Fine structure and radiation resistance in Acinetobacter: A comparison of a range of strains

Studies on the use of ionizing radiation for the preservation of food have led to the isolation of many strains of Acinetobacter (earlier called Achromobacter) which vary considerably in resistance to radiation (Thornley, Ingram & Barnes, 1960; Thornley, 1963). The observation of an inverse relationship between radiation resistance and penicillin resistance suggested that some component of the cell wall might influence both of these properties. A comparative study of the fine structure of 9 strains of Acinetobacter has therefore been made with particular emphasis on the mode of division and the organization of the surface layers. Detailed results for one strain, strain 199 A, have already been published (Thornley & Glauert, 1968), and the present paper compares these results with those obtained for 8 other strains.

The strains considered here were included in a taxonomic study made by Thornley (1967) and details of the general properties of the strains are described in that paper.

The bacteria were isolated from poultry carcases as described by Thornley et al. (1960); those obtained from experiment F₄ were labelled MJT/F₄, followed by the strain numbers, 3/16, 8/3, 11/5 and 18/30, while those from experiment F₅ were labelled MJT/F5/5, 14, 199 A and ₂8₄. The numbers are the same as those used by Thornley (1967) in the taxonomic study.

The bacteria were grown on Difco Heart Infusion broth and agar; 0·01 % (w/v) CaCl₂ was added to the broth in cultures grown for electron microscopy. The cultures were incubatedfor 18–24 h at 25 °C, with aeration, in rotating flasks (Mitchell, 1949) containing 1 1. of broth.

Cultures for irradiation were grown for 2 days at 20 °C on slopes to provide cells in the stationary phase of growth, and these were suspended in broth for irradiation. Immediately after irradiation the suspension was diluted and plated for viable counts to be made. The number of viable bacteria was estimated by the method described by Matsuyama, Thornley & Ingram (1964) in which samples are spread on the surface of plates and incubated at 20 °C; colonies are counted after 2 and 3 days.

The bacteria were irradiated with gamma rays in the 360-Curie cobalt-60 source described by Coleby (1960). Dose rates varied from 2720 to 2270 J kg^{−1 −1} (272–227 krd/h) during the course of the experiments. Air was bubbled through the bacterial suspensions during irradia tion, which was done at a temperature of 10–13 °C.

The methods for the preparation of cell walls were described in detail by Thornley & Glauert (1968). They consist of mechanical disruption of the cells in the disintegrator designed by Nossal (1953), followed by separation of walls from intact cells by centrifugation. The walls were washed a total of 8 times, first in 1 M sodium chloride solution and then in 0·1 M phosphate buffer and water.

The cells and walls of strain 5 stuck together in clumps when the bacteria were broken in water and to avoid this the cells were suspended in 0·2 M NaCl during disintegration and during the separation of cells and walls. The walls were then washed by the usual procedure.

The amino acids and sugars present after hydrolysis of preparations of freeze-dried cell walls were studied qualitatively by paper chromatography (Thornley & Glauert, 1968).

Preparations of freeze-dried cell walls were digested with papain and lysozyme as described by Thornley & Glauert (1968); incubation times were either 3 or 18 h for papain, and 3 h for lysozyme. One preparation of intact cells of strain 5 was treated with lysozyme under the same conditions.

The effect of the routine washing procedure on the appearance of the cell walls was studied in an experiment with strain 5; freshly isolated cell walls were examined by electron microscopy before and after washing by the routine procedure.

Freshly isolated cell walls of strain 5 were treated with 0·2 % sodium dodecyl sulphate (SDS) in 0·5 % sodium chloride for 15 min, followed by the routine washing procedure.

Intact cells and cell walls were fixed by the glutaraldehyde-osmium procedure described by Glauert & Thornley (1966), except for the preparation of cells illustrated in Fig. 15 which was fixed by the Ryter-Kellenberger method. After embedding in Araldite, thin sections showing silver-grey interference colours were cut on an A. F. Huxley or LKB Ultro tome III microtome. Sections were stained with uranyl acetate and/or lead citrate.

Equal quantities of a suspension of cells or cell walls in water and 2 % potassium phosphotungstate, pH 7·0, were mixed together. A small drop of the mixture was placed on the collodion-carbon support film of an electron microscope grid with a fine pipette, and excess fluid removed with filter paper. Alternatively, a small drop of the suspension in water was placed on a grid and dried with filter paper; then a small drop of 1 % ammonium molybdate, pH 7·0, was added and this was dried with filter paper.

Electron micrographs were taken with a Siemens Elmiskop I or an AEI EM 6 B electron microscope, operating at 60 kV with a 50-μm objective aperture at instru mental magnifications of 5000 to 40000.

All the strains included in the taxonomic survey made by Thornley (1967) were examined by light microscopy, using phase-contrast or Gram staining, and the observations are reported in detail in that paper. The strains classified as Acinetobacter were non-motile cocci, coccoid rods or short rods, and were often in pairs; they were Gram-negative, but appeared rather more darkly stained than is usual for Gram negative bacteria.

The appearance of 3 strains in the phase-contrast microscope is shown in Figs. 6–8. Electron micrographs of intact cells of 2 other strains (8/3 and 5), negatively stained with potassium phosphotungstate, are included in the paper of Thornley (1967), and show the presence of fimbriae; in addition, strain 8/3 has a thick layer of capsular material which retains a dense layer of phosphotungstate.

After fixation with glutaraldehyde, followed by osmium tetroxide, the surface layers of the bacteria remain closely apposed to the cytoplasm and there are no intracellular vacuoles or other signs of serious damage to the fine structure. The cytoplasmic ribosomes are often very prominent (Figs. 11, 14, r), while the fibrils of the nuclear regions are not easy to distinguish at low magnifi cation. These fibrils are located in the areas of lesser density (Figs. 9, 11, 12, n) and are clearly visible at higher magnification (Fig. 15, n).

Bacteria were fixed at various stages of division and it was there fore possible to deduce a sequence of changes during division for each strain. Division occurred either by simple constriction, with the formation of a very slight septum or none at all, or by the formation of a thick septum accompanied by constriction.

Examples of these 2 modes of division are illustrated in a comparison of strain 14 (Figs. 9–11) with strain 284 (Figs. 12–14). The modes of division of the other strains are shown in Table 1.

The division of cells of strain 14 is initiated by the formation of a constriction around the circumference of the cell, at which all the surface layers invaginate to gether (Fig. 9). This constriction deepens (Fig. 10) and finally the cell is completely separated into 2 daughter cells (Fig. 11); these remain attached by a little, excess wall material (Fig. n) for a time. Simple mesosomes (Figs. 9, 10, m) composed of 2 or 3 concentric layers of membranes are seen near the site of division.

The first sign of division of cells of strain 284 is the appearance of a thick ingrowing septum around the circumference of the cell (Fig. 12, s). The septum is formed by an invagination of the plasma membrane and a growth of a thick layer of material that appears to be associated with the inner dense layer of the cell wall (see next section).

The outer layer of the cell wall (the outer membrane) does not appear to be involved in the formation of the septum. Subsequently the thick septum grows inwards, the cell shows slight constriction (Fig. 13) and eventually a complete septum is formed be tween the 2 daughter cells(Fig. 14). There are considerable variations in the amount of constriction shown by different cells in the same culture. Simple mesosomes (Fig. 12,m) are associated with the invaginations of the plasma membrane during division by the formation of a septum. This type of division is observed in strain 199 A and has already been described in detail (Thornley & Glauert, 1968). The mesosomes of strain 199 A are more complex than those observed in the other strains of Acinetobacter studied.

Two membranes, each with the general appearance of a ‘unit membrane’, and an intermediate dense layer are observed at the surfaces of all strains of Acinetobacter (Fig. 1). The inner of the 2 membranes is the plasma membrane and is closely apposed to the surface of the cytoplasm of healthy cells (Fig. 16, pm). The outer membrane (Figs. 15, 16, om) and the intermediate dense layer (Figs. 15, 16, d) are both components of that part of the surface that is usually described as the ‘cell wall’ (Glauert & Thornley, 1969). These 3 components of the surface structure are most clearly visible in cells which appear to have undergone slight plasmolysis (Fig. 15); they can be distinguished in unplasmolysed cells (Fig. 16) but they are less distinct, since some lightly staining material is present between the outer membrane and the dense intermediate layer (Fig. 16). In some bacteria the outer membrane is fairly smooth and follows the contours of the dense intermediate layer (Figs. 15, 17), while in other cells the outer membrane is convoluted and has a wavy appearance in sections (Fig. 16).

Thin sections of isolated ‘cell walls’ washed by the routine procedure (see Materials and Methods) show that the plasma membranes are removed and that the walls are usually free from cytoplasmic contamination. The outer membrane and dense inter mediate layer are both visible (Figs. 18, 19, om and d), and have the same appearance as in sections of intact cells.

All the strains examined showed the same general surface structure (see Table 1) except for strain 5 in which an additional surface layer (Figs. 2, 20, a) was seen in sections of intact cells. This extra layer appears as a dense line with granular material on its outer surface and lies outside the outer membrane. It is particularly clearly visible in sections of unwashed isolated walls (Fig. 21, a) and is still present in walls that have been washed completely in the routine way (Fig. 22, a). Treatment of walls of strain 5 with sodium dodecyl sulphate, followed by the routine washing pro cedure, causes the outer membrane to fragment and separate from the dense inter mediate layer which remains as an intact sheet (Fig. 23, d). The fragments of the outer membrane roll up with the extra layer still present on the outer surface (Figs. 3, 23, arrows).

The earlier study of the structure of Acinetobacter strain 199A (Thornley & Glauert, 1968) showed that the examination of isolated cell walls by the negative-staining technique reveals the presence of a layer of wrinkled material which appears to be located between the outer membrane and the dense intermediate layer (Figs. 4, 24). Consequently other strains of Acinetobacter were examined in the same way to see if similar material is present. Wrinkles were observed in strains 18/30, 284, 3/16, 11/5 and 8/3 (see Table 1); they were frequently much less closely packed than those in strain 199A (compare Figs. 24 and 25). Some of the present observations were made on preparations of cell walls which had been freeze-dried for storage and it is possible that this treatment loosens the structure of the wrinkled material.

In three strains, 5, 14 and 96, wrinkles are absent, but a few folds are sometimes seen, and, in some cell walls, a layer of finely granular material appears to occupy the position of the wrinkles (Fig. 26, arrow).

Digestion with the proteolytic enzyme papain removes the wrinkled material from cell walls of Acinetobacter strain 199 A, leaving the outer membrane cleanly separated from the underlying dense intermediate layer. The outer membrane may initially remain more or less intact, but eventually it breaks down into tubes and vesicles (Thornley & Glauert, 1968). As soon as the majority of the wrinkled material has been removed the dense intermediate layer becomes visible in negatively stained preparations as a smooth sheet lying within the outer membrane. Similar treatment of the other strains of Acinetobacter with papain does not produce such a clean separa tion of the layers of the cell wall. The wrinkled material (e.g. strain 8/3, Fig. 27) and the granular elements (e.g. strain 14, Fig. 28) are no longer present, so that the inner dense layers and their associated septa (Fig. 27, s) become visible, but the outer membrane usually fragments into vesicles, many of which become completely de tached from the cell wall (Fig. 27). In some strains, however, such as strain 14 (Fig. 28), most of the fragments of the outer membrane remain associated together in a sheet surrounding the inner dense layer.

From these observations it seems likely that the granular material observed in some cell walls occupies a similar position to the wrinkled material in strain 199 A, that is between the outer membrane and the dense intermediate layer.

An array of peg-like subunits is visible on the surface of the outer membrane of Acinetobacter strain 199 A in negatively stained preparations of isolated cell walls (Thornley & Glauert, 1968). These subunits were not seen in all preparations and it seems possible that they may be detached or distorted during the preparation of the walls or that their presence is masked by capsular material. Recent observations suggest that the array of subunits can best be studied in spheroplasts prepared by controlled digestion with lysozyme.

In the present study a vague indication of regular surface patterns was sometimes obtained, and in preparations of strain 5 a clear picture of an array of projecting sub units was visible (Fig. 29). The individual subunits are best seen at the folded edges of isolated cell walls, where they appear to be regularly spaced with a centre-to-centre distance of about 11 nm (Fig. 29, arrow). The arrangement of the subunits on the membrane is not clear in Fig. 29 as a result of superposition of detail from the 2 layers of the cell wall and the consequent formation of moire patterns (Glauert, 1966).

Treatment of cells of strain 5 with lysozyme removes the inner layers of the walls and causes lysis of many cells. The outer membrane fragments into small vesicles (Fig. 30), but the subunits remain attached to the membrane during this process and are still clearly visible on the surfaces of the vesicles; the subunits appear to be in an hexagonal array (Fig. 30, arrows).

The treatment of cell walls of strain 5 with sodium dodecyl sulphate also causes a detachment and fragmentation of the outer membrane, but the dense intermediate layers are not digested (Fig. 23) and are still visible in negatively stained preparations, together with large aggregates of SDS. The vesicular fragments of the outer mem brane are similar to those found during digestion with lysozyme (compare Figs. 30 and 31) and the subunits are still present. A comparison of negatively stained pre parations (Fig. 31) and thin sections (Fig. 23) suggests that the prominent projecting subunits of strain 5 may be associated with the extra layer observed outside the outer membrane (compare Figs. 2 and 3). There appears to be some difference in stability of the outer membranes of strain 5 and strain 199A, since the membranes of strain 199 A are no longer visible after treatment with SDS and appear to break down into irregular granular material. Furthermore, the extra layer (Fig. 3, a) is not seen in sections of cells of strain 199 A.

The projecting subunits on the outer surfaces of strain 5 and strain 199 A are no longer visible after treatment with papain, suggesting that the subunits are composed of protein, or are attached to the outer membrane by a link involving protein (Thornley & Glauert, 1968).

Qualitative results on amino-acid composition were obtained for freeze-dried cell walls of four strains, 5, 14, 96 and 8/3, in addition to those for strain 199A (quoted by Thornley & Glauert, 1968, table 3). The results for the 4 strains resembled those for strain 199 A in that a wide range of amino acids, in cluding diaminopimelic acid, was present in all the cell walls, and sulphur-containing amino acids were absent. The only differences from the results for strain 199 A were the failure to detect proline in 2 of the strains, and tyrosine in 1 strain.

The sugar composition of the cell walls was examined in the 4 strains listed above and also in strain 3/16. The walls of strain 199A contained glucose and galac tose; the walls of all the other 5 strains contained glucose, but galactose was found only in strains 5 and 3/16; a larger quantity appeared to be present in strain 3/16. One other hexose, probably mannose, was present in the walls of strain 8/3. No pentoses were found in any of the cell wall preparations.

After aerobic irradiation with gamma rays all the strains showed sigmoid survival curves (Fig. 5), with a shoulder followed by an exponential portion. A considerable range in resistance is shown by the strains illustrated. Many replicate results were obtained for strains 5, 14, 8/3 and 199A, at the 2 extremes of the scale of resistance, and from these typical curves have been selected; the other strains were tested only once or twice.

The D₁₀ (or dose required to reduce the fraction of cells surviving by a factor of 10) was calculated from the exponential parts of the curves, and found to be 72 J kg’⁻¹ for strain 14 (described as very sensitive, see Table 1), while for strain 199 A it was 460 J kg^-1 (described as resistant, see Table 1). By this criterion the strains differed in resistance by a factor of about 6.

The bacteria described here fall into a group whose classification has caused much confusion in the past. Thornley (1967) compared many isolates from poultry with named strains of various genera, and suggested the name Acinetobacter as most appropriate for the bacteria included in Phenons 2, 3, and 4 of that study. Within the very large Phenon 4, three small phenons, 4i, 4ii and 4.iii, were found, while some strains remained ungrouped within the phenon. The 9 strains of this paper are all in Phenon 4, and their arrangement in the small phenons is shown in Table 1. Four strains of Moraxella hvoffi were also placed in Phenon 4, but not in the subgroups just mentioned.

The fine structure of bacteria named Acinetobacter has not been described by earlier workers. However, Ryter & Piechaud (1963) studied 2 varieties of Moraxella duplex and 2 strains of M. Iwoffi; the former are oxidase-positive and the latter oxidase-negative (see below). All the strains they studied were alike in most aspects of structure, and in section showed the cell wall structure typical of Gram-negative bacteria. All divided by constriction, giving an appearance very similar to that of Acinetobacter strain 14. The present study shows that division by the formation of a thick septum, unusual in Gram-negative bacteria, occurs among acinetobacters. Some species of Achromobacter resemble Acinetobacter, and it seems possible that the marine strains described by Weibe & Chapman (1968) are in this category; they also show thick septa.

It seems to be now generally accepted that ‘acinetobacters’—that is, aerobic Gram-negative bacteria which are non-motile coccoid rods or cocci, with a DNA composition between 37 and 46 moles %GC—should be separated generically from other bacteria. It is not so clear whether they should be grouped into one genus, called either Moraxella (Lwofr, 1964) or Acinetobacter (Thornley, 1967), or separated into two genera, with Moraxella reserved for oxidase-positive strains and Acinetobacter for those which are oxidase-negative (Henriksen, 1960; Baumann, Doudoroff & Stanier, 1968). The evidence on cell division so far provided by electron microscopy does not reinforce the case for separation, since a comparison of our data (see Table 1, p. 22) with those of Ryter & Piechaud (1963) shows that both the oxidase-positive and oxidase-negative categories contain some strains which divide by constriction and others which divide by septation.

This study of a range of strains of Acinetobacter makes it possible to attempt a correlation of morphological and other characters with radiation resistance.

The most striking difference between the strains of Acinetobacter is in their mode of division, although even in this the distinction is not clear cut. Division by the formation of thick septa is often accompanied by some constriction, while strains which divide mainly by constriction often have a few cells with thin septa. However, the production of thick septa characterizes strains 284, 11/5, 8/3 and 199A, and the last 3 of these are the most radiation-resistant, while strain 284 is of intermediate resistance, similar to that of strain 3/16, which does not show thick septa.

It can thus be concluded that the mode of division may be an important factor in resistance to radiation, but that radiation resistance also depends on other factors so that there is no complete correlation. Studies on Escherichia coli have shown that certain strains are more sensitive to radiation than other closely related strains, because their cell division mechanism becomes disorganized, leading to the formation of long filaments (Witken, 1967). The mode of division could be important in de termining the ease of production of such a lesion.

Strain 5 differs from all the other strains of Acinetobacter that we have studied in having an extra surface layer which is visible in thin sections outside the outer membrane. This layer clearly has no influence on radiation resistance since it is lacking in strain 14 which has an almost identical survival curve to strain 5 (Fig. 5)-

The wrinkled material, between the outer membrane and the intermediate dense layer, is observed in the 6 more radiation-resistant strains, while its position appears to be occupied by granular material in the 3 more sensitive strains, 5, 14 and 96 (see Table 1 and Fig. 5). The wrinkled material, therefore, could possibly have some function in radiation resistance, although its nature remains a matter of speculation.

It is now generally accepted that damage to DNA is primarily responsible for the inactivation of bacteria by u.v. and X-rays, and variations in the ability to repair this damage seem the most likely cause of the wide range of resistance of strains of Acinetobacter. This ability may well be controlled by the extent of the permeability changes induced by the radiation. Such changes are a common effect of irradiation (Bacq & Alexander, 1961), although in bacteria the evidence suggests that they are not a primary cause of the lethal effects. However, if they are important in controlling the repair of DNA, a wide difference in radiation resistance would be expected among strains with different surface structures.

Extensive damage to the outer membrane of the cell wall of strain 199 A is seen in negatively stained preparations after 10⁴ J kg^{− 1} of aerobic irradiation (M. J. Thornley and A. M. Glauert, unpublished observations), and it seems likely that lower doses may produce enough damage to alter significantly the permeability properties of the cell, assuming that the same effect occurs in intact cells. It is clear that the structure and composition of the different layers of the cell wall may be critical in determining changes in permeability properties and, in particular, the wrinkled layer, found only in the resistant strains of Acinetobacter, may be of great importance.

Some data from an earlier taxonomic study of strains of Acineto bacter (Thornley, 1967) are included in Table 1 of the present paper for comparative purposes. It can be seen from Table 1 that the radiation resistance of the strains is related to their grouping, the most sensitive belonging to Phenon 4111, the most resistant being in, or close to, Phenon 41, while those of intermediate resistance are in, or similar to, Phenon 411. This taxonomic grouping is the result of many associated characters (see Thornley, 1967, table 11), which separate the phenons on a quanti tative basis. However, few characters are 100% positive or negative for any phenon and, when examined individually, most properties show little correlation with radia tion resistance. As mentioned earlier, penicillin sensitivity, as assessed by a qualitative test, shows the closest links with radiation resistance, but when the minimum in hibitory concentration was measured (Table 1), the correlation was not quite com plete; the most penicillin-resistant strain, 96, although sensitive to radiation, was not the most sensitive.

The DNA composition varies between phenons (Thornley, 1967, table 11), and this character also appears to be linked with radiation resistance (Table 1), the molar percentage of guanine + cytosine (%GC) varying from 37 or 38 for the most sensitive strains to 45 for the most resistant. Kaplan & Zavarine (1962) studied various bacteria showing a wide range in DNA composition and found that a correlation exists between a low %GC and higher radiation resistance. Among Acinetobacter strains, the variation in DNA composition is comparatively small, the radiation resistance extends to much higher values than those studied by Kaplan & Zavarine, and the correlation is in the opposite sense. It seems unlikely therefore that the differences found in Acinetobacter have any connexion with resistance to radiation.

The observations reported in this paper suggest that morphological characters may be involved in the resistance of Acinetobacter strains to radiation. In particular, one of the surface layers, the wrinkled layer, has been found only in the more resistant strains. It is clearly necessary, however, to have some evidence on the possible bio chemical mechanisms underlying this resistance before claiming that there is a meaningful correlation with any particular structure. Consequently studies are now in progress using ultraviolet irradiation, where some DNA repair systems are well understood, to find out whether any of the suggested mechanisms appear to function.

We are grateful to the International Atomic Energy Agency and to the Science Research Council for support for one of us (M.J.T) and to the Wellcome Trust for loan of the AEI EM 6B electron microscope. One of us (A.M.G.) is a Sir Halley Stewart Research Fellow.We gratefully acknowledge the skilled technical assistance of Mr R. A. Parker.