INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Much progress has been made in the last 20 years in the understanding of the biochemistry of bacterial methanogenesis. The recognition of methanogens as archaebacteria (16) made them very interesting from a broader biological perspective as well. A recent comprehensive review (44) covers the molecular biology of methanogens. Unfortunately, their slow growth and strict anaerobiosis have tended to discourage experimental approaches that require much culture manipulation. Among methanogens, Methanobacterium and Methanococcus are the two best-studied genera. They differ in several respects, the most striking being the absence of a true cell wall in Methanococcus which, like numerous other archaebacteria, has nothing but a thin S-layer covering the plasma membrane. Concerning Methanococcus voltae, the object of the present work, a comprehensive review up to 1989 is available (29). Later developments may be traced from recent publications (2, 27, 32) (see also references 8 and 58 for the related M. maripaludis).

Strain PS of M. voltae has been known to undergo low-frequency, natural DNA transformation (6). Another, more efficient natural gene transfer mechanism, discovered several years ago (4, 5), is examined in this study. The agent responsible for this activity is referred to here as VTA (voltae transfer agent). The phenomenon may be described as generalized transduction in which, however, the bacteriophage component has not been detected and is probably absent, defective, or different in size from the transducing particles or, in any case, is present in very small amounts.

A variety of defective lysogenic systems among eubacteria have been described. Of particular interest in connection with VTA are systems that accomplish transduction in the absence of "true," i.e., lytically self-reproducing, viral particles, as first found in Rhodopseudomonas capsulata (now Rhodobacter capsulatus) (37, 66) and later found in Desulfovibrio desulfuricans (43), Serpulina hyodysenteriae (24), and possibly Myxococcus xanthus (53). The Rhodobacter example, for which the term "capsduction" has been introduced, is particularly interesting because of the small size of the bacterial DNA fragments transferred, almost equal to that observed here for VTA. Also, capsduction has recently found application as a tool in the analysis of the Rhodobacter genome sequence (33). A more thorough comparison of VTA with the above-mentioned systems can be found in the Discussion.

The only other example of generalized transduction among archaebacteria has been reported for Methanobacterium thermoautotrophicum (39) and involves a typical, medium-sized bacteriophage.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Anaerobic techniques, storage of bacterial strains, medium composition, bacterial DNA extraction, and optical density (OD) measurements are described elsewhere (6). Whenever a method evolved over time in the course of the work, only the most satisfactory version is described.

Bacterial strains. Most bacterial strains were derivatives of strain PS of M. voltae and are listed in Table 1. M. voltae C1, C2, C4, and A3 (62, 64) were obtained from W. B. Whitman, University of Georgia, Athens; M. vannielii (52) was obtained from J. N. Reeve, Ohio State University, Columbus.

Bacteriophages. X174 (50) and P2 lg del1 del2 (7) were used as markers in sedimentation. For mixtures of the two phages, Escherichia coli C-1172, a P2-resistant mutant of C-1055 (63), was used to detect X174, and E. coli K-221 (23) was used to detect P2.

Cultures. M. voltae cultures (5 ml in Bellco anaerobic culture tubes or 25 ml in 125-ml serum bottles) were grown in WM medium plus needed supplements at 30 to 33°C as described previously (6). They were used for experiments at an OD at 600 nm (OD₆₀₀) of approximately 0.3 to 0.7. In calculations, a quantity of 3.5 × 10⁷ total bacteria per ml for an OD₆₀₀ of 0.1 was assumed. For the strains most frequently used, fully grown cultures were stored at room temperature in an anaerobic hood for several weeks, repressurized occasionally with H₂-CO₂ to delay lysis (6), and used as inocula (diluted 1:30 to 1:300) for new cultures. Stock cultures of each strain were stored frozen and retrieved when needed as described previously (6). Viable, easily accessible inocula have been successfully maintained by this method for more than 12 years.

Media. WM medium (6, 61) was generally used, but without resazurin. L-Histidine (final concentration, 0.5 mM), hypoxanthine (final concentration, 0.2 mM), and cyanocobalamin (final concentration, 100 nM) were added at inoculation as needed for the growth of strains carrying the mutation his, pur, or cbl (6). 2-Bromoethanesulfonate (Na salt) (BES) was added for some selections, but BES-resistant strains were generally grown without it. For frozen storage and recovery of cultures and, on occasion, to facilitate growth from old inocula or in the large tumbler cultures, the medium was supplemented with 1/20 to 1/200 volume of a concentrated solution of Difco Casamino Acids, yeast extract, and L-tryptophan (10, 5, and 0.1% [wt/vol], respectively).

Viable counts. Viable counts were obtained by the thin-layer, soft-agar, pour plate technique (6), which does not require preparing plates ahead of time. The agar concentration was critical; satisfactory results were obtained with 0.5% Difco Noble agar as well as with 0.34% Fisher Scientific laboratory-grade agar (6). Before inoculation, each tube of melted soft agar (already containing added sulfide solution and any other needed supplements) was removed from the 45 to 48°C heating block so it could cool down for 3 to 4 min. The plates were incubated for 10 to 12 days at 30 to 33°C. At variance from the method used earlier (6), no CaCl₂ was added to the canister. Also, instead of H₂S gas, an open container with a few milliliters of a 20% Na₂S solution was placed at the bottom of the canister (9). Colony size was strongly affected by crowding; when present in large numbers, colonies were quite small and, except for a few spreaders, did not merge with each other. For some crowded plates, colony counts were obtained under a microscope at an appropriate magnification for a number of randomly chosen fields on the plate, averaged, and multiplied by the known plate/field area ratio.

Resistance to BES. M. voltae PS is sensitive to BES; a stationary-phase inoculum grew with a 1-day lag in a liquid culture with 1 µM BES. At 10 and 100 µM, the growth rate was reduced four- and eightfold, respectively, and there was no growth at 1 mM BES. When samples were plated in agar with 1 mM BES, the efficiency of colony formation was usually well below 10⁵.

3

VTA assays. VTA assays were done in essentially the same way as viable-count assays. In early experiments, recipient bacteria and filtrate samples were introduced directly into the melted-soft-agar tube by use of disposable syringes and Becton-Dickinson 22-gauge needles and by measurement of volumes in drops (holding the syringe vertically gives approximately 14 µl per drop). Later (this is now the preferred method), a measured volume of donor filtrate (between 50 and 200 µl, diluted if necessary) was placed in a 1.5-ml Eppendorf disposable vial, which was kept on ice in air and transferred to an anaerobic hood minutes before plating. A drop of a fresh culture of recipient bacteria (OD₆₀₀, 0.4 to 0.7) was delivered by needle and syringe to a (10-ml) melted-soft-agar tube through the stopper. Immediately afterward, by use of a new needle and syringe, about 0.5 ml was removed from the tube and gently blown into the sample in the vial. The whole sample was sucked up again into the syringe and delivered through the stopper to the tube. To completely deliver the sample, the syringe was rinsed out into the tube by filling it with melted agar and emptying it once more. The stopper was removed with forceps, and the agar was poured into a petri dish. Recipient cultures, kept at room temperature and repressurized with H₂-CO₂, were sometimes used again in new platings; after a few days, however, the efficiency of gene transfer decreased significantly.

Sterile filtrates. Fresh cultures were filtered by hand pressure with either Gelman Sciences Acrodisc (0.2-µm-pore size, 26-mm-diameter) polysulfone filter cartridges or Schleicher & Schuell UNIFLO (0.2-µm-pore size) cellulose acetate filter cartridges; they were equally satisfactory. Filters became clogged after passage of 5 to 15 ml of culture, depending on the bacterial concentration. Some other types of commercial membrane filters, also of 0.2-µm porosity, did not yield as much VTA as the two mentioned above. Filters of 0.45-µm-pore size clarified the cultures more rapidly but occasionally allowed passage of a small fraction (below any visible turbidity) of Methanococcus cells and were therefore avoided. The filters were not usually washed before use. The error in filtrate absorbance readings due to material leached from the membrane filters was negligible.

VTA stability measurements. VTA preparations were usually titrated the day they were made. Later assays supplied estimates of VTA stability upon storage and/or dilution. VTA inactivation was expressed as the number of lethal hits, H = lnS, where S is the VTA survival ratio (titer at time t₂ to titer at time t₁). More specifically, H₀ is the estimate based on the first titration after the beginning of storage (or after dilution into a different suspension medium) compared with the initial titer, whereas H_mo = (365/12)(lnS/d) (where d = t₂  t₁ days of storage) is the rate (lethal hits per month) based on any two successive titrations past the initial one. When more than one H_mo estimate was available for a given preparation, an average of the n estimates, H_mo(n), was calculated. Alternatively, H_mo(n) was calculated as the linear regression coefficient of the logarithm of the VTA titer versus days of storage.

VTA preparation and concentration. The simplest method for preparing VTA was direct filtration of cultures. Since most of the VTA in a culture was found to be associated with the cells (see Results), another simple method consisted of sedimenting the bacteria, resuspending them in a buffer, and filtering them. At earlier stages of the work, however, numerous attempts were made to separate VTA from the bacteria by what was presumed would be desorption, as in the following example. Donor bacteria were grown (e.g., two bottles, 53-ml total volume) to an OD₆₀₀ of 0.3 to 0.5 (about 3 days of incubation of a 1:50 inoculum). They were transferred in a hood to polypropylene centrifuge tubes and spun anaerobically for 5 min at 10,000 rpm (Sorvall SS-34 rotor) in the cold. The tubes were opened in the hood, and the supernatants were discarded; the tubes were drained well. The pellets were gently resuspended with a Squeezette (soft plastic transfer pipette) and pooled in a total volume of 10 ml of cold 0.3 M NaCl with 5 mM sodium citrate. The suspension was spun again as described above. The new supernatant was collected and filtered (F₁). The new pellet was resuspended and centrifuged again as before. The next supernatant was collected and filtered (F₂). This process could be repeated. In the end, the pellet was resuspended as before and filtered (F_p). For this last filtration, because of the high bacterial concentration, more than one filter often had to be used. Bovine serum albumin (BSA) (final concentration, 1 or 2 mg/ml) was added to the filtrates, and they were stored on ice. Of the total VTA recovered, about half was in the last filtrate (see Fig. 3). The total active VTA yield, measured as his⁺ transfer, varied between 2 × 10⁴ and 3 × 10⁵ per ml of culture.

Short-cut preparation of VTA DNA. Bacteria from two tumbler cultures were pelleted anaerobically at 9,000 × g (Sorvall GSA rotor, 7,000 rpm, 10 min). All supernatant fluid was removed. The bacteria were resuspended in 1 ml of 0.3 M NaCl-20 mM Tris (pH 7.7) with BSA (1 mg/ml), pooled, and lysed by the addition of 200 ml of ice-cold distilled water, with rapid mixing. Small volumes of NaCl, MgCl₂, and DNase (Sigma DN-25) solutions were added to obtain final concentrations of 50 mM, 1 mM, and 2 µg/ml, respectively, and the mixture was warmed to room temperature. After 20 min, RNase was added to a final concentration of 25 µg/ml. Forty minutes later, EDTA was added to a final concentration of 2 mM. VTA and bacterial debris were immediately pelleted at 44,000 × g (Spinco SW28 rotor, 18,000 rpm, 4.5 h, 8°C), resuspended in 3 to 4 ml of STE2 buffer (100 mM NaCl, 10 mM Tris, 2 mM EDTA [pH 8.0]), extracted with phenol (with hydroxyquinoline and -mercaptoethanol) through chloroform (45), and concentrated by precipitation with cold ethanol (45) in a thick-walled glass centrifuge tube. Mussel glycogen was added at 20 µg/ml before the ethanol as a carrier (19, 56). After 1 day at 20°C, the ethanol mixture was centrifuged (Sorvall SS-34 rotor, 10,000 rpm, 1 h, near 0°C), and the pellet was dried and suspended in the smallest possible volume of STE buffer (45). No residual DNase activity was detectable in the final preparation. With strains PS-1 and PS-13, total yields of about 1 µg of VTA DNA could be reproducibly obtained. VTA DNA was further purified and concentrated by cutting bands out of agarose gels after electrophoresis and extracting the DNA by the QIAGEN method.

Other methods. Other nucleic acid methods were standard (11, 45). Gel electrophoresis was usually performed with 0.8% agarose gel slabs (25 ml, 95 by 65 mm) horizontally submerged in buffer containing 40 mM Tris, 5 mM sodium acetate, and 1 mM EDTA (pH 7.4) at 100 V (8-V/cm gradient) for 80 min. Some gels (see Fig. 6G and Fig. 7) contained 1% agarose and were run at 50 V for 200 min. The gels were stained in a solution of ethidium bromide (1 µg/ml) for 20 min and destained with 1 mM MgSO₄ for 40 min before being photographed. Restriction enzymes were used as recommended by the commercial enzyme suppliers. DNase represents DNase I and RNase represents RNase A throughout.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

VTA activity. When the histidine-requiring PS-3 mutant strain of M. voltae was grown in mixed culture with the purine-requiring PS-6 mutant strain in fully supplemented medium, large numbers (e.g., 10⁴/ml) of His⁺ Pur⁺ colonies were obtained upon plating in the appropriate selective medium. The frequency of prototrophs in pure cultures of either strain was extremely small (<50/ml). Large numbers of prototrophs were also found in mixed cultures to which DNase had been added at inoculation; such prototrophs could hardly have resulted from the low level of DNA-mediated, natural transformation that was demonstrated earlier for M. voltae PS (6). These observations suggested the presence of some other kind of genetic recombination process and were extended as follows.

View larger version (19K): [in this window] [in a new window]   FIG. 1.   Histidine-independent colonies resulting from a constant amount of filtrate and variable recipient bacterial input. The donor strain was PS-13, and the recipient strain was PS-3. The standard pour plate technique was used (10 ml of agar per plate).

View larger version (11K): [in this window] [in a new window]   FIG. 2.   VTA-mediated transfer of bes to BES-sensitive bacteria. The recipient strain was PS-1 (about 1.6 × 106 bacteria per plate). VTA was from strain PS-2. Each point represents the colonies from one plate. A 5-ml agar overlay with BES was added after 2 days of incubation, establishing upon diffusion a 1.1 mM BES concentration. Two control plates without VTA yielded 12 and 18 resistant colonies; the data as plotted are corrected for this background. The broken line represents the best fit for direct proportionality.

Mode of VTA liberation. The bulk of VTA is associated with bacterial cells, and very little is free in culture supernatants (see above). A larger fraction of VTA could be recovered in supernatants through "washing" (pelleting of cells in a centrifuge, resuspension in buffer, pelleting again, and filtration of the supernatant). Although salt solutions of diverse compositions and concentrations were tried (0.15 to 1.5 M NaCl, with or without 15 to 60 mM sodium citrate, 8 to 60 mM Tris, 0.5 to 5 mM EGTA, and pH 6 to 8), there were no striking differences in the results; even with repeated washings, longer waiting times or higher temperatures after resuspension, or lower cell concentrations (to reduce possible readsorption of VTA), only rarely (e.g., Table 2, I4) could more than half of the total VTA be recovered by this procedure, the rest being recoverable only upon filtration of the bacteria themselves (Fig. 3). For 26.5-ml cultures of strain PS-2 or PS-13 (1 × 10⁸ to 3 × 10⁸ bacteria per ml), this procedure resulted in a total his⁺ VTA yield of 5 × 10⁵ to 8 × 10⁶ (average, 3.4 × 10⁶ [n = 17], or approximately 6 × 10⁴ his⁺ VTA per cell).

View larger version (19K): [in this window] [in a new window]   FIG. 3.   VTA cumulatively recovered in successive extractions (washings) of a culture of M. voltae PS-2. The recipient strain was PS-3. Selection was for histidine independence. F0, filtered culture supernatant; F1 through F5, filtrates of supernatants obtained in five washing cycles; Fp, filtrate of the last resuspended pellet. The culture (52 ml, 2 × 108 cells per ml), was resuspended each time in 10 ml of 0.3 M NaCl-0.015 M sodium citrate. The total his+ VTA recovered was 4.3 × 106.

View larger version (13K): [in this window] [in a new window]   FIG. 4.   Relationship between VTA titer and concentration of UV-absorbing material in the filtrates shown in Fig. 3 (F0 excluded). The mean A280/A260 ratio was 0.55 (range, 0.51 to 0.59).

Other properties of VTA. Most of the experiments shown in Table 2 were done with freshly prepared culture filtrates. The VTA activity in such preparations, even under refrigeration, decayed quite rapidly. At first, VTA was routinely stored under anaerobic conditions. It was soon realized, however, that the gain in stability over that obtained with storage in air, if any, was small. All preparative operations with VTA following filtration were thereafter done without anaerobic precautions.

4

View larger version (29K): [in this window] [in a new window]   FIG. 5.   Isopycnic banding of his+ VTA and two bacteriophage markers in a CsCl density gradient (Spinco SW50.1 rotor, 41,500 rpm, 27 h, 8°C). The expected buoyant densities for P2 lg del1 del2 and for X174 are, respectively, 1.42 (3, 7) and 1.40 (50) g/ml. The VTA sample was prepared from strain PS-2 by the PEG-bag method; hence, high concentrations of NaCl (final concentration in the centrifuge tube, 0.5 M), BSA (tentatively 8 mg/ml), and added glycerol (0.6 M) were present in addition to cell components set free by filtration. The solid line for density is from refractometer readings for a control centrifuge tube containing only CsCl in pure water. The broken line shows the gravimetrically determined densities for reconstructed solutions of corresponding concentrations of CsCl with added NaCl and glycerol. A comparison of refractometer readings for such reconstructed solutions with those taken over the experiment gradient (data not shown) suggests that BSA and/or cell proteins were beginning to band in the upper part of the gradient. The recoveries of activities were 1% of input for VTA, 65% for X174, and >28% for P2.

VTA DNA. Gel electrophoresis of total DNA preparations of M. voltae PS revealed, in addition to high-molecular-weight, chromosomal DNA, an extremely faint, rather broad secondary band at about 4.4 kbp (as calculated for linear, double-stranded DNA) (Fig. 6A). DNase eliminated both the chromosomal DNA and the 4.4-kbp band. Both were unaffected by RNase. The band could be demonstrated in six DNA preparations (from strains PS-1, PS-6, PS-9, and PS-12 and two subisolates) of eight made by slightly different methods over a number of years. The presence of such a band had not been noted in work from other laboratories (28, 49, 51, 62) that used M. voltae PS DNA for a variety of purposes. Presumably, the amount of DNA in the band is affected by the preparation method and by the history of the culture. For two DNA preparations of known concentrations, made from unwashed bacterial pellets, the amount of DNA in the 4.4-kbp band, roughly estimated by comparison with the DNA size standards, amounted to less than 1% of the total DNA (0.1 and 0.4%, respectively, or approximately one copy of 4.4-kbp DNA, on average, per M. voltae chromosome of 1,880 kbp [51]).

View larger version (97K): [in this window] [in a new window]   FIG. 6.   Agarose gel electrophoresis of phenol extracts showing the VTA DNA band. Size markers (lanes 1 in A, B, C, D, and E and lane 2 in F) are fragments of a HindIII digest of phage  DNA (23.1, 9.4, 6.6, 4.4, 2.3, 2.0, and 0.6 kbp). (A) Lane 2 contains DNA from whole bacteria (strain PS-6). (B) Lane 2 contains an extract from partially purified (see Results) VTA (strain PS-6). (C) Lane 2 contains an extract from partially purified (procedure 1 in Materials and Methods) VTA (strain PS-13). (D) Lane 2 contains an extract from a concentrated filtrate (PEG-bag method) (strain PS-2). The sample applied was the equivalent of about 8 ml of culture. (E) Lane 2 contains DNA from a short-cut preparation (see Materials and Methods) (strain PS-13). (F) Lane 1, VTA DNA (strain PS-1) exposed to 85°C for 3 min; lane 3, unexposed control VTA DNA; lane 4, HindIII digest of phage  DNA exposed to 85°C for 3 min. (G) Lane 1, unexposed control VTA DNA (strain PS-1); lanes 2 through 5, VTA DNA exposed for 1 min to 78, 80, 82, and 84°C, respectively. See Fig. 7C, lane 2, for size markers for panel G (part of the same gel slab). Note that in panel G, the agarose concentration was higher and the voltage applied was lower than those in panels A through F.

View larger version (113K): [in this window] [in a new window]   FIG. 7.   Fragment size distribution determined by agarose gel electrophoresis of M. voltae DNA and concentrated VTA DNA after exhaustive digestion with restriction endonuclease AluI. Size markers are as in Fig. 6. Electrophoresis conditions are as in Fig. 6G. (A) Lanes 1, 2, and 3, decreasing amounts (4:2:1 ratio) of M. voltae DNA; lane 4, size markers; lane 5, concentrated VTA DNA. (B) Lane 1, VTA DNA; lane 2, size markers; lane 3, undigested VTA DNA (the amount applied is approximately 1/100 that in lane 1); lane 4, M. voltae DNA. (C) Lane 1, concentrated VTA DNA; lane 2, size markers; lane 3, same as lane 1 but exposed to RNase before application to the gel.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Based on the evidence presented here, M. voltae PS spontaneously generates DNase-resistant particles, referred to here as VTA, that contain 4.4 kbp of DNA and are able to transfer to other cells any of four genetic markers tested. It is fair to assume that VTA will transfer almost any genetic marker on the M. voltae chromosome. It was shown elsewhere (15) that partially purified preparations of VTA contain bacteriophage-like (in the morphological sense) particles with an isometric head and a tail, the head being of an appropriate size for harboring a 4.4-kbp molecule of DNA.

Is this a case of generalized transduction, where the appropriate host strain for bacteriophage plaque formation is missing? Circumstantial evidence made this idea unlikely and discouraged launching the serious effort required to try to isolate a host strain that may be attacked and lysed by a presumptive bacteriophage accompanying VTA.

First, in generalized transduction (36, 38, 55), as exemplified by bacteriophages P1 and P22, there is typically a mixed population of morphologically identical particles, some of which, the plaque formers, contain a viral genome, while others, the potentially transducing particles, contain similarly sized fragments of the donor host chromosome. The proportion R of potentially transducing particles is generally small. This fact also applies to the mechanistically different transduction by bacteriophage Mu. [Based on physical measurements, R values are 0.3% for P1 vir1 (25), 6% for P1 kc (17), and at most 2% for wild-type P22 (13). For phage SP10 in Bacillus subtilis (41), an R value of 14% may be calculated. With specially selected phage mutants, higher R values may be obtained (47). More commonly, the information available for the numerous examples of generalized transduction across the bacterial spectrum is at best the ratio of transductants to phage-producing, singly infected cells. Technically, this measurement may be affected by a low efficiency of selection for the markers used and by killing of potential transductants by secondary phage infection. As an estimate of R, such a measurement would have to be corrected for abortive transduction, uneven sampling in DNA packaging along the host chromosome, and the size of the transducing fragments relative to the host chromosome, the latter being a ratio that may vary from 1/27 for CR30 in Caulobacter crescentus to 1/170 for Mx4 and Mx8 in Myxococcus xanthus (as calculated from sources quoted in reference 40). Raw transduction frequencies across the spectrum of materials are very variable but are rarely (39) higher than 10⁴. Higher values may be obtained (10, 47, 59) if special mutations are introduced.] For the purpose of discussion, the R value for VTA in M. voltae may be estimated as follows. VTA DNA corresponds in size to 1/423 of the 1,880-kbp (51) M. voltae chromosome. A culture of 2 × 10⁸ bacteria per ml yields (by the short-cut preparation method) about 2 ng of VTA DNA per ml, equivalent to 4.2 × 10⁸ VTA copies. Such a culture would produce an activity (as his⁺ transfer) of about 1.3 × 10⁵ per ml (an average for several similar experiments of the type exemplified in Fig. 3 for strains PS-2 and PS-13). The efficiency of transfer by VTA particles for his⁺ may then be calculated as 1.3 × 10⁵/4.3 × 10⁸, or 0.031% and, by correction for a sampling factor of 423 (see above) (tentatively assumed to be constant over the bacterial chromosome) to have an R value of 13%. The corresponding calculation for the transfer of another auxotrophic marker, pur⁺, would result in an R value of 4.1%. These figures do not take into account the possibility of inefficient integration (still undemonstrated in this system) of the transferred DNA fragments. If the efficiency of integration is assumed to be on the order of 1/10, as in P22 (42), the above estimates of R would encompass 100%, a result clearly unlike that for typical generalized transduction.

Second, the DNA molecules carried by VTA particles are quite small (4.4 kbp) compared to those of any double-stranded DNA bacteriophage (known range, 12 to 750 kbp). The genes necessary for the specification of particle structure, the control of replication, the sizing of host DNA fragments, and their incorporation into the capsid could hardly fit into 4.4 kbp of DNA. (However, note the arguable counterexample of polyomavirus and simian virus 40, which have 5 kbp of circular DNA and where "pseudovirions" may contain host cell DNA fragments.)

Third, routine examination of VTA DNA after digestion with a variety of restriction enzymes suggested at first that it was nothing but a random sample of the bacterial DNA of M. voltae, and the possibility that VTA might represent a new case of capsduction, described (37, 60, 66) for strains of R. capsulata, was considered. These strains spontaneously produce particles (called GTA, for gene transfer agent) with 30-nm isometric heads and short tails, containing 4.5 kbp of DNA, and able to transduce any bacterial marker. Only bacterial DNA was found in the GTA particles (66). Their production would seem to be controlled and specified by bacterial genes, without any differential DNA replication. A similar, but rather poorly characterized, example was later described (53) for Myxococcus. Additional examples were described for strains of D. desulfuricans (43) and for the spirochete S. hyodysenteriae (24), although in these the particles are larger, containing 13.5 and 7.5 kbp, respectively, of DNA. The interesting question of whether capsduction evolved in the absence or independently of "true" (i.e., infectious, virally reproducing) bacteriophage was raised (66). Which came first: capsduction and then bacteriophage or bacteriophage and then transduction, followed, through loss of function, by capsduction?

Defective phage systems have been known for many years, although they usually have been detected (e.g., in many Pseudomonas and Bacillus strains) by evidence of viral or subviral particles (35), recognized either morphologically by electron microscopy or through their bactericidal activity, rather than by evidence of gene transfer experiments. For the much-studied PBSX defective prophage (approximately 33 kbp in size) present in B. subtilis and in other species of the genus Bacillus (1, 48, 54, 65), heterologous cell killing seems to have supplanted the gene transfer function, even though the particles contain a random sample of bacterial DNA (in fragments of 13 kbp) which, after extraction, is capable of transfection. Similar examples are those of PBLA (22) in B. licheniformis (producing particles containing 51-kbp host DNA fragments) and of PBND8 (57) in B. natto (particles with 8-kbp DNA fragments). In the absence of comparative studies of particle formation and its control, the difference between such defective lysogeny and capsduction may be one of degree only, depending on the extent to which viral functions are present.

An effort was made to test stringently to what extent VTA DNA was representative of M. voltae DNA. When VTA DNA was concentrated as much as possible and then digested with AluI (which statistically may be expected to cut M. voltae DNA approximately once every 360 bp), a new band of 0.9 kbp was observed (band ; Fig. 7); this band was not visible in similarly digested bacterial DNA preparations. This band presumably represents a DNA sequence that is either amplified through replication relative to the rest of the bacterial genome or preferentially packaged in the course of the production of VTA particles. It remains an open question whether it is part of a presumptive prophage sequence.

Both PBSX of B. subtilis and GTA of R. capsulata are present in a majority of the strains classified within the respective species. The fact that VTA-like DNA could be demonstrated in three other isolates of M. voltae, C1, C2, and A3 (note, however, that C1 and C2 were isolated [62] from the same site), suggests that VTA may likewise be fairly typical of M. voltae strains. The negative results for attempted gene transfer from these three strains to PS would suggest the presence of exclusion mechanisms, probably DNA restriction and modification. Since the chromosomal DNA of strain PS includes (64) some sequences that hybridize with the DNA of a plasmid present in strain A3, further work ought to consider the possibility that this plasmid (or parts of it) is the repository for the genes necessary for VTA production.

VTA has been rather unstable under all conditions tested to date. From Table 3, the average H_mo (see Materials and Methods) at ordinary refrigerator temperatures is 1.8. This value seems to be on the same order of magnitude as that for the least stable E. coli bacteriophages, e.g., Mu. The comparable H_mo for a moderately stable E. coli bacteriophage, wild-type P2, would be 0.04 (unpublished data).

The mechanism of VTA production is very puzzling. Most of the VTA in a culture is found somehow adsorbed to, or complexed with, the cells (or with large, easily sedimentable cell debris), and much of it is efficiently set free only upon filtration, presumably as a result of shear and/or cell disruption. This situation might reflect either the process by which VTA is produced or what happens to VTA after it is produced. In bacteria with an external S-layer, bacteriophages specifically adsorb to it (14, 21, 26, 34), but the reversibility of this step has not been studied. Perhaps injection of the DNA (an irreversible step) occurs only when the S-layer is locally in proper contact with the plasma membrane. Such a situation would be compatible with the recovery of VTA activity upon desorption or filtration. Observations reported elsewhere (15) make it very probable that VTA consists of tailed, bacteriophage-like particles. The buoyant density of VTA in CsCl gradients is exceptionally low compared with those reported in the literature for bacteriophage particles with similar structures. This fact might be explained if the particles are easily but still reversibly entrapped by cell surface structures. This hypothesis would also explain the stubborn association of the bulk of VTA with the cells. More speculatively, M. voltae might be capable of generating membrane vesicles, for example, like the ones that have been experimentally obtained (20), with VTA inclusions. These might then remain, at least for a time, associated with the cells that produced them. Such particles would obviously have a significantly lower buoyant density than compact, phage-like particles. Vesicular structures derived from cell membranes and other surface material are known for several eubacteria: blebs in Neisseria (12), transformosomes in Haemophilus (31), predatory membrane vesicles in Pseudomonas (30), and others. Some of these structures are known to protect DNA from DNase and to participate in the transfer of DNA to other cells.

Cases of apparent reversion to the wild type, noted in the course of the isolation and purification of auxotrophic mutants from M. voltae (see Discussion in reference 6), may well have been due to the then-unsuspected presence of VTA.

From the point of view of microbiological techniques, the setting free by filtration of a gene transfer agent largely bound to the cells is relevant to the interpretation of cases of gene transfer as being due to conjugation; the control experiments in such cases ought to use a culture filtrate and not simply (the filtrate of) a culture supernatant.