vol. in u.s.a. structure and function of vacuolesalgal suspensions suggestedto himthat the gas was...

BACTEMIOLOGIcAL REVIEws, Mar. 1972, p. 1-32Copyright 0 1972 American Society for Microbiology

Vol. 36, No. 1Printed in U.S.A.

Structure and Function of Gas VacuolesA. E. WALSBY

Department of Botany, Westfield College, London, United Kingdom, and Department of Botany,University of California, Berkeley, California 94720'

INTRODUCTION ............................................................. 1

GAS VESICLES .............................................................. 2STRUCTURE OF THE MEMBRANE ....... .................................. 5Membrane Thickness........................................................ 5Rib Spacing ................................................................ 6Subunit Structure ......................................................... 7

CHEMISTRY OF THE MEMBRANE ....... ................................... 8Methods of Isolation ....................................................... 8

Cell lysis ................................................................. 8Centrifugation............................................................ 9

Chemical Analysis ......................................................... 9Relationship of Membrane Structure and Protein Component .... ............. 11

Implications of the Subunit Model ....... .................................... 12PHYSICAL PROPERTIES OF THE MEMBRANE ..... ........................ 13Permeability to Gas ....................................................... 13Rigidity.................................................................... 14

Surface Properties.......................................................... 17FORMATION AND SHAPE OF GAS VESICLES ..... ......................... 17FUNCTIONS OF GAS VACUOLES .......................................... 19Providing Buoyancy ...................................... 19Buoyancy in the halobacteria ...................................... 21In other heterotrophic bacteria ...................................... 23In blue-green algae ...... ................................ 23

Regulating Buoyancy ...... ................................ 24In blue-green algae....................................... 24In photosynthetic bacteria ........... ........................... 25In heterotrophic bacteria .......... ............................ 25

Light Shielding ...................................... 25Light shielding in suspensions en masse .................................... 28

Increasing Cell Surface Area to Volume Ratio................................ 28CONCLUDING REMARKS ........ .............................. 28LITERATURE CITED ...................................... 29

INTRODUCTION

Although many types of cells grow in directcontact with the atmosphere and may dependon the air for their supply of carbon, oxygen,and nitrogen, gas-filled spaces are rarely foundinside the protoplasm of living cells. There are,however, certain prokaryotic organisms whichpossess gas-filled structures of a more or lesspermanent nature. These structures are knownas gas vacuoles. They were first reported to-wards the end of the last century as occurringin a few species of bacteria (50) and blue-greenalgae (22). Their bright, refractile qualities andreddish appearance under the light microscopeseemed to distinguish them from other gran-ules which had been observed in these organ-isms, though for some time the nature of theircontents remained a matter of dispute, and a

I Present address.

variety of synonyms [Hihlungen (117) or hollowbodies, aerosomes (67), pseudovacuoles (59,63), and Schwebekbrperchen (67) or flotationbodies] were used to describe them. Evidencesupporting the idea that they contained gaswas first provided in 1895 by three Germanmicrobiologists, Ahlborn (1), Klebahn (45), andStrodtmann (91), who showed, in the now clas-sical "Hammer, Cork and Bottle" experiment,that gas vacuoles could be made to disappearon the application of a moderate pressure.When a stout bottle was filled to the brim witha suspension of gas-vacuolate algae, a corkapplied and struck sharply with a hammer, thepressure generated on the suspension causedthe immediate disappearance of the structures.This was indicated by the sudden decrease inturbidity of the suspension, and confirmed bysubsequent microscope examination. If thealga had initially been in a floating condition,as was usually the case when gas vacuoles were

1

on June 7, 2020 by guesthttp://m

mbr.asm

.org/D

ownloaded from

http://mmbr.asm.org/

BACTERIOL. REV.

abundantly present in the cells, then loss ofbuoyancy invariably accompanied the disap-pearance (see Fig. 1 and 13).In a series of more detailed experiments that

he carried out some years later with highlyconcentrated suspensions of alga gatheredfrom waterblooms, Klebahn (46-48) demon-strated that the application of pressure re-sulted in a permanent volume decrease whichhe took to be equal to the volume of vacuoleswhich had been destroyed. By comparing thischange with the concomitant increase in spe-cific gravity of the algal material, he concludedthat the content of the gas vacuoles must havebeen of very low density, similar to that of agas, and he pointed out that the compressi-bility of the structures also indicated gaseouscontents. Analyses which Klebahn performedon the gas he collected from gas-vacuolatealgal suspensions suggested to him that the gaswas probably nitrogen. However, while theresults of his experiments are substantiated byrecent investigations into the nature and originof the vacuole gas (105), they now require freshinterpretation (see below).

It was apparent to Klebahn that gas vac-uoles differed from simple gas bubbles in sev-eral respects. Firstly, he observed that theywere rarely spherical and usually had an irreg-ular outline. Secondly, the pressure of the gasthey contained was, he estimated, substan-tially less than would exist in bubbles of com-parable size. And thirdly, they persisted indef-initely in material placed under a vacuum for

prolonged periods of time. For these reasons,he proposed that the vacuole gas must be en-closed in a membrane (46) whose rigidity pro-vided protection from the effects of surfacetension and a certain degree of externally ap-plied pressure (a concept which still stands),and whose impermeability prevented the gasfrom diffusing away (an idea which is nolonger supported, see below).

Further details of Klebahn's experiments,and of other contemporary physiologicalstudies on gas vacuoles in blue-green algae,can be found in the review made by Fogg (22)in 1941. With the many uncertainties left bythese studies, he was unable to reach any firmconclusion as to the functional significance ofgas vacuoles in these organisms. This presentreview takes up the story where he left it andattempts to see whether the information thathas since been acquired on the structure, com-position, physical properties, and formation ofgas vacuoles in blue-green algae and bacteriaprovides support for any of the functions withwhich these curious structures have been at-tributed.

GAS VESICLESKlebahn's suspicions on the existence of a

membrane were confirmed in a rather unex-pected way by electron microscope studiesthat were carried out many years later. In 1965Bowen and Jensen (4, 5), using thin-sectioningtechniques, found that, in five species of blue-green algae they investigated, the gas vacuoles

a bFIG. 1. The "Hammer, Cork and Bottle" experiment, showing two results of collapsing the gas vacuoles in

the blue-green alga Microcystis aeruginosa. (a) The decrease in turbidity after striking the cork in the left-hand bottle (the bottle on the right is untreated). (b) The loss of buoyancy apparent in the treated sampleafter allowing it to stand for a couple of hours.

2 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

were made up of stacks of cylindrical, elec-tron-transparent structures which they termed"gas vesicles." Hollow, and bounded by singlemembranes only 2 nm thick, the vesicles werereported to have the form of cylinders, of con-stant diameter (75 nm) but variable length(200 to 1,000 nm), with conical ends. Bowenand Jensen could not detect these cylindricalstructures in algae which had been subjectedto a pressure sufficient to destroy their gasvacuoles, and they correctly concluded thatthe vesicles had collapsed flat to the double,membranous elements which they found intheir place.

Simultaneously, Jost (40), investigating thefine structure of another blue-green alga, Os-cillatoria rubescens, by the newly-developedfreeze-etching technique, produced very finepictures of similar structures which, presumingthey were filled with water, he called "Hohls-pindeln" (hollow spindles). They were subse-quently recognized to be identical with gasvesicles, however (44). The gas vacuoles ofthirteen species of blue-green algae from eightgenera have now been studied by electron mi-croscopy (3, 4, 40, 44, 66, 80, 84, 96, 101, 114),and in each case gas vesicles of the same basicpattern have been found (see Table 1). Thestructures have been variously termed "gasvacuoles" (101), "gas cylinders" (85, 112), and"gas vacuole subunits" (105), but the originalterm gas vesicle is used in this review, as it isunambiguous and makes no assumptions aboutshape when comparing with homologous struc-tures from other groups of organisms.Although it is in the blue-green algae that

gas vesicles have recently attracted most at-tention, they were first seen by electron mi-croscopy in the extremely halophilic bacteriumHalobacterium halobium by Houwink in 1956(34). His studies showed the gas vacuoles to bemade up of hollow structures, much smallerand more numerous than the optical micro-scope revealed. In the absence of a satisfacto-rily developed thin-sectioning technique,stereo-pairs were used to demonstrate thethree-dimensional structure of the vesicles,which plainly survived without collapse in thedesiccated bacteria. The vesicles were reportedto be lens-shaped, though this form was reck-oned to have resulted from distortion of aninitially spherical morphology. Apart fromthese original contributions, Houwink also cor-rectly interpreted the collapse of gas vesiclesby the pressure developed on centrifugation.Regrettably, Houwink's ingenious work andsound observations remained rather neglected,

perhaps because the techniques available tohim did not permit him to demonstrate thepresence of the enclosing membranes. Indeed,this was not done until 1967, when Larsen,Omang, and Steensland (57) revealed the ex-istence of the membranes by thin-sectioningand commented on their fundamental simi-larity to the gas vesicles of blue-green algae. Atthe same time a much more extensive investi-gation was reported by Stoeckenius and Rowen(90) describing many facets of gas vesicles ofHalobacterium halobium, as shown by thin-sectioning, negative staining, and metal shad-owing. Unfortunately all of their work was per-formed on material which had been centri-fuged, with the result that the vesicles wereseen only in the collapsed state. Stoeckeniusand Rowen referred to them as "intracyto-plasmic membranes." They recognized thatthey were "related to the formation of gas vac-uoles" in the halobacteria cells, and that theyhad features in common with the structurescharacterized as gas vesicles in blue-greenalgae (4), but it was not until a later publica-tion (89) that their true identity was conceded.

Gas vesicles of homologous form have nowbeen demonstrated in several other bacterialgroups, including some of those in which gasvacuoles had been reported by earlier workerson the basis of light microscope observations.A comprehensive list is given in Table 1. At-tention is drawn to the possibility that in cer-tain of the forms described in the earlier litera-ture, other refractile granules may have beenmistaken for gas vacuoles. For example, thecriteria of certain optical properties and solu-bility in various reagents used to identify thesestructures in species of Thiothrix (117) Rhodo-capsa (67), Peloploca and Pelonema (58, 59)are really inadequate. It is worth noting thatthe occurrence of the structures in each of thelast three examples was originally reported byMolisch and Lauterborn who were sceptics ofKlebahn's "gas vacuole theory" and that Rho-docapsa suspensa (67) is now incorporated, viaRhabdomonas gracilis (see Bergey's Manual,7th ed.), in the genus Chromatium, which issaid to be without gas vacuoles [N. Pfennigand H. G. TrUper, in Bergey's Manual, 8thed., in preparation]. There are, however, incertain of the reports from this era (e.g., 50, 70,71), descriptions of the disappearance underpressure of the structures in question. Thisbehavior is diagnostic only of gas vacuoles inprokaryotic organisms and can therefore beregarded as providing unambiguous identifica-tion of these structures where they occur.

3VOL. 36, 1972


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

TABLE 1. Prokaryotic organisms in which gas vacuoles have been reported

Shape and width (in nm)Organisma of gas vesicles, References

if reported"

Schizophyceae CohnChroococcales Wettstein

Chroococcaceae NageliMicrocystis aeruginosa Kutzing emend:

ElenkinCoelosphaerium sp. Nageli

Nostocales GeitlerOscillatoriaceae Kirchner

Spirulina platensis (Gom.) GeitlerOscillatoria rubescens D.C.

0. princeps Vaucher0. agardhii Gomont0. redekei Van Goor

Trichodesmium erythraeum EhrenbergPhormidium sp. KutzingLyngbya sp. Agardh

Nostocaceae KutzingAnabaena flos-aquae (Lyngb.) Breb

A. spiroides KlebahnAnabaenopsis sp. MillerAphanizomenon flos-aquae (Linn.) RalfsNostoc muscorum Agardh

N. sphaericum VaucherScytonemetaceae Rabenhorst

Tolypothrix sp. KutzingRivulariaceae Rabenshorst

Calothrix sp. Agardh.Gloetrichia echinulata (J. E. Smith)

P. RichterSchizomycetes NageliPseudomonadales Orla-Jensen

Thiorhodaceae MolischLamprocystis roseopersicina (Kiitzing)

SchroeterThiodictyon sp. Winogradsky

Thiopedia rosea WinogradskyAmoebobacter roseus Winogradsky

(= Rhodothece conspicua)(Rhodocapsa suspensa = Rhabdomonas

gracilis = Chromatium sp.)cChlorobiaceae Truper and Pfennig

Pelodictyon clathratiforme (Szafer)Lauterborn

Clathrochloris sulfurica Geitler

Pseudomonadaceae Winslow et al.Halobacterium halobium Petter

H. strain Delft

H. strain 5

cyl. w. 71

cyl.cyl. w. 65cyl.cyl. w. 70cyl.cyl. w. 75-85

cyl. w. 70, 75cyl. w. 75

cyl. w. 75cyl.cyl.

cyl. w. 70

spin., short cyl. w. < 120'spin., short cyl. w. 80-90,<120

cyl. w.75

mostly spin., w. <250d;cyl. w. < 180d

mostly spin., w. <250;cyl. w. < 137

mostly spin., w. <300,<198

41

77

31, 664080841149610, 1459

84, 10843141003

10

1084

Bergey's Manual, 8th ed.,in preparation


781310867

73


89

13

57108

a Bacteria (Schizomycetes) classified as in Bergey's Manual of Determinative Bacteriology, 7th ed., exceptfor Thiorhodaceae and Chlorobiaceae, classified according to Pfennig and Truper (Bergey's Manual, 8thed., in preparation). Blue-green algae (Schizophyceae) classified according to Desikachary (14).

bspin. = spindle shaped; cyl. = cylindrical; w. = width; < denotes largest dimension recorded.c Criteria for gas vacuoles doubtful and require confirmation (see text).d Estimated from micrographs in publication.

4 WALSBY


mbr.asm

.org/D

ownloaded from


TABLE 1-Continued

Shape and width (in nm)Organisma of gas vesicles, References

if reported"

Spirillaceae MigulaMicrocyclus aquaticus Orskov cyl. w. 68-125 97

Chlamydobacteriales BuchananPeloplocaceae Beger

Peloploca sp. Lauterbornc 58Pelonema sp. Lauterbornc 51, 59

Eubacteriales BuchananAchromobacteriaceae Breed

?Achromobacter sp. Bergey et al. 98

Micrococcaceae PribramSarcina ventricula Goodsir emend. Beijerinck 33, 50

Beggiatoales BuchananBeggiatoaceae Migula

Thiothrix tenuis Winogradskyc 115Classification as yet undecided

Prosthecomicrobium pneumaticum Staley short cyl., spin. w. <200d 87Ancalomicrobium adetum Staley short cyl., spin. w. <200"

STRUCTURE OF THE MEMBRANESupport for the idea that the gas vesicles in

these widely separated groups of prokaryoticorganisms are homologous structures comesfrom the striking similarity of the membranein each case. It is markedly thinner than thetypical cell unit membranes, being only 2 nmwide; it has a characteristic striated appear-ance, being banded by ribs 4.5 nm wide run-ning at right angles to the long axis of thestructure; and the ribs seem to be made up ofparticles. The evidence for this, presented indetail below, is drawn from published accountsof electron microscope studies, and from X-raydiffraction studies carried out by A. E. Blaur-ock, W. Wober and W. Stoeckenius (at SanFrancisco Medical Center) on the gas vesiclesof Halobacterium halobium, and subsequentlyby A. E. Blaurock and myself (at LondonUniversity) on those of Anabaena flos-aquae.

Membrane Thickness

In thin-sectioned material the membranes inall the organisms investigated have showngood preservation in osmic acid (4, 13, 84) butnot in permanganate (84) unless post-stainedin uranyl acetate (90) which itself proves a

good initial stain (57). Prefixation in glutaral-dehyde (13) or formaldehyde (90) gives goodresults and is essential in the case of the halo-bacteria (13, 57) which otherwise show imme-diate lysis. If the gas vesicles remain intactduring preparation, and are cut normal to ei-

ther the major or minor axes, the stainedmembrane appears as a single line, or perhapsas a series of touching dots, of mean thicknessgiven variously as 2 nm (4, 73, 84), 2 to 3 nm(13), or 3 nm (57, 84). Because of the danger ofgas vesicles collapsing flat on centrifuging,several workers have preferred to use mem-brane filtration as a means of concentratingmaterial during the fixing and staining proce-dures (13, 57, 89). Centrifugation may be usedhowever, provided low accelerations are used(see below). The material may concentrate atthe surface during the initial stages of prepara-tion, due to the buoyancy imparted to it by thegas vesicles, though it invariably sinks afterabsorption of heavy stain.

In section, collapsed membranes appear asthree-layered structures, the narrow centralregion of low electron density indicating eitherthat the two adpressed inner surfaces contactone another over a relatively small proportionof their area, or that the stain does not pene-trate completely. The half-thickness of the col-lapsed structure should give a better estimateof the membrane's mean width. The totalthickness of the collapsed structure has beengiven as 8.0 nm in the halobacteria (90), 6.0nm in the blue-green algae (4) and 5.0 to 5.5nm in the photosynthetic bacteria (13), thoughall of these values should be considered asbeing only approximate. Jost and Jones (41)point out that the dimension may be affectedby staining and report a value of 2.8 i 0.4 nmfor the thickness of the Microcystis gas vesiclemembrane, from measuring the lengths of

VOL. 36, 1972 GAS VACUOLES 5


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

shadows cast by vesicles collapsed flat onFormvar-coated grids. Even with this methoderrors may arise due to uncertainties in theshape of the profile at the edge of the collapsedvesicle, and to the difficulty of estimating thethickness of shadowing material deposited onthe specimen itself, and it is emphasized thatthese estimates are also challenged by the res-olution of the technique.The best estimate of the wall thickness

comes from X-ray diffraction studies. The firstof these was made with the gas vesicle mem-branes of H. halobium by A. E. Blaurock andhis collaborators, who produced a layeredstructure by drying an aqueous dispersion ofcollapsed vesicles onto a smooth surface andanalyzed the X-ray reflections recorded onphotographic film. The X-ray beam was tan-gent to the surface, and the photographic filmwas placed perpendicular to the beam a fewcentimeters from the dried dispersion. The dif-fraction pattern obtained indicated a series ofsheets layered at a repeating interval of 3.8nm. By comparison with the pattern obtainedfrom an aqueous dispersion of collapsed vesi-cles, it was concluded that this measurementrepresented the thickness through the twoopposed walls of the collapsed vesicle and thatthe mean thickness of the vesicle wall is there-fore 1.9 nm. However, metal-shadowed replicasof freeze-fractured (40, 101) and isolated (41)gas vesicles give evidence that both the innerand outer surfaces are corrugated (see also Fig.12a). Consequently, the overall thickness maybe greater than 1.9 nm if the corrugations ofopposing surfaces interdigitate, as indicated inFig. 2a. Studies in progress on gas vesicles iso-lated from Anabaena flos-aquae have shownthat the molecular structure is similar to thatof the H. halobium vesicles, no significant dif-ference having yet been found in corre-sponding X-ray diffraction patterns.

a bFIG. 2. Diagram showing two ways in which the

corrugated surfaces of successive layers of collapsedgas vesicles might, in theory, pack together; (a) in-terdigitating, and (b) not interdigitating. Arrow rep-resents the repeating interval (3.8 nm) that would begenerated in both cases. R denotes the ribs in crosssection. Wavy edge represents the inner, and smoothedge the outer surface of the collapsed vesicle.

Rib SpacingJost (40) was first to show by freeze-frac-

turing that the inner surface of the membranewas clearly striated, ribs running round thestructure approximately at right angles to itslong axis and at a period given as 4 nm. It isalways the inner surface which is exposed byfracture of frozen specimens, but the outer sur-face may be exposed by subliming away theencasing ice, and this reveals a similar ribbedtopography (41, 99). The ribs show with lesscontrast on the outside, perhaps because ofcontamination by nonvolatile material left af-ter etching, but then the contrast is also low inmetal-shadowed preparations of isolated ves-icles (D. Branton, unpublished data; see fig-ures in reference 41). More contrast is revealedby uranyl acetate (13) and phosphotungstate(112) negative stain, which presumably lodgesin troughs between the ribs on the outer sur-face. The rib period appears somewhat larger,5 nm (86, 90, 112) on staining, however. Thepresence of 4-nm (13, 57) or 5-nm (90) wideribs is also seen in sections parallel to thelong axis of the vesicle. Recently, we have at-tempted to obtain a more accurate estimateof the rib spacing by analyzing the electronmicroscope images of gas vesicles, isolatedfrom Anabaena flos-aquae, by the optical dif-fraction methods described by Klug and Ber-ger (49), and we have obtained values of 4.5nm from freeze-fractured material, 4.9 nmfrom vesicles metal-shadowed after drying flaton Formvar-coated grids, and between 4.8 and4.9 nm with various negative stains. The valuefrom freeze-fracturing would seem to be themost dependable as the other methods appearto be accompanied by distortion during dryingand staining. Moreover, the value 4.5 nm isthe same as that obtained by direct X-ray dif-fraction analysis of the untreated intact vesi-cles of both Halobacterium halobium andAnabaena flos-aquae.

Stoeckenius and Kunau (89) have pointedout that the ribs indicate a structure com-prising either a stack of hoops or "one or sev-eral continuous strands wound into a conicalhelix of low pitch," in the halobacteriummembrane (and by extension, into a cylin-drical helix in the central region of the blue-green algal type of gas vesicle). Observingshort striae resembling dissociated ribs in oldpreparations of disintegrated gas vesicles, theyfavored the hoop model. Jost and Jones (41) onthe other hand, observing that some ribs liber-ated from algal vesicles exceeded in length thecircumference of the vesicle, preferred thehelix model, and they claimed to be able to

6 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

detect the calculated angle of inclination (2.50)of the ribs to the long axis, from stereoscopicmicrographs, though the evidence for this hasnot yet been formally presented. It might bethought that the controversy could be settledby analyzing the images of negatively stainedcollapsed membranes. Assuming that thestructures were stained on both sides, and thatapart from being collapsed flat they were nototherwise distorted, hooped and spiralled cyl-inders when flattened would be expected togive the sort of images depicted in Fig. 3a and3b, respectively. Even though it might not bepossible to resolve the two sets of ribs in case(b) due to the narrow angle 0 (= 2.50) and smallrib period (4.5 nm), (a) and (b) should bereadily distinguishable because in (a) the in-tensity of the banding would be the same rightacross the width of the structure (as is ob-served, see Fig. 12c) whereas in (b) the con-trast would be greater at the edges where theribs at the back and front overlap. Superfi-cially then, the homogeneous appearance ofthe collapsed structure would seem to lendsupport to the hoop model. However, theseconsiderations presuppose that the basichooped or spiralled format is retained on col-lapse, and in fact this may not happen. Theribs must anyway be broken along the edges ofthe collapsed structure, and this would givethe two halves of each rib the freedom to repo-sition themselves. For example, were the ribsto be turns of a spiral, on collapse those in onehalf of the membrane might come to lie oppo-site those in the other half (as in Fig. 3a) or inthe troughs between them (as in Fig. 3c) wherethey would give closer packing. X-ray diffrac-tion studies indicate that the latter occurs, asthe first-order reflection generated by the 4.5-nm rib period disappears on collapsing themembrane.We are thus unable to settle the simple

matter of how the ribs are wound in the struc-ture at present. The unravelling of thisproblem is, of course, essential to the solutionof the gas vesicle structure and to an under-standing of gas vesicle assembly.

Subunit StructureMicrographs of freeze-etched gas vesicles

suggested to some that the ribs might be madeup of globular subunits (40, 41, 100), and thesame has been claimed from their negativelystained appearance (112), though in the viewof Stoeckenius and Kunau (89) the evidence isunconvincing. Nevertheless, Jost and Jones(41) have pursued this idea, claiming that par-ticles occur at intervals of between 2.8 and 3.5

nm. From these measurements and theirmeasurements of membrane thickness (2.8 ±0.4 nm) and rib period (4.0 ± 0.2 nm), theypropose that the gas vesicle membrane of theblue-green alga Microcystis aeruginosa is madeup of ellipsoid particles whose three principalaxes measure 3.0, 2.8 and 4.0 nm, respectively.Now, if the major axis is only as large as therib period (4.0 nm), then it must be assumedthat the ellipsoids in one rib are aligned withthose in the two adjacent ribs, as indicated inFig. 4a. If the particles are ellipsoids, there isno reason why they should not interdigitate, asshown in Fig. 4b. With this hexagonal stackingarrangement,_ the length of each ellipsoidwould be 2/\/3 times larger than the rib period(giving a major axis of 4.6 nm).The model proposed by Jost and Jones may

seem plausible at first sight, but it is open tocertain objections (see below). The dimensionsunder consideration are not much in excess ofwhat is generally reckoned to be the limit of

a

b

e,m

e m

e,m

FIG. 3. Diagram showing three ways in which,theoretically, the ribs might be arranged in the col-lapsed vesicle; face views and profiles, e at edge, mat middle. Solid lines denote ribs at front; brokenlines at back; the two coincidental in (a).

a bFIG. 4. Diagram showing two ways in which hypo-

thetical ellipsoid particles might be stacked, (a) rec-tangularly, and (b) hexagonally. The regular "rib"period, indicated by the arrows, is the same in bothcases, even though the particles in (b) are larger.

7VOL. 36, 1972


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

resolution of biological structures by electronmicroscopy (about 2 nm) due to the granu-larity imposed by staining and metal-shad-owing techniques. Even the smooth Formvarsupport may look granular when slightly un-derfocused (as in Fig. 12b), and this could leadto misinterpretation when superimposed onlinear structures like the ribs. The images ofthe membranes need detailed analysis by thesort of averaging techniques which have beenused in determining the substructures of cer-tain viruses (e.g., 49, 65). As yet our opticaldiffraction studies on the images of shadowed,stained, and freeze-fractured gas vesicles fromAnabaena flos-aquae (D. Branton and A. E.Walsby, unpublished data) have failed to re-veal any repeating subunit in the plane of ribs.And a second cautionary note should besounded here. Certain authors (13, 57, 86) haveproposed a globular or beaded substructure onthe basis of the dotted appearance of themembranes in thin section. However, this ap-pearance is only given when the vesicle is cutacross the ribs (as described above). Of course,the lack of a dotted appearance in sectionsrunning in the plane of the rib does not, byitself, prove that the ribs are not composed ofglobular subunits. Globules of the dimensionsunder consideration (-3 nm) would only bedetected in 30-nm "thin" sections if they hap-pened to be arranged in ranks perpendicular tothe plane of the section. It seems to invalidatethe model portrayed in Fig. 4a, however, asthis would give rise to just such a ranked ar-rangement.

Recently, Blaurock and I have preparedsamples of partially orientated gas vesicleswhich have permitted X-ray studies of thesubunit structure of the ribs. These sampleshave been obtained by drying drops of a highlyconcentrated, aqueous suspension of intactAnabaena flos-aquae vesicles between twoglass points, the vesicles apparently tending toalign side by side as a consequence of having ahigh axial ratio. According to Blaurock, thediffraction patterns obtained indicate thatthere are subunits packed in a crystalline fash-ion, but it is not clear whether they have anycorrespondence with the subunits suggested bythe electron microscope studies. The interpre-tation of these results is summarized in a latersection as it presupposes knowledge of thechemical composition of the gas vesicle mem-brane.

CHEMISTRY OF THE MEMBRANEThe investigations of early workers (46, 67),

reviewed in detail by Fogg (22), did not give

any exact information on the chemical compo-sition of gas vacuole membranes, and in factKlebahn's suggestion that the disappearance ofgas vacuoles in the presence of various hydro-carbons indicated a lipoid structure (46) hassince proved misleading. The unusual stainingproperties of gas vesicles prepared for electronmicroscopy, while indicating a chemistry dif-ferent from that of typical unit membranes (4,84), again provided no positive information;this had to await the development of tech-niques for isolating and purifying these struc-tures.

Methods of IsolationJost and Matile correctly reasoned that, like

gas vacuoles, the constituent gas vesiclesshould have a specific gravity substantiallyless than that of water, and that consequentlyit should be possible to separate them fromother cell components by centrifugally acceler-ated flotation (43). They reported the isolationof gas vesicles from the blue-green alga Oscil-lktoria rubescens by two-stage centrifugationof a cell brie obtained by grinding the algawith glass beads. However, the procedure theyemployed has been questioned on the groundsthat (i) the structures they isolated did notpossess the characteristic shape and fine struc-ture (86, 89) which gas vesicles have beenshown to retain on isolation (112); (ii) the frac-tion they obtained was of a very differentcomposition from the well-characterized prep-arations subsequently obtained (38, 89, 111), inhaving lipophilic rather than proteinaceouscomponents; (iii) gas vesicles would not havesurvived the pressures generated by the highcentrifugation speeds they employed (89). In-tact gas vacuoles have now been obtained fromblue-green algae and halobacteria by severalgroups, using various methods which ensurethat these sensitive structures are not sub-jected to damaging pressures during cell dis-ruption and the subsequent purification proce-dures.

Cell lysis. The problem of breaking opengas-vacuolate cells without subjecting gas vac-uoles to pressure was first solved in principleby Petter (71) who found that the bodies werereleased intact when Halobacterium halobiumwas killed on diluting the culture medium withwater. Stoeckenius and Kunau (89) employeda modification of this method, gradually dilut-ing the medium by dialyzing the culture againstdistilled water. Larsen et al. (57) found thatthe cells of halobacteria would lyse spontane-ously on elevating the pH of the culture me-dium by adding sodium hydroxide. Neither of

8 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

these methods are effective with other gas-vacuolate bacteria or blue-green algae, but wehave found that algae of Anabaena and Nostoctype lyse rapidly in strong solutions of sucrose(0.7 M) which cause osmotic shrinkage of theprotoplasts (111). This method does not workwell with the more robust unicellular forms,though these can be broken by osmotic shockafter impregnation with glycerol, as long as thecell walls have been weakened by incubationin penicillin (200 units per liter) in the pres-ence of the magnesium ion (38). Some otherblue-green algae lyse much more readily (13),allowing gas vesicles to be collected from theautolysates of old cultures.

Centrifugation. Both Larsen et al. (57) andStoeckenius and Kunau (89) found that gasvesicles could be collected by centrifugallyaccelerated flotation, but the difficulties weresuch that they preferred to use membrane fil-tration (of intact vesicles, 57) and density gra-dient ultracentrifugation (of collapsed ones,89) in their purification, though it appears thatneither method will by itself yield preparationsof very high purity (38). However, we havefound (9, 111) that it is possible to obtain cen-trifugation conditions which give good yields(50% recovery) of gas vesicles in a high degreeof purity (97.5%). This depends on first deter-mining the maximum pressure that can beapplied to the gas-vacuolate suspensionwithout causing any collapse (about 2.5 atmwith blue-green algae, but only 0.4 atm withthe halobacteria; see Fig. 9), and then seeingthat the pressure developed by centrifugation(equal to hpg, where h is the depth of suspen-sion, p its density, and g the centrifugal accel-eration) does not exceed this value (111). Byrepeatedly centrifuging the gas vesicles to thesurface, ideally through distilled water orbuffer layered over the suspension, the vesiclesare washed free of soluble material and sepa-rated from particulate components which ei-ther precipitate or do not float up as quickly(9). This technique has been coupled withliquid polymer partitioning (which removescertain contaminants) and a final ultracentri-fugation (permitting recovery of intact vesiclesfrom a shallow layer at the surface) (38), butthese additional procedures do not seem to beessential to the purification process (9). Amore detailed account of the various isolationprocedures is to be found elsewhere (110).Two groups have used radioactive tracing

techniques to assess the per cent purity ofthe gas vesicle preparations given by their iso-lation procedures (9R, 38). This was done by re-isolating previously purified, nonlabeled vesi-

cles after mixing them with (supposedly) ho-mogeneously labeled cell lysates of known spe-cific activity which had been subjected to pres-sure to destroy any (labeled) gas vesiclespresent (8). The degree of contamination of thefinal preparation was then calculated from itsspecific activity. This procedure probably givesa fairly good indication of purity, though itwould not detect contamination by substanceswhich become adsorbed irreversibly to the gasvesicle surfaces at sites complexed during theinitial isolation.

Chemical AnalysisThe proteinaceous nature of the halobac-

terium gas vesicle membrane was first alludedto by Larsen et al. (57); Stoeckenius andKunau also thought protein to be the majorcomponent, the amino acids released by acidhydrolysis accounting for 70% of the total dryweight (89) even without allowing for hydrol-ysis losses. A small amount of hexosamine wasalso found to be present, but no lipid was de-tected. Krantz (52) reports an even higherproportion of protein, 78% by weight, withnon-ashable material accounting for much ofthe residue. This may represent salts of theprotein or salt unremoved by his dialysis pro-cedure. He also records that phosphate andgalactose account for a small proportion (2%and 1.5%, w/w, respectively) of his prepara-tions though these were of uncertain purity.However, the fact that the phosphate was notseparated from the vesicles by isoelectric fo-cussing, and the galactose was not separatedby gel filtration, suggested to him that thesecomponents may form an integral part of thestructure. If this is the case, then these vesicleswould seem to differ from the algal ones,where neither phosphorus (38) nor sugar (9,38) has been detected.

Gas vesicles from blue-green algae have also,on reinvestigation, proved to be highly protein-aceous (8, 9, 38, 111). Careful analyses haveshown that no lipids or carbohydrates (theother majo'r components of unit membranes)are present (8, 38). On the basis of this, Jonesand Jost (38) have concluded protein is theonly component of the membrane. This viewis supported by their quantitative amino acidanalyses showing protein accounting for 95%of a sample having about 95.5% purity.

Jones and Jost have proceeded to suggestthat not only is protein the sole component ofthe gas vesicle membrane, but also that onlyone species of protein may be present (38).

VOL. 36, 1972 9


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

They found that the membranes were not sol-uble in many of the detergent and other sys-tems used to dissolve proteins, but that theywere partly soluble in 80% formic acid andphenol-acetic acid-water (2: 1: 1). The solubi-lized fraction ran as a single band on poly-acrylamide gel electrophoresis at low pH; theinsoluble fraction which remained at the originof the gel apparently comprised more of thesame protein molecule, as an eluate of it in thesame solvent system produced a similar elec-trophoretogram, with an identical mobile bandand, again, a proportion remaining at the or-igin (36, 39). By comparison with the electro-phoretic behavior of other proteins, Jonespostulated a molecular weight for the proteinof about 14,000, though he points out that thisfigure may be in error due to differences incharge density (36).

Substantially similar results have been ob-tained in our laboratory with the gas vesiclesof Anabaena flos-aquae after solublizing in aphenol-urea-acetic acid-water solvent system(8, see Fig. 5). These results have brought intoquestion a previous investigation by Smith,Peat and Bailey (86) using the same organism.They obtained a preparation containing col-lapsed gas vesicles (and various other subcel-lular structures) from the alga and a compara-ble fraction from another, but non-gas-vacu-olate, alga also designated A. flos-aquae byTischer (94), and analyzed them by electro-phoresis. They proposed that one prominentprotein band present only in the gas-vacuolatesample must have come from the gas vesicles,and suggested it represented a protein sub-unit of 22,000 molecular weight. However,Buckland has since found that the gas vesicleprotein of this alga is insoluble in the solvent

0 F

FIG. 5. Polyacrylamide gel electrophoretogram ofthe gas vesicles isolated from Anabaena flos-aquae,showing single mobile band. Material staining atfront, F, an impurity from the gel, also encounteredin unloaded controls. (Reproduced from B. Buck-land, reference 8.) The material at the origin, 0, isthought to be more of the same protein which hasprecipitated in the surface layers of the gel [seeD. D. Jones (36) and Jones and Jost (39)]. Photo-graph by courtesy of Barbara Buckland.

system they employed (8), and it is concludedthat they must have happened upon a compo-nent of some other structure. It seems prob-able that the molecular weight of Anabaenagas vesicle protein is, in fact, of the same orderas the Microcystis protein. This we have con-cluded principally from a quantitative aminoacid analysis which shows that most of theamino acids occur in integer ratios, suggestingan empirical formula of 143 residues with atotal molecular weight in the region of 15,100(Table 2). The amino acid composition of theAnabaena protein is very similar to that ofthe Microcystis protein (38) though it is notpossible to make exact estimates of residuedifferences because with the latter the molarratios do not yield numbers close to integersin the same way, assuming a single specieshaving a molecular weight in the region of14,000. This is also the case with the halobac-terium gas vesicle protein (89) which thoughshowing an overall similarity to the algal pro-teins, has markedly more aspartate, a fea-ture of other halobacterium proteins and pos-sibly an adaptation to the saline environment(56, 88), and less leucine. The notable featurescommon to all three membranes are the ab-sence of any sulphur-containing amino acidsand the relatively high proportion of residueswith hydrophobic side chains, which mightdetermine the postulated hydrophobicity ofthe inner membrane surface (89, 105) andhydrophobic interactions of the subunits (seebelow).As yet, no estimates of molecular weight

have been obtained for the halobacterium gasvesicles. Krantz (52) has remarked that, likethe algal gas vesicles, those from Halobacte-rium halobium are not solubilized in solventswhich are generally regarded as being suitablefor proteins. Hydrochloric acid (12 N), andanhydrous formic acid did produce relativelyclear suspensions of collapsed vesicles, buteven with these solvents it is doubtful thatthe membrane was resolved into its individualcomponents, as it was found that the proteincould be brought down relatively easily fromformic acid suspension by centrifugation.The successful solubilization of the mem-

brane is, of course, of great importance, notonly for the electrophoretic analysis of the pro-tein composition, but also for studies on theproduction and accumulation of the precursorsubunits prior to their assembly into gas vesi-cles. The difficulties encountered in solubi-lizing these structures may be connected withthe fact that the gas-facing and water-facing

10 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

surfaces of the protein molecules have suchdifferent properties (see below). At the same

time, we should entertain the possibility thatthe protein molecules are held together bycovalent cross-linkages.

Relationship of Membrane Structure andProtein Component

Both the thickness of the membrane and therib period are dimensions which could bespanned by individual protein macromole-cules, and it seems likely that the proteincomponent of the membrane indicated bychemical analysis may be arranged as a regu-lar, repeating subunit along the ribs.The X-ray diffraction studies on the Halo-

bacterium gas vesicles by Blaurock and his col-laborators have shown that the membrane incross section consists of two similar parts cen-tered 10 nm apart, and a computed profileshows the peaks of electron density expectedfor the backbones of two polypeptide chains.In addition there have been observed severalprominent wide-angle reflections which char-acterize an extensive #-protein structure. Thewide-angle reflections recorded on the photo-graphic film lie on an axis parallel to the planeof the wall and so confirm the presence ofchains running in the wall. Another feature ofinterest is that the electron density profile ofthe wall is asymmetric, as predicted assumingthat hydrophobic amino acid residues, whichare generally less electron-dense than the aver-

age, predominate on the inner (gas-facing) sur-face, whereas hydrophilic residues, which are

generally the more electron-dense, predomi-nate on the outer (cytoplasm-facing) surface.These findings correlate well with the resultsof investigations into the physical propertiesof the vesicle membrane (see below). Accord-ing to Blaurock, the X-ray study of partiallyoriented A. flos-aquae gas vesicles (see above)indicates a repeating "unit cell" from whosedimensions a molecular weight of 7,800 is cal-culated. Assuming a single species of subunitmolecule, the molecular weight is a small mul-tiple of this value, either about 16,000 or

23,000. It appears that some part of the poly-peptide chain of each subunit molecule is inthe cross-, conformation such that, along anyrib, the molecules are joined by hydrogenbonding between the backbones of the chains.There may be two layers of the cross-,B struc-ture extending along each rib in such a way as

to provide a stiffened structure, which mayaccount for the rigidity of the membrane (seebelow).

TABLE 2. Amino acid composition of gas-vesicleproteins

Arnabaena Micro- Halobac-flos-aquae cystisa teriumc

Amino acid aerugi- halobiumEmpir- nosa

icformula (mole%) (mole%)Alanine ........ 22.7 15.9 18.6 14.7Valine .......... 18.0 12.6 11.6 8.8Glutamic acid ... 17.7 12.4 11.6 8.6Isoleucine ...... 14.4 10.1 7.3 9.1Leucine ........ 14.3 10.0 11.2 0.9Serine .......... 14.1 9.9 9.3 8.1Aspartic acid ... 8.0 5.6 6.4 11.6Threonine ...... 7.0 4.9 5.2 6.5Lysine ......... 6.7 4.7 5.4 4.7Glycine ........ 6.1 4.3 3.4 9.4Arginine ........ 6.0 4.2 4.9 5.5Tyrosine ....... 4.0 2.8 3.3 2.3Proline ......... 2.0 1.4 trace 4.2Tryptophan ..... 1.0 0.7 1.5Phenylalanine ... 0.9 0.6 0.4 2.5Histidine ....... 0 0 trace 2.5Methionine ..... 0 0 0 0Cysteine ........ 0 0 0 0(Amide-N) 7.1 5.0

a Analysis by H. Larsen, P. Falkenberg (Tron-dheim), B. Buckland and A. E. Walsby (WestfieldCollege London).

b Data from Jones and Jost (38).c Data from Stoeckenius and Kunau (89). (Table

reproduced from B. Buckland, Ph.D. thesis, Univ. ofLondon, 1971.)

The model proposed by these studies de-parts from the simple globular subunit model(Fig. 4a) in several respects, though it has yetto be verified by an accurate, independentdetermination of the subunit's molecularweight. In the meantime it is worthwhile re-viewing the comparisons that Jones and Jost(38) have drawn between the protein they haveseparated by electrophoresis and the particlesthey have postulated from their electron mi-croscope observations. Assuming that eachparticle is a perfect ellipsoid of volume 4/3 ir(2.8 x 4.2 x 3.0)/8 nm (see above), they havecalculated its molecular weight, by multiplyingby the Avogadro number and the density.They estimated the density of the Microcystisgas vesicle protein from the amino acid com-position and arrived at a value of 1.34 g cm-3.This compares with a value of 1.29 g cm-3 de-termined directly by density gradient centrifu-gation of the collapsed membranes from thisalga (38) [and values of 1.23 g cm-3 for themembranes of Anabaena flos-aquae (86) andHalobacterium halobium (89)]. I would suggest

VOL. 36, 1972 11


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

that the overall density of the collapsed mem-branes is less than that of the component pro-teins because a small amount of space is leftbetween the two membrane faces (as shown inFig. 3), and that consequently the larger esti-mate, which corresponds more closely with thedensities of many other proteins, is probablythe better one. Using this figure Jones andJost arrive at a molecular weight of 14,300.They point out that this coincides with themolecular weight of the gas vesicle protein in-dicated by its electrophoretic mobility, about14,000, suggesting that the particles that areseen are the individual protein macromole-cules. However, taken at its face value, thiscomparison demonstrates that, at the outside,there is only just enough protein to permit agrazing, tangential contact between the parti-cles stacked in the least space-filling confor-mation (Fig. 4a). Such an arrangement ishardly likely to satisfy the requirement themembrane has for structural rigidity (see be-low). This would only be achieved in the con-text of their model by increasing the area ofcontact between the particles (e.g., as repre-sented ini Fig. 4b); and this, in turn, wouldrequire that the particles are smaller in thedimensions through the membrane (as the X-ray studies suggest) or along the rib.With the idea of corroborating their particu-

late concept of the vesicle membrane, Jost,Jones, and Weathers (42) have obtained esti-mates of the average weight of a single gas ves-icle. By counting vesicles sprayed in smalldroplets, whose volume was determined by theaddition of a standardized suspension of latexparticles, they found that 3.61 x 1011 vesiclesof mean length 370 nm had a protein contentof 56 ,gg, i.e., an average vesicle contained 1.55x 10-10 yg of protein. [Using an entirely dif-ferent approach we have obtained a similarvalue for the weight of the Anabaena flos-aquae gas vesicle. We measured the totalvolume occupied by the gas-filled space in asuspension of purified vesicles using a mano-metric technique (105) and estimated that 31.9pliters of space was enclosed in 5.3 mg ofmembrane material (A. E. Walsby and B.Buckland, unpublished data). From estimatesof the mean dimensions of the average gas ves-icle (a cylinder of length 227 nm and diameter68 nm to the center of the membrane, cappedby right cones of altitude 50 nm), we calculatethe volume of space in a single vesicle to be9.45 x 105 nm3. From this it is estimated that3.38 x 1013 vesicles were present in the sus-pension and that the average weight was there-

fore 1.57 x 10-1 gg. These gas vesicles are, ofcourse, rather shorter than those of Microcystisaeruginosa. ]

Jost and his collaborators have calculatedthat the amount of protein they find in asingle gas vesicle would be sufficient for 6,700molecules of 14,000 molecular weight and thatthis compares with the 6,900 molecules theyestimate would be required to cover a rightcylinder of length 370 nm, again tacitly assum-ing the arrangement described in Fig. 4a. If wecalculate the area of a Microcystis gas vesiclemore precisely, taking account of the conicalshape of the ends (see below), then the valueobtained, 7.14 x 104 nm2, can be covered byabout 6,000 of Jost's particles, or by a uniformlayer of 1.62 or 1.68 nm (assuming densities of1.34 and 1.29 g cm-3, respectively). Thesevalues are rather less than the figures given byX-ray diffraction, which is to be expected ifthere are projections on the membrane whichprevent complete contact. Similar calcula-tions on the figures for the Anabaena vesicleyield a value of about 1.90 nm, the same as theX-ray diffraction value. In both cases the de-gree of correspondence is encouraging andsuggests that it would be worthwhile obtainingmore exact estimates of all of the parametersof shape, size, and density used in these cal-culations.

Implications of the Subunit ModelThe general model suggested by these inves-

tigations, of a structure built up of just onerepeating protein macromolecule, has implica-tions in various fields: firstly, in the processesof gas vesicle formation which may be resolvedinto those concerned with synthesis of the pro-tein subunits and their assembly into mem-branes; secondly, in the interactions betweenthe subunits, which determine the overallshape of gas vesicles; and thirdly, in under-standing of the physical and chemical proper-ties of the membranes.

Before going on to examine these facets ofthe gas vesicle in more detail, it is, perhaps,worth digressing to reconsider the descriptionof it as a membrane. The classical unit mem-brane has two essential ingredients, proteinand lipid (and a third, carbohydrate, may alsobe present). The lipid bilayer has been consid-ered as providing the organizational backboneon which the protein covering is stabilized,and in lacking lipid the gas vesicle would ap-pear to be a very different kind of structure.However, it has been demonstrated that somecharacteristics of membrane structure remain

12 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

after extraction of the lipoid components (21),and the existence of proteinaceous subunits invarious types of unit membrane is now beingrecognized (6). Jones (36) has rightly pointedout that protein-protein interactions will there-fore be of general importance and that the gasvesicle may provide an ideal model for study-ing them.An alternative view of the gas vesicle has

been presented by Smith et al. (83, 86), whodraw attention to its similarity to various viralstructures which also consist of polymerizedprotein subunits. In some instances the over-production of viral coat proteins leads to theformation of empty shells and tubes (65), someof which rather resemble gas vesicles. Smithhas made the interesting suggestion that gasvesicles may have originated from viral infec-tions, an attractive hypothesis but one that isdifficult to test unless similar processes arecontinuing in present times. Perhaps when wehave gathered more data on the amino acidcomposition of gas vesicle proteins from dif-ferent organisms, we may be in a better posi-tion to discuss their phylogenetic relationships.

PHYSICAL PROPERTIES OF THEMEMBRANE

The unique feature of the gas vesicle mem-brane is that it encloses a hollow, gas-filledspace, and this poses three questions: how doesthe gas get inside the membrane; what pre-vents the membrane from collapsing into thehollow space; and what prevents water fromaccumulating inside it? Some answers to thesequestions have been provided by the followingstudies into the physical properties of themembrane.

Permeability to GasThe original idea that the vacuole gas must

be enclosed in a gas-tight membrane was putforward by Klebahn (46) to meet objections tohis "gas vacuole theory" (45) that were madeby Molisch (67) on observing that gas vacuolesdo not disappear on prolonged evacuation. Theidea that the membrane prevented the con-tained gas from diffusing away persisted untilrecently, and was extrapolated to the gas vesi-cles when they were seen by electron micros-copy (see 4, 43, 57, 74). However, now that thesituation has been investigated it turns outthat the gas vesicle membrane is highlypermeable to gases (105). The evidence ob-tained for this is as follows.

(i) By use of a Warburg apparatus whichhad been modified to enable the gas exchangecapacity of a liquid system to be measured, it

was found that, when the pressure of the over-lying gas phase was reduced, more gas wasevolved from a gas-vacuolate suspension thanfrom an equivalent quantity of a suspension inwhich the gas vacuoles had been collapsedunder pressure. When, after equilibrating thesuspensions under reduced pressure the systemwas returned to atmospheric pressure, the gas-vacuolate suspension took up a proportionatelygreater amount of gas. The extra quantity ofgas exchanged in these experiments was equalto the product of the pressure change and thevoid volume of the gas vacuoles present (asdetermined independently from the change inspecific gravity of the suspension, on col-lapsing the gas vacuoles). Details of the experi-ment are summarized in Fig. 6 and 7.

(ii) By use of the same apparatus, it wasfound that after equilibration of the gas-vacu-olate suspension under atmospheric pressure,no pressure change was recorded in the closedsystem on collapsing the gas vacuoles with anultrasonic pulse from a transducer locatedunderneath the Warburg flask and then re-equi-librating. Since the gas released was beingtransferred to an overlying gas phase whosevolume was increased by a volume equal tothat of the gas vacuoles, the lack of pressurechange indicated that the gas had been presentin the vacuoles at atmospheric pressure. Aparallel experiment indicated that after equili-bration under a reduced pressure, the gas pres-sure in the gas vacuoles had fallen to the samelevel as that of the overlying gas phase.

(iii) Mass spectrometric analyses showedthat after evacuation of a gas-vacuolate sus-pension to remove dissolved gases, all gaseshad also been removed from the gas vacuolespresent.

(iv) Gas vesicles collapse instantaneously ata well defined range of pressure, when this isapplied to them in aqueous suspension as ahydrostatic pressure (108). The required pres-sure may be generated by the overlying gasphase, but if this is allowed to build up butslowly, the gas dissolves in the surface layersof the suspending medium and diffuses intothe gas vesicles there, preventing the differen-tial pressure required to deflate them frombeing established across the membrane (105).In fact, by using gas vesicles which have beenfreeze-dried in an intact state (9), it has beenfound virtually impossible to collapse them inthis way, even under rapid rates of gas pres-sure rise, because the gas leaks inside them soquickly as the pressure builds up.As yet it has not been possible to give a

VOL. 36, 1972 13


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

a bFIG. 6. Diagram summarizing the use of a modi-

fied Warburg apparatus to demonstrate the perme-

ability of gas vesicle membranes to gas, and to deter-mine the volume occupied by the vacuole gas in thesample. (a) Gas-vacuolate suspension equilibrated atatmospheric pressure, Warburg apparatus shaking.(b) Apparatus brought to rest. Pressure of gas phaseis reduced rapidly to minimize the loss of gas fromthe suspension by diffusion. (c) Manometer tapclosed. Apparatus set shaking. Gas evolved from thesuspension measured by manometry (see reference105).

80-

40-

a

_0

time (min)

FIG. 7. Gas exchanged by (0) gas-vacuolate sus-pension of Anabaena flos-aquae, (-) similar suspen-sion in which gas vacuoles have been collapsed, (a)following reduction in pressure of the overlying gasphase, as described above in Fig. 6, (b) after re-

turning overlying gas phase to atmospheric pressure.Volume of the vacuole gas can be calculated fromthe difference between the two curves in each case(redrawn from reference 105).

figure to the gas permeability constant of themembrane, but it is evident that the mem-

brane offers far less resistance to gaseous diffu-sion than does even a thin film of surroundingwater. From the results of manometric and

pressure-rise experiments that have been car-ried out with N2, O2, CO2, CO, H2, Ar, andCH4 with gas vesicles of blue-green algae andbacteria (108), it seems likely that the mem-branes are freely permeable to all types ofgases.

Until we know more about the molecularstructure of the membrane we can only specu-late on the cause of its extreme permeability.However, it is known that the amino acid resi-dues in protein macromolecules, whose struc-tures have been analyzed by X-ray crystallog-raphy, can pack together so tightly that verylittle space is left between them (60, 92). Infact, there may be only a few spaces in a pro-tein macromolecule large enough to accommo-date even the smallest gas molecule. It seemsvery likely that this will also be true of certainregions of the gas vesicle protein. Perhaps thegases diffuse mainly through spaces betweenthe protein molecules: indeed were the proteinsubunits perfect ellipsoids stacked rectangu-larly, and exhibiting only tangential contact(cf. Fig. 4a), the intervening spaces would oc-cupy as much as (4 - /r)4 (or 21%) of themembrane's surface area.With the discovery that the membranes are

permeable to gases, the identity of the vacuolegas ceases to be even of academic interest! Itscomposition and pressure will reflect that ofthe gases dissolved in the surrounding me-dium, and will normally be air at the ambientpressure, albeit modified by the metabolic ac-tivities of the cells. Two important conse-quences of the permeability are firstly that thevacuoles cannot be used to store gas, and sec-ondly that they cannot be inflated by gas (105).

RigidityArguments for the rigidity of gas vacuole

membranes were first put forward in 1922 byKlebahn (46), who observed that the pressure

of the gas inside gas vacuoles, only one atmos-phere, must be considerably less than pres-sures exerted on the structures by cell turgorand surface tension. He correctly deduced thatthe pressure difference must be borne by an

enclosing membrane and postulated that thedisappearance of gas vacuoles which occurs onapplication of greater pressures must resulteither from water being forced into the hollowspaces through cracks in the membrane, orfrom collapse of the structure. Electron mi-croscopy has shown the latter always to be thecase (e.g., 4, 89), the constituent gas vesiclescollapsing flat in a plane at right angles to theplane of the ribs (108, 112). Theoretical con-siderations show that the flattening of themembrane must entail its tearing (108) and, as

14 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

this seems to occur preferentially between ribsrather than across them (see micrographs inreferences 89 and 112 and Fig. 12c), results inthe end caps of the cylindrical vesicles split-ting off, and spindle-shaped vesicles of thehalobacteria splitting at the middle (38, 89).Consideration of the pressures required to col-lapse gas vesicles leads to the conclusion thatthey are insufficient to force out the gas as abubble from the collapsing vesicle, and that itmust, rather, escape by diffusion through themembrane (108).The pressure relationships of gas vesicles

inside prokaryotic cells are summarized in Fig.8. The vesicle collapses when the sum of pres-sures acting on the outside less the pressure ofthe gas inside exceeds a certain value, knownas the critical pressure. The true critical pres-sures can best be investigated by using gasvesicles isolated from the cells (so that theyare free of turgor pressure) and in a suspensionequilibrated with the air (so that g, the pres-sure in the gas vacuoles, balances a, the atmos-pheric pressure). Investigations have shown(108) that gas vesicles exhibit a wide, but pre-cisely defined, range of critical pressure, whichvaries markedly in the different groups of or-ganisms (Fig. 9). In seeking an explanation forthis variation, it is worth drawing comparisonswith the critical pressures of engineering struc-tures such as pipes and boilers whose criticalpressures are determined by the strength ofthe wall material, their shape and wall thick-ness relative to overall length, and, particu-larly, diameter (2). The same factors probablydetermine the critical pressures of gas vesicles;

la

FIG. 8. Diagram summarizing the sources of pres-sure which may act on a gas vesicle inside a prokar-yotic cell: (a) atmospheric pressure; (h) hydrostaticpressure due to depth; (t) cell turgor pressure; (c)pressure from surface tension at the cell wall; (s)pressure from surface tension at the gas vesicle wall(g) pressure of the vacuole gas. (Reproduced fromreference 108 with permission from The Royal So-ciety, London.)

0o

40

80[

0o

I'U

8

._

a?

a

ONp

40

80

HALOBACTERIUN

400

AMOEBOBACTER

0

200 400 600

~~ \ < ~~ANABAENA

,I

pressure applied (kN mr2)

FIG. 9. Collapse of gas vesicles with pressure inHalobacterium sp. strain 5, Amoebobacter rosea andAnabaena flos-aquae. (0) Cells suspended in culturemedium; (0) cells suspended in medium supple-mented with sucrose to remove turgor pressure. Ineach case the turgor pressure of the cells is given bythe mean distance between the two curves (redrawnfrom references 15 and 108).

it has been found that the gas vesicles of thehalobacteria, which tend to be considerablywider than those of the alga (Fig. 10) are alsomuch weaker. The case of the vesicles in aphotosynthetic purple bacterium is interme-diate between these two (108). However, theinverse correlation between width and strengthhas yet to be rigorously tested. There seems tobe no significant (inverse) correlation betweenthe length and strength of the blue-green algalvesicles (which do not vary in diameter), andthis suggests that, unlike cylinders of homoge-neous construction, little support is derived

VOL. 36, 1972 15

VI

40


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

Wa~Ta Ca t - __

FIG. 10. Electron micrographs of freeze-fractured cells of (a) Halobacterium sp. strain 5, (b) Amoebobacterrosea, and (c) Anabaena flos-aquae, demonstrating range of size and shape of the gas vesicles, whose inner(gas-facing) membrane faces are exposed. No etching; all x45,000. (Specimens prepared in collaboration withD. Branton and S. Whytock, Univ. of California, Berkeley.)

from the ends along their length, i.e., the mainstrength of the structure resides in the indi-vidual ribs and there is little lateral supportfrom one rib to the next. The variation in crit-ical pressure of the algal vesicles probably re-flects, therefore, variations in the strength ofthe wall material. Irregularities in assembly ofthe structure may be one cause of this; anotheris that the protein is weakened by attack fromproteolytic enzymes, as has been demonstratedby experiments with isolated vesicles (9, 37).

Whatever the causes of variation in critical

pressure are, it is clear that the blue-greenalgae require gas vesicles which are strongerthan those of the halobacteria. Firstly, theyinhabit deeper waters than the shallow brinepools occupied by the salt bacteria. But, moreimportant, they generate cell turgor pressureof several atmospheres. This has been demon-strated, in the absence of any other suitablemethod for prokaryotic organisms, by usingthe gas vesicles themselves as devices formeasuring hydrostatic pressure. The turgorpressure can be estimated from the difference

16 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

in pressure that has to be applied to collapsegas vesicles in cells suspended in a hypertonicsucrose solution, and in the normal culturemedium (108). In this way turgor pressures ofup to 4.5 atm have been recorded for the blue-green algae (15) and up to 1.5 atm for a pur-

ple sulfur bacterium, whereas no turgor pres-sure has been detected in cells of halobacteria(108), as is shown in Fig. 9. Thus it appearsthat the gas vesicles in these different groupsof organisms are adapted to withstand thepressures to which they are likely to be sub-jected.As is to be expected for a structure whose

only component is a protein, its strength isaffected by a range of chemical and physicalfactors which affect the conformations and in-teractions of protein molecules. The vesiclesbecome progressively weaker in temperaturesin excess of 40 C, in the presence of agentswhich compete for hydrogen bonds, and others(such as chloroform) which destroy hydro-phobic bonding (9). Electron paramagneticresonance studies indicate that irreversibleconformational changes take place under sim-ilar sets of conditions (37). The gas vesicles ofboth blue-green algae and halobacteria showoptimal stability at around pH 7.5 and demon-strate marked weakening and even sponta-neous collapse at pH values above 10 andbelow 4 (8, 9). They show different behaviorsin solutions of varying salt concentration, how-ever, for while those of the alga are unaffectedby concentrated solutions of monovalent cat-ions, those of the halobacteria become con-

siderably stronger. This obviously representsan adaptation to the salinity of the cell en-vironment (see above).

Surface PropertiesThe inner and outer surfaces of gas vacuoles

are almost certainly of a contrasting nature,for while the two opposed inner surfaces of col-lapsed membranes show no tendency to sepa-rate, the outer surfaces show little tendency toaggregate in aqueous suspension (89).

It is to be expected that the outer surfacehas a hydrophilic nature which minimizes thepressures generated on the structure by surfacetension (105). With bubbles of the size of gasvesicles, surface tension would generate pres-sures in the order of 20 atm or more, well inexcess of the observed critical pressures. Ofcourse it could be argued that with gas vesi-cles, the critical pressure observed is thatwhich the structure will stand over and abovethat already generated by surface tension.However, it seems that the pressure developed

in this way must be very small as no signifi-cant increase in critical pressure can be de-tected on adding surface-active agents (9).Other observations indicate a surface readilywettable by water, and the fact that intact gasvesicles show a partition coefficient of 1.0:0.0between water and olive oil confirms the hy-drophilic nature of their outer surfaces (108).There are reasons for thinking that the inner

surface of the gas vesicle is, on the other hand,hydrophobic (89, 105), and that by this mecha-nism water is prevented from accumulatinginside the structure. It seems likely that thegas vesicle membrane is as freely permeable towater vapor as it is to other gases. (It is pos-sible that it might discriminate against wateron account of its polar qualities, though themembranes were not found to be any lesspermeable to another polar gas-carbon mon-oxide-albeit having only one-tenth of water'sdipole moment.) However, if pores in themembrane were to open into a hydrophobicsurface it would require a very large pressure(equal to cos 0 -2 -y/r, where 0 is the contactangle, -y the coefficient of surface tension, andr the pore diameter) to force liquid water in-side the structure, and such a surface wouldalso prevent water from accumulating insideby condensation (105).

FORMATION AND SHAPE OF GASVESICLES

The original ideas associated with the for-mation of gas vacuoles were that they resultedfrom the metabolic production of some gas (see22) and recently Bowen and Jensen (4), whoobserved the reappearance of gas vesicles incells which had been subjected to pressure,proposed that they can reversibly collapse andexpand, suggesting erection of the structure byinflation with gas. However, considering thepermeability of the vacuole membranes togases, it seems most unlikely that this couldoccur. I have proposed that the vesicle must,therefore, be a self-erecting structure, and sug-gested that it starts life as a cluster of parti-cles, which are orientated in such a way that asmore are added, a hollow structure is formed.The energy required to increase the size of thestructure under the pressures impinging on itwould have to be derived from the assemblyprocess, but the gas would enter the enclosedspace by passive diffusion (105). Perhaps thecontrasting hydrophobic and hydrophilic sur-faces presented at the inner and outer mem-brane surfaces, respectively, may be importantin correctly aligning the particles and in the

VOL. 36, 1972 17


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

process of their insertion into the membrane(108).The existence of juvenile stages in gas ves-

icle production (105) lent some support to theabove hypothesis, but the first good evidencethat gas vesicles are synthesized in a de novofashion was presented by Waaland andBranton (101). They found that gas vesicle for-mation in the blue-green alga Nostoc mus-corum could be induced on dilution of its cul-ture medium. The first gas vesicles to formwere small biconical structures, which devel-oped into short cylinders with conical ends;with time, the length of the cylinders in-creased but the ends remained the same. Thissequence suggested to them that the new com-ponents of the membrane might be added atthe middle of the cylinder as it increased inlength, and they remarked that in freeze-etched material there is often a rather moreprominent rib in this region, which might rep-resent the growing point. Jost and Jones (41)say that some vesicles appear to have morethan one, while others have none of these ribs,and they preferred to interpret them asstacking faults or the result of a deformationwhere they do occur. However, we have foundthat gas vesicles which collapse flat on dryingonto Formvar grids often split in two abouthalf-way along their length (111) or show a dis-continuity between ribs in this region (D.Branton and A. E. Walsby, unpublished data;see Fig. 12c). These observations seem to indi-cate that each gas vesicle may be assembled intwo halves. This being so, it would be expectedthat the two halves are identical, rather thanmirror images, and this must mean that thecontact relationship between the ribs at theend of each half differs from that between anytwo ribs within each half (Fig. 11) explainingthe weaknesses observed at the junction.

Evidence has now been presented by Leh-mann and Jost (61, 62) that the reappearanceof gas vesicles which occurs within a few hoursof collapsing existing vesicles under pressurefollows a course essentially similar to that ob-served on inducing Nostoc muscorum and con-firms the idea that collapsed membranes are

A"TV<EIKU; 1 Ts1WII>FIG. 11. Diagram demonstrating that, if a gas ves-

icle is made up of two identical halves, the contactrelationship between the two ribs at the center of thestructure must be different from those between anytwo ribs within each half.

not reinflated (105). It is still possible, ofcourse, that the subunits of the collapsedmembranes might be disassembled and reusedin the construction of new vesicles. Apparently,in the first hour or so the rate at which newvesicles both form and elongate is greater thanthat at steady state. Lehmann and Jost spec-ulate that this might indicate differences inthe assembly processes of the conical end andcylindrical center pieces, as in the early stagesof formation the juvenile forms possess only theend pieces. However, if the assembly processis normally limited by the availability of pre-cursor subunits, then the observed increasesare to be expected. This is because in thesteady state the growth of a large number ofexisting gas vesicles has to be sustained. Whenthese are effectively removed by pressuriza-tion, they no longer compete for the precur-sors, all of which may then be employed in theconstruction of the juvenile forms. This shouldbe taken into account in analysis of the as-sembly kinetics of the different parts.The shape of a gas vesicle must ultimately

be determined by the shape and arrangementof its constituent protein subunits. If there isbut one species of protein in the algal vesicles,then the mutual relationships of the subunitsmust be somewhat different in the conical endcaps and the cylindrical central portion of thestructure; or perhaps they are quasi-equivalentin the sense used by Caspar and Klug (12) indescribing polyhedral virus structures. Theshape of the vesicle might then be the outcomeof their being a particularly stable conforma-tion generated by a certain number of subunitsper rib, which ensures that the growing juve-nile, double-coned structure changes its modeof expansion to produce a cylinder on at-taining the critical size, around 70 nm. Such amechanism might give rise to structures ofhighly uniform diameter, perhaps constant tothe nearest subunit. In this respect, it wouldbe worthwhile obtaining better information onthe variability of this dimension in the vesiclesof a given organism. It would also be nice toknow if the small differences in mean diameterreported for the vesicles of different blue-greenalgae and certain photosynthetic bacteria (13)are genuine, and if so whether they are relatedto the size of the protein component. Certainlysome of the estimates of vesicle diameter arein error due to distortion which occurs in nega-tive staining (108).The gas vesicles of the halobacteria do not

have such a uniform shape. They apparentlycontinue to grow as some approximation of adouble-coned structure, indicating that no

18 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

strong restraint exists on their rib diameter.Cylindrical vesicles are occasionally encoun-tered, but these are of variable diameter. Thesame appears to be true of those in the purplesulfur bacteria (see Fig. 10b) and certain heter-otrophic bacteria (87, 98), though the range ofsize is smaller in these cases.The end caps of cylindrical vesicles, wher-

ever they occur, are invariably referred to asbeing cone-shaped (e.g., 4, 13, 41). This is theimpression usually given by freeze-etching andby sectioning, but not by negative staining,where the cap usually appears to be morerounded (41) and to end in a mucronate point(see Fig. 12a) or even a short stalk (see Fig.12b). Jost and Jones (41) find the end angle tobe 700 in the vesicle of Microcystis aeruginosa(which gives a cone of altitude 35 nm x cot 700,or about 50 nm), though calculations fromtheir measurements of flattened cones suggesta substantially wider angle. Again, it would beworthwhile obtaining more detailed informa-tion on the exact shape and variability of theseregions. Details of the extreme tip would beparticularly welcome, as it presumably reflectsthe form of the vesicle in its earliest stages andmay hold clues to the processes of its as-sembly.

FUNCTIONS OF GAS VACUOLES

Before proceeding to analyze in detail thefunctions that gas vacuoles might perform, weshould at least entertain the possibility thatthey confer no advantages on the cells of thebacteria or algae which possess them. Twosorts of structure might fall under this cate-gory. The first are those which might belong tosome parasite, as is the case with the viralstructures discussed above. Of course, thesestructures are themselves essential to the vi-ront (in protecting nucleic acid in the case ofviral coat protein). The second are those struc-tures which might be produced by the cell, butare incapable of fulfilling any useful function,or which perhaps once fulfilled a function, inan ancestral form, which has now become re-dundant. It is generally considered that suchstructures would not be retained in the face ofselection pressures on evolving organisms. Oneis forced back to consider these possibilitiesseriously only if no convincing function forthe structure can be found.

In setting out to assess the functions which astructure might perform, one cannot do betterthan to begin by quoting the four postulates ofWilliams and Barber (116).

(i) "The structure is necessary for successful

growth of the plant (or organism) in competi-tion with others.

(ii) The structural provision is adequate forthe requirements of the function it is supposedto serve.

(iii) These requirements could not have beenmet with markedly greater economy by someother available means.

(iv) Provision is not markedly more than isnecessary to fulfil the functional require-ments."

In clearly contravening postulate (ii) thefunction of storing gases such as nitrogen (46)or the products of fermentation (10, 50) can beruled out, as the membranes are far too perme-able to prevent any particular gas from dif-fusing away (105). It seems unlikely that theextra amount of gas "contained" in a gas vac-uole can be of any advantage for metabolicpurposes, even in the short term, as the timefor which a gas molecule is retained by themembrane is probably very small. Moreover,any gas which does enter the structure mustfirst diffuse through the surrounding cyto-plasm. It would appear, therefore, that, unlessthe gas vesicle membrane possesses propertiesother than those discussed above, any functionperformed by gas vacuoles must depend on thespaces they contain being kept free of solid orliquid, rather than filled with gas (105, 108).

Providing BuoyancyIntact gas vacuoles can provide buoyancy

because they have an overall density much lessthan that of the aqueous medium in which thevarious organisms grow. The probable densityof the constituent gas vesicles can be calcu-lated from estimates of the density of themembrane, its thickness, and the overall shapeof the structure. Assuming the gas vesicles ofblue-green algae to be cylinders of mean length370 nm and width 71 nm capped by rightcones with a solid angle of 70° (41), and themean thickness of the membrane to be 1.62nm (see above), then the ratio of membranevolume to the total volume of the structure iscalculated to be 1.0:10.6. The shape of thehalobacterium gas vesicles is more variable,but if we assume an oblate spheroid of length216 nm and width 133 nm (108) then the ratiois 1.0:16.4. Then taking 1.34 g cm-3 as thedensity of the membrane (see above), the over-all density of gas vesicle is 0.126 g cm-3 in thealga and 0.082 g cm-3 in the halobacterium.Clearly, the first two postulates are satisfied,since no solid or liquid substance is knownhaving a density of such a low order. Indeed, ifwe accept that a constraint is placed on the

VOL. 36, 1972 19


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

iTi.w.'..

'A-.&.M. 1.

.,: I i rll.---,..w

N1;

X.llI%..

aeob<S

Z., .,:.'

.A-~

FIG. 12. Gas vesicles of Anabaena flos-aquae negatively stained with phosphotungstate. (a) Intact vesicleshowing corrugated profile of the ribbed structure, x300,OOO. Micrograph by Dennis Hodge. (b) Intact vesicleshowing short stalk which is occasionally encountered. Underfocussed, x200,OOO. (c) Collapsed vesiclesshowing separation of the cone-shaped ends, and discontinuities at the central region, x200,OOO. Micrographby Daniel Branton.

diameter a gas vesicle may reach by the pres-sure it must withstand, as suggested above,this explains why buoyancy is apparently pro-vided with "more marked economy" by thehalobacterium vesicles (which are subjected to

less pressure) than the algal ones. In fact it canbe demonstrated that the cylindrical form ofthe algal vesicle is the most efficient space-storing shape when rib diameter is limited bythe impinging pressure (108) and that some of

20 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

the lost advantage is made good by the cylin-ders stacking together with less interveningspace than spheroids (109).

Naturally, gas vesicles will only provide acell with buoyancy if they occupy a sufficientproportion of the cell volume. The possessionof gas vacuoles is often correlated with cells ina floating condition, not only in the alga (e.g.,22, 27, 30, 46, 75) and halobacteria (34, 57, 70)but also in the prosthecate (87) and photosyn-thetic bacteria (73). In almost all cases thecells can be shown to be dependent on gasvacuoles for their buoyancy, which is lost whenthe vacuoles are collapsed by pressure (e.g.,4, 22, 46, 50, 71, 104; but compare 7, 107).However there are some algae (14) and photo-synthetic bacteria (P. Hirsch, personal com-munication) which, though possessing gasvesicles for much or all of the time, are seldomor never buoyant.Few direct measurements have been per-

formed on the actual volume occupied by gasvacuoles in cells. From the changes in specificgravity which occurred when the alga was sub-jected to pressure, Klebahn estimated that thegas-filled spaces of the vacuoles occupied 0.8%of the cell volume in Gloeotrichia echinulata,and that 0.7% by volume was the minimumproportion which would be required to makethe alga float (46). Using the same method, Ihave found that with Anabaena flos-aquae thevolume occupied by gas varies from 1.8 to9.8%, when grown under varying culture condi-tions, and that about 2.1% is required torender it buoyant (107). Smith and Peat, in-vestigating the same two algae by electron mi-croscopy, concluded that the gas vacuoles oc-cupy 22% of the cell volume, and in a thirdalga Oscillatoria agardhii, 39% (84). However, Iunderstand from them that these figures in-clude the volume occupied by the gas vesiclemembranes and the spaces between them, sothat comparisons with estimates of gas volumeare not valid.

Petter observed that by light microscopy thecells of the halobacteria appeared to be almostfilled with gas vesicles (70), and, although elec-tron microscopy shows this view to be some-what exaggerated (34), I have found that thevacuole gas occupies a substantially greaterproportion of the cell volume than in the alga.Estimates of gas vesicle void volume, by themanometric method (46), and packed cell vol-ume, by centrifugation, indicate a value ofnearly 15% (unpublished data).The excesses of buoyancy indicated by these

limited studies may, at first sight, appear tocontravene the fourth postulate dealing with

"works of supererogation" (116). However, flo-tation is a time-dependent activity whose ratevaries in proportion with the difference be-tween the densities of the floating object andof the suspending medium, according toStokes' law. For a sphere U = 2 g r2 (p _ p')/9 1,where U is the sinking rate (negative when p< p'), g the gravitational acceleration, r theradius of the sphere, p its density, p' the den-sity of the medium, and q its coefficient of vis-cosity. Plainly, to maintain a given rate of rise,p - p' must have a greater (negative) value fora small object than a large one, and this putsin perspective the higher degree of gas vacu-olation in the halobacteria. In the same con-text, it is interesting to note that, while mostother planktonic algae tend towards the uni-cellular habit or small, divided forms givingslow sinking rates, many gas vacuolate blue-green algae have adopted streamlined colonialforms which give high rates of flotation; viz.,the flakes of Aphanizomenon (69), balls ofMicrocystis and Coelosphaerium (75), andrafts of Anabaena. Reynolds has shown thatthe observed flotation potential of blue-greenalgae gathered from waterblooms is correlatedabove all else with colony size (76, 77).Buoyancy in the halobacteria. It has

proved relatively simple to demonstrate thatthe putative buoyancy-providing role of gasvacuoles satisfies the last three postulates ofWilliams and Barber; but it is not quite so easyto decide about the first postulate, dealingwith the necessity of the structure. In doing so,it is felt that affirmatives to the four questionsset out below, should be sought. Although weare not yet in a position to supply all the an-swers, an examination of the need for buoyancyin the halobacteria seems to provide an idealtest case.

(i) Are there reasons for thinking that thefunction, as and when performed by the struc-ture, should aid the survival or growth of theorganism, at least under some of the condi-tions to which it may be exposed? Petter(71) has pointed out that these bacteria inhabitsaturated brine pools, in which the solubility(and hence the diffusive transfer rate) of oxy-gen is very low, and gas vacuoles enable thebacteria, which are obligate aerobes, to floatto the surfaces of the pools, thus ensuringtheir access to air (57).

(ii) If the structure is removed from the orga-nism, or rendered ineffective, does it have det-rimental effects on the growth or survival ofthe organism? This might be difficult to testwith some structures but it is extremelysimple with the gas vacuole, which may be col-

VOL. 36, 1972 21


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

lapsed by pressure. It would be interesting tocompare the growth rates of the vacuolate andnonvacuolate halobacteria in static cultures(where buoyancy should provide advantage)and shaken cultures (where no advantagewould be conferred).

(iii) If a stable mutant lacking the structureoccurs, is it selected against under conditionswhich permit the execution of the function?Both Halobacterium halobium (89) and Halo-bacterium sp. strain 5 (57) form mutants whichlack gas vacuoles. In the latter, this mutantarises quite frequently, and no back mutationis observed (unless induced by mutagens, seereference 89). Larsen et al. (57) reported thatwhen this strain is repeatedly subcultured onagar, the proportion of the gas-vacuolate mu-tant progressively increases. Buckland hasfound that when serial subcultures are madefrom static liquid cultures, there was no suchincrease (8). However, we are reluctant to con-clude, at this stage, that this demonstrates theselection of the gas-vacuolate form due to itsbuoyancy, since the increase has not been ob-served in shaken cultures, where buoyancyshould be of no consequence.

(iv) If the structure is an inducible character,do the conditions which promote its formationcorrelate in any way the need for the functionit performs? It is not clear whether gas vac-uoles can be regarded as inducible systems inthe halobacteria, but it has been observed thatthey are most abundant in the late exponentialand stationary phases of growth and absent, orof low frequency, during the early exponentialphase (8, 57, 89). One condition that might becorrelated (negatively) with this sequence ofevents is oxygen tension which will fall, evenin shaken cultures as the population densityrises. There are indications that high oxygentensions inhibit gas vacuole formation (8), butagain the evidence is, as yet, inconclusive.In other heterotrophic bacteria. As yet we

have very little information on the physiologyof the other heterotrophic bacteria listed inTable 1, but it is clear that in certain of theforms which have been observed in culture (87,97) that the gas vacuoles render the cellsbuoyant such that they float at the water sur-face. For the aerobic organisms, buoyancy islikely to have the same significance as thatproposed for the halobacteria, increasing theavailability of oxygen. Not all of the hetero-trophic forms described are aerobes, however.The possibility and significance of gas vacuolesbeing used to regulate buoyancy in the anaer-obic forms is discussed, briefly, below.

In blue-green algae. Paradoxically, severalof the early reports of gas-vacuolate algae wereof forms occupying the sapropel or bottommuds (14, 22). It was suggested that the gasvacuoles might form specifically under theanaerobic conditions prevailing in these habi-tats, causing the cells or filaments to float tothe water surface (50). It has also been claimedthat gas vacuole production is induced in somealgae by a change in the concentration of min-eral salts in the culture medium (10, 101) or achange in light intensity (14, 63, 95, 103). How-ever, in at least certain of these cases gas vac-uole production is restricted to particulartypes of cell whose differentiation may be theprimary event triggered by changing condi-tions. For example, Canabaeus (10) reportedthat, in both Tolypothrix rivularia and Cal-othrix epiphytica, gas vacuole formation wasrestricted to the hormogonia into which thetapering ends of the filaments differentiated,and similar observations have recently beenmade with Gloeotrichia ghosei (81). The hor-mogonia are buoyant when they are releasedand this may be important in the dispersal ofspecies having sedentary stages.The majority of truly planktonic blue-green

algae possess gas vacuoles at all times andmay aggregate at the surfaces of natural waterswhere they form waterblooms. Reynolds, whoinvestigated the formation of several suchblooms on meres of the Shropshire-CheshirePlain, in England, reported that alga takenfrom the lake surfaces always possessed gasvacuoles, and owed its buoyancy to them (76).By taking depth profiles of algal concentrationin the water, he demonstrated that the suddenappearance of a bloom did not result from anexplosive increase in algal numbers, but was aconsequence of the "telescoping" of algaegrowing throughout the water column to thelake surface (75). This only happened in calmweather when turbulence, which tended to mixthe alga back down into the water mass, sub-sided (64, 75).

Superficially, it might seem that the condi-tions existing at a lake surface would be thosewhich were most suitable for algal growth. Ithas been suggested that carbon dioxide is lim-iting in some waters (54) and that this mightbe more available to algae floating in contactwith air at the surface (107). The mean lightintensity, too, is higher at the top of the watercolumn, and should sustain higher rates ofphotosynthesis. However, surface light intensi-ties may, of course, become high enough toinhibit growth, and even cause death (see be-

22 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

low) though the sensitivity of different speciesseems to vary considerably in this respect. Forexample, studies in progress on Clear Lake,Northern California, have revealed that whileAphanizomenon flos-aquae may not survive formore than a few hours at the lake surface,Microcystis aeruginosa apparently thrives inthis situation (A. J. Home, personal communi-cation). Another factor which may lead to thedemise of the algal population at the watersurface is the exhaustion of nutrients by theconcentrated algal mass (24). When lightwinds fan the surface of the water without ac-

tually inducing turbulence in it, the alga may

be swept to the leeward shore, and produce athick surface scum (77). Once subjected tothese conditions, the alga rarely remains via-ble (20, 25, 75; but contrast 107). Probably, formany species, the appearance of a waterbloommarks the end of an algal population ratherthan its climax, though as with most goodshows the climax may come very close to thefinale.

It thus seems likely that the buoyancy pro-

vided by gas vacuoles has some more subtlesignificance to many planktonic blue-greenalgae than merely that of carrying them to thewater surface. The same must be true of gas

vacuoles in the photosynthetic bacteria, sincethese organisms are obligate anaerobes whichwould not survive exposure to light in the aer-

ated medium of water surfaces (72). Lund (64)has pointed out that the buoyancy provided bygas vacuoles will produce "forced convection"as the alga (or bacterium) moves with respectto the water mass, and that as with sedi-menting algae this may be of importance inmaintaining around the cells steep diffusiongradients of nutrients (68).Lund also remarked on the paradox of algae

with gas vacuoles maintaining a positionsomeway below the lake surface. That thiseven occurred in the waters under ice, in whichturbulence is completely absent, demonstratedthat the alga could not be lighter than water(64) and provided the first suggestion that gas

vacuoles might be important in regulatingbuoyancy.

Regulating BuoyancyVariation in the ratio of cell volume to gas-

vacuole volume may be brought about bychanges in the respective rates at which cellmaterials and intact gas-vacuoles (i) are pro-duced, and (ii) disappear. Such changes will,in theory, result in variation of the cell densityand will determine whether the cell sinks or

floats, and how rapidly it does so. If thechanges are correlated with physical or chem-ical conditions which form vertical gradientsin natural waters, they may provide the cellswith a means of regulating their position in thevertical water column. There are a number ofconditions which may vary with depth, and, inparticular, light intensity, temperature, salin-ity, oxygen tension and pressure may formstable gradients.

In blue-green algae. The degree of gas vac-uolation in various blue-green algae has beenreported to respond in different ways (contrastreference 105 with 103) to light intensity, butin three planktonic forms which have been in-vestigated, Anabaena flos-aquae (15, 105, 108),Oscillatoria redekei (114), and Microcystisaeruginosa (15), gas vacuoles are recorded asbeing more abundant in alga grown under lowlight. Two mechanisms have been proposedto explain this.The first is that the formation of gas vac-

uoles may proceed independently of light in-tensity, with the result that as this increasesthey become diluted out by the greater vegeta-tive growth that this sustains (107). Smith andPeat (85) found that gas-vacuolation of Ana-baena flos-aquae was least in the early expo-nential phase when growth of the alga wasmost rapid; and Lehmann and Jost (61) dem-onstrated a similar response in Microcystisaeruginosa by the gas vesicle-counting method(42). They concluded that the decrease in gasvesicle numbers must have resulted from dilu-tion by cell growth, regarding the disappear-ance of these structures as being unlikely be-cause of their "high stability." It has beenpointed out, on the other hand (105), that thefall in gas-vacuolation may be a direct re-sponse (perhaps by the second mechanism) ofthe cells to the higher light intensities whichexist in young cultures.The second mechanism is that the alga dis-

poses of intact gas vesicles in high.light inten-sity by collapsing them (108). It has been foundthat when Anabaena flos-aquae is grown underlow light intensity its cells have a lower turgorpressure and a proportion of the gas vesicleswith a lower critical pressure than when thealga is grown under high light (108). When thealga was transferred from low to high light in-tensity, the cell turgor pressure was found toincrease in instances from less than 3 to morethan 4.5 atm (290 to 460 kN m-2) and at a rateof nearly 1 atm (94 kN m- 2) per hr. The rise inturgor pressure, which results from the in-creased accumulation of photosynthate, was

23VOL. 36, 1972


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

found to be sufficient to collapse enough of thealga's gas vesicles to destroy its buoyancy. Theloss of buoyancy was often quite rapid, beingdetectable within half an hour, and nearlycomplete at two hours (15). Sometimes rathermore gas vesicles appeared to be lost thancould be accounted for by the rise in turgoralone and while this may have been an artifactof measurement (15) it is also possible thatsome vesicles collapsed as a result of their be-coming weakened in a changing cellular envi-ronment. We have found that several factorsaffect the strength and stability of isolated gasvesicles, and two of these, decrease in pH (asmight occur with the production of organicacids) and exposure to proteolytic enzymes,might be important in weakening them insidecells (9). This possibility requires further in-vestigation.

It is plain that any response which results indecreased gas-vacuolation with increasing lightintensity could enable the alga to select a posi-tion on a vertical light gradient (108), and thereare several accounts of blue-green algae doingthis in freshwater lakes (18, 19, 64, 119; J. W.G. Lund, quoted in reference 26). Fogg (25)has argued that the conditions which exist inthe discrete layers occupied by these algaemay be those most suitable for their growth. Infact it has been observed that as inorganic nu-trients become depleted in certain freshwaterlakes, the population maxima of gas-vacuolateblue-green algae (but not other forms) movedown (18, 119). This suggests that the algaestation themselves at a depth where the meanlight intensity does not exceed that required tomaintain growth at a level which is limited byother factors (26). It has been proposed thatjust such a response would be provided by themechanism described above in which the risein turgor pressure results from the amount ofcarbon fixed in photosynthesis exceeding thatwhich can be assimilated by the cells (15).The collapse of gas vacuoles under rising

turgor pressure is sufficiently rapid to explainanother curious behavior pattern of blue-greenalgae, that of diurnal migration to and fromthe water surface during night and day, respec-tively (30, 82). This behavior would also seemto be connected with avoiding high incidentlight intensities, particularly in the case of thetropical lake described by Ganf (30).In photosynthetic bacteria. It has been

suggested that the sensitivity of blue-greenalgae to high light intensities, particularly inthe presence of oxygen (26), bears the mark oftheir affinities with photosynthetic bacteria(107). These organisms show a requirement for

that strange combination of light and anaer-obic conditions provided only below layers ofwater, which although transparent to lightprovides a diffusion barrier to oxygen. Thus,whereas recent observations concerning thegreat suitability of layers beneath the surfacesof fresh-water lakes for blue-green algal growth(25) may be met with some surprise, parallelobservations form part of the well establishedlore of the photosynthetic bacteria. Pfennig(72) has recently reviewed the literature docu-menting cases of purple and green photosyn-thetic sulfur bacteria occupying the anaerobiczones at or below the thermoclines of variouslakes, down to depths as great as 35 meters.Many of the bacteria concerned are motile"and able to migrate by their own flagellar ac-tivity," but others (e.g., Rhodothece, Thiope-dia, Lamprocystis, and Pelodictyon) althoughimmotile, possess gas vacuoles, and "will risepassively as long as they are less dense thanwater, which is mostly true for water at 4 to 8C." This statement suggests a simple mecha-nism whereby, in possessing gas vacuoles,cells achieve a minimum density that is lessthan that of cold water beneath the thermoclinebut greater than that of the warmer water aboveit. However, subsequent investigations byPfennig and Cohen-Bazire (73), who foundthat cells of Pelodictyon clathratiformeformed gas vacuoles and became buoyantwhen grown in dim light at between 4 and8 C, but not at higher temperatures, indi-cate a more complex mechanism. This re-sponse may be regarded as parallel to that ofthe blue-green algae and indicates a similarbuoyancy-regulating mechanism. Studies withthe purple sulfur bacterium Rhodothece con-spicua [= Amoebobacter roseum (Bergey'sManual, 8th ed., in preparation)] have shownthat the capacity to collapse gas vesicles undercell turgor pressure exists in these organisms(108) and it is possible that a rise in turgor maybe promoted by high light intensity and ele-vated temperatures. Some of these organismsoccur abundantly at the more permanent ther-mocline in meromictic lakes (13, 72), and it istempting to speculate on the outcome of a pos-sible turgor pressure rise resulting from thecells floating from the saline to the overlyingfreshwater layers in this region.

In heterotrophic bacteria. Van Ert andStaley (98) have recently isolated a gas-vacu-olate, heterotrophic bacterium (possibly aspecies of Achromobacter) which is a faculta-tive anaerobe. They reported that gas vacuoleproduction by this organism is sporadic butseems to be favored when the cells are grown

24 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

in liquid culture media under reduced oxygentension. Perhaps this organism, which was col-lected from the oxygen-depleted thermoclineof a eutrophic lake, possesses a buoyancy-reg-ulating mechanism which responds to the con-

centration of oxygen in the water. It wouldalso be interesting to know if such a responseexisted in Sarcina ventriculi (50), which is an

obligate anaerobe (33).Hirsch and Pankratz (32) have described

other gas-vacuolate bacteria, some of whichmay be heterotrophic organisms, occurring inthe Thiopedia layer on the mud surface of a

shallow pond. Presumably these cells eitherare anchored to a substrate which preventsthem from rising (they were selected by theirability to attach to Formvar-coated grids im-mersed in the pond) or have insufficient gas

vesicles to render them buoyant. The role ofgas vacuoles in these forms is not immediatelyapparent. It may be that these organisms alsostratify at the anaerobic thermoclines ofdeeper waters in which situation buoyancyregulation is important, and do not have thecapacity to lose their gas vacuoles completelyunder the equivalent conditions offered in thesapropel of shallow pools.

Light ShieldingThe idea that gas vacuoles might perform a

light shielding function was first suggested byLemmermann (63), on noticing that they were

invariably present in the algae of surface wa-

terblooms, which might be exposed to high,perhaps damaging, light intensities. This con-

cept has been supported by the observationthat optical changes accompany the destruc-tion of gas vacuoles by pressure, and perhapsalso by the reddish appearance of the struc-tures under the light microscope.Now that gas vesicles have been isolated and

purified, it is plain from the results of chem-ical analysis, and the white appearance of in-tact gas vesicles in suspension, (38, 89, 111)that they are free from pigments. Thus, theirreddish hue must be a structural color (118),and any shielding they perform must be a re-sult of structural interaction with light. Fourtypes of interaction have to be considered: re-

fraction, reflection, interference, and scatter-ing. Apart from interference, which can resultin the annihilation of light rays, these structuralinteractions will result only in the redirection oflight through a suspension of cells, and not a

diminution of its intensity. However, the gas

vesicles may be so located and oriented in thecells that they direct light away from under-lying light-sensitive structures. Refraction was

suggested to be the cause of the red appear-ance of gas-vacuoles by Petter (71). She arguedthat red light would be bent less than blue onpassing across the air-water interfaces of thevacuole and consequently, at certain levels offocus, more red light would be collected by theobjective lens. In this way, gas vacuoles mightprovide a measure of protection from light ofshorter wavelength when they are interposedin the path of incident light rays. Analysis ofgas vacuoles under the interference micro-scope by Fuhs (28, 29) shows their overall re-fractive index to be close to 1.1, the value ex-pected considering the combination of theirgaseous contents (which would have a refrac-tive index of close to 1.0) and membranoussubstructure.

Internal reflection provides another way inwhich light might be redirected back out of thecell, as proposed by van Baalen and Brown(96). The same authors propose that theremight also be "an interference effect perpen-dicular to the long axis" and that, for gas vesi-cles of average diameter 78 nm, "zero orderreflection will occur [at a wavelength] around320 nm." No explanation is tendered. As withreflection, the angle the incident light raymakes with the gas vesicle will be extremelyimportant. The strongest interference effectsare created by planar surfaces (rather than cy-lindrical ones). I have observed that thin filmsof gas vesicles spread on water, like oil films,produce rainbow effects, indicative of interfer-ence, but there is no evidence that this effectcan produce light shielding of any value insidecells, even considering the particular configu-rations which gas vesicles adopt.On account of their minute size and the

large difference in refractive index betweentheir gaseous contents and the aqueous sus-pending medium, gas vesicles produce verystrong light scattering (37, 106, 107). Over mostof the visible range, the intensity of the lightscattered shows the relationship i cc X-4 char-acteristic of Tyndall or Rayleigh scattering(106, 107). Gas vacuoles in cells show basicallysimilar characteristics, though because of thepresence of other scattering and absorbing ma-terial, and also for reasons discussed below, itis less easy to analyze. There seems little doubtthat gas vesicles singly or stacked together canscatter light away from underlying objects,though just what proportion of the incidentirradiation of any given wavelength a singlelayer of gas vesicles will scatter has yet to bedetermined. One feature of interest, however,is that the shorter wavelengths, which are themost damaging, will be scattered away much

VOL. 36, 1972 25


mbr.asm

.org/D

ownloaded from


BACrERIOL. REV.

...

*

^~ r-: /

/^e'9.'

FIG. 13. Light micrographs of filaments of Anabaena flos-aquae (a, b) with gas vacuoles; (c, d) filamentsfrom the same sample whose gas vacuoles have been collapsed under pressure. (a, c) Bright field; (b, d) posi-tive phase-contrast illumination. (The clear cells are heterocysts.) Fuhs (28) points out that gas vacuoles areexcellent phase objects because they have a refractive index much lower than the surrounding cytoplasm.When the difference in optical path length exceeds the retardation of the phase plate, this results in falsereversal, and they appear brighter than the background (as in b). They do not absorb light and thus are pooramplitude objects, almost invisible under bright field with high resolution optics; but with low numericalapertures they are characterized by diffraction fringes (as in a).

more strongly than the longer ones (which areperhaps more useful in photosynthesis). Thisdifferential scattering effect also probably con-tributes to the red appearance of gas vacuolesviewed by transmitted light.

Plainly, the theoretical aspects of the waysin which gas vesicles interact with light havereceived but scant attention to date. It is feltthat, of the effects considered, light scatteringis the most important, but that a much moredetailed analysis of the problem is required.

In each of the above considerations it isemphasized that the gas vacuoles should beorientated between the incident light and thelight-sensitive structure. The latter comes intwo categories: the photosynthetic apparatus

whose pigments may catalyze or be susceptibleto photooxidations (93), and proteins and nu-cleic acids which can be damaged by ultravi-olet radiation (35). Gas vesicles are usuallylocated in the center of the cell or throughoutthe cytoplasm in most blue-green algae (22)and bacteria (13). However, there are accountsof three blue-green algae which, at least undercertain conditions, have gas vacuoles distrib-uted at the periphery of their cells. The first isTrichodesmium erythraeum, a marine algawhich, floating at the surfaces of tropical seas,is exposed to very high illumination. Its vac-uoles form a hollow tube running nearly theentire length of each cell surrounding most,but not all, of the photosynthetic lamellae (96).

26 WALSBY


mbr.asm

.org/D

ownloaded from


V,GAS VACUOLES

The second is a strain of Nostoc muscorum inwhich gas vesicle formation is induced onlyunder certain specific conditions, viz., on dilu-tion of the culture medium, when they formthroughout the cell (101), and on transfer to a

higher light intensity (>4500 lx), when theyformed parietally in the cell (102). The third isAnabaena flos-aquae, in which the gas vac-

uoles, while showing a quantitative decreasewhen the alga is transferred to a higher lightintensity (15, 105), do exhibit a different dis-tribution, forming at the cell periphery, ratherthan throughout the cytoplasm (H. Shear andA. E. Walsby, manuscript in preparation).Taken with the theoretical considerations out-lined above, these observations appear to sup-port the idea that gas vesicles provide a lightshielding function; recently two empirical ap-proaches have been made in an attempt to seeif they do.In the first, the Waalands and Branton (103)

examined the absorption spectra of Nostocmuscorum using a spectrophotometer whichcorrected for a proportion of the light scat-tering by samples (i.e., the photocell collectedmuch of the scattered as well as the trans-mitted light). When isolated vesicles wereadded to a suspension of nonvacuolate alga,the difference between the absorption spectrawith the vesicles intact and with them col-lapsed was a smooth curve, nearly the same asthat given by isolated vesicles in the absenceof the alga. However, the difference betweenthe spectra of alga with intact and with col-lapsed vesicles in the cells showed distincttroughs at the position of the major pigmentabsorption peaks (in the blue and red). Theyconstrued this as demonstrating that when thegas vesicles were present in the cells (but notin suspension outside of them) they were scat-tering light away from the cells in such a waythat it was not available for absorption by theunderlying structures, and in this manner theyshielded the photosynthetic apparatus. Thefact that, on collapse of the gas vacuoles, oneof the absorption peaks actually increased alittle, though only by 3%, does demonstratethat there must have been some shielding.However, the occurrence of the troughs in thedifference curve must be due in part to an arti-fact of measurement (Shear and Walsby, inpreparation). Duysens (17) has shown that lesslight is absorbed by a sample when the pig-ment it contains is present in suspended parti-cles than when the pigment is homogeneouslydistributed in solution, and that this effect isgreatest at the absorption peaks. Conse-quently, on removal of the pigment from the

particles, the decrease in absorption at thepeak wavelengths is less than would be ex-pected from measurements on the extractedpigment. We have shown that a comparablesituation arises at the absorption maxima onremoval of the light-scattering vacuoles (again,using a spectrophotometer which is not totallycorrected for light scattering) and occurs ir-respective of whether the gas vesicles arepresent in the center or periphery of the cells(Shear and Walsby, in preparation). Thus, inorder to assess the degree of shielding by spec-trophotometry it would be necessary first toestimate the contribution of this effect byapplying the theoretical corrections describedby Duysens (17).

Shear and I have attempted to estimate thedegree of shielding directly by comparing therates of various photochemical events in cellsof Anabaena flos-aquae with and without in-tact gas vacuoles. We investigated photosyn-thesis over a range of limiting light intensities;the photooxidation of phycocyanin (which didnot show photochemical equivalence) underintense illumination; and the survival of thealga after exposure to ultraviolet light (whichshould be scattered very strongly). In no casewere significant differences recorded betweenalgae with gas vacuoles, and those without,either with low-light-adapted (with central gasvacuoles) or with high-light-adapted alga (withperipheral vacuoles). There were differencesbetween the two sets of alga, though thesemust have been due to some other cause, pre-sumably pigmentation (manuscript in prepa-ration). There is thus no evidence that gas vac-uoles provide a significant degree of shieldingin this alga. The different orientation of thegas vacuoles under high-intensity light remainsa mystery at present but holds out a warningto the dangers of jumping to conclusions fromteleological evidence.

It is suggested that it would be a simplematter to test the effectiveness of gas vacuolesin other algae (especially the Nostoc musco-rum) and halobacteria (which inhabiting thesurfaces of brine pools are also subjected tohigh and perhaps damaging light intensities)by measuring the resistance of these organismsto ultraviolet light. Pending these results it iswondered how the light-shielding function willstand in relation to the third postulate of Wil-liams and Barber (116). Carotenoids, whichapparently fulfil this function (53), occur abun-dantly in blue-green algae (23), photosyntheticbacteria (72) and halobacteria (16). However,in comparing the economy of gas vacuoles andcarotenoids in providing light shielding, it

VOL. 36, 1972 27


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

should be remembered that if gas vacuoles arealready fulfilling another function, such asproviding buoyancy, then the additional ex-pense involved in their providing lightshielding (say by reorientating the structures)may be small, representing an overall savingeven if they shield relatively inefficiently.Moreover, if they are effective, they should beparticularly so in the ultraviolet wavelengthswhere carotenoids offer no protection (55).

Light shielding in suspensions en masse.Even if gas vacuoles do not provide shadingdirectly for other structures within a given cell,their presence may significantly affect the pas-sage of light through a suspension of such cells.If the suspension is illuminated from above, asin direct sunlight, then back-scattering of lightfrom the superficial layers will serve to de-crease the amount of light entering the suspen-sion. On the other hand, scattering might alsoincrease the effective path length of lightthrough the suspension, and so increase theeffective absorption within a given layer. Thisin turn will result in the mean light intensityfalling off more rapidly with depth, so thatcells with gas vacuoles in surface layers mightbe considered as providing sacrificial shadingfor cells in the layers below them.

Increasing Cell Surface Area to VolumeRatio

Houwink has pointed out that the presenceof gas vacuoles will result in an increase in theratio of cell wall area to cytoplasm volume,though he questioned whether this would beprofitable to the organism (34). It would, theo-retically, bring about an increase in the rate ofexchange of substances by the cell, and itmight also allow the accommodation of agreater quantity of a metabolically active sur-face membrane.There are, or course, other ways of in-

creasing the surface area to volume ratio (byconvoluting the surface, or decreasing cell size)and while this does not prove that gas vacuolesare unimportant in this respect, it is empha-sized that their capacity for doing this is verylimited. Assuming a spherical cell geometry isretained, introducing gas vacuoles to 15% ofthe cell volume (the largest proportion ob-served, see above) would increase the surfacearea by only about 10%.

CONCLUDING REMARKSGas vesicles are clearly most unusual struc-

tures without parallel in the biological or phys-ical world. They have, at the same time a raresimplicity which deserves thorough investiga-

tion and promises the reward of perhaps amore complete understanding than is likely tobe obtained with any other subcellular organ-elle in the near future.The fact that gas vesicles retain their unique

properties on isolation and purification notonly facilitates such investigations but alsosuggests that they may be put to work inspheres well removed from biology. Insidecells, their usefulness in measuring cell turgorpressure has been demonstrated (15) and theymight be employed in the measurement ordetection of pressures generated in abiotic sys-tems. It has been suggested that they couldbe used to measure accelerations, decelera-tions, and the hardness of materials, from theobservation that their collapse can detect pres-sures developed momentarily in collisions(108). Their extremely high light-scatteringproperties may also have application in opticalfields.To date, the structure and functions of gas

vacuoles have been little investigated in thevarious bacterial groups, compared with thosein the algae, and parallel studies are clearlyrequired. The prokaryotic groups in which gasvacuoles have been discovered are widely sepa-rated on the phylogenetic tree and it may wellbe that they are of even more widespread oc-currence. Aquatic habitats, including the sea,from which no gas-vacuolate forms have yetbeen reported, would seem to be the best placeto look (13). Van Ert and Staley (98) have ex-pressed the view that gas-vacuolate bacteriaare common inhabitants of lake waters andattribute the fact that so few have been iso-lated to the inability of these bacteria to formgas vacuoles or even to grow in "typical" labo-ratory media. It is also possible that gas-vacu-olate forms have been overlooked in the scru-tiny of water samples by the standard sedi-mentation procedures, either because the cellsfloat or because their gas vacuoles are de-stroyed on addition of preservatives. Methodsof collection (such as sealing without an airspace in screw capped bottles) and concentra-tion (by filtration or centrifugation) which canresult in the generation of pressure on samplesshould also be avoided. Considering the re-cently popularized theory of the prokaryoticorigin of chloroplasts and mitochondria (11,79), perhaps we might apply the same precau-tions in looking for gas vacuoles in these or-ganelles of eukaryotic organisms (113). How-ever, it is doubted that they could ever occupya sufficient proportion of the total cell volumein these cells to provide useful buoyancy, andthis may have militated against their selection.

28 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

Finally, it is suggested that the informationwe have gathered on the stability of gas vesi-cles under various conditions (9) might also beemployed in their destruction. If gas vacuolesare so important to the success of planktonicblue-green algae which form waterblooms, wemight be able to control these nuisance orga-

nisms by collapsing their vacuoles. Pressuresgenerated by explosions have been found effec-tive in this respect (104), and field trials are inprogress (107); but it is hoped that fundamen-tal studies on these curious structures mightlead to less catastrophic solutions.

LITERATURE CITED1. Ahlborn, F. 1895. Uber die Wasserblute Byssus

Flos-aquae und ihr Verhalten gegen Druck.Verh. Naturwiss. Ver. Hamburg III 2:25.

2. Biezeno, C. B., and J. J. Koch. 1938. The buck-ling of a cylindrical tank of variable thick-ness under external pressure. Proc. Fifth In-ternat. Congress of Applied Mechanics.

3. Bisalputra, T., D. L. Brown, and T. E. Weier.1969. Possible respiratory sites in a blue-green alga Nostoc sphaericum as demon-strated by potassium tellurite and tetranitro-blue tetrazolium reduction. J. Ultrastruct.Res. 27:182-197.

4. Bowen, C. C., and T. E. Jensen. 1965. Blue-green algae: fine structure of the gas vacuoles.Science 147:1460-1462.

5. Bowen, C. C., and T. E. Jensen. 1965. Finestructure of gas vacuoles in blue-green algae.Amer. J. Bot. 52:641.

6. Branton, D. 1969. Membrane Structure. Annu.Rev. Plant Physiol. 20:209-238.

7. Brook, A. J. 1959. The waterbloom problem.Proc. Soc. Water Treat. Exam. 8:133-137.

8. Buckland, B. 1971. Studies on gas vacuoles inblue-green algae and bacteria. Ph.D. Thesis,University of London, England.

9. Buckland, B., and A. E. Walsby. 1971. A studyof the strength and stability of gas vesiclesisolated from a blue-green alga. Arch. Mikro-biol. 79:327-337.

10. Canabaeus, L. 1929. Uber die Heterocysten undGasvakuolen der Blaualgen und ihre Bezie-hung zueinander. Pflanzenforschung, Vol. 13.Jena.

11. Carr, N. G., and I. W. Craig. 1970. The rela-tionship between bacteria, blue-green algaeand chloroplasts p. 119-140. In, J. B. Har-borne (ed.), Phytochemical phylogeny. Aca-demic Press Inc., London.

12. Caspar, D. L. D., and A. Klug. 1962. Physicalprinciples in the construction of regular vi-ruses. Cold Spring Harbor Symp. Quant.Biol. 27:1-24.

13. Cohen-Bazire, G., R. Kunisawa, and N.Pfennig. 1969. Comparative study of thestructure of gas vacuoles. J. Bacteriol. 100:1049-1061.

14. Desikachary, T. V. 1959. Cyanophyta. IndianCouncil of Agricultural Research, New Dehli.

15. Dinsdale, M. T., and A. E. Walsby. 1972. Theinterrelations of cell turgor pressure, gas-

vacuolation and buoyancy in a blue-greenalga. J. Exp. Bot., in press.

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

17. Duysens, L. N. M. 1956. The flattening of theabsorption spectrum of suspensions as com-pared to that of solutions. Biochim. Biophys.Acta 19:1-12.

18. Eberly, W. R. 1967. Problems in the laboratoryculture of planktonic blue-green algae, p. 7-34. In Environmental requirements of blue-green algae. Federal Water Pollution ControlAdministration, Northwest Region, Corvallis,Oregon.

19. Findenegg, I. 1947. Uber die Lichtenspricheplanktischer Susswasseralgen. S. B. Akad.Wiss. Wien Math-Naturwiss. K. 155:159-171.

20. Fitzgerald, G. P. 1967. Discussion, p. 97-102. InEnvironmental requirements of blue-greenalgae. Federal Water Pollution Control Ad-ministration, Northwest Region, Corvallis,Oregon.

21. Fleischer, S., G. Brierly, H. Klouwen, and D. B.Stautterback. 1962. Studies of the electrontransfer system. XLVII. The role of phospho-lipids in electron transfer. J. Biol. Chem. 237:3264-3272.

22. Fogg, G. E. 1941. The gas-vacuoles of the Myx-ophyceae (Cyanophyceae). Biol. Rev. (Cam-bridge) 16:205-217.

23. Fogg, G. E. 1956. The comparative physiologyand biochemistry of the blue-green algae.Bacteriol. Rev. 20:148-165.

24. Fogg, G. E. 1965. Algal cultures and phyto-plankton ecology. University WisconsinPress, Madison.

25. Fogg, G. E. 1969. The physiology of an algalnuisance. Proc. Roy. Soc. Ser. B 173:175-189.

26. Fogg, G. E., and A. E. Walsby. 1971. Buoyancyregulation and the growth of planktonic blue-green algae. Mitt. Int. Ver. Limnol. 19:182-188.

27. Fritsch, F. E. 1945. The structure and reproduc-tion of the algae, vol. 2. Cambridge Univer-sity Press, Cambridge.

28. Fuhs, G. W. 1968. Cytology of blue-green algae:light microscope aspects, p. 213-233. In Al-gae, man and the environment. SyracuseUniversity Press, Syracuse, N.Y.

29. Fuhs, G. W., and N. Y. Albany. 1968. Interfer-enzmikroskopische Beobachtungen an denPolyphosphatk6rpern und Gasvakuolen von

Cyanophyceen. Osterr. Bot. Z. 116:411-422.30. Ganf, G. G. 1969. Physiological and ecological

aspects of the phytoplankton of Lake George,Uganda. Ph.D. Thesis, University of Lan-caster, England.

31. Geitler, L. 1932. Cyanophyceae. In L. Raben-

VOL. 36, 1972 29


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

horst's Kryptogamen-flora. Akademische Ver-lagsgesellschaft, Leipzig.

32. Hirsch, P., and S. H. Pankratz. 1970. Study ofbacterial populations in natural environmentsby use of submerged electron microscopegrids. Z. Allg. Mikrobiol. 10:589-605.

33. Holt, S. C. 1966. The comparative ultrastruc-ture of Sarcina maxima and Sarcina ventric-uli. Antonie van Leeuwenhoek J. Micro-biol. Serol. 32:451.

34. Houwink, A. L. 1956. Flagella, gas vacuoles andcell wall structure in Halobacterium hal-obium; an electron microscopic study. J.Gen. Microbiol. 15:146-150.

35. Jagger, J. 1967. Introduction to research in ul-traviolet photobiology. Prentice-Hall, NewJersey.

36. Jones, D. D. 1970. Isolation and characteriza-tion of gas vacuole membranes from Micro-cystis aeruginosa Kuetz. emend. Elenkin.Ph.D. Thesis, Michigan State University,East Lansing.

37. Jones, D. D., A. Haug, M. Jost, and D. R.Graber. 1969. Ultrastructural and conforma-tional changes in gas vacuole membranes iso-lated from Microcystis aeruginosa. Arch.Biochem. Biophys. 135:296-303.

38. Jones, D. D., and M. Jost. 1970. Isolation andchemical characterization of gas vacuolemembranes from Microcystis aeruginosaKuetz. emend. Elenkin. Arch. Mikrobiol. 70:43-64.

39. Jones, D. D., and M. Jost. 1971. Characteri-zation of the protein from gas vacuole mem-branes of the blue-green algae Microcystisaeruginosa. Planta 100:277-287.

40. Jost, M. 1965. Die ultrastruktur von Oscilla-toria rubescens D.C. Arch. Mikrobiol. 50:211-245.

41. Jost, M., and D. D. Jones. 1970. Morphologicalparameters and macromolecular organizationof gas vacuole membranes of Microcystisaeruginosa Kuetz. emend. Elenkin. Can. J.Microbiol. 16:159-164.

42. Jost, M., D. D. Jones, and P. J. Weathers. 1971.Counting of gas vacuoles by electron micros-copy in lysates and purified fractions of Mi-crocystis aeruginosa. Protoplasma 73:329-335.

43. Jost, M., and P. Matile. 1966. Zur Charakteri-sierung der Gasvacuolen der Blaualge Oscilla-toria rubescens. Arch. Mikrobiol. 53:50-58.

44. Jost, M., and A. Zehnder. 1966. Die Gasvaku-olen der Blaualgen Microcystis aeruginosa.Schweiz. Z. Hydrol. 28:1-3.

45. Klebahn, H. 1895. Gasvakuolen, ein Bestandteilder zellen der Wasserbluitebildenden Phy-cochromaceen. Flora (Jena) 80:241-282.

46. Klebahn, H. 1922. Neue Untersuchungen Uberdie Gasvakuolen. Jahrb. Wiss. Bot. 61:535-589.

47. Klebahn, H. 1925. Weitere Untersuchungenfiber die Gasvakuolen. Ber. Deut. Bot. Ges.43:143-159.

48. Klebahn, H. 1929. Uber die Gasvakuolen der

Cyanophyceen. Verh. Int. Ver. Limnol. 4:408-414.

49. Klug, A., and J. E. Berger. 1964. An opticalmethod for the analysis of periodicities inelectron micrographs, and some observationson the mechanism of negative staining. J.Mol. Biol. 10:565-569.

50. Kolkwitz, R. 1928. Uber Gasvakuolen bei Bak-terien. Ber. Deut. Bot. Ges. 46:29-34.

51. Koppe, F. 1923. Die Schlammflora der osthol-steinischen Seen und des Bodensees. Arch.Hydrobiol. (Plankt.) 14:619-762.

52. Krantz, M. J. 1971. Studies on gas vacuoles.Ph.D. Thesis. University of California,Berkeley.

53. Krinsky, N. I. 1966. The role of carotenoidpigments as protective agents against photo-sensitized oxidations in chloroplasts, p. 423-430. In T. W. Goodwin (ed.), Biochemistry ofchloroplasts. Academic Press Inc., London.

54. Kuentzel, L. E. 1969. Bacteria, carbon dioxideand algal blooms. J. Water Pollut. Contr.Fed. 41:1737-1747.

55. Kunisawa, R., and R. Y. Stainer. 1958. Studieson the role of carotenoid pigments in a chemo-heterotrophic bacterium Corynebacteriumpoinsettiae. Arch. Mikrobiol. 31:146-156.

56. Larsen, H. 1967. Biochemical aspects of ex-treme halophilism, p. 97-132. In A. H. Roseand J. F. Wilkinson (ed.), Advances in micro-bial physiology, vol. 1. Academic Press Inc.,London.

57. Larsen, H., S. Omang, and H. Steensland.1967. On the gas vacuoles of the halobacteria.Arch. Mikrobiol. 59:197-203.

58. Lauterborn, R. 1913. Zur Kenntnis einiger sap-ropelischer Schizomyceten. Allg. Bot. Z. 19:97-100.

59. Lauterborn, R. 1915. Die sapropelische Le-bewelt. Naturhist-Med. Vereins Z. Heidel-berg, N. F. 13:396-480.

60. Lee, B., and F. M. Richards. 1971. The inter-pretation of protein structures: estimation ofstatic accessibility. J. Mol. Biol. 55:379-400.

61. Lehmann, H., and M. Jost. 1971. Kinetics ofthe assembly of gas vacuoles in the blue-green alga Microcystis aeruginosa Kuetz.emend. Elekin. Arch. Mikrobiol. 79:59-68.

62. Lehmann, H., and M. Jost. 1972. Assembly ofgas vacuoles in a cell free system of the blue-green alga Microcystis aeruginosa Kuetz.emend Elenkin. Arch. Microbiol. 81:100-102.

63. Lemmermann, E. 1910. Kryptogamenflora derMark Brandenburg. Algen, vol. 1. Leipzig.

64. Lund, J. W. G. 1959. Buoyancy in relation tothe ecology of the fresh water phytoplankton.Brit. Phycol. Bull. 1(7):1-17.

65. Markham, R. 1968. Deviations in nitrogen me-tabolism associated with viruses, p. 203-216.In E. J. Hewitt and C. V. Cutting (ed.), Re-cent aspects of nitrogen metabolism inplants. Academic Press Inc.. London.

66. Marty, P. F., and F. Busson. 1970. Donneescytologiques sur deux Cyanophyc6es: Spiru-

30 WALSBY


mbr.asm

.org/D

ownloaded from


GAS VACUOLES

lina platensis (Gom.) Geitler et Spirulinageitleri J. de Toni. Schweiz. Z. Hydrol.32:559-565.

67. Molisch, H. 1903. Die sogennanten Gasvaku-olen und das Schweben gewisser Phycochro-maceen. Bot. Z. 61:47.

68. Munk, W. H., and G. A. Riley. 1952. Absorp-tion of nutrients by aquatic plants. J. Mar.Res. 11:215-240.

69. O'Flaherty, L. M., and H. K. Phinney. 1970.Requirements for the maintenance andgrowth of Aphanizomenon flos-aquae in cul-ture. J. Phycol. 6:95-97.

70. Petter, H. F. M. 1931. On bacteria of saltedfish. Proc. Acad. Sci. Amsterdam 34:1417-1423.

71. Petter, H. F. M. 1932. Over roode en anderebakterien van getzouten visch. DoctoralThesis, University of Utrecht, Holland.

72. Pfennig, N. 1967. Photosynthetic bacteria.Annu. Rev. Microbiol. 21:285-324.

73. Pfennig, N., and G. Cohen-Bazire. 1967. Someproperties of the green bacterium Pelodictyonclathratiforme. Arch. Mikrobiol. 59:226-236.

74. Pringsheim, E. G. 1966. The nature of pseudo-vacuoles in Cyanophyceae. Nature (London)210:549.

75. Reynolds, C. S. 1967. The breaking of theShropshire meres. Shropshire ConservationTrust Bull. 10:9-14.

76. Reynolds, C. S. 1969. The 'breaking' of themeres-some recent investigations. Brit.Phycol. J. 4:214.

77. Reynolds, C. S. 1971. Investigations on the phy-toplankton of Crose Mere, and other stand-ing waters in the Shropshire-Cheshire Plain.Ph.D. Thesis, University of London, Eng-land.

78. Rheinheimer, G. 1971. Mikrobiologie derGewAsser. Fischer Verlag, Jena. (micrographby P. Hirsch).

79. Sagan, L. 1967. On the origin of mitosing cells.J. Theoret. Biol. 14:225-274.

80. Shinke, N., and K. Ueda. 1956. A cytomorpho-logical and cytochemical study of Cyano-phyta. 1. An electron microscope study ofOscillatoria princeps. Mem. Coll. Sci. Univ.Kyoto, Ser. B 13:101-104.

81. Singh, R. N., and D. N. Tiwari. 1970. Frequentheterocyst germination in the blue-green algaGloeotrichia ghosei Singh. J. Phycol. 6:172-176.

82. Sirenko, L. A., N. M. Stetsenko, V. V. Arendar-chuk, and M. I. Kuz'menko. 1968. Role ofoxygen conditions in the vital activity of cer-tain blue-green algae. Mikrobiologiya 37:239-244.

83. Smith, R. V. 1969. The gas-vacuole membrane.Brit. Phycol. J. 4:215.

84. Smith, R. V., and A. Peat. 1967. Comparativestructure of the gas-vacuoles of blue-greenalgae. Arch. Mikrobiol. 57:111-122.

85. Smith, R. V., and A. Peat. 1967. Growth andgas-vacuole development in vegetative cells

of Anabaena flos-aquae. Arch. Mikrobiol. 58:117-126.

86. Smith, R. V., A. Peat, and C. J. Bailey. 1969.The isolation and characterization of gas-cyl-inder membranes and a-granules from Ana-baena flos-aquae D 124. Arch. Mikrobiol. 65:87-97.

87. Staley, J. T. 1968. Prosthecomicrobium andAncalomicrobium: new prosthecate fresh-water bacteria. J. Bacteriol. 95:1921-1942.

88. Steensland, H., and H. Larsen. 1969. A study ofthe cell envelope of the halobacteria. J. Gen.Microbiol. 55:325-336.

89. Stoeckenius, W., and W. H. Kunau. 1968. Fur-ther characterization of particulate fractionsfrom lysed cell envelopes of Halobacteriumhalobium and isolation of gas vacuole mem-branes. J. Cell Biol. 38:336-357.

90. Stoeckenius, W., and R. Rowen. 1967. A mor-phological study of Halobacterium halobiumand its lysis in media of low salt concentra-tion. J. Cell Biol. 34:355-395.

91. Strodtmann, S. 1895. Die Ursache desSchwebevermogens bei den Cyanophyceen.Biol. Zentralbl. 15:113-115.

92. Stryer, L. 1968. Implications of X-ray crystallo-graphic studies of protein structure. Annu.Rev. Biochem. 37:25-50.

93. Thomas, J. B. 1965. Primary photoprocesses inbiology. John Wiley, New York.

94. Tischer, R. G. 1965. Pure culture of Anabaenaflos-aquae A-37. Nature (London) 205:419-420.

95. Troitzkaja, 0. W. 1921. Sur la phototaxie et 1'-influence de la lumiere sur Oscillatoria Agar-dhii Gom. J. Soc. Bubs Russ. 6:121-136.

96. Van Baalen, C., and R. M. Brown, Jr. 1969.The ultrastructure of the marine blue-greenalga, Trichodesmium erythraeum, with spe-cial reference to the cell wall, gas vacuoles,and cylindrical bodies. Arch. Mikrobiol. 69:79-91.

97. Van Ert, M., and J. T. Staley. 1971. Gas-vacu-olated strains of Microcyclus aquaticus. J.Bacteriol. 108:236-240.

98. Van Ert, M., and J. T. Staley. 1972. A new gasvacuolated heterotrophic rod from freshwa-ters. Arch. Mikrobiol. 80:70-77.

99. Waaland, J. R. 1969. The ultrastructure anddevelopment of gas vacuoles. Ph.D. Thesis,University of California, Berkeley.

100. Waaland, J. R., and D. Branton. 1968. Ultra-structure of gas vacuole membranes in blue-green algae and halophilic bacteria. Amer. J.Bot. 55:709.

101. Waaland, J. R., and D. Branton. 1969. Gas vac-uole development in a blue-green alga. Sci-ence 163:1339-1341.

102. Waaland, J. R., and S. D. Waaland. 1970. Lightinduced gas vacuoles and their effect on lightabsorption by the blue-green alga Nostocmuscorum. J. Phycol. 6:Supplement 7.

103. Waaland, J. R., S. D. Waaland, and D.

31VOL. 36, 1972


mbr.asm

.org/D

ownloaded from


BACTERIOL. REV.

Branton. 1971. Gas vacuoles. Light shieldingin blue-green algae. J. Cell Biol. 48:212-215.

104. Walsby, A. E. 1968. An alga's buoyancy bags.New Sci. 40:436-437.

105. Walsby, A. E. 1969. The permeability of theblue-green algal gas-vacuole membrane togas. Proc. Roy. Soc. Ser. B 173:235-255.

106. Walsby, A. E. 1969. Studies on the physiologyof gas-vacuoles. Brit. Phycol. J. 4:216.

107. Walsby, A. E. 1970. The nuisance algae: curios-ities in the biology of planktonic blue-greenalgae. Water Treat. Exam. 19:359-373.

108. Walsby, A. E. 1971. The pressure relationshipsof gas vacuoles. Proc. Roy. Soc. Ser. B 178:301-326.

109. Walsby, A. E. 1971. Gas-filled structures pro-viding buoyancy in photosynthetic organisms.Symp. Soc. Exp. Biol. 26, in press.

110. Walsby, A. E. The isolation of gas vesicles fromblue-green algae. In S. Fleischer, L. Packer,and R. Easterbrook (ed)., Methods in enzy-mology, Academic Press Inc., New York, inpress.

111. Walsby, A. E., and B. Buckland. 1969. Isolationand purification of intact gas vesicles from ablue-green alga. Nature (London) 224:716-717.

112. Walsby, A. E., and H. H. Eichelberger. 1968.

The fine structure of gas-vacuoles releasedfrom cells of the blue-green alga Anabaenaflos-aquae. Arch. Mikrobiol. 60:76-83.

113. Whitton, B. A., N. G. Carr, and I. W. Craig.1971. A comparison of the fine structure andnucleic acid biochemistry of chloroplasts andblue-green algae. Protoplasma 72:325-357.

114. Whitton, B. A., and A. Peat. 1969. On Oscilla-toria redekei Van Goor. Arch. Mikrobiol. 68:362-376.

115. Wille, N. 1902. Ueber Gasvakuolen bei einerBakterie. Biol. Centralbl. 22:257-262.

116. Williams, W. T., and D. A. Barber. 1961. Thefunctional significance of aerenchyma inplants. Symp. Soc. Exp. Biol. 16:132-144.

117. Winogradsky, S. 1888. Zur Morphologie undPhysiologie der Schwefelbacterien. ArthurFelix, Liepzig.

118. Wright, W. D. 1964. Structural colours of bio-logical material. In Colour and life. Symposiaof the Institute of Biology, vol. 12.

119. Zimmermann, U. 1969. Okologische und phy-siologische Untersuchungen und der plank-tonischen Blaualge Oscillatoria rubescensD.C. unter besonderer Berucksichtigung vonLicht und Temperatur. Schweiz. Z. Hydrol.31: 1-58.

32 WALSBY


mbr.asm

.org/D

ownloaded from


vol. in u.s.a. structure and function of vacuolesalgal suspensions suggestedto himthat the gas was...

Documents