![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| DIALOG |
Laboratoire d'Analyse Ultrastructurale, Biophore, Université de Lausanne, CH-1015 Lausanne, Switzerland
Recently, we reported results of the structural analysis of Deinococcus radiodurans nucleoid observed by cryoelectron microscopy of vitreous sections (CEMOVIS) (15). We were able to visualize the arrangement of DNA molecules in the nucleoid of this extraordinary radioresistant bacterium. In previous work, Minsky and colleagues proposed that the nucleoid of D. radiodurans is organized as a densely packed DNA toroid. According to them, this structure could play a crucial role in the radiation resistance of D. radiodurans and, potentially, other species of Deinococcaceae (16, 24). According to our observations, the DNA arrangement in the nucleoid of D. radiodurans Sark differed from a dense toroidal spooling, suggesting that the model of Minsky and colleagues is not generally correct for this bacterium.
Soon after publication, our article faced criticism from Dr. Minsky. His major points, as we understand them, were the following: (i) the quality of the images obtained by our method is insufficient to draw any serious conclusions and (ii) there was inadequate discussion of some of our results. We attribute these criticisms to the newness of our technique and to the attractiveness of his toroidal model, which our results do not support. We see that supplementary explanation and discussion are necessary.
Below we explain why the micrographs of vitreous sections we have presented are faithful and detailed representations of native biological material but must be interpreted in a manner different than that for conventionally fixed, stained, and embedded sections. Furthermore, we discuss several aspects of our results in relation to what has previously been published by other researchers.
CEMOVIS
Micrographs obtained by CEMOVIS must be viewed using criteria different from those we have been accustomed to using during 50 years of electron microscopy of stained and dry specimens. We will try to convince the reader that the micrographs we present contain more, and more reliable, information than that obtained from resin-embedded sections. We will also explain why the cutting artifacts inherent to CEMOVIS are well controlled in good sections; they do not impair image interpretation.
We will refer to Fig. 1, which is a general view of Fig. 1D in our published article at reference 15, as well as to other figures that appeared in that article. This image was obtained after the following preparatory steps: the bacterial suspension was rapidly mixed with 30% dextran, resulting in about a 20% final dextran concentration. The reason for adding dextran is to ensure that, as regards vitrification and cutting, the extracellular medium behaves in approximately the same way as does the interior of the bacteria. Dextran does not penetrate the cell, it has a negligible osmotic effect, and it does not impair survival. D. radiodurans grows well in TGY broth (0.8% Bacto tryptone, 0.1% glucose, 0.4% Bacto yeast extract) containing 20% dextran (7; M. Eltsov, unpublished data). The bacterial suspension was vitrified by high-pressure cooling in a thin copper tube, in which a pressure of 2000 bar was applied at the instant that it was cooled from outside by jets of cryogen. The procedure took ca. 0.1 s and resulted in the vitrification of the whole sample (31). The specimen was cut into thin sections in a cryo-ultramicrotome below –140°C. To this end, the specimen in the tube was prepared with a trimming knife in order to present a small, ca. 100- by 100-µm pyramid from which the sections were cut with a diamond knife. The sections formed small, freely suspended ribbons above the face of the knife, from which they were transferred with an eyelash onto the electron microscopy grid (4). Still keeping the temperature well below –140°C, the grid was transferred into a cryoelectron microscope, where images were recorded with minimum electron exposure. Details of the method were previously presented elsewhere (1).
|
(i) The native biological material is observed instead of the stain deposited on it. The density recorded on the micrograph is directly related to the real local density in the cell. The first consequence is that the cell image looks globally homogeneous (except for special inclusions, such as the dense granules discussed in the article) because there are no regions where the biological material is much more densely packed than in others as is expected for macromolecules floating in a solution. Even condensed chromatin or chromosomes in eukaryotic cells, which by definition are heavily stained regions in conventional preparations, are only a few percentage points denser than the rest of the cyto- or nucleoplasm. In bacteria, it is observed that the nucleoid is generally slightly less dense than the rest of the cell, as is the case for the ribosome-free region in Fig. 1.
(ii) The fine structure of the cell appears at the molecular level. In Fig. 1, we see that the density of a membrane differs very notably from its surroundings, thus producing a visible contrast when the membrane is viewed along the right direction. A membrane appears fuzzy or is invisible, e.g., in septum, when it is tilted with respect to the viewing direction. Many other structures are well visible, as for example, various layers in the cell envelope (S layer), ribosomes, or any large enough, compact macromolecular arrangements. Electron microscopists will appreciate that the visibility of fine structures relies almost exclusively on focus-dependent phase contrast with the associated problem of a nonconstant transfer function. For example, ribosomes are practically invisible in images recorded exactly at focus (pure amplitude image). Nevertheless, optimal use of phase contrast makes it possible to see isolated double-stranded DNA molecules floating in water (14), reconstruct the structure of ribosomes down to an 0.8-nm resolution (3), and observe many details in sections such as those presented here.
(iii) Consequently, the biggest difficulty in working with a vitreous section is not the lack of information in the image, but just the contrary. Structures in the nanometer range can be visualized when they stand alone in water, but how is it possible to single them out in the crowded 50 to 100 nm of the section thickness? This problem will eventually be solved with computerized electron tomography (26). Considerable effort is being made in this direction, but high-resolution reconstruction of vitreous sections is still in its infancy. Pending the advance of tomography, the detailed analysis of conventional high-resolution, two-dimensional images of vitreous sections must not be neglected. Currently, molecules of known shapes and dimensions can be identified, provided that their characteristic projections have sufficient signal-to-noise ratios: a good example is DNA. A single randomly oriented DNA molecule is clearly visible in a micrograph of a thin vitrified layer of low-salt buffer (14). These conditions, however, are vastly different from the situation inside the cell, where DNA molecules are crowded and intermixed with other molecules with similar electron-scattering properties. This is what occurs in the bacterial nucleoid, which contains approximately 9% of proteins (33). In order to estimate the traceability of DNA fibers in the nucleoid, we generated the model represented in Fig. 2. According to the model, randomly oriented overlapping DNA molecules cannot be resolved in vitreous sections of the nucleoid. Fortunately, when straight DNA filaments are oriented along the viewing direction, the accumulation of density over the section thickness produces contrast that is much higher than the noise threshold (Fig. 2A and B). DNA molecules then appear in the image as strongly contrasted 2-nm dots, which are grouped in case of cross-section of bundle parallel filaments (Fig. 2C). The same bundle seen at a tilted angle (Fig. 2A and B) also has a distinguishable contrast. In this case, the overlapping projections of DNA filaments will produce a stripy texture in the image (Fig. 2C), whereas the projection of individual tilted filament cannot be identified (Fig. 2A and B). The "dotted" and "stripy" patterns are frequently seen in the nucleoid of stationary-phase D. radiodurans, revealing bundles of parallel DNA filaments in its nucleoid (15). Similar features are observed in images of other samples containing parallel DNA arrangements, such as DNA crystals (6, 23) and mammalian spermatozoids (30).
|
Deformation is a characteristic feature of CEMOVIS, but it is not just any kind of deformation; to a good approximation, it is homogeneous deformation. By this we mean that a plane remains a plane, and a straight line remains straight. This is evidenced by the preservation of fine details. In the circled region of Fig. 1 (see Fig. 1D of reference 15), the periodic structure seen in the S layer corresponds to 10 nm and the thickness of the cell membrane is about half that much. If we keep in mind that the section thickness is around 70 nm, we understand that the membrane is still remarkably planar in the section and the lines of the S layers have remained straight. Other cases are documented where the deviation from a homogeneous deformation is less than 2% in the major part of the section thickness (30). Homogeneous deformation can be quantitatively described by a mathematical model which makes it possible to calculate back to the original geometry of the specimen from the geometry of the vitreous section (2). Several features of the model have been tested, and work is currently in progress to quantitatively evaluate the accuracy of the homogeneous deformation approximation.
The above description of a vitreous section and how it is produced enables us to appreciate that CEMOVIS makes it possible to observe the native biological material down to molecular dimensions, taking into account the homogeneous deformation that results from the cutting process. We also see that a detailed observation of the nucleoid regions reveals single DNA fragments and their arrangements when they are adequately oriented.
BACKGROUND AND STATEMENT OF THE PROBLEM OF RADIATION RESISTANCE AND DNA COMPACTION IN DEINOCOCCACEAE
The idea of linking the extraordinary DNA radiation resistance of Deinococcaceae to dense ordered compaction of genomic DNA was proposed by Minsky and colleagues (16, 24). They reported an unusual ring shape of nucleoids observed in sections of cryofixed and freeze-substituted D. radiodurans R1 (24). It was suggested that this ring shape was indicative of a dense toroidal packing of genomic DNA similar to that which was observed in vitro in DNA toroids (24). Such DNA toroids are condensates which can be obtained by exposing diluted solutions of DNA to various packing agents (21). A cryoelectron microscopy study of these toroids showed a dense, often hexagonal, packing of laterally ordered DNA filaments (20). The distance between the filaments is around 2.8 nm. Minsky and colleagues assumed that toroidal nucleoids act as rigid matrices in which the diffusion is reduced enough to keep the free ends of radiation-induced DNA fragments close together (16, 24). This geometry could favor high-fidelity repair of double-stranded DNA breaks, especially during the first phase of recombination-independent repair (16). Using the same method, Minsky and colleagues demonstrated that two additional Deinococcaceae, namely, D. radiopugnans and D. radiophilus (16), have ring-shaped nucleoids. The toroidal organization of nucleoids was assumed to be a general strategy to enhance DNA repair efficiency (16).
Results from other groups, however, made the connection between the ring shape of nucleoids and the radiation resistance of bacteria less clear-cut. Zimmerman and Battista examined the nucleoid shape in stationary-phase cells of seven extremely radioresistant, known species of Deinococcaceae by means of in vivo epifluorescence and deconvolution light microscopy (34). They confirmed the ring shape aspect for D. radiodurans R1 nucleoids and for four other species of Deinococcaceae. However, nucleoids of two remaining species, among which is D. radiopugnans, lack any specific shape (34). In addition, by means of a conventional fixation-embedding transmission electron microscopy (TEM) study, Daly et al. demonstrated that the prevalence of ring shapes for nucleoids in D. radiodurans cultures depends on growth conditions (10). They reported that ring-like nucleoids predominate in cultures growing in a defined minimal medium, but these cultures are less radioresistant than cultures growing in a rich medium, in which, however, ring-like nucleoids were less frequent (10).
These data argue against the importance of ring shape in nucleoids as a prerequisite for radiation tolerance; but does this suggest that radiation resistance is unrelated to DNA compaction? Apparently not, because at least at the stationary phase, all radiation-resistant species of Deinococcaceae as well as an unrelated but very radiation-resistant bacterium, Rubrobacter radiotolerans, keep nucleoids highly condensed in contrast to radiosensitive species, which exhibit dispersed nucleoids (17, 34).
In order to understand the structure of the nucleoid, we have to consider two levels of organization: the general shape and the local arrangements of the DNA. A toroidal shape of the nucleoid suggests a toroidal packing of the DNA, but it does not prove it. Conversely, the finding that the nucleoid is not toroidal does not tell about the local arrangement of the DNA. Clearly, the only way to know for sure how DNA is arranged is to look at it directly. This was not possible with conventional electron microscopy methods. It is now possible with CEMOVIS.
STRUCTURE OF THE NUCLEOID IN D. RADIODURANS SARK
We used CEMOVIS to study exponentially growing and stationary-phase cells of D. radiodurans Sark (15). This strain is closely related to, and is as radioresistant as, strain R1 (8), which was used by Minsky and colleagues. Sark was chosen because the structure of the S layer of the cell wall has been studied, down to molecular details, by freeze-etching and atomic force microscopy (5, 22, 29). The fine structure of the S layer is not well preserved by conventional embedding and sectioning techniques, but it is with CEMOVIS, thus serving as a marker of the excellent preservation of cell structure in general.
IN GLOBAL VIEW, RING-LIKE NUCLEOIDS ARE NOT ABUNDANT IN VITREOUS SECTIONS OF D. RADIODURANS SARK
In a CEMOVIS section, we recognize the nucleoid at low resolution as a ribosome-free area of slightly less than average density (Fig. 1). In favorable cases, this identification is confirmed at the molecular level by the characteristic pattern of bundles of DNA filaments (Fig. 1 or, better, Fig. 3 of reference 15). The three-dimensional shape of a nucleoid, in particular, a toroidal shape, cannot be unambiguously determined from a single thin section. Good indications can, however, be obtained from a favorable section plan. This could be the case if the toroid is cut through its central plane. However, the resulting circular shape could also result from a section through a hollow spherical body. A section along the axis of a toroid would show a characteristic pair of similar domains. The presence of a single domain as in Fig. 1 does not completely exclude the possibility of a toroidal shape of the nucleoid, but it is compatible with a toroidal organization of the DNA only if the DNA filaments are approximately oriented in the plan of the section (see below).
In exponentially growing cells, we observed dispersed shapeless ribosome-free areas (15). This structural aspect was similar to that shown by CEMOVIS in other bacteria. It corresponds to the coralline shape of the nucleoids observed by using DNA-specific osmium amine staining in other studies (19, 27).
In stationary phase, the confined, roundish shape of ribosome-free areas of Sark cells corresponds to the well-defined, condensed nucleoid which is a general feature of Deinococcaceae in stationary phase (34). In contrast to what is reported by Minsky and colleagues for the R1 strain (24), ring shapes of ribosome-free areas were rarely found.
Why this difference? We think it is unlikely to be a consequence of our preparation method. After all, the global structure of bacteria obtained by cryofixation and freeze-substitution is very similar to that observed by CEMOVIS (18, 26). The simplest hypothesis is that the difference reflects a difference between strains R1 and Sark. We have also speculated, and we do it with all due prudence, that the abundance of ring shapes reported is linked to dense granules. These are the spherical bodies which are located in the center of the cell and surrounded by the nucleoid but, according to our observation, do not contain much DNA. The dense granules are ubiquitous in exponentially growing D. radiodurans Sark and also found in stationary phase (15), suggesting that the dense granules can be still very abundant at early stationary phase. Together with beginning DNA compaction, which is characteristic for stationary phase, this can result in the abundant ring aspects of nucleoids observed by Zimmerman and Battista with DAPI (4',6'-diamidino-2-phenylindole) epifluorescence in D. radiodurans R1 (34). In addition, it is known that the dense granules tend to get lost during sectioning of resin-embedded samples, thus leaving a region free of biological material in the nucleoid (32). It could mimic a toroidal structure.
SEEN AT HIGH RESOLUTION, DNA ARRANGEMENT IN D. RADIODURANS SARK IS NOT COMPATIBLE WITH TOROIDAL PACKING
As discussed above, the detailed information on the local molecular organization of the DNA is certainly present in vitreous sections but it is revealed only when the molecular filaments are adequately oriented. This is the case when straight portions of DNA molecules are seen along or close to the viewing direction. The pattern is then very characteristic (Fig. 2). The nucleoid area in Fig. 1 is an example. Other examples are given in the article (see Fig. 3 of reference 15). Many randomly oriented DNA filaments overlapping the thickness of the section are expected to result in a smooth, nearly structureless image.
With exponentially growing D. radiodurans Sark cells, ordered DNA arrangement was never observed. This suggests that DNA was not segregated from the cytoplasm; thus, its concentration was not enough to form condensed, ordered DNA phases. We did not observe different levels of compaction between the nucleoids of cells in the same tetrad in exponentially growing R1 (24) as postulated by Minsky and colleagues. The ribosome-free areas, which could be interpreted as rings, were seldom observed in exponentially growing Sark. Dense toroidal DNA packing, however, was not a feature of these structures.
Our data strongly suggest that radiation resistance of exponentially growing D. radiodurans Sark is independent of ordered DNA packing. This does not exclude the possibility that it is supported by other structural particularities of nucleoids, e.g., an alignment of genomes, which Daly and colleagues suggested could contribute to the efficiency of DNA repair (11, 13). Alternatively, the high-radiation resistance at the growing phase may rely on only nonstructural mechanisms (for a recent review, see reference 9).
In the stationary phase, ribosome-free areas frequently show the characteristic pattern of bundles of locally parallel DNA oriented close to the viewing direction. Refer to Fig. 3A in reference 15 and the accompanying image in this paper (Fig. 1). In order to be compatible with the toroid model, such areas of aligned DNA must appear pairwise, corresponding to the two sections of a torus cut along its axis. This has never been observed.
In the regions where the DNA is aligned with the viewing direction, the average distance between filaments is ca. 4.8 nm. The same preferential interfilamentous distance is revealed by Fourier transform in every region with the characteristic pattern of aligned filaments. This distance is almost twice as much, and thus DNA concentration is about four times less than that in compact DNA toroids. Similar interfilamentous distance is observed in DNA solutions in vitro at the transition from isotropic to liquid crystalline phases (25). It is probable that the mobility of the DNA molecules remains high in this low-density liquid crystalline phase.
The local order and nucleoid compaction observed at the stationary phase may impose certain limitations on the diffusion of DNA fragments. It is difficult to believe, however, that the observed nucleoid organization could have the same impact on the immobilization ends of DNA fragments as does tight, toroidal DNA packing. Instead, we expect that at the stationary phase, DNA fragments in the nucleoid would be rather mobile. The mobility of free DNA ends is supported by the data of Daly and Minton (12). In D. radiodurans R1, they detected radiation-induced cyclization of fragments of duplicated insertions during the first RecA-independent stage of repair (12), when according to the toroidal model, the ends of DNA fragments are kept immobile inside the toroids (24).
We speculate that, besides other extraordinarily efficient mechanisms of DNA protection and repair, it is the segregation of the nucleoid from the cytoplasm that is important rather than the local order. In the stationary-phase cell, this confined nucleoid can provide a certain environment in which DNA degradation is decreased because of a concentration of factors inhibiting nucleolytic DNA damage. The recently identified DdrA protein protecting the 3' ends of a DNA fragment, a deletion of which strongly reduces viability in starving cells, could be one of these factors (18). In addition, a decrease of the volume in which the DNA fragments can move increases, to a certain extent, the concentration of the free DNA ends. This could help the single-strand annealing and RecA-dependent screening during DNA repair by homologous recombination (28).
ACKNOWLEDGMENTS
This work was supported by the 3D-EM Network of Excellence within Research Framework Programme 6 of the European Commission.
 |
FOOTNOTES |
|---|
REFERENCES
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Appl. Environ. Microbiol. | Infect. Immun. | Eukaryot. Cell |
|---|---|---|
| Mol. Cell. Biol. | J. Virol. | Microbiol. Mol. Biol. Rev. |
| ALL ASM JOURNALS |
2 µm). Note that the section is devoid of crevasses and has only a moderate compression. Such characteristics of the section, together with proper defocusing, enable observation of the fine details of cell structure, such as a periodical organization of the S layer of the cell envelope (outlined white arrowheads) and the specific texture of the nucleoid (N). Outer membranes (OM) and cytoplasmic membranes (CM) have high contrast when they are viewed along the right direction (black arrowheads) and appear fuzzy (white arrowheads) or are invisible, e.g., in septum (Sm), when they are tilted. R, ribosomes. Large white arrows, knife marks. The circled region was previously published (reprinted from reference 