Abstract We describe here aspects of the anatomy oftwo “Epulopiscium” morphotypes, unusually large bacte-ria that are not yet cultured and that reproduce by the in-ternal generation of two or more vegetative daughter cells.Two morphotypes, A and B, which are enteric symbiontsof several species of herbivorous surgeonfish (Acanthuri-dae), were collected around the Great Barrier Reef ofAustralia, preserved there, and later stained for light mi-croscopy. Some samples were examined by electron mi-croscopy. In both morphotypes, countless discrete nucleo-plasms or nucleoids were found to occupy a single shal-low layer just beneath the surface all around these organ-isms. At each end of the morphotype B cells, a membrane-bound compartment containing dense cords of chromatinwas observed. When these were found at each end ofgrowing daughter cells, no polar compartments were thenfound in their mother organism. Electron micrographs ofsections of morphotype A symbionts show that their out-ermost region is composed of tightly packed coated vesi-cles, each surrounded by a thin, dense, spacious capsule.Near the surface of type A organisms the remains of bro-ken vesicles, broken capsules, and a finely fibrous matrixfuse to form a fabric that serves as the cell wall. Morpho-type B organisms, however, were observed to have a dis-tinct, morphologically continuous outer wall.

Key words ”Epulopiscium” · Nucleoids · Polar chromatin · Coated vesicles · Unusual composite wall · Daughter cells

Abbreviations DAPI 4′,6-Diamidino-2-phenylindole · PBS Phosphate buffered saline

Introduction

”Epulos” has been used as the name of members of a largeand varied group of uncommonly large bacteria (Clementset al. 1989). The first such organism was reported fromthe gut of the herbivorous surgeonfish Acanthurus nigro-fuscus from the Red Sea. It was given the name “Epulo-piscium fishelsoni” and was placed in the kingdom Protista(Fishelson et al. 1985; Montgomery and Pollak 1988).Various organisms sharing some of its properties werelater encountered in surgeonfish from Australia’s GreatBarrier Reef (Clements et al. 1989). Electron microscopyof sections of the large organisms showed that they con-tain typically prokaryotic strands of chromatin and havebacterial-type flagella. Their large size apart (some ofthem attain a length of more than 500 µm), these bacteriaare unusual in that they multiply (or merely rejuvenatethemselves) in a viviparous mode of internal formation ofone, two, or more daughter cells that escape to the outsidethrough a tear in the wall of the mother organism (Mont-gomery and Pollak 1988).

Sequence comparisons of the 16S rRNA gene (Angertet al. 1993) have placed “Epulopiscium” among the anaer-obic, gram-positive, spore-bearing clostridia. More re-cently, Angert et al. (1996) have reported that the unusu-ally large Metabacterium polyspora (Chatton and Pérard1913), which is a symbiont of the cecum of guinea pigsand other rodents and which forms two or more refractileendospores per cell, is also located in this branch of thegenealogical tree. Spore-bearing cells of Metabacteriumspp. superficially resemble “Epulopiscium” morphotypesengaged in the formation of vegetative daughter cells. Herewe describe observations made with the light microscopeon nucleoids and other forms of chromatin in “Epulopis-cium” morphotypes. A brief account is also provided ofelements of the complex outermost layer or “cortex” of typeA organisms as seen in electron micrographs of sections.

Carl Robinow · Esther R. Angert

Nucleoids and coated vesicles of “Epulopiscium” spp.

Arch Microbiol (1998) 170 :227–235 © Springer-Verlag 1998

Received: 3 December 1997 / Accepted: 11 June 1998

ORIGINAL PAPER

C. Robinow (Y)Department of Microbiology and Immunology, Health Sciences Centre, University of Western Ontario, London, Ontario, Canada N6A 5C1e-mail:crobinow@mni.uwo.ca

Tel. +1-519-661-3427; Fax+1-519-661-3499

E. R. AngertDepartment of Molecular and Cellular Biology, Harvard University, Cambridge, MA 02138, USAe-mail: eangert@biosun.harvard.eduTel.+1-617-495-0532; Fax+1-617-496-4642

Materials and methods

Sample collection

Numerous samples of gut contents of surgeonfish were collectedon several occasions on Lizard Island (a part of the Great BarrierReef, Australia) by K.D. Clements. Samples were preserved therein several different ways appropriate for 4′,6-diamidino-2-phe-nylindole(DAPI)-enabled fluorescence of DNA, the staining ofbacterial chromatin for light microscopy and electron microscopy(see below).

DAPI fluorescence

Samples of gut contents were fixed and stored in 80% ethanol in1993 and were recently prepared for the detection of DNA in thefollowing manner. Fixed samples were applied to poly-L-lysine-coated coverslips and air-dried. The coverslips were then im-mersed in a solution of 0.5 µg/ml DAPI in phosphate-bufferedsaline (PBS) for approximately 2 min; subsequently, they wererinsed with PBS. The coverslips were next mounted over a PBS/50% glycerol solution. Slides were viewed and photographed us-ing an Olympus BX 60 epifluorescence microscope as previouslydescribed by Harry et al. (1995). Cells were visualized using an ex-citation cube unit (U-MNU) appropriate for viewing DAPI fluo-rescence with a narrow band-pass (360–370 nm) excitation filter anda long band-pass (420 nm) barrier filter. Photographs of fluores-cent preparations were obtained on Kodak T-MAX film, ASA 400.

Transmission light microscopy

The fixative for light microscopy of the large bacteria from sur-geonfish was a mixture of two parts of a saturated aqueous solu-tion of mercuric chloride with one part of absolute alcohol andthree parts water with acetic acid added to the mixture to make it2% of the total volume. Fixation time was 10–15 min, after whichthe sample was washed with and stored in 70% ethanol. Nucleoidstaining was achieved by both the Feulgen procedure and the HCl-Giemsa method (Piekarski 1937). Hydrolysis for 8–10 min with 1M HCl at 60°C was followed by staining with the Giemsa mixtureof dyes. We used Gurr’s Giemsa “Improved 66” (British DrugHouses, Pool, England) at one drop of stain per ml of distilled wa-ter that had been given a pH of 6.8 by the use of Gurr’s buffertablets (British Drug Houses). The progress of staining waschecked with a × 40 water immersion lens. Photographs weretaken of wet, stained preparations mounted over buffer of pH 6.8.

Electron microscopy

Samples intended for electron microscopy were prepared at the siteof collection in 2.5% glutaraldehyde/0.2 M cacodylate buffer (pH7.2). Samples were fixed for 30–45 min, then washed with andstored in 0.2 M cacodylate (pH 7.2) at 4°C. On arrival at the Lon-don laboratory, the deposit of centrifuged sample was gelled inNoble agar and post-fixed for 1 h in 1% OsO4 in 0.2 M cacodylate(pH 7.2), washed with water, and dehydrated via a series ofethanol solutions of a concentration gradually increasing from 30to 100%. Dehydrated samples in agar were next placed into Spurrresin that was polymerized overnight at 60°C. Sections were cut ona hand-operated Porter Blum microtome, were stained for 5 minwith uranyl acetate (2%) followed by 1 min in 1.4% lead cit-rate/1.8% sodium citrate, and were then examined in a Philips 300electron microscope. The fixation procedures just described haveroutinely provided satisfactory images of the membranes, nu-cleoids and ribosomes of bacteria studied in our laboratory, but theinterval between primary fixation of the giant symbionts in Aus-tralia and secondary fixation with osmium 5 weeks later proved tohave been too long. Only in sections of the A morphotype was de-

tail found to have been passably well-preserved, and this only inthe outermost regions of these symbionts.

Results

The A morphotypes in our samples measured 200–250µm in length, were slender, and tapered towards their tips.Some of them bore one or two daughter cells, mostly atthe very early stages of development. B morphotypeswere only 100–185 µm long; they were straight cylinderswith rounded ends or were of slender cigar shape. Manyof these organisms bore one or two daughter cells.

Nucleoids of morphotype B

As Fig. 1 (A–E) and Fig. 2 (A,B) clearly show, B morpho-types have innumerable, closely-packed, small nucleoidsspaced all over their periphery in a shallow layer just be-neath their surface. In DAPI preparations, brightly fluo-rescing polar compartments were invariably found at thetips of free, mature individuals or in the tips of internallyarising offspring. These will be discussed below. At thehigher magnification of the Giemsa-stained preparation inFig. 3, we found that the nucleoid layer is composed oflarge numbers of thin, flexible entities, nucleoids or nu-cleoplasms, packed together in single-layered, flat patchesof widely differing size and shape. Neighboring patchesare connected with strands or narrow ribbons also com-posed of nucleoids. Overall, the arrangement here is thatof a reticulate sheet – to borrow a term used by Smith(1956) to describe the structure of the chloroplast of thegreen alga Oedogonium. The nucleoids of B symbiontsvary in length and on the whole appear to be somewhatsmaller than those of common Bacillus species. AfterFeulgen-style acid hydrolysis, the nucleoids of B morpho-types displayed an affinity for purple components of theGiemsa stain, as in the case of other bacteria, but they didso rather weakly. Oddly enough, they also proved to beonly weakly, if quite distinctly, Feulgen-positive in prepa-rations in which the nuclei of protozoa, which were oftenscattered among the giant symbionts, gave the expectedstrongly positive response to the Feulgen procedure. Inretrospect, it seems possible that a shorter time spent inacid hydrolysis would have evoked a stronger positive re-sponse of the B-type nucleoids to the Feulgen treatment,particularly since DAPI scans showed the nucleoids of Bsymbionts fluorescing as brightly as the nucleoids of otherbacteria contained in the same samples.

Densely chromatinic cords in membrane-enclosed space at the tips of B-type symbionts

The weakly positive response of B morphotypes to theFeulgen procedure is all the more remarkable because inour samples, mature singles or pairs of daughter cells stillenclosed in their mother organism displayed, in the ma-

228

jority of instances, at both poles small lens-shaped com-partments containing several cords (which may be part ofa single knot) that were strongly Feulgen-positive andalso stained deeply after HCl-Giemsa (Fig. 4, A–C). Weshall refer to these structures as terminal or polar chro-matin compartments. The cords of these compartmentsare much larger and denser than individual nucleoids ofthe superficial net shown in Fig. 2B. We believe that Fig.

5D of Montgomery and Pollak (1988) of a stout, twistedcord of nucleoplasm in the center of what the authors re-gard as a developing daughter cell may represent part ofa knot of cords of chromatin within its compartmentsimilar to what we have observed at the tips of B morpho-types.

We can but speculate what may be the function of thisorganelle that is generally present at both poles of the Bmorphotype. Aware that daughter cells of some morpho-types of “Epulopiscium” arise in the tips of mother cells[see, e.g., Fig. 6 of Montgomery and Pollak (1988)], wepropose that the terminal compartments with their promi-nent cords of chromatin may be primordia of daughtercells. Among the DAPI-stained B-morphotype cellsscanned, many harbored long, slender daughters that in-variably carried strongly fluorescing chromatin compart-ments at both ends, while no such compartments werefound in the remaining cytoplasmic space of their motherorganism even when, as in Fig. 1C, the mother cell’s ownsuperficial nucleoids were still vividly fluorescing. That iswhat would be expected if, as suggested above, activatedterminal compartments transform themselves into daugh-

229

Fig. 1A–E Epifluorescence of B morphotypes stained with 4′,6-diamidino-2-phenylindole. A and B show median- and upper-sur-face optical sections, respectively, of the same cell. A The layer ofnucleoids just under the cell cortex is seen as a bright broken line.DNA associated with putative daughter cell primordia is seen atboth poles. B A surface view of the network of islands of nucleoids(A,B bars 50 µm). C and D show examples of B morphotypes withmature daughter cells. Note the fine broken line of mother cell nu-cleoids surrounding the brighter daughter cell nucleoid layer. CTwo daughter cells in this image show coalesced DNA at the polesand at regions along the side of each daughter cell. These may rep-resent “granddaughter” primordia. D The unusual instance of asingle daughter cell. Coalesced DNA is seen at the poles of thedaughter cell but not in the mother cell. E A surface view of a lat-eral chromatin cluster of cords of DNA (C–E bars 20 µm)

ter cells. For direct evidence of this proposed course ofevents it would, however, be necessary to examine B-types from much earlier phases of the growth cycle thanthose represented in the samples that have been availableto us.

The DAPI preparations have also revealed instances ofcells that had, in addition to polar chromatin compart-ments, one or more brightly fluorescing clusters of chro-matinic cords along their sides (Fig. 1, C and E). Such lat-eral clusters could perhaps account for certain B morpho-types mentioned by Clements et al. (1989) that harbormore than two daughter cells. A look at 3-year-old Feul-gen preparations in London revealed there, too, three in-stances of solidly Feulgen-positive, lateral chromatinclusters among twenty B organisms. DAPI fluorescencealso drew our attention to the fact that even fully growntype-B twin daughter cells often continue to cling togetherfor some time within the faintly illuminated remains ofthe cell wall of their mother organism. Such remains are

seen clinging to much of the surface of the morphotype Bof Fig. 2B.

Nucleoids of morphotype A

In A morphotypes, the nucleoids were evenly distributedover the entire subsurface plane of the cells (Fig. 5, A andB). The nucleoids were not arranged there in any particu-lar order, but were a remarkably uniform distance fromeach other and quite possibly exist as distinct and separateentities; differences in the size of individual nucleoidsmay reflect their having been preserved at different stagesof their replication cycle. Near the tip of A symbionts, nu-cleoids tend to be more closely packed together than overthe rest of the body of these organisms, and some may ex-tend from the plane of the main nuclear layer and enterdeeper regions of the cytoplasm. The nucleoids werestrongly Feulgen-positive, and most of them were slightlylarger than nucleoids of cells from fast growing culturesof Bacillus mycoides and Bacillus megaterium [see Robinowand Kellenberger (1994)].

Chromatin of internally generated daughter cells

The developing type-A daughter cells we have encoun-tered were of two kinds. There were those whose chro-matin presented as a tubular shell of seemingly solid chro-matin perforated by large irregular holes (Fig. 5B). In the

230

Fig. 2A, B Morphotype B (HCl-Giemsa). Views of the same organism at two different levels of focus. A Median optical sec-tion. Note the straight cell wall of even density. The nucleoids ap-pear as a line of pieces of chromatin just beneath the cell wall. Thedark, lens-shaped body at the tip is of a kind regularly found atboth poles of growing and of mature B morphotypes. Normallythese compartments contain stout cords of chromatin (see Fig. 4,A–C). However, during the short interval between taking the pho-tographs of A and B, the dark stain diffused away from the com-partment and revealed in B – for reasons unknown – that the com-partment was empty. B Overview of the superficial reticulate sheetof nucleoids. The shreds of membranous materials clinging to thiscell are probably remains of the wall of its mother organism (bars5 µm)

Fig. 3 Morphotype B (HCl-Giemsa). Superficial net of clusters oftightly packed nucleoids. The group indicated with an arrow is inbest focus (bar 5 µm)

other kind, which we regard as having been in a more ad-vanced state of development, the chromatin was disposedin much the same way as it was in the surface layer of themother cell except that the daughter’s chromatin tended tostain more deeply than that of its mother organism. Thefluorescence of the chromatin of B morphotypes harbor-ing juvenile cells seemed approximately equally bright inmother organisms and offspring (Fig. 1C).

Light microscopy of the structure of the surface of B and A morphotypes

Morphotypes B and A differ markedly in the nature of theboundary between themselves and their environment. Op-tical sections such as that of Fig. 2A show that the type-Bsymbiont is surrounded by a proper wall of geometricalneatness and unvarying density. A narrow, translucentzone intervenes between the wall and the nucleoid layer.In contrast to this unexceptional pattern, we find the Amorphotypes of Fig. 5 (A and B) surrounded by a thick“cortex”, to borrow a term adopted by Montgomery andPollak (1988) for the membranous outer boundary of “E.fishelsoni” as a sign that they found it differed signifi-cantly from the cell walls of plants and bacteria. Note thatin our usage “cortex” will refer to all of the region exteriorto the layer of nucleoids. The type-A cortex of Fig. 5 (Aand B) seems to be of the nature of a soft rind that lacksthe sharply defined outer and inner contours of a properwall, varies slightly in thickness along the length of thecell, and is not set off by a distinct gap from the nucleoidlayer. The complexity of this cortex was revealed by elec-tron microscopy (Figs. 6, 7).

231

Fig. 4A–C B morphotypes A,C HCl-Giemsa and B Feulgen. Examples of polar compartments containing stout cords of chro-matin (bars 5 µm)

Fig. 5A, B Two A morphotypes (Feulgen). A Surface view ofshallow layer of evenly spaced nucleoids. Note along both edgesthe rind-like, thick cortex. Out of focus in the interior is the densechromatin of a developing daughter cell. B Optical section of an-other A morphotype. In the polar region some nucleoids have en-tered deeper layers of the cytoplasm. Over the rest of the cell thenucleoids are still seen as forming a shallow layer just beneath thethick, translucent cortex. In the interior is seen the fenestrated tubeof dense chromatin of a developing daughter cell (bars 5 µm)

Electron microscopy of the cortex of A morphotypes

An overview of the periphery of a typical A morphotypeis provided in Fig. 6. Up from the micrograph’s loweredge are seen profiles of nucleoids of low density and ir-regular contours similar to what has been found in E. colipreserved only with gluteraldehyde, where, according toHobot et al. (1985), “… coarse aggregates of DNA werepresent within a seemingly empty dispersed nucleoid.”Continuing upwards, three levels of the outer cortex canbe seen. Prominent is an extensive layer of labyrinthinedesign reminiscent of the maze of interconnected spacesfound by Clements and Bullivant (1991) just inside asymbiont’s cell wall. Its upper reaches here are formed bya shallow layer of open spaces whose lining membranesare much denser than those of the main labyrinth below.Higher still, a region of stacked ovoid or spherical struc-tures is observed, and beyond these a layer of tightlypacked, ill-defined, perhaps fibrous materials are seen. Fi-nally, exterior to this blurred horizon, a mat of bent and

broken filaments are observed; these have also been en-countered at the periphery of “Epulopiscium” morpho-types studied by Clements and Bullivant (1991) and, onthe basis of their fine-structure, have been identified bythem as bacterial flagella.

Figure 7 shows a stretch of cortex similar to that of theprevious figure but at higher magnification. What is seenin the higher levels of this micrograph and was found reg-ularly in other sections of long stretches of uniformlystructured cortex of A morphotypes, we regard as profilesof stacks of spherical or ovoid vesicles, each one enclosedwithin a spacious, thin, dense membranous capsule.Finely fibrous material appears to fill the space betweenneighboring capsules. The vesicles within, with diametersin the order of 0.11 µm, we regard as belonging to the“decorated” or “bristle-coated” kind familiar to cell biolo-gists. The closely packed decorated vesicles resemble the“fluid segregating organelles” of Paramecium (McKanna1976). These take the form of fascicles of narrow tubules,blind at one end, that are situated proximal to the canals

232

Fig. 6 Electron micrograph ofa section of the cortex of an-other A morphotype. Thecloud-like, faintly speckledshapes near the bottom are pro-files of nucleoids (n). Abovethese is a labyrinth of intercon-nected spaces. Further up thereis yet another labyrinth wherethe open spaces are lined withmuch denser membranes (d)than are those of the mainlabyrinth below. The outer bor-der of the cortex appears to beformed by roughly circularprofiles of irregularly stackedvesicles (bar 1.0 µm)

conducting fluid to the pulsating vacuoles. The con-stituent tubules of the fascicles bear an array of minutepegs that endow cross-sections of them with a strong like-ness to the bristle-coated vesicles of the outer cortex ofthe A morphotype.

Discussion

Mode of reproduction of the A morphotype

The A morphotypes, like “E. fishelsoni” itself, reproduceby the internal generation of daughter organisms (Fishel-son et al. 1985; Montgomery and Pollak 1988; Clementset al. 1989), but to date we have not found out where thisprocess initiates. Most probably the development ofdaughter morphotypes starts at some point of the superfi-cial nucleoid layer with its profusion of discrete nucleo-plasms, but unambiguous evidence of such an event hasnot been encountered in our samples. Among A morpho-types are found many that bear a single, short daughtercell in the cytoplasm of the middle region of the motherorganism. Thus, a count performed on two Feulgen-stained slides of A morphotypes from the same sampleyielded 27 that had single daughters varying from 28 to 40µm in length and only 13 that bore two, long, overlappingdaughter cells. As to the origins of these pairs, therewould seem to be two possibilities. One scenario wouldbe that of two short daughter cells of the same length,hence probably initiated at the same time (of which wehave seen examples), lying some distance apart, bothgrowing, eventually overlapping, and occupying most ofthe available space. Another conceivable origin of over-lapping pairs would be their having arisen from a single

short cell that had grown to great length and had, in theend, divided by binary fission before extrusion from themother organism. To allow for the possibility of binaryfission of a single internally generated daughter cellwould not be unreasonable. It has, in fact, been observedto take place in the cells of long, segmented filamentousbacteria attached to the lining of the ileum of mice (Fer-guson and Birch-Andersen 1979). Then again, two ofMontgomery and Pollak’s (1988) striking electron micro-graphs show a single, snake-like cell emerging from a tornwall of the mother organism. If these scenes had been partof a process of reproduction, then binary fission would,presumably, in such instances have taken place soon afterextrusion. If not, what is observed here are instances ofmere rejuvenation and not propagation since the motherorganism is destroyed during the process. Uncertain isalso the life story, past and future, of the occasionally en-countered instances of B morphotypes bearing a singledaughter within a mother organism whose own nucleoidlayer is still visible outside that daughter cell, as in Fig. 1D.

Coated vesicles of “E. fishelsoni”

The A morphotypes from surgeonfish around the GreatBarrier Reef (Australia), are not the only ones that areequipped with coated or decorated vesicles. In electronmicrographs accompanying their ground-breaking paperon “E. fishelsoni,” the giant enteric symbiont of surgeon-fish of the Red Sea, Montgomery and Pollak (1988) de-scribe “reticulate membranes and tubules at the cell pe-riphery.” A close look at their work has convinced us thatthe structures referred there to are what used to be called“alveolate” vesicles or tubules. Some vesicles there have

233

Fig. 7 High magnification of ashort stretch of the outermostregion of the cortex showsstacks of the characteristic pro-files (thin arrows) of “deco-rated” or “bristle-coated” vesi-cles, each one surrounded bythe profile of its thin but dense,spacious membranous capsule.Below, thick arrowheads pointto vesicles preserved in theprocess of being encapsulated(bar 0.2 µm)

a spiny circumference. Both aspects have been describedby Leedale (1967) in his writing about alveolate vesiclesnear the pulsating vacuole of Euglena spirogyra. As heput it: “Alveolate vesicles are distinctive structures, theirwalls carrying a well defined alveolate patterning whichin sectioned vesicles appears as hairs radiating from thesurface”. We are aware that in current work in the field ofmolecular cell biology, alveolated/reticulated vesicleswould be referred to as being “clathrin-coated,” but forthe time being we feel it appropriate to refer to such struc-tures in the manner of the authors whose work we are herediscussing.

Coated vesicles are known to perform a variety offunctions not only in protists but also in the cells of cer-tain tissues of plants and animals. Electron micrographsdo not reveal what coated vesicles contribute to the main-tenance of the giant bacteria. Tentatively we propose thattheir function is excretory, that in A morphotypes newvesicles are steadily produced at the base of the vesiclezone (Fig. 7) and are gradually pushed up to more periph-eral levels and disintegrate there. The employment of thebristle-coated kind of vesicles for the removal of wasteproducts may be a strategy that allows the A morphotypeto overcome obstacles that Koch (1996) sees as prevent-ing bacteria from attaining large size. The continuous,outward-moving supply of broken vesicles and fragmentsof their membranous capsule, all seemingly embedded ina fibrous matrix, may provide the cohesion required of alayer that has to function as the equivalent of a cell wall.

Similar events seem to be happening in “E. fishelsoni.”Montgomery and Pollak (1988) mention that the numer-ous tubules of this symbiont “frequently touch the innercortical surface” (i.e., the inner surface of the organism’sbounding membrane), and so they do. But close scrutinyof the relevant electron micrographs has provided clearevidence that at several points the reticulated/alveolatedtubules actually fuse with and enter the substance of thecortex and (perhaps only temporarily) impart to stretchesof it an alveolar pattern. Here, then, is another example ofvesicle/tubule-coating materials contributing somethingto the building of a structure capable of performing thefunctions of a wall. The course of events proposed here astaking place near the surface of two kinds of enteric sym-biont morphotypes may well seem improbable, but (inmorphological terms) it resembles the dynamics of thesteady loss of cell wall materials to the outside and theirrenewal from below that has been demonstrated in gram-positive, rod-shaped bacteria (Koch and Doyle 1985;Beveridge and Kadurugamuwa 1996). We have not ob-tained unambiguous electron micrographs of B morpho-types, but have looked at enough of them to allow us tosay that the outer region of the cortex there is not com-posed of coated vesicles but of more sophisticated durablestructures and that B morphotypes have a distinct, two-ply, coherent wall.

The samples of intestinal symbionts we examined werefrom those segments of their life cycle through which theypass during the hours of daylight. Thus, the few observa-tions that we have been able to make leave much of the

life cycle of these giant symbionts unaccounted for. Wehope that others will collect over a stretch of 24 h a set ofanatomical/cytological observations on “E. fishelsoni” andsome of its morphotypes that would complement the cir-cadian set of numerical data on growth and reproductioncollected by Montgomery and Pollak (1988). Meanwhile,it is obvious that the nucleoids of the symbiont morpho-types we studied resemble those of common bacteria,those of type A more obviously so than the strangelysmall ones of the type-B symbionts. The validity of oursuggestion that the polar compartments containing stoutcords of chromatin may be primordia of daughter cells de-pends on the proof or disproof that can only be providedby future life cycle studies. Until this information is avail-able, we find support in the fact that, despite their largesize, these polar compartments resemble the passing stagein the development of spores in Bacillus cells in whichdistinct and variously coiled cords of chromatin are foundwithin a small compartment at the tip of rod-shaped cells(Robinow 1960). Lastly, while the task of alveolate vesi-cles and tubules of “E. fishelsoni” and bristle-coated vesi-cles of the A-type cortex can at present only be guessedat, it is of interest to have found that the employment ofsuch organelles is not limited to eukaryotes.

Acknowledgements The authors are much indebted to K. D.Clements, now at the University of Auckland (New Zealand), forcatching surgeonfish and sending us samples for genetic analysisand microscopy. For electron micrographs, the authors thank D.Moyles, now at the University of Guelph (Ontario, Canada). C.Robinow owes thanks to R. G. E. Murray for putting him on thetracks of E. Angert’s work with the giant bacteria.

References

Angert ER, Clements KD, Pace NR (1993) The largest bacterium.Nature 362:239–241

Angert ER, Brooks AE, Pace NR (1996) Phylogenetic analysis ofMetabacterium polyspora: clues to the evolutionary origin ofEpulopiscium spp. the largest bacteria. J Bacteriol 178 :1451–1456

Beveridge TJ, Kadurugamuwa JL (1996) Periplasm, periplasmicspaces and their relationship to bacterial wall structure: novelsecretion of selected periplasmic proteins from Pseudomonasaeruginosa. Microb Drug Resistance 2:1–8

Chatton É, Pérard C (1913) Schizophytes du caecum du cobaye. 2.Metabacterium polyspora n.g., n.s. C R Hebd Soc Biol (Paris)74:1232–1234

Clements KD, Bullivant S (1991) An unusual symbiont from thegut of surgeonfishes may be the largest known prokaryote. J Bacteriol 173:5359–5362

Clements KD, Sutton DC, Choat JH (1989) Occurrence and char-acteristics of unusual protistan symbionts from surgeonfishes(Acanthuridae) of the Great Barrier Reef, Australia. MarineBiol 102:403–412

Ferguson DJP, Birch-Andersen A (1979) Electron microscopy of afilamentous, segmented bacterium attached to the small intes-tine of mice from a laboratory animal colony in Denmark. ActaPathol Microbiol Scand B 87:247–252

Fishelson L, Montgomery WL, Myrberg AA (1985) A uniquesymbiosis in the gut of tropical herbivorous surgeonfish (Acan-thuridae: Teleostei) from the Red Sea. Science 229:49–51

Harry EJ, Pogliano K, Losick R (1995) Use of immunofluores-cence to visualize cell-specific gene expression during sporula-tion in Bacillus subtilis. J Bacteriol 177:3386–3393

234

Hobot JA, Villiger W, Escaig J, Maeder M, Ryter A, KellenbergerE (1985) Shape and fine structure of nucleoids observed onsections of ultrarapidly frozen cryosubstituted bacteria. J Bac-teriol 162:960–971

Koch AL (1996) What size should a bacterium be? A question ofscale. Annu Rev Microbiol 50:317–348

Koch AL, Doyle RJ (1985) Inside-to-out growth and the turnoverof the gram-positive rod. J Theor Biol 117:137–157

Leedale GF (1967) Euglenoid flagellates. Prentice Hall, Engle-wood Cliffs, NJ

McKanna JA (1976) Fine structure of the fluid segregation or-ganelles of Paramecium contractile vacuoles. J Ultrastruct Res54:1–10

Montgomery WL, Pollak PE (1988) Epulopiscium fishelsoni n.g.,n.s., a protist of uncertain taxonomic affinities from the gut ofan herbivorous reef fish. J Protozool 35:565–569

Piekarski G (1937) Zytologische Untersuchungen an Paratyphus-und Coli Bakterien. Arch Mikrobiol 8:428–439

Robinow CR (1960) Morphology of bacterial spores, their devel-opment and germination. In: Gunsalus IC, Stanier RY (eds) Atreatise on structure and function. Academic Press, New YorkLondon, pp 207–248

Robinow CR, Kellenberger E (1994) The bacterial nucleoid revis-ited. Microbiol Rev 58:211–232

Smith GM (1956) Cryptogamic botony, vol 1, 2nd edn. McGrawHill, New York

235