TABLE 1. Strains of C. michiganensis subsp. sepedonicusused in this study

In our previous studies using contour-clamped homogeneous electric field (CHEF) gel analysis of high-molecular-weight DNA, the minimal genome size of C. michiganensis subsp. sepedonicus was estimated to be 2.5 to 2.6 Mb (4). High genetic similarity values (>88%) were obtained by hierarchical analysis of HindIII and EcoRI genomic fingerprints (4). These findings were consistent with previous claims that C. michiganensis subsp. sepedonicus is a homogeneous taxon (5, 8, 18, 23). Differences between the genomic fingerprints of virulent and avirulent stains of C. michiganensis subsp. sepedonicus were also detected (4). Here we present evidence that the genome of C. michiganensis subsp. sepedonicus contains a previously unreported extrachromosomal element and that this element is a linear plasmid.

By using protocols described for genomic fingerprinting of digested high-molecular-weight DNA, two bands were consistently detected in the undigested genomic DNA of all strains of C. michiganensis subsp. sepedonicus (Fig. 1) (4). One was a faint band that corresponded to unresolved genomic DNA. The other was a bright band that did not correspond in size to any element reported previously for C. michiganensis subsp. sepedonicus. The genome of most strains of C. michiganensis subsp. sepedonicus contains a single 50-kb circular plasmid, designated pCS1 (6, 11, 21). The bright band present in CHEF gels of undigested DNA was much larger than 50 kb. In 10 of the 11 strains, the size of the bright band was about 90 kb, while the size of a similarly intense band in the remaining strain was about 140 kb (Fig. 1). Even undigested DNA from strains P45 and BCP45, which reportedly lack pCS1 and were not amplified by pCS1-specific PCR primers, contained a 90-kb band (4, 21, 26).

FIG. 1. Undigested genomic DNA of C. michiganensis subsp. sepedonicus separated by CHEF gel electrophoresis. Lanes 1 and 13 contain λ markers; lanes 2 to 12 represent strains ATCC 33113, ATCC 33111, ATCC 9850, NCPPB2140, CIC31, Cs3, Cs3NM, CIC242, CIC243, (more ...)

Restriction enzymes known to digest pCS1 into a particular number of fragments were evaluated to determine if the bright 90- and 140-kb bands were perhaps forms of pCS1. Closed circular supercoiled plasmids migrate slowly in pulsed-field gels, while open circular forms typically remain trapped in the sample well (2, 3, 19). Linear plasmids, in contrast, migrate in a pulse-time-dependent manner and at a rate that allows accurate size estimation. If, for example, the 90 kb was due to a circular form of pCS1, digestion with XbaI would be expected to produce a 50-kb linear fragment, since pCS1 contains a single XbaI site (21). Genomic DNA prepared in agarose blocks of selected strains was digested with the restriction enzyme XbaI and separated alongside an undigested sample by using previously described CHEF conditions (4). In all cases, the bright bands did not shift in size after XbaI digestion (data not shown).

BamHI and ClaI were evaluated also, because these cut pCS1 into known restriction patterns (21). Undigested genomic DNA was separated by CHEF, and slices containing the 90- and 140-kb bands were removed from the gel. The DNA was digested with restriction enzymes while still in the gel slices (4). The gel slice was either loaded directly into the well of a new gel or digested under standard conditions for gelase (Epicentre Technologies), and the fragments were precipitated from the mixture. Precipitation of DNA in gel slices produced poor quality and low yields of DNA. The best preparations were obtained by loading the treated plugs or slices directly into the wells of fresh gels followed by separation under CHEF conditions. ClaI, an AT-rich recognition site enzyme, cuts pCS1 at only a few locations (21). In contrast, ClaI digestion of the bright bands from CIC31, ATCC 33113, and P45 produced multiple (>10) bands of greater size than those reported for pCS1 (data not shown). Similar results were obtained with BamHI, a GC-rich recognition site enzyme; several fragments not reported for pCS1 were detected after BamHI digestion of 90- and 140-kb bands (data not shown).

The bright band in CHEF profiles of undigested genomic DNA of C. michiganensis subsp. sepedonicus could not be attributed to pCS1, because strains reported previously to lack pCS1 contained the 90-kb band, and digestion with BamHI, ClaI, and XbaI did not produce fragments of the predicted sizes. Furthermore, digestion of genomic DNA with S1 nuclease, which can be used to linearize circular plasmids for sizing by CHEF electrophoresis, failed to change the size of the 90- or 140-kb bands (1). A faint 50-kb band corresponding to pCS1 was apparent, however, after S1 nuclease treatment of genomic DNA from ATCC 33113 (data not shown). In addition, pCS1 reportedly does not contain any HindIII or EcoRI sites (21), yet the 90- and 140-kb bands were absent or were of reduced intensity in genomic fingerprints obtained with either HindIII or EcoRI (Fig. 2) (4).

FIG. 2. Genomic fingerprints of C. michiganensis subsp. sepedonicus obtained by CHEF gel electrophoresis of high-molecular-weight genomic DNA. (A) Genomic DNA from strains CIC31 (lanes 3, 4, and 5), BCP45 (lanes 6, 7, and 8) and NCPPB2140 (lanes 9, 10, and 11) (more ...)

The intensity and reproducibility in size of the 90- and 140-kb bands in repeated CHEF gels would be predicted if the bands were from linear plasmids (2, 15). The plasmid is present at about five copies per cell, as estimated from measurements of the amount of DNA represented by the plasmid in relation to the total amount of DNA present in the corresponding agarose plug. Assuming the plasmids are linear, the sums of the sizes of all fragments obtained by restriction digests should add up to the same sizes as the unrestricted bands. Digestion of the 90- and 140-kb bands with SpeI gave fragments suitable for sizing, and the sum of the SpeI fragments agreed with the size of the undigested bands (Table 2). The sum of fragments obtained with BclI was about 20 kb less than the size of the undigested plasmid, presumably due to the presence of multiple overlapping fragments not resolved by the CHEF conditions (Table 2). BclI digests revealed differences between the 90-kb linear plasmids from different strains. The BclI restriction pattern of the plasmid from the virulent strain ATCC 33113, in particular, was not the same as those obtained from the avirulent strains P45 and BCP45, even though these plasmids have identical SpeI patterns (Fig. 3). The BclI and SpeI patterns of the putative linear plasmids were clearly different from those reported for pCS1 (21).

TABLE 2. Sizes of SpeI and BclI fragments of linear plasmids from C. michiganensis subsp. sepedonicus

FIG. 3. CHEF gels of BclI-digested linear plasmids from C. michiganensis subsp. sepedonicus strains CIC31, BCP45, and ATCC 33113 (lanes 2, 3, and 4, respectively). Size markers are in lane 1.

The 90- and 140-kb plasmids were apparently related in that they contained some of the same-size SpeI or BclI fragments (Table 2 and Fig. 3), and homology between the plasmids was detectable by Southern hybridization (Fig. 4). The 90-kb band from undigested genomic DNA of ATCC 33113 was purified from CHEF gels, labeled, and detected with the DIG (digoxigenin) Chem-Link labeling and detection set from Roche Molecular Biochemicals. Strong hybridization signals corresponding to the 90- and 140-kb bands were seen in the undigested DNA (Fig. 4). No hybridization signals to unresolved genomic DNA migrating at the compression zone were detected (Fig. 4). If the bands at 90 and 140 kb were actually linear forms of circular plasmids, then the circular forms should have been detected elsewhere in the gel. The lack of any hybridization signals other than those corresponding to the 90- and 140-kb bands was again evidence that these bands represented linear plasmids.

FIG. 4. (A) Undigested genomic DNA of C. michiganensis subsp. sepedonicus strains ATCC 33113, CIC31, P45, and BCP45 (lanes 2, 3, 4, and 5, respectively) separated by CHEF gel electrophoresis. Midrange size markers are in lane 1. (B) Southern blot of undigested (more ...)

We conclude that the genome of C. michiganensis subsp. sepedonicus contains an extrachromosomal element in addition to pCS1 and that this element is a linear plasmid. The linear plasmid is designated pCSL to distinguish it from the circular plasmid pCS1. We propose the terms pCSL1 and pCSL2 for the 90- and 140-kb plasmids, respectively. These are the first linear plasmids to be described for the genus Clavibacter. It is not surprising that pCSL has not been detected in previous studies characterizing plasmids from C. michiganensis subsp. sepedonicus. Most of those studies used plasmid purification protocols that would greatly reduce or eliminate the presence of linear molecules (11, 16, 21). The lack of a band corresponding to pCS1 in CHEF gels of undigested genomic DNA is also not surprising, because of the inherent limitations associated with pulsed-field gel electrophoresis of large circular molecules (2, 3, 19).

The role of linear plasmids in the biology and evolution of Clavibacter remains to be determined. Linear plasmids, which occur frequently in Actinomycetales, are receiving considerable attention because of their importance in virulence, promotion of genetic variability, and production of antibiotics (7, 17, 25, 28). Complete genome sequencing of C. michiganensis subsp. sepedonicus is currently under way. Once obtained, the complete sequence is expected to reveal putative functions associated with the linear plasmids from C. michiganensis subsp. sepedonicus and to enable direct sequence comparisons between pCSL1 and pCS1 and other linear actinomycete plasmids.

We thank S. De Boer, M. Metzler, C. Orser, and A. Oleson for kindly providing strains. We acknowledge the technical assistance of A. Reilley.

We also acknowledge grant support provided by the Colorado Agricultural Experiment Station, USDA Special Research Grants, and the Colorado Potato Administrative Committee.