ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS Vol. 205, No. 2, December, pp. 422-427, 1980

Role of Hydrophobic Interactions in Collagen Fibril Formation: Effect of Alkylureas in Vitro’

GERARD0 SUAREZ, MARTA VELIZ, AND RONALD L. NAGEL

Division of Experimental Hematology, Department of Medicine, Albert Einstein College of Medicine, Bronx, New York 104.61

Received February 15, 1980

Alkylureas inhibit the rate of in vitro fibril formation at 10 mM range of concentrations. There is a direct correlation between the extent of inhibition and the length of the alkyl chain (degree of hydrophobicity). When the alkylureas are added during the lag phase, the extent of inhibition depends on the time, after the onset of polymerization, in which the alkylurea is added. The effect of alkylureas is reversible since after dialysis the rate of fibril formation is normal. In conditions in which the lag phase is very short or not observable, the rate of fibril formation is not affected by the alkylureas. Ethylurea inhibits the rate of fibril formation but the extent of polymerization appears to be unaffected. In the presence of alkylurea there is an increase in the activation energy. It is concluded that hydrophobic interactions are signifi- cantly involved in the stabilization of intermediates formed during the lag phase.

Collagen fibril formation has been described as a specific aggregation process which leads to a supramolecular packing characteristic of the in vivo deposition of the fiber. Electron microscopic studies have shown that using the appropriate conditions it is possible to reproduce in vitro the native type of assembly (1). In addition, collagen offers advantages over other aggregating systems, such as actin and tubulin (2), because the number of molecular species involved in the construction of the fibril is restricted to tropocollagen and the process is independent of an energetic coupling (3). Furthermore, the aggregation does not require a previous conformational transition as in the case of the gelation of sickle hemoglobin (4).

The temperature dependence of collagen self-assembly is well established (5). Cooper has analyzed the temperature dependence of the equilibrium parameters and has con- cluded that collagen aggregation of the native type is an entropy-driven process. The entropy gain could be derived from a decrease in the organization of water mole- cules surrounding amino acid residues as a

1 Supported by Grant HL21016-03.

result of the fiber formation. Earlier dila- tometric studies by Cassel and Christensen (6) are also consistent with this view.

The kinetic aspects of fibril formation have been studied extensively in the past two decades (7- 13). Most authors have con- cluded that hydrophobic, as well as ionic interactions stabilize the assembly of the fibril. The time course of fibril formation reveals a lag phase and a growth phase but the relative contributions of the two types of interactions mentioned above in these steps of aggregation have not been determined. In an effort to solve this question we have investigated the effect of alkylureas on the self-assembly of collagen. It has been previ- ously demonstrated that these reagents are capable of dissociating the hemoglobin tetramer into dimers under nondenaturing conditions (14) as well as reducing the extent (15) and the rate of polymerization (16) of deoxy-sickle-hemoglobin in direct proportion to their hydrophobicity. Therefore, we have selected the alkylureas as suitable chemical probes to investigate the contribution of hydrophobic interaction in the tempera- ture-dependent native type aggregation of collagen.

0003-9861/80/140422-0$02.00/O Copyright 6 1980 by Academic Press, Inc. All rights of reproduction in any form reserved.

422

COLLAGEN FIBRIL FORMATION 423

MATERIALS

Methylurea, ethylurea, and n-butylurea were pur- chased from Pfalz Bauer, Inc., Stamford, Connecticut. N-propylurea was from Eastman, Rochester, New York, and was recrystallized from ethanol before use. TES was purchased from Calbiochem (Le Jolla, Calif.). Unlabeled glycine ethyl ester hydrochloride and 1-ethyl,3-(3-diaminopropyl)carbodiimide were from Sigma Chemical Company (St. Louis, MO.). Glycine ethyl ester hydrochloride [I-“‘ClGlycine, in ethanolic solution at a specific activity of 41 mCi/mmol, was obtained from New England Nuclear, Boston, Massachusetts.

Acid-soluble collagen was extracted from the skin of an unborn calf essentially as described by Gallop and Seifter (17). The skin was procured through a local slaughter house.

All chemicals not mentioned above were of reagent grade.

METHODS

To prepare stock solutions of collagen, the lyophilized material was dissolved in cold 0.005 M acetic acid by gently stirring overnight at 5°C. To minimize the pres- ence of unspecific aggregates the suspension was cen- trifuged at 110,OOOg for 3 hat 4°C. The collagen concen- tration in these stock solutions varied from 0.9 to 2.7 mg/ml and the solutions were discarded after about 4 weeks of storage at 4°C. The collagen preparation was characterized by polyacrylamide gel electrophoresis which revealed (Y,& and y components (18). In view of the possible variations in the content of the different crosslinked species of collagen, the same batch was utilized throughout a series of experiments.

The buffer used for the aggregation experiments had the same composition as that described by Williams et al. (1). At the final dilution the ionic strength was 0.225, the pH 7.5, and the phosphate concentration 0.03 M. The main buffering component was TES which was present in the mixture at a concentration of 0.05 M. The buffer stock solution was prepared double strength and is designated “aggregation buffer.” Collagen fibril formation was induced as follows: 0.7 ml of collagen solution in 0.005 M acetic acid and 0.7 ml of “aggrega- tion buffer” were preequilibrated to the required temperature in separate cuvettes placed in a spectro- photometer to which a thermoregulated Haake water circulator was attached. After at least 5 min the two solutions were mixed and the course of aggregation was monitored by recording the increase in optical density at 313 nm or, when less sensitivity was required, at 540 nm, in a Gilford 2000 recording spec- trophotometer. The alkylureas, when needed, were dissolved a few minutes before the experiments in the

1 Abbreviation used: TES, N-(tris(hydroxymethyl)- methyl-2-amino)ethanesulfonic acid.

“aggregation buffer.” In the experiments which required the addition of the alkylureas after the induc- tion of fibril formation the corresponding amount of each of them was carefully weighed on an empty cuvette preequilibrated at the required temperature. In a matched euvette a mixture containing collagen was taken to conditions suitable for aggregation, and, at the indicated times, it was transferred to the cuvette containing the alkylurea as powder and dissolved by gentle mixing, and the subsequent aggregation monitored as described. In control experiments per- formed by transferring the aggregating mixture to empty cuvettes, it was observed that this procedure did not affect the rate of aggregation per se.

In view of the well-known difficulties encountered in applying the conventional methods of protein deter- mination to collagen solutions, their assay was per- formed by the estimation of the free carboxyl groups according to Carraway and Koshland (19). During the coupling step, %-labeled glycine ethyl ester was used as the blocking nucleophile at a concentration of 0.7 M and a specific activity ranging from 4.35 to 4.9 x lo6 cpm/mmol. The protein concentration was calculated by assuming a carboxyl content of 220 mol/mol of calf skin collagen (20) and determining the radioactivity present in the dialyzed sample.

RESULTS

Figure 1 shows the effect of various alkyl- ureas on the time course of aggregation. The inhibition of the kinetics becomes more pronounced as the chain length of the alkyl group and, consequently, its hydrophobic- ity, increases. In this particular experi- ment the control sample had a lag phase of about 2 min and a temperature of 37°C was chosen to enable the comparison of all rates on a linear graph. The final turbidity change approaches a similar value in all the inhib- ited samples, which does not differ signifi- cantly from the control if the aggregation is

FIG. 1. Rate of collagen fibril formation in the pres- ence of various alkylureas at a concentration of 0.125 M at 3’7°C. Turbidity was assayed as optical density at 540 nm. The collagen concentration was 0.63 mg/ml. The delay time of the control was about 2 min.

424 SUAREZ, VELIZ, AND NAGEL

Log Time Immutes)

FIG. 2A. Semilogarithmic plot of the rate of fibril formation in the presence of different alkylureas at a concentration of 0.0625 M. The temperature was kept at 25°C and the collagen concentration was 0.252 mg/ml. (01 control; (0) methylurea; (0) ethylurea; (A) propyl- urea.

allowed to proceed for longer periods (see below).

If the temperature for fibril formation is decreased to 25°C the differences of the inhibitory capacity between the particular alkylureas become more striking. For instance, the t5,,, i.e., the time required to reach half the value of the final turbidity, is 32 min, in the absence of alkylurea, at 25”C, while it is 52, 110, and 144 min, if the fibril formation occurs in the presence of 0.0625 M methyl-, ethyl-, and propylurea respec- tively. If the turbidity in these experiments is plotted versus the logarithm of time, as shown in Fig. ZA, the slopes at the midpoints appear quite similar within experimental error.

If the conditions are chosen so that the lag phase approaches zero, the kinetics of aggregation appears to be insensitive to the alkylureas tested, even though their con- centrations were doubled. In order to assess the validity of the view that these results reflect a preferential effect of the alkylureas during the lag phase, these additives were introduced in the aggregating mixture at various time lapses after the induction of fibril. formation (Fig. ZB). A significant reduction in the extent of inhibition is observed if ethylurea is added after 2.5 min have elapsed from the initiation of fibril formation. The reduction of the inhibition

shown in the semilogarithmic plot is more striking if the ethylurea is added 5 min after the induction. The significance of this influ- ence on the inhibition of the kinetics is emphasized if it is pointed out that the delayed additions of ethylurea were done during the lag phase which lasted about 10 min, and the time required to reach the final turbidity was about 270 min in the control sample, i.e., the system was free of ethyl- urea only less than 3% of the total polymer- ization time.

In some experiments of fibril formation ethylurea was present only during the lag phase and then the reaction mixture was dialyzed overnight against 0.005 M acetic acid. The rate of fibril formation of the dialyzed treated sample was indistinguish- able from a control mixture preincubated in the absence of ethylurea.

In several experiments performed at dif- ferent initial concentration of soluble colla- gen only an inhibition of the kinetics, expressed as increase of the tsO, was observed. The total turbidity change is not affected by the presence of the alkylurea. Table I summarizes these findings over temperatures ranging from 25 to 3’7°C and at two different concentrations of ethyl-

1.2 1.4 1.6 I.8 20 2.2 2.4 2.6 2.8 3.0

Log Time (minutes)

FIG. 2B. Semilogarithmic plot of the rate of fibril formation in the presence of ethylurea (0.0313 M) added at different intervals during the lag phase. The tem- perature was 25.5% and the collagen concentration 0.226 mg/ml. The lag time in the control was about 10 min. (A) control; (0) added at 0 min; (0) added at 2.5 min; (A) added at 5 min. See text for experimental details.

COLLAGEN FIBRIL FORMATION

TABLE I DISCUSSION

COMPARATNEEFFECTSOFETHYLUREAONRATEAND EXTENTOF COLLAGENFIBRIL FORMATION"

Collagen Ethylurea Tempera- concentra- concentra-

ture tion tion Rate, AA, (“C) (mg/ml) (M) Rate,=

25 0.630 0.1250 0.18 0.96 37 0.252 0.0625 0.76 0.88 30 0.252 0.0625 0.86 1.06 25 0.252 0.0625 0.59 1.03 27 0.126 0.0625 0.31 1.00 28 0.126 0.0625 0.52 0.90 31 0.126 0.0625 0.56 1.04

” Abbreviations: e, in the presence of ethylurea; c, control; rate, (tJ’; AA, total turbidity change.

We have demonstrated here a correlation of the extent of inhibition of the rate of fibril formation with the aliphatic chain length of the alkylureas. Since the chain length is in close accordance with the hydrophobicity, we conclude that these findings support the notion that hydrophobic interactions are involved in the process of fibril formation. An analogous increasing effectiveness of the alkylureas as inhibitors of deoxy-sickle- hemoglobin gelation reported previously by Elbaum et al. (15) was explained in terms of the same principle. Earlier findings by Hayashi and Nagai (10) of the inhibition of the rate of the heat precipitation of collagen by urea were similarly interpreted.

urea. In this table the fourth column refers to the ratio of rates, expressed as the reciprocal of tsO, in the presence and absence of ethylurea and the last column lists the ratios of the final turbidity change in an analogous way. These results might suggest that the alkylureas do not affect the final equilibrium of the collagen fibril, their effects being mainly associated with kinetic parameters.

The alkylureas appear to affect the rate of the process without altering the mechanism in view that the slope of the increase of turbidity versus the logarithm of time in the presence of various alkylureas remains invariable. This criterion for identity of mechanism is compatible with several models of fibril formation (2, 21).

To provide a further explanation for the kinetic effects of the alkylureas the influence of the temperature on the kinetics of aggregation was investigated. Figure 3 is an Arrhenius plot of in vitro fibril formation in the absence and in the presence of ethyl- urea at a concentration of 0.0313 M. The kinetic constant was assumed to be of the first order with respect to the initial collagen concentration and was expressed as the reciprocal of the t,, divided by the molarity of tropocollagen according to treatments followed by Williams et al. (1). The plot clearly shows a difference in the slope between the control and the inhibited sample. In three series of experiments the plot was linear between 25 and 35°C and beyond these limits it was anomalous. The calculated activation energies for fibril formation on this temperature range were 28.36 and 45.69 kcallmol in the absence and presence of ethylurea, respectively.

The time course of fibril formation is generally described in terms of a lag phase, when no change in turbidity is apparent, and a growth phase, in which the turbidity increases sharply. Based on the lack of temperature dependence of the lag phase, if a short induction period at relatively high temperatures is allowed, Gelman et al. (22) have argued that in this phase filaments would be formed and stabilized in large part by electrostatic forces. In contrast, our

o-

-l-

FIG. 3. Arrhenius plot of collagen fibril formation within the temperature range 25.5-35°C. The collagen concentration was kept at 0.226 mg/ml. (A) control; (0) in presence of 0.0312 M ethylurea.

1

J 3’ I

426 SUAREZ, VELIZ. AND NAGEL

observations on the dependence of the inhibition by ethylurea on the time of addi- tion after the aggregation has been initiated appears to suggest that hydrophobic forces are operating during the lag phase. This, of course, does not exclude a partial additional involvement of electrostatic interaction. If hydrophobic interactions were acting only during the growth phase, the rate of polym- erization would be independent of the time of addition of the alkylurea within the lag phase in contradiction with our experi- mental observations. However, as a result of a systematic analysis of the temperature dependence along the course of fibril for- mation, Gelman et al. (21) have pictured the lag phase as a succession of two stages, the first of which would be temperature dependent (initiation step). If this model is correct it would be tempting to hypothesize that during this initiation step self-assem- bled intermediates are generated via hydro- phobic interactions.

The effect of the alkylureas during the lag phase appears to be reversible because the kinetics of fibril formation does not differ from that of the untreated collagen after the samples are dialyzed to remove the reagent.

Under the described experimental condi- tions fibril formation from our soluble colla- gen preparations is only kinetically inhib- ited. The final turbidity increase has been shown to be linearly dependent on the amount of precipitated material (7,9) and is considered a good parameter to quantitate the fibril formed. The lack of effect of the alkylureas on the final turbidity raises the possibility that these reagents do not modify the equilibrium of the fibril with the soluble tropocollagen molecule.

The absence of an effect on the final tur- bidity change also affords strong evidence against any denaturing effect of the alkyl- urea on collagen on the basis that the native triple helical conformation is a prequisite for the formation of the fibril. The restoration of the capability to form fibril after removal of the alkylurea by dialysis also points to the same conclusion. A renaturation during the time of dialysis is an unlikely event because this process is known to be very slow and requires days for its completion (23).

The increase in the activation energy of the overall process of fibril formation in the presence of ethylurea suggests that these reagents act as destabilizers of intermedi- ates. These could be considered formally as the transition state in the pathyway of solu- ble collagen to fibril. This view would become more tenable if the lack of an effect on the final turbidity is a true reflection of the intrinsic equilibrium of the self-associa- tion. Under suitable experimental condi- tions Bensusan and Hoyt (8) and Comper and Veis (24) have found that the energy of activation is insensitive to changes in the ionic strength. This finding gives additional evidence that the intermediates formed in rate-limiting steps are stabilized by hydro- phobic interactions,

We have found lower energies of activa- tion than those reported by other investi- gators (1). A likely explanation for this dif- ference could be found in the method of purification of collagen. Williams et al. (1) used collagen extracted at acid pH, as ours, but later subjected to several cycles of salt precipitation (25). Our preparation might well have contained a higher proportion of aggregates which could promote an increase in the rate of polymerization (26).

We have found the power dependency of the rate on initial collagen concentration to be close to 1 in controls as well as in alkyl- urea-treated samples. This finding does not invalidate previous efforts of constructing a mechanistic model for fibril formation in terms of nucleation and growth (27,28) but renders the formation of hypothetical nuclei a very fast step which is not rate limiting. Furthermore, recent electron micrographs of collagen samples obtained while in the lag phase have revealed microfibrillar struc- tures (1,29). These initial aggregates could well be generated through an alkylurea- sensitive step.

REFERENCES

1. WILLIAMS, B. R., GELMAN, R. A., POPPKE, D. C., AND PIEZ, K. A. (1978) J. Biol. Chem. 254, 6578.

2. OOSAWA, F., AND ASAKURA, S. (19’75) Thermody- namics of the Polymerization of Protein, Academic Press, New York.

COLLAGEN FIBRIL FORMATION 427

3. WOOD, G. C. (1964) Int. Rev. Connect. Tissue Res.

21, 1. 4. BOOKCHIN, R. M., AND NAGEL, R. L. (1974)

Semin. Hematol. 11, 577.

5. COOPER, A.. (1970) &o&em. J. 118, 355. 6. CASSEL, J. M., AND CHRISTENSEN, R. G. (1967)

Biopolymers 5, 431.

7. GROSS, J., AND KIRK, D. (1958)J. Biol. Chem. 233, 355.

8. BENSUSAN, H. B., AND HOYT, B. L. (1958) J. Amer. Chem. Sot. 80, 719.

9. WOOD, G. C., AND KEECH, M. K. (1960) Biochem. J. 75, 588.

10. HAYASHI, T., AND NAGAI, Y. (1972) J. Biochem. (Japan) 72, 749.

11. HAYASHI, T., AND NAGAI, Y. (1973) J. Biochem.

(Japan) 74, 253. 12. HONYA, M. J. (1975) J. Biochem. (Japan) 75, 1037. 13. FESSLER, J. H., AND TANDBERG, W. D. (1975) J.

Supramol. Struct. 3, 17. 14. ELBAUM, D., AND HERSKOVITS, T. T. (1974) Bio-

chemistry 13, 1268. 15. ELBAUM, D., NAGEL, R. L., BOOKCHIN, R. M.,

AND HERSKOVITS, T. T. (1974) Proc. Nat. Acad.

Sci. USA 71, 4718.

16. ELBAUM, D., HARRINGTON, J. P., BOOKCHIN, R. M., AND NAGEL, R. L. (1978) Biochim. Biophys. Acta. 534, 228.

17. GALLOP, P. M., AND SEIFTER, S. (1963) in Methods

in Enzymology (Colowick, S. P., and Kaplan, N. O., eds.), Vol. 6, p. 635, Academic Press,

New York. 18. NAGAI, Y., GROSS, J., AND PIEZ, K. A. (1964)

Ann. N. Y. Acad. Sci. 121, 494. 19. CARRAWAY, K. L., AND KOSHLAND, D. E. (1972) in

Methods in Enzymology (Hirs, C. H. W., and

Timasheff, S. N., eds.), Vol. 25, p. 616, Aca- demic Press, New York.

20. PIES!, K. A., AND GROSS, J. (1960) J. Biol. Chem.

235, 995. 21. CASSEL, J., MANDELKERN, L., AND ROBERTS,

D. E. (1962) J. Amer. Leather Chem. Assoc. 57,

556. 22. GELMAN, R. A., WILLIAMS, B. R., AND PIEZ,

K. A. (1979) J. Biol. Chem. 254, 180.

23. VON HIPPEL, P. H. (1967) in Treatise on Collagen (Ramachandran, G. N., ed.), Vol. 1, p. 302, Academic Press, New York.

24. COMPER, W. D., AND VEIS, A. (1977)Biopolymers 16, 2113.

25. CHANDRAKASAN, G., TORCHIA, D., AND PIEZ, K.

(1976) J. Biol. Chem. 251, 6062. 26. WOOD, G. C. (1962) Biochem. J. 84, 429.

27. WOOD, G. C. (1960) Biochem. J. ‘75, 598. 28. COMPER, W. D., AND VEIS, A. (1977) Biopolymers

16, 2133.

29. TRELSTAD, R. L., HAYASHI, K., AND GROSS, J. (1976) Proc. Nat. Acad. Sci. USA 73, 4027.