258 PERMEATION MECHANISMS PERMEATION MECHANISMS IN BACTERIAL MEMBRANES BY P. MITCHELL AND J. MOYLE Zoology Department, University of Edinburgh Received 30th January, 1956 Bacteria possess a plasma-membrane which acts as an osmotic barrier between the internal medium of the cell and the external medium for many solutes, but establishes osmotic linkage between the internal and external media for the transport of nutrient and waste between the cell and its environment. The specificity and kinetics of phosphate transport in Staph. aurezis resembles that of enzyme-linked reactions and the temperature characteristics suggest a complex movement of components of the plasma-membrane during the passage of phosphate from one side to the other. The material of the plasma- membrane is a complex lipo-protein, of which there is sufficient to form about one mono- layer of lipid and one of protein.The protein component includes the cytochrome system and a number of enzymes. It is suggested that some of these enzymes may themselves be the carriers of the substrates which are found to pass through the membrane during metabolism and in some cases, where exchange diffusion occurs, also during rest. The plasma-membrane of living organisms performs a dual function : it causes the separation of the internal medium of the cell from the external medium with respect to many solutes, but for other solutes (notably the nutrients and end- products of metabolism) it allows osmotic connection or acts as a specific osmotic link between the internal and external media. The two questions which we would like to answer are (i) what physical and chemical properties of the plasma-membrane determine its effectiveness as an osmotic barrier, and (ii) what is the molecular mechanism of osmotic linkage through the membrane?P .MITCHELL AND J. MOYLE 259 1. PLASMA MEMBRANE AS AN OSMOTIC BARRIER At the end of the last century, the botanist Alfred Fischer 1 carried out a care- fully controlled series of studies on the penetration of salts and polyhydric alcohols into bacteria. He observed that the protoplasm of many bacteria was caused to retract from the rigid outer cell-wall by salt and sugar solutions of high osmotic pressure but not by glycerol, urea or chloral hydrate solutions of the same osmotic pressure, and obtained some evidence that the protoplasm did not possess intrinsic rigidity, He therefore suggested that although bacteria are so small- having a volume of the order of lp3-the protoplasm is nevertheless covered by a delicate plasma-membrane which is an effective barrier to the free diffusion of salt and sugar molecules between the outer medium and the interior of the cell.Fischer's view met with considerable opposition 2 because the very nutrients which, according to his plasmolysis experiments, could not pass through the plasma-membrane into the cell, were rapidly metabolized under appropriate conditions. However, evidence in favour of the existence of a plasma-membrane of low permeability in bacteria gradually accumulated. Some of this evidence has been reviewed by Knaysi,3 Mitchell,4 Weibull5 and Mitchell and Moyle.6 Six main methods have been employed to measure the passive permeability of bacterial plasma-membranes : (i) Microscopic observation of the plasmolysis and rate of deplasmolysis of individual cells suspended in hypertonic solution.(ii) Macroscopic measurement, by light scattering, of the dependence of the shrinkage and swelling of the cells on solute concentration and time. (iii) Microscopic measurement of the change of volume of individual cells suspended at different solute concentrations. (iv) Macroscopic measurement, by light scattering, of the rate of lysis of sus- pensions of naked bacterial protoplasts in initially isotonic solution. (v) Measurement of the net rate of efflux of a solute by serial chemical analysis of the " internal medium " obtained by subsequent treatment of the cells with trichloroacetic acid, organic solvents, detergents or heat.(vi) Measurement of the net rate of influx of a solute by serial chemical analysis of the suspension medium and/or " internal medium " in suspensions adjusted so that the total volumes of water on either side of the plasma- membrane are approximately equal. NON-ELECTROLYTES. Methods (i) and (ii) supplemented by method (iii) have been applied to Bacterium coli (American strain B).7 In suspension media buffered at neutral pH with 0.02 M (Na2HP04 + NaH2P04) at 20" C, the apparent permeability of the plasma-membrane to the following non-electrolytes is too low to be measured : sucrose, lactose, D-glucose, D-fructose, D-rnannose, D-galactose, D-sorbose, D-sorbitol, L-rhamnose, L-arabinose, D-xylose.But for some other non-electrolytes the approximate times for half equilibration across the membrane are as follows : erythritol and pentaerythritol 15 min, D-ribose 5 min, glycerol less than 1 min. The rates of penetration of ribose and the polyhydric alcohols are not materially affected when carbohydrate metabolism is inhibited by mM mercuric chloride, mM sodium iodoacetate or mM potassium cyanide. In the presence of 7 mM K+, however, the metabolism of D-glucose, D-galactose, D-niannose or D- ribose causes a limited rise in the internal osmotic pressure of the cells, which can be abolished by mM mercuric chloride, mM sodium iodoacetate, mM potassium cyanide or 10-1 mM sodium dinitrophenate.Methods (iv), (v) and (vi) have been applied to Staphylococcus aureus (strain Duncan),*. 99 10 Micrococcus Iysodeikticus (NCTC 2665) and Sarciizn Iutea (labora- tory strain),ll and methods (iii) and (vi) to Bacillus megateriunz (strain KM).12, 13 In Sfaph. auyeus, M. Zysodeikticus, and S. Zutea the permeability of the plasma- membrane in media buffered at neutral pH with 0-02 M(Na2HP04 + NaH2P04) at 20" C is very low for sucrose, D-glucose, D-fructose, D-mannose, D-galactose and260 PERMEATION MECHANISMS D-sorbose. The permeability of B. megaterium to sucrose is also very low. The half equilibration times for the diffusion of some other non-electrolytes across the plasma-membrane of Staph. aweus, M. lysodeiktus and S. lutea are approximately as follows : D-sorbitol 60 min, L-arabinose 30 min, D-ribose 5 min, erythritol 20 sec, glycerol 3 sec.Urethane equilibrates rapidly across the membrane of B. megaterium. It will be noted that in all the organisms studied, the pentoses, arabinose and xylose, penetrate more slowly than ribose although the only difference between the sugars is in the configuration of the OH groups. Also, while ribose penetrates at about the same rate in all the organisms studied, glycerol, erythritol, arabinose and sorbitol penetrate the plasma-membrane of the cocci more than an order of magnitude faster than they penetrate the membrane of B. coli, the ratio of the surface area to volume of these organisms being about the same (cn. lop-1). A half equilibration time of 7 min corresponds to a permeability coefficient of ca.1 Ajsec. There is no doubt that the hexoses, glucose, mannose and galactose, which can be metabolized by all the organisms studied at a rate of about 1 mole/l. wet cell volume per hour, nevertheless do not approach osmotic equilibrium across the plasma-membrane to a significant extent (when present initially in the external medium at a concentration between 0.1 and 0.5 M) either in normal cells or in cells treated with the metabolic inhibitors, dinitrophenol, iodoacetate, cyanide or mer- curic chloride. The passage of the hexoses through the plasma-membrane (either without chemical change or after chemical transformation on the outer surface of the plasma-membrane) must therefore be linked in some way to carbohydrate metabolism and cannot be due to an independent passive permeability of the plasma- membrane as appears to be the case for D-ribose.We can suggest two distinct types of mechanism whereby linkage between the penetration of the hexoses and their metabolism might occur. (a) The membrane may be impermeable to the hexoses themselves, but permeable to a metabolic product of the hexoses which is formed on the outer surface of the plasma-membrane. (b) The membrane may be permeable to the hexoses through a specific carrier mechanism of the type studied by Rosenberg and Wilbrandt 14 in the red blood corpuscle membrane. If the sugar molecules pass through the membrane only as a specific carrier complex, as the concentration of the sugar on the side of the mem- brane towards which it is diffusing rises across the dissociation constant of the carrier complex, the net rate of diffusion drops rapidly and becomes effectively zero when the carrier is saturated on both sides of the membrane.If thedissociation constant of the carrier complex were in the region of that often observed for enzyme- substrate complexes, namely ca. lO-4M, the entry of sugar might readily be con- trolled by its rate of metabolism within the cell. We should, perhaps, point out that in the permeability experiments described earlier, one measures the osmotic equilibration of solutes at a concentration of 0.1 to 0-5 M across the plasma- membrane. If specific carrier mechanisms exist for the diffusion of solutes across the plasma-membrane, we would expect to be able to observe them only when the concentrations of the solutes used were not much higher than the effective dissocia- tion constants of the carriers.This poses a difficult practical problem. It is, however, an important step to have become aware of it (see 5 2). Methods (i), (ii) and (vi) applied to B. C O Z ~ , ~ and methods (iv), (v) and (vi) applied to Staph. azcreus,9. 10 S. lutea and M. lysodeikticcrs 11 have shown that in media buffered at neutral pH with 0.02 M(Na2HP04 + NaH2P04) at 20" C, the plasma-membranes of these organisms are practically impermeable to the following salts : NaCl, KCl, NH4C1, MgC12, KBr, Na acetate (pH 9), K acetate (pH 9), (NazHP04 + NaHzP04), (K2HP04 + KHzPO~), Na2S04. We have observed that 0.01 M glucose causes a limited rise in the internal osmotic pressure of cells suspended in saline media containing 0.007 M K+.When bacteria have been reported to be permeable to salt solutions on the grounds that deplasmolysis occurs in the course of about an hour, the " permeability " was ELECTROLYTES.P . MITCHELL AND J . MOYLE 26 1 probably due to the presence of low concentrations of Kf and glucose or other metabolites. B. coli is practically impermeable to NaCNS, KCNS and NH4CNS, but Staph. aureus is permeable to these salts. For KCNS at pH 6.8, pH 6-1 and pH 5-0 the times for half equilibration across the plasma-membrane are respectively about 10 min, 5 min and 1 min. The dependence of the rate of penetration of NaCNS on pH is similar to that of KCNS, the rate scale being reduced by a factor of 0-8. These observations show that the plasma-membrane of Staph.aureus is permeable to the cations Na+ and K + in the presence of CNS- . The pH dependence of the permeability to the alkali thiocyanates suggests that the anion and cation permeate separately and that the rate of penetration of NaCNS and KCNS is determined by the rate of CNS- penetration through a titratable, charged plasma-membrane. However, although it is certain that the CNS- ion does not cause a general rise in membrane permeability, there being no loss of internal inorganic phosphate under the conditions of our experiments, it is possible that the presence of CNS- may cause an increase in the permeability to cations, or alternatively that the alkali and ammonium thiocyanates may penetrate in the unionized state.Unlike Staph. aureus, B. coli is not significantly permeable to the alkali thio- cyanates. It is also impermeable to NH4C1 at pH 7 or 8 and to Na acetate and K acetate at pH 7 or 6, provided that traces of glucose or other metabolites are absent. On the other hand, NH4 acetate at pH 6 to 8 equilibrates across the mem- brane in a few seconds. The NH4 acetate does not damage the plasma-membrane, for the cells may subsequently be plasmolysed by the addition of NaCl to the NH4- acetate-containing medium. These observations suggest that the plasma-membrane of B. coli is impermeable to the cations K+ and Na+, and to the anions C1- and CNS-. One may perhaps ask why we have attempted to measure the permeability of the membranes of bacteria by observing net solute transport instead of making use of up-to-date tracer techniques.15 The reason for this is that the flux rates measured by tracers may include an exchange diffusion component of unknown magnitude.With phosphate, for example (with which we shall deal in more detail later) the flux rate measured with tracers is high, but the net flux of phosphate or the permeability of the membrane in the old sense is negligibly low. The net flux and the mutual exchange of solute molecules across a membrane may represent two quite different processes between which we wish to distinguish. In order to determine whether the net flux of an ion through a membrane is possible, it is necessary to allow an electric charge to pass across the membrane. For membranes which are accessible from both sides, this may be done as Ussing has shown 27 by making a suitable electrical connection between the media on either side of the membrane.As the medium on the inside of the bacterial plasma- membrane is not accessible to electrode systems, the permeability of the membrane to one ion can only be measured in conjunction with that of another ion. As pointed out above, isotope exchange methods will not yield the information which we require. One therefore searches for a salt to both ions of which the membrane is permeable so that either ion may then be used to investigate the permeability of the membrane to other ions of opposite sign. PARTICIPATION OF A ' LIPID ' PHASE IN THE PLASMA-MEMBRANE. The permeability properties of the plasma-membranes of the bacteria described above are generally in accord with the view that the obstacle to the free diffusion through the plasma- membrane may be a thin hydrophobic or ' lipid ' layer, for the rates of penetration of the solutes tend to decrease rapidly as the hydration is increased. However, there is no doubt from the results described that the configuration of the solutes also plays an important part in determining the rate of penetration, and that factors other than lipid solubility must be concerned in the interaction between solute and plasma-membrane during penetration.CHEMICAL coMPosinoN OF THE PLASMA-MEMBRANE. Determinations of the weight, morphology and chemical composition of fragments of mechanically262 PERMEATION MECHANISMS disintegrated Staph. aureus, segregated into morphologically homogeneous fractions by differential centrifugation, indicated that the plasma-membrane of this organism, which readily disintegrates into small particles, is a complex lipo-protein containing 41 % by weight protein and 22.5 % lipid.17 Semi-quantitative amino-acid analysis of the protein component, by paper chromatography, showed the presence of a high content of the non-polar amino acids, glycine and alanine, and of the acidic amino acid, glutamic acid.The lipid component contained 1.85 % P and 1.3 % N- about half the phosphorus content and rather more than half the nitrogen content of lecithin. The material of the “ small particle fraction ” has been found to account for some 10 to 15 % of the dry weight of the cells.17~ 18 Since the cells are ca.0 . 7 ~ in diameter and have a ratio of wet to dry weight of ca. 3, it can readily be calculated that the material of the “ small particle fraction ” would form a layer ca. 5 rnp thick if unhydrated. This would correspond to about a monolayer of lipid and a monolayer of protein. Work which is at present in progress has shown that the composition of morphologically intact protoplast membranes from Staph. aureus correspond fairly closely to that of the “ small particle fraction ”, and there can be little doubt that the material described above corresponds to that of the plasma-membrane of normal intact cells. 2. THE PLASMA-MEMBRANE AS AN OSMOTIC LlNK It has been suggested that exchange diffusion might represent the trans- location reaction of active transport uncomplicated by the activity of the coupled reactions which normally drive i t ; and that the study of the characteristics of exchange-diffusion might shed light upon the mechanism of active transport.19 We would like to add that when the permeability of the membrane to a particular solute is caused by a specific carrier, although net transport effectively ceases when the solute concentration is such as to saturate the carrier on both sides of the mem- brane, exchange diffusion would be expected to continue at its maximum rate, and to be very strictly coupled.Also, we suggest that if the passage of a solute through the plasma-membrane is facilitated by the mutual occupation of hydro- philic groups of a “ carrier ” component of the membrane and the solute, the move- ment of the “ carrier ” across the membrane may be as dependent upon the presence of the specifically carried solute as the movement of the solute is dependent upon the “carrier”.The study of exchange diffusion reactions may therefore be expected to play an important part in research on the mechanisms of membrane permeability and active transport. TRANSPORT OF PHOSPHATE. Some of the evidence for the participation of an exchange diffusion-like reaction in the exchange of phosphate across the plasma- membrane of resting Staph. aureus has already been reviewed.19 We shall summarize this evidence 8 9 20921 and discuss it critically in the light of recent observations.lo.22 (i) The plasma-membrane of resting Staph. aureus is apparently impermeable to the H2PO; and HPOj- ions when they are present in solution as the alkali salts.This might be due either to impermeability of the membrane to cations or to impermeability to phosphate ions or both. However, the phosphate ions do not exchange across the membrane with acetate, arsenite, azide, bicarbonate, bromide, chloride, chromate, cyanide, fluoride, glutamate, iodide, molybdate, nitrate, nitrite, oxalate, pyroantimonate, succinate, sulphate, thiocyanate, thio- sulphate, p-toluenesulphonate, tungstate or versenate : they exchange strictly with arsenate. The membrane may therefore be impermeable to all the above anions including phosphate, or be specifically permeable to phosphate and arsenate alone and impermeable to the associated cations. While there is a slow permeation of the membrane by NaCl estimated as chloride and a rapid permeation by NaCNS estimated as thiocyanate, there is no significant exchange of phosphate for C1- or CNS-.We may therefore consider two possibilities : (a) 2 he membrane is impermeable to HzPOi and HPOZ- but slightly permeable to C1- and very permeable to CNS-. (6) The strong electrolytes NaCl and NaCNS permeateP . MITCHELL AND J. MOYLE 263 unionized and the membrane may be specifically permeable to H2P0, and/or HPOZ- (but not to the unionized salts), and impermeable to the associated cations. (ii) By labelling the inorganic phosphate of the medium internal or external to the plasma-membrane of Staph. nureus with 32P, a mixing of the phosphate of the media on either side of the membrane is observed. The rate of this mixing varies with the salinity, the pH and the phosphate concentration of the external medium and with temperature.The dependence of the rate of phosphate exchange on pH and external phosphate concentration indicates that the H2POT.ion and not the HPO2- ion takes part in the exchange. The rate of exchange (P) may be described in terms of the external H2PO4- concentration ([HzPO, ] E ) by the equation in which ISmax. stands for the maximum value of 9 and Kis the value of [H2P0& at which P = Pma.J2. When P is expressed in ,u mole phosphate/g cell dry weight min, and [H2P04 ] E is expressed in mM, K has a value of 0.8 & 0.1 mM between pH 5.5 and 8-5 ; and PmaX. has a value of ca. 10 pmole/g min at pH 7 and a positive slope of some 5 pmole/g min pH unit.The hyperbolic form of eqn. (1)-which is formally identical to the enzyme kinetics equation of Michaelis and Menten 23-shows that a saturation phenomenon occurs in phosphate exchange and implies that one stage in the movement of phos- phate through the plasma-membrane involves a specific spatial or bonding relation between a component of the membrane and the phosphate molecule. (iii) The rate of phosphate exchange is very sensitive to certain inhibitors, notably phenyl-Hg+ and other compounds which combine with thiols of low re- activity. The relationship between the degree of inhibition of the exchange reaction and the concentration of phenyl-Hgf may be represented by (2) M being the amount of phenyl-Hg+, n the percentage activity of P and K' a constant. This indicates a reaction of the type M + X + MX, X representing the sites which when combined with inhibitor (as MX), cause inactivation of a corresponding number of units controlling phosphate exchange.The number of these sites can be estimated to correspond to not more than 4.4 p mole phenyl- Hg+/g cell dry weight. (iv) From the dependence of P on temperature it can be calculated that the total heat of activation for the exchange movement of phosphate across the membrane is 37,40Ocal/mole. Using the above estimate of the number of exchange sites, the absolute value of P gave a maximum value for the free energy of activation of 19,700 cal/mole, leaving an entropy of at least 17,700 cal/mole. The thermo- dynamic data suggest that the movement of phosphate across the plasma-membrane is accompanied by a molecular disturbance quite out of proportion to that which would be expected unless the phosphate moves in relation to some larger molecule or molecules within the plasma-membrane.21 The resemblance of the data to those of reversible protein denaturation is perhaps significant. There can be no doubt that whether the plasma-membrane is permeable to phosphate ions (H2POiions) or not, the movement of phosphate groups across it is dependent upon the existence of a highly specific reaction mechanism.There are three simple alternatives for this mechanism : (a) The membrane allows a net transport of H2POQ through the specific reaction mechanism, and the distribution of phosphate ions across the membrane is determined by the distribution of cations, to which the membrane is supposed to be impermeable.The exchange mechanism may, under these conditions, be identical to that proposed by Ussing for Naf exchange.16 (b) The membrane will not allow a net transport of H2P04. Since there is negligible net movement of phosphate across the membrane even when the concentration on the outside is as low as 0.08 mM (an order of magnitude less K' = Mn/(lOO - n),264 PERMEATION MECHANISMS than the apparent dissociation constant K of the carrier implicit in eqn. (l)), the distribution of phosphate ions across the membrane in resting cells must be main- tained by a strict one-to-one carriage of phosphate inwards and outwards by the exchange mechanism. This might be accomplished by the type of carrier considered by Ussing if it is assumed either that the carrier can move only with a phosphate passenger or that the carrier is always occupied with a phosphate group and that this group may exchange with phosphate ions in the media on either side of the membrane.(c) It is possible that no carrier is involved, but that the media on each side of the plasma-membrane are connected by a " pore " at either end of which there is an adsorption site accessible to phosphate ions in the medium on that side only. One of the adsorption sites is assumed to be always occupied by a phosphate group which is supposed to be able to move from one end of the pore to the other and to be able to exchange with a phosphate group in the medium at the appropriate end of the pore, but not to be able to leave the pore unoccupied.Analysis of the kinetics will not distinguish between the alternative mechanisms proposed above. The fact that the membrane is only one or two molecules thick encourages one to consider that the carriage of solutes across it may occur by a thermal rotation of protein, lipid or other component in a manner similar to that visualized by Langmuir 24 in surface films and by Lundegardh 25 in plant cell mem- branes. The formal similarity of the kinetics of the phosphate exchange reaction to enzyme-linked reactions does not, of course, show that enzymes are involved. There are, however, other circumstances which suggest that the phosphate exchange reaction may be coupled to enzyme reactions. When glucose is present there is a net transport of phosphate inwards through the osmotic barrier.This occurs, not as a result of an increase in the rate of influx but as a result of a decrease in the rate of outflux below that of resting cells. It has therefore been suggested that the phosphate exchange reaction of resting cells represents the active transport reaction operating reversibly because it is not being driven by coupling with carbohydrate metabolism. This concept has been supported by the observation that all the in- hibitors of phosphate exchange are also inhibitors of active phosphate uptake. Staph. aureus is not peculiar in possessing the phosphate exchange system, for a similar system has been demonstrated in B. coli. As might have been anticipated, exchange diffusion of phosphate does not occur across the plasma-membranes of the strict aerobes S.Iutea and M. lysodeikticrrs, under semi-anaerobic conditions. ENZYMES OF THE PLASMA-MEMBRANE. The material of the plasma-membrane of Staph. aureus, isolated as the " small particle fraction " described above, contains more than 90 % of the total activity of an acid phosphatase which acts at the outer side of the plasma membrane in intact cells. It also contains at least 90 % of the cytochrome, measured by the total extinction at a wavelength of 425 mp, and some 90 % of the total succinic dehydrogenase activity, as well as potent lactic dehydrogenase activity, the latter showing the characteristics of the cytochrome blinked enzyme.22 The cytochrome spectrum shows the presence of components with extinction maxima at 604, 558 and 528mp.The plasma-membrane of B. megnteriunz has also been reported to contain the cytochrome system.26 These studies show that part of the protein of the plasma-membrane of Staph. aureus is constituted of enzymes, and we are at present investigating the possible participation of these enzymes in the reactions causing phosphate exchange and accumulation across the plasma-membrane. We suggest that the occurrence of the cytochrome blinked enzymes in the plasma-membrane may be connected with the fact that they are the last members of the enzymic chain and deal with the end- products of metabolism, namely lactic, succinic and formic acids which must be specifically carried through the plasma-membrane. We are indebted to the Scottish Hospital Endowments Research Trust for personal grants in support of this work, and to the Rockefeller Foundation for a grant for equipment.P. MITCHELL AND J . MOYLE 265 1 Fischer, Vorlesungen iiber Bakterien (Fischer, Jena, 2nd ed., 1903). 2 Topley and Wilson, The PrincQles of Bacteriology and Immunity (Amold,London, 3 Knaysi, Efements of Bacterial Cytology (Comstock Publ. Co., Ithaca, New York, 4 Mitchell, The Nature of the Bacterial Surface, ed. Miles and Pirie (Blackwell, Oxford, 5 Weibull, Symp. SOC. Gen. Microbioi., 1956, 6, 111. 6 Mitchell and Moyle, Symp. SOC. Gen. Microbiol., 1956, 6, 150. 7 Mitchell and Moyle, in press, 1956. 9 Mitchell and Moyle, in press, 1956. 11 Mitchell and Moyle, J. Gen. Microbiol., 1956, 15, 512. 12 Weibull, Expt. Cell. Research, 1955, 9, 139. 13 Weibull, Expt. Cell. Research, 1955, 9, 294. 14 Rosenberg and Wilbrandt, Expt. Cell. Research, 1955, 9, 49. 15 Roberts, Abelson, Cowie, Bolton and Britten, Studies of Biosynth. in Escherichia coli 17 Mitchell and Moyle, J. Gen. Microbiol., 1951, 5, 981. 18 Mitchell and Moyle, unpublished. 19 Mitchell, Sywp. SOL Expt. Biof., 1954, 8, 254. 20 Mitchell and Moyle, J . Gen. Microbiol., 1953, 9, 257. 21 Mitchell, J. Gen. Microbiol., 1954, 11, 73. 22 Mitchell, J . Gen. Microbiol., 1954, 11, x. 23 Michaelis and Menten, Biochem. Z., 1913, 49, 333. 24 Langmuir, Science, 1935, 87, 493. 25 Lundegardh, Lantbr. Hogsk. Ann., 1930, 8, 233. 26 Weibull, J. Bact., 1953, 66, 696. 1929). 2nd ed., 1951). 1949), chap. 4. 8 Mitchell, J. Gen. Microbiol., 1953, 9, 273. 10 Mitchell and Moyle, in press, 1956. (Carnegie Inst., Washington, 1955). 16 Ussing, Nature, 1947, 160, 262. 27 Ussing, Physiol. Rev., 1949, 29, 127.