Volatile buffers can override the “pH memory” of subtilisin catalysis in organic media

Abstract

The protonation state and activity of enzymes in low-water media are affected by the aqueous pH before drying (“pH memory”). However, both protonation and activity will change if buffer ions can be removed as volatile or organic-extractable weak acids or bases. With NH₄OOCH buffers, in which both ions can be removed, pH memory disappears completely for subtilisin-catalyzed transesterification in hexane. Only weak pH memory is found with buffers having one volatile component, NH₄-phosphate and NaOOCH. The changes in ionization state result from proton exchanges like Protein-COO⁻NH₄⁺ → Protein-COOH + NH₃ (g) and Protein-NH₃⁺HCOO⁻ → Protein-NH₂ + HOOCH (g). An equivalent, complementary picture is that net charges on the protein and buffer ions must remain equal and opposite. With NaOOCH buffers, loss of some HCOO⁻ ions gives a more negative net charge on the protein, balanced by the excess Na⁺. With NH₄-phosphate buffers, loss of NH₃ gives protein with a more positive net charge. The resulting catalytic activities were high and low, respectively, similar to those after drying from Na-phosphate buffers of optimal (8.5) and acid pH. All of the above effects have been demonstrated for both covalently immobilized subtilisin and the lyophilized free enzyme. Subtilisin lyophilized from NH₄OOCH buffers gave pH ≈4 after redissolution in water, probably because removal of HCOO⁻ counterions remains incomplete. The resulting catalytic activity was low. The effects are discussed in relation to the possible locations, in low-dielectric media, of the positive charge that balances the net negative catalytic triad in active subtilisin.

A central theme of enzymology is that protein ionization, which is pH-dependent, is a key determinant of enzyme activity. In low-water media (1, 2, 3), the initial demonstration of a “pH memory” effect (4) caused great interest. Enzymic activity was found to be critically dependent on the aqueous pH before drying, even when there is no true aqueous phase under the assay conditions. Another manifestation of this phenomenon is the observed effect on the solid state stability of proteins of ionization of labile side chains (5).

pH memory has been attributed to a fixation of protein catalytic group ionizations after drying of the biocatalyst preparation. The standard model for this fixation process is the maintenance of all of the ionization states present (protein and buffer species) before the freezing of the preparation (6). The various effects of additives on the ionization state of low-water proteins has been reviewed recently (7).

The most systematic studies of pH memory have all used Na or K phosphate buffers (4, 8, 9). Many buffers that might be used are derived from weak acids or bases that are volatile enough to be removed during drying (or hydrophobic enough to be extracted into an organic solvent). Removal of buffer ions in this way will irreversibly change the protonation state of the enzyme. We show here how this leads to the disappearance of most pH memory and its replacement by a dependence on whether both buffer ions, or only one of them, are removable in this way.

MATERIALS AND METHODS

Chemicals.

Propan-1-ol (specified laboratory reagent) was a product of Fisons (Loughborough, U.K.). Hen egg white lysozyme (L-6876, lot:53H7145), horse heart myoglobin (M-1882, lot:106H7035), subtilisin Carlsberg (P-5380, lot:24H0868), and N-acetyl-l-phenylalanine ethyl ester were Sigma products. The 4A molecular sieve (product 540054S), n-hexane, formic acid (98+%), ammonia solution (35%), (NH₄)₂HPO₄, NaH₂PO₄⋅2H₂O, Na₂HPO₄⋅12H₂O, and KCl (all AnalaR grade) were BDH/Merck products. NaOOCH (American Chemical Society grade 99+%) and NH₄OOCH (97%) were from Aldrich. The water used throughout this study was from a Milli-U10 (Millipore) apparatus.

Drying Agent Preparation.

Molecular sieves were heated to 350°C for at least 1 h and were cooled before use as drying agents for solvents and biocatalysts.

Lyophilization.

After flash-freezing in liquid N₂, the samples were lyophilized on a Modulyo freeze drier (Edwards, Crawley, United Kingdom) overnight.

Pre-Preparation of Stock Lyophilized Proteins.

All proteins used in this study were first dialyzed against NH₄OOCH to exchange out the bulk of residual nonvolatile counterions. To this end, myoglobin, lysozyme, and subtilisin were dissolved in 2.0 mM NH₄OOCH buffer of pH 9.5, 9.5, and 8.4 (at 4°C) and to concentrations of 5.0, 5.0, and 1.0 mg/ml, respectively. These then were dialyzed against 20 volumes of the same solution for 4 h at 4°C, were lyophilized, and then were stored over molecular sieve.

Preparation of Lyophilized Subtilisin for Activity Tests.

Stock lyophilized subtilisin Carlsberg was dissolved at a concentration of 1.0 mg/ml in a 2.0 mM solution of each salt tested: Na₂HPO₄; NaOOCH; NH₄OOCH; and (NH₄)₂HPO₄. The pH was adjusted with diluted H₃PO₄, HCOOH, or NH₃, as appropriate; then, the solution was immediately frozen in liquid nitrogen and was freeze-dried.

Preparation of Immobilized Subtilisin.

Preparation of subtilisin Carlsberg covalently immobilized to PolyHipe was accomplished as described (10). The immobilized preparation was washed with several volumes of various 20 mM buffers of known pH at room temperature. The preparations then were recovered by filtration and were dried under vacuum over 4A molecular sieves for 3 days. The preparations then were removed and stored over 4A molecular sieves until required for further processing.

Enzymatic Reactions.

An organic phase consisting of n-hexane with 1.0 M 1-propanol, 20 mM N-acetyl-l-phenylalanine ethyl ester, and 3.0 mg/ml tetracosane internal standard was preequilibrated to the appropriate water activity (a_w) at room temperature overnight. The enzyme preparations were equilibrated to the same a_w for at least 3 days. Reactions were carried out in 23- × 58-mm (14 ml) glass vials at 23°C with shaking at 500 oscillations/minute. Samples were analyzed by GLC. Initial rates were determined by monitoring the increase in Ac-Phe-OPr and were normalized to the total weight of enzyme preparation. All rates were measured for at least duplicate reactions (reproducible to within 5%). The principal effects of buffer type also have been demonstrated with a different batch of immobilized enzyme.

Reactions with the immobilized preparations were carried out at a_w 0.84. At high a_w values, there is a tendency for lyophilized enzyme powders to form a highly hydrated mass that does not suspend freely in the organic phase. Because this results in a sharp drop in catalytic rate, we decided to use a lower a_w value (0.75) for the measurement of catalytic activity of lyophilized powders.

RESULTS

We have examined a set of buffers composed of both volatile and nonvolatile components over the range of pH values relevant to catalytic activity with subtilisin Carlsberg. Both covalently immobilized and lyophilized subtilisin preparations were examined. By using an immobilized enzyme preparation, it is possible to reduce the effects of the type and quantity of buffer remaining in lyophilized enzyme preparations after drying, which can significantly affect enzymatic properties (7, 11, 12).

Catalytic Activity of Covalently Immobilized Subtilisin.

When the immobilized subtilisin is dried from Na-phosphate buffer, there is a strong pH memory effect (13). In water, subtilisin Carlsberg is known to display maximal activity at pH values between 8.0 and 8.5. The highest activity in hexane also was observed at pH 8.5 (13). Catalyst dried from lower pH values was less active, in accordance with the known pH profile of the enzyme in aqueous solution and the findings of several studies using lyophilized subtilisin in organic solvents (4, 8, 9).

When buffers containing a volatile component were used, however, the pH memory effect was either absent or much reduced (Table 1). With NH₄OOCH buffers, in which both ions are derived from volatile molecules, there was no detectable effect across the pH range examined. With only one volatile component, there was a small but significant remaining effect of previous aqueous pH. The two buffers of this type gave very different rates, being uniformly high for NaOOCH and uniformly low for NH₄-phosphate. Experiments also were performed with other volatile or extractable buffer components, making comparisons after drying from a fixed pH of 7.8. NH₄-acetate and triethylammmonium acetate gave preparations of similar activity to that from NH₄OOCH whereas sodium acetate was similar to NaOOCH (data not shown).

Table 1.

The effect of buffering compounds and pH on the catalytic activity of covalently immobilized subtilisin Carlsberg suspended in n-hexane

Buffer type	Relative initial rate, %
	pH 5.0	pH 7.0	pH 8.5
Na-phosphate	51*	84*	100
NH4-phosphate	13.6	16.1	18.2
NaOOCH	82	112	121
NH4OOCH	27	28	28

The immobilized enzyme was washed with 20 mM buffer of the type and pH shown and then was dried. The solvent mixture and biocatalyst were equilibrated separately to a_w 0.84. The reaction was carried out at 23°C with shaking at 500 oscillations per minute. The values have been normalized to the sample prepared from Na-phosphate, pH 8.5, 0.33 nmol⋅min⁻¹ (milligrams of biocatalyst)⁻¹. The immobilized enzyme used had an aqueous activity (on either Ac-Tyr-OEt or N-succinyl-Ala-Ala-Pro-Phe-p-NA) equivalent to 0.89 mg of subtilisin per gram of biocatalyst. 

^*Taken from ref. 13. 

Lyophilized Subtilisin Preparations.

We sought to confirm the generality of these findings by studying lyophilized subtilisin preparations. Because buffer type and concentration have been shown to significantly influence rates with lyophilized powders (7, 11, 12, 14), we used a lower buffer concentration for their preparation (based on 2.0 mM salts) than for the immobilized enzyme (20.0 mM).

The initial rate data for these lyophilized preparations (Table 2) revealed effects analogous to those seen with the immobilized subtilisin preparation. pH memory clearly is abolished when the enzyme is prepared from NH₄OOCH. Some reduced pH memory remains with NaOOCH, and rather more remains with NH₄-phosphate, though this gives very slow rates. The NaOOCH buffers again give rates similar to that with Na-phosphate of the optimal pH, and the rates are much lower with NH₄OOCH and lower still with NH₄-phosphate. The relative rates with these NH₄ salts are, however, significantly lower for the lyophilized enzyme than for the immobilized preparation. One reason for such differences may be the action of residual salt on the activity of most of the lyophilized powders, either as a “support” or in other ways (7, 12).

Table 2.

Effect of buffer type and pH on the catalytic activity of lyophilized subtilisin Carlsberg samples

Buffer type	Relative initial rate, %
	pH 5.0	pH 8.5
Na-phosphate	12*	100
NaOOCH	101	117
NH4-phosphate	1.0	3.3
NH4OOCH	8.4	8.7

The enzyme was lyophilized from buffer of the type and pH shown. The solvent mixture and enzyme powder were equilibrated separately to a_w 0.75. The reaction was carried out at 23°C with shaking at 500 oscillations per minute. The rates were expressed in terms of the content of subtilisin in each preparation (correcting for the different amounts of residual salt after drying the enzyme from different buffers) and then were normalized to the sample prepared from pH 8.5 Na-phosphate buffer, 10.7 nmol⋅min⁻¹ (milligrams of salt-free subtilisin)⁻¹. 

^*Determined by Yang et al. (9) under similar conditions. 

On preparation of lyophilized powders from these buffers, differences were observed in relative masses of the preparations after drying (Table 3). These reflected the extent to which buffer components remained in the dried powders. With the enzyme prepared from NH₄OOCH, no additional mass other than that contributed by the protein was observed, confirming that virtually all buffer ions had been removed. With all other preparations, however, extra mass was found, as expected when part of the buffer cannot be removed through the gas phase. In such cases, most of the volatile buffer component must also remain in the solid powder because electroneutrality requires it to balance most of the nonremovable ion. The excess of the latter cannot be greater than the opposite sign net charge on the protein.

Table 3.

Percentage excess mass over that contributed by protein in lyophilized and further dried subtilisin samples after drying

Buffer type	pH 5.0	pH 8.5
Na-phosphate	n.d.	36%
NaOOCH	7%	12%
NH4-phosphate	39%	17%
NH4OOCH	−1%	1%

Stock lyophilized subtilisin (9.2–9.7 mg) (known precisely in each case and the mass equal to 100%) was weighed and redissolved to 1.0 mg/ml in 2 mM buffer. After pH adjustment, the samples were lyophilized and then were dried further over molecular sieve before weighing. n.d., not determined. 

With Na-phosphate at pH 8.5, the buffer ions would be expected to contribute an extra mass of ≈31% (mainly as solid Na₂HPO₄). The NaOOCH buffer of pH 8.5 should contribute essentially a stoichiometric amount of NaOOCH, with a mass increase of 13.6%. In the case of NH₄-phosphate, we might expect 26.4% extra mass at pH 8.5 [as (NH₄)₂HPO₄] and 46% at pH 5.0, after adding extra H₃PO₄ to get NH₄H₂PO₄. These are all in reasonable agreement with the data.

Measured pH Values of Redissolved Proteins.

To further investigate the loss of pH memory in proteins dried from buffers with volatile components, we redissolved them in water and measured the resulting pH. Table 4 shows that subtilisin again loses most of its pH memory with such buffers. The pH values are relatively high for enzyme dried from NaOOCH and relatively low for enzyme dried from NH₄-phosphate. With NH₄OOCH, loss of all nonprotein ions as volatiles would be expected to give a pH on redissolution near the pI value for that protein. However, the measured values are significantly below the pI of subtilisin Carlsberg (9.4). To clarify the effect, we also studied two other proteins of widely differing pI, horse heart myoglobin (pI = 6.8) and hen egg white lysozyme (pI = 11.1). After lyophilization from 20 mM NH₄OOCH, pH values on redissolution were found to be 3.9–4.1 and 4.2–4.4, respectively. In both cases, there was no trend with prelyophilization pH for the values tested (2, 4, 6, 8, or 10).

Table 4.

pH values measured after redissolution of lyophilized subtilisin in pure water

	Original pH 5.0	Original pH 8.5
Na-phosphate	5.6	8.4
NaOOCH	7.1	8.0
NH4-phosphate	3.9	3.9
NH4OOCH	4.8	4.5

Subtilisin was lyophilized from buffers based on 2 mM solutions of the salts shown. Samples were redissolved at 1 mg/ml in purified water, which had a measured pH of 5.35. 

DISCUSSION

Because their structure and mechanisms are well described and their reactions are conveniently followed, serine proteases have become model enzymes in nonaqueous enzymology. Primarily, the catalytic activity of subtilisins depends on deprotonation of the catalytic triad. Thus, subtilisins are inactive in water up to pH 5.5 and are fully active around pH 8.0–8.5 (15). Similarly, they show a strong pH memory effect when assayed in organic media, with highest activity after drying from aqueous pH values of 8.0 or higher (8, 9). To understand when pH memory may be reduced or eliminated, it is useful to consider how the ionization state of a protein molecule may change during drying. We explain here the possible mechanisms, and the restrictions on them.

When and How a Dried Protein can Lose its pH Memory.

Let us start with the protein exposed to aqueous solution, at the point when all further water removal will occur via the vapor phase. In the case of lyophilization, this will be in the solution just before freezing (or in the residual unfrozen liquid before drying). In the case of a solid-adsorbed protein to be air dried, this will be when water removal by drainage is complete.

The initial ionization state of the protein is clearly set by the pH of this aqueous solution. Except at the isoelectric point, the protein molecule will have a net charge. This must be compensated by an imbalance in the numbers of positive and negative counterions in the surrounding solution. In the simplest case (e.g., with Na⁺ and Cl⁻), these ions cannot be changed or removed during the drying process. Hence, the imbalance will persist into the dried state. The net charge of the counterions must continue to be balanced by a net charge on the protein molecule, thus restricting changes in its ionization state. Therefore, the pH memory can be seen to reside in the unequal numbers of counterion charges, as much as in the state of the protein. This picture also can be applied to the weak pH memory in enzyme catalysts (e.g., cross-linked crystals) dried by organic solvent rinsing (16). The rinsing process will remove counterions from the preparation, allowing a change in net protein charge.

Fig. 1 shows various possibilities that subsequently can change the ionization state of the protein. Removal of water and consequent reduction in dielectric will tend to enforce ion-pairing. Exchanges of H⁺ may occur internally between protein groups (17), without change in net charge. Similar exchanges can occur with weakly acidic or basic counterions, derived from buffer species such as TrisH⁺ or HPO₄²⁻. These lead to equal and opposite changes in the net charge of the protein and the counterions. All of these exchanges will be reversible on rehydration but may well affect protein function (e.g., catalytic activity) in the dry state.

If the weak acid or base formed from the counterion is volatile, then a complete loss of pH memory is possible. Once the neutral weak acid or base has been lost, the process becomes impossible to reverse (unless fresh acid or base is supplied). Thus, the net charge of the protein molecule, even after rehydration, will be changed. If the readily removed counterions are of only one sign, then the protein will tend to gain a net charge opposite to that on the counterions that remain. If all counterions can be removed, the final state should be a protein with zero net charge. Hence, on rehydration, the protein should give the isoelectric pH.

Finally, we should note that, if dried proteins are placed in organic or similar media, further protonation changes may result from exchanges with acids or bases in the surrounding fluid phase. The agents involved must be able to exchange counterions as well as H⁺, in line with the picture above. This applies equally to buffering systems added deliberately to override pH memory and/or to control acid–base conditions (10, 13, 16, 18).

Fully Volatile Buffers: Ammonium Formate.

With the NH₄OOCH buffers used in this study, both ions should be removable, by processes 1 and 2 shown in Fig. 2. Together, these should result in the same form of the protein, whatever the initial aqueous pH. The pH measured on redissolution of powders dried from NH₄OOCH buffers indicates that the proteins retain a net positive charge, which must be balanced by some residual HCOO⁻ anions. This is consistent with the relatively low catalytic activity in hexane, as expected for subtilisin in such an ionization state. We can suggest two reasons why HCOO⁻ is removed less completely than NH₄⁺. Most obviously, NH₃ is much more volatile than formic acid. However, the added organic phase should fairly readily extract formic acid. A second factor concerns relative acidities and basicities. Taking aqueous pKs as a rough guide to these, we can compare the ease of the proton transfer steps that are required before the most strongly bound counterions are removed. Transfer of H⁺ from a protein guanidinium to a HCOO⁻ counterion is opposed by an aqueous ΔpK of some 8 units. In contrast, transfer from an NH₄⁺ counterion to a protein carboxylate needs to overcome an aqueous ΔpK of only ≈4 or 5 units.

To our knowledge, there is only a single report of use of volatile buffers in preparing enzymes for nonaqueous reactions (14). The maximum activity of bilirubin oxidase in water-rich cetyltrimethylammonium bromide water-in-oil microemulsions occurs at pH 8.4. When the enzyme was lyophilized from NH₄-acetate pH 8.4 and was suspended in a chloroform-n-heptane 50:50 (vol/vol) medium, it was inactive, however. We now can provide a reasonable explanation for the observed complete loss of activity, assuming the NH₄-acetate buffer behaved in a similar way to NH₄OOCH. The pH “remembered” by the dried enzyme preparation then would be near to 4.0, which is far from the optimum for the enzyme. (When the enzyme was lyophilized from citrate and 2-[N-morpholino]ethanesulfonate (Mes) buffers of pH 4.5 and 6.0, respectively, the enzyme was inactive in the same reaction system.)

Buffers with One Removable Component.

When the buffer used contains only one volatile component (e.g., NH₄-phosphate or NaOOCH), then the appropriate counterions can be lost from the protein, as discussed above. It might be thought that protein groups associated with the nonremovable counterion could not change their protonation state. However, there normally will be some excess of the buffer salt in the initial dried preparation, so that reactions 3 and 4 shown in Fig. 2 are possible. Such processes will tend to bring all of the protein groups to the same final state, independent of the previous aqueous pH. Hence, we see the much reduced pH memory when using this type of buffers. However, the state reached will be very different, depending on which of the buffer ions is removable. Table 5 illustrates the possibilities. Removal of formic acid will tend to leave a relatively unprotonated form of the protein, whatever the initial aqueous pH. This can be seen as attributable to removal of H⁺ from the protein by HCOO⁻ before loss of neutral formic acid. The equivalent picture is that HCOOH loss leaves an excess of positive counterions (e.g., Na⁺), which must be balanced by a net negative charge on the protein. The resulting deprotonated subtilisin has a high catalytic activity, as when it is dried from Na-phosphate of the optimal pH. The slightly higher activity observed from NaOOCH (Tables 1 and 2) may reflect some adverse effect of the residual phosphate ions when dried from this buffer. The analogous processes on drying from NH₄-phosphate (Table 5) lead to a relatively protonated form of the protein, with balancing phosphate ions. This shows low activity, as expected for subtilisin. The rather higher rate after drying from NH₄OOCH (Tables 1 and 2) presumably reflects the much lower quantities of HCOO⁻ remaining with the protein because much has been removed. The measured pH values on redissolution in water (Table 4) confirm what happens with these buffers having one volatile component.

Table 5.

Changes in protein ionization state during drying from various buffer types

State of protein groups		Additional process involved, see Fig. 2
After initial drying	After further drying
Na-phosphate buffer
−NH3+H2PO4−	−NH3+H2PO4−
−NH2	−NH2
−COO−Na+	−COO−Na+
−COOH	−COOH
NaOOCH buffer
−NH3+HCOO−	−NH2	HCOOH loss (2)
−NH2	−NH2
−COO−Na+	−COO−Na+
−COOH	−COO−Na+	Exchange with HCOO−Na+ (3)
NH4-phosphate buffer
−NH3+H2PO4−	−NH3+H2PO4−
−NH2	−NH3+H2PO4−	Exchange with NH4+H2PO4− (4)
−COO−NH4+	−COOH	NH3 loss (1)
−COOH	−COOH

The possible processes should be general for all ionizable protein groups and all types of buffer ions. They are illustrated by using examples to represent any protein basic group (−NH₂); any protein acidic group (−COOH); any removable buffer ions (NH₄⁺ and HCOO⁻); and any nonremoveable buffer ions (Na⁺ and H₂PO₄⁻). The numbers of the additional processes refer to equation numbers in Fig. 2. 

Charge Compensation in the Subtilisin Active Site.

The catalytic triad in the active state has a negative charge, formally placed on the Asp-32 carboxyl, which must be balanced by some positive charge. In water, this can be far away, but, in low dielectric media like hexane, close approach of the compensating charges will be strongly favored. It is instructive to consider their possible locations by using the known three-dimensional structure of subtilisin Carlsberg in acetonitrile (21). There are no obvious candidate positive groups on the protein surface surrounding the active site. The nearest is His-67, whose nitrogens are between 0.7 and 1.1 nm (7–11 Å) from the electronegative atoms of the triad. However, the His-67 imidazole is rather buried and seems unlikely to be protonated under the relatively basic conditions at which the enzyme is active. The next nearest is Lys-94, whose epsilon-N is 0.97 nm from one of the Asp-32 oxygens.

Another possibility in low-water proteins is that helix dipoles (or other particularly polarized or polarizable parts of the protein) may serve to stabilize the placing of the counterion at greater separation. The subtilisin structure has two helices with their positive (N-terminal) ends near the active site. The helix beginning at residue 63 has positive charge centers (amide N and carbonyl C) <0.4 nm from one Asp-32 oxygen. The helix beginning at Thr 220 has similar distances to the active site Ser 221 hydroxyl oxygen. Both of these helices cross the core of the protein, so a reasonably favorable location for the countercation to the active site might be near the far ends of these helices. The C terminus of the first helix (at residues 72–74) is somewhat buried, but the negative charge centers on the carbonyl oxygens are fairly close (0.5–1.0 nm) to the Ca²⁺ and the epsilon N of Lys-22. The other helix terminates at residues 236–238, with possible positive charges on the side chains of His 238 (≈0.5 nm), Lys 237 (≈0.8 nm), and Arg 249 (≈0.9 nm), though the latter two are also quite close to the potential negative charge of the protein C terminus.

If none of these possibilities provides charge balance, a positive counterion must be bound close to the active site itself. This obviously carries the risk of interference with the catalytic process, either by steric or electrostatic effects. This might be another factor reducing the catalytic activity of subtilisin in low-dielectric media, to add to the list already compiled (1).

Interference by a required counterion also may explain two other acid–base effects on the catalytic activity of subtilisin under low-water conditions. Appreciable relative activity after drying from pH 5 or so has been reported in several studies (e.g., refs. 8 and 9), even though subtilisin is essentially inactive in aqueous solution at such low pH. The electrically neutral active site at acid pH will have a much lower intrinsic catalytic power, but it needs no counterion, so there will be no interference from this source. The change in the ratio of activity at low and high pH results from a greater reduction at the optimal pH because of interference by the counterion. A second effect is the higher activity of subtilisin pretreated so as to remove most counterions (22). The immobilized enzyme was washed in dilute HCl, which should remove most cations, and then was exposed to a basic organic-soluble buffer, which would remove Cl⁻ together with H⁺. The result may be a form of the enzyme with very few counterions available, so only intramolecular charge neutralization is possible. With no counterion interfering with the active site, the activity is enhanced.

CONCLUSION

The phenomenon of pH memory in dried proteins can be seen equally to result from the net charge of the buffer ions present in the powder. When buffer ions are derived from weak acids or bases that can be removed during drying, pH memory is reduced or absent. The protonation state of the dried protein (and, hence, its catalytic activity) will depend instead on the extent of removal of the buffer ions of different signs. These processes will affect the ionization state of any protein, but what is optimal for enzyme activity will be case-specific. Nevertheless, drying from a suitable, partly volatile buffer may often be a convenient method to prepare an enzyme in near-optimal form, as with subtilisin from sodium formate.

Figure 1.

A. Ion-pairing on reduction in dielectricProtein-NH₃⁺ + X⁻ == Protein-NH₃⁺X⁻ Protein-COO⁻ + M⁺ == Protein-COO⁻M⁺ B. Internal proton transfer, always possible Protein-COO⁻ + Protein-NH₃⁺ == Protein-COOH + Protein-NH₂ C. If counter-ion is from weak acid or base Protein-NH₃⁺⋅A⁻ == Protein-NH₂⋅HA Protein-COO⁻⋅HB⁺ == Protein-COOH⋅B D.  If weak acid or base is volatile Protein-NH₂⋅HA → Protein-NH₂ + HA ↑ Protein-COOH⋅B → Protein-COOH + B ↑  Fig. 1. Processes that can change the ionization state of low-water proteins. −NH₂ and −COOH are used to represent any basic and acidic group in the protein. An equivalent process to D may occur if the dried protein is suspended in an organic medium and HA or B is extracted into the organic phase. Note that the proton exchanges represented as B and C may consist of just a very small movement of an H nucleus. This applies where the charged groups exist as an ion pair and the neutral forms exist as a hydrogen bonded complex. These differ only in the exact location of an H nucleus, slightly closer to the base or the acid.

Figure 2.

Loss of volatile counter-ions Protein-COO⁻NH₄⁺ → Protein-COOH + NH₃ (g) (1)  Protein-NH₃⁺HCOO⁻ → Protein-NH₂ + HOOCH (g) (2) Exchanges with partly volatile salt  Protein-COOH + Na⁺HCOO⁻ → Protein-COO⁻Na⁺  + HCOOH (g) (3)  Protein-NH₂ + NH₄⁺H₂PO₄⁻ →  Protein-NH₃⁺H₂PO₄⁻  + NH₃ (g) (4)  Fig. 2. Reactions affecting protein ionization state in our experiments. NH₃ boils at −34°C whereas HCOOH has a vapor pressure of 10 mmHg (1 mmHg = 133 Pa) at 2°C (19). Their salt NH₄OOCH is readily sublimed under vacuum (20).

ABBREVIATION

a_w

appropriate water activity