The Reaction of Hydrogen Chloride with Dry Proteins
Part 3—The Adsorption of Water Vapour on Hydrochlorides
of Collagen and of Silk Fibroin
[Read before the Otago Branch, November 14. 1950: received by the Editor, November 30, 1950]
A knowledge of the behaviour of collagen in the presence of water and acids was early recognised as fundamental to the science of leather chemistry, so that, in some of its aspects, the problem has been widely studied. In aqueous acids collagen absorbs water and swells until, at the maximum point, it is found to have taken up about five times its own weight of water, together with a quantity of acid approximately equivalent to the basic groups of the protein. These changes have been fairly satisfactorily explained as a Donnan membrane effect modified by the chemical composition and mechanical rigidity of the collagen fibres.
However, when neither water nor acid is present in excess, the protein-acid-water system assumes a very different form, which appears to have been entirely neglected. The adsorption of water vapour on collagen has already been discussed in detail by the present author (1948), who has also shown in Parts 1 and 2 of this series (1950, 1952), that dry collagen and dry silk fibroin can each combine with anhydrous hydrogen chloride to form a series of stable derivatives in which the amount of combined acid can again be related to the amino-acid composition of the protein. It is now proposed to combine these two general reactions and consider the adsorption of water vapour on several protein hydrochlorides.
Collagen was used in the form of standard hide powder. Pure silk fibroin was from the same sample as that used in Part 2. Calfskin was subjected to the normal tannery processes of soaking, liming, fleshing and scudding, partial deliming and bating. It was then completely delimed in the laboratory with ammonium chloride solution, thoroughly washed with distilled water, dehydrated with four changes of acetone, and cut into strips 1·5 cm. wide. The product may be regarded as similar in composition to the hide powder, but probably received a milder enzyme treatment and so may still have been associated with some elastin.
All these materials, after drying at 100° C. over phosphorus pentoxide in vacuo, were converted to hydrochlorides by means of
anhydrous ethereal hydrogen chloride by the method previously described (Green, 1950).
Silk fabric was identical with that used in the earlier work.
Water vapour adsorption isotherms were measured at 25° with the isopiestic apparatus of Robinson and Sinclair (1934), all experiments being made in duplicate.
The effect of water adsorption on the dimensions of calfskin was studied by using the same apparatus to condition strips of dry skin 1·5 cm. wide and about 12 cm. long. Each strip was fitted with a small clamp near either end, the distance between the jaws being adjusted to 10 cm. at the beginning of the experiment. When the dry specimens had been conditioned for a further three days over concentrated sulphuric acid, they were quickly removed and the terminal clamps were attached to a measuring device which applied a constant tension and, by means of a rack and pinion, indicated changes of length to within 0·1 mm. After the zero reading for each strip had been noted, a series of observations were made by equilibrating the strips for three days at each relative humidity (aw) and measuring the new length.
At each aw the electrical conductivity of the calfskin was determined by clamping a specimen between two disc-shaped electrodes 1 cm. in diameter. The discs, held against the two surfaces of the specimen by a constant pressure, were connected to a bridge circuit of the type used for measuring electrolytic conductance. The thickness of the specimen was measured with a micrometer. This arrangement, while not designed to give absolute values, was suitable for indicating any important changes in conductance.
Since proteins appear to vary in their reaction to different methods of drying, and since the water content at any aw may depend on the method chosen, before proceeding to an account of the adsorption isotherms, we report here some observations on drying silk fibroin.
Silk fibroin in equilibrium with the air of the laboratory was dried by three methods—viz., dehydration over sulphuric acid in the evacuated rocking desiccator at 25°; heating in the air oven at 103°; heating in vacuo at 100°. Constant weight was achieved in six days by the first method and in three days by the other two. It was found that the first two methods gave identical values for the water content, but that the dry product of either lost a further 0·27 g. per 100 g. fibroin when heated in vacuo at 100°. Placing the vacuum-dry sample in the air oven caused its weight to increase by the same small amount, which was again lost if the fibroin was re-heated in vacuo. It seems probable that the difference in weight loss observed between heating in air and in vacuo was due to the fact that in the air oven the sample is in equilibrium, not with dry air, but with air of relative humidity about 0·01.
Water vapour adsorption isotherms were measured for the following modifications and derivatives of silk:—
Silk fabric dried in the air oven at 103°.
Silk fibroin dried in the air oven at 103°.
Silk fibroin dried at 25° over concentrated sulphuric acid.
Silk fibroin soaked in water and dehydrated to aw = 0·05 only.
Silk fibroin dried in vacuo at 100°.
Silk hydrochlorides containing respectively 1·12%, 2·26% and 2·65% combined hydrogen chloride.
About eighteen points were measured on each curve, the points being spaced rather more closely in the lower humidity range. The experimental results, interpolated for round values of aw, are presented in Tables I and II, the adsorbed water being expressed as g. per 100 g. of the appropriately dried adsorbent. The figures for the hydrated specimen refer to the amount of water which could be removed on drying at 25°. Table III shows similar data for a series of collagen hydrochlorides and for collagen dried in vacuo at 100°. The latter had been omitted from the previous study of adsorption on collagen and was needed here for direct comparison with the hydrochlorides. Analysis showed that there was no loss of acid from the hydro-chlorides, even at high aw. This is, of course, an indispensable condition for the interpretation of gains in weight as adsorption of water.
The general equation of Brunauer, Emmett, and Teller (1938) for multimolecular adsorption was applied to each isotherm, using the range between aw = 0·05 and aw = 0·40 to calculate the constants vm and C. For each adsorbent, this range included from eight to ten experimental points, which were subjected to the well-known B.E.T. transformation and fitted to a linear regression line. vm is interpreted as the weight of adsorbate filling the first molecular layer of 100 g. adsorbent; and C is related to the heat of adsorption of the same layer. The value of n, the maximum number of adsorbed molecular layers, was calculated from each curve between aw = 0·60 and aw = 0·70. In Table IV, we present the B.E.T. constants for all the adsorbents examined, and in Fig. 1 we compare the experimental points with the B.E.T. plots for three isotherms.
|aw||g. H2O per 100 g. dry adsorbent|
[Footnote] *—Oven-dried silk fabric
[Footnote] *—Oven-dried silk fibroin
[Footnote] *—Fibroin dried at room temperature
[Footnote] *—Hydrated fibroin, dehydrated to 2.9% water at aw = 0.05
|aw||g. H2O per 100 g. adsorbent dried in vacuo at 100°|
|1.12% HCl||2.26% HCl||2.65% HCl|
|aw||g. H2O per 100 g. adsorbent dried in vacuo at 100°|
|1.60% HCl||2.10% HCl||3.33% HCl||3.77% HCl||4.80% HCl||5.25% HCl|
|Adsorbent||Method of drying||HCl content %||vm g. H2O/100 g. adsorbent||v′m g. H2O/100 g. original protein*||C||n|
|Silk fabric||Air oven, 103°||–||3.97||4.0||5.3|
|Silk fibroin||Air oven, 103°||–||4.20||5.4||5.9|
|In vacuo, 100°||–||4.19||7.0||6.1|
|Over H2SO4, 25°||–||4.18||14.7||6.3|
|To aw = 0.05, 25°||–||4.12||25.2||6.6|
|Silk hydrochloride||In vacuo, 100°||1.12||4.11||4.16||7.1||6.1|
|Collagen||In vacuo, 100°||–||9.43||6.2||5.6|
|Collagen||In vacuo, 100°||1.60||8.39||8.53||6.8||6.0|
[Footnote] Protein content of hydrochloride is assumed to be total dry weight minus HCl.
Fig. 1—Water vapour adsorption isotherms at 25°. Continuous lines represent B.E.T. plots
calculated from the constants in Table IV. Broken lines—experimental curves at high aw.
In appearance the dry hydrochlorides were indistinguishable from the parent proteins, but in some of the higher collagen hydrochlorides this appearance underwent a marked change at high aw. Above aw=0·7 they became damp and the individual particles appeared to have shrunk and to be adhering together. At still higher humidity, the hydrochlorides were very moist and gelatinised; and this effect increased with increasing hydrogen chloride content of the specimen. The change was irreversible, the gelatinised material drying out to a hard, dark brown, horny mass on reducing the humidity. No such phenomenon was observed with silk hydrochlorides or with collagen hydrochlorides containing less than about 3·5% hydrogen chloride.
Fig. 2—Length changes of strips of calfskin, and of calfskin containing
combined HCl, plotted against aw.
The behaviour of collagen hydrochloride was reflected in the results obtained with calfskin. Changes in length with aw were recorded for strips of calfskin, and of calfskin containing combined hydrogen chloride to the extent of 2·2% and 5·2%, respectively. Figure 2 shows that the specimen containing the greater proportion of hydrogen chloride increased slightly in length at low humidities, but began shrinking fairly suddenly at about aw = 0·7; whereas the other two samples continued to increase in length over the whole humidity range. Gelatinisation appeared soon after shrinkage had begun and increased rapidly, so that at aw = 0·92 the tensile strength of the higher hydrochloride was too low to permit a measurement to be taken. Changes in electrical conductance of the same samples are shown in Table V.
|aw||Specific Conductance (ohm−1cm.−1 × 10−6)|
|Calfskin||Calfskin with 2.2% combined HCl||Calfskin with 5.2% combined HCl|
|0.732||Less than 1||Less than 1||Less than 1|
As a complement to these experiments at high aw, samples of collagen and silk hydrochlorides were immersed in distilled water. The collagen derivatives all swelled and became more or less gelatinised, but the appearance of the silk fibroin hydrochlorides remained unaltered. Hydrochloric acid could be detected immediately in the water and, by sufficiently prolonged washing on a sintered glass filter, all the combined chlorine could be removed. The collagen samples were free from combined chlorine in twenty-four hours, but the same result with silk required a week's continuous washing.
The above experimental results are best considered in two sections; the region of lower aw, from which the constants vm and C are derived, and the region of high aw, where the observed isotherm diverges from the B.E.T. curve.
Table I shows that silk fibroin qualitatively resembles collagen, elastin, and casein, discussed in our earlier papers (1948, 1949) in that its adsorptive power is diminished by severe drying. Reference to Table IV shows that the B.E.T. analysis attributes this lessened affinity for water to changes in C, related to the energy of adsorption, and that the quantity of water in the first molecular layer, vm, scarcely deviates from a mean of 4·17 g. per 100 g. protein. The values of 4·07 for vm and 12·78 for C found by Bull (1944) will be seen to fit well into the series in Table IV, when it is remembered that Bull's data represent the mean of two isotherms, one obtained by adsorption on silk dried at 80° in vacuo, and one obtained by desorption from silk conditioned over water for 24 hours.
All these experimental values for vm come very near the figure 3·97 which can be derived by reckoning one water molecule adsorbed for each polar side chain, using the amino-acid composition summarised by Cohn and Edsall (1943). A later analysis reported by Tristram (1949) differs in including appreciable amounts of glutamic and aspartic acid residues and leads to the much higher estimate of 5·15. The observed values of vm therefore fall well within the range demanded by the analytical figures, and may be said to support the hypothesis of equality between the number of polar groups and the number of water molecules adsorbed. Whether each water molecule is adsorbed on a given group or whether, in the complexity of the protein molecule, some kind of “activation” arising from the polar group finally results in the binding of a water molecule at a nearby peptide group or elsewhere is a question which can have little meaning at the present time. We believe that there is a one-to-one correspondence between polar groups in any protein and water molecules in the first adsorbed layer; but that this may be modified by steric factors which either reduce adsorption, or, as has already been suggested for the open structure of collagen, permit some additional water molecules to be co-ordinated on the polypeptide chain. In the case of the compact silk fibroin, there seems to be no need to postulate adsorption beyond that controlled by the polar side chains.
The constants for collagen dried in vacuo at 100° are almost identical with those previously reported (Green, 1948) for the same material dried at 103° in the air oven.
Before considering adsorption on the hydrochlorides, we may recall what is known of their probable constitution. Some of the acid is, of course, held by salt formation at the —NH2 and other basic groups of the side chains; but it was shown in Part 2 that, in all silk hydrochlorides containing more than 0·2–0·3% HCl and in all collagen hydrochlorides containing more than 3·3%HCl, some of the chlorine is directly linked to carbon in the β-position of the side chains originally represented by serine and threonine,
The properties of these β-chloro-α-aminoacid residues are unusual. The chlorine can be removed quantitatively by titration with alkali and even by prolonged washing with water. On the other hand, free β-chloro-α-aminopropionic acid forms a solution in water which is stable to 2N lithium hydroxide (Fischer and Raske, 1907). It can only be concluded that the ease of hydrolysis of chloraminoacid residues in protein hydrochlorides results from their close association with the polypeptide chain and the centres which co-ordinate water.
The hydrochlorides of silk were found to adsorb no less water than the similarly dried silk itself, and the vm values, when calculated on the protein moiety of the adsorbent, were not significantly different from those for the parent protein. The two other B.E.T. constants were likewise unchanged by the presence of combined hydrogen chloride. As there was no visible difference in physical properties between silk and its hydrochlorides throughout the adsorption, it appears that all the points of the original molecule which adsorbed water retained that power after reaction with hydrogen chloride. This is natural enough for the basic side chains, but leads to the rather unexpected conclusion that the chlorine atoms linked directly to carbon were just as effective in the co-ordination of water as were the hydroxyl radicals which they had replaced. However, it is consistent with the unusual reactivity toward water which we have already noted in the β-chloraminoacid residues, and emphasises the danger of assuming that atoms bound to a protein molecule will react as they would in a small molecule, such as an amino acid. If any other example were needed to illustrate this point, the continuous sigmoid water vapour adsorption isotherm, exhibited by all proteins but by no amino acid, should suffice.
A quite different behaviour was observed in collagen, where even the lowest hydrochloride had markedly less adsorptive power than the original similarly dried protein, and adsorption was still further depressed by greater proportions of combined chlorine. Throughout the series, the energy term, C, scarcely differed from the value shown for collagen, and almost the whole change in adsorption was associated with a diminished vm. Using the conclusions already drawn from the simpler data for silk, we shall assume that the ability of the polar side chains of collagen to co-ordinate water is unchanged by reaction with hydrogen chloride. The sharp fall in vm occasioned by even small quantities of combined HCl must therefore be assigned to reduced adsorp-
tion on peptide groups, which have been regarded as contributing about 40% of the active points of the first layer (Green, 1948). This draws attention to the fact that, for the three lowest hydrochlorides, the mean values of vm, corrected for protein content, and of C are 8·4 and 6·1, respectively, almost identical with 8·2 and 6·6 reported for oven-dried denatured collagen in the earlier paper. The coincidence suggests that, in the process of preparing the hydrochloride, which involves the adsorption and subsequent desorption of large quantities of loosely held HCl, a change akin to denaturation may take place. The excess adsorbed hydrogen chloride must certainly disturb the various salt linkages and secondary bonds in the protein, and, on its removal, hydrogen bonds between peptide groups may appear in greater numbers than before, so that in the final product there are fewer free centres available for the co-ordination of water.
The value of vm exhibits a second sharp change on passing to hydrochlorides containing more than 3·3% combined HCl. We shall not attempt an explanation of this change beyond pointing out that this is the very region where the collagen begins to contain more combined chlorine than can be accounted for by all the basic groups. It may be more than a coincidence that a change in vm appears at the point where we have demonstrated a difference in the state of the combined chlorine.
The third parameter, n, of the B.E.T. equation, which assumes importance in the moderately high humidity range, has a mean very close to 6·0 for all the adsorbents studied. This may be regarded as the maximum number of molecular layers of adsorbate which can be built up without restriction by the capillary walls of the adsorbent.
At still higher humidities, the experimental plot and the general B.E.T. curve begin to diverge, through uptake of more water than can be explained by multimolecular adsorption theory. This has often been ascribed to condensation of liquid, beginning in the smallest capillary spaces of the adsorbent and extending to the larger pores as aw increases. If that is so, Figure 1 shows that new phenomena, arising from the presence of liquid water, may be looked for when aw exceeds 0·7.
We have seen that all the combined acid can be removed from a protein hydrochloride by prolonged washing with water, so that we may assume that some acid will be removed from combination as soon as any liquid water appears in the capillaries. Any changes that ensue will now depend on the amount of combined HCl and the amount of liquid water present, and may be governed by equilibria similar to those approached from the opposite direction when a protein is titrated with aqueous hydrochloric acid. If the hydrogen chloride content of the specimen is less than equivalent to all the basic groups, the concentration of the acid in the aqueous phase will be low; but if the basic group equivalent (3·30% HCl for collagen) is exceeded, all the excess acid will pass into solution in the condensed water with which it comes in contact. As the aqueous vapour pressure is increased, the internal area covered by liquid water will spread, bringing more of the combined acid into equilibrium with an aqueous phase. Therefore, at all points above aw = 0·7, the protein hydrochlorides will contain more or less extensive spaces filled with hydrochloric acid solution, of
hydrogen ion concentration depending on the relative proportions of basic groups and acid.
These considerations can be used in a qualitative explanation of the physical changes observed at high aw. The shrinkage and gelatinisation occurring in the higher collagen hydrochlorides are more severe than ordinarily produced by the action of aqueous acid on collagen, and more closely resemble the effects of hot water or of treatment with strong solutions of calcium chloride in the cold. Reference to the titration curve of collagen (McLaughlin and Theis, 1945a) suggests that the condensed water will reach an acid concentration, differing in capillaries of unequal size, as high as pH 1·0. At this pH there is very little osmotic swelling, and the structural cohesion of the fibres is greatly reduced, so that shrinkage occurs at room temperatures, as has already been observed when collagen strips are placed in aqueous acid solutions (McLaughlin and Theis, 1945b). The loss of stability at low pH values is generally attributed to breaking of the short links or hydrogen bonds between peptide groups in adjacent chains. However, we may perhaps speculate on the part played by co-ordination of aliphatic β-hydroxyl groups—absent from the higher hydrochlorides—in stabilising collagen.
In the hydrochlorides of collagen containing less than 3·3% HCl, some free basic groups remain, and the aqueous titration curve shows that the equilibrium pH lies in the range 2–4. Here acid swelling is fairly strong, but the shrinkage temperature is as high as 45–55°. The condensed water in these adsorbents is insufficiently acid to disorganise the fibre structure or induce gelatinisation, and no shrinkage is observed.
Although the silk hydrochlorides all contained an excess of acid over the basic groups, they underwent no apparent change at high aw, and retained the integrity of their fibres even on soaking in water. This is in accord with the known behaviour of silk fibroin and its compact structure.
The conductivity changes also agree with the concept of gradually extending volumes of condensed water containing hydrogen and chloride ions. Measurable conductivity appeared first in the highest hydrochloride, near aw = 0·75, but could not be detected in the lower hydrochloride until aw = 0·85. The rapid increase in conductivity of the former was probably associated with the radical breakdown of fibre structure and the early establishment of a continuous aqueous phase. Collagen itself and the lower hydrochlorides retained their fibrous organisation even at the highest humidity measured, and this fact is reflected in their much higher electrical resistance.
The above discussion provides strong support for the use of the B.E.T. equation in analysing water vapour adsorption isotherms for protein materials. Both the remarkable constancy of vm for some proteins, such as silk and casein, and its equally remarkable sensitivity for collagen can be reasonably explained in terms of the molecular structures of these adsorbents, which convinces us that vm measured in this way is a characteristic with real physical significance for each adsorbent.
The extended equation can be made to fit the-observed isotherms to aw = 0·7, and, at the very point where divergence appears, the collagen hydrochlorides begin to exhibit abnormal properties which are best explained by postulating the presence of liquid water.
Water vapour adsorption isotherms at 25° have been measured for several modifications of silk and for hydrochlorides of silk and collagen.
The curves have been analysed by the method of Brunauer, Emmett and Teller, and the derived constants have been shown to support the hypothesis of adsorption on polar groups of the protein molecule.
The power of basic groups and of aliphatic hydroxyl to co-ordinate water molecules is unaffected by reaction with hydrogen chloride.
The higher hydrochlorides of collagen undergo irreversible shrinkage and gelatinisation when the relative aqueous vapour pressure exceeds 0·7. At the same time their electrical conductivity increases rapidly with aw. These changes have been shown to support the explanation already advanced concerning the nature of collagen hydrochlorides. They also confirm the hypothesis of capillary condensation of water in the region of moderately high aw where the B.E.T. equation becomes inadequate.
The author wishes to thank Professor F. G. Soper for permission to use the facilities of his laboratory, and the Royal Society of New Zealand for a grant from their Research Fund.
Brunauer, S., Emmett, P. H., and Teller, E., 1938. Adsorption of Gases in Multimolecular Layers. J. Amer. Chem. Soc., vol. 60, p. 309.
Bull, H. B., 1944. Adsorption of Water Vapor by Proteins. Ibid., vol. 66, p. 1499.
Cohn, E. J., and Edsall, J. T., 1943. Proteins, Aminoacids and Peptides, p. 358. Reinhold Publishing Corp., New York.
Fischer, E., and Raske, K., 1907. Verwandlung des 1-Serins in d-Alanin. Berichte, vol. 40, p. 3717.
Green, R. W., 1948. The Adsorption of Water Vapour on Collagen and Elastin. Trans. Roy. Soc. N.Z., vol. 77, p. 24.
—— 1949. The Adsorption of Water Vapour on Casein. Ibid., vol. 77, p. 313.
—— 1950. The Reaction of Hydrogen Chloride with Dry Proteins, Part 1. Ibid., vol. 78, p. 291.
—— 1951. The Reaction of Hydrogen Chloride with Dry Proteins, Part 2. Ibid., vol. 79, p. 485.
McLaughlin, G. D., and Theis, E. R., 1945a. The Chemistry of Leather Manufacture, p. 109. Reinhod Publishing Corp., New York.
—— 1945b. Ibid., p. 125.
Robinson, R. A., and Sinclair, D. A., 1934. The Activity Coefficients of the Alkali Chlorides and of Lithium Iodide in Aqueous Solution from Water Vapour Pressure Measurements. J. Amer. Chem. Soc., vol. 56, p. 1830.
Tristram, G. R., 1949. Amino-acid Composition of Purified Proteins. In Anson, M. L., and Edsall, J. T., Advances in Protein Chemistry, vol. 5, p. 142.