Maintenance of {Delta}pH by a butanol-tolerant mutant of Clostridium beijerinckii

In the obligately anaerobic clostridia glycolytically generated ATP energizes the extrusion of H⁺ by the H⁺-translocating ATPase (Hasan & Rosen, 1979; Clarke et al., 1979), establishing and maintaining a transmembrane electrochemical gradient of protons. This gradient, which exerts the protonmotive force (Δp), consists of a transmembrane electrical potential difference, Δψ, and a chemical gradient of protons, ΔpH, equal to the difference in pH between the interior of the cell and the extracellular medium (Mitchell, 1961, 1963, 1966; see reviews by Harold, 1972, 1986; see also Nicholls & Ferguson, 2002). In the solvent-producing species Clostridium beijerinckii and Clostridium acetobutylicum, the relative contributions of Δψ and ΔpH to the Δp vary depending on the pH of the external medium (Terracciano & Kashket, 1986). Thus, C. acetobutylicum cells incubated in buffer at pH 5 maintained a ΔpH of approximately 1·7, which was dissipated by the addition of butanol. In contrast, the Δψ was low and relatively unaffected by butanol.

The main mechanism by which butanol exerts its toxic effects on C. beijerinckii, C. acetobutylicum and other solvent-producing clostridia has generally been taken to be its chaotropic effect on the integrity of the cell membrane. Butanol addition leads to leakage of ions and other solutes from clostridia (Hutkins & Kashket, 1986); with excessive ion leakage the cell cannot maintain its transmembrane ion gradients by ion pumping (Terracciano & Kashket, 1986). However, butanol toxicity is not necessarily a limiting factor in solvent production by these organisms. Mutants that produce high concentrations of solvent (>13 g butanol l¹, levels that previously were thought to be limiting) have been isolated without direct selection for butanol tolerance. The mutants include the hyperamylolytic C. beijerinckii NCIMB 8052 mutant, BA101 (Formanek et al., 1997) and a mutant of C. acetobutylicum ATCC 824 with an inactivated solR gene (Harris et al., 2001). The existence of these strains suggests that butanol at the concentrations that accumulate in cultures may not massively damage the cell membrane, and raises the possibility that the most sensitive target of butanol toxicity is not the phospholipid bilayer per se, but one or several specific cellular functions. For example, a target of butanol appears to be a function(s) involved in septation/cell separation, as suggested by the formation of filamentous forms when low concentrations of butanol were added to growing C. beijerinckii cells (E. R. Kashket, unpublished data).

The isolation of C. beijerinckii mutants that are more tolerant of butanol than the wild-type offered the opportunity to investigate whether specific functional components of the cell are especially sensitive targets of butanol toxicity. In the present studies we focused on the activities of the cell membrane which are required for maintaining transmembrane ion gradients. We report that a butanol-tolerant mutant, strain BR54 (Liyanage et al., 2000), undergoes less dissipation of the H⁺ gradient than the wild-type at a given concentration of butanol. We also provide evidence that the dissipation of ΔpH by Na⁺/H⁺ antiporter activity can be specifically blocked by low concentrations of butanol under conditions of limited cellular Mg²⁺ concentration. Finally, observations of differences in the content of arginyl residues between mutant BR54 and wild-type membrane proteins suggest that the proteins of the former are more extensively glycated by cellular methylglyoxal, which is consistent with the decreased ability of the mutant to detoxify metabolically generated methylglyoxal (Liyanage et al., 2001). The mutant's apparent greater Na⁺/H⁺ antiporter activity may be a result of an increased degree of protein glycation compared to the wild-type.

Growth of bacteria.
Clostridium beijerinckii NCIMB 8052 and the butanol-tolerant mutant, strain BR54, were routinely grown at 37 °C in medium TYG containing 0·5 % (w/v) glucose, as described previously (Liyanage et al., 2000). Where indicated, methylglyoxal (MG) was added to a concentration up to 4 mM to cultures at approximately OD₆₀₀ 0·8 (measured in a Spectronic 21 spectrophotometer, Milton Roy Co., 1 cm path length), and growth was continued for another 12 h. The growth of the cells and the experiments were carried out in an anaerobic chamber (Sheldon Manufacturing).

ΔpH measurements.
The incubations were carried out at pH 5·0, since at this external pH the protonmotive force of the clostridia consists essentially of the ΔpH (Terracciano & Kashket, 1986), and it is therefore not necessary to measure Δψ, the transmembrane electrical potential. The ΔpH was measured by the accumulation of low concentrations of a radioactively labelled, nonmetabolizable organic acid, [¹⁴C]benzoic acid (Baronofsky et al., 1984; Terracciano & Kashket, 1986). Weak organic acids such as benzoic acid or acetic acid distribute themselves across the membrane according to their pK_a' values and the difference in pH (ΔpH) on the two sides (reviewed by Kashket, 1985). The cells were harvested in late exponential phase (OD₆₀₀ 0·9), washed and suspended in the buffer. The buffers used were citric acid (12·5 mM) adjusted to pH 5·0 with either sodium phosphate (NaCP buffer) or potassium phosphate (KCP buffer), and supplemented with 1 % (w/v) glucose. Clostridial Basal Medium (CBM) (O'Brien & Morris, 1971) containing 1 % (w/v) Casamino acids (Difco) was added to the ΔpH assay mixture where indicated. The ΔpH values were determined in triplicate and are reported as the means and standard deviations of at least three cultures per growth condition. P values were calculated by Student's t test.

Measurement of arginine residues in membrane proteins.
Protoplasts were prepared by treating washed cell pellets from 100 ml cultures suspended in 1·0 ml of a mixture containing 2·5 g sucrose l¹, 50 mM Tris buffer, pH 8·0, and 55·5 mM EDTA, pH 8. Lysozyme (Sigma-Aldrich) was added to a final concentration of 2·5 mg ml¹, and the cell suspension was incubated at 35 °C for 30 min, at which time conversion to round forms was found to be complete, as monitored by phase-contrast microscopy. The protoplasts were collected by centrifugation at 4000 r.p.m. (1900 g) for 10 min in Servall centrifuge, SS34 rotor, at 4 °C. The resulting pellet was suspended in 30 ml 10 mM sodium phosphate buffer, pH 7·0, and approximately 1 mg crystalline DNase (Sigma-Aldrich) was added at room temperature. The suspension clarified within a few minutes and became more yellow in colour. The lysed protoplasts were centrifuged at 9500 r.p.m. (10 800 g) for 20 min in a Servall SS34 rotor at 4 °C to remove any remaining intact protoplasts and debris. The supernatant fluid was centrifuged at 30 000 r.p.m. (166 000 g) for 1·5 h in a Servall Ultracentrifuge 4 °C with a Surespin 630 rotor. The resulting membrane pellet was suspended in 10 mM sodium phosphate buffer, pH 7·0, to a final volume of 400 µl. Part of the suspension (10 µl) was assayed for protein by the Folin-phenol (Lowry) method. A 250 µl aliquot of the membrane suspension was acid-hydrolysed under N₂ with an equal volume of concentrated HCl at 105 °C, for 24 h. The hydrolysate was dried at 70 °C under N_2, dissolved in 500 µl water, dried again, and dissolved in 250 µl 100 mM CAPS (Sigma-Aldrich) buffer, pH 9·0. The pH was readjusted to 9 with 1 M NaOH. Aliquots of 1030 µl of the hydrolysates were assayed for ornithine (Chinard, 1952), using CAPS buffer, pH 9·0, with and without treatment with beef liver arginase (EC 3.5.3.1; MP Biomedicals) for 60 min at room temperature, followed by heating at 100 °C for 10 min. The increase in the ornithine content of samples after arginase treatment reflected unmodified arginine residues, since arginine modified by MG at the guanidino group is not subject to hydrolysis by the enzyme (Greenberg, 1960).

The ΔpH of mutant BR54 is more tolerant of butanol than wild-type C. beijerinckii
The working hypothesis, that the ability of the C. beijerinckii cell membrane to maintain transmembrane gradients is a sensitive target of butanol toxicity, was studied by comparing the capacity of wild-type and butanol-tolerant cells to maintain transmembrane proton gradients when challenged with butanol (Fig. 1). Wild-type cells incubated in NaCP buffer (citric acid adjusted to pH 5·0 with sodium phosphate) and supplemented with MgCl₂ and glucose maintained a ΔpH of 1·41±0·12 (mean±SD) (Fig. 1a). The ΔpH of the butanol-tolerant mutant, BR54, was 1·46±0·14. In KCP buffer the ΔpH values of both strains were higher (Fig. 1b): after 20 min the wild-type cells had a ΔpH of 1·64±0·07 and the mutant cells a ΔpH of 1·77±0·09.

(15K): Fig. 1. Effect of butanol on the ΔpH of wild-type C. beijerinckii () and butanol-tolerant mutant BR54 (•). The cells were suspended in NaCP buffer (a) or KCP buffer (b) supplemented with glucose, and 1·5 mM MgCl2; butanol and [14C]benzoate were added at zero time. Samples were taken after 20 min incubation at 23 °C and assayed as described in Methods. The values shown are means±SD (n=46 experiments). P values were calculated for the data indicated by asterisks.

In both buffers the ΔpH decreased when butanol was added to fermenting cells (Fig. 1). In NaCP buffer the ΔpH of the wild-type decreased from 1·41 at 0 % butanol to 0·79 in the presence of 1·5 % (g per 100 ml final concentration) added butanol (Fig. 1a). With 1·75 % butanol the ΔpH was completely dissipated. In contrast, the ΔpH of mutant BR54 decreased from 1·46 at 0 % butanol to 1·25 at 1·5 % butanol. At 1·75 % butanol the ΔpH of BR54 cells was 0·63, and at 2 % butanol it was dissipated completely. Thus, the ΔpH values of the butanol-treated wild-type and mutant cells were significantly different when 1·0 %, 1·25 % and 1·5 % butanol was added (P<0·01).

The differences in butanol tolerance seen in wild-type and mutant cells were more pronounced in KCP buffer than in NaCP buffer (Fig. 1b). The ΔpH of the wild-type was lower with increasing butanol concentrations, and was completely dissipated at 1·5 % butanol. In the mutant the ΔpH decreased more gradually than in the wild-type, from 1·77 at 0 % butanol to 1·37 at 1·5 % butanol, and to 0·27 at 1·75 % butanol. The difference in butanol sensitivity between the strains was significant when 0·51·5 % butanol was added (P<0·001). Thus, the butanol-tolerant mutant, compared to the wild-type, was able to maintain a higher electrochemical gradient of H⁺ ions when challenged with butanol.

Effect of Mg²⁺ deprivation on ΔpH
Wild-type C. beijerinckii and butanol-tolerant mutant BR54 cells incubated in MES buffer, pH 5·2, and supplied with glucose as fermentable energy source, maintained a ΔpH of 1·63±0·02 (n=3) and 1·67±0·13 (n=4), respectively, for at least 20 min. However, when the cells were incubated in citrate/phosphate buffer, either NaCP or KCP, the ΔpH gradually dissipated. Thus, in mutant BR54 cells incubated in KCP buffer supplemented with glucose the ΔpH decreased significantly from 1·75 after 3 min to 1·27 at 30 min, a decrease of 0·48 pH units (Fig. 2). When Mg²⁺ was included in the incubation mixture, however, the ΔpH was maintained over at least 30 min. In the experiment shown in Fig. 2, Mg²⁺ was supplied by substituting 25 % of the KCP buffer with growth medium CBM (equivalent to 0·8 mM Mg²⁺). The ΔpH decreased to 1·69, a decrease of only 0·06 pH units. The components of CBM medium (potassium phosphate, cysteine, p-aminobenzoic acid, biotin, thiamine, MnSO₄, FeSO₄ and MgSO₄), as well as casamino acids, were tested separately for ability to prevent or lessen the decrease in the proton electrochemical gradient in glucose-supplemented citrate/phosphate buffer. Only Mg²⁺ was found to lessen the dissipation of the ΔpH (data not shown). The same effect of Mg ²⁺ deprivation on ΔpH was seen in NaCP buffer and with wild-type cells (data not shown). We attribute the decrease in ΔpH in the absence of added Mg²⁺ to the chelating effects of citrate and phosphate and the resulting depletion of cellular Mg²⁺. Without adequate Mg²⁺, nucleotide-dependent reactions of energy metabolism, such as ATP-powered H⁺ extrusion, are apparently insufficient to replenish the proton electrochemical gradient in these anaerobes.

(9K): Fig. 2. Effect of growth medium on ΔpH of mutant BR54 cells incubated in KCP buffer. Cells were prepared as described in Methods, and suspended either in KCP buffer + 1 % glucose () or in KCP buffer + glucose in which growth medium CBM was substituted for 25 % of the volume of the buffer, as a source of Mg2+ (•). The values shown are means±SD (n=34 experiments); the SD values for the cells with added growth medium were <0·3 and are not plotted.

Effect of low concentrations (0·8 %) of butanol on ΔpH dissipation
In NaCP buffer, in the absence of Mg²⁺, the ΔpH of mutant BR54 decreased steeply, from ΔpH 1·87 at 3 min of incubation to ΔpH 0·19 after 30 min (Fig. 3a) and a less pronounced ΔpH dissipation was seen with the wild-type (Fig. 3c). In KCP buffer the dissipation of ΔpH of BR54 cells was slower than in NaCP buffer: the ΔpH decreased from 1·74 to 1·02 between 3 and 30 min (Fig. 3b). In the wild-type the dissipation was similar in the two buffers, falling from 1·74 to 1·03 between 3 and 30 min (Fig. 3d).

(19K): Fig. 3. Effect of K+ and Na+ and of butanol on ΔpH. Mutant BR54 (a, b) and wild-type C. beijerinckii (c, d) were incubated in NaCP buffer (a, c) or KCP buffer (b, d) supplemented with 1 % glucose, but without added Mg2+. Open symbols, no butanol added; filled symbols, 0·8 % (w/v) butanol added. The ΔpH values shown are means±SD (n=57).

Surprisingly, the addition of 0·8 % butanol inhibited the dissipation of ΔpH of Mg²⁺-deprived cells in Na⁺ buffer (Fig. 3a). With added butanol the ΔpH decreased from 1·49 at 3 min to 1·08 after 30 min, a decrease of 0·41 pH units, compared to a decrease of 1·68 units in the absence of added butanol. Inhibition of ΔpH dissipation by butanol was also seen in wild-type cells (Fig. 3d), although to a lesser degree. In contrast, butanol had no effect on ΔpH dissipation in KCP buffer, either in wild-type or in mutant cells (Fig. 3b, d). However, the addition of 0·8 % butanol to cells in KCP buffer in the absence of added Mg²⁺ lowered the ΔpH by approximately 0·4 pH units, compared to cells without butanol added (Fig. 3b, d). This effect is similar to the ΔpH-dissipating effect of 0·8 % butanol seen in cells incubated in the presence of Mg²⁺ (Fig. 1).

Effect of MG on cell growth and ΔpH
The difference between the wild-type and the butanol-tolerant mutant, BR54, is the degree of expression of the glycerol dehydrogenase gene, gldA, resulting in increased MG levels in the mutant cells (Liyanage et al., 2001). We postulated that the differences in butanol tolerance and rate of ΔpH dissipation in citrate/phosphate buffer seen in the mutant and the wild-type are due to differences in the extent of protein glycation by MG in the two strains. It is important to note that the concentrations of free MG measured in the cultures were in the micromolar range (Liyanage et al., 2001). However, since MG is known to react with many constituents in the medium as well as in the cells, we tested the effects of micromolar (low) and millimolar (high) MG concentrations on ΔpH dissipation. When we added MG at the same time as [¹⁴C]benzoate (zero time) there was no effect on the rate of ΔpH decrease in either NaCP buffer or KCP buffer, in both mutant and wild-type cells (data not shown). When MG was added to high concentrations (millimolar) to growing cells for 12 h before harvest the culture continued to increase in optical density, albeit somewhat more slowly (not shown). The most striking effect on cell growth was the inhibition of cell separation. One hour of exposure of BR54 cultures to 3 mM MG resulted in a significant increase (P<0·0003) in the proportion of chains consisting of three or more cells that had not separated from each other, and a decrease in the proportion of single-cell lengths. Wild-type cells behaved in a similar way (not shown). When MG was added to cultures earlier in exponential phase (OD₆₀₀<0·3), the chains were longer, consisting of four cell-lengths or more, and there were fewer in one or two cell-lengths (not shown).

The addition of MG to growing cultures 1 h before harvesting of both wild-type and mutant BR54 cells inhibited the dissipation of the ΔpH of cells assayed in citrate/phosphate buffer in the absence of added Mg²⁺ (Fig. 4). The MG effect on ΔpH was more pronounced at higher MG concentrations, reaching a maximum at 3 mM added MG. The inhibiting effect of MG was seen both in mutant BR54 cells and in the wild-type cells. Thus, in the absence of added Mg²⁺, MG-treated cells retained their ΔpH, particularly in KCP (Fig. 5). The addition of 0·8 % butanol to MG-treated cells only had the usual small ΔpH-dissipating effect that is seen in Mg⁺-replete cells (see Fig. 1).

(11K): Fig. 4. Effect of growth with MG on the ΔpH of C. beijerinckii. MG was added to the cultures to the indicated concentrations 2 h before harvesting. Cells of the wild-type (white bars) and mutant BR54 (black bars) were assayed for ΔpH after 30 min incubation in NaCP buffer plus glucose, but without added Mg2+. The ΔpH values shown are means±SD (n=3).

(18K): Fig. 5. Effect of K+ and Na+ and of butanol on the ΔpH of MG-treated C. beijerinckii. MG (3 mM) was added to the cultures 1 h before harvesting. Mutant BR54 (a, b) and wild-type C. beijerinckii (c, d) were incubated in NaCP buffer (a, c) or KCP buffer (b, d) supplemented with 1 % glucose, but without added Mg2+. Open symbols, no butanol added; filled symbols, 0·8 % (w/v) butanol added. The ΔpH values shown are means±SD (n=57).

Wild-type membrane proteins contain more arginine residues than mutant BR54 cells or either strain grown with added MG
The hypothesis that the proteins of the butanol-tolerant mutant BR54 are more glycated than those of the wild-type cells was tested by comparing the content of unmodified arginine residues of membrane proteins of the two strains, and in cells grown without and with added MG. Wild-type membrane proteins were found to contain a higher concentration of arginine residues [0·31±0·02 µmol (mg protein)¹, mean±SD (n=8)] than proteins of mutant BR54 membranes [0·27±0·03 µmol (mg protein)¹, mean±SD (n=9)]. This difference is significant (P<0·01). The latter value was similar to wild-type or mutant BR54 membranes of cells grown with added 3 mM MG [0·27±0·02 µmol (mg protein)¹ (n=3) and 0·27±0·01 µmol (mg protein)¹ (n=4), respectively]. The higher arginine content of wild-type membrane proteins is consistent with a lower degree of glycation of arginine residues in membrane proteins compared to cells exposed to increased MG levels, due either to addition of MG to growing wild-type cells or to endogenous MG in mutant BR54 cells. Butanol exerts at least two effects on C. beijerinckii. One is the chaotropic effect of the alcohol on the integrity of the cell membrane phospholipid bilayer. We showed previously that perturbation of the clostridial membrane with high concentrations of butanol leads to leakage of ions, as well as other solutes (Hutkins & Kashket, 1986). With excessive ion leakage the cell cannot maintain its transmembrane ion gradients by ion pumping. The present results showed that dissipation of ΔpH, which reflects the transmembrane electrochemical gradient of H⁺ ions, was complete in K⁺ buffer at high butanol concentrations (1·5 %, w/v, for the wild-type C. beijerinckii and 1·75 % for the butanol-tolerant mutant, strain BR54). At a lower concentration (0·8 %) butanol had a smaller effect on ΔpH, decreasing it by approximately 0·20·4 pH units under most conditions of incubation, a result that is consistent with limited chaotropic damage to the cell membrane.

The second effect of added butanol manifested itself under conditions of Mg²⁺ deficiency. Under such conditions the cells cannot maintain their transmembrane proton electrochemical gradient by Mg²⁺-dependent H⁺-translocating ATPase, leading to a decrease in ΔpH with time, catalysed by Na⁺/H⁺ or K⁺/H⁺ antiporter activity. However, in Na⁺-containing citrate/phosphate buffer the addition of 0·8 % butanol prevented ΔpH dissipation despite cellular Mg²⁺ deficiency. This ΔpH-sparing effect of 0·8 % butanol was not seen in Mg²⁺-deficient cells incubated in K⁺ buffer. In fact, only the ΔpH-dissipating activity of 0·8 % butanol was clearly seen in cells incubated in K⁺ buffer (Fig. 3b, d). The simplest explanation for these observations is that butanol inhibits a Na⁺/H⁺ exchanger that ordinarily catalyses the extrusion of Na⁺ by exchange with H⁺, by analogy to antiporters identified in other bacteria (West & Mitchell, 1974; Padan et al., 2001; Terracciano et al., 1987). Under Mg²⁺-deficient conditions, the ΔpH cannot be maintained by the H⁺-translocating ATPase and the proton gradient is dissipated in Na⁺ buffers. In K⁺ buffers, ΔpH dissipation would occur by exchange of K⁺ and H⁺ ions, and the protein catalysing this reaction is apparently not sensitive to butanol. These findings point to the ability of low concentrations of butanol to bring about a selective effect on a specific metabolic function.

An additional finding was that the dissipation of ΔpH in mutant BR54 was faster than in wild-type cells in Na⁺-containing buffer under conditions of Mg²⁺ deficiency. The single transposon insertion in mutant BR54 results in decreased expression of the gldA gene and a 25 % reduction in glycerol dehydrogenase activity (Liyanage et al., 2000). This essential enzyme is apparently involved in the detoxification of MG; indeed, the mutant is more sensitive to growth inhibition by MG addition than the wild-type and contains higher cellular levels of the toxic compound. The simplest explanation for the apparently higher Na⁺/H⁺ exchange activity in the mutant strain is that the mutant protein is more highly glycated than the antiporter in the wild-type cells. Since proteins, as well as nucleic acids, are glycated by MG (Thornalley, 1996; Kalapos, 1999; Oya et al., 1999; Uchida et al., 1997), one would expect that many cell proteins, in addition to the putative Na⁺/H⁺ antiporter, would be more extensively glycated in the mutant than in the wild-type. The present experiments showed that the concentration of arginine residues unmodified by glycation in membrane fractions was indeed significantly higher in the wild-type strain than in the mutant. The mechanism of butanol tolerance may be an indirect result of the elevated glycation of cell proteins in the mutant strain.

Growth of the cells with high levels of added MG resulted in a decrease of the concentration of unmodified arginine residues in wild-type membranes. Interestingly, the addition of MG at concentrations of 14 mM, which are considerably higher than the micromolar concentrations of unbound MG found in wild-type and mutant cultures, resulted in the inhibition of Na⁺/H⁺ and K⁺/H⁺ exchange activities in Mg²⁺-deficient buffers. Presumably cellular proteins, or other glycation targets, are more extensively glycated at the high concentrations of the added MG than when MG is derived only from cellullar metabolism.

This project was supported by the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service, grant number 2001-35504-10670.