish first to correlate the state of intermolecular bonding in native steer corium collagen with its hydrazine reactivity and chemical modification; second, we wish to relate our data to the studies of others on the hydroxylamine and hydrazine reactivity of systems containing denatured fibrous collagens or gelatins. Because of possible chemical differences in collagens from different sources, and because of the differences in reaction conditions, this latter

comparison may be considered more speculative in nature. The ensuing discussion is, therefore, somewhat artificially divided into two sections

Reaction of Native Fibrous Collagen with Hydra&e

Evidently, three reactions take place when fibrous undena- tured collagen is treated with aqueous hydrazine: deamidation, peptide bond hydrazinolysis, and deguanidination of the argi- nine. The first two reactions result in the formation of protein- bound hydrazide, while the deguanidination reaction does not. The extent of deamidation is directly proportional to the hy- drazine solution concentration over the entire range examined. The peptide bond hydrazinolysis becomes important only when H- = 14 and is complete, at room temperature, when H- reaches 15. The deguanidination reaction becomes significant when H- > 15. As the hydrazinolysis reactions proceed, the fibers begin to dissolve, and they pass rapidly into solution at H-. > 16. The course of each of these reactions is summarized in Figs. 7 and 8 for the native and limed collagens, respectively.

At hydrazinc concentrations <40% (H- < 14) both collagens show a direct, and in limed collagen an almost one-to-one, equiva- lence between amide removal and the formation of protein- bound hydrazide. As noted in the previous section, the amide loss from the native purified collagen is slightly greater than the hydrazide formation at low hydrazine concentrations, indicating that some of the most reactive amide groups are undergoing hydrolysis rather than hydrazinolysis. The near equivalence of amide loss and hydrazide formation excludes the possibility that interchain imides are involved, since the breakage of such link- ages would result in the formation of 2 moles of collagen-bound hydrazide for each mole of ammonia removed. Similar quanti- tative considerations also seem to exclude the possibility of hydrazine reaction with ester linkages. This kind of reaction would have been indicated in the native collagen if the plot of bound hydrazide in Fig. 7 had been above, rather than below, the amide removal curve.

The sharp rise in formation of protein-bound hydrazide at hydrazine > 400/& and the concomitant appearance of new

32 1 I I 4 I I

I -IGO

HYDRAZlNE CONCENTRATION , %

FIG. 8. A summary of the modification of limed collagen. All points except open circles refer to left-hand ordinate. O--O, moles of bound hydrazide, t-W, moles of amide removed; V-V, moles of arginine destroyed; O--O, dissolution of the collagen.

by guest on Novem

ber 5, 2020http://w

ww

.jbc.org/D

ownloaded from

NH*-terminal residues, indicates the presence of some particu- formed on the reaction of aldehydes with proteins. Landucci, larly labile peptide bonds in native collagen. There must be a Pouradier, and Durante (46) claim that aldehydes occur in all limited number of such bonds since the enhanced hydraeide collagens and gelatins. The major portion of these aldehydes formation levels off at hydrazine concentrations > 55% and are thought to be bonded to the peptide chains and in alkali- since the limed collagen, in which these bonds have presumably labile cross-linkages. The existence of very stable inter-TC and already been cleaved, shows no such increase in bound hydrazide. intra-TC cross-linkages also fits in well with the data of Courts The peptide bond hydrazinolysis occurs at an H- of 14 to 15 and Stainsby (47) and Veis, Anesey, and Cohen (11,48). These and that the basicity of the liming solutions approximates 14. workers have each presented evidence that both alkali- and acid- It is possible that the enhanced hydrazide binding includes precursor gelatins of high molecular weight can be obtained hydrasides formed from the cleavage of ester linkages. The under fairly drastic conditions and that these high molecular NH*-terminal amino acid residue content was not determined weight gelatins are cross-linked multichain structures. Some quantitatively or directly. However, the hydrazinolysis reac- peptide bonds were cleaved during the preparation of these gela- tions in the H- 14 to 15 range do not lead to the marked solu- tins (47), but the chain cross-linkages were still present. bilization of the native collagen which would be expected if ester In summary, the data presented here suggest that the inter- cross-linkages were severed. Correspondingly, the limed colla- TC cross-linkages are rather strong bonds, more stable than at gen did not become markedly more soluble in this hydrazine least one set of weak peptide bonds in the collagen backbone concentration range, and the lack of solubility of the limed chains. The resistance of these interchain cross-linkages to collagen provides evidence that the cross-linkages in the limed base-catalyzed hydrolysis or to hydrazinolysis at high basicity collagen were intact. This is in spite of the fact that approxi- argues against their being considered as of “weak ester-like” mately 20 X 10v5 mole of peptide bonds per g had been cleaved nature. On the other hand, a portion of the amide groups in in both collagens at this point. It is difficult to imagine that native collagen are readily reactive with hydrazine to form weak ester cross-linkages between TC units would be less ac- protein-bound hydrazide. cessible to general basic hydrolysis or hydrazinolysis than would Alkali-pretreated collagen contains many severed peptide peptide bonds within the organized framework of the TC-back- chains in the polar regions of the molecule. This hydrolysis bone chains. does not lead to dissolution of the collagen unless some further,

The peptide chains in collagen contain segments alternately more stable interchain bonds are hydrolyzed. These bonds are rich in the nonpolar and imino acid residues and in the polar equivalent in resistance to hydrazine to that holding the guani- amino acids. The appearance of NH&erminal aspartic and dino group to arginine. Native collagen is more reactive with glutamic acids on hydrazinolysis or basic hydrolysis and the hydrazine than limed collagen only insofar as the weakest amide lack of appearance of NHz-terminal proline or hydroxyproline groups are removed and the weakest peptide bonds are severed. following exhaustive liming of collagen (40) provides evidence These reactions reduce native collagen to the state of limed that the polar regions are most susceptible to basic hydrolysis. collagen and do not render it any more soluble than the limed The imino acid-rich gelatin chain segments, on the other hand, collagen, until at H- > 16 the deguanidination reaction begins. are stable to basic hydrolysis and, in addition, give to collagen One possibility for a strong but alkali-labile bond is that of a and gelatin gels their particular chain folding and organization. bond joining an c-amino group of lysine to a neighboring amide Gallop et al. (13) have suggested that ester linkages are located group or guanidino group by condensation of an aldehyde on in the polar regions of the gelatin molecule. Since, in fibrous the e-amino group to form an aminomethylol compound that collagen, these regions are the most readily deformed and swollen subsequently condenses to hydrolyze out water and join the two and the most susceptible to general basic hydrolysis, it is difficult groups with a methylene or substituted methylene bridge. to see why ester linkages in these regions would not have been Intrachain cross-linkages of this type have been introduced into attacked by hydrazine. Hence, we consider the presence of native ichthyocol tropocollagen showing that in the native inter-TC ester linkages as being rather unlikely. structure the appropriate cross-linking groups are sterically in

The most evident correlation in Figs. 7 and 8 is that the de- the proper configuration for reaction (49). guanidination of arginine is paralleled by an increase in the solubility of both the limed and native collagens. Arginine has

Reaction of Denatured Collagen and Gelatin with Hydrazine.

been shown to be of importance in the ordering of tropocollagen Presence of Ester Linkages in Collagens

units to form the native structure (41) and in the hydrogen bond Quantitatively, the data presented previously for the forma- stabilization of the ordered fibrils (42). Similarly, Grabar and tion of bound hydraaides at 2 to 3 M hydrazine concentrations Morel (43) and Janus (44) have shown the importance of the are in good agreement with the value reported by Bello (14) for guanidino group in the setting and gelation of gelatin. The re- the number of collagen-bound hydroxamic acids following the ciprocal relationship between solubility and arginine content reaction of denatured fibrous collagen with hydroxylamine, but leads one to consider the possibility that, in addition to their are lower than the values reported by Hormann (15) and Gallop participation in electrostatic chain interactions, the guanidino et al. (7). Both Gallop et al. (7, 13) and Hijrmann et al. (15-17) groups in collagen may be involved in the covalent cross-linkages used more vigorous reaction conditions to secure reaction of that stabilize mature collagen. At least, the covalent cross- their gelatins or denatured collagens with the hydroxylamine. linkages must have approximately the chemical stability of the Hormann, Reidel, Altenschopfer, and Klenk (16); for example, bond linking the guanidino group into the arginine molecule. heated their reaction mixtures at 37” and pH 9.55 for 10 hours. A suitable cross-link might be that joining an e-amino group to Under these conditions, even in the absence of hydroxylamine, an amide or to a guanidino group via a methylene bridge result- the isoelectric pH of acid precursor gelatin or native collagen is ing from an aldehyde condensation reaction. Fraenkel-Conrat shifted to lower values indicating the loss of amide nitrogen. and Olcott (45) have shown that such cross-linkages are readily Pentide bond hvdrolvsis is also enhanced at nH 9.55. the first I ” I

January 1963 R. de la Burde, L. Peckham, and A. Veis 195

by guest on Novem

ber 5, 2020http://w

ww

.jbc.org/D

ownloaded from

196 Action of Hydra&e on Collagen. I Vol. 238, No. 1

order rate constant for gelatin peptide bond hydrolysis being number of ester bonds in denatured collagens and gelatins by some 8 times greater at pH 9.5 than at pH 7.1, at which the rate hydroxylamine or hydrazine reactivity. constant has its minimal value (50). It is difficult to imagine that some amide loss and peptide bond hydrolysis did not occur

REFERENCES

in the presence of the hydroxylamine at elevated temperature 1. HARRINGTON, W. F., AND VON HIPPEL, P. H., in C. B. ANFIN-

and high pH. In addition, our experiments indicate the ready SEN, JR., M. L. ANSON, K. BAILEY, AND J. T. EDSALL (Edi-

reactivity of approximately one-half the amide groups in native tors), Advances in protein chemistry, Vol. 16, Academic Press, Inc., New York, 1961, p. 1.

collagen even at room temperature. Therefore, the results re- 2. VEIS, A., AND COHEN, J., J. Am. Chem. Sot., 76, 6238 (1956);

ported here, -2 x lop5 equivalent of bound hydrazide per g of J. Phys. Chem., 62,459 (1958).

collagen at 3.0 M hydrazine, must be considered as the base line 3. VEIS, A., ANESEY, J., AND COHEN, J., J. Am. Leather Chem.

for chemical modification during hydrazinolysis, excluding carbo- Assoc., 55, 548 (1960).

4. MECHANIC, J., AND LEVY, M., J. Am. Chem. Sot.. 81, 1889 hydrate or ester reactions. Gallop el al. (7) reported that an (1959).

I

alkali precursor gelatin that had lost 50% of its amide groups 5. JOSEPH, K., AND BOSE, S., Bull. Central Leather Research

during pretreatment also formed 50% less bound hydrazide. Inst., 7, 97 (1960).

The agreement of this finding with our results seems more than 6. PIEZ, I?., AND ‘GROSS, J., J. Biol. Chem., 235,995 (1960).

coincidental. 7. GALLOP, P., SEIFTER, S., AND MEILMAN, E., Nature, 183, 1659

(1959). Direct chemical evidence for the occurrence of ester linkages in 8. GRASSMANN, W., ENDRES, H., AND STEBER, A., 2. Natur-

collagens has been given by Grassmann et al. (8) and Konno forsch., 9b, 513 (1954).

and Altman (9). Recently, Gallop et al. (13) described the 9. KONNO, K., AND ALTMAN, K. I., Nature, 181,994 (1958).

isolation of amino alcohols following lithium borohydride reduc- 10. GUSTAVSON, K. H., The chemistry and reactivity of collagen,

tion of gelatin. These amino alcohols could only have arisen Academic Press, Inc., New York, 1956, p. 167.

11. VEIS, A., ANESEY, J., AND COHEN, J., Arch. Biochem. Bio- from (Y- and ,&aspartyl ester bonds. Quantitatively, they were phys., 94, 20 (1961).

able to recover only 30% of the amount expected on the basis of 12. GRASSMANN, W., HANNIG, K., AND ENGEL, J., 2. Physiol.

their hydroxylamine assay, i.e. they recovered only -2 x 10e5 Chem., 324, 284 (1961).

equivalent per g of ester reaction products, which would corre- 13. GALLOP, P. M., BLUMENFELD, O., FRANZBLAU, C., AND SEIF-

TER, S., Abstr. Meeting of Am. Chem. Sot., 1962,‘~. 9C. spond to the glucose and galactose analyses of Hormann (17). 14. BELLO, J., Nature, 184, 241 (1960).

Thus, while there seems to be little doubt that ester-linkages are 15. H~RMANN, H., Das Leder, 11, 173 (1960).

present in collagen, the number of such linkages is. probably on 16. H~RMANN, H., RIEDEL, A., ALTEN&HO&ER, T., AND KLENK,

the order of 2 x 10W6 mole per g and the higher values reported M.. Das Leder. 12. 175 (1961).

for the hydroxylamine-sensitive bonds in denatured collagens and 17. H~R~ANN, H., ~bst&cts,‘Seve~th Congress of the International

Union of Leather Chemists Societies, Washington, D. C., gelatins probably include some bound hydroxamate resulting 1961, p. 16.

from the cleavage of weak amide and peptide bonds. 18. VEIS, A., AND COHEN, J., Nature, 186, 720 (1960). 19. COURTS, A., Nature, 191, 1097 (1961); Biochem. J., 74, 238

SUMIllARY (1960).

20. WARD, A. G., Nature, 190, 1960.

1. Native fibrous collagen was treated at 25” with aqueous 21. BOWES, J. H., AND KENTEN, J., J. Sot. Leather Trades Chem.,

hydrazine solutions ranging in concentration from 5 to 70% 33, 365 (1949).

hydrazine (-1.5 to 19 M). The insoluble fibrous residues were 22. NEUMAN, R., AND LOGAN, M., J. Biol. Chem., 184, 299 (1950).

freed of excess hydrazine, and the hydrazide content was deter- 23. BOWES, J. H., J. Sot. Leather Trades Chem., 33,176 (1949). 24. CASSEL, J., AND KANAGY, J., J. Research Natl. Bur. Standards,

mined. Similar studies were made with limed collagen. 42, 557 (1949).

2. Analyses were performed to determine other possible reac- 25. ROSENBERG, H., ENNOR, A. H., AND MORRISON, J. F., Bio-

tions, particularly amide loss, peptide bond hydrolysis, and them. J., 63, 153 (1956).

conversion of arginine to ornithine. The solubility of the chemi- 26. FEARON, W. R., Biochem. J., 33, 902 (1933).

cally modified fibers was also determined. 27. PESEZ, M., AND PETIT, A., Bull. Sot. Chem. France, 37, 122

(1947).

3. In the low hydrazine concentration range (< 407,), the 28. CUY, E., AND BRAY, W., J. Am. Chem. Sot., 46, 1786 (1924).

amide loss was found to be slightly greater than hydrazide for- 29. GILBERT, E., J. Am. Chem. Sot., 51, 2744 (1929).

mation in the native collagen. Limed collagen formed bound 30. AUDRIETH, L., AND OGG, B., The chemistry of hydrazine, John

hydrazide and lost amide nitrogen in exactly equivalent amounts. Wiley and Sons, Inc., New York, 1951, p. 141.

31. FEIGL, F., Qualitative analysis by spot test, Ed. 3, Elsevier The bound hydrazide content of the limed collagen was one-half Publishing Company, New York, 1946, p. 186.

that of the native collagen in this range. At hydrazine concen- 32. SEIFTER, S., GALLOP, P., MICHAELS, S., AND MEILMAN, E.,

trations in the range from 40 to SO%, peptide bonds were cleaved J. Biol. Chem., 235, 2619 (1960).

in the native collagen, yielding approximately 19 x 10-b equiva- 33. HIMES, J., AND METCALFE, L., A&. Chem., 31, 1192 (1959). 34. LEVY. A. L.. Nature. 1’74, 126 (1954).

lent. of bound hydrazide. This reaction did not occur with the 35. DEN&, N. C:, J. Am: Chem. Xoc., 74, 2039 (1952). limed collagen, indicating that these peptide bonds were also 36. HAMILTON, P. B., AND ANDERSON, R. A., J. Biol. Chem., 211,

alkali-labile. At higher hydrazine concentrations, arginine was 95 (1954).

converted quantitatively to ornithine. Dissolution of the fibers 37. BRADBURY, J. H., Biochem. J., 68,475 (1958).

appeared to accompany thr deguanidination reaction. 38. REES, E. D., AND SINGER, S. J., Arch. Biochem. Biophys., 63,

144 (1956). 4. The lack of collagen dissolution, under conditions in which 39. BRADBURY, J. H., Biochem. J., 68,482 (1958).

amide and peptide bond hydrazinolysis occur, suggests that the 40. COURTS, A:, Biochem. J., 58, $4 (1954):

intermolecular cross-linkages in mature collagens are not ester- 41. BENSUSAN. H. B.. Biochemistru. 1. 215 (1962).

like in character. The ready removal of amide further suggests 42. RICH, A., AAD CR&K, F. H. C., 7. MoZecuZar L&ok, 3, 483 (1961).

that amide loss should be taken into account in assessing the 43. GRABAR, P., AND MOREL, J., Bull. Xoc. Chim. Biol., 32, 643 ,..-..

(1950).

by guest on Novem

ber 5, 2020http://w

ww

.jbc.org/D

ownloaded from

January 1963 R. de la Burde, L. Peckham, and A. Veis 197

44. JANUS, J. W., in G. STAINSBY (Editor), Recent advances in 47. COURTS, A., AND STAINSBY, G., in G. STAINSBY (Editor), Re- gelatine and glue research, Pergamon Press, Inc., London, cent advances in gelatine and glue research, Pergamon Press, 1958, p. 214. Inc., London, 1958, p. 100.

45. FRAENKEL-CONRAT, H., AND OLCOTT, H. S., J. Am. Chem. 48. VEIS, A., ANESEY, J., AND COHEN, J., in G. STAINSBY (Editor), Sot., 70, 2673 (1948). Recent advances in gelatine and glue research, Pergamon

4G. LANDUCCI, J. M., POURADIER, J., AND DURANTE, M., in G. Press, Inc., London, 1958, p. 155. STAINSBY (Editor), Recent advances in gelatine and glue re- 49. DRAKE, M. P., AND VEIS, A., Federation Proc., 21, 405 (1962). search, Pergamon Press, Inc., London, 1958, p. 62. 50. CROOME, R. J., J. Appl. Chem., 3, 280 (1953).

by guest on Novem

ber 5, 2020http://w

ww

.jbc.org/D

ownloaded from

Roger de la Burde, Leonard Peckham and Arthur VeisThe Action of Hydrazine on Collagen: I. CHEMICAL MODIFICATION

1963, 238:189-197.J. Biol. Chem. 

  http://www.jbc.org/content/238/1/189.citation

Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/238/1/189.citation.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on Novem

ber 5, 2020http://w

ww

.jbc.org/D

ownloaded from