Physical and Genetic Map of the Obligate Intracellular Bacterium Coxiella burnetii

Abstract

Pulsed-field gel electrophoresis and PCR techniques have been used to construct a NotI macrorestriction map of the obligate intracellular bacterium Coxiella burnetii Nine Mile. The size of the chromosome has been determined to be 2,103 kb comprising 29 NotI restriction fragments. The average resolution is 72.5 kb, or about 3.5% of the genome. Experimental data support the presence of a linear chromosome. Published genes were localized on the physical map by Southern hybridization. One gene, recognized as transposable element, was found to be present in at least nine sites evenly distributed over the whole chromosome. There is only one copy of a 16S rRNA gene. The putative oriC has been located on a 27.5-kb NotI fragment. Gene organization upstream the oriC is almost identical to that of Pseudomonas putida and Bacillus subtilis, whereas gene organization downstream the oriC seems to be unique among bacteria. The physical map will be helpful in investigations of the great heterogeneity in restriction fragment length polymorphism patterns of different isolates and the great variation in genome size. The genetic map will help to determine whether gene order in different isolates is conserved.

Coxiella burnetii, an obligate intracellular bacterium propagating in the phagolysosomes of eucaryotic cells, is the only species of the genus Coxiella. In comparison to other members of the family Rickettsiaceae, C. burnetii demonstrates high resistance to chemical and physical agents, making it possible for the organism to remain infectious after years outside the host cell (2). C. burnetii infection of humans manifests itself as acute Q fever with influenza-like symptoms. However, infection may also result in the therapy-resistant chronic form of Q fever with endocarditis, granulamatous hepatitis, or osteomyelitis (37).

Though C. burnetii has been recognized worldwide as an important pathogen, little is known about the genes that contribute to virulence, particularly tissue invasiveness and intracellular persistence. Virulence potential is correlated with lipopolysaccharide (LPS) content of the organism (55). When propagated in nonimmunocompetent systems such as embryonated chicken eggs or persistently infected cell cultures, C. burnetii converts from phase I to phase II particles. Phase shift is accompanied by a drastic change in LPS content and structure of the outer membrane, with a significant decrease in virulence potential (34). Initial studies of LPS phase variation focused on possible alteration of plasmid-encoded genes. However, cloning and sequencing of the entire QpH1 plasmid (45) and plasmid sequences integrated into the chromosome of plasmidless C. burnetii Scurry Q217 (51) revealed no evidence for genes involved in phase variation. Furthermore, plasmid type had been correlated to the type of disease (39), but this has been refuted by several authors (44, 58). These findings suggest virulence factors to be chromosomally encoded. Restriction fragment length polymorphism (RFLP) patterns demonstrated a great heterogeneity of C. burnetii isolates (21, 24, 46) which may be associated with virulence potential. A physical map of C. burnetii Nine Mile would provide a means to evaluate the reason(s) for this great heterogeneity and the different virulence potentials of isolates.

Due to restrictions imposed by the intracellular nature of C. burnetii, genetic studies have always been performed with Escherichia coli as the vehicle for gene expression. Nevertheless, gene organization of an organism can be studied only with an existing genetic map. Classical genetic maps are constructed by well-established methods such as transduction, transformation, conjugation, or transposon mutagenesis. But since obligate intracellular bacteria are not amenable to conventional linkage mapping, the only way to construct a genetic map—apart from sequencing—is to first establish a physical map and then locate genes on the physical map by hybridization techniques. A first step toward generation of C. burnetii mutants was recently undertaken by Suhan et al. (42), with the successful transformation of C. burnetii to ampicillin resistance. Here we present the physical and genetic map of C. burnetii Nine Mile constructed mainly by two methods, pulsed-field gel electrophoresis (PFGE) and PCR.

MATERIALS AND METHODS

Bacterial strains.

C. burnetii Nine Mile RSA 493 was obtained from L. P. Mallavia, Washington State University, Pullman. C. burnetii NotI/Sau3a and NotI/EcoRI fragments were shotgun cloned in phagemid vector pBluescript II KS(+) (Stratagene, Heidelberg, Germany) and transformed into E. coli XL1 Blue (Stratagene).

PFGE sample preparation.

Heat-inactivated (15 min, 85°C) C. burnetii organisms were diluted to a concentration of 2 × 10⁹ particles/ml. One volume of this suspension was mixed with 1 volume of solubilized 1% InCert agarose (Bio-Rad, Munich, Germany) at 50°C. Agarose-embedded C. burnetii was lysed overnight at 56°C with proteinase K (500 μg/ml) and washed twice with 10 volumes of 1× TE (10 mM Tris-HCl, 1 mM EDTA) for 30 min. Proteinase K was inactivated by phenylmethylsulfonyl fluoride (1 mM) at 50°C. C. burnetii total DNA embedded in agarose plugs was digested with 10 U of NotI/ml in 400 μl of 2× Universal buffer (Stratagene) overnight at 37°C.

PFGE.

All PFGE gels (1% MBC agarose; Bio-Rad) were run on a CHEF (contour-clamped homogeneous electric field) mapper apparatus (Bio-Rad). The following parameters were applied to separate NotI fragments of the indicated size ranges: (i) 2 to 270 kb, pulse time of 0.1 to 11.0 s for 8 h, linear gradient and then 9.0 to 24.0 s for 12 h, linear gradient at constant voltage (6 V/cm) and constant angle (120°); (ii) 50 to 270 kb, pulse time of 14.23 to 16.08 s for 23 h 22 min, linear gradient at constant voltage (6 V/cm) and constant angle (120°), ramping factor of −1,357; and (iii) 2 to 55 kb (field inversion gel electrophoresis [FIGE]): pulse time of 0.05 to 0.12 s for 9 h 26 min, forward voltage of 9 V/cm, reverse voltage of 6 V/cm.

DNA probes.

Biotinylated gene probes were prepared by PCR with biotin-21-dUTP (Amersham, Braunschweig, Germany), using different dTTP/biotin-21-dUTP ratios (3:1, 6:1, and 9:1). From cloned DNA fragments, plasmid DNA was prepared and biotin labeled by nick translation (Nick translation biotinylation kit; Serva, Heidelberg, Germany).

Southern hybridization.

To dissect large NotI fragments, PFGE gels were exposed to UV light (Stratalinker; Stratagene) with an energy of 60 mJ/cm². Southern blotting was performed as downward blotting (8). Fragments were denatured (0.5 M NaOH–1.5 M NaCl, two cycles of 30 min each) and transferred to Biodyne B membranes (Pall; Dreieich, Germany) with 20× SSC (3 M NaCl, 0.3 M sodium citrate) as the transfer buffer. Transferred DNA was immobilized by exposure to UV light (120 mJ/cm²). Hybridization experiments were carried out at 60°C with 200 ng of denatured probe. Since probed DNA was always the same (C. burnetii Nine Mile) and only probes differed, the membrane was cut into stripes (3 by 130 mm) and hybridizations were performed in a specially constructed hybridization chamber. Stringency washes and chemiluminescence detection of biotinylated DNA were carried out as instructed by the manufacturer (Serva).

XL PCR.

XL (extra-large) PCR was developed for the amplification of DNA fragments of up to 40 kb (7). XL PCR was performed on a Perkin-Elmer thermal cycler (model 9600; Perkin-Elmer/ABI, Weiterstadt, Germany) in a total volume of 50 μl consisting of 1× XL buffer, 1.1 mM magnesium acetate, 0.4 μM each primer, 200 μM each deoxynucleoside triphosphate, 1 U of XL polymerase, and 10⁴ to 10⁶ DNA templates. To amplify the 27.5- and 11-kb fragments, the following conditions were applied (values for the 11-kb fragment are in parentheses): 15 cycles consisting of denaturation at 94°C for 15 s, annealing and extension at 68°C for 15 min and 30 s (60°C, 7 min); 15 cycles consisting of denaturation at 94°C for 15 s, annealing and extension at 68°C for 15 min and 30 s, with an autoextension time of 15 s per cycle (60°C, 7 min plus 15 s per cycle). Prior to PCR, template DNA was allowed to completely denature at 94°C for 1 min. After PCR, an additional extension step at 72°C for 10 min was applied to maintain fully double-stranded DNA.

Sequencing.

Nonradioactive sequencing reactions were performed with a PRISM Ready Reaction Dye-Deoxy Terminator Cycle Sequencing kit (Perkin Elmer/ABI) as recommended by the manufacturer.

Sequence analysis.

DNA sequence analysis was performed with the DNASTAR software package (DNASTAR Inc., London, England). Primers were designed with the OLIGO software program (Medprobe, Oslo, Norway) and synthesized on a model 381A DNA synthesizer (Perkin-Elmer/ABI).

RESULTS

Mapping strategy.

The physical map of C. burnetii Nine Mile has been constructed by a top-down approach in combination with PCR technology.

PFGE was applied to separate fragments after NotI digestion of C. burnetii total DNA. PFGE gels were blotted and used in Southern hybridization experiments. Simultaneously, NotI/EcoRI and NotI/Sau3A fragments respectively of C. burnetii were shotgun cloned and sequenced. Sequence data were analyzed for open reading frames (ORFs) containing the NotI restriction site. Polypeptides deduced from ORFs were compared to protein database entries. Two polypeptides showing homology to the same database entry were considered as being adjacent. Proximity was proven by PCR with total C. burnetii DNA as the template. Amplicons from positive PCR results were digested with NotI to verify the contiguity of the two fragments.

Proximity of cloned fragments without any homology to database entries was examined by checkerboard PCR. For this purpose primers deduced from sequenced fragments and directed to the NotI site were combined in pairs and subjected to PCR with C. burnetii total DNA as the template. In all other cases where adjacent fragments were missing, we applied adaptor PCR (54).

Physical map.

PFGE has been shown to be the method of choice for constructing macrorestriction maps of bacterial genomes (10) provided that appropriate rare-cutting restriction enzymes are available. Criteria for the selection of restriction enzymes have been developed by McClelland et al. (31). Most obviously, in bacterial genomes with G+C contents above 45%, the tetranucleotide sequence CTAG is extremely rare. Similarly, trinucleotides CCG and CGG are rare in genomes with G+C contents of less than 45%. The G+C content of the C. burnetii genome has been determined to be 43 to 45% (36). Therefore, we selected restriction enzymes SfiI (GGCCNNNN′NGGCC), NotI (GC′GGCCGC), FseI (GGCCGG′CC), and SmaI (CCC′GGG). Finally we applied restriction enzyme NotI, which produces 29 fragments, 25 of which are distinguishable on ethidium bromide-stained CHEF-PFGE gels (Fig. 1B). For optimal resolution, portions of the molecular weight size range where DNA fragments accumulated were expanded. Resolution of NotI fragments around the 200-kb region was enhanced by extending the pulse times (Fig. 1A). The 30-kb region was expanded by FIGE (Fig. 1C). The average resolution of the macrorestriction map is 72.5 kb, with 2.1 kb being the smallest and 268 kb being the largest NotI fragment. Altogether, the size of the C. burnetii chromosome is 2,103 kb (Fig. 2).

FIG. 1.

CHEF-PFGE of NotI-digested total DNA. Different CHEF-PFGE parameters (see text) were applied to separate NotI fragments of 2 to 270 kb (B), 55 to 270 kb (A), and 6.7 to 55 kb (C). Lines indicate regions where resolution of NotI fragments has been increased. Panel A demonstrates fragments ranging in size from 210 to 268 kb to be clearly separated. Nevertheless, the two 227-kb NotI fragments are not distinguishable. Resolution enhancement is even more striking in panel C, where the 52.3-kb double fragment (B) divided into 53.8- and 50.8-kb NotI fragments. Panel C also demonstrates the 27.5- and 32.1-kb NotI fragments to be double fragments, as indicated by increased band intensities. In contrast, band intensity of the QpH1 plasmid suggests that there is only a single copy of the plasmid in C. burnetii Nine Mile.

FIG. 2.

Physical and genetic map of the chromosome of C. burnetii Nine Mile. NotI fragments are indicated by numbers on the outer circle (sizes in kilobases): 1, 7.3; 2, 50.8; 3, 13.8; 4, 26.1; 5, 227; 6, 227; 7, 103; 8, 27.5; 9, 151; 10, 16.9; 11, 268; 12, 9.4; 13, 3.9; 14, 210; 15, 19.2; 16, 6.7; 17, 168; 18, 6.7; 19, 28; 20, 53.8; 21, 12.4; 22, 27.5; 23, 32.1; 24, 118; 25, 5.3; 26, 32.1; 27, 8.2; 28, 2.1; 29, 240. NotI recognition sites are indicated by lines connecting the inner and outer circles. ¹, contains the NotI recognition site. ², hybridization signal intensity suggests IS1111a to exist at least once on the 50.8- and 53.8-kb NotI fragments (in Fig. 1B, lane 1, indicated as the 52-kb fragment). ∗, Could not be related to one of the 227-kb NotI fragments; hybridization signal intensity suggests IS1111a to exist at least once on both 227-kb NotI fragments (Fig. 1B, lane 1). ∗∗, putative genes located around the oriC region: fmu, glysAB, ygi2, ygi1, omp, gidB, gidA, 50K, 60K, and 9K.

In total, 3 NotI fragments (2.1, 5.3, and 6.7 kb) were sequenced completely, and partial sequence information was obtained from 3 NotI, 59 NotI/EcoRI, and 15 NotI/Sau3A fragments. Neighbors of the 3.9-kb NotI fragment had already been determined (EMBL accession no. X75627 and X70045), and one neighbor of the 6.7-kb NotI fragment was extracted from the EMBL database (accession no. L33409). From 87 clones sequenced, 33 fragments (3 NotI, 28 NotI/EcoRI, and 2 NotI/Sau3A) differed in sequence. In total, 50,282 bp including those extracted from the EMBL database were analyzed. The average G+C content has been determined to be 44%. The tetranucleotide sequence CTAG occurs about 23 times less frequently (0.08%) than AAAA (1.88%), whereas trinucleotides CGG and CCG, which are parts of the NotI recognition site show no trend. Assuming a random distribution of nucleotides, NotI sites appear every 131,070 bp. Thus, statistically 16 NotI fragments would be generated upon restriction of the C. burnetii chromosome (2,103 kb) with NotI. This is in contrast to experimental data (29 NotI fragments), meaning that the frequency of trinucleotides CGG and CCG is higher as expected by randomness.

We analyzed sequences for ORFs containing the NotI restriction site and found putative polypeptides deduced from 20 ORFs to be homologous to database entries. In each case, two polypeptides show homology to the same database entry. Hence, 10 NotI linking fragments (Table 1, amplicons 1, 5, 8, 13, 14, 18, 19, 23, 24, and 27) were identified simply as a result of homology to database entries. Remaining linking fragments were identified by checkerboard PCR (Table 1, amplicons 2, 4, 6, 10, 12, and 21) and adaptor PCR (Table 1, amplicons 3, 7, 9, 11, 15, 16, 22, and 25). Sequence data obtained from one adaptor fragment revealed a region containing an unusual G+C stretch of 17 bp with two nested NotI sites (GCGGCCGCGGCCGCGCC; first site in boldface and second site underlined).

TABLE 1.

Linking fragments as identified by PCR

Amplicon no.	Primer 1/primer 2	Amplicon size (bp)a	EMBL accession no.	Protein homologyb (Swiss-Prot)	NotI linking fragments (kb)
1	12_1/1454	363	AF064944	PUR9_ECOLI	12.4/53.8
2	1406/1408	240	AF064945	NH	27.5/151
3	1411/1411NB	328	AF064946	RL22_ECOLI	32.1/118
4	1412/1465	495	AF064947	NH	27.5/103
5	1455/516F	361	AF064948	MMSA_PSEAE	16.9/268
6	1463/5KS15	517	AF064960/X79075	NH	5.3/32.1
7	1467/1467NB1	548	AF064949	NH	227/227
8	1468/1629	336	AF064950	YAAC_PSEFL	28/53.8
9	1469/1469NB	≈1,800	AF064962/AF064963	NH	19.2/210
10	1473/1488	346	AF064951	CMLA_PSEAE	9.4/268
11	1474/1474NB	≈1,300	AF064964/AF064965	NLPD_ECOLI	26.1/227
12	1482/1607	492	AF064952	NH	103/227
13	1496/1487	501	AF064953	RUVB_ECOLI	7.3/50.8
14	1501/542NSF	349	Y10436	GIDA_PSEPU	27.5/32.1
15	1510/1510NB	≈1,100	AF064966/AF064967	ND	6.7/168
16	1632/1632NB	631	AF064954	ECOLYSU	6.7/28
17	200_1/4_1	748	X75627/X70045	SP3E_BACSU	3.9/210
18	20T3/1475	407	AF064955	PUR2_ECOLI	12.4/27.5
19	2KS5/1451R	745	X78969c	HEM1_SALTY	2.1/8.2
20	4_2/10_1	584	X75627/X70045	SYS_ECOLI	3.9/9.4
21	526F/1456	314	AF064956	ND	13.8/26.1
22	526R/526NB	224	AF064957	NH	13.8/50.8
23	543NSF/1452	200	AF064958	PNTB_ECOLI	8.2/32.1
24	565/1459	518	AF064959	COM2_BACSU	16.9/151
25	5T716/5T716NB	488	AF064961/X79075	PSECRC	5.3/118
26	6KSN/6KS6307	591	L33409	DHSA_ECOLI	6.7/168
27	6SK946/1470F1R	1,331	X77919/U07789	SUCC_ECOLI	6.7/19.2
28	Bicyclo1/2	669	X78969c	BCR_ECOLI	2.1/240

^aAmplicons 9, 11, and 15 have not been sequenced completely; therefore, the exact amplicon size is not known. 

^bFrom Swiss-Prot database. ORFs have been detected from amplicons 2, 4, 6, 9, and 12, but deduced polypeptides showed no homology (NH) to database entries. ORFs from amplicons 15 and 21 were not detected (ND). 

^cReference 53. 

Southern hybridizations were used to relate the cloned DNA fragments (NotI/EcoRI and NotI/Sau3a) to the corresponding NotI fragment. Nevertheless, four NotI fragments (227, 32, 27.5, and 6.7 kb) appeared to be double fragments. One of the 6.7-kb NotI fragments has been sequenced completely (EMBL accession no. X77919). The 227- and 32-kb doublets were distinguished by Southern hybridization using a PFGE blot of partially FseI- and FseI/NotI-digested total C. burnetii DNA.

The 27.5-kb NotI double fragment was differentiated by XL PCR. For this purpose, primers deduced from fragments adjacent to the 27.5-kb NotI fragments were combined in pairs and subjected to XL PCR with C. burnetii total DNA as the template. With one primer combination, we achieved a positive result with an amplicon size of about 28 kb. Hybridization with probes derived from the four DNA fragments constituting the ends of the 27.5-kb NotI fragments revealed positive signals with only two of them.

Genetic map.

ORFs with homology to database entries (putative genes) and genes were located on the physical map by Southern hybridization. Hybridization with the omp gene as a probe revealed a hybridization signal with the 27.5-kb NotI double fragment. To relate the omp gene to one of the 27.5-kb NotI fragments, XL PCR was performed with primers deduced from the omp gene and the ends of the 27.5-kb fragments. The positive PCR result with one primer combination was confirmed by hybridization and sequencing of the 11-kb amplicon. In total, we mapped 54 genes and putative genes, of which 39 (Table 2; Table 1, amplicons 17, 20, and 26) were extracted from the EMBL database and 16 putative genes (Table 1, amplicons 1, 3, 5, 8, 10, 11, 13, 14, 16, 18, 19, 23, 24, 25, 27, and 28) were newly identified during the mapping experiments. Nineteen putative genes with homology to database entries contained the NotI recognition sequence. Several putative polypeptides demonstrated only low homology to database entries (bcr, cmlA, and nlpD), though hydrophobicity plots were very similar. One gene, recognized as transposable element (23), was found to hybridize with at least nine NotI fragments.

TABLE 2.

Genes assigned to the physical map

Gene	Polypeptide	Accession no.a	Reference(s)
rrs		M21291	1
qrsA	Sensor-like protein	U07186	30, 32
IS1111a	Transposase	M80806	23
fmub	Hypothetical protein	U10529
glysAb	Glycyl-tRNA synthetase (α chain)	Y10435
glysBb	Glycyl-tRNA synthetase (β chain)	U10529/Y10435	43
ygi2b	32.4-kDa protein	Y10436
ygi1b	28.9-kDa protein	Y10436
omp (com1)	Outer membrane protein	Z11828/M88613	5, 22
gidBb	Glucose-inhibited division protein B	Y10436
50Kb	50-kDa protein	Y10436
60Kb	60-kDa protein	U10529	43
9Kb	9-kDa protein	U10529	43
rnpAb	RNase P	U10529	43
rpmHb	Ribosomal protein L34	U10529	43
mip	Macrophage infectivity potentiator protein	U14170	9, 33, 41
rnC	RNase III	L27436	60
recO	Repair and recombination protein	L27436	60
era	GTP-binding protein	L27436	60
sdhB	Succinate dehydrogenase (small subunit)	X77919	20
sdhCb	Succinate dehydrogenase	L33409
sdhD	Succinate dehydrogenase	X77919	20
gltA	Citrate synthase	M36338	16, 19
sucAb	2-Oxoglutarate dehydrogenase	X77919
sucBb	Dihydrolipoamide succinyltransferase	X77919
sucDb	Succinyl-coenzyme A synthetase	X77919
scvA	Small cell variant protein	L49019	18
trxBb	Thioredoxin reductase	X75627
gyrAb	DNA gyrase (subunit A)	S82903	35
aldCb	α-Acetolactate decarboxylase	U14393
sodB	Superoxide dismutase	M74242	17
htpAB	Heat shock proteins	M20482	49
mucZ	Mucoidy-inducing protein	L42518	59
dutb	dUTP nucleotide hydrolase	X79075
algC	Phosphomannomutase	X79075
pyrEb	Orotate phosphoribosyltransferase	X79075

^aEMBL/GenBank database entry. 

^bPutative gene for which there is no experimental evidence; designation corresponds to homologous genes in the database. 

16S rRNA genes appear only once on the C. burnetii chromosome.

The putative oriC of C. burnetii (43) is located on a 27.5-kb NotI fragment. Gene organization upstream the putative oriC of C. burnetii is identical to that of Pseudomonas putida and nearly identical to that of Bacillus subtilis (Fig. 3). Most strikingly, sequences downstream the oriC of C. burnetii demonstrated a gene organization unique among bacteria. The gyrA gene, which has been found to be clustered in some bacteria with the dnaA, dnaN, recF, and gyrB genes, has been recently identified (35) and was mapped on the 210-kb NotI fragment.

FIG. 3.

Gene organization around the putative oriC of C. burnetii (B) compared to that of P. putida (A) and B. subtilis (C). The region upstream the oriC is almost identical to that of P. putida and B. subtilis (indicated by black boxes). Hypothetical proteins YGI1 and YGI2 of C. burnetii are homologous to YR55 and SpoJ of B. subtilis. Gene organization downstream the putative oriC of C. burnetii is quite different from that of P. putida and B. subtilis, as indicated by white boxes. Whereas in P. putida and B. subtilis genes around the oriC are transcribed in opposite direction, genes in C. burnetii are all transcribed in the same direction (indicated by arrows). Question marks above the white boxes indicate polypeptides without designation and with unknown function.

DISCUSSION

Since the introduction of PFGE (40), strategies for constructing physical maps changed from bottom-up (25, 57) to top-down techniques (10). With bottom-up techniques, genome maps are established by sorting clones from a genomic library. But this procedure is tedious and may become ambiguous if one enters regions of repetitive DNA. Moreover, regions which are difficult to clone represent one of the major problems encountered in bottom-up genome mapping. In contrast, top-down approaches are easily to perform and only few restriction fragments have to be examined for their natural order. Once the physical map is established, locations of a wide variety of other restriction enzymes can be placed on the existing map with relatively little effort.

The macrorestriction map of C. burnetii Nine Mile has been constructed mainly by two methods, PFGE and PCR. Southern hybridization was used to relate cloned fragments to the corresponding NotI fragments, and PCR was used to identify linking fragments.

The size of the C. burnetii Nine Mile chromosome has been determined from PFGE data to be 2,103 kb and thus has been underestimated by several authors (14, 36). Myers et al. (36) used renaturation techniques to determine the chromosome size, and Frazier et al. (14) applied PFGE. Both groups calculated a size of 1,600 kb for the C. burnetii chromosome, which is about 25% less than found in this study. Most obviously, Frazier et al. did not recognize double fragments and/or failed to notice fragments in regions where DNA fragments accumulated. Double fragments in this study were primarily identified by increased band intensities in PFGE gels compared to single bands of similar size. Duplication of these fragments was proven by hybridization to a PFGE blot of C. burnetii DNA digested with a second enzyme, by partial digestion, or by XL PCR. Other mapping strategies use two-dimensional PFGE to circumvent the need for detection of double fragments. However, one general intrinsic problem with two-dimensional PFGE is the reliable detection of smaller fragments due to unfavorably low fluorescence signals of the ethidium bromide stain. Smaller fragments may even diffuse out the agarose plug during digestion.

In most cases, genomes were mapped by two different techniques whereby most often partial digests were combined with the linking clone approach. In this study, linking clones were identified by two PCR methods (checkerboard and adaptor) instead of screening conventional DNA libraries or applying cross-hybridization techniques. The PCR methods circumvent some of the problems associated with cloned linking fragments. For instance, if two unlinked NotI-containing fragments are coligated, aberrant linking clones can be obtained. Cross-hybridization techniques use gel-purified probes to identify linking fragments. Nevertheless, gel-excised fragments contain impurities of randomly sheared DNA, leading to increased background hybridization signals. This increased background may hamper the unambiguous detection of faint hybridization signals generated from probes with very short overlaps. Moreover, in regions where DNA fragments accumulate, it is almost impossible to excise gel plugs without contamination with neighboring bands. Another restriction may be the amount of DNA in this fragment required to produce a reliable probe. A prerequisite to identify linking clones by PCR methods is the availability of sequences to construct primers. Sequences herein were obtained from shotgun cloning experiments. To make sure that clonable sizes are generated after digestion of C. burnetii total DNA, we chose as the second restriction enzyme (apart from EcoRI) Sau3A, which produces DNA fragments below 2 kb in size.

It remains uncertain whether the C. burnetii chromosome is linear or circular. However, several experimental results indicate the presence of a linear chromosome. Thus, it was not possible to clone the ends of the two fragments which would constitute the right and left ends of the linear chromosome, nor was it possible to amplify one of the putative ends (the 7.3-kb fragment) by XL adaptor PCR. In addition, undigested C. burnetii total DNA migrates as a single band on PFGE gels (data not shown), whereas open circular DNA molecules larger than 15 kb fail to migrate in PFGE (28). To date, linear chromosomes have been found in Borrelia burgdorferi (13), Streptomyces lividans (27), Agrobacterium tumefaciens (3), and Rhodococcus fascians (11), but nothing is known about the mechanism of replication in bacteria with linear chromosomes.

The oriC of C. burnetii has not yet been determined unequivocally. Gene organization around the oriC varies according to the taxonomic classification of the organism. Nevertheless, certain bacterial genes involved in replication are clustered. Whereas in all other organisms investigated so far, genes dnaA, dnaN, and gyrB are clustered and in the vicinity of rpmH, none of these genes is present downstream the oriC of C. burnetii (Fig. 3). This unusual gene organization reflects a replication mechanism quite different from that observed in bacteria with circular chromosomes. Therefore, further investigations are feasible to characterize the nature of the ends (e.g., single-stranded loops and telomeric structure) of the linear chromosome. The gyrA gene occupies a central position on the physical map of C. burnetii Nine Mile as has also been demonstrated for the gyrA genes of, for example, Borrelia burgdorferi, Clostridium perfringens, B. subtilis, and Leptospira interrogans (10). Nevertheless, presence of the dnaA gene, which is essential for replication and which is clustered with the gyrA gene in above-mentioned bacteria, remains to be confirmed.

The genetic map of C. burnetii revealed one gene which is recognized as a transposable element to be evenly distributed over the whole chromosome. Hoover et al. (23) demonstrated IS1111a to be present at least 19 times on the C. burnetii Nine Mile chromosome. We have mapped nine copies of the IS1111a gene, indicating that the gene appears more than once on several NotI fragments. Primers derived from the IS1111a element were used to establish a diagnostic PCR (52). Housekeeping genes belonging to the Krebs cycle show gene organizations identical to that of, for example, E. coli (12) or Azotobacter vinelandii (50).

On the C. burnetii chromosome there is only a single copy of the 16S rRNA gene, located on the 118-kb NotI fragment. This finding is in agreement with results published by Afseth and Mallavia (1) and correlates with the slow growth rate of C. burnetii. Though no direct evidence relates the number of rrn operons in an organism to growth rate regulation, slowly growing organisms such as Spiroplasma citri (56), Mycoplasma pneumoniae (26), and Chlamydia trachomatis (4) have been shown to have fewer copies.

As has been shown by DNA solution hybridization (29), C. burnetii isolates have many commonalities. However, RFLP analysis demonstrated a great heterogeneity in restriction patterns (21, 24, 46), possibly due to missense mutations and/or restriction site redistributions. Redistributions again are a result of significant chromosomal rearrangements (e.g., translocations, inversions, insertions, or deletions) which may play an important role in pathogenesis or virulence (47, 48). Similar rearrangements and exchange of DNA blocks have been described for Pseudomonas aeruginosa (38), leading to a 10% variation in chromosome size. Variation in chromosome size is even more striking in C. burnetii isolates ranging from about 1,500 to 2,400 kb. The high degree of DNA homology among C. burnetii isolates suggests that deletions resulting from recombinational (homologous or heterologous) events generated smaller genomes from larger ones. The now existing library of NotI linking fragments will facilitate comparisons of genome organization of different isolates and thus help to elucidate the reason(s) for the great heterogeneity in RFLP patterns. Furthermore, linking fragments may help to clarify which parts of the chromosome are the most conserved or variable ones and may help to answer the question of whether smaller C. burnetii genomes resulted from deletions in larger genomes. If so, this would provide a means for extensive epidemiological studies to evaluate the ubiquitous abundance of C. burnetii. The genetic map of C. burnetii Nine Mile may serve as a basis to identify genes or gene clusters affected by this tremendous loss of genetic information which seems to be no longer necessary for the organism to survive. It has been shown for Borrelia species that gene order is conserved, though nearly all isolates investigated demonstrated unique RFLP patterns (6). Whether this is also true for C. burnetii will be evaluated through comparison of gene order between isolates with different RFLP patterns.