INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

In 1985, an unusual unicellular organism was reported from the gut of brown surgeonfish, Acanthurus nigrofuscus (Acanthuridae: Teleostei), in the Red Sea (14). These organisms were named Epulopiscium fishelsoni and tentatively placed in the kingdom Protoctista by Montgomery and Pollak (23) on the basis of their maximum size (>600 µm), great variation in length (~30 to >600 µm) and volume (>2,000-fold) both within single hosts and during a daily cycle, daily variability in nucleoid and cytoplasm structure, and a peculiar mode of reproduction (daughter cells form within parental organisms and eventually emerge as mobile cells from the maternal envelope). Similar and apparently related organisms representing at least 10 different morphotypes were later collected from surgeonfishes from the Hawaiian Islands, French Polynesia, Tuvalu, Guam, southern Japan, Papua New Guinea, the Great Barrier Reef, and South Africa (8, 13a, 15a, 22, 24a). None of these organisms (here termed "epulos" to distinguish them from other groups of bacteria) has been cultured, so all descriptions of distribution or functions within hosts and descriptions of diel changes in size or stages in the life cycle are based on samples from populations of epulos collected from hosts sacrificed at different times of day and night.

Initial electron microscopy of epulos revealed a complex ultrastructure lacking standard eukaryotic organelles (8, 14, 23). Following suggestions by Clements and Bullivant (9) that the fine structure of cytoplasm, cell wall, and flagella indicated that epulos were in fact prokaryotic, Angert et al. (2) isolated and sequenced the gene encoding the 16S rRNA subunit, placing these giant microorganisms in a group of low-G+C gram-positive bacteria related to Clostridium. In situ hybridization with fluorescein-labelled oligonucleotide probes based on cloned rRNA sequences confirmed the source of the rRNA gene. E. fishelsoni represents, therefore, the largest bacterium so far described (2, 9).

Cell size also varies more than in other bacteria. If one estimates the volume of a cigar-shaped E. fishelsoni cell as roughly the volume of two cones, each with height and radius of the base equal to half of cell length and half of its maximal diameter, respectively (V = 2/3r²h), the volume of a very large E. fishelsoni (~354,000 µm³ [500 by 52 µm]) is approximately 3,000-fold greater than that of a very small E. fishelsoni (~125.6 µm³ [30 by 4 µm]). The volume of a large epulo can, therefore, exceed the volume of a bacterium such as Escherichia coli (~2 µm³ [1 by 2 µm]) (25) by at least 5 orders of magnitude.

Cellular or molecular mechanisms that may support and control such exceptional variability in dimension, volume and surface/volume ratios of E. fishelsoni are unclear. However, cells are highly mobile, vary in mean size and structure during a 24-h period, affect the pH of the host's gut fluids differentially during day and night (suggesting metabolic changes on a diel cycle), and construct mobile daughter cells within the parental cell (9, 14, 23, 24). The active metabolism and corresponding expectation of genomic activity supporting synthesis of macromolecules led us to predict correlations between DNA quantity or condensation state and cell size or stage in the daily life cycle.

Because E. fishelsoni and related organisms have not been cultured, we relied on collections of cells from host fishes sacrificed at different times of day and night, fluorescence cytochemistry, microfluorometry, and transmission electron microscopy to describe changes in the functional state and distribution of DNA and other cell constituents during the microbe's life cycle. Our primary objective was to study relationships between DNA quantity and cell volume, as well as possible changes in DNA distribution and functional activity of the nucleoid during the life cycle of E. fishelsoni. In this paper, we initially demonstrate that DNA quantity is directly proportional to cell volume rather than to length or surface area, and that large quantities of DNA are almost equally divided and distributed to developing daughter cells. We then describe how coordinated changes in DNA condensation state and distribution, nucleoid expansion, and cell wall deposition lead to the formation of daughter cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Specimens of the host surgeonfish, A. nigrofuscus, were collected by net or hand spear near the Interuniversity Institute, Steinitz Marine Biological Laboratory, Eilat, Red Sea, Israel (see reference 21 for details of the site and the collection techniques). Because surgeonfish are active only during the day, host fish intended for sampling at night were held in flowing seawater prior to sacrifice; samples obtained during the day were taken shortly after fish were collected. Epulos have not been cultured, so each fish provided a single sample of an epulo population at the time of sacrifice.

Specimens were sacrificed by a blow to the head and immediately dissected. E. fishelsoni cells were collected from the central intestine of the fish within ~15 min of host sacrifice. Cells were either examined live at the marine laboratory or were fixed (in absolute methanol for fluorometry and in other fixatives, described below, for light and electron microscopy) at various times of day and night for subsequent analysis at our home institutions.

Lengths of epulos varied within a single host fish, and epulo size distribution changed with time of day or night. In the field, we used an ocular grid in a Zeiss binocular microscope at ×100 to assign live cells collected at different times to ~50-µm-interval length categories (<50 µm, 50 to 101 µm, and so on [see Table 1]). Sixty haphazardly chosen cells from each of two fish were measured per time period (total n = 120 per period). Because we were unable to measure rapidly moving cells precisely with the grid available to us at our field laboratory, we created a frequency distribution among categories and calculated an approximate mean length for each time period, assuming that all cells in a particular category were the median length for that category (see Table 1).

Preserved cells intended for fluorometry (see below) were returned to Tel Aviv University. There, cells falling into several narrow size ranges (5 to 10 µm for cells <250 µm long, 20 to 30 µm for cells >300 µm long [see Table 2]) were selected from samples. Samples representing different collection times were treated independently. Initially, six size categories (groups I through VI of Table 2) were established, such that cells in each of groups II through VI had volumes approximately twice those of cells in the preceding smaller group (see Table 2). A seventh group (group 0) was later added to expand the size of range of surveyed cells and included specimens of the smallest easily assayed E. fishelsoni (~30 µm long). Cells with lengths between ~30 and ~150 µm were excluded due to limitations in equipment available to make precise measurements.

Because epulos collected at different times of day and night show marked variability in the distribution and condensation state of DNA and in the size and location of nucleoids, initial measurements of total DNA in nucleoids were made on cells from a sample taken at approximately 0800 h. Nucleoids in this material were compact, round in shape, and located at the poles of the cell. We studied 60 cells from each group. Surveys of a variety of cells sampled at different times of day or night were then made in order to describe correlations in state or distribution of DNA and other cell components with time or stage in the epulo life cycle. We used fluorescence cytochemistry, microfluorometry, and digital analysis of fluorescently stained cells, the most sensitive methods to measure amount and state of DNA and RNA in single prokaryotic and eukaryotic cells (3, 4, 10-13, 32, 33).

Fluorescence cytochemistry had never been applied to E. fishelsoni, so we used two classical fluorescence methods to estimate relative amounts of DNA in cells: the Fuelgen reaction with acridine yellow-Schiff-type reagent (Sigma), and berberine sulfate (Sigma) staining after ribonuclease treatment (3, 4, 10). Schiff reagent reacts with aldehyde groups of free deoxyribose nucleotides liberated by acid hydrolysis. Because the intensity of the Fuelgen reaction depends on temperature and time of hydrolysis, we standardized our preparation by hydrolyzing with 6 N HCl for 10 min at room temperature before treating with Schiff reagent for 30 min. After washing (in 1 N HCl-10% NaHSO₃-distilled water [1/1/18 by volume] followed by distilled water), stained bacteria were mounted in nonfluorescent glycerine for measurements. In the second technique, we stained with a 0.01% solution of berberine sulfate in ethanol for 20 min after pretreatment with ribonuclease; berberine sulfate is an intercalating agent used without the hydrolysis required by the Fuelgen technique. A conventional fluorescence objective (10 by 0.40) and measuring diaphragms corresponding to the structure being measured (whole cell or nucleoid only) were used for measuring relative amounts of DNA (Fuelgen and berberine procedures); units were arbitrary (microamperes of current). Results from these two techniques (see Table 2) validate their application in this system.

We used both morphological and cytochemical criteria following staining with acridine orange to map the locations of different cell components during different stages in the cell and life cycles. Acridine orange has been applied widely as a bacterial stain to detect and characterize soil or sediment bacteria (15, 16, 27, 31), and the metachromic red or green fluorescence of acridine orange has been used to assess viability and physiological activity of both bacteria and bacterial spores (7, 18-20, 28).

Cells stained with acridine orange were mapped according to the classes of molecules distinguished (those with condensed DNA or decondensed DNA, RNA enriched, general cytoplasm, cell wall [see Fig. 3 through 6]). The morphology of E. fishelsoni is well known from both light and electron microscope studies (9, 14, 23) (see Fig. 1 and 2), and we scanned cells stained with acridine orange under permanent visual control. We could, therefore, distinguish fluorescent signals from wall, cytoplasm, and nucleoid under the light microscope.

The fluorescent signals we used for this purpose are the ratios of red and green fluorescence (red fluorescence/green fluorescence × 1,000 [R/G ratio]) of acridine orange excited at 380 to 420 nm. In all treatments and stages in the life cycle (see Table 4), ratios for wall materials remained stable at 150 to 154 and thus served as a check on the comparability of the technique among treatments. Variation among samples in ratios measured for cytoplasm and nucleoids was due primarily to either RNA enrichment or the condensation state of DNA. Acridine orange fluoresces green (maximum at 530 nm with excitation at 380 to 420 nm) when intercalated near A-T (or A-U) base pairs of double-stranded nucleic acids (e.g., DNA and duplex sections of tRNAs) but fluoresces red (maximum at 620 nm) when associated with single-stranded nucleid acids. (Red and green autofluorescence of fixed, unstained cells was <5% of values for stained cells and is ignored here; red fluorescence of epulos [reported in reference 14] occurred when live cells were excited at 510 to 560 nm). Thus, nucleoids with highly condensed DNA will generate lower R/G ratios than nucleoids with decondensed DNA, those with relatively high frequency of single-stranded DNA segments, or those enriched with RNA; in fact, any sites enriched with RNA will exhibit high R/G ratios (11-13, 33).

In order to control for RNA content in nucleoids and cytoplasm of epulos, cells taken at the same life history stage and from the same samples as those scanned for Fig. 3 through 6 were treated with RNase (DNase-free RNase A; Sigma). Cells were incubated in RNase solution (100 U/ml) for 2 h at 37°C, washed in cold phosphate buffer, stained with acridine orange, and scanned as described above. This control demonstrates the impact of RNA enrichment on ratios from various samples and explains why ratios may appear to overlap for different regions of the cells. For example, R/G ratios for cytoplasm prior to RNase treatment ranged from 174 to 210 but dropped to 0 after RNase treatment as red fluorescence was eliminated. For nucleoids, ranges of ratios pre- and post-RNase treatment were 92 to 210 and 96 to 139, respectively, demonstrating that ratios for DNA did not overlap with those of cytoplasm and cell wall.

We used a special microfluorometer equipped with both conventional and contact fluorescence objectives for this work (5, 6). This included an epifluorescence illuminator (OI-30; Leningrad Optical and Mechanical Corporation) designed for work with both contact objectives (tube length, 190 mm) and conventional objectives (tube length, 160 mm), a focusing system for contact objectives, changeable filters, rectangular changeable measuring diaphragms, and a Nikon photomultiplier. To create digital images which show distribution and state of cellular components at various stages of the cell cycle, we scanned longitudinal optical sections through the central plane of cells. A fluorescence contact objective (60 by 1.15), a rectangular measuring diaphragm (5 by 10 µm), and dual-wavelength microfluorometry at 530 and 620 nm allowed us to calculate the R/G ratio and plot values for each 5-by-10-µm frame. Digital images were thus produced for uninucleoid (those with a single nucleoid) and binucleoid (those with two nucleoids) cells of approximately equal size from each separate time sample. These were then scanned into a computer, ranges of ratios representative of different compounds were replaced by distinctive shadings, and edges were smoothed to present a more realistic image (see Fig. 3 and legend for details).

Microbes intended for light and electron microscopy were fixed (2% glutaraldehyde-0.25 M sucrose-0.05 M cacodylate in distilled water or 2% glutaraldehyde-0.05 M cacodylate in filtered seawater). Light micrographs were taken on Kodak Technical Pan film on a Leitz Orthomat E microscope with differential interference contrast (Nomarsky) optics. Specimens for transmission electron microscopy were postfixed in 1% OsO₄, embedded in EMBED 812 resin, thin sectioned, and viewed on a JEOL 1200exII electron microscope.

Descriptive statistics were calculated, and statistical tests were performed with Systat (version 5.03) and Sigma Plot.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Life cycle of E. fishelsoni. Key events in the E. fishelsoni life cycle include changes in mean cell size as well as changes in number, size, shape and location of nucleoids and incipient daughter cells (Table 1; Fig. 1) (23). Because cell replication events occur against a backdrop of consistent, daily changes in apparent cell function and size, we distinguish the cell cycle (restricted to cell division events) from the life cycle (which includes events throughout a 24-h period). We term cells with a single nucleoid uninucleoid and those with two nucleoids binucleoid. We have encountered E. fishelsoni cells with three nucleoids, but their extreme rarity (e.g., a single trinucleoid cell was seen among 480 cells surveyed in 1988) leads us to ignore them here.

View larger version (124K): [in this window] [in a new window]   FIG. 1.   Light micrographs of stages in the life cycle of E. fishelsoni taken under differential interference contrast illumination. Bars = 50 µm. (A) A 237-µm-long cell with two compact, apical nucleoids (arrowheads). The cell was collected at 0640 h. (B) A 45-µm binucleoid (arrowheads) cell collected at 0640 h. (C) Uninucleoid and binucleoid cells with expanded nucleoids. The larger cell is 60 µm long. Unlike larger cells, small nucleoids (arrowheads) of small binucleoid cells do not overlap within the parent cell. Cells were collected at 2000 h. (D) Very small binucleoid cells with nonoverlapping nucleoids (arrowheads). The smallest cell (arrow) is 23 µm long with nucleoids that are <10 µm long. The cells were collected at 2000 h. (E) A 216-µm binucleoid cell with oval nucleoids (between arrowheads). Note the presumed spirillum (arrow) with a length of ~18 µm. Cells were collected at 0915 h. (F) A 195-µm-long uninucleoid cell exhibiting presumed "caps" (see text) of condensed DNA (arrowhead) at apices of maximally enlarged nucleoid. Cell was collected at 1640 h. (G) A 225-µm-long uninucleoid cell with optically distinct daughter cell found only in night samples (compare with Fig. 1F). Cell was collected at 2200 h. (H) A 184-µm-long binucleoid cell with optically distinct daughter cells. Cells were collected at 0130 h.

Structure and DNA content of compact nucleoids. E. fishelsoni cells collected at 0800 h included both binucleoid cells, with two compact, round nucleoids located near the poles of the cell (Fig. 2 and 3A), and uninucleoid cells, with a single compact nucleoid located at one pole (Fig. 3B). To avoid variation in dispersion and state of DNA that might be caused by comparing different stages in the cell cycle, we initially measured total DNA on cells from this morning collection (Table 2). As noted above, cells in size categories II through VI were approximately twice the volume of cells within the previous, smaller category.

View larger version (122K): [in this window] [in a new window]   FIG. 2.   Transmission electron micrographs of compact nucleoids of E. fishelsoni collected at 0800 h. Note presumptive condensed DNA in chromosome-like bodies (arrowheads), delicate cross-striations on some of these bodies, and separation of nucleoidal material from remaining cytoplasm by structures (arrows) continuous with similar materials below the cell wall at the tip of the cell. Bars = 1 µm. (A) Cell fixed in 2% glutaraldehyde-0.05 M cacodylate in filtered seawater. (B) Cell fixed in 2% glutaraldehyde-0.25 M sucrose-0.05 M cacodylate in distilled water.

View larger version (87K): [in this window] [in a new window]   FIG. 3.   Digital images and graphical representation of the same images for binucleoid (A) and uninucleoid (B) E. fishelsoni cells collected at 0800 h and containing compact nucleoids similar to those shown in Fig. 1A and B. Both cells measured ~520 by 65 µm. Cells were stained with acridine orange as described in the text. Values in digital images are R/G fluorescence ratios. Numbers in italics are values consistent with glycoprotein of cell walls; bold and underlined numbers are values consistent with condensed DNA or condensed DNA plus RNA; remaining values are consistent with RNA-enriched cytoplasm, with higher values reflecting higher concentrations of RNA. For visual clarity, ranges of values were identified for condensed DNA, decondensed DNA, general cytoplasm, RNA-enriched cytoplasm, and cell wall materials. Values were then replaced by distinctive shading or hatching patterns, and the edges were smoothed to more closely represent cell configuration. Subsequent figures present only the graphical images; copies of the original digital images are available from the authors and are posted on W. L. Montgomery's web site (http://www2.nau.edu/~wlm). In some figures generated from digital images, cell wall material appears unusually thick along sides and, particularly, apices of cells. This is due to readings of thick optical sections along strongly curved portions of whole cells. More accurate representations of walls occur in electron micrographs (Fig. 2) (14, 23).

Changes in state and distribution of DNA with stages of the cell cycle. The following sections are best interpreted in light of the general effects of RNase treatment on samples (Table 4). R/G ratios of the periphery of cells, interpreted as cell wall materials, remained unchanged at 150 to 154. Ratios for cytoplasm dropped from high values to zero due to the loss of red fluorescence from RNA. Ratios for compact, apical nucleoids declined slightly (128.3 to 124.8) and those for apical caps remained essentially unchanged (111.4 to 112.1), indicating little single-stranded nucleic acid. In contrast, the mean ratio for enlarged nucleoids declined from 145.2 to 136.7, indicating that single-stranded nucleic acid is common in such nucleoids.

(i) Morning. As noted above, in early morning most cells contain compact, apical nucleoids (Fig. 1A and B). The DNA in these nucleoids exhibited R/G ratios characteristic of condensed DNA. Ratios for nucleoids in the samples collected at 0800 h averaged 128.3 (95% confidence interval, ±1.0 [Table 4]). Shortly before this, in samples collected at 0640 h, average R/G ratios were even lower (109.8 ± 2.5), suggesting that DNA in the 0800-h samples was partially decondensed.

(ii) Late morning. In most cells collected between 1000 and 1100 h, nucleoids in both uninucleoid and binucleoid cells had enlarged and shifted from the apices toward the center of the cell (Fig. 4). Nucleoids were characterized by decondensed DNA (reflected in an increase of the R/G ratio to 140 to 148; mean, 145.2 ± 0.4 [Table 4]) distributed evenly around the periphery of the nucleoid and core areas rich in RNA (R/G ratios of 267 [Fig. 4]). This is likely the stage described and shown by Robinow and Kellenberger (26). RNA-rich areas, with ratios in the 250 to 259 range, also occurred in the cytoplasm external but adjacent to the nucleoids. The cytoplasmic mean R/G ratio (210.6 ± 2.3) was higher than that for earlier periods when DNA was condensed, suggesting more active RNA synthesis and transport to the cytoplasm with decondensation of DNA. R/G ratios for peripheral portions of the cell remained unchanged from earlier samples (151.6 ± 0.2 [Table 4]).

View larger version (46K): [in this window] [in a new window]   FIG. 4.   Binucleoid and uninucleoid E. fishelsoni collected at 1000 h. Dimensions: ~320 by 45 µm (A) and ~330 by 55 µm (B). Other features are as described in the legend for Fig. 3. Note expansion of nucleoids, their shift toward the center of the cells, the concentration of decondensed DNA along periphery of nucleoid, and the high concentration of RNA in core of nucleoids.

(iii) Early afternoon. Elongation of nucleoids continues until those of both uninucleoid and binucleoid cells reach approximately 75% of cell length (23). During this period of elongation, which in our samples extended until at least 1430 h, R/G ratios remained much as described for the samples from 1000 to 1100 h (Fig. 4). DNA remained decondensed (ratios of 140 to 149) and evenly dispersed around the periphery of the nucleoid. The interior core of the nucleoids, as well as areas immediately external to the nucleoidal DNA layer, continued to exhibit R/G ratios indicating RNA enrichment (245 to 255), cytoplasm external to these enriched areas remained high at ~184 to 187, and the cell edge was unchanged at ~150.

(iv) Evening. Late in the day and into night (1640 and 2200 h), nucleoids within both uninucleoid and binucleoid cells were surrounded by material producing ratios previously seen only at the edge of cells (R/G ratio, ~150) and were separated from the parent cell's periphery by a layer of material yielding ratios characteristic of general cytoplasm (R/G ratio, 182 to 186 [Fig. 5]). Shifting between fluorescence and transmitted light microscopy demonstrated that the areas with R/G ratios of ~150 were thickened zones surrounding nucleoids and that they were separated from, but structurally and visually similar to, the parental cell wall. This is consistent with electron micrographs which show nucleoids surrounded with thick, laminar material (23) (Fig. 3 and 5 and our unpublished results). We interpret this new layer as cell wall formation around maturing daughter cells.

View larger version (36K): [in this window] [in a new window]   FIG. 5.   Formation of mature daughter cells in binucleoid and uninucleoid E. fishelsoni collected at 1640 h. Dimensions: ~520 by 65 µm (A) and ~530 by 65 µm (B). Other features are as described in the legend for Fig. 3. Note formation of presumptive cell wall material around nucleoids and the separation of this wall material from the parental wall by a thin layer of cytoplasm. Cells for this figure were collected at the same time as those shown in Fig. 6A through C, emphasizing that stages in the epulo life cycle are not precisely synchronized in time and between individual host fish.

(v) Early morning. In some daughter cells collected between 1640 and 0500 h, the DNA located at one or both apices of the cell was very highly condensed (R/G ratio, 96 to 115; mean, 111.4 ± 1.9 [Table 4]) compared to the peripheral DNA (R/G ratios, 130 to 140; mean, 138.3), and DNA near the apices appeared thicker than that along the lateral walls (Fig. 6B C). We termed these condensed, thickened areas of apical DNA "caps."

View larger version (44K): [in this window] [in a new window]   FIG. 6.   Maturing daughter cells of E. fishelsoni collected at 1640 h (A through C) and 0430 h (D through E). Other features are as described in the legend to Fig. 3. (A) Daughter cell (~390 by 45 µm) lacking caps and with decondensed DNA evenly dispersed below the cell wall. (B) Daughter cell (~350 by 45 µm) with two caps. (C) Daughter cell (~350 by 45 µm) with single cap. (D) Daughter cell (~360 by 45 µm) with two caps and DNA almost completely separated. (E) Daughter cell (~360 by 45 µm) with two caps and DNA completely separated. Note mid-cell constriction of DNA in panels B and C that appears unrelated to additional wall formation (contrast with Fig. 4) and condensation of DNA in caps while that along periphery and in center of cell remains decondensed.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The morning nucleoid. Previously published electron micrographs (14, 23) show specimens fixed in late afternoon or evening and show considerable complexity of epulo ultrastructure, including a trend for the nucleoid regions to be clearly separated from surrounding materials. Where a structure responsible for this separation has been visible, it has generally appeared thick and occasionally laminar and has been interpreted as wall material. Specimens used for Fig. 2 were fixed shortly after the first appearance of apical nucleoids of condensed DNA in early morning and demonstrate that even at this early stage in the life cycle, the nucleoid appears to be separated from the cytoplasm. This is clearest in Fig. 2A, where a fine, smoothly curving line surrounds the nucleoid. The identity of the delineating feature and its significance remain unclear, but spore formation in Bacillus subtilis involves the surrounding of an incipient spore by maternal cell membrane and the subsequent formation of a spore coat (17). Angert et al. (1) suggest that similar events occur during daughter cell and endospore formation in a close relative of E. fishelsoni, Metabacterium polyspora.

Cell size, volume, and DNA content. E. fishelsoni appears to be not only the largest known eubacterium but the most variable in size as well. Most of our data derive from studies on large cells (length, >150 µm), supplemented by work with one much smaller group (30 to 35 µm). Four observations make it clear that group 0 cells in fact are small E. fishelsoni. First, the production of one or two nucleoids or daughter cells within the parent cell as well as the elongate cigar shape of E. fishelsoni cells is consistent throughout the size range and may, in fact, extend well below this range (Fig. 1C and D).

DNA content and distribution to daughter cells. Flow cytometric studies of E. coli show that DNA content varies systematically among other bacterial cells, and to some degree with cell size, but this variation usually represents only two- to eightfold increases over the amount of DNA within a single genome (29). We do not know the size of a unit genome in E. fishelsoni, but copy number in large cells appears to be very great. There are large quantities of DNA in large cells, there is localization of DNA-specific stains to nucleoids, and fluorescence ratios indicate condensation of DNA when electron micrographs show darkly staining, elongate structures in the nucleoids.

Changes in state and distribution of DNA with stages of the cell cycle. Clearly, a complex series of events attends growth and maturation of epulos. Quantitative fluorescence cytochemistry demonstrates that DNA in the largest epulos exceeds amounts in the smallest cells by 4 to 5 orders of magnitude and probably exceeds amounts in most other bacteria by an even greater margin. Such large quantities of DNA could pose problems for a cell undergoing the dynamic processes of DNA replication, condensation and decondensation, and systematic distribution to duplicate nucleoids.