microbial water stress806 brown solute value to,t requires an absolute value for le that is not...

BACTIIIOWGICAL Rzvzws, Dec. 1976, p. 803-846 Vol. 40, No. 4Copyright C 1976 American Society for Microbiology Printed in U.S.A.

Microbial Water StressA. D. BROWN

Department of Biology, University of Wollongong, Wollongong, N.S.W. 2500, Australia

INTRODUCTION.............................................................. 803PHYSICOCHEMICAL PARAMETERS .............. ........................... 804SOME ECOLOGICAL ASPECTS OF MICROBIAL WATER RELATIONS ..... ... 807Microbial Water Relations and Food .............. ............................ 807The Saline Environment ..................................................... 807The Nonionic Environment ......... .......... 809

HYPOTHETICAL EXPLANATIONS OF THE TOLERANCE OF LOW WATERACTIVITY ..................... 809

HALOPHILIC BACTERIA..................... 811Cell Envelope and Lipid Biochemistry ........................................ 811Nutrition..................................................................... 813Miscellaneous Characteristics........................................... 813Gas vacuoles .......................................... 813"Satellite" DNA .......... ................................ 813Halophil bacteriophages .................. ........................ 813

Light Reactions of the Red Halophils ......................................... 814Halophilic Characteristics .................. ........................ 815Intracellular Physiology ................ .......................... 818

XEROTOLERANT YEASTS .................. ........................ 821General Biology .......................................... 821Physiology.................................................................. 822

HALOPHILIC ALGAE ............ .............................. 825Intracellular Composition .................. ........................ 826Physiological Role of Glycerol........................................... 827Regulation of Glycerol Production .......................................... 828Physiological Basis of Algal Halophilism...................................... 828Algae Other than D nalie........................................... 829

COMPATIBLE SOLUTES .............. ........................... 829Electrolytes................................................................. 829Nonelectrolytes.............................................................. 833Physicochemical Mechanism of Nonelectrolyte Action ......... ................ 835Regulation of Compatible Solute Accumulation ................................ 837

LOOSE ENDS. 838Survival..................................................................... 840

SUMMARY ............... 841LITERATURE CITED................ 841

INTRODUCTION

A fellow of my acquaintance, on seeing acolleague drink undiluted water (55.5 molal),has been known to comment in disapprovalthat water at such a concentration should notbe used for that purpose and that its mainfunction is for putting around the outside ofboats. He conceded that dilution with a littlesalt is acceptable for boats but for no otherpurpose. The proponent ofthis philosophy is nota biologist and it is unlikely that many biolo-gists would accept his generalization withoutsome qualification. Nevertheless, it is a point ofview.Another point of view with which all biolo-

gists might not agree, at least initially, is onewhich I wish to advance in this review. It is

that, notwithstanding the indispensability ofwater in living systems and the unique proper-ties of solvent water, quantitative variations inthe amount of water available are of less directmicrobiological significance than is generallyconceded.

Discussions of microbial water relations usu-ally emphasize the stresses associated with lim-iting water availability rather than with anexcess of water. A major reason why this is so,as already pointed out (19), is that any distribu-tion curve of microbial biomass or number ofmicrobial species against salt concentrationwould be asymmetrical, with the peak or peaksin the freshwater-seawater region. Similarly,plant and animal physiologists usually treatwater stress as a stress of deficiency ratherthan of abundance. This review is no exception,

103

on Novem

ber 23, 2020 by guesthttp://m

mbr.asm

.org/D

ownloaded from

http://mmbr.asm.org/

804 BROWN

and emphasis is placed on those microorga-nisms that tolerate and sometimes require anenvironment in which the amount of thermo-dynamically available water is greatly reduced.Tolerance, in this context, refers to the abilityto thrive, not merely survive.

In very general terms, there is probably notmuch difference in the range of habitats, asdefined by water availability, tolerated by ap-propriate representatives of higher plants, ani-mals, and microorganisms. The mechanisticdetails of adjustment to water stress are quitedistinct in the case of microorganisms, how-ever, although microbial adaptive mechanismsare sometimes encountered to a limited extentin higher organisms.The nutritional physiology of the protista is

itself sufficient to cause a distinctive type ofadaptation mechanism. Since prokaryotes, andmany unicellular eukaryotes, can acquire theirnutrients only from solution, they can be grownonly in direct contact with liquid water (thereare some marginal situations associated withice). A desiccated environment for a growingprokaryote is therefore a concentrated solutionwith which the microbial cell must come di-rectly to thermodynamic terms. Although it istrue that plants also obtain their nutrientsfrom solution, land plants, at least, have alarge aerial component. Their water relationsare affected profoundly by intermittent wettingand the problem ofreducing water loss by evap-oration. This is achieved partly by control ofgas exchange by mechanical means such asclosing stomata. The same kind of generaliza-tion is applicable to animals that, in manyexamples of adaptation to a desert situation,conserve water by excreting concentrated ur-ine, coprophagy, and by various dodges thatreduce evaporation.As the following quantitative treatment ex-

plains, however, no plant or animal is knownwhose cells can tolerate solute concentrationseven remotely approaching the extreme levelsat which some groups of microorganisms canthrive and which some exceptional microorga-nisms even require. This review is concernedprimarily with such organisms.

PHYSICOCHEMICAL PARAMETERSBiological water relations have been tradi-

tionally and, to a large extent still are, dis-cussed in terms of osmotic pressure. This pa-rameter is useful when turgor and related phe-nomena are under consideration, but it is muchless useful in relation to the rigid-walled pro-karyotes in which turgor pressure cannot bereliably measured by current techniques and

BACTERIOL. REV.

for which attempts to derive them indirectlyhave been handicapped by many assumptionsand oversimplifications (19). Nevertheless, os-motic pressure, as a colligative property of solu-tions, can be validly employed if its limitationsare recognized.By strict definition, osmotic pressure is not a

pressure that can be measured directly. If asolution is placed in an osmometer, however, ahydrostatic pressure will develop, and thispressure can be measured. With a pure idealsolvent on one side of the osmometer mem-brane, the hydrostatic pressure necessary toprevent a flow of solvent is, by definition, nu-merically equal to the osmotic pressure of thesolution.

Notwithstanding textbook definitions, thereis confusion and some disagreement about thephysical implications of osmotic pressure. Twoschools ofthought can be identified among biol-ogists. The first (group 1) considers, in essence,that osmotic pressure is an intrinsic property ofa solution that exists independently of any os-mometer; as already stated, the osmometer re-veals a derived hydrostatic pressure. The otherschool (group 2) considers that osmotic pressureis not an intrinsic property and is manifest onlywith the involvement of an osmometer.Statements such as "the osmotic pressure of

the suspending solution was increased," whichare common in microbiological publications,will mean different things to the two groups. Togroup 1, the cell is subjected to an increasedexternal pressure; to group 2, the cell is sub-jected to a decreased internal pressure. Fur-thermore, the expression "osmotically activesolute" should logically have a different mean-ing to the two groups. In fact, it seems to meanmuch the same thing to everyone, suggesting abreakdown in the logic of group 1 and, onceagain, confusion between osmotic and hydro-static pressures. To group 1, all solutes must,by definition, be osmotically active, but in prac-tice they are so described only if they are effec-tively retained by a membrane and can contrib-ute thereby to the generation of a hydrostaticpressure in an osmometer. Considerations ofthis type have led some cell physiologists to usethe term "osmotic potential" to denote a capac-ity to develop a specified hydrostatic pressureunder appropriate conditions (97).Osmotic pressure is related positively to the

concentration of solutes. For a dilute ideal solu-tion, the Van't Hoff relation states:

(1)i= RT -j Vw-nw

in which IIL is osmotic pressure (the subscript s

on Novem


mbr.asm

.org/D

ownloaded from


MICROBIAL WATER STRESS 805

being used to denote an ideal solvent), R is theuniversal gas constant, T the absolute temper-ature, fj the number of moles of species j, V,,the partial molal volume of the solvent (wa-ter), and n,, is the number of moles of water.The summation, I, denotes that all solutes in

the system are considered. The expressionnj/ V,,,n,,, is a statement of concentration andcan be substituted by a symbol for concentra-tion, c,. Thus, equation 1 becomes

H = RTEc, (2)

(The derivation of this equation and those thatfollow is given in a number of books on thephysical chemistry of solutions. Those byRobinson and Stokes [1151 and Nobel [97] areof particular relevance, the latter being ori-ented throughout to biological systems.)

In a solution of "real" (that is, nonideal)solutes in an ideal solvent, equation 2 be-comes:

H = RT E yjcj = RT Y2 a, (3)i j

where yj is the activity coefficient (1 in anideal solution, <1 in nearly all other cases),and aj is the activity of the solute.There is no evidence that osmotically gener-

ated hydrostatic pressure is a factor of anysignificance in distinguishing the physiologyof, say, a halophilic bacterium from one thriv-ing in a dilute environment. Although thenatural environment of halophils would de-velop an enormous pressure in an osmometer,their internal contents are also very concen-trated (see below) and, as a consequence, theseorganisms are subjected to a net hydrostaticpressure that is probably less than that devel-oped in freshwater microorganisms. The evi-dence generally is that the only serious require-ment of hydrostatic pressure is that it be keptwithin those limits sufficient to maintain acell in a satisfactory state of turgor. There isno evidence, of which I am aware, that varia-tions in hydrostatic pressure within the limitsoperable in a functioning cell, or within thelimits encountered between healthy cells ofdifferent habitats, have any measurable effecton enzyme function. Of course, hydrostaticpressure can be a factor of great significance ina transient situation, such as a change in sol-ute concentration of the environment of a mi-croorganism. Under such conditions hydro-static pressure sufficient to burst susceptiblecells can be generated, but this is not the kindof phenomenon with which we are concernedin this review.

Nevertheless, between dilute and concen-trated environments there are fundamentalphysicochemical differences that requirequantitative description. The other colligativeproperties of a solution, namely, elevation ofboiling point, depression offreezing point, andthe lowering of vapor pressure of the solvent,are no less useful than osmotic pressure inproviding a numerical expression related to athermodynamic property of the solution.These measures are rarely used in biology,however, presumably because they lack theapparent direct physiological significance thatis commonly attributed to osmotic pressure.This independence could be advantageous.There are two fundamental properties of so-

lutions directly related to all four colligativeproperties. These are the chemical potential ofa solvent and the thermodynamic solvent ac-tivity. The chemical potential, u, of substancej is defined as the partial molal Gibbs freeenergy and is denoted thus,

u = (an) ni, T, P,E, h (4)

where F is the Gibbs free energy, nfj the num-ber of moles of substance j, ni the number ofmoles of all other substances present, and T,P, E and h being temperature, pressure elec-trical potential, and height, respectively.Chemical potential is calculated from the

equation,

itj = hi° + RTlnaj + VjP (5)+ ZiFE + migh

where AO is the chemical potential of sub-stance j in a standard state that can be arbi-trarily chosen, aj is the thermodynamic activ-ity of substance j, Vj is the partial molal vol-ume of substance j, ZjFE is a term represent-ing the effect of electrical potential on chemi-cal potential, and mjgh is a gravitational termthat is of importance, for example, in relationto sap movement in trees but of negligiblesignificance for microorganisms. The electri-cal term can be omitted for nonelectrolytesand for water, and thus the chemical potentialof water in a situation where gravitationaleffects are ignored can be written:

g4,0 = gO~j° + RTlnaO + 'VP (6)where al,, is the thermodynamic activity of wa-ter. The dimensions of chemical potential areexpressed as energy per mole, the units com-monly used being calories per mole or joulesper mole.Equation 6 shows that assignment of an ab-

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


806 BROWN

solute value to ,t requires an absolute valuefor le that is not obtainable. This is no prob-lem in biological situations, however, in whichthe significance of a chemical potential, likethat of other potential notations, lies in poten-tial difference and the value for the standardstate is constant.The potential difference, (A, - p.,10)/(V,), is

sometimes called the water potential and isgiven the symbol /. Thus, from equation 6:

RTln,V W +P (7)vw

The parameter, 4i, has received little attentionfrom microbiologists but is frequently used byplant physiologists, although not always cor-rectly. Thus Hsiao (61) equated i withlnaw(RT/V), which is, in fact, an expressionfor the determination of osmotic pressure (seebelow).Over recent years, microbiologists have

tended to discuss microbial water relations interms of water activity, aw, as used in thepreceding equations. This practice received amajor impetus in food microbiology, largelyfrom the work of W. J. Scott (see, for example,reference 121). As the previous equationsshow, lnaw is related directly to water poten-tial and there are some practical advantagesin its use. Among these advantages are itsrelative ease of experimental determination,its simple mathematical manipulation, and itsdirect relation to some other easily recognizedand measured properties of solutions.

In the foregoing discussion, aj, the activityof substance j, can be defined on a molar ormolal concentration scale or as a mole frac-tion. The activity of a solvent is usually basedon its mole fraction. Water activity, aw, isdefined as vyNw where yw is the activity coeffi-cient and Nw is the mole fraction of water =nw/(nw + ni) in which nw is the number ofmoles of water and ni is the number of moles ofall solutes. This definition, of course, has thesame form for any other solvent. Therefore, ina pure ideal solvent, ni = 0 and aw = 1.One form of expressing Raoult's law is

shown in equation 8:

P=

nwPo nw +N (8)

where P and PO are the vapor pressures ofsolution and solvent, respectively. Thus, wa-ter (or solvent) activity is numerically equal tothe vapor pressure of the solution relative tothat of the pure solvent.A convenient numerical example sometimes

used to illustrate equation 8 is that of a 1 molal

ideal aqueous solution. Since the molecularweight of water is 18.016, 1 kg of water con-tains 55.51 mol; in other words, pure water is55.51 molal. For a 1 molal solution, the right-hand side of equation 8 is 55.51/56.51 = 0.9823.Furthermore, it is apparent that such a solu-tion will equilibrate with an atmosphere of98.23% relative humidity. The water activityof a solution is therefore numerically equal to10-2x the equilibrium relative humidity of asolution when the latter is expressed as a per-centage.

In an ideal solution, a, is independent oftemperature. The effect of temperature issmall for dilute nonideal solutions and, forconcentrated nonideal solutions, is significantto the extent that it affects the activity coeffi-cient of solvent or solute.

Finally, the fundamental relation betweenosmotic pressure, 11, and a, is shown in equa-tion 9:

-RT lna.Vw

(9)

The negative sign demonstrates that osmoticpressure responds oppositely to changes in a,.

Table 1 shows a relation between water ac-tivity and the ability of microorganisms togrow. Some other biological and biophysicalphenomena, as well as representative food-stuffs, are included in the table as a series ofreference points. Conspicuous among thesereference points is the high at,, at which plantswilt. No vascular plant is known whose cellscan tolerate levels of water activity as low asthose of a wide range of common bacteria, letalone yeasts or molds.This observation not only exemplifies a fun-

damental physiological difference in the waterrelations of plants and microorganisms, but italso illustrates an ecological aspect of micro-bial water relations mentioned in the intro-ductory comments of this review. When soilsdry out, the water that remains eventuallybecomes discontinuous, the zones most resist-ant to evaporation being occluded within soilparticles. Plants can obtain this occluded wa-ter via their root hairs. Since water obtainedin this fashion will prevent wilt, it follows thatthe occluded soil water must be more dilutethan the wilt point of plants; in other words, itmust have a water activity greater than about0.98 a.. These and other aspects of plant waterrelations have been discussed extensively byplant physiologists (for example, Slatyer[127]). In turn, this emphasizes that the waterrelations of soil bacteria are not commonly amanifestation of growth at low a, but are,

BACTZRaOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



TABLE 1. Approximate limiting water activities for microbial growth"Water ac- Reference points Foods Bacteria Yeasts Fungitivity (a,,)

1.00 Blood JVegetables CaulobaterPlant wilt {Meat, fruit Spirillum spp.Seawater

0.95 Bread Most gram rods Basidiomycetous Basidiomycetesyeasts

0.90 Ham ~~~~~~~~(Mostcocci (Fusarium0.90 | Ham {Lactobacillus Ascomycetous yeasts I MucoralesWBacillus

0.85 Salami Staphylococcus Saccharomyces rouxii,(in salt)

Debaryomyces (in salt)0.80 Fruit cake S. baijii (in sugars) Penicillium

Conserves0.75 Salt lake Salt fish Halophils Wallemia

Aspergillus{Cereals Chrysosporum

0.70 ConfectionarytDried fruit

0.65 Eurotium

0.60 S. rouxii (in sugars) Xeromyces bisporus

0.55 DNA disordered

a This table was obtained in its essentials from J. I. Pitt. With the exception of the halophilic bacteria, the organismslisted had a tolerance range from a,, 1.00 (approximately) down to their tabulated level. The halophilic bacteria, togetherwith halophilic species ofthe alga Dunaliella and the halophilic actinomycete Actinospora halophila, have an upper limit aswell (see text). The growth characteristics of halophilic bacteria suggest that they are limited at 0.75 a,, by the solubility ofsalt rather than by their physiology.

instead, a consequence of physical discontinui-ties of liquid water. Under such circum-stances, the problem confronting a soil bacte-rium is one of surviving equilibration with air,or whatever the gas phase might be, in a soilmicroenvironment. Survival under these cir-cumstances is a very complex phenomenonand involves more than peculiarities of micro-bial physiology. Thus, for example, the clay,montmorillonite, enhances the survival dur-ing desiccation of fast-growing strains of Rhi-zobium but not the slow-growing strains (91).

SOME ECOLOGICAL ASPECTS OFMICROBIAL WATER RELATIONSMicrobial Water Relations and Foods

Human foods, which are usually reasonablynutritious for heterotrophic microorganisms,even if not always for humans, are preservedessentially by eliminating viable microorga-nisms and preventing readmission (as in can-ning) or by adjusting physicochemical condi-tions so that microbial growth is retarded orprevented. Lowering water availability is oneof the major processes in this second category.It can be achieved by freezing, by the physicalremoval of water, or by the addition of solutes,as in curing.As already indicated, the use of the parame-

ter, a, in microbial physiology and ecologyreceived a major impetus from food microbiolo-gists. Notwithstanding the comments later in

this review in which the direct physiologicalsignificance of a,,, is questioned, there can be nodoubt that a,, is a very useful, perhaps the mostuseful, parameter so far employed to describemicrobial water relations in a complex environ-ment, such as a foodstuff. This is reflected inTable 1, which, among other things, puts sal-monellosis and staphylococcal food poisoning,as well as outright food spoilage, in perspectiveon a water activity scale and in relation tospecific types offoods. The table also shows thatsome of the extreme types of microorganisms,such as the halophilic bacteria and xerotolerant(osmophilic) yeasts, are potential spoilage orga-nisms for various types of relatively dry foods.A detailed discussion of this topic, however,

is beyond the scope of the present review. Addi-tional information is contained in articles byIngram (62), Scott (121), Christian and Waltho(40), Pitt and Christian (109), and Pitt (108).

The Saline EnvironmentTable 1 states, in essence, that the prokar-

yotes that can tolerate the lowest a,. are theextremely halophilic bacteria that grow well at0.75 a,,,, the value for a saturated solution ofsodium chloride. The physiology of these bacte-ria is dominated by an absolute requirement forsodium chloride at a high concentration; theywill not grow at less than about 2.9 M sodiumchloride. Although some other salts such ascalcium chloride are tolerated at moderate con-

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


808 BROWN

centration, sodium chloride, as Brown (19) haspointed out, is the only readily soluble salt,which is known to be able to support some formof life over its entire concentration range. Noother salt of any monovalent cation is knowneven to approximate that capability.A requirement for salt at concentrations

above 3 M is, of course, associated with a verydistinctive ecology. Highly concentrated and of-ten saturated solutions of salt occur naturallyin marine pools concentrated by solar evapora-tion, the Dead Sea (total salt concentration, 23to 33%, wt/vol) and various inland lakes, forexample in Australia and the United States, inwhich salt concentration reaches saturation.Halophilic microorganisms have been isolatedfrom all of these sources. Some soils are quitesaline but, as already pointed out, uneven dis-tribution of water, both temporally and spa-tially, presents special problems of microbialsurvival in soils over and above ability to thrivein an environment of specified composition.There is little reliable information about theoccurrence of halophilic microorganisms insoils. Halophils are well known in some indus-tries. The preparation of salt by solar evapora-tion of seawater is sometimes accompanied bythe appearance of a red coloration attributed tothe growth of halophilic bacteria. The develop-ment of this color has been recognized sinceancient times and frequently used as a guide tothe degree of concentration of the brine (7).Brines used for curing meat and fish productsby modern techniques usually contain about12% (wt/vol) sodium chloride, together withsome other salts such as sodium nitrate. Thereis also a significant amount of organic nutrientextracted from the food being processed. Suchbrines provide excellent growth media for mod-erately halophilic and salt-tolerant microorga-nisms; according to Ingram, counts of 107 to 108/ml are common (49). Some of these bacteriahave a role in the curing process by reducingnitrate and contributing to the color and flavorof the foodstuff. An account of the microbiologyof curing brines is contained in a symposiumedited by Eddy (49). A salt-tolerant strain ofDesulfovibrio has been implicated in thespoilage of brines used for pickling olives (110).

Since high salt concentrations constitute apowerful selective influence, the ecology of saltlakes is inevitably simpler (that is, there arefewer species) than that of marine or freshwaterenvironments. Early reports (7) that brineshrimps, worms, etc., die out as salt concentra-tion approaches saturation do not seem to havebeen disproved. I am not aware ofany reports ofanimals or vascular plants in saturated-saltlakes. Aspects of salt tolerance by some ani-

mals have been discussed by Bayly (10). Never-theless, such lakes commonly harbor microbialcommunities, which implies either chemo-trophs sustained entirely by elution of nutrientfrom the surrounding land or else an ecosystemthat contains both photosynthetic and hetero-trophic microorganisms. The latter condition isthe general one, although there has been onereport (94) of an Antarctic pond with a freezingpoint of about -480C, which apparently con-tained bacteria and yeasts but no viable algae.(One diatom frustule was observed.) The pre-dominant cation in the pond was Ca2+. Theauthors reported culturing several bacteria andyeasts, but the significance of this is uncertain,since they apparently used nonsaline growthmedia. On the other hand, K. Kerry (personalcommunication) observed bacteria, yeasts,blue-green algae, and eukaryotic algae in an-other Antarctic lake which contained salts inthe same proportion as in seawater but whichremained unfrozen at -180C.

In warmer climates, evaporation from saltlakes can proceed to the point of saturationwith sodium chloride. The concentration proc-ess is accompanied by sequential changes in theviable population of the lakes; in the laterstages, a red or pink color is sometimes devel-oped in the water or on the salt crystals that aredeposited on the foreshores. Apparently, such acolor can be caused by ferric oxides or hydrox-ides (7), but it can also indicate a microbialbloom. There seem to be differences of opinionas to whether the blooms are predominantlyalgal (such as Dunaliella salina) or bacterial,and it can reasonably be assumed that bothtypes occur. The bloom expresses a selection bysalt concentration of extremely halophilic orhalotolerant microorganisms. It also implies, ofcourse, an unusually high nutrient concentra-tion. Evaporation would increase nutrient con-centration in at least the same proportion as itincreased salt concentration. In fact, the pro-portion is likely to be greater, since the increas-ing salt concentration would normally kill offsome organisms present in the "dilute" lakeand convert them, thereby, from biomass tononviable organic matter available to the sur-viving heterotrophs. There is also evidence (seebelow) that some extremely halophilic bacteria(species ofHalobacterium) are capable ofphoto-phosphorylation, in which case they might beexpected to make very efficient use of organicnutrients for biosynthesis under appropriateconditions.Few types of halophilic microorganisms have

been identified. Those that have been are soubiquitous in highly saline environments thatthere can be little doubt that they are indeed

BACTERIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from


VOL. 40, 1976

the major representatives of the halophils. Onthe other hand, it is unlikely that they are theonly types. Among the algae, the genus Duna-liella is outstanding and, indeed, is commonlythe only algal species encountered in saturated-salt lakes (17). This genus has marine and halo-philic representatives. The three best knownhalophilic species are Dunaliella salina, Duna-liella viridis, and Dunaliella parva; their saltrelations are a little different from each otherand substantially different from those of theextremely halophilic bacteria (see below).Nonetheless, at least the first two can grow insaturated solutions of sodium chloride. A halo-philic photosynthetic bacterium, Ectothiorho-dospira halophila, has been isolated (113); itssalt relations are closer to those of the halo-philic Dunaliella than to the extremely halo-philic bacteria. Recently, an extremely halo-philic actinomycete, Actinopolyspora halo-phila, has been isolated and described (54).The extremely halophilic bacteria are hetero-

trophic,- aerobic, and usually highly pig-mented-either red or pink. There are two ma-jor types. One is rod shaped and lacks a "con-ventional" bacterial wall (see below) andshould therefore be classified as gram negative.(Direct Gram stains on halophils are of doubt-ful value unless adequate precautions aretaken against disruption of the bacteria on theslide.) This genus is Halobacterium, the majorspecies being Halobacterium halobium, Halo-bacterium salinarium, and Halobacterium cu-tirubrum. The validity of the species differen-tiation is questionable. The other major type isa thick-walled (i.e., gram-positive) coccus forwhich the nomenclature is less clear than forthe halobacteria. The generic names Sarcina,Micrococcus, and Halococcus have been used(70). The name Halococcus will be used hence-forth in this review. In addition there are ap-parently fragile (gram-negative) cocci thathave not been investigated systematically. Wehave encountered such organisms in watersamples from Australian salt lakes.

I am not aware of reports of extremely halo-philic chemotrophic anaerobes, nor reliable re-ports of yeasts or molds in saturated salt lakes.Table 2 lists the approximate salt relations of anumber of microorganisms.A curious aspect of the ecology ofthe extreme

halophils is their ubiquity. The natural habitatsof these organisms are discrete and widely sep-arated, yet, so far as we know, all salt lakes thatare not otherwise inhibitory contain halophilicmicroorganisms. Presumably, this ubiquity isthe result of dissemination rather than inde-pendent evolution in each lake. Halophils, es-pecially the bacteria, are extremely sensitive to


dilution (the halobacteria are killed at salt con-centrations less than about 2.8 M) and to vir-tual dissolution in a dilute environment. Theywould seem, therefore, to be highly vulnerableto rain, especially during transport by wind,birds, etc.

The Nonionic EnvironmentThere are few natural habitats in which a

low water activity occurs predominantly be-cause of a high concentration of a nonelectro-lyte. To some extent, of course, this generaliza-tion is modified by one's definition of a naturalhabitat. When such environments do oc-cur, their major microbial inhabitants are thesugar-tolerant yeasts and fungi. Sugar-tolerantyeasts are normally found in nectar and aretransmitted by bees among nectaries and tohoneycomb (62). When honey is spoiled by fer-mentation, these yeasts are usually responsi-ble.The circumstances under which sugar- or

salt-tolerant yeasts are most commonly encoun-tered by microbiologists are associated with thefood industry, in which they frequently appearas spoilage organisms and the agents of unde-sirable fermentations. Some indication of theirspoilage potential, as a function of water activ-ity, is given in Table 1. Since most knownsugar-tolerant yeasts and molds are tolerant of,but do not require, high solute concentrations(there are some exceptions; see below), they areable to grow over a very wide range of wateractivity. Thus, they have been associated withthe spoilage of honey, wine must, maple syrup,fruit juices, dessert wines, dried fruits, molas-ses, malt extract, conserves, etc. Salt-tolerantyeasts frequently inhabitat pickling brines andare needed in the preparation of some orientalfermented foods such as soy sauce and misopaste. The biology, distribution, and nomencla-ture ofthese yeasts are described by Scott (121),Ingram (49, 62), Onishi (106), and Pitt (108).Scott's and Pitt's reviews include two of the fewaccounts of aspects of the biology of xerophilicmolds.

Regardless of their ecological and industrialsignificance, however, these organisms war-rant physiological study because, as Table 1implies, they are able to thrive at water activi-ties far lower than any other known organismor, more specifically, the cells of any other typeof organism.HYPOTHETICAL EXPLANATIONS OFTHE TOLERANCE OF LOW WATER

ACTIVITYMicrobiologists with even a passing ac-

quaintance with the extreme types of microor-

on Novem


mbr.asm

.org/D

ownloaded from


TABLE 2. Approximate salt (sodium chloride) relations of some representative microorganismsaOrganism Salt tolerance (M) Salt optimum (M)

Non-halophilic bacteriaGram-negativeSpirillum serpensS. undulaEnterobacteriaceae

Pseudomonas IAchromobacterSerratia )Vibrio parahaemolyticusMarine bacteria

Gram-positiveAnaerobic sporers

Aerobic sporers

Lactobacillus plantarumStreptococcus faecalisCocciStaphylococcus aureus

Moderately halophilic bacteriaVibrio costicolusMicrococcus halodenitrificans

Halophilic and extremely halophilic bacteriaEctothiorhodospira halophilaHalococciHalobacteria

Marine and halophilic algaeDunaliella tertiolectaD. viridis

Halophilic actinomyceteActinospora halophila

0-0.2

0-0.70-1.4

0-0.7

0.2-0.3

0.2-0.3

0.2-1.50.2-0.70.2->1.40.3-0.7

0.70.7->1.4

0-0.90-1.60-2.60-3.40-1.60-1.70-20->3

0.2-4.0

1.5-5.12-saturated

2.6-saturated

0. 17-1.51.7-saturated

2-saturated

0.2

0.2

LowLowLow0.2

1.0

1.9-3.83.4-5.03.4-5.0

0.171.7

a The salt relations listed in this table are simplifications and are intended primarily as a general guide.Salt tolerances vary with other environmental factors such as temperature, pH, and nutrition. The listing ofmore than one tolerance range for a type of organism implies that subgroups have been recognized. Whengiven as a single concentration, salt optima should be treated even more cautiously than the tolerances, butthe optima do indicate which end of the tolerance range supports better growth. A tolerance limit of 0indicates that growth can occur, at least in some media, without the specific addition ofsodium chloride. Theinformation listed in this table was compiled from various sources, including my observations and refer-ences 15, 38, 39, 54, 75, 92, and 113.

ganisms under discussion have generally' beenintrigued by the physiological basis of theirunusual environmental tolerances. The ex-tremely halophilic bacteria have attracted themost attention. It is of some interest that theU.S. National Aeronautics and Space Adminis-tration (NASA) supports a research programinto these bacteria, presumably because thatthey are among the most extraordinary micro-organisms on earth, at least in their environ-mental requirements, and therefore morelikely than any others to resemble any extra-terrestrial organisms which might be encoun-tered in NASA's exploratory activities.

For many years biologists were reluctant toaccept the idea that any living system couldoperate in a saturated salt solution and, for thisreason, sought explanations of halophil physi-ology in a supposedly dilute interior. It is nowwell known that the interior is not dilute and,moreover, that there is no mechanism availa-ble to bacteria that could make it so (seeBrown, reference 19).A microorganism responds to a new physico-

chemical situation essentially in two stages.The first occurs when the organism, whichmight or might not be growing, is transferredto a new environment. The changes that occur

8 10 BROWN BACTERIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from


VOL. 40, 1976

in stage 1 are associated with the cell's thermo-dynamic adjustment to the new conditions andare comparatively rapid. When, as in the cir-cumstances under discussion, the environmen-tal change involves a change in water activity,the thermodynamic adjustment always in-volves a transient osmotic stress. The adjust-ment process can be modified if the organismhas a source of energy.The second stage of adaptation occurs when

the organisms have adlusted thermodynami-cally and then grow.'Stige 2 involves altera-tions in the organism's metabolism, levels ofenzyme activity, etc., by the new environmen-tal conditions. The processes are more complexand much slower than those of stage 1. Initiallythey involve changes in enzyme activities andin the details of enzyme regulation (each ofwhich can be relatively fast) and subsequentlymodification ofbiosyntheses and changes in thedetails of control of enzyme formation. Theselatter processes can be expected to have a timescale similar to that of a generation of theorganism.When an organism is transferred to a new

environment, it faces, in stage 1, a simple, bi-nary choice, death or survival. If it survives,the limits of its environmental tolerance arethen determined by the nature and extent ofthe adaptive processes that occur in stage 2.Once an organism is through stage 2 and into aphysiological steady state, or something like asteady state, there are two basic types of expla-nation that are theoretically possible for a tol-erance of or requirement for a low a,,. Theseare: mechanism I, the proteins of a tolerantorganism are fundamentally and generally dif-ferent from those of a nontolerant organismand are intrinsically better able to functionunder the extreme environmental conditions;or mechanism H, the proteins of tolerant andnontolerant organisms are essentially similarbut, in the tolerant species, enzymes can func-tion because intracellular conditions are modi-fied so that the inhibitory effect of the environ-ment is diminished. As discussed below, halo-philic bacteria fit into class I; a functional halo-phil enzyme can be selected at random andrecognized by its response to salt. On the otherhand, although bacterial endospores do not rep-resent the type ofbiological phenomenon underdiscussion, they do provide a good example ofa type H situation, that is, enzyme protectionby a highly specialized set of intrasporal (intra-cellular) conditions. The water relations of "os-mophilic" yeasts, "xerophilic" molds, and halo-philic algae can also largely be explained on thebasis ofmechanism II (see below). Of course, it


is an oversimplification to suggest that onlytwo mechanisms are possible or that they aremutually exclusive. For example, notwith-standing the salt tolerance ofhalophil enzymes,it will be shown that halophilic bacteria dependalso on mechanism II.A variation or combination of both mecha-

nisms can be invoked if, for example, tolerantorganisms use distinctive metabolic pathways.Metabolic peculiarities might be relevanteither because they are catalyzed by enzymesthat, in any organism, have intrinsically favor-able water (or solute) relations (mechanism IA)or because the end product of the distinctivepathways can modify intracellular conditionsso as to diminish environmental inhibition(mechanism IIA). We are not aware of anyexamples of IA but, as will be shown, HA isused by all known xerotolerant eukaryotic mi-croorganisms.

HALOPHILIC BACTERIAThe extremely halophilic bacteria grow in

sodium chloride at concentrations betweenabout 2.8 M and saturated (6.2 molal). Thelower limit varies slightly with genus (Halo-bacterium and Halococcus), with nutritionalconditions, and with temperature. The growthcurves suggest that these bacteria would growat sodium chloride concentrations above 6.2 mo-lal were it possible to achieve them. The gen-eral biology of the extreme halophils is welldocumented (19, 70, 75, 76) and does not war-rant extensive repetition here. There are sev-eral distinctive and intrinsically interestingcharacteristics of these organisms, however,which are presumably a result of environmen-tal selection but which, on present evidence, donot appear to have a deterministic role in halo-phil salt relations.

Cell Envelope and Lipid BiochemistryOne such characteristic is the absence of bac-

terial peptidoglycan, not only in the envelopesofthe halobacteria (19, 27) but also in the wallsof the gram-positive halococci (24). The latterbacteria, as already mentioned, have very thickwalls of the gram-positive type, the polymericdetails of which have not yet been determined.Acid hydrolysates of halococcal walls containsugars and amino sugars, together with sub-stantial amounts of glycine, but no muramicacid (24). Presumably it is an inability to syn-thesize this compound that is responsible forthe lack ofpeptidoglycan in halophilic bacteria.Recent evidence suggests that the major halo-coccal wall polymer is a complex sulfated het-

on Novem


mbr.asm

.org/D

ownloaded from


812 BROWN

eroglycan (131). On the other hand, the halo-philic actinomycete Actinopolyspora halophiladoes contain diaminopimelic acid and is lyso-zyme sensitive (54).The extreme halophils are also noteworthy

for their membrane lipids, which do not containesterified fatty acids in any more than traceamounts, but which have, instead, hydrocar-bon side chains bound to glycerol by an etherbridge. The definitive work on halophilic bacte-rial lipids has come from Kates and his associ-ates and has been collated and reviewed byKates (66). The major phosphatides of halobac-terial lipids are derivatives of 2,3-di-O-phy-tanyl-sn-glycerol. The predominant halobacter-ial phospholipid is the diether analogue ofphos-phatidyl glycerol phosphate, which can accountfor more than 70% of lipid phosphorus in H.cutirubrum (122) and as much as 80% in H.halobium (89). The next most abundant phos-phatide in both these species is the ether ana-logue of phosphatidyl glycerol, which accountsfor 6.0 to 7.5% of lipid phosphorus in-each spe-cies. Phosphatidyl choline (lecithin, diether an-alogue) has been detected in trace amounts;phosphatidyl choline has not been reported. Themembrane lipids of the halococci are similar(149), but the halophilic actinomycete A. halo-phila, in contrast, contains esterified fattyacids, not phytanyl ether-linked lipids (54).Ether lipids are not unique to halophils (129),

but they are unusual in bacteria. They havebeen reported, however, in some other bacteriawith extreme environmental tolerances. Theseare the acidophilic thermophilic organisms,Thermoplasma acidophilum, which requires agrowth medium at pH 2.00 or less and 590C ormore, and the sulfur oxidizer, Sulfolobus aci-docaldarius, which requires a pH of 3.0 or lessand a temperature in the range 65 to 90°C.Neither of these species produces fatty acids toany significant extent, and both contain astheir major lipid component isopranol (C40H0)glycerol diethers (71, 72, 128). Sulfolobus acido-caldarius is also distinguished by some unu-sual glyco- and phosphoglycolipids (71).Halophil lipids are very acidic, a property

that is of direct relevance to the salt relations ofthese organisms (see below). The synthesis oftheir ether lipids, however, seems to be almosta chance result of a selection process ratherthan an essential mechanism of adaptation to ahighly saline environment. The synthesis ofthese lipids provides an interesting biochemicalexample of"where there's a will there's a way."H. cutirubrum does, in fact, have the genetic

specification of a fatty acid-synthesizing systemand does, in fact, produce enzymes that can

synthesize fatty acids from malonyl-coenzymeA. The enzymes are very salt sensitive, how-ever, and indeed, are inhibited by salt to agreater extent than is the corresponding fattyacid-synthesizing system from Escherichia coli(I). Apart from any of its other implications,this curious set of circumstances argues thathalophils evolved from organisms with muchlower salt requirements rather than by anyother process. From the point of view of fattyacid synthesis, the adaptive process might besaid to be incomplete but, since the bacteriamanage well enough without fatty acids, thereis no apparent evolutionary disadvantage inthis.Most extreme halophils are highly pig-

mented with carotenoids which, indeed, are themajor nonpolar lipid constituents. The bio-chemical ingenuity of halophil lipid metabo-lism lies in the adaptation of a pathway usedfor the biosynthesis of the isoprenoid chain ofcarotenoids to the production of the phytanylchain of the lipid. This pathway includes thefollowing steps (66):

Acetate -, --* Mevalonate --+ Isopentenyl pyrophosphate

ll~Phytanyl pyrophosphate 4- 4-- Dimethyl allyl pyrophosphate

The metabolic peculiarities of the extremehalophils which culminate in their lack of mur-amic acid and fatty acids are perhaps reflectedin an aspect of the antibiotic sensitivity patternof these bacteria. Neither the halobacteria northe halococci are affected by the so-called wallantibiotics (i.e., those, like penicillin, thatspecifically inhibit peptidoglycan synthesis)with the notable exception of bacitracin which,at a concentration of 20 ug/ml, completely in-hibits the growth of H. halobium and H. sali-narium (unpublished observation from S. Cum-ming in our laboratory). Recently, bacitracinhas also been reported to cause sphere forma-tion in a Halobacterium species (93).The biosynthesis of peptidoglycan and some

other polysaccharides associated with bacterialcell envelopes involves the transfer of a sugaror sugar complex from anhydride combinationwith a nucleotide (as, for example, in a uridinediphosphate sugar) to the growing polymerthrough its intermediate attachment to a lipid-soluble compound known as glycosyl carrierlipid (GCL) (117). According to the type of poly-mer being synthesized, attachment ofthe sugarto the GCL is either through a pyrophosphate(sugar-P-P-GCL) or a monophosphate (sugar-P-GCL) group. The sugar or sugar complex isthen transferred to the growing polymer with-

BACTICRIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



out the pyrophosphate (or phosphate), whichremains bound to the GCL as GCb-P-P (orGCL-P). The function of the GCL in this sys-tem, as its name suggests, is the catalytic one ofa carrier that is used repeatedly in a cycle. Toaccept another molecule of sugar from the nu-cleotide, the GCL must be available as themonophosphate (GCL-P). In those pathwaysthat involve a GCL-P-P-sugar, therefore, a de-phosphorylation is required after the GCL hasreleased its sugar:

GCL-P-P -- GCL-P + P1It is this reaction that is inhibited by bacitra-cin.The GCL has been identified in other bacte-

rial systems as a C55-polyisoprenoid alcohol.This might be fortunate for the halophils that,as stated, do not have an effective means ofmaking fatty acids but do synthesize otherpolyisoprenoid compounds. Both the mem-brane and outer layer of the halobacterial cellenvelope yield sugars and amino sugars onacid hydrolysis. It can reasonably be assumedthat these sugars are incorporated into enve-lope polymers, whether glycoproteins or poly-saccharides, with the mediation of a GCL ofthe type similar to that found in nonhalophilicbacteria.

NutritionExtreme environmental conditions should

be expected, a priori, to select against somemetabolic pathways. Such a selection is re-flected in the cell envelope and lipid composi-tion already discussed and also, presumably,in halophil nutritional requirements. The hal-obacteria and halococci are somewhat fastidi-ous nutritionally. They make little use ofsugars as an energy source and do not usuallyproduce, from carbohydrates, carboxylic acidsin quantities large enough to be measured byconventional diagnostic bacteriological meth-ods. Halobacterium salinarium produces aglucose dehydrogenase and a glucose-6-phos-phate dehydrogenase only when induced, andthen the enzymes do not have a high specificactivity (1).

Synthetic media have been devised for halo-philic bacteria. Dundas et al. (48) used a me-dium containing 10 amino acids and cytidylicacid which supported limited growth. Onishiet al. (107) used a medium containing 15amino acids, 2 nucleotides, glycerol, andeither asparagine or NH4+, which gave betterresults, although Gochnauer and Kushnerhave pointed out that the concentration of K+in this medium is suboptimal and, under some

conditions, limiting for both growth rate andyield of the bacteria (53). The two sets of au-thors differed as to which amino acids wereessential for growth. Dundas et al. concludedthat valine, methionine, isoleucine, and leu-cine were essential for the growth of H. sali-narium. Their results showed, among otherthings, that lysine was not essential. On theother hand, Onishi et al. (107) found the aminoacids essential to H. cutirubrum to be argi-nine, leucine, lysine, and valine. A. Markides,in our laboratory, using the medium of Onishiet al., has confirmed that lysine is essential forall the halophils in our collection, which in-cludes species of Halobacterium and Halococ-cus. Dundas et al. (48) diluted but did notwash their inocula; the discrepancy can possi-bly be attributed to this.

Miscellaneous CharacteristicsGas vacuoles. An unusual feature of some

strains ofHalobacterium is the presence of gasvacuoles. They were first reported by Hou-wink (60); they are unusual among microorga-nisms but are not unique to halophils. Theirformation requires the presence of a specifictype of thin, lipid-free protein membrane. Mi-crobial gas vacuoles have been discussed atlength by Walsby (145).

"Satellite" DNA. Extreme halophils haveanother peculiarity that is not related in anyobviously directly deterministic manner totheir salt requirements but which might haveimportant indirect implications. This is a "sat-ellite" deoxyribonucleic acid (DNA) originallydescribed by Joshi et al. (64) in preparationsof DNA from H. salinarium and H. cutiru-brum and later by Moore and McCarthy (95)for a number of other extreme halophils, in-cluding the halococci. The "satellite" compo-nent accounted for 11 to 36% ofthe DNA in thevarious preparations. DNA prepared by Mooreand McCarthy from two moderate halophilsand from E. halophila was homogeneous.The satellite DNA was, in all cases, less

dense than the major component and had adifferent base composition from it. For reasonsoutlined previously (19), there is a possibilitythat the satellite DNA is an artifact of prepara-tion. If it is not an artifact, then it is notunique. Satellite DNAs have been discussed inthe context of information theory with interest-ing evQlutionary implications (51).

Halophil bacteriophages. Until recently novirus of halophilic bacteria had been reported.Even the halophils have been overtaken byinevitability, however, and two reports now at-test to the existence of salt-dependent bacterio-

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


814 BROWN

pha~ges able to infect species of Halobacterium(138, 143).

Light Reactions of the Red HalophilsMost extremely halophilic bacteria are heav-

ily pigmented with carotenoids that might rea-sonably be expected to modify halophil physiol-ogy, especially in a brightly illuminated en-vironment such as constitutes the common nat-ural habitat of these bacteria. Larsen (75) hasdiscussed a possible role of carotenoids in pro-tecting the organisms against damage by visi-ble radiation and, in collaboration with Dundas(47), has shown a colorless mutant of H. sali-narium to be more vulnerable to visible lightthan the pigmented parent strain. More re-cently, however, Hescox and Carlberg (58)questioned this simple interpretation and pro-duced evidence that the carotenoids are in-volved in an energy transfer capacity that facil-itates photoreactivation by visible light afterexposure to damaging doses of ultraviolet.The evidence is now clear that radiant en-

ergy trapped by membrane pigments is used byhalophilic bacteria. First, at a simple level, H.halobium produces more pigment when grownin the light than in the dark (A. D. Brown,unpublished observation). In this respect it re-sembles other carotenoid-pigmented bacteria(136).The membrane of H. halobium contains a

pigmented protein complex that in many re-spects resembles rhodopsin (101). This proteinhas been named "bacteriorhodopsin"; it occursin the membrane bound to vitamin A aldehyde.The production of bacteriorhodopsin is en-hanced by growing the organism at reducedoxygen tensions. The pigment forms patches of"purple membrane" that can account for asmuch as 50% of the membrane surface. If thebacteria are incubated anaerobically in thedark, their adenosine 5'-triphosphate (ATP)content decreases sharply but it can be restoredby 02 or by anaerobic incubation in the light.Respiratory-chain inhibitors abolish the re-sponse to oxygen but not to light. Uncouplers ofphosphorylation that function as proton trans-locators abolish the light response (44).

Unlike rhodopsin, bacteriorhodopsin is notbleached by light. Instead, it responds to a lightflash by a reversible change in its absorptionpeak from 560 to 415 nm and a concomitantrelease of protons. Illumination of whole bacte-ria leads to proton excretion, inhibition ofrespi-ration, and an increase in intracellular ATP.These effects can be observed to best advantagein starved bacteria that are heavily pigmented.

If bacteriorhodopsin is illuminated continu-

ously, the pigment apparently oscillates be-tween the two forms identified by their absorp-tion peaks. When it is incorporated into a mem-brane, the oscillation is accompanied by a vec-torial release and uptake of H+, which resultsin a net outward flow of protons from the bacte-rium (44). The direction of the proton flux (out-wards) is the same as that which occurs duringrespiration ofthe bacteria or ofmitochondria. Itis the opposite of the light-induced proton fluxof chloroplasts. The proton flux causes an elec-trochemical gradient which, according toDanon and Stoeckenius (44), can be used inphosphorylation within the framework ofMitchell's chemiosmotic mechanism.Racker and Stoeckenius (112) reported the

incorporation of bacteriorhodopsin into phos-pholipid vesicles which then transported pro-tons in the opposite direction from that whichoccurred in whole bacteria. In other words, theyaccumulated protons in the light and releasedthem in the dark. Uncouplers of oxidative pho-tophosphorylation abolished the proton uptake.Inclusion of a mitochondrial adenosine triphos-phatase in the vesicles, however, enabled themto catalyze phosphorylation. Furthermore, illu-minated cell envelope vesicles of H. halobiumactively accumulate leucine. The leucine trans-port system is not dependent on ATP hydrolysisnor is it a function of the proton gradient. Itdoes respond to membrane potential, however,and there is some evidence that leucine trans-port is facilitated by the associated transport ofNa+ (85). Cyclic electron flow also energizesamino acid transport in vesicles from Rhodo-pseudomonas spheroides (56).

The overall process of photophosphorylationin halobacteria is not yet well understood, butit apparently involves a cyclic electron flowinasmuch as there seems to be no need for anelectron donor (in isolated vesicles), nor arethere reports of the production of reduced pyri-dine nucleotides. Recently, Oesterhelt (100) de-scribed the whole phenomenon in some detail.Illumination also elicits a motor response in H.halobium; the response is apparently mediatedby the purple membrane (59). A three-dimen-sional model at 0.7-nm resolution of the purplemembrane has been obtained by electron mi-croscopy (57).There is no evidence yet that their capacity

for photophosphorylation has any determiningrole in their salt requirements or water rela-tions, but presumably it gives the bacteria anecological advantage in their peculiar environ-ment. It should enable them to function anaero-bically in the day time, and it would also lessentheir demand for organic energy sources. This

BACTrzWOL. Ricv.

on Novem


mbr.asm

.org/D

ownloaded from



last factor is likely to be of importance in thehalophil blooms that are reported from time totime (see above). Their capacity for photophos-phorylation might also reflect a phylogeneticrelationship with photosynthetic bacteria, al-though there is a substantial difference in DNAbase composition between the red halophils andtwo photosynthetic bacteria, R. spheroides andEctothiorhodospira halophila (95).

Halophilic CharacteristicsUp to this point the properties that have been

discussed, although generally characteristic ofhalophilic bacteria and of intrinsic interest, donot have an obvious directly deterministic rolein the salt or water relations of the organisms.One exception is the acidic nature of the mem-brane phospholipids (see below). Indeed, thehalobacterial and, by inference, the halococcalmembranes in toto do have a major function indetermining the environmental requirementsof these bacteria. This is to be expected, ofcourse, when a microorganism has an absolutespecific requirement for a high concentration ofa solute and where, as is usually the case, thereis effective exclusion of that solute from thecell.The halobacterial envelope comprises a lipo-

protein membrane with an outer glycoproteinlayer (19, 27, 32, 90, 133). If the halobacterialenvironment is diluted, the bacteria, predicta-bly, will burst. Osmosis is a major factor intheir bursting but, in addition, there is anequally significant change in the cell envelopeitself; this change is not a result of osmoticfactors. If isolated envelopes are exposed to pro-gressively more dilute suspending solutions,they soon lose the outer glycoprotein layer andlater disaggregate to give what, for most practi-cal purposes, is a lipoprotein solution (18, 25,26, 73, 90).Although the halococcal membrane has not

yet been examined in these terms, it probablybehaves similarly to the halobacterial mem-brane. Halococcal lipids are similar to halobac-terial lipids (149) and, although the halococci donot burst in dilute solutions, they are disorgan-ized internally.The osmotic disruption of halobacteria has its

counterpart in any sufficiently fragile microor-ganism that equilibrates with a concentratedsolution and is then transferred to a dilute one.But for the inherent instability of the envelopeit would be reasonable to assume that (in mor-phological terms) halobacteria could be"trained" to accept the simple osmotic conse-quences of a dilute suspending medium or thereplacement of salt by sucrose. In this sense,

part of the halophilic salt relations can be seenas a tolerance, rather than a requirement.

This aspect of halophil physiology fallswithin the province of microbial water relationsin the sense in which that expression is com-monly used. On the other hand, the instabilityof the cell envelope reflects a very specific sol-ute requirement that can be described as "wa-ter relations" only by severely stretching defi-nitions and concepts. The role of salt in main-taining the structural integrity of halophil en-velopes has been discussed by Brown (19) interms of the possible contributions of the salt towater structure, hydrophobic bonding, andelectrostatic forces; it was attributed largely tothe electrostatic effects (18-21) of salts in neu-tralizing excess charges on the membrane. Asimilar explanation had been invoked earlierby Baxter (9) to explain the inactivation indilute solution of a halophil lactate dehydro-genase. The electrostatic events in the mem-brane were attributed by Brown to a net nega-tive charge caused by an excess of aspartic andglutamic acids in the membrane protein. Morerecently, Lanyi (74) has argued persuasivelythat hydrophobic interactions are a major fac-tor in the salt relations of halophil proteinsgenerally.

Lanyi's argument, however, is not that elec-trostatic effects do not occur but, rather, thatthey do not need very high concentrations ofsalt to overcome them. He points out (validly)that, as a general rule, the net charge on poly-anions is overcome by salt concentrations of 0.5M or less. He also points out that the a-helicalstructure of polyglutamic acid, which is highlycharged, becomes unstable at very high saltconcentrations. He suggests that the electro-static phenomena attributed to halophil pro-teins do, in fact, occur at relatively low saltconcentrations (0.5 M or less) and that the de-mand for very high concentrations is probably areflection of other factors such as hydrophobicinteractions. His evidence includes specific dif-ferences of various anions, an effect which hepoints out is difficult to reconcile with a simplecharge-shielding hypothesis. For example, theorder of effectiveness of various sodium salts instabilizing the envelope of H. cutirubrum wasNaCl > NaNO3 > NaClO4 (73). Those compo-nents that remained particulate at less thanabout 0.7 M salt, however, had little specificityin their salt requirements. The order of effec-tiveness of the salts in stabilizing the envelopeis also the order of increasing "salting in" capa-bility. Lanyi's results have an experimentalerror large enough to cast some doubt on thedifferences attributed to sodium chloride and

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


816 BROWN

sodium nitrate, but there was no doubt aboutthe difference of sodium chlorate from the othertwo.Other evidence relevant to the involvement

of electrostatic forces and hydrophobic bonds indetermining the characteristic properties ofhalophil proteins, and especially the halophilenvelope, is (briefly) as follows.

First, there is a theoretical basis, in the De-bye-Huckel theory, for stabilization of polyionsand polyionic aggregates by salts although,strictly, this theory is quantitatively applicableonly to dilute solutions. A more recent andmathematically more versatile theory of ionicsolutions has been advanced by Olivares andMcQuarrie (104). Second, the disintegrationof halophil envelopes responds quantitativelyto pH to give titration curves with pK valuesvery close to those of the (8- and y-carboxylgroups of aspartic and glutamic acids (18). Fur-thermore, there is a substantial excess of theseamino acids (over the basic amino acids) in themembrane proteins (18, 90). Structural changesin the envelopes of halobacteria occur in re-sponse to changes in salt concentration withinany range, but the response is steep at lowconcentrations (21). The outer layer, which ac-counts for about 15% of the envelope, is lostcompetely in 1 M sodium chloride (90). Thesequence of events in envelope disintegrationhas been described in some detail by Lanyi (73).Magnesium chloride, at a concentration of

0.02 M, however, will substitute for 4 M sodiumchloride in stabilizing the membrane, but itdoes not prevent loss of the outer layer (28, 90).Stabilization of the membrane by 0.02 M mag-nesium chloride is effective even to the extentof keeping the respiratory chain intact (31). Bi-and multivalent cations can form bridges be-tween neighboring negative charges and, be-cause of this, they are much more effective thanDebye-Huckel atmospheres of monovalentcounterions in neutralizing charges on the sur-face of a polymer or aggregate. When a phe-nomenon of this kind, that is, effective replace-ment of a high concentration of a monovalention by a low concentration of a divalent one, isencountered, it provides strong evidence thatelectrostatic forces are involved in the struc-tural changes under investigation. Bivalentcations are also able to form coordination com-plexes. Since the magnesium in a halophil en-velope can be displaced by suitably high con-centrations of sodium chloride, however, coor-dination complexes are not likely to be impli-cated to a major extent in membrane stabiliza-tion by magnesium chloride. It is thus verydifficult to avoid the conclusion that unneutral-

ized charges provide the forces that disinte-grate the halophil envelope. Of course, this con-clusion does not exclude a role of hydrophobicinteractions as one of the types of bond thatstabilizes or tends to stabilize the envelopeagainst the disruptive forces generated by re-moving neutralizing counterions. Indeed, in alipoprotein membrane, hydrophobic interac-tions are axiomatic. The question is whetherthese bonds assume abnormal significance in ahalophil and whether their strength is substan-tially affected by changes of salt concentrationabove about 0.5 M.The relative ease with which the outer glyco-

protein layer of H. halobium is lost from theenvelope on lowering salt concentration mightreflect a particular involvement of hydrophobicbonds in its attachment. The outer layer ismarginally more acidic than the membraneproteins, but it contains no lipid and it has aslightly lower content of amino acids with non-polar side chains (90). Furthermore, such ac-cess as it might have to the membrane lipid andto the nonpolar side chain of the membraneproteins is likely to be restricted. Its attach-ment to the envelope is easier to reconcile withpolar bonds of some kind (hydrogen bonds, ionpairs, or salt bridges) than it is with hydropho-bic bonding.Other evidence for a role of surface charges in

membrane disaggregation was given by succi-nylating the envelope of a marine pseudo-monad (20). This treatment, which substituted-COO- for -NH3+ groups, increased the sur-plus of carboxyls in the envelope to a level onlyslightly less than that occurring naturally inthe membrane of H. halobium. Succinylationeliminated the autolytic properties that charac-terized the untreated pseudomonad envelopeand endowed it with an ability to disaggregatein a similar manner to the halophil envelope.Salt inhibited disaggregation, but at lower con-centrations than required by halophil enve-lopes. Furthermore, acetylation, which, bymerely masking -NH3+ groups, caused asmaller surplus of excess carboxyls than didsuccinylation, produced an intermediate levelof salt dependence.The obvious rigidity of whole halobacteria

and isolated envelopes in high salt concentra-tions requires, in the absence of peptidoglycan,an explanation that is almost certain to involveenvironmental salt. A plausible explanation,based on "relaxation effects" is available (19);relaxation effects are electrostatic phenomena.It can also be assumed that any increase in thestrength of hydrophobic interactions that oc-curs at high salt concentrations would also

BACTERIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



stiffen an aggregated structure such as a mem-brane and thus supplement the electrostaticstiffening.A factor of perhaps crucial importance in the

membrane, however, is the effect of lipid phos-phate on the charge density of protein carbox-yls. This factor is frequently ignored in thosearguments based on the simple numerical ex-cess of acidic over basic amino acids in themembrane proteins. It has already beenpointed out that halophil membrane lipids arevery acidic and that there is almost a completeabsence of neutral phospholipids. Brown (21)showed that essentially all the carboxyl groupson the membrane proteins could be titratedwith hydrogen ions (in fact, more carboxylstitrated'on the envelope than in the dispersedlipoproteins, a phenomenon attributed to con-formational stabilization on the membrane).The majority of basic groups, namely, the E-amino groups of lysine, the phenolic hydroxylsof tyrosine and, by inference, the guanidiniumion of arginine, did not titrate in the envelope.This apparent "burial" of the basic groups wasattributed to ion pairing with lipid phosphate.Such an association is not, in itself, sufficient toprevent titration, but it is a condition necessaryfor the group to remain "buried" and inaccessi-ble to the titrant; burial ofan unmasked chargerequires an expenditure of 40 to 400 kJ of freeenergy/mol (137). In the membrane, the groupsmost likely to be associated with the basicgroups on the protein are lipid phosphate. Nei-ther the titer ofphosphate nor the guanidiniumion of arginine could be determined accuratelyin these experiments (21), but some calcula-tions suggest that all the lipid phosphategroups are likely to be involved in ion pairs inthe intact membrane. Under the relevant ex-perimental conditions, the membrane of H.halobium contains about 0.8% P (89). This isequivalent to four to five atoms of P/100 aminoacid residues, assuming an average molecularweight of 150 for the amino acids. One hundredmoles of amino acids include about 6.5 mol ofarginine + lysine (90). In each molecule ofphosphatidyl glycerophosphate, there are two Patoms, with a total of three dissociable groups.If all the lipid phosphorus were present in thisform, four atoms ofP would be equivalent to sixdissociable groups, which is approximately thenumber needed to neutralize the basic aminoacids. In fact, about 80% of the lipid P occurs asphosphatidyl glycerol phosphate, with another7.5% as phosphatidyl glycerol, which has onedissociable group/P atom. It is thus evidentthat the membrane contains phosphate andbasic amino acids in about the proportions

needed for neutralization.The immediate effect of this arrangement is

to increase the effective net surplus of acidicgroups in the membrane protein from about 19to about 26 mol/100 mol. Of probably greatersignificance, however, is the effect the ion pair-ing has on protein orientation. The carboxylgroups, which are not compensated by basicgroups, are held firmly on the membrane sur-face to give a charge density substantiallygreater than they could possibly achieve in thedispersed lipoprotein.Furthermore, there are at least two types of

bonds involved in stabilizing the intact mem-brane. One of these is ionic, which disruptswhen the membrane disintegrates; this type ofbond was detected only because it broke ondisaggregation of the membrane. The other is anoncovalent and probably nonpolar associationof lipid with protein, which is highly stable andremains intact when the membrane disaggre-gates into dispersed lipoprotein (21).The conclusion seems inescapable that the

forces which are directly responsible for thegross disaggregation of halophil membranesoriginate predominantly in protein carboxylgroups on the membrane surface and that theexposure of their charges is effected by the re-moval of neutralizing cations. It is equally ines-capable that, since there are demonstrably hy-drophobic interactions in a lipoprotein mem-brane, such interactions must supply part ofthe total bond strength that holds the mem-brane together. There is a body of indirect evi-dence and some good argument that, in halo-phils, hydrophobic bonds are intrinsicallyweaker than in nonhalophils and that they re-quire the "salting out" effect of high concentra-tions of sodium chloride to bring them up to"normal" strength. It is therefore reasonable toassume that the structural changes that occurin halophil membranes in response to a slightlowering of salt concentration are primarily theresult of weakening cohesive bonds, whereasthe structural disruption that occurs on lower-ing the salt concentration further is primarilythe result of generating disruptiveforces. Un-fortunately, there is insufficient evidence to en-able this statement to be rephrased quantita-tively, nor is it easy to distinguish betweenconformational changes of membrane proteinsand the separation of lipoprotein complexesfrom one another.

Quantitative information could probably beobtained by exploiting the different relative ef-fects of temperature and pressure on hydropho-bic bonds, hydrogen bonds, and electrostaticbonds (ion pairs); increasing temperature

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


818 BROWN

strengthens hydrophobic bonds but weakensthe other two; increasing pressure has the oppo-site effect (99). To my knowledge, there is noinformation about effects of hydrostatic pres-sure on halophil membrane stability and onlylimited data about temperature. Such evidenceas there is does not, in fact, support a signifi-cant quantitative role of hydrophobic bonds indetermining the response of halophil mem-branes to changes of sodium chloride concentra-tion within the range 0.31 to 1.25 M. Withinthat range, the extent ofmembrane disaggrega-tion was much greater at 300C than at 00C (18),but it is of interest and relevance to Lanyi'sargument that the proportional differencebetween the extent of disaggregation at the twotemperatures was greatest at the intermediatesalt concentration of 0.62 M.There is clearly room for additional measure-

ments of the type suggested but, inasmuch as itis relevant to the main theme of this review,the disaggregating properties of the halophilmembrane give no support whatever to any sup-position that they are a response to a, ratherthan the type and concentration of salt in thesuspending solution.

Intracellular PhysiologyThe intracellular physiology of extreme halo-

phils is dominated by the massive accumula-tion of K+ and Cl- and by the effective exclu-sion of Na+. This was first demonstrated byChristian and Waltho (37), who found that H.salinarium, when grown to stationary phase ina medium containing 4 M sodium chloride and0.03 M potassium chloride, accumulated potas-sium to a concentration of about 4.5 molal andhad apparent sodium and chloride contents ofabout 1.4 and 3.6 molal, respectively. They alsoreported that Sarcina (Halococcus) morrhuaecontained about 2 molal potassium and 3 molalsodium after similar growth conditions. It ispossible that the sodium content reported forH.salinarium was too high because of leakageduring centrifugation and technical difficultiesof accounting accurately for Na+ in interstitialwater. It is virtually certain that the sodiumand potassium contents reported for H. mor-rhuae were substantially too high and low, re-spectively, because of errors associated withocclusion of "extracellular" salts in the enor-mously thick cell walls of these bacteria (seereference 24). Nevertheless, the basic observa-tion of massive potassium accumulation andeffective sodium exclusion by halophilic bacte-ria has been amply confirmed in many labora-tories. For example, C. E. Armstrong (Ph.D.thesis, University of New South Wales, Syd-

ney, 1975) reported potassium concentrations of4.5, 5.5, and 7.5 molal in mid-exponential-phaseH. salinarium grown in media containing 3.4,4.3, and 5.1 M sodium chloride, respectively.Sodium content was apparently 0.4 to 0.8 molalin all cases. It is noteworthy that 7.5 molalexceeds the solubility of KCl. Conditions of thissort naturally raise many questions about thedetails of enzyme function within the cell.The discussion that follows is confined to

those features of the salt relations of halophilenzymes that are of immediate relevance to themain theme of this review. A broad spectrum ofhalophil enzymes has been discussed in somedetail by Lanyi (74). From the outset, however,a clear distinction should be made between ef-fects of salt on the activity, on one hand and, onthe other, the stability of an enzyme. Condi-tions that allow an enzyme to function vigor-ously might also allow a significant rate ofinactivation. Conversely, some enzymes arestable under conditions that severely inhibitactivity. Thus, a halophil isocitrate dehydro-genase is quite stable in but severely inhibitedby 4 to 5 M sodium chloride (3).

This section is confined almost entirely toenzyme activity, since it is reasonable to as-sume that, in the viable organism, enzyme sta-bility is not a major variable in determining theresponse of halophils to the environment. Nev-ertheless, the different effects that salts canhave on stability and activity of an enzyme arerelevant to the interpretation of enzymologicalresults and to arguments about cell physiologythat might arise from such interpretation.Thus, a question that is seldom answered is towhat extent the history of an enzyme prepara-tion might affect its properties as measured inthe ensuing experiments.A special case of an interaction of stability

with measured activity is to be found in theinteresting work of Louis and Fitt (79-83) on aDNA-dependent ribonucleic acid (RNA) polym-erase ofH. cutirubrum. According to Louis andFitt, this halophil enzyme comprises two sub-units, each of about 18,000 molecular weight.The subunits, designated a and A, have verydifferent degrees of stability on removal of saltby dialysis. The a subunit retains virtually fullactivity for up to 24 h after dialysis against asalt-free buffer and 50% activity after 20 days ofdialysis. The 8 subunit, on the other hand, iscompletely and irreversibly inactivated afterdialysis against a salt-free buffer. In a "high-salt buffer" (2.5 M potassium chloride plus 1.0M sodium chloride) the subunits are stable, butthey do not aggregate, nor do they catalyze thesynthesis of RNA unless bivalent cations are

BACTERIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from


VMICROBIAL WATER STRESS 819

present. The addition of manganous chloride(10 mM) to the high-salt buffer causes the sub-units to dimerize to a complex of molecularweight 36,000, which is the active form of theenzyme. Magnesium will not substitute formanganese in the dimerization but, even whendimerized, the enzyme will not function with-out the addition of Mg2+ (100 mM), which isneeded for attachment of the enzyme to theDNA template.The effect of salt concentration on template

specificity is of special interest. At a high saltconcentration (1.5 M potassium chloride + 0.6M sodium chloride), halobacterial DNA is tran-scribed, whereas calf thymus DNA is not. Inthe absence of added salt the converse is true:calf thymus DNA promotes RNA synthesis,whereas halobacterial DNA is inactive.The effects of salt on substrate specificity are

not correlated with the guanine + cytosine con-tent of the DNA. The assay used by Louis andFitt lasted 1 h at 3700, during which time the (3-protein would presumably have been exten-sively inactivated under conditions of low saltconcentration. In fact, this is apparently thecause of the changed template specificity thatoccurs at low salt concentration. The a-proteinalone can synthesize RNA from 'a DNA tem-plate provided the reaction is primed with asuitable dinucleoside phosphate that suppliesthe first diester bond of the new polymer. Thea-protein alone will not initiate polymerizationwithout a primer and, in this respect, it resem-bles the much larger core enzyme of Esche-richia coli RNA polymerase. Furthermore, in aprimed reaction, the a-protein alone shows notemplate specificity with any of four differenttypes ofDNA, nor does it show any salt depend-ence. The ,8-protein, which is unstable in theabsence of salt, cannot catalyze RNA polymeri-zation, but it is entirely responsible for chaininitiation and template specificity in the com-plete enzyme. In these respects, it resemblesthe o- factor of the E. coli RNA polymerase.Thus, the halophil RNA polymerase has

some properties that distinguish it sharplyfrom other halophil enzymes on one hand, andfrom the corresponding polymerase from E. colion the other.

Halophil enzymes have now been sufficientlywell studied to enable a working generalizationof their salt requirements to be advanced; ofcourse, there are exceptions. In general, en-zymes associated with the cell membrane havean optimum in the region of 4 M NaCl or KCI;ribosomal enzymes have a specific requirementfor KC1 at a sharp optimum of about 4 M.Soluble enzymes of intermediary metabolism

collectively have a wider range of salt optima,but there are many, such as the nicotinamideadenine dinucleotide phosphate (NADP)-spe-cific isocitrate dehydrogenase, which have anoptimum in the vicinity of 0.5 to 1.0 M salt(NaCl or KCI).Thus, from the outset, halophilic bacteria

can be recognized as having a physiology thatinvites explanation by both the hypotheticalmechanisms already advanced. The high saltoptima must reflect proteins intrinsically dif-ferent from those of other organisms, whereasthe accumulation of K+ and virtual exclusion ofNa+ denotes an intracellular environment withenzymological implications that are likely to bevery different from those of the bacterium'shabitat. An apparent anomaly lies in the rela-tively low salt optimum of some soluble cyto-plasmic enzymes, with the implication thatthose enzymes function under conditions of in-hibition in the intact cell.A simple explanation of this is shown for

isocitrate dehydrogenase in Fig. 1. At the opti-mal salt concentration, enzyme activity washigher in either lithium chloride or sodiumchloride than in potassium chloride, but theactivity fell off very sharply in the first twosalts at higher concentrations. On the otherhand, the inhibition caused by potassium chlo-ride at concentrations above optimal was rela-

100

80-

60-

~40-"4.

20

20of

0 1.0 2.0 3.0 4.0Salt concentration (X)

FIG. 1. Effects ofsalts on the activity ofa halophilNADP-specific isocitrate dehydrogenase. Symbols:0, LiCi; 0, NaCL; A, KCI; 0, NH4Cl. Results ofAitken et al. (3) are reprinted with permission.

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


820 BROWN

tively slight. Thus, although potassium chlo-ride was less effective than either sodium chlo-ride or lithium chloride in activating the en-

zyme, it was also less effective as an inhibitorat concentrations approaching physiologicalconditions for the organism. Figure 1 allowstwo other points to be made. First, there is no

evidence whatever that au is a determinant ofenzyme activity under the experimental condi-tions, although the range of salt concentrationsused was great enough to cause substantialchanges in a,. This is scarcely surprising. Withmost biological systems and, especially withcell-free enzyme preparations, we are accus-tomed to thinking in terms of direct interac-tions between electrolytes and proteins ratherthan indirect interactions in which the electro-lyte has a nonspecific role in merely changingau. This does not exclude a possible indirecteffect of the salt on water structure and hence,presumably, on hydrophobic bonding withinthe protein molecule(s); indeed, there are rea-

sons already discussed for supposing that, athigh concentrations, salts act partly in thisway.Second, generalizations about the relative ef-

fects of salts on enzymes are of little valueunless the comparison makes due allowance forsalt concentration. Clearly, a comparisonwithin Fig. 1 would lead to a set of conclusionswhich were completely different at, say, 3.0 Msalt on the one hand, and 1.0 M salt on theother. For this reason comparisons like thatgiven, for example, in Lanyi's (74) Table 2 are

of doubtful value.

(a) (b)

A similar but more sophisticated comparisonis shown in Fig. 2: a series of values of apparentVmaj are plotted against salt concentration.The comparison between the effects of potas-sium chloride and sodium chloride is qualita-tively similar to that of Fig. 1, but it differsquantitatively and introduces a new parame-

ter, inasmuch as the concentration of fixed sub-strate affects the comparison. This is anotherreason why simple comparisons of salt effectsare difficult to defend. Figure 1 shows that it isthe effect of sodium chloride on enzyme activitywhich is conspicuously susceptible to fixed sub-strate concentration. In potassium chloride thesubstrate effect was minor.Thus, the apparent anomaly of an enzyme's

having a salt optimum well below the salt con-

centration within the cell can, at this stage, begiven a simple explanation. The organism vir-tually excludes Na+ and accumulates such mas-sive concentrations of KCI that its intracellularcontents are dominated by it. Although theseconcentrations are well above the optimum forsome enzymes, KCI is a poor inhibitor and,even at those concentrations, produces only a

minor degree of inhibition. Both from the pointof view of its accumulation and its effective"protection" of enzymes at high salt concentra-tions, potassium chloride is a physiologicallycompatible substance for halophilic bacteria.The use, by halophilic bacteria, of both the

physiological mechanisms proposed above isnow evident. The proteins are intrinsically dis-tinctive and nearly all require a salt concentra-tion that is very high by nonhalophilic stan-

(C) (d)

so I0L.

-5

0

0 2 3 4 50 1 2 3 4 50 2 3 4 SO 2 3 4 5

Concn. of salt (M)

FIG. 2. Effects of salt concentration on apparent Vm~a of a halophil NADP-specific isocitrate dehydrogen-ase. The results were obtained at several concentrations ofeach fixed substrate. In the top panels (a-d) NADP+was used at fixed concentrations of (left to right) 0.80, 0.60, 0.40, and 020 mM. In the lower panels (e-h)sodium isocitrate was used at fixed concentrations of (left to right) 1.0, 0.50, 0.25, and 0.125 mM. Symbols:0, Sodium chloride; 0, potassium chloride. Results ofAitken and Brown (2) are reprinted with permission.

o s X.

(e) (f) (s) (h)

o10

I . . IC

BACTERIOL. REV.

a

a

a

on Novem


mbr.asm

.org/D

ownloaded from



dards. This is true even of enzymes which, likeisocitrate dehydrogenase, have a salt optimumat 1.0 M or less. (It it not true of the fatty acidsynthetase system, which is not functional inthe viable organism.) On the other hand, thosecytoplasmic enzymes that are inhibited by highsalt concentrations can function because theintracellular composition is such that the im-pact of the organism's environment is softened.If potassium chloride were not accumulated,the intracellular au and, hence, total soluteconcentration would still be about the same. Itsthermodynamic adjustment to the environmentwould be achieved either by loss of water, inwhich case the major intracellular soluteswould be "pool" intermediary metabolites, mis-cellaneous salts, etc., which, for reasons dis-cussed below would be severely inhibitory atsuch concentrations, or else the bacteria wouldaccumulate sodium chloride which, as shownabove, is also inhibitory.Potassium chloride thus has a physiological

function in halophilic bacteria which is very

similar to that of the polyhydric alcohols inxerotolerant yeasts and halophilic algae (seesections, Xerotolerant Yeasts and HalophilicAlgae). We have called such substances "com-patible solutes" (29). Their essential propertiesand mode of action are discussed in the section,Compatible Solutes.

XEROTOLERANT YEASTSGeneral Biology

The adjective "osmophilic" was first used byRichter in 1912 (106). The word has been useful,but it is misleading in both of its components.For reasons already outlined, osmotic pressure

is not a major factor in the peculiar physiologyof these yeasts, and most known species (notall, however) are tolerant, not "-philic" of highsolute concentrations. The term "sugar-toler-ant" is used and has been advocated (5), butthere is some evidence for the existence of a

group ofyeasts distinguished by their salt toler-ance. The dominant sugar-tolerant genus isSaccharomyces. Two species and one varietyhave been clearly identified; they are Saccharo-myces rouxii (Table 1), Saccharomyces rouxiivar. polymorphus and Saccharomyces mellis(Lodder and Kreger-van Rij). Yeasts associatedwith moderate salt concentrations include sev-

eral genera among which Saccharomyces(rouxii), Debaryomyces, Hansenula, and Pichiaare prominent. The nomenclature and distribu-tion of the tolerant yeasts have been discussedby Onishi (106).A term that involves minimal bending of

descriptions already in use and is generally

applicable to the group as a whole is "xerotoler-ant"; it will be used henceforth in this review.The subdivisions of sugar- and salt-tolerantwill also be used where appropriate.A definition of xerotolerance will not be at-

tempted at this stage since any definition basedon growth characteristics will be either vagueor inaccurate. The major physiological charac-teristics of the group should become clearer inthe discussion which follows. Scarr and Rose(119) defined "osmophilic" yeasts as those thatcan grow in sugar solutions at concentrationsabove "65° Brix" (65%, wt/wt) at 200C; this defi-nition can be used as a useful working descrip-tion of the sugar-tolerant group.A reliable definition based on tolerance at

low a, cannot be advanced because the waterrelations of these organisms vary greatly withsolute used to adjust a,. This is revealed inTable 1, for example, which attributed widelydifferent water relations to S. rouxii in sugarand salt solutions. There are many other exam-ples of this kind of phenomenon, some of whichare cited by Onishi (106). The differences, how-ever, are not confined to those between saltsand nonelectrolytes. One interesting examplereported by Anand (Ph.D. thesis, University ofNew South Wales, 1969) occurred with a sugar-tolerant strain that could not grow in a mediumadjusted with sucrose to 0.85 a,, but did grow inthe same medium when the water activity waslowered even further (to 0.80) by the supple-mentary addition of glycerol. Furthermore,growth in media adjusted to specified levels ofwater activity with polyethylene glycol (molec-ular weight, 200) generally diminished the dif-ference in minimal levels of water activity tol-erated by sugar-tolerant and nontolerantyeasts (Fig. 3 and 4).

Results of this type illustrate very clearly, atleast for a simple environment, that any directdeterministic effect of a,, on the yeast is over-whelmed by the type and concentration of sol-ute used to adjust a,,,. Figures 3 and 4 also showa fairly consistent difference in maximalgrowth rates between tolerant and nontolerantstrains; the tolerant strains have exponentialgrowth rates which, on average, are about halfthose of their nontolerant counterparts. (Thepaper by Anand and Brown [5] should be con-sulted for comments on the "water relations" ofsome of the intermediate strains.)There is, moreover, a complication intro-

duced by temperature. Opposing effects on bio-logical processes of increasing temperature, onthe one hand, and increasing solute concentra-tion, on the other, are common (19). With somestrains of xerotolerant yeast these effects canbecome large enough to cause a change from a

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


822 BROWN

0 ~~0 NG0.2 YD YE YF

0 0 O '000*6

o 2 C E YG YH CK0-2

098 096 0-94 092 098 096 094 092 098 096 094 0-92Water activity (a)

FIG. 3. Response ofexponential growth rate of nine xerotolerant yeasts to water activity in media adjustedwith polyethylene glycol (molecular weight, 200). Results ofAnand and Brown (5) are reprinted with per-mission. The original paper should be consulted for the identity of the yeasts.

0-6 Y14 Y31

'0 S~~~~~

04 0

0

02

0-6

~~~~Y4Y400-4~~

02

0-6

Y41 Y43

0-4

0-2

0-98 0-96 0-94 0-92 0-98 0-96 0-94 0-92

Water activity (a.,)

FIG. 4. Response ofexponential growth rate ofsixnonxerotolerant yeasts to water activity in media ad-justed with polyethylene glycol (molecular weight,200). Results ofAnand and Brown (5) are reprintedwith permission. The original paper should be con-sulted for the identity of the yeasts.

tolerance of to a requirement for a high soluteconcentration. For example, Onishi (105) hasreported that, at 3000, the yeast Torulopsishalonitratophila would not grow in a dilutemedium but did grow in the presence of6% (wt/vol?) sodium chloride. At 2000, however, it grewin dilute medium. Thus at the higher tempera-ture the organism was halo- or xerophilic,

whereas at 20TC it was xerotolerant. Onishi(106) has cited several other examples of thistype of phenomenon. In our own experience,the yeast Zygosaccharomyces nectarophilus(5), which will not grow in a dilute basal me-dium at 3000 (Fig. 3), will do so in the tempera-ture range of 16 to 2300 (M. Edgley, unpub-lished data).

PhysiologyThe available evidence will not support many

generalizations about the nutritional require-ments of xerotolerant yeasts. One of the fewsupplementary growth factors that appears tobe generally required is biotin (130). The strainof S. rouxii used extensively in this laboratoryrequires biotin and pantothenate (A. J. Mar-kides, unpublished data).Some effects of solute concentration on the

ability of yeasts to assimilate sugar have beenreported and summarized by Onishi (106). Forexample, a strain of S. rouxii assimilated glu-cose readily in a dilute medium or in the pres-ence of 18% sodium chloride. It also grew read-ily in basal medium with galactose or maltoseas sole carbon source but very poorly on eitherof those sugars in 18% sodium chloride. Onishihas added the comment that very few strainswould assimilate or ferment these sugars inhighly saline media. Observations of this kindare to be expected. They need reflect no morethan differential effects of au. (or solute concen-tration) on the kinetics of the relevant reac-tions. Similar effects occur with other physico-chemical factors such as temperature.Although the xerotolerant yeasts can thrive

at levels ofwater activity even lower than thosetolerated by the extremely halophilic bacteria,

BACTRIUOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



the two types of organism are fundamentallydifferent, inasmuch as the yeasts have no abso-lute requirement for a specific solute. Neitherdo they generally require a low a,, although, asstated, there are some exceptional strains thatdo appear to have such a requirement, which istemperature dependent. Partly for this reasonand partly because ofthe great strength oftheirthick walls, xerotolerant yeasts do not have theproblems of structural integrity which beset thehalophilic bacteria in dilute solutions.There was, therefore, no a priori reason to

look initially to the cell membrane of theyeasts, as was done with the halophils, for di-rect explanations of their environmental pecu-liarities. Of course, the yeast membrane is in-volved in the overall deterministic physiology(see below) but, on present evidence, not in anyspecific sense. Instead, it seemed probable thatthe primary explanation of their water rela-tions would lie inside the cell within the broadframework of mechanism I or II as outlinedabove.Ofimmediate relevance to these mechanisms

was a comparison of the kinetics, water rela-tions, and electrophoretic properties of anNADP-specific isocitrate dehydrogenase fromthe tolerant S. rouzxii and the nontolerant S.cerevisiae. This comparison, which was madeby Anand (Ph.D. thesis, University of NewSouth Wales, 1969) and extended by Brown (un-published data), failed to show any significantdifference, by the criteria used, between theenzyme preparations from the two yeasts. Al-though the systematic comparison was confinedto one enzyme it was, in our view, sufficient todiscount mechanism I (above) as an explana-tion of the sugar tolerance of S. rouxii. Asalready noted, any functional halophilic bacte-rial enzyme has distinctive salt requirements.

It followed, therefore, that mechanism II, orsome modification of it, should operate and thatsome intracellular property such as composi-tion should be substantially and consistentlydifferent in the two types of yeasts. Broadlyspeaking, intracellular composition will be de-termined by the 'uptake of solutes from theextracellular fluid and by the retention of me-tabolites. Since the different water relations ofthe tolerant and nontolerant strains show upclearly in high sugar concentrations, any differ-ences in composition arising from solute uptakemight reasonably be expected to show up withsugars and other nonelectrolytes. Evidence forsignificant penetration of some solutes lies inthe effects of different solutes on the growthresponses of the yeasts to a,, (see above). It isdifficult, for example, to explain the effects ofsupplementary addition of glycerol to a me-

dium containing sucrose (above) without invok-ing significant penetration of at least one of thesolutes.Brown (22) found consistent differences be-

tween S. cerevisiae and S. rouxii in the amountof solute taken up as a function of the extracel-lular solute concentration when the yeasts wereincubated in buffered solutions ofany of severalsugars or glycerol. S. rouxii had a lower capac-ity for the nonelectrolytes than S. cerevisiae.The way in which the two species responded tosucrose is of some practical significance, sincethis sugar is often encountered at high concen-trations in foods and because the differenceswere consistent between all sugar-tolerant and-nontolerant strains examined. Saccharomycescerevisiae produces an invertase; S. rouxii doesnot. As a consequence, solute uptake by S.cerevisiae was an active process dominated bymetabolism to the extent that no free sucrosewas recovered from the yeast. In S. rouxii,however, sucrose uptake occurred by diffusionand, under the experimental conditions, thestrain used (YA) equilibrated with the suspend-ing solution in such a way that the intracellularconcentration was about 40% of that of the ex-tracellular fluid (22). In this sugar, then, thereare major qualitative differences in the intra-cellular composition of the two species, differ-ences attributed directly to metabolism in onetype but not the other.

Quantitative differences between the speciesattributable simply to uptake of nonmetabo-lized sugars can be illustrated by lactose which,in S. cerevisiae, equilibrated at about 70% ofthe extracellular concentration and, in S.rouxii, at about 10% (which probably meansexclusion from the protoplast, since the wallscan take up about as much as this). Glycerolequilibrated in S. cerevisiae at about the sameconcentration as in the suspending solutionwith an upper limit to the intracellular accu-mulation of about 1 M. S. rouxii, on the otherhand, approximated an equilibrated conditionat 35 to 40% of the extracellular concentration(22).These differences were minor, however, and

doubtless themselves partly determined by an-other major difference in intracellular composi-tion between the two types of yeast. All ourxerotolerant strains contained high intracellu-lar concentrations of a polyhydric alcohol; thenontolerant strains did not. With one excep-tion, the major polyol accumulated was arabitol(presumably the D-isomer) when the yeastswere grown in a basal medium. The exceptionwas a small unidentified yeast, YO (29), whichaccumulated mannitol. Another variant wasYE (5), the temperature-dependent xerophilic

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


824 BROWN

yeast (see above). In addition to arabitol, thisyeast accumulated significant quantities ofglycerol and a hexitol, presumably mannitol.The concentration of polyol within S. rouxii

(strain A, reference 5) when growing exponen-tially in a basal nutrient medium containing(initially) 0.2% glucose is normally within therange 0.6 to 0.9 molal (22). At mid-exponentialphase, the growth medium is likely to containabout 0.002 M polyol, although polyol yields arevariable in response to nutritional and otherenvironmental factors (106, 130). Nevertheless,the quantities cited indicate a concentrationfactor within the yeast of several hundred.One of the environmental parameters that

affects both polyol production and intracellularpolyol accumulation is the water activity (orthe solute concentration) of the growth me-dium. Onishi (106) and Spencer (130) have dis-cussed effects of salt on (extracellular) polyolproduction by various species and strains ofsalt-tolerant yeasts. Onishi reported a range ofresponses to sodium chloride (18%, wt/vol) byvarious strains from a doubled yield of polyolwith S. rouxii strain N28 to a halved yield withS. acidifaciens var. halomembranis. The re-sponses were also affected by aeration.Spencer (130) has summarized related results

from various sources, among which is a reportof a shift in the type of polyol produced byPichia miso in response to potassium chloridewithin the range 0 to 3 M. Increasing concen-trations of this salt shifted polyol productionfrom predominantly erythritol to predomi-nantly glycerol. M. Edgley (unpublished data)has observed an increase in the proportion ofintracellular glycerol (relative to arabitol) ongrowing S. rouxii in elevated concentrations ofpolyethylene glycol (molecular weight, 200) orespecially sodium chloride. The reviews of Oni-shi (106) and Spencer (130) should be consultedfor their extensive tabulation of polyol yieldsunder various growth conditions.As already intimated, however, the pub-

lished information refers only to extracellularpolyol. The intracellular polyol content is moredifficult to determine accurately, especiallyafter growth of the yeast in a high concentra-tion of sugar.There are, nevertheless, some clear trends in

the response ofintracellular polyol content ofS.rouxii to the water activity of the growth me-dium. The basic trend is that a decrease in a,causes an increase in the intracellular polyolconcentration. This happens not only when asugar is the solute, under which conditionssuch an effect might be expected as a directconsequence of extensive metabolism of the

sugar, but also with nonmetabolites such aspolyethylene glycol. Polyol can reach a concen-tration of at least 5 molal within S. rouxii inmedia adjusted to 0.95 a, or lower (22). Theamount of polyol that leaks out on washing isalso affected by the growth conditions, by polyolcontent, and by the solute concentration in awashing fluid. Thus yeast grown at a low a,and containing polyol at a high concentrationleaks more when washed in water in 00C thanyeast grown in a dilute medium and containingless polyol. The result of this is that, at leastunder the experimental conditions used byBrown (22), S. rouxii grown in several media at300C and then washed with water at 00C con-tained, after the washing, polyol in the rangeapproximately 9 to 15% of the dry mass of thewashed yeast.As already stated, there is no evidence that

mechanism I (above) explains the water rela-tions of the xerotolerant yeasts, but their intra-cellular composition, which is consistently andsubstantially different from that of their nontol-erant counterparts, provides the basic require-ment for mechanism II. Since the difference incomposition obviously arises from a metabolicpeculiarity, mechanism IIA is implied.What remains to be demonstrated is a role of

the polyol in yeast water relations. Workingwith nonelectrolytes, Anand (Ph.D. thesis,University of New South Wales, 1969) showedthat the extent of enzyme inhibition at reducedwater activity was dependent on the soluteused to adjust a,. Specifically, he showed differ-ences in the rates of the reaction catalyzed by ayeast isocitrate dehydrogenase in solutions ofpolyethylene glycol (molecular weight, 200) andof sucrose. He also obtained preliminary evi-dence that glycerol was far less inhibitory thanthe other two solutes. This second observationwas confirmed by Brown and Simpson (29) andis illustrated in Fig. 5. Glycerol and sucrosefunctioned solely as inhibitors; there was noevidence of their activating the enzyme as salt,at low concentrations, activated the corre-sponding halophil enzyme.Thus, in solutions of nonelectrolytes as well

as electrolytes, water activity (a,) is neither asimple nor a direct determinant of enzyme ac-tivity. Simpson (Ph.D. thesis, University ofNew South Wales, Sydney, 1976) has since ex-tended these comparisons to a wide range ofnonelectrolytes; some aspects of her findingsare discussed in the section, Compatible Solutes.

Briefly, then, polyols make two major physi-cochemical contributions to the physiology ofxerotolerant yeasts; they function (i) as an "os-moregulator" and (ii) as a compatible solute, a

BACTEMUOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



100

._tETo 0

' a

*-:100

ciE

o 40 80| [Solute] (g!100 g water)

1.' .0 096Water activity (a,,)

FIG. 5. Activity ofan NADP-specific isocitrate de-hydrogenase from the xerotolerant yeast, Saccharo-myces rouxii, measured in various concentrations ofglycerol (0) and sucrose (a). The upper and lowerpanels show the same results with solution propertiesrepresented in two different ways. Curves showingenzyme activity as a function of molal solute concen-

tration were basically similar to those in the lowerpanel. Results of Brown and Simpson (29) are re-printed with permission.

substance that at high concentration protectsenzymes against inhibition or inactivation. Inaddition, polyols can be expected to serve as areserve food supply, but there are some con-straints on this role as discussed for halophilicalgae (below). The physiological role is thusvery similar to that of K+ (and KCI) in halo-philic bacteria; it is discussed in greater detailin the section, Compatible Solutes. Fungal wa-ter relations have recently been discussed byPitt (108). The physiological basis of xerotoler-ance by fungi imperfecti has not been studied inany detail, but there is reason to assume ageneral similarity with that of the xerotolerantyeasts. Polyol accumulation by fimgi is com-mon (78).

HALOPHILIC ALGAEThe terms "halophyte," as applied to higher

plants, and "halophil," as applied to unicellularalgae, are commonly used loosely to denotesome degree of salt tolerance without necessar-ily giving a very clear indication ofthe range ofconcentrations tolerated. This interpretation istoo broad for present purposes and the noun,halophil, will be used to apply, somewhat arbi-trarily, to those algae with a requirement forsalt, usually at concentrations of about 1.5M ormore. In fact its use, even in this way, is ques-tionable. As will be shown, there are no eu-karyotes known to have absolute and clearly

defined salt requirements similar to those ofthe extremely halophilic bacteria. There areindeed halophilic algae with a salt requirementof the order of 1.5 M and a tolerance of up tosaturated sodium chloride but, for reasonswhich will become apparent, the designation ofeven these organisms as halophilic is open tochallenge.

Halophilic algae are found within the phy-lum Chlorophyta, order Volvocales. The princi-ple genera with halophilic species are Duna-liella and Chlamydomonas. Both are flagel-lates; the major morphological distinction be-tween them is the lack of a cell wall in Duna-liella. The halophilic algae that have been sub-jected to the closest scrutiny are species ofDun-aliella. Little detail is known of the salt rela-tions of Chlamydomonas, but ecological evi-dence (17) suggests that the genus is much lesshalotolerant than Dunaliella. The best knownhalophilic species ofDunaliella are Dunaliellaparva, Dunaliella viridis, and Dunaliella sal-ina which, in that order, appear to have in-creasing tolerance of high salt concentrations.In addition, the marine species Dunaliella ter-tiolecta has also been the subject of a number ofinvestigations and has been used for compari-son with the halophilic algae.The species named above are characterized

collectively and individually by a remarkablywide range of salt tolerance, probably the wid-est known for any unicellular organisms. D.salina has been reported (L. A. Loeblich, Ph.Dthesis, University of California, San Diego,1972) to grow throughout the range 0.3 M tosaturated sodium chloride and to be able towithstand sudden substantial changes in salin-ity (87, 88, 139). Ben-Amotz and Avron (14)have reported that D. parva can withstand achange in salt concentration from 1.5 to 0.6 Msodium chloride without apparent leakage ofcell contents.The halophilic D. viridis grew over the range

of approximately 1.6 M to saturated sodiumchloride when the culture medium was inocu-lated directly with a suspension grown in 3.4 Msodium chloride, with the maximum growthrate occurring close to the bottom ofthe concen-tration range (15). Similarly, the optimum saltconcentration for the marine species Dunaliellatertiolecta was at the bottom of its effectivetolerance range (approximately 0.17 to 1.5 Msodium chloride) when the inoculum was grownin 0.17 M sodium chloride (15). The range ofsalt concentrations tolerated and the location ofthe optimum each differentiate Dunaliellaclearly from the halophilic bacteria.A more fundamental distinction from halo-

I I

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


826 BROWN

philic bacteria lies in the ability of the algae tobe "trained" to extend their limits of salt toler-ance, both up and down, with the implicationthat these limits are determined in part by thehistory of the organism. Thus, McLachlan (86)reported that D. tertiolecta grew within therange approximately 0.06 to 2 M sodium chlo-ride. Craigie and McLachlan (43) extended thisto about 2.6 M by serial subculturing throughintermediate salt concentrations. Latorella andVadas (77) adapted this species to 3.6 M salt.Conversely, D. S. Kessly (personal communica-tion) has trained the halophil, D. viridis, againby serial subculture, down to 0.3 M sodiumchloride.Another biological difference from halophilic

bacteria lies in the ability ofD. viridis to acceptreplacement of a major part of its salt require-ment by a nonelectrolyte such as sucrose. Suc-cessful transfer to a sucrose medium can bemade in one step, but growth is much slowerthan in purely saline media of the same wateractivity (D. S. Kessly, personal communica-tion). In its ability to accept substitution of saltby a nonelectrolyte, D. viridis has, at least qual-itatively, a property in common with the xero-tolerant yeasts. There is, nevertheless, an ap-parent minimal requirement by the species forsodium chloride. In the presence of 1.0 M su-crose this is about 0.05 M sodium chloride (D. S.Kessley, personal communication).At yet another level, the fundamental differ-

ence between halophilic algae and bacteria isreflected in the complete absence of any evi-dence of halophilic characteristics in algal pro-teins. Thus, although the Dunaliella surfacecarries a net negative charge (A. C. T. Jokela,Ph.D thesis, University of California, SanDiego, 1969), as indeed do most microbial sur-faces, they show none of the instability at re-duced salt concentrations which is so character-istic of halobacterial membranes. Again, al-though halophilic algal enzymes have not beenstudied nearly as extensively as their halo-philic bacterial counterparts, ofthose that havebeen studied, there is no evidence whatever ofany unusual salt requirements. Halophilic al-gal enzymes are sharply inhibited by salt con-centrations well below that encountered in thegrowth medium (63, 15). Furthermore, the saltrelations of two enzymes, glucose-6-phosphatedehydrogenase and glycerol dehydrogenase, arefunctionally identical in preparations from themarine species D. tertiolecta and the halophilD. viridis (15). Admittedly, these comparisonsare limited but, as pointed out elsewhere in thisreview, a random selection of a functional halo-bacterial enzyme will reveal its peculiar saltrequirements.

Intracellular Composition

The inevitable debate over the internal com-position of halophilic algae has resembled insome respects the early speculation about halo-philic bacteria, although there does not seem tohave been explicit advocacy of a "dilute inte-rior." The debate has centered predominantlyon whether or not salt of any kind is accumu-lated to a concentration to match that outsidethe cell. Technical problems of estimating saltconcentrations within the cells have facilitatedthe expression of opposite opinions. These prob-lems are substantial with Dunaliella, which isa mechanically delicate organism and suscepti-ble to leakage and salt exchange during centrif-ugation.The evidence for and against high concentra-

tions of salt within halophilic species ofDuna-liella has not, in general, been based on directanalyses. Thus, Trezzi et al. (139) drew infer-ences from volume changes in D. salina inresponse to changes in environmental salinity.They concluded that the plasma membrane isfreely permeable to salt and that the algaeshould therefore contain salt at a concentrationclose to that of the growth medium. Similarconclusions were drawn by Marre and Servet-taz (88) and Ginzburg (52). On the other hand,Johnson et al. (63) argued from the salt sensi-tivity of a number of cell-free enzyme prepara-tions from D. viridis that this species excludessalt. Ben-Amotz and Avron (12), however, in-terpreted similar findings, together with thesalt sensitivity of isolated chloroplasts, as indic-ative ofcompartments of low salt concentrationwithin the cell. Borowitzka and Brown (15) con-firmed the salt sensitivity of two enzymes fromD. viridis and D. tertiolecta (and, as alreadystated, demonstrated the functional identity ofthe corresponding enzymes from the two spe-cies).Of more immediate relevance to the question

of intracellular composition was the demonstra-tion that species ofDunaliella accumulate glyc-erol. Craigie and McLachlan (43) had demon-strated glycerol production by D. tertiolecta in1964. Later, Ben-Amotz and Avron (14) showedthat glycerol accumulated to a concentration ofabout 2 M in D. parva when the alga wasadapted to 1.5 M sodium chloride. Ben-Amotzand Avron recognized glycerol as an "osmore-gulator," as had Wegmann (147), inasmuch asits concentration responded positively to envi-ronmental salinity changes. Similarly, Boro-witzka and Brown (15) demonstrated glycerol ac-cumulation in direct proportion to extracellularsalt concentration in D. tertiolecta and D. viri-dis. In the latter species, glycerol reached a

BACTEINOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



concentration of about 4.4 molal when the orga-nism was grown in 4.25 M sodium chloride.This value is similar to the potassium concen-tration in halophilic bacteria and to total polyolin xerotolerant yeasts when grown at compara-ble levels of a,

In simple terms, the physiological implica-tion of glycerol accumulation is that it does notleave room for much salt in the algal cell. Theexperimental results suggest that glycerol con-centration is sufficient to bring cellular waterpotential to about the same level as that outsidethe cell. Moreover, if salt did enter the cell to aconcentration similar to its external concentra-tion, the delicate membrane of Dunaliellawould not be able to withstand the osmotic/hydrostatic consequences of the considerablereduction in internal water potential caused bythe additive effect of the two solutes. Thus,with glycerol at the stated concentrations, saltuptake should be no more than a minor eventsufficient to make up any discrepancy betweeninternal and external water potentials (withallowance, of course, for the slightly lower in-ternal water potential required for turgor).This limited uptake is likely to be met by potas-sium chloride rather than sodium chloride; ex-

perimental results with D. tertiolecta and D.viridis support this assumption (15).

Polyol production is common in plants gener-ally, including algae, and has been reviewedextensively by Lewis and Smith (78), althoughthose authors specifically omitted glycerol fromconsideration. Mannitol is a common algal po-lyol, particularly among the brown seaweeds.Lewis and Smith have also pointed out theubiquity of polyols at high concentrations inlichens and noted the possible function of thesecompounds as osmoregulators in marine algaeand seaweeds.

Physiological Role of GlycerolGlycerol accumulation can be assumed to

have three major functions in the physiology ofmarine and halophilic species of Dunaliella.The first and most obvious function is that ofanosmoregulator, that is, a substance whose con-centration responds positively to extracellularsolute concentration (negatively to a,,), whichmaintains thereby approximate parity betweeninternal and external a,, (or water potential)and which therefore minimizes osmotic stressesand dehydration to which the cell would other-wise be subjected. Although this is a vital func-tion, it is not particularly specific. Any soluteretained within a cell will contribute to its os-

motic status and, in the absence of a dominantosmoregulatory solute, thermodynamic adjust-ment to a low water potential will be achievedby a water flux.

The second role is that of a compatible solute,or protector of enzyme activity. Glycerol hasnot been compared with other nonelectrolytesin its effect on algal enzymes, as it has for ayeast enzyme, and as potassium chloride hasbeen compared with sodium chloride for halo-philic bacterial enzymes (see sections, Halo-philic Bacteria and Xerotolerant Yeasts). Somesimple comparisons of glycerol and salt havebeen made using algal enzymes (15, 55). Never-theless, its protective action on yeast isocitratedehydrogenase (sections Xerotolerant Yeastsand Halophilic Algae), its failure to inhibitDunaliella glucose-6-phosphate dehydrogenaseat concentrations below about 4 M (15) and thesimple proposition that the algae would notgrow if the glycerol were not protective makeits compatible nature in algae a virtual cer-tainty.

Third, it can act as a food reserve under someconditions. Lewis and Smith (78) have dis-cussed polyols as food reserves, but it should benoted that extensive consumption of the polyolwould deplete an algal cell of its compatiblesolute and leave it with diminished protectionagainst environmental salinity. It is likely,therefore, that glycerol becomes available as.acarbon source in significant quantities only as aresult of a drop in salt concentration.The biosynthetic pathway of glycerol inDun-

aliella has not yet been studied in any detail.Since glycerol is an early product of photosyn-thesis (43, 147) and can also be formed in thedark from accumulated starch (D. S. Kessly,personal communication), it is possibly pro-duced in the chloroplasts. Furthermore, thepresence of an active NADP-linked dehydro-genase (11, 13, 15) suggests that the final stepsof the sequence are:

Triose phosphate -t Dihydroxyacetone = GlycerolPi

Glycerol could be expected to diffuse readilyfrom the chloroplast into the cytosol. Normally,dihydroxyacetone phosphate is the major prod-uct of photosynthesis which passes from chloro-plast to cytosol (144). If glycerol were producedin the cytosol from triose phosphate, we couldexpect glycolytic enzymes to catalyze the proc-ess, in which case the sequence would probablybe:

Dihydroxyacetone phosphate = Glycerol-3-phos-phate -< Glycerol

P.

In this case the reduction step would requireNADH, not NADPH. This is essentially themechanism suggested by Wegmann (146, 147),but it does not seem to happen that way.

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


828 BROWN

Regulation of Glycerol ProductionThe regulation by extracellular solute con-

centration of production and accumulation ofcompatible solutes is a complex process that ispoorly understood. It is apparent, however,that glycerol content varies as a result of syn-thesis or degradation, not simply by concentra-tion or dilution caused by water fluxes (11, 15).The NADP-specific glycerol dehydrogenase

of Dunaliella has a high apparent Michaelisconstant for glycerol (1.5 to 3.5 M) and a lowapparent Michaelis constant for dihydroxyace-tone (0.8 to 2.2 mM) (13, 15; L. J. Borowitzka,Ph.D. thesis, University of New South Wales,1974). In spite of this big difference between thetwo constants, the enzyme can be assumed tofunction as a freely reversible dehydrogenase inthe cell, since the glycerol concentrations are ofthe same order as the Michaelis constants; thispoint has already been emphasized (15). More-over, the enzyme does not have simple Michae-lis-Menten kinetics. Under some circumstan-ces, notably with glycerol as variable substrate,it gives parabolic double reciprocal plots (15).These, and related results described by Boro-witzka (Ph.D thesis, University of New SouthWales, 1974), suggest that glycerol adds twicein the reaction sequence, once as a substrateand once as an effector. The enzyme is thuslikely to be crucial in the overall process ofregulating glycerol production.Borowitzka (Ph.D. thesis, University ofNew

South Wales, 1974) has argued that thereshould be two levels at which glycerol concen-tration is regulated. One of these is the homeo-static process by which glycerol is maintainedmore or less at a constant concentration in anyone set ofenvironmental conditions. The secondlevel is that of the direct response of glycerolcontent to environmental salinity.Borowitzka has suggested that the nonlinear

kinetics are important at the first level since, atlow glycerol concentrations corresponding tothose normally encountered in D. tertiolecta,reaction velocity changes sharply in response toglycerol concentration. On the other hand, athigh glycerol concentrations, such as occur inD. viridis, reaction velocity responds relativelyslightly to changes in glycerol concentration. Inother words, the enzyme kinetics suggest afairly coarse control of glycerol oxidation byglycerol concentration in the marine speciesand a fine control in the halophil. In turn, thissuggests that glycerol concentration should os-cillate to a greater extent in D. tertiolecta thanin D. viridis. It is not yet known if, in fact, thishappens. The problem of regulating glycerolconcentration in response to environmental sa-

linity has many features in common with theregulation of polyol production in xerotolerantyeasts and, so far, is largely unexplored. Someadditional comments on regulation are made inthe section, Compatible Solutes.

Physiological Basis of Algal HalophilismWe have argued elsewhere (15, 29), and in

this review, that cells that thrive at biologicalextremes of low water activity must accumu-late a compatible solute. Glycerol accumulationcan explain in large measure the tolerance byDunaliella of high salt concentrations but, onpresent evidence, it is insufficient to explainthe apparent requirement ofD. viridis for 1.5 Msodium chloride. Nor does it explain the differ-ent salt relations of D. tertiolecta and D. viri-dis, since each can be trained to grow in theother's domain and, when it does, it containsglycerol at a concentration appropriate to thatdomain (D. S. Kessly, personal communica-tion).The essential function of a compatible solute

is to confer a tolerance, not a requirement. It istrue that a mechanically delicate alga such asDunaliella might, for purely osmotic reasons,require a certain minimal salt concentration tomaintain cellular integrity, but this is a"chicken-and-egg" argument. The presence ofglycerol might explain why D. viridis needstraining to grow at a lowered salt concentra-tion, but it does not explain why D. tertiolectaneeds training to grow at higher salt concentra-tions.

In fact, there is no significant information ofwhich I am aware that can explain the basi-cally different salt relations of D. tertiolectaand D. viridis. Certainly the differences be-tween these species do not appear to lie in anygeneralized intrinsic properties of their en-zymes. As already stated, there is no apparentdifference in their ability to adjust their abso-lute glycerol content to a specific salinity.A notable difference between the two species,

however, is observed when D. tertiolecta istrained to grow in the salt concentration rangeofD. viridis. Under those conditions it (D. terti-olecta) still has a substantially higher growthrate than the halophil (D. S. Kessly, personalcommunication). Kessly's experiments weredone under conditions of continuous illumina-tion, and it is relevant, therefore, that D. terti-olecta is reported (77) to require continuousillumination when, after training, it grows atsalt concentrations greater than 2.5 M.The difference in growth rate between the

two species is reminiscent of the generally lowgrowth rates of the xerotolerant yeasts (5) and

BACTERIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



suggests that a fundamental difference be-tween the two algal types might be found intheir energy metabolism. Thus, if the halophilwere to divert a substantially greater propor-tion of its carbon to glycerol production it wouldhave less carbon and NADPH available forother biosynthetic processes. The greateramount of glycerol so produced would impose aneed for a higher salinity for the osmotic rea-sons already mentioned, but this is unlikely tobe the whole explanation. Inasmuch as the sol-ute requirement ofD. viridis is not absolutelyspecific for salt, the alga resembles those xero-tolerant yeasts such as Torulopsis halonitrato-phila (105) and Zygosaccharomyces nectarophi-lus (strain YE, reference 5) which have an ap-parent requirement for a lowered a,. Further-more, both these yeasts lose this requirement.with a reduction in temperature (105; M. Edg-ley, personal communication). The effect oftemperature has not yet been satisfactorily ex-plained nor, to my knowledge, have the effectsof temperature on halophilic algal salt require-ments been investigated.

Algae Other than DunaliellaThe genus Chlamydomonas has species with

a minimal requirement of about 0.34 M sodiumchloride, an ability to grow in 1.7 M sodiumchloride, but an uncertain upper limit of con-centration (102, 140). Blue-green algae havebeen isolated from the Dead Sea and, ofthese, aspecies ofAphanocapsa was reported to have aminimum salt requirement of 1.0 M and togrow best between 1.5 and 3 M sodium chloride(142). The Dead Sea, like concentrated marinepools and the saline Arctic pools mentionedearlier, contains substantial concentrations ofMg2+, which is itself a significant factor in adefinition of salinity tolerance.

Little is known of the physiological basis ofsalt tolerance in any of these algae. Chlamydo-monas is reported to contain about 0.2 M Na+and 0.05 M K+ when grown in 1.7 M sodiumchloride (103). If these estimations are reasona-bly accurate, then there is an obvious need foran additional osmoregulatory solute(s). Onemight fairly assume that such a solute wouldbe a polyol or something chemically similar. Acompatible solute of some kind can be assumedto accumulate in all algae with significant salttolerance. This is discussed further in the sec-tions, Compatible Solutes and Loose Ends.

COMPATIBLE SOLUTESElectrolytes

The results discussed so far have shown thatall microorganisms capable of growth at biolog-

ically extremely low levels of au. and appropri-ately studied accumulate compatible solutes toa concentration of a similar order to that of theextracellular solute(s). The solutes are K+, Cl-,and KCl in halophilic bacteria and one or morepolyhydric alcohols in the other organisms wehave studied. It is likely that all other protistawith a conspicuously xerotolerant physiologyaccumulate either a polyol or a closely relatedhydroxylated organic metabolite. Compatiblesolutes function partly as osmotically activesubstances and partly as protectors of enzymeactivity. The mechanism of osmotic adjustmentis obvious and relatively nonspecific; it does notwarrant detailed discussion.Enzyme protection, however, is less obvious.

A semantic distinction might be made betweenan intracellular solute such as potassium chlo-ride when it functions as an activator as, forexample, with halophil ribosomal enzyme sys-tems and as an inhibitor as, for example, withthe isocitrate dehydrogenase under halophilicphysiological conditions. The compatible na-ture of potassium chloride is apparent underthe latter conditions because it is a very poorenzyme inhibitor. On the other hand, our lim-ited experimental evidence supplemented witha generous helping of intuition suggests thatthe polyols function under inhibitory conditions(i.e., at or above any optimum that might exist)with nearly all enzymes other than those in-volved directy in their metabolism. Like potas-sium chloride, they are very poor enzyme inhib-itors and, at high concentrations, display theircompatible properties because of this.Potassium chloride inhibits halophil isocit-

rate dehydrogenase not only less severely thandoes sodium chloride, but it does so in a muchsimpler fashion. Figures 6 and 7 illustrate thisstatement in kinetic terms. Secondary plots ofslope and intercept show potassium chloride tobe a simple linear noncompetitive inhibitor,whereas sodium chloride gives a complex non-linear inhibition pattern. The nonlinearity canbe overcome by increasing the concentration ofNADP+ (3).Nonlinear kinetics of this kind imply a coop-

erative action of the inhibitor which, in turn,means multiple addition of the inhibitor in thereaction sequence and/or a change in state ofthe enzyme caused by the inhibitor. It is likelythat a change in state is involved since (i) theeffect can be modified by high substrate concen-trations amd (ii) the enzyme has a higher appar-ent molecular weight in sodium chloride thanin potassium chloride. Apparent molecularweights are: in a low ionic strength buffer with-out added potassium chloride or sodium chlo-ride, 70,900; in 1.0 M KCl, 122,000; in 4.0 M

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


830 BROWN

20

a.2

01

(a)

[NADP+]1

IJ

2 -I 0 1 2 3 4

[KCe (M)

40(c) 0

0

IC (~~~~~Ilsocitrat]J

i4 201-

t:lI

(b)[NADP+]

GAG..&~

-2 -I 0 1 2 3 4

[KCI (M)

(R(d)

-2 -I 0 1 2 3 43 4

[KCIJ (M)

FIG. 6. Halophil NADP-specific isocitrate dehydrogenase. Secondary plots ofslopes and intercepts againstconcentration ofpotassium chloride. Top panels (a and b): NADP+ used at fixed concentrations of (top tobottom) 0.20, 0.40, 0.60, and 0.80 mM. Bottom panels (c and d): sodium isocitrate used at fixed concentra-tions of (top to bottom) 0.125, 0.25, 0.50, and 1.0 mM. Results ofAitken and Brown (2) are reprinted withpermission.

(.) [NADP

L ..2 3 4 5

[NaCI] (M)

s 20

t

¢Il= 501

(b) [NADP+]-w

-I O 2 3 4 S

[Naa] (M)

40 (4) [ISo

* _ 0 7_VI.%

[Naal (M) (Naal (a)FIG. 7. Halophil NADP-specific isocitrate dehydrogenase. Secondary plots ofslopes and intercepts against

concentration ofsodium chloride. Top panels (a and b): NADP+ used at fixed concentrations of(top to bottom)0.20, 0.40, 0.60, and 0.80 mM. Bottom panels (c and d): sodium isocitrate used at fixed concentrations of(topto bottom) 0.125, 0.25, 0.50, and 1.0 mM. (At high concentrations ofsodium chloride and low fixed substrateconcentration, the kinetics were not Michaelis-Menten.) Results ofAitken and Brown (2) are reprinted withpermission.

-2 -I 0 1 2

[KCIJ (M)

10Us

- I

BACTZRIOL. REcV.

_;

-2 -1 0 1 2 3 4 5

on Novem


mbr.asm

.org/D

ownloaded from


VOL. 40, 1976

KCl, 135,000; in 1.0 M NaCl, 224,000; in 4.0 MNaCl, 251,000 (2). Thus, the enzyme probablyhas a monomeric molecular weight of about71,000, dimerizes in 1 to 4 M potassium chloride(the physiological condition), and forms a tri-mer or tetramer in 1 to 4 M sodium chloride.The higher apparent molecular weight in so-dium chloride is consistent with the "saltingout" characteristics of this salt, as discussed byLanyi (74).There is evidence that salt concentration af-

fects the reaction mechanism of the halophilisocitrate dehydrogenase, but under physiologi-cal conditions it has a sequential mechanism,normal for pyridine nucleotide-linked dehydro-genases, in which NADP+ is the first substrateadded and NADPH is the last product released.The evidence also shows, however, that theeffects of salt in activating the enzyme, and athigh concentrations inhibiting it, can be inter-preted in a kinetic sense as if salt were a thirdsubstrate, causing substrate inhibition at highconcentrations. The reaction mechanism can berepresented as in reaction (10):

&

.< .0

E I I L

-Salt

E-Salt

c)

0.Central T E (10)Complexes

Thus, salt adds in a substrate-like mannerbetween NADP+ and isocitrate and also, in aninhibitory manner, to form a dead-end complexbefore addition ofNADP+. Whatever the physi-cochemical explanation of this associationmight be, there are extensive implications forthe physiology of halophilic bacteria in the factthat a relatively nonspecific substance such asa common inorganic salt can modify an enzymein such a way that the kinetics suggest a fairlyprecise interaction with the active site of theenzyme.The kinetics also allow a calculation of the

dissociation constants of the constants of theenzyme-salt complexes. The reaction mecha-nism shown above implies two salt-enzyme dis-sociation constants. One constant, that of theenzyme-"substrate salt" complex is formallysimilar to a Michaelis constant (Ki). Theother, that ofthe dead-end complex, is a normalinhibitor constant (Kd). The values of theseconstants for both salts are shown in Table 3.The table shows that the two salts give iden-


tical "Michaelis constants," and there is noth-ing at this level to distinguish between them asenzyme activators. The magnitude of the "Mi-chaelis constants" is normal for a substrate. Onthe other hand, the inhibitor constants showsome conspicuous and important differences.First, for both salts they are orders of magni-tude greater than the "Michaelis constants." Adifference in this direction is predictable fromthe simple fact that the salts do activate at lowconcentrations and do inhibit at high concen-trations. The difference in magnitude is notnecessarily so obvious. The outstanding differ-ence, however, is between the two salts, the K,(KCl) being about 10 times as great as the Ki(NaCl) and well in excess of the solubility ofeither salt.

In simple terms, therefore, the "compatible"nature of potassium/chloride for a halophil en-zyme is a direct consequence of its very lowaffinity for the enzyme at the "inhibition site";sodium chloride is a much more powerful inhib-itor because of its tighter binding. The aggrega-tion ofthe enzyme is associated with the tighterbinding of sodium chloride.The physicochemical processes underlying

this kinetic explanation are undoubtedly com-plex. From the outset, the assumption can bemade that in the low (activating) range of saltconcentration the salts are extensively ionizedand the quantitative differences between thesalts under these conditions are attributable todifferences between K+ and Na+. Thus, there isessentially no difference between the two cat-ions as reflected by the "Ki" of the enzyme-saltcomplex. There are differences, however, whichare shown in the relations between the enzymeand its substrates. In the low (activating) rangeof salt concentration, K. (isocitrate) was con-sistently higher in sodium chloride than in po-tassium chloride, but the opposite was true ofKm (NADP). Furthermore, at low salt concen-trations apparent Vma. was higher in sodiumchloride than in potassium chloride except with

TABLE 3. "Dissociation constants" of salt (or cation)complexes ofhalobacterial isocitrate dehydrogenasea

Constant NaCi KCl"X^"pp 31-32 mM 30-31 mMKi 0.9 M 9.5 M

a The values were calculated from data of Aitkenet al. (3). In its role as an activator, expressed in the"K,i," the salt probably exerted its effect largelythrough the cation. On the other hand, the inhibitorconstant (Ks) must be assumed to reflect an interac-tion with a substantial proportion of undissociatedsalt (see text).

on Novem


mbr.asm

.org/D

ownloaded from


832 BROWN

very low fixed concentrations ofisocitrate (2).The differences between the two cations as

well as Li+ and NH4+ (at low concentrations,Fig. 1) can be interpreted superficially in termsof radii or, inversely, of hydrated volumes oftheions. Not only are explanations of this typesuperficial, but they are also serious oversim-plifications that are likely to be wrong on sev-eral counts. Although it is true that the ionicradii, or hydrated volumes, do correlate withand can predict the behavior of ions under somecircumstances, these circumstances are usuallyrestricted by a number of factors including arelatively narrow concentration range of thesalts.Of far greater predictive value is the thermo-

dynamic explanation developed largely by Ei-senman. A description of this theory is beyondthe scope of the present review, but it has beendiscussed in detail in Diamond and Wright (45).Some comment is appropriate, however. In anaqueous environment the relative affinities of asite on a protein, a membrane, etc., for twodifferent cations will be a function of the freeenergies of hydration and of binding to the site;

AFNa+(site) - AFK+(site) (11)- (AFNa+h - AFK+h)

where AFx (site) is the free energy of interac-tion between the cation, X, and the site; AFXhis the free energy of hydration of the cation, X.Thus, the "K." values for the two salts,

which we assume actually reflect cation bind-ing to the enzyme, suggest that the sum ofexpression (11) is about the same for Na+ andK+. Perhaps this is coincidental, since the freeenergies of hydration are certainly not thesame (116). A complication, of course, is thatthe "K." is not an equilibrium constant but, aswith any conventional Ki, it reflects a steady-state phenomenon. Even though the cationdoes not change in the reaction, it cannot dis-sociate (after the reaction) from the same formof the enzyme as that to which it initiallyattached. This is evident from reaction 10.There can be no recurrence of the enzyme-NADP+ complex from which the cation woulddissociate to give a true equilibrium. Never-theless, it is inconceivable that the thermody-namic considerations discussed by Diamondand Wright (45) do not apply in some measureto an enzymic situation such as this. Clearly,it is an area for detailed investigation.In the high, inhibitory range of salt concen-

tration the situation is more complex because,among other things, the salts become propor-

tionately less dissociated with increasing con-centration. Furthermore, the activity coeffi-cients of the two salts respond differently toconcentration: that for sodium chloride in-creases over the relevant concentration range,whereas that of potassium chloride remainsrelatively constant (Fig. 8). This basic differ-ence in solution properties of the two saltspossibly contributes to the complex inhibitionpatterns caused by sodium chloride, but it isunlikely to provide the whole explanation. Forone thing, it cannot explain the differences inapparent molecular weight determined in thetwo salts nor, for another, does it overcome thedifferences in apparent Vmax encountered athigh salt concentration, although a correctionfor activity does reduce those differences sub-stantially (Fig. 9).A physicochemical explanation of the spe-

cial properties of compatible solutes shouldalso take into account, wherever possible, therelevant properties of solvent water. Solventproperties and the thermodynamics of solva-tion assume greater relative importance whena solute has an affinity for an enzyme that isnot very different from the affinity of water forthe enzyme. In the extreme case in which sol-ute and water are bound to a polymer in thesame proportion as they occur in solution, thesolute effectively has zero affinity for the poly-mer and, by definition, is not adsorbed (30).There is now very good experimental evi-

dence, based largely on nuclear magnetic reso-nance and dielectric dispersion measure-

1I-r

0-8zI-

,0-

>-°-4>

90 2

. O..." I

1 2 3 4 5SALT CONCN (molal)

FIG. 8. Activity coefficients of sodium chloride(0) and potassium chloride (@) as a function ofconcentration. Plotted from data in Robinson andStokes (115).

BACTERIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



40

I01

aU

o0.

__ 1 2 3SALT ACTIVITY (M)

FIG. 9. Halophil NADP-specific isocitrate dehy-drogenase. V,,, as a function of salt activity insodium chloride (0) and potassium chloride (@). Re-plotted from results ofAitken et al. (3).

ments, for three "types" of water in anaqueous solution of a protein. Cooke andKuntz (42) have described these three types inthe following way. Type I -"bulk water," therotational and translational properties ofwhich do not appear to be appreciably alteredby binding to macromolecules. A protein solu-tion (20%, wt/vol) has about 90% of its water inthis form. Type II-"bound water." The rota-tional motions and freezing point of this waterare substantially modified by interaction withthe surface of macromolecules. This waterseems to exist as one to two monolayers and ispresent in the proportion ofabout 0.3 to 0.6 g ofwater/g of protein. A 20% protein solution hasabout 10% of its water in this form. Type III-"irrotationally bound water," in which watermolecules are essentially bound to specificsites on the polymer for periods of microsec-onds. A 20% protein solution has about 0.1% ofis water in this form.

It can be assumed that the dissolution of saltto a high concentration will nonspecificallyaffect the amount or activity of type I water,will specifically affect type II water in someway, and probably will not affect type III.Physicochemical information about salt-pro-tein interactions at high salt concentrations islimited, but what information there is seemsto be consistent with the preceding commentsand with the properties of compatible solutes.For example, Bull and Breese (30) have stud-ied water and solute binding to egg albuminover a salt concentration range of approxi-

mately 1 to 3 molal. Their results suggest that,under these conditions, the chloride of the al-kali cations are only slightly bound to theprotein and that their major effect is on theamount of bound water. The values cited byBull and Breese suggest that their "bound wa-ter" is the type II water as defined by Cookeand Kuntz (above). For example, they quote740 mol of water bound/mol of egg albumin at97% relative humidity. This is equivalent to0.3 g of water/g of protein. In solution with nobinding of salt, about 1,350 mol of water isbound per mol of protein (0.5 g/g). These val-ues are within the range of type II water.Water bound to the protein over a salt con-

centration range of 1 to 3 molal was constantin both potassium chloride and sodium chlo-ride but different for each salt: 211 mol/mol(0.08 g/g) for sodium chloride and 360 mol/mol(0.14 g/g) for potassium chloride.

It thus seems likely that the compatible na-ture of potassium chloride is expressed in partthrough the limited disturbance that it causesto the type II bound water. It would clearly bedesirable to examine the inhibition of a halo-phil enzyme caused by rubidium chloride andcesium chloride since, according to Bull andBreese, they cause even less dehydration ofegg albumin than does potassium chloride.Furthermore, sodium sulfate actually en-hances the degree of solvation; kinetic studiesof its effects should also be informative.

It is also likely that salts modify enzymicreactions through their effect on "activationvolume," that is, the change in volume thatoccurs when an enzyme-substrate complexchanges from the ground state to the transi-tion state. Effects of salt on this process, how-ever, are generally restricted to salt concen-trations below about 300 mM (84).

NonelectrolytesThe compatible solute role of polyhydric al-

cohols was established in the first instance bycomparative studies of the effects of sucrose,on the one hand, and glycerol, on the other, onthe kinetics of the NADP-specific isocitratedehydrogenase of S. rouxii. In a number ofrespects this investigation was complemen-tary to that of Aitken and Brown (2) on theeffects of salts on the halophil isocitrate dehy-drogenase. At the time of writing, the greaterpart of the nonelectrolyte studies is unpub-lished but can be found in the Ph.D. thesis ofJ. R. Simpson (University of New SouthWales, 1976).

Interpretation of the kinetics of this enzymewas complicated by nonlinear double-recipro-

VOL. 40, 1976

I 1-

on Novem


mbr.asm

.org/D

ownloaded from


834 BROWN

120F

Glycerol

0 55 221 4.44

098 0-94

80jSuc

40F

0

///rose

iwi

S03A0-53 1 92 3-47

098WATER ACTIVITY

0-94

FIG. 10. Yeast NADP-specific isocitrate dehydrogenase. Secondary plots ofintercept against water activityin solutions adjusted with glycerol (left panel) and sucrose (right panel). The small numerals above theabscissa indicate the concentration (molal) ofthe solutes at selected points. Previously unpublished data ofJ.R. Simpson and A. D. Brown.

cal plots (concave down). Associated with thenonlinearity was gel-electrophoretic evidencethat the enzyme occurred in several forms,probably signifying different states of aggre-gation. Multiple forms of isocitrate dehydro-genases have been reported from other sources(114, 118, 123). Because of the nonlinearity ofthe double-reciprocal plots, secondary plots ofslope were not possible, although interceptcould be so treated. Figure 10 illustrates suchplots and gives a striking comparison with thecorresponding analysis of the halophil isocit-rate dehydrogenase in potassium chloride andsodium chloride (Fig. 6 and 7). Briefly, thecomparison shows intercept replots for glyc-erol to be linear with a small slope (i.e., inhi-bition changed only slightly with inhibitorconcentration), whereas the correspondingplot for sucrose is steep and nonlinear (inhibi-tion increased sharply with concentration andthe rate of increase was concentration depend-ent).

In this respect glycerol, a compatible solutein halophilic algae and some xerotolerantyeasts, resembles potassium chloride, the com-patible solute for halophilic bacteria. There isan equally striking similarity between the non-electrolyte sucrose, which is severely inhibi-tory and partly excluded by xerotolerantyeasts, and the electrolyte sodium chloride,also severely inhibitory and largely excluded byhalophilic bacteria. Furthermore, by extrapo-lating secondary plots of the type shown in Fig.

10, inhibitor constants can be derived for thetwo nonelectrolytes. Obviously, any constantderived in this way for sucrose can be, at best,an approximation. Nevertheless, even a casualinspection of Fig. 10 reveals a major differencein the Ki of the two inhibitors, that for sucrose

being by far the smaller. The numerical valuesderived for these constants are: Ki (glycerol),13 molal; Ki (sucrose), 1.5 molal.Thus, there is again a striking resemblance

to the values obtained for the halophil enzyme(Table 3). It is evident that, with nonelectro-lytes as well as electrolytes, compatible solutesare distinguished as inhibitors with a very lowaffinity for an enzyme.To some extent, however, the selection of

sucrose for these comparisons was fortuitous.Simpson has extended her investigation tocover a range of nonelectrolytes to include thefully hydroxylated polyols up to hexitols, in-completely hydroxylated di- and triols, and a

range of sugars.In all cases, inhibition was assessed from the

slope of plots of reciprocal velocity (1/v) againstinhibitor concentration; the steeper the slopes,the greater the response of inhibition to inhibi-tor concentration.When measured in this way, the inhibition

caused by the fully hydroxylated alcohols waspositively correlated with chain length (molec-ular weight), but the relation was sigmoidal(Fig. 11). Thus, although chain length domi-nated the inhibitory properties of the polyols,

60F

%_ 40a.wuw

z

20

BACTERIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



I-I

-a

(A

1 2 3 4 5 IC CHAIN LENGTH

FIG. 11. Yeast NADP-specific isocitrate dehydro-genase. Inhibition, expressed as the slope of double-reciprocal plots, as a function of the chain length ofacyclic fully hydroxylated polyhydric alcohols. TheC1 alcohol was methanol. Previously unpublisheddata ofJ. R. Simpson and A. D. Brown.

there was another determining factor(s) aswell. Some indication of the level at which thisadditional factor operates was obtained by plot-ting slope against the chromatographic Rf ofthe polyol. Values of Rf were obtained for thesolvent butan-1-ol + acetic acid + water (6:1:2,by volume) although, since the polyols are non-electrolytes, virtually any solvent mixture ca-pable of reflecting an "oil"-water partition coef-ficient might be expected to illustrate the point.When this was done an essentially linear rela-tion was obtained between slope and Rf, thehighest Rf correlating with the lowest slope.The probable implications of this are discussedbelow.The situation with sugars is far more com-

plex in that there was no correlation betweenslope (as used above) and molecular weight.Stereochemical factors apparently exert a ma-jor influence and several sugars, including su-crose, gave nonlinear plots of 1/v versus inhibi-tor concentration. The generalizations that canbe made about inhibition by sugars are briefly:(i) inhibition caused by straight-chain aldoses,glyceraldehyde and erythrose, is substantiallyirreversible, implying complexing between thealdehyde group and the enzyme; (ii) hexoseswere generally less inhibitory than othersugars, the least inhibitory being fructosewhich gave about the same slope as glycerol (itis probably no coincidence that some of thelowest values ofa, supporting growth of micro-organisms have been obtained in fructose); (iii)

some sugars, notably sucrose and ribose, gavenonlinear plots of 1/v versus inhibitor concen-tration, the slope increasing with inhibitor con-centration. Moreover, there was no correlationof slope with the chromatographic Rf of thesugars.The incompletely hydroxylated diols and

triols (propane and butane diols and butanetriol) all gave biphasic plots of 1/v versus polyolconcentration, the plots being divisible into twostraight lines, with the greater slope occurringat the higher polyol concentration. The order ofincreasing inhibition caused by these di- andtriols was: propane-1,2-diol; propane-1,3-diol;butane-1,2,4-triol; butane-2,3-diol; butane-1,4-diol; and butane-1,3-diol.

Physicochemical Mechanism ofNonelectrolyte Action

To this point the evidence is clear that at aseries of specified levels of a,,, an enzyme be-haves differently in solutions of different non-electrolytes. The conclusion is inescapable thata,, cannot account for the differences shown bythe various solutes and is thus unlikely to bethe major determinant in the total quantitativeeffect produced by the nonelectrolyte solutions.This should scarcely be surprising for solutionswhich might contain 10 or 15% less thermody-namically available water but several orders ofmagnitude more solute than "conventional" di-lute biological solutions. It is equally evidentthat nonelectrolytes exert a direct action onthe enzyme. There is insufficient information toenable any firm conclusions to be drawn aboutthe action mechanism, although there are someuseful pointers available from various sources.

Nonelectrolytes might affect enzyme actionby changing the viscosity, pH, ionic strength,or dielectric constant of a solution, by changingwater "structure" or by directly reacting withthe protein molecule in any of several possibleways. A partial correlation of enzyme inhibi-tion with viscosity suggests that this might, atmost, be a supplementary factor in determininginhibition (Brown, unpublished data). The mi-nor effects of the solutes on pH do not explaindifferences in inhibition (J. R. Simpson, Ph.D.thesis, University of New South Wales, 1976).Although ionic strength is increased in directproportion to the lowering ofa, and thus mightcontribute to the total inhibition encounteredin nonelectrolyte solutions, the increase shouldbe the same for all solutes. Dielectric constant(4) correlates with inhibition caused by at leastsome polyols and one sugar (glucose), but notwith sucrose (J. R. Simpson and A. D. Brown,unpublished data). The limited informationavailable for dielectric constant, together with

VOL. 40, 1976

0-2.

on Novem


mbr.asm

.org/D

ownloaded from


836 BROWN

that given by the correlation with Rf (seeabove), suggests a possible involvement of hy-drophobic interactions in the inhibition causedby polyols and some sugars.Another possible mechanism is the modifica-

tion of water "structure" by the solute. In thiscase, it can be assumed from the outset thatany relevant changes in water structure wouldoccur in the type II bound water describedabove.There is evidence that the solutes can modify

the aggregation state of a protein. Thus, Simp-son has shown by gel electrophoresis that am-monium sulfate fractionation appears to depo-lymerize the yeast isocitrate dehydrogenase,and she has obtained some evidence that glyc-erol (2 molal) has a similar effect. Ifthis is so, itagrees with earlier findings of Contaxis andReithel (41) that glycerol depolymerizes an en-zyme. These authors made a series of compari-sons of propane-1,2-diol, ethylene glycol, andglycerol on the physical chemistry and enzymeactivity of jack bean urease. They observedthat, on exposure to aqueous solutions (90% vol/vol) of each of the alcohols, urease dissociated toa molecular weight of 240,000, half the normalvalue. The rate of dissociation increased in theorder: glycerol, propane diol, ethylene glycol.There was no evidence ofany significant confor-mational change in the 240,000-molecular-weight monomers during the dissociation, andenzyme activity remained at 80 to 82% of theuntreated enzyme. Moreover, the "monomer"reassociated to the dimer on removal of thealcohol. On prolonged exposure (2 days) to eth-ylene glycol or propane diol, the enzyme reasso-ciated in a manner different from that de-scribed above to give an enzymically inactivepolymer. This did not happen in glycerol, inwhich the enzyme remained active on pro-longed exposure. (-Lactoglobulin also disso-ciates to a "monomer" in about 10% ethyleneglycol and reassociates (dimerizes) in about20% glycol (69).

Since the dissociation of urease did not ap-parently cause a conformational change in theprotein, Contaxis and Reithel assumed that theassociation was confined to a restricted zone onthe surface of the 240,000-molecular-weightsubunit. They also assumed that the alcoholspromoted dissociation by modifying waterstructure and by "competing with water forhydrogen bonding." A change in hydration waspresumed to occur near hydrophobic areas onthe protein and, by virtue ofthe entropy changethat accompanied it, to contribute to the nega-tive free energy of dissociation in the polyolsolution.Douzou, who, for some years, has studied

enzyme activity at subzero temperatures, hasrecently reviewed the theory and methodologyof this field of investigation (46). High soluteconcentrations are needed to prevent freezingat the temperatures of Douzou's experiments,and it is scarcely a coincidence that polyols,notably, ethylene glycol, propane diol, andglycerol, were the solutes used for this purpose.Douzou comments that, at normal ambienttemperatures, these polyols are far less effec-tive "denaturing" agents than monohydric alco-hols, including ethanol, and that many proteinsare stable in 1:1 (by volume) aqueous solutionsof the polyols. At first sight this observationconflicts with Simpson's finding (above) thatthe inhibitory action of methanol on isocitratedehydrogenase was (marginally) less than thatof ethylene glycol or glycerol. This apparentdiscrepancy might be explained by the distinc-tion between inhibition and denaturation;Simpson's measurements of enzyme activitywere limited to periods of a few minutes.

If nonelectrolytes do affect enzyme activityby disturbing water structure, it is probablytype II water, as suggested already, which is soaffected. It is possible, however, the nonelectro-lytes can act by more than one mechanism. Thenegative correlation of inhibition by polyolswith chromatographic Rf (Rf being used as aconvenient measure of a hydrophobic/hydro-philic partition coefficient), together with thekinetic evidence of tight binding by inhibitors(weak binding by compatible solutes), suggestthat the greater the affinity for a hydrophobicregion on a protein the more effective the inhi-bition. Conversely, compatible solutes appar-ently have a lower affinity for the hydrophobicregions and a correspondingly greater affinityfor water (a low free energy of hydration). Al-though it is phrased somewhat differently, thisinterpretation is in general accord with the as-sumptions made by Contaxis and Reithel. It isalso consistent with the interpretation of No-zaki and Tanford (98) of thermodynamic as-pects of protein denaturation in which ethyleneglycol was considered to be much less effectivethan urea in reducing the free energy of hydra-tion of nonpolar groups on the protein. More-over, their interpretation agrees with the ac-tion KCl proposed for as a compatible solute forhalophil enzymes (see above).This reasoning cannot be easily extended to

the sugars, however, whose stereochemistryappears to dominate their inhibitory efficiencyand for which there is no correlation with ahydrophobic/hydrophilic distribution coeffi-cient. Generalizations about the significance ofsugar configuration or conformation are diffi-cult, partly because of inadequate understand-

BACTERIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



ing of effects of concentration on the equilib-rium conformer composition of sugars inaqueous solutions.As already stated, the severe inhibition

caused by glyceraldehyde and erythrose ispartly irreversible, from which a direct, possiblycovalent reaction between the carbonyl groupand a reactive group on the protein might beassumed. In dilute aqueous solution, glyceral-dehyde equilibrates to a mixture that containsabout 20% free aldehyde. Glyceraldehyde issubstantially more inhibitory than erythrosewhich, at equilibrium, contains about 1% freealdehyde. Ribose contains about 0.01% free al-dehyde, whereas the other sugars used havenegligible proportions (S. J. Angyal, personalcommunication).

Sucrose was the most inhibitory of thesugars giving strictly reversible inhibition.The severity of its inhibition cannot be ex-plained simply on the basis of its molecularweight since another disaccharide, maltose,was much less inhibitory and, in fact, was lessinhibitory than the pentose, arabinose. Thenext most inhibitory sugar was ribose, whichshared with sucrose the property of causing"nonlinear" inhibition, the plot of 1/v versusinhibitor concentration being concave up inboth cases. The nonlinearity of ribose inhibi-tion might be attributed to variations in ano-meric and isomeric composition, which isknown to change with concentration, but thesame explanation cannot be applied to sucrose.This sugar is apparently conformationally sta-ble over a wide range of concentration inaqueous solution (S. Angyal, personal commu-nication).The fructose moiety of sucrose is in the fura-

nose form; ribose, in dilute aqueous solutions,contains about 24% furanose (a + 13) (6). On theother hand, free fructose contains about 31% (3-furanose (50). Fructose was the least inhibitorysugar being roughly equivalent to glycerol.There is no less difficulty in attempting to cor-relate inhibitory properties with the propor-tions of the 1C and C1 forms of the sugars.Other aspects of sugar conformation in relationto inhibition are discussed by Simpson (Ph.D.thesis, University of New South Wales, 1976).

Finally, it cannot be assumed that all en-zymes will respond quantitatively to nonelec-trolytes exactly as described for the yeast isocit-rate dehydrogenase. For example, Heimer (55)reported substantial differences in glycerol in-hibition of a nitrate reductase from D. parva,Chlorella pyrenoidosa, and XD cells oftobacco.In every case, however, glycerol was far lessinhibitory than sodium chloride, the only solutewith which it was systematically compared. On

the other hand, generalizations about nonelec-trolyte action will probably be broadly applica-ble to virtually any enzyme that does not havea specific interaction with any of the solutes.

Regulation of Compatible SoluteAccumulation

The direct proportionality between potas-sium concentration in halophilic bacteria andsodium chloride concentration in their growthmedium can probably be explained physico-chemically on the basis ofmaintaining approxi-mate parity between internal and external wa-ter potential, coupled with the relatively imper-meant nature of Na+. A detailed discussion ofthis mechanism is beyond the scope of this re-view.The eukaryotic polyol content also varies di-

rectly with external solute concentration (in-versely with water activity). In this case, al-though water movement must contribute to thefinal internal polyol concentration, the majorvariable is the actual proportion of polyol of thedry cell mass. This is shown clearly in theresponse ofthe glycerol content ofDunaliella toexternal salt concentration (15). (The polyolstatus of the eukaryotic organelles is notknown. Like bacteria in a concentrated me-dium, the nucleus, mitochondria, and chloro-plasts have no mechanism for maintaining adilute interior against a concentrated cyto-plasm. We must assume that they contain acompatible solute that might or might not bethe same as in the cytoplasm.)

It is thus evident that a control mechanism ofsome kind operates across the plasma mem-brane: the concentration of an intracellular sol-ute regulates the biosynthesis of an intracellu-lar solute. Such a regulatory process has exten-sive implications for cell physiology generallybut, at present, little if anything is known of itsmechanism.The following observations are relevant to a

study and ultimate explanation ofthe control ofcompatible solute production.

(i) A similar response is shown by glycerol inalgae and by polyols in xerotolerant yeasts.

(ii) Internal polyol content varies broadly inresponse to aw, with little effect of the chemicalnature of the external solute on the total polyolaccumulation. Thus, in Dunaliella, glycerol re-sponds to sucrose or sodium chloride (Kessly,personal communication); in S. rouxii, polyolcontent responds to glucose, polyethylene gly-col or salt (22; M. Edgley, personal communica-tion).There is, however, a specific aspect of the

effect of salt on polyol accumulation by S.rouxii. It has long been recognized that S.

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


838 BROWN

rouxii gives increased yields of (extracellular)glycerol in response to increased salt concentra-tion of the growth medium (106). The earlierwork did not distinguish clearly between thetypes of polyols produced; polyol yields werecommonly expressed as "glycerol." Neverthe-less, the basic observation that more glycerol isformed in response to increased salinity is trueand conspicuously includes intracellular polyol.When grown in basal medium, S. rouxii accu-mulates arabitol plus traces of glycerol (22, 29).Increasing the salt concentration of the growthmedium causes an increase of total intracellu-lar polyol, but the increase is entirely attribut-able to glycerol and is, moreover, accompaniedby a slight diminution in arabitol content (M.Edgley, personal communication).

(iii) Dunaliella contains an NADP-linkedglycerol dehydrogenase (13, 15). S. rouxii hasan NADP-dependent ability to dehydrogenateD-arabitol and xylitol as well as an NAD-de-pendent ability to dehydrogenate xylitol, ribi-tol, sorbitol, and mannitol. It is apparently un-able to dehydrogenate glycerol, at least via apyridine nucleotide dehydrogenase, but it candehydrogenate butane-1,2,4-triol and butane-2,3-diol, both with NAD+ (J. R. Simpson, Ph.D.thesis, University of New South Wales, 1976).

(iv) Dunaliella responds to increased saltconcentration in the dark, producing more glyc-erol from accumulated starch (Kessly, personalcommunication).

(v) Dunaliella produces extra glycerol in re-sponse to increased salt concentration underconditions of nitrogen starvation, suggestingthat the regulatory mechanism might not re-quire the induction or repression of an enzyme(Kessly, personal communication).S. rouxii requires aerobic conditions for ara-

bitol production and reverts to an ethanolicfermentation under anaerobic conditions (106).Unlike S. cerevisiae, S. rouxii is not subject tocatabolite repression of its respiratory capabil-ity by high concentrations of glucose in thegrowth medium (23). It is not known whetherthis reflects a fundamental difference in thecontrol of gene expression in the two species orwhether it represents another manifestation ofprotection by the cell's compatible solute. It hasbeen shown, however, that glycerol, at concen-trations up to about 2 M, will substitute forcyclic adenosine 5'-monophosphate and cyclicadenosine 5'-monophosphate receptor by stimu-lating transcription of the gal operon in E. coli(96).

LOOSE ENDSThere are many microorganisms with a sol-

ute tolerance substantially less than those of

BACTERIOL. REv.

the extreme types already discussed. Marinemicroorganisms tolerate about 3.5% (0.6 M) so-dium chloride, plus smaller quantities of othersalts, and there are some bacteria, commonlydescribed as moderately halophilic, which re-quire or tolerate salt at a concentration of 1 to1.5 M or sometimes higher. Furthermore, thegenus Staphylococcus is well known for strainswith a wide range of salt tolerance, although inno sense are these bacteria halophilic.The physiological basis of intermediate sol-

ute (or a,,,) tolerance is not understood, al-though there are some pointers. Moderatelytolerant cells, like their extremely tolerantcounterparts, must adjust thermodynamicallyto reduced a, by sustaining a reduced internala,. This means that those solutes which areretained within the cell will increase in concen-tration and, by definition, be osmotically ac-tive. Furthermore, since enzymes continued tofunction under these conditions, it also followsthat the solutes which are present at such in-creased concentrations are not excessively in-hibitory. If the content (per dry mass of cell, asdistinct from the concentration) of any soluteincreases under these conditions it should besuspected as a possible intermediate type ofcompatible solute.

In fact, there are solutes that accumulate invarious organisms under these conditions.These substances lie within a limited range ofchemical types but occur within quite a widerange of organisms. They accumulate in directresponse to a lowering of external a,,, and, al-though they have not yet been studied gener-ally as enzyme inhibitors, their occurrence un-der these conditions suggests that they do, infact, act as compatible solutes at moderate lev-els of water stress.

Polyols or their derivatives accumulate ineukaryotic protista. Thus, as already described,the marine alga D. tertiolecta accumulatesglycerol, but to a lower concentration than itshalophilic relative, D. viridis. The freshwateralga, Ochromonas malhamensis, produces a-galactosyl glycerol in response to the relativelysmall changes in solute concentration which itcan tolerate (67, 68).

Bacteria, however, do not apparently accu-mulate polyols or carbohydrate derivatives inthis manner. A positive correlation between K+accumulation from a standard low-salt growthmedium and salt tolerance of some 32 bacterialstrains was demonstrated by Christian andWaltho (36). Their experimental conditions in-cluded values for a, down to 0.88 (plus one salt-tolerant coccus that grew at a, 0.84); potassiumcontents ranged between about 200 and 1,050,tmol/g (dry mass). The water content of bacte-

on Novem


mbr.asm

.org/D

ownloaded from



ria growing at 0.90 a,,, is about 0.83 g/g (drymass) (39). Assuming no additional uptake,relative to cell mass, at the low levels ofa,, de-hydration would result in a K+ concentrationwithin the tolerant species of about 1.3 molalwhich, for KCl, is roughly equivalent to 0.95 to0.96 a,,, (see reference 115). It is not knownwhether any supplementary active uptake ofK+ occurs in nonhalophils in response to thelowering of external aw, but Christian andWaltho (39) have reported that the increasedK+ concentration in Staphylococcus aureus islargely the result of dehydration. If that is so,the value of 1.3 molal should be reasonablyaccurate for bacteria grown at the low levels ofa.,, and a substantial concentration of addi-tional solute(s) would be needed to meet thethermodynamic requirements of the situation.Furthermore, although the accumulation ofK+is reminiscent ofthe situation in halophilic bac-teria, we do not know whether K+ acts to anyappreciable extent as a compatible solute forenzymes of salt-tolerant but nonhalophilic bac-teria.Some bacteria accumulate specific metabo-

lites in response to water stress and it is likelythat at least one such compound, proline, mightfunction as a compatible solute. In fact, threeamino acids correlate, like K+, in two wayswith a tolerance of reduced a,. The first corre-lation is between "intrinsic" amino acid contentand inherent or potential tolerance of waterstress. The second correlation is the response ofintracellular amino acid content to changes inaw. There are actually few reports of intrinsi-cally high levels of these amino acids in toler-ant bacteria. I have, however, observed a muchhigher concentration of "pool" aspartate, gluta-mate, and proline in salt-tolerant strains ofStaphylococcus than in nontolerant strains(unpublished data). On the other hand, there isample documentation of responses of theseamino acids to changes in a.. Tempest et al.(135) report that, in continuous culture, suddenincreases in the salinity of the growth mediumcause extensive and rapid increases in the"pool" free glutamate concentration in gram-negative bacteria and similar, but slower, re-sponses in gram-positive bacteria. Whether ornot glutamate functions as a compatible soluteis uncertain. Tempest et al. (135) reported a

glutamate concentration of about 0.1 M inAerobacter aerogenes growing at a dilution rateof 0.3 h-' in 4% sodium chloride. This was thehighest concentration obtained for any aminoacid in A. aerogenes, although slightly higherconcentrations were achieved in some otherbacteria. In fact, their reported concentrationsare almost certainly low by a factor of about 3.

A water content of 4 g/g (dry mass) had beenassumed, whereas under their growth condi-tions a value of about 1.5 g/g is to be expected(see, for example, reference 39; my experienceagrees with Christian and Waltho in this re-spect). Thus, a glutamate concentration ofabout 0.3 M was apparently achieved in bacte-ria growing in 4% (about 0.6 M) sodium chlo-ride. Glutamate could also serve as a counter-ion for K+ and would presumably promote theaccumulation ofthat ion. Glutamate thus madea significant contribution to the overall osmoticbalance ofA. aerogenes and, at a concentrationof 0.3 M, it is unlikely to be a very effectivegeneral enzyme inhibitor.There are also reports of enhanced proline

accumulation by bacteria in response to low-ered a,, (16, 35). Christian and Hall (35) showedthat proline increased linearly in Salmonellaoranienburg with decreasing a,, to reach avalue of 1.5 mmol/g (dry mass) at about 0.95 a,,.This is approximately equivalent to a concen-tration of 1.36 molal.Measures (92) has also demonstrated an ac-

cumulation of glutamate, y-aminobutyrate, orproline in bacteria in response to increased saltconcentration. There was a qualitative trend inthe type ofamino acid accumulation; glutamatepredominated in the least salt-tolerant, whereasproline predominated in the most tolerant bac-teria of the group studied.Prima facie evidence that proline might func-

tion as a compatible solute was provided in 1955by Christian (33, 34) who showed that, in adefined medium, this amino acid was essentialfor the growth ofS. oranienburg at values ofa,,less than 0.97. Cristian and Waltho (40) alsoshowed that proline stimulated the respirationof this and other bacteria at reduced a,. It isrelevant to this interpretation that the use ofglycerol to adjust the a,, of the suspending solu-tion did not appreciably inhibit respirationdown to about 0.96 a., (38).

Proline also accumulates in plants as a partof their response to water stress. Contents ofupto 3.9 mg of proline/g of dry tissue have beenreported for barley plants subjected to -10 to-20 bars of osmotic potential, compared withabout 0.25 mg/g in watered plants (125). Up to 5mg/g was detected in excised leaf laminae ex-posed to polyethylene glycol (molecular weight,4,000), producing an "osmotic potential of -20bars" (126). This is equivalent to a proline con-centration of about 10 mmolal (assuming 4 g ofwater/g of dry tissue), which is negligible bymicrobial standards.

Proline concentration in halophytes can be alittle higher. Stewart and Lee (132) have re-ported that proline in the halophyte Triglochin

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


840 BROWN

maritima is equivalent to 113 umol/g of wettissue or about 10% of the dry mass of shoottissues of plants collected in the field. Together,these figures imply a water content of about 6.7g/g of dry tissue which, if true, suggests a sub-stantial contribution by vascular tissue, withthe implication that the proline was unlikely tohave been distributed uniformly in the mate-rial that was analyzed. Taken at their facevalue, however, the figures imply a mean pro-line concentration of about 17 mmolal, againnegligible by microbial standards. The generaltrend of these results is entirely consistent withthe high a, at which plants, in general, wilt(Table 1).Nevertheless, Stewart and Lee have provided

some enzymological evidence of a possible func-tion of proline as a compatible solute. Up to aconcentration of 0.7 M it did not inhibit gluta-mate dehydrogenase, acetolactate synthase, ni-trate reductase, or glutamine synthetase fromT. maritima. Sodium chloride (0.6 M), on theother hand, caused more than 50% inhibition ofthe enzymes.There are many reports of osmoregulation by

metabolites, frequently amino acids, in a widerange of plants, animals, and microorganisms(e.g., 8, 65, 120, 134, 141, 148). Most, if not all,of the multicellular organisms so consideredhave a very limited range of cellular toleranceofa,. EVen under the relatively mild conditionsthat prevail for the cells of multicellular orga-nisms, however, thermodynamic adjustment("osmoregulation") must occur and solute(s)must concentrate to achieve that adjustment toreduced a,. If salts are inhibitory, as they fre-quently are even at relatively low concentra-tions, then the continued activity of the orga-nism will demand that osmotic balance beachieved by a noninhibitory solute. At concen-trations up to about 0.5 M, few common metab-olites are likely to be general enzyme inhibi-tors, although some would obviously inhibitspecific enzymes very effectively.Thus, amino acids might well act as osmo-

regulators under mild conditions and functionas low-grade compatible solutes. It is not sur-prising that aspartate and glutamate accumu-late in some organisms, since these amino acidsare at the beginning of several biosyntheticpathways and are thus unlikely to have anywidespread function as feedback inhibitors.Proline seems to be in a somewhat differentclass, however, and warrants a reasonably de-tailed enzymological study.

Finally, it is to be expected that there will beoccasional circumstances in which a specificsubstance at low concentration can relievesome of the major effects of water stress in a

way reminiscent of the modification by a sub-strate of effects of high salt concentrations onan enzyme (see section, Halophilic Bacteria). Asubstance acting in this way should not be re-garded as a compatible solute. A possible exam-ple involving whole bacteria lies in some inter-esting work of Avi-Dor and associates. In arecent report (124), it was shown that betaine(0.5 mM) relieved the salt inhibition of growthand respiration of a halotolerant bacterium.Betaine was slowly accumulated (to an intra-cellular concentration of about 800 mM whenthe bacteria were suspended in 2.0 M sodiumchloride), but only extracellular betaine waseffective in relieving salt inhibition. Appar-ently, the action of betaine is on the cell mem-brane.

SurvivalAlthough this review is concerned with mi-

crobial activity, not survival, some brief com-ments on the latter situation might not be outof order. It is a long- and well-established prac-tice to assist the preservation of freeze-driedmicroorganisms by the inclusion of a nonelec-trolyte such as glycerol or a sugar. This practiceapparently reduces mortality during dehydra-tion, storage, and rehydration. It is virtuallycertain that, in this kind of circumstance, non-electrolytes function by directly substituting asa "solvating" molecule for the water that isremoved on dehydration. Hydration of the po-lar sites on organic molecules commonly in-volves hydrogen bonds; removal of the solventcan, and frequently does, lead to the substitu-tion of site-solvent H bonds by site-site Hbonds, which can be intra- or intermolecular.This is exactly what happens in the manufac-ture of a sheet of paper and, presumably, is thesignificance of the use of glucose for obtainingthe three-dimensional 0.07-nm structure of thepurple membrane of H. halobium in the anhy-drous environment of an electron microscope(57).The formation of intra- or intermolecular H

bonds under conditions of excessive dehydra-tion, however, causes irreversible changes insome proteins which lead to enzyme inactiva-tion. The addition of hydroxylated compoundsprovides an alternative source of H bonds thatcan prevent the formation ofthe intersite bondsand thereby prevent the protein inactivationwhich accompanies them.Thus, although there are some superficial

similarities between the function of glycerol,etc., in this context and the role we have pro-posed for compatible solutes, there is an impor-tant basic difference in the mechanism of actionin the two situations. As discussed, our evi-

BACTERIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from



dence suggests that in aqueous solutions com-patible solutes have an unusually low affinityfor enzyme proteins, or at least for those re-gions of the enzyme molecules where bindingwould be inhibitory. The affinity of compatiblesolutes is for solvent water.

SUMMARY

Microbial water stress has been discussedprimarily in relation to three distinctive groupsof microorganisms, namely, the extremely hal-ophilic bacteria, the halophilic algae, and thexerotolerant ("osmophilic") yeasts. Brief refer-ence was also made to other xerotolerant fungi.The halophilic bacteria are distinguished by

an absolute, specific requirement for high con-centrations of sodium chloride. The halophilicalgae (notably Dunaliella), notwithstanding acertain minimal salt requirement, have a phys-iology that is generally much more characteris-tic of a tolerance than of an absolute require-ment for salt. Moreover, the halophilic algaecan be trained to grow outside their normal salttolerance ranges and they can accept replace-ment of a substantial part of the salt require-ment by a nonelectrolyte. In both respects, thehalophilic algae differ fundamentally from thehalophilic bacteria. In the second respect, theyare reminiscent of the xerotolerant yeasts. Thexerotolerant yeasts are generally distinguishedby a remarkable tolerance of low water activity(a,,), although the tolerance range is dependenton the solute used to adjust au.. Occasionally, arequirement for a reduced water activity is en-countered among the yeasts; such a require-ment is usually temperature dependent.The salt requirements of the extremely halo-

philic bacteria are determined by a need for saltpartly to maintain membrane integrity and thefunction of those enzyme systems associatedwith the cell membrane and with ribosomes.Their overall salt relations are also affected bythe ability of some cytoplasmic enzymes to tol-erate salt at intracellular concentrations. Theseenzymes have an optimum in the region of 0.5to 1 M salt. Their function is possible becausethe bacteria accumulate potassium chlorideand effectively exclude sodium chloride. Inmarked contrast to sodium chloride, potassiumchloride is a poor enzyme inhibitor at high con-centrations. Because of this, it protects the rele-vant enzymes against the inhibition that wouldoccur at the same water activity in its absence(that is, if au. were determined by sodium chlo-ride or by common intermediary metabolites).Salt inhibition is discussed in some detail inrelation to an isocitrate dehydrogenase.

Halophilic algae and xerotolerant yeasts do

not apparently produce enzymes with a distinc-tive tolerance of low au. (or increased soluteconcentration). The water relations of these or-ganisms are largely determined by the accumu-lation of polyhydric alcohols to a concentrationcommensurate with extracellular a,. As potas-sium chloride does in halophilic bacteria, thepolyols function, not only as osmoregulators,but also as protectors of enzyme activity. Sub-stances that function in this way have beencalled compatible solutes.The enzymological and physiological signifi-

cance of compatible solutes was discussed. It isevident that enzyme function responds, not tochanges in a, per se, but to the nature andconcentration of the solute used to adjust a,.Nevertheless, a, remains a valuable parameterto describe microbial water relations in a com-plex medium.The role of "intermediate" or "low-grade"

compatible solutes in cells with intermediatelevels of xerotolerance was considered briefly.Perhaps the most notable solute of this type isproline, which accumulates in response to wa-ter stress in some higher plants as well as insome bacteria.

ACKNOWLEDGMENTS

The expansion of the author's research activitiesfrom halophilic bacteria to include eukaryotic micro-organisms was facilitated in the first instance by aresearch grant from the Rural Credits DevelopmentFund of the Reserve Bank of Australia and contin-ued with support from the Nuffield Foundation(Australia). The author wishes to thank Lesley J.Borowitzka for many helpful comments and forbringing to his attention a number of valuable pub-lications. Thanks are also due to J. I. Pitt for thesubstance of Table 1.

ADDENDUM

P. S. Low and G. N. Somero (Proc. Natl. Acad.Sci. U.S.A. 72:3305-3309, 1975) have extended theirevidence and argue that salts affect the velocities ofenzyme-catalyzed reactions by changing the activa-tion volume of an enzyme-substrate complex (Seesection, Compatible Solutes and reference 84). Theydemonstrated a linear relation between activationvolume and reaction rate, but pointed out that therelation is not causal. Their interpretation supple-ments the comments made in Compatible Solutes onthe role of water. It is important to note, however,that the effects reported by Low and Somero areconfined to salt concentrations less than about 300mM.

LITERATURE CITED

1. Aitken, D. M., and A. D. Brown. 1969. Citrateand glyoxylate cycles in the halophil, Halo-bacterium salinarium. Biochim. Biophys.

VOL. 40, 1976

on Novem


mbr.asm

.org/D

ownloaded from


842 BROWN

Acta 177:351-354.2. Aitken, D. M., and A. D. Brown. 1972. Proper-

ties of halophil nicotinamide-adenine dinu-cleotide phosphate-specific isocitrate dehy-drogenase. Biochem. J. 130:645-662.

3. Aitken, D. M., A. J. Wicken, and A. D. Brown.1970. Properties of a halophil nicotinamide-adenine dinucleotide phosphate-specific iso-citrate dehydrogenase. Biochem. J. 116:125-134.

4. Akerlof, G. 1932. Dielectric constant of someorganic solvent-water mixtures at varioustemperatures. J. Am. Chem. Soc. 54:4125-4139.

5. Anand, J. C., and A. D. Brown. 1968. Growthrate patterns of the so-called osmophilic andnon-osmophilic yeasts in solutions of poly-ethylene glycol. J. Gen. Microbiol. 52:205-212.

6. Angyal, S. J. 1969. The composition and confor-mation of sugars in solution. Angew. Chem.Int. Ed. Eng. 8:157-166.

7. Baas-Becking, L. G. M. 1931. Historical noteson salt and salt manufacture. Sci. Monthly32:434-446.

8. Baginski, R. M., and S. K. Pierce. 1975. An-aerobiosis: a possible source of osmotic solutefor high-salinity acclimation in marine mol-luscs. J. Exp. Biol. 62:589-598.

9. Baxter, R. M. 1959. An interpretation of theeffects of salts on the lactic dehydrogenase ofHalobacterium salinarium. Can. J. Micro-biol. 5:47-57.

10. Bayly, I. A. E. 1972. Salinity tolerance and os-motic behaviour of animals in athalassic sa-line and marine hypersaline waters. Annu.Rev. Ecol. Syst. 3:233-267.

11. Ben-Amotz, A. 1975. Adaptation of the unicel-lular alga Dunaliella parva to a saline envi-ronment. J. Phycol. 11:50-54.

12. Ben-Amotz, A., and M. Avron. 1972. Photosyn-thetic activities of the halophilic alga Duna-liella parva. Plant Physiol. 49:240-243.

13. Ben-Amotz, A., and M. Avron. 1973. NADPspecific dihydroxyacetone reductase fromDunaliella parva. FEBS Lett. 29:153-155.

14. Ben-Amotz, A., and M. Avron. 1973. The role ofglycerol in the osmotic regulation of the hal-ophilic alga Dunaliella parva. Plant Physiol.51:875-878.

15. Borowitzka, L. J., and A. D. Brown. 1974. Thesalt relations of marine and halophilic spe-cies of the unicellular green alga, Duna-liella. The role of glycerol as a compatiblesolute. Arch. Microbiol. 96:37-52.

16. Britten, R. J., and F. T. McClure. 1962. Theamino acid pool in Escherichia coli. Bacte-riol. Rev. 26:292-335.

17. Brock, T. D. 1975. Salinity and the ecology ofDunaliella from Great Salt Lake. J. Gen.Microbiol. 89:285-292.

18. Brown, A. D. 1963. The peripheral structuresof Gram-negative bacteria. IV. The cation-sensitive dissolution of the cell membrane of

the halophilic bacterium, Halobacteriumhalobium. Biochim. Biophys. Acta 75:425-435.

19. Brown, A. D. 1964. Aspects of bacterial re-sponse to the ionic environment. Bacteriol.Rev. 28:296-329.

20. Brown, A. D. 1964. The development of halo-philic properties in bacterial membranes byacylation. Biochim. Biophys. Acta 93:136-142.

21. Brown, A. D. 1965. Hydrogen ion titrations ofintact and dissolved lipoprotein membranes.J. Mol. Biol. 12:491-508.

22. Brown, A. D. 1974. Microbial water relations:features of the intracellular composition ofsugar-tolerant yeasts. J. Bacteriol. 118:769-777.

23. Brown, A. D. 1975. Microbial water relations.Effects of solute concentration on the respi-ratory activity of sugar-tolerant and non-tol-erant yeasts. J. Gen. Microbiol. 86:241-249.

24. Brown, A. D., and K. Y. Cho. 1970. The wallsof extremely halophilic cocci. Gram-positivebacteria lacking muramic acid. J. Gen. Mi-crobiol. 62:267-270.

25. Brown, A. D., and A. Netschey. 1967. Sedimen-tation, viscosity and partial specific volumesof membrane proteins and lipoproteins. Bio-chem. J. 103:24-28.

26. Brown, A. D., and R. F. Pearce. 1969. Prelimi-nary fractionation by gel electrophoresis ofthe membrane proteins and lipoproteins ofHalobacterium halobium. Can. J. Biochem.47:833-837.

27. Brown, A. D., and C. D. Shorey. 1963. The cellenvelopes of two extremely halophilic bacte-ria. J. Cell Biol. 18:681-689.

28. Brown, A. D., C. D. Shorey, and H. P. Turner.1965. An alternative method of isolating themembrane of a halophilic bacterium. J. Gen.Microbiol. 41:225-231.

29. Brown, A. D., and J. R. Simpson. 1972. Waterrelations of sugar-tolerant yeasts: the role ofintracellular polyols. J. Gen. Microbiol.72:589-591.

30. Bull, H. B., and K. Breese. 1970. Water andsolute binding by proteins. I. Electrolytes.Arch. Biochem. Biophys. 137:299-305.

31. Cheah, K. S. 1970. The membrane-bound car-bon monoxide-reactive haemoproteins in theextreme halophils. Biochim. Biophys. Acta197:84-86.

32. Cho, K. Y., C. H. Doy, and E. H. Mercer. 1967.Ultrastructure of the obligate halophilic bac-terium Halobacterium halobium. J. Bacte-riol. 94:196-201.

33. Christian, J. H. B. 1955. The influence of nutri-tion on the water relations of Salmonellaoranienburg. Aust. J. Biol. Sci. 8:75-82.

34. Christian, J. H. B. 1955. The water relations ofgrowth and respiration of Salmonella oran-ienburg at 30°C. Aust. J. Biol. Sci. 8:490-497.

35. Christian, J. H. B., and J. M. Hall. 1972. Waterrelations of Salmonella oranienburg: accu-

BACTERRIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from


VOL. 40, 1976

mulation of potassium and amino acids dur-ing respiration. J. Gen. Microbiol. 70:497-506.

36. Christian, J. H. B., and J. A. Waltho. 1961.The sodium and potassium content of non-halophilic bacteria in relation to salt toler-ance. J. Gen. Microbiol. 25:97-102.

37. Christian, J. H. B., and J. A. Waltho. 1962.Solute concentrations within cells of halo-philic and non-halophilic bacteria. Biochim.Biophys. Acta 65:506-508.

38. Christian, J. H. B., and J. A. Waltho. 1962.The water relations of staphylococci and mi-crococci. J. Appl. Bacteriol. 25:369-377.

39. Christian, J. H. B., and J. A. Waltho. 1964.The composition of Staphylococcus aureus inrelation to the water activity of the growthmedium. J. Gen. Microbiol. 35:205-213.

40. Christian, J. H. B., and J. A. Waltho. 1966.Water relations of Salmonella oranienburg;stimulation of respiration by amino acids. J.Gen. Microbiol. 43:345-355.

41. Contaxis, C. C., and F. J. Reithel. 1971. Stud-ies on protein multimers. II. Mechanism ofurease dissociation in 1,2-propanediol: com-parative studies with ethylene glycol andglycerol. J. Biol. Chem. 246:677-685.

42. Cooke, R., and I. D. Kuntz. 1974. The proper-ties of water in biological systems. Annu.Rev. Biophys. Bioeng. 3:95-126.

43. Craigie, J. S., and J. McLachlan. 1964. Glyc-erol as a photosynthetic product in Duna-liella tertiolecta Butcher. Can. J. Bot. 42:777-778.

44. Danon, A., and W. Stoeckenius. 1974. Photo-phosphorylation in Halobacterium halo-bium. Proc. Natl. Acad. Sci. U.S.A. 71:1234-1238.

45. Diamond, J. M., and E. M. Wright. 1969. Bio-logical membranes: the physical basis of ionand nonelectrolyte selectivity. Annu. Rev.Physiol. 31:581-646.

46. Douzou, P. 1974. The use of subzero tempera-tures in biochemistry, p. 401-512. In D. Glick(ed), Methods of biochemical analysis, vol.22. John Wiley & Sons, Inc., New York.

47. Dundas, I. D., and H. Larsen. 1962. The physi-ological role of the carotenoid pigments ofHalobacterium salinarium. Arch. Mikrobiol.44:233-239.

48. Dundas, I. D., V. R. Srinivasan, and H. 0.Halvorson. 1963. A chemically defined me-dium for Halobacterium salinarium strain 1.Can. J. Microbiol. 9:619-624.

49. Eddy, B. P. (ed.). 1958. The microbiology offish and meat curing brines. Her Majesty'sStationer's Office, London.

50. Eliel, E. L., N. L. Allinger, S. J. Angyal, andG. A. Morrison. 1965. Conformational analy-sis. Interscience Publishers, New York.

51. Gatlin, L. L. 1972. Information theory and theliving system. Columbia University Press,New York.

52. Ginzburg, M. 1969. The unusual membrane


permeability oftwo halophilic unicellular or-ganisms. Biochim. Biophys. Acta 173:370-376.

53. Gochnauer, M. B., and D. J. Kushner. 1969.Growth and nutrition of extremely halo-philic bacteria. Can. J. Microbiol. 15:1157-1165.

54. Gochnauer, M. B., G. G. Leppard, P. Komara-tat, M. Kates, T. Novitsky, and D. J. Kush-ner. 1975. Isolation and characterization ofActinopolyspora halophila, gen. et sp. nov.,an extremely halophilic actinomycete. Can.J. Microbiol. 21:1500-1511.

55. Heimer, Y. M. 1973. The effects of sodium chlo-ride; potassium chloride and glycerol on theactivity of nitrate reductase of a salt-tolerantand two non-tolerant plants. Planta 113:279-281.

56. Hellingwerf, K. J., P. A. M. Michels, J. W.Dorpema, and W. N. Konings. 1975. Trans-port of amino acids in membrane vesicles ofRhodopseudomonas spheroides energized byrespiratory and cyclic electron flow. Eur. J.Biochem. 55:397-406.

57. Henderson, R., and P. N. T. Unwin. 1975.Three-dimensional model of purple mem-brane obtained by electron microsocpy. Na-ture (London) 257:28-32.

58. Hescox, M. A., and D. M. Carlberg. 1972. Pho-toreactivation in Halobacterium cutirubrum.Can. J. Microbiol. 18:981-985.

59. Hildebrand, E., and N. Dencher. 1975. Twophotosystems controlling behavioural re-sponses of Halobacterium halobium. Nature(London) 257:46-48.

60. Houwink, A. L. 1956. Flagella, gas vacuolesand cell-wall structure in Halobacteriumhalobium; an electron-microscope study. J.Gen. Microbiol. 15:146-150.

61. Hsiao, T. C. 1973. Plant responses to waterstress. Annu. Rev. Plant Physiol. 24:519-570.

62. Ingram, M. 1957. Micro-organisms resistinghigh concentrations of sugars or salts, p. 90-133. In R. E. 0. Williams and C. C. Spier(ed.), Microbiol ecology. Cambridge Univer-sity Press, Cambridge.

63. Johnson, M. K., E. J. Johnson, R. D. Mac-Elroy, H. L. Speer, and B. S. Bruff. 1968.Effects of salts on the halophilic alga Duna-liella viridis. J. Bacteriol. 95:1461-1468.

64. Joshi, J. G., W. R. Guild, and P. Handler.1963. The presence of two species of DNA insome halobacteria. J. Mol. Biol. 6:34-38.

65. Kaneshiro, E. S., G. G. Holz, and P. B. Dur-ham. 1969. Osmoregulation in a marine cil-iate, Miamiensis avidus. H. Regulation ofintracellular free amino acids. Biol. Bull.137:161-169.

66. Kates, M. 1972. Ether-linked lipids, p. 351-398.In F. Snyder (ed.), Ether lipids. Chemistryand biology. Academic Press Inc., New York.

67. Kauss, H. 1967. Isofloridosid und Osmoregula-tion bei Ochromonas malhamensis. Z. Pflan-

on Novem


mbr.asm

.org/D

ownloaded from


844 BROWN

zenphysiol. 56:453-465.68. Kauss, H. 1973. Turnover of galactosyl glycerol

and osmotic balance in Ochromonas. PlantPhysiol. 52:613-615.

69. Kientz, M. L., and C. C. Bigelow. 1966. Theeffect of ethylene glycol on the structure of f3-lactoglobulin. Biochemistry 5:3494-3500.

70. Kushner, D. J. 1968. Halophilic bacteria. Adv.Appl. Microbiol. 10:73-99.

71. Langworthy, T. A., W. R. Mayberry, and P. F.Smith. 1974. Long-chain glycerol diether andpolyol dialkyl glycerol triether lipids of Sul-folobus acidocaldarius. J. Bacteriol. 119:106-116.

72. Langworthy, T. A., P. F. Smith, and W. R.Mayberry. 1972. Lipids of Thermoplasmaacidophilum. J. Bacteriol. 112:1193-1200.

73. Lanyi, J. K. 1971. Studies of the electron trans-port chain of extremely halophilic bacteria.VI. Salt-dependent dissolution of the cell en-velope. J. Biol. Chem. 246:4552-4559.

74. Lanyi, J. K. 1974. Salt-dependent properties ofproteins from extremely halophilic bacteria.Bacteriol. Rev. 38:272-290.

75. Larsen, H. 1963. Halophilism, p. 297-342. In I.C. Gunsalus and R. Y. Stanier (ed), Thebacteria, vol. 4. Academic Press Inc., NewYork.

76. Larsen, H. 1967. Biochemical aspects of ex-treme halophilism. Adv. Microbiol. Physiol.1:97-132.

77. Latorella, A. H., and R. L. Vadas. 1973. Salin-ity adaptation by Dunaliella tertiolecta. I.Increases in carbonic anhydrase activity andevidence for a light-dependent Na+/H+ ex-change. J. Phycol. 9:273-277.

78. Lewis, D. H., and D. C. Smith. 1967. Sugaralcohols (polyols) in fungi and green plants.I. Distribution, physiology and metabolism.New Phytol. 66:143-184.

79. Louis, B. G., and P. S. Fitt. 1971. Nucleic acidenzymology of extremely halophilic bacteria.Halobacterium cutirubrum deoxyribonucleicacid polymerase. Biochem. J. 121:621-627.

80. Louis, B. G., and P. S. Fitt. 1971. Nucleic acidenzymology of extremely halophilic bacteria.Halobacterium cutirubrum ribonucleic acid-dependent ribonucleic acid polymerase. Bio-chem. J. 121:629-633.

81. Louis, B. G., and P. S. Fitt. 1972. Isolation andproperties of highly purified Halobacteriumcutirubrum deoxyribonucleic acid-dependentribonucleic acid polymerase. Biochem. J.127:69-80.

82. Louis, B. G., and P. S. Fitt. 1972. The role ofHalobacterium cutirubrum deoxyribonucleicacid-dependent ribonucleic acid polymerasesubunits in initiation and polymerization.Biochem. J. 127:81-86.

83. Louis, B. G., and P. S. Fitt. 1972. Purificationand properties of the ribonucleic acid-de-pendent ribonucleic acid polymerase fromHalobacterium cutirubrum. Biochem. J.128:755-762.

84. Low, P. S., and G. N. Somero. 1975. Activationvolumes in enzymic catalysis: their sourcesand modification by low-molecular-weightsolutes. Proc. Natl. Acad. Sci. U.S.A.72:3014-3018.

85. MacDonald, R. E., and J. K. Lanyi. 1975.Light-induced leucine transport in Halobac-terium halobium envelope vesicles: a chem-iosmotic system. Biochemistry 14:2882-2889.

86. McLachlan, J. 1960. The culture ofDunaliellatertiolecta Butcher-a euryhaline organism.Can. J. Microbiol. 6:367-379.

87. Marre, E., 0. Servettaz, and F. Albergoni.1958. Sul meccanismo di adattamento a can-dizioni osmotiche estreme in Dunaliella sal-ina. I. Reazioni fisiologiche a variazionidell'ambiente osmotico. Atti. Accad. Naz.Lincei Rend. Cl. Sci. Fis. Mat. Natur. Ser. 825:567-575.

88. Marre, E., and 0. Servettaz. 1959. Sul meccan-ismo di adattamento a candizioni osmoticheestreme in Dunaliella salina. II. Rapportofra concentrazioni del mezzo esterno e com-posizione del succo cellulare. Atti. Acad.Naz. Lincei Rend. Cl. Sci. Fis. Mat. Natur.Ser. 8 26:272-278.

89. Marshall, C. L., and A. D. Brown. 1968. Themembrane lipids of Halobacterium halo-bium. Biochem. J. 110:441-448.

90. Marshall, C. L., A. J. Wicken, and A. D.Brown. 1969. The outer layer of the cell en-velope of Halobacterium halobium. Can. J.Biochem. 47:71-74.

91. Marshall, K. C. 1964. Survival of root-nodulebacteria in dry soil exposed to high temper-atures. Aust. J. Agric. Res. 15:273-281.

92. Measures, J. C. 1975. Role of amino acids inosmoregulation of non-halophilic bacteria.Nature (London) 257:398-400.

93. Mescher, M. F., and J. L. Strominger. 1975.Bacitracin induces sphere formation in Halo-bacterium species which lack a wall peptido-glycan. J. Gen. Microbiol. 89:375-378.

94. Meyer, G. H., M. B. Morrow, 0. Wyss, T.Berg, and J. L. Littlepage. 1962. Antartica:the microbiology of an unfrozen saline pond.Science 138:1103-1104.

95. Moore, R. L., and B. J. McCarthy. 1969. Char-acterization of the deoxyribonucleic acid ofvarious strains of halophilic bacteria. J. Bac-teriol. 99:248-254.

96. Nakanishi, S., S. Adhya, M. Gottesman, and I.Pastan. 1974. Activation of transcription atspecific promoters by glycerol. J. Biol.Chem. 249:4050-4056.

97. Nobel, P. S. 1970. Introduction to biophysicalplant physiology. W. H. Freeman and Co.,San Francisco.

98. Nozaki, Y., and C. Tanford. 1965. The solubil-ity of amino acids and related compounds inaqueous ethylene glycol solutions. J. Biol.Chem. 240:3568-3573.

99. Oakenfull, D. G. 1975. Temperature and pres-sure as a means of assessing the significance

BACTERRIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from


VOL. 40, 1976

of hydrophobic interactions in determiningprotein structure. Proc. Aust. Biochem. Soc.8:23.

100. Oesterhelt, D. 1975. The purple membrane ofHalobacterium halobium: a new system forlight energy conversion, p. 147-165. In CIBAFoundation Symposium 3; Energy transfor-mation in biological systems. Elsevier, Am-sterdam.

101. Oesterhelt, D., and W. Stoeckenius. 1971. Arhodopsin-like protein from the purple mem-brane of Halobacterium halobium. Nature(London) New Biol. 233:149-152.

102. Okamoto, H. 1962. Interaction between cellsurface and several types of solutes with spe-cial reference to the osmotic regulationmechanism in a moderately halophilic strainofChlamydomonas. Z. Allg. Mikrobiol. 2:32-44.

103. Okamoto, H., and Y. Suzuki. 1964. Intracellu-lar concentration of ions in a halophilicstrain of Chlamydomonas. I. Concentrationof Na+, K+ and Cl- in the cell. Z. Allg. Mik-robiol. 4:350-357.

104. Olivares, W., and D. A. McQuarrie. 1975. Onthe theory of ionic solutions. Biophys. J.15:143-162.

105. Onishi, H. 1960. Studies on osmophilic yeasts.Part IX. Isolation of a new obligate halo-philic yeast and some consideration on halo-philism. Bull. Agric. Chem. Soc. Jpn.24:226-230.

106. Onishi, H. 1963. Osmophilic yeasts. Adv. FoodRes. 12:53-94.

107. Onishi, H., M. E. McCance, and N. E. Gib-bons. 1965. A synthetic medium for ex-tremely halophilic bacteria. Can. J. Micro-biol. 11:365-373.

108. Pitt, J. I. 1975. Xerophilic fungi and the spoil-age of foods of plant origin, p. 273-307. In R.B. Duckworth (ed.), Water relations of foods.Academic Press Inc., London.

109. Pitt, J. I., and J. H. B. Christian. 1968. Waterrelations of xerophilic fungi isolated fromprunes. Appl. Microbiol. 16:1853-1858.

110. Postgate, J. R. 1965. Recent advances in thestudy ofthe sulfate reducing bacteria. Bacte-riol. Rev. 29:425-441.

111. Pugh, E. L., M. K. Wassef, and M. Kates.1971. Inhibition of fatty acid synthetase inHalobacterium cutirubrum and Escherichiacoli by high salt concentrations. Can. J. Bio-chem. 49:953-958.

112. Racker, E., and W. Stoeckenius. 1974. Recon-stitution of purple membrane vesicles cata-lyzing light-driven proton uptake and adeno-sine triphosphate formation. J. Biol. Chem.249:662-663.

113. Raymond, J. C., and W. R. Sistrom. 1969. Ec-tothiorhodospira halophila: a new species ofthe genus Ectothiorhodospira. Arch. Mikro-biol. 69:121-126.

114. Reeves, H. C., B. A. Brehmeyer, and S. J. Ajl.1968. Multiple forms of bacterial NADP-spe-


cific isocitrate dehydrogenase. Science162:359-360.

115. Robinson, R. A., and R. H. Stokes. 1965. Elec-trolyte solutions, 2nd ed. Butterworths, Lon-don.

116. Rosseinsky, D. R. 1965. Electrode potentialsand hydration energies. Theories and corre-lations. Chem. Rev. 65:467-490.

117. Rothfield, L., and D. Romeo. 1971. Role oflipids in the biosynthesis of the bacterial cellenvelope. Bacteriol. Rev. 35:14-38.

118. Rowe, J. J., and H. C. Reeves. 1971. Electro-phoretic heterogeneity of bacterial nicotina-mide adenine dinucleotide phosphate-spe-cific isocitrate dehydrogenases. J. Bacteriol.108:824-827.

119. Scarr, M. P., and D. Rose. 1966. Study of osmo-philic yeasts producing invertase. J. Gen.Microbiol. 45:9-16.

120. Schobert, B. 1974. The influence of waterstress on the metabolism of diatoms. I. Os-motic resistance and proline accumulation inCyclotella meneghiniana. Z. Pflanzenphy-siol. 74:106-120.

121. Scott, W. J. 1957. Water relations of food spoil-age microorganisms. Adv. Food Res. 7:83-127.

122. Sehgal, S. N., M. Kates, and N. E. Gibbons.1962. Lipids of Halobacterium cutirubrum.Can. J. Biochem. Physiol. 40:69-81.

123. Self, C. H., and D. J. Weitzmann. 1972. Theisocitrate dehydrogenases of AcinetobacterIwoffi. Separation and properties of the twonicotinamide-adenine dinucleotide phos-phate-linked isoenzymes. Biochem. J.130:211-219.

124. Shkedy-Vinkler, C., and Y. Avi-Dor. 1975. Be-taine-induced stimulation of respiration athigh osmolarities in a halotolerant bacte-rium. Biochem. J. 150:219-226.

125. Singh, T. N., L. G. Paleg, and D. Aspinall.1973. Stress metabolism. I. Nitrogen metabo-lism and growth in the barley plant duringwater stress. Aust. J. Biol. Sci. 26:45-56.

126. Singh, T. N., D. Aspinall, L. G. Paleg, and S.F. Boggess. 1973. Stress metabolism. II.Changes in proline concentration in excisedplant tissues. Aust. J. Biol. Sci. 26:57-63.

127. Slatyer, R. 0. 1967. Plant-water relationships.Academic Press Inc., London.

128. Smith, P. F., T. A. Langworthy, W. R. May-berry, and A. E. Hougland. 1973. Character-ization of the membranes of Thermoplasmaacidophilum. J. Bacteriol. 116:1019-1028.

129. Snyder, F. (ed.). 1972. Ether lipids. Chemistryand biology. Academic Press Inc., New York.

130. Spencer, J. F. T. 1968. Production ofpolyhydricalcohols by yeasts, p. 1-42. In D. J. D. Hock-enhull (ed.), Progress in industrial microbi-ology, vol. 7. J. and A. Churchill Ltd, Lon-don.

131. Steber, J., and K. H. Schleifer. 1975. Halococ-cus morrhuae: a sulfated heteropolysacchar-ide as the structural component of the bacte-

on Novem


mbr.asm

.org/D

ownloaded from


846 BROWN

rial cell wall. Arch. Microbiol. 105:173-177.132. Stewart, G. R., and J. A. Lee. 1974. The role of

proline accumulation in halophytes. Planta120:279-289.

133. Stoeckenius, W., and R. Rowen. 1967. A mor-phological study ofHalobacterium halobiumand its lysis in media of low salt concentra-tion. J. Cell Biol. 34:365-393.

134. Stoner, L. C., and P. B. Dunham. 1970. Regu-lation of cellular osmolarity and volume inTetrahymena. J. Exp. Biol. 53:391-399.

135. Tempest, D. W., J. L. Meers, and C. M. Brown.1970. Influence of environment on the con-tent and composition of microbial free aminoacid pools. J. Gen. Microbiol. 64:171-185.

136. Thierry, 0. C., and J. J. Cooney. 1966. Physico-chemical factors influencing growth and pig-ment synthesis by Micrococcus roseus. Can.J. Microbiol. 12:691-698.

137. Timasheff, S. N. 1970. Polyelectrolyte proper-ties of globular proteins, p. 1-64. In A. Veiss(ed.), Biological polyelectrolytes. MarcellDekker Inc., New York.

138. Torsvik, T., and I. D. Dundas. 1974. Bacterio-phage of Halobacterium salinarium. Nature(London) 248:680-681.

139. Trezzi, M., G. Galli, and E. Bellini. 1965. Theresistance of Dunaliella salina to osmoticstresses. A. Bot. Ital. 72:255-263.

140. Van Auken, 0. W., and I. B. McNulty. 1968.Some growth characteristics of a halophilicChlamydomonas. Proc. Utah Acad. Sci.45:312-313.

141. Virkar, R. A., and K. L. Webb. 1970. Free

amino acid composition of the soft-shellclam, Mya arenaria in relation to salinity ofthe medium. Comp. Biochem. Physiol.32:775-783.

142. Volcani, B. E. 1944. The microorganisms oftheDead Sea, p. 71-85. In Papers collected tocommemorate the 70th anniversary of Dr.Chaim Weizmann. Daniel SieffResearch In-stitute, Rehovoth, Israel.

143. Wais, A. C., M. Kon, R. E. MacDonald, and B.D. Stollar. 1975. Salt-dependent bacterio-phage infecting Halobacterium cutirubrumand H. halobium. Nature (London) 256:314-315.

144. Walker, D. A. 1974. Chloroplast and cell, p. 1-49. In D. H. Northcote (ed.), M. T. P. inter-national review of the science of biochem.,series vol. 11. Butterworths, London.

145. Walsby, A. E. 1972. Structure and function ofgas vacuoles. Bacteriol. Rev. 36:1-32.

146. Wegmann, K. 1969. On the pathways of CO2fixation in Dunaliella. Progr. Photosyn. Res.3:1559-1564.

147. Wegmann, K. 1971. Osmotic regulation of pho-tosynthetic glycerol production in Duna-liella. Biochim. Biophys. Acta 234:317-323.

148. Willbrandt, W. 1963. Transport through bio-logical membranes. Annu. Rev. Physiol.25:601-630.

149. Woodrow, G. C., H. T. Cheung, and K. Y. Cho.1973. Phospholipid of an extremely halo-philic bacterium, Sarcina morrhuae. Aust.J. Biol. Sci. 26:787-792.

BACTrBRIOL. REV.

on Novem


mbr.asm

.org/D

ownloaded from


microbial water stress806 brown solute value to,t requires an absolute value for le that is not...

Documents