|Home | About | Journals | Submit | Contact Us | Français|
Borrelia hermsii and Borrelia burgdorferi, two closely related spirochetes, are the etiological agents of tick-borne relapsing fever and Lyme disease, respectively. Previous studies have shown the loss of infectivity of B. burgdorferi is associated with in vitro cultivation. This diminished infectivity of B. burgdorferi has occurred as early as three in vitro passages, and the loss of plasmids have been observed with these less virulent to noninfective cultures. The effects of long-term in vitro cultivation on B. hermsii have not been investigated. However, understanding the degree of genomic degradation during in vitro cultivation is important for investigating pathogenic mechanisms of spirochetes. In this study, we analyzed the effects of continuous in vitro cultivation on the genomic composition and infectivity of B. hermsii and B. turicatae. We report that all seven isolates of B. hermsii and the one isolate of B. turicatae examined retained infectivity in mice after 1 year of continuous in vitro cultivation. Furthermore, there were few apparent differences in the plasmid profiles after long-term cultivation. Two isolates of B. hermsii remained infective after high passage despite losing a portion of the 200-kb linear plasmid containing the fhbA gene encoding the factor H binding protein. Also, sequence analysis of multiple B. hermsii isolates demonstrated two types of fhbA with complete congruence with the two genomic groups of B. hermsii spirochetes. Therefore, these results suggest that relapsing fever spirochetes are genetically stable during in vitro cultivation, and the fhbA-containing segment of DNA that is lost during cultivation is not required for infection.
In 1971 Kelly developed a liquid medium that supported the continuous growth of the North American relapsing fever spirochetes Borrelia hermsii, Borrelia turicatae, and Borrelia parkeri (Kelly 1971; Pickett and Kelly 1974). However, Stoenner noted the inefficiency of Kelly medium for cultivating B. hermsii directly from mouse blood (Stoenner 1974). He improved the medium, which allowed cultures to arise from a single organism of B. hermsii, and he renamed the medium fortified Kelly medium (Stoenner et al. 1982).
In late 1981, Burgdorfer and coworkers used fortified Kelly medium to cultivate spirochetes found in the midgut of Ixodes scapularis ticks, which were suspected to be the causative agent of Lyme disease (Burgdorfer et al. 1982), which was later confirmed (Benach et al. 1983; Steere et al. 1983). In 1983, Barbour et al. isolated a spirochete from Ixodes ricinus ticks in a modified form of fortified Kelly medium, changing the name of the medium to Barbour-Stoenner-Kelly medium (BSK; Barbour et al. 1983).
The discovery that Lyme disease was caused by Borrelia burgdorferi opened a new arena of biomedical research. However, Johnson et al. observed that B. burgdorferi lost virulence in Syrian hamsters quickly during cultivation in BSK medium (Johnson et al. 1984), while Barbour noted plasmid loss in an isolate of B. burgdorferi during prolonged in vitro growth (Barbour 1988). Shortly thereafter, Schwan et al. reported that plasmid loss during in vitro cultivation was associated with the loss of infectivity in white-footed mice (Schwan et al. 1988).
The sequence of the B. burgdorferi B31 genome included the identification of 21 plasmids (Casjens et al. 2000; Fraser et al. 1997), facilitating the detection of specific plasmids and genes associated with infectivity, as well as those lost during in vitro cultivation (Labandeira-Rey and Skare 2001; Purser et al. 2003; Purser and Norris 2000). Genomic instability of B. burgdorferi during serial cultivation is problematic for understanding pathogenic mechanisms of this spirochete.
A clue into the genomic stability of relapsing fever spirochetes was first noted during Kelly's initial description of B. hermsii in cultivation, in which he observed that the spirochetes retained infectivity in mice after 8 months of continuous in vitro cultivation (Kelly 1971). This early report of B. hermsii remaining infective after prolonged serial cultivation is strikingly different from B. burgdorferi. However, other than B. burgdorferi, there are few reports investigating the effects of long-term cultivation in other borrelia (Kelly 1971; Levine et al. 1990). Since B. hermsii was first cultivated (Kelly 1971), we now have defined two genomic groups in the species throughout western North America (Porcella et al. 2005; Schwan et al. 2007). Herein, we report a comparative analysis of genomic DNA from short- and long-term subcultures with multiple isolates from both genomic groups of B. hermsii and one isolate of B. turicatae. We also demonstrate that relapsing fever spirochetes retained infectivity in mice after 1 year of continuous cultivation in vitro.
Seven uncloned isolates of B. hermsii (DAH, CoN, MAN, FRO, YOR, HAN, and REN) and one uncloned isolate of B. turicatae (91E135) were used in the study (Porcella et al. 2005; Schwan et al. 2005). The eight isolates were tested for infectivity in mice as previously reported (Porcella et al. 2005), then grown continuously in liquid BSK medium (Barbour 1984) with 104 passages for 1 year. Low passage cultures were subjected to fewer than 10 in vitro passages except for CON, whose initial passage history is unknown. Isolates were grown at 35°C in 15 mL Falcon tissue culture tubes (Becton Dickinson Labware, Franklin Lakes, NJ) to approximately 108 spirochetes per milliliter, and every 3.5 days (twice a week), 300 μL were passed into 9 mL of fresh BSK medium. Because the stationary cultures contained approximately 108 spirochetes per milliliter, the 300 μL used to inoculate each passage contained approximately 3.2 × 106 spirochetes. Based on this inoculum, each passage displayed five doublings, primarily during the exponential phase. Therefore, the 104 passages during the 1 year of continuous cultivation represented approximately 520 generations.
To determine the infectivity of high passage isolates, B. hermsii isolates and B. turicatae were injected intraperitoneally into three Rocky Mountain Laboratory (RML) mice (a closed colony used at the RML since 1937 that originated from outbred Swiss-Webster mice) per isolate at 1 × 105 spirochetes per mouse. For 3 days postinoculation, blood was obtained by tail clip. Spirochetemias were quantified by bright-field microscopy (Schwan et al. 2003). All animal work was in compliance with the Rocky Mountain Laboratories Animal Care and Use Committee (Protocol # 2006–10).
Whole-cell lysates of the low- and high-passaged B. hermsii isolates were prepared as previously described (Schwan et al. 1989). Proteins were separated by one-dimensional sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) with Laemmli buffer (Laemmli 1970). Proteins were stained in one gel with Coomassie brilliant blue and blotted onto nitrocellulose membranes with Towbin buffer (Towbin et al. 1979) and a transblot cell (Bio-Rad Laboratories, Hercules, CA). The membranes were blocked overnight at room temperature with TSE-Tween (50 mM Tris [pH 7.4], 150 mM NaCl, 5 mM ethylenediamine-tetraacetic acid, 0.05% Tween 20). Membranes were incubated with either rabbit anti-B. hermsii DAH hyperimmune serum (1:100), a convalescent serum sample (1:100) from a human relapsing patient, or antibodies specific for the variable tick protein. Bound antibodies were detected with the ECL™ Western Blotting Detection Reagent (GE Healthcare, Buckinghamshire, UK).
Genomic DNA samples were prepared from the isolates prior to and after 520 generations grown in vitro (Simpson et al. 1990) and compared by reverse-field agarose gel electrophoresis as previously described (Porcella et al. 2005).
Total genomic DNA was separated by reverse-field agarose gel electrophoresis (Porcella et al. 2005) and transferred to a MagnaGraph Nylon Transfer Membrane (Osmonics Inc., Minnetonka, MN) by the method of Southern (Southern 1975). Primers (Invitrogen, Carlsbad, CA) for hybridization probes were designed using B. hermsii DAH genomic sequence (Schwan and Porcella, unpublished) for resT, pncA, chbC, fhbA, and blyA (Stevenson et al. 2000; Table 1). Primers for the vsp and vlp families were described previously (Hinnebusch et al. 1998). Additionally, the vsp and vlp families were chosen because they are located on multiple linear plasmids in the 25- to 35-kb range (Kitten and Barbour 1990; Plasterk et al. 1985).
Hybridization probes were produced from genomic DNA of B. turicatae (resT, pncA, and chbC), B. hermsii DAH (blyA), and B. hermsii YoR (fhbA) with the polymerase chain reaction (PCR) DIG (digoxigenin) probe synthesis kit as specified by the manufacturer (Roche Applied Science, Indianapolis, IN). PCRs began with heating at 96°C for 5 minutes, followed with amplification of 35 cycles at a melting temperature of 94°C for 30 seconds, annealing temperature of 55°C for 30 seconds, and an extension temperature of 72°C for 2 minutes. After the 35th cycle, an additional extension was done at 72°C for 5 minutes. Southern blots were probed and developed as previously described (Schwan et al. 2005) with the CDP-Star chemiluminescence kit (Roche Applied Science, Indianapolis, IN). Southern blots for the vsp and vlp loci involved pooling the hybridization probes.
PCR analysis and sequencing of this locus was performed to further analyze the genomic instability and heterogeneity of the factor H binding protein gene fhbA as previously reported (Hovis et al. 2006). A total of 35 isolates was used (34 B. hermsii isolates and one B. turicatae isolate) including the isolates used elsewhere (Hovis et al. 2006). The entire fhbA was PCR amplified as stated above from total genomic DNA with primers flanking the gene (Table 1). PCR amplicons were visualized in agarose gels stained with ethidium bromide, and reaction products that contained a single amplicon of the predicted size were processed with the Qiagen PCR Purification Kit (QIAGEN, Inc., Valencia, CA). Sequencing was performed as previously described (Schwan et al. 2005). Briefly, the mixtures were run with a PTC-225 DNA Engine Tetrad Thermal Cycler (MJ Research, Inc., Waltham, MA). Nucleotide sequences were analyzed with the Sequencher 4.1 (Gene Codes Corp., Ann Arbor, MI) and MacVector 6.0 (oxford Molecular, Beaverton, oR) software packages. DNA alignments were constructed with the Megalign and ClustalV programs in the Lasergene software package (DNASTAR, Madison, WI), and transferred into the MacClade program for manual correction (Maddison and Maddison 2003). MacClade output files were opened in PAUP (Swofford 1998), where maximum-likelihood neighbor-joining trees were created. A full heuristic search with 1,000 bootstrap replicates was performed to test the robustness of clade designations.
Nucleotide sequences of the entire fhbA gene of B. turicatae 91E135 and 33 isolates of B. hermsii have been deposited in the GenBank database under accession numbers EF411135 through EF411169. OurfhbA sequence for B. hermsii YOR was identical to that determined previously (Hovis et al. 2004) (DQ020176); therefore, we did not duplicate this sequence in the database.
All eight isolates of relapsing fever spirochetes retained their infectivity in mice after 1 year of continuous serial cultivation by producing detectable spirochetemias in most mice 2 days postinoculation (Table 2). The REN isolate produced the highest spirochetemia with an average of 7.02 × 105 spirochetes per milliliter of blood, while HAN produced the lowest detectable spirochetemia with an average of 1.0 × 104 spirochetes per milliliter of blood. One mouse inoculated with CoN did not yield a detectable spirochetemia 2 days postinoculation, but spirochetes were detected the next day, indicating the bacteria had retained infectivity.
Whole-cell lysates of the low- and high-passaged cultures of the seven isolates of B. hermsii showed few changes during the 1 year of continuous cultivation. In all analyses, the only obvious changes in two of the isolates were an increase during cultivation of the Vtp, the variable tick protein, which was shown previously to arise during in vitro growth (data not shown; Schwan and Hinnebusch 1998; Stoenner et al. 1982). We had expected to see a loss of one protein, FhbA, in the high-passage REN culture, but our immune sera apparently were not reactive with this protein.
Few changes in plasmid profiles were observed in spirochetes after 1 year of continuous cultivation (Fig. 1). High-passage cultures of B. hermsii REN and B. turicatae contained linear plasmids of 170 kb and 56 kb, respectively, indicating rearrangements or deletions of genomic DNA during long-term subcultures. Also, high-passage MAN contained an additional circular plasmid (Fig. 1).
Several genes encoded on different plasmids were used to examine plasmid retention after 520 generations in vitro. Southern blot analysis of resT (Fig. 2A), pncA (Fig. 2B), chbC (Fig. 2C), and fhbA (Fig. 2D) confirmed the retention of plasmids encoding these genes in high-passage cultures. Retention of the 30-kb circular plasmid was confirmed by detection of blyA, with no apparent discrepancies between low-and high-passage cultures (data not shown).
Previous analysis of the vsp and vlp gene families indicated multiple plasmid locations of these genes within the B. hermsii genome (Hinnebusch et al. 1998; Kitten and Barbour 1990; Plasterk et al. 1985), making them ideal candidates for analysis of plasmid retention between low- and high-passage isolates. Detection of the vsp and vlp loci confirmed the retention of the linear plasmids encoding these sequences in high-passage cultures (data not shown). Also, the 25- to 35-kb linear plasmids were retained in all highpassage isolates except REN (Fig. 1), which may not contain vsp or vlp loci on these given plasmids. This finding in REN was consistent with a previous report (Hinnebusch et al. 1998).
resT mapped to the 53-kb linear plasmid in most isolates. However, in HAN, YOR, and B. turicatae, resT was located on plasmids of differing size. Additionally, in B. turicatae, resT mapped to two locations in both low- and high-passage cultures (Fig. 2A), suggesting a population of bacteria that contained resT on plasmids of different sizes.
pncA was located on a linear plasmid of approximately 25 kb and was retained in all the high-passage relapsing fever spirochete cultures (Fig. 2B). In B. turicatae, pncA was similar to resT in that it appeared to have copies on plasmids of two sizes in both low- and high-passage cultures.
chbC was located on the 200-kb linear plasmid in most B. hermsii isolates (Fig. 2C). In B. hermsii CON and B. turicatae, chbC was located on a 170- and 52-kb linear plasmid, respectively. In low-passage REN, chbC was located on the 200kb linear plasmid, but in the high passage culture it was on a 170-kb linear plasmid (Fig. 2C), indicating a reduction in size of the 200-kb linear plasmid.
fhbA was also present on the 200-kb linear plasmid in all low-passage cultures and in most high-passage cultures (Fig. 2D). However, it was undetectable in B. hermsii CON and REN after prolonged in vitro cultivation. Additionally, in low-passage CON and high-passage B. turicatae, detection of fhbA via Southern blot was diminished compared to other isolates, suggesting it was lost during cultivation as previously reported (Hovis et al. 2006).
Given the apparent loss of the fhbA locus in some isolates during cultivation, the fhbA locus of 34 B. hermsii isolates was sequenced, which identified only three alleles, one fhbA1 allele and two fhbA2 alleles. All low-passage isolates contained fhbA1 or fhbA2. Alignment of the sequences identified two deeply branched B. hermsii fhbA gene clusters as previously demonstrated (Hovis et al. 2006; Fig. 3). However, we found no incongruity between fhbA types and genomic groups; all fhbA1 types belonged to GGI and all fhbA2 types belonged to GGII.
Although early evidence indicated that B. hermsii retained infectivity during in vitro cultivation (Kelly 1971), we present the first detailed analysis of plasmid retention and infectivity of relapsing fever spirochetes. In contrast to Lyme disease spirochetes, which lose infectivity after several subcultures (Barbour 1988; Busch et al. 1997; Johnson et al. 1984; Levine et al. 1990; Norris et al. 1995; Purser and Norris 2000; Schwan et al. 1988; Simpson et al. 1990), relapsing fever spirochetes retained their ability to infect mice after 520 generations in vitro, with little evidence of plasmid instability. B. turicatae and B. hermsii REN and CON underwent plasmid alterations, where in REN, the 200-kb linear plasmid was truncated to a 170-kb linear plasmid. Although fhbA was no longer present in the high-passage culture of REN, the presence of chbC on the 170-kb linear plasmid indicated that the change in plasmid size was due in part to the loss of a segment of DNA that included fhbA. B. hermsii CON also lost fhbA during cultivation, but unlike REN, the low- and high-passage cultures of CON had chbC on the 170-kb linear plasmid. This was not completely unexpected because prior to our acquisition of CON, this isolate had been passed an unknown number of times, and a subpopulation of CON may have begun to lose a portion of the 200-kb linear plasmid. These findings with CON and REN suggest that truncation of the full-length 200-kb linear plasmid occurred during cultivation. Interestingly, high-passage isolates that lost fhbA were still infective, although Hovis et al. suggested that fhbA is required for B. hermsii to persistently infect mice (Hovis et al. 2008). Additionally, Colombo and Alugupalli reported that FhbA in B. hermsii is a surface antigen stimulating an immunoglobulin M response that produces a protective immune response (Colombo and Alugupalli 2008).
Although specific probes were not available for all plasmids, probes for chbC, fhbA, blyA, pncA, resT, and the vsp and vlp families provided significant insight into plasmid stability in relapsing fever spirochetes during prolonged cultivation. However, in high-passage MAN, a second hybridizing band appeared above the 30-kb circular plasmid, indicating possible plasmid dimerization or conversion of a linear plasmid to a circular plasmid, an observation previously reported in B. hermsii (Ferdows et al. 1996).
Although resT was retained in the high-passage cultures, in B. turicatae there were copies of this gene on different-sized plasmids, indicating genomic rearrangements during cultivation. Also, the larger resT-bearing plasmid increased in abundance during the year of continuous cultivation (Figs. 1 and and2).2). During passage, recombination of additional DNA may have occurred with the 53-kb linear plasmid, producing the larger resT-bearing linear plasmid. Such a rearrangement between plasmids occurred in B. hermsii HS1 during serial cultivation, with a portion of the largest linear plasmid recombining with the 30-kb linear plasmid, producing a 40-kb linear plasmid (Hinnebusch et al. 1998).
The ability of relapsing fever spirochetes to retain their plasmids and infectivity after prolonged in vitro cultivation contrasts with Lyme disease spirochetes (Barbour 1988; Busch et al. 1997; Johnson et al. 1984; Levine et al. 1990; Norris et al. 1995; Purser and Norris 2000; Schwan et al. 1988; Simpson et al. 1990). Although B. hermsii and B. burgdorferi have similar-sized genomes, B. burgdorferi contains 21 plasmids (12 linear and nine circular plasmids; Casjens et al. 2000; Fraser et al. 1997), although genomic group I isolates of B. hermsii contain only 12 plasmids (seven linear and five circular plasmids; Dai et al. 2006; Stevenson et al. 2000). The condensed plasmid portion of the genome of relapsing fever spirochetes may have resulted in essential genes such as resT and pncA that are required for growth and infectivity, respectively (Byram et al. 2004; Purser et al. 2003), being distributed on all plasmids.
Alternatively, relapsing fever spirochetes may have a unique mechanism for genomic stability. B. hermsii grows repeatedly to cell densities up to 108 spirochetes per milliliter in the blood of infected animals (Barbour et al. 2006; Coffey and Eveland 1967a, 1967b; Stoenner et al. 1982). This cyclic pattern of infection is mimicked during in vitro growth, when a small portion of the population of spirochetes is repeatedly grown to high densities. In contrast, the number of Lyme disease spirochetes in the blood is low and only detectable by quantitative PCR or cultivation with no sign of cyclic growth, comparable to B. hermsii (Benach et al. 1983; Dolan et al. 2004; Maraspin et al. 2001; Schwan et al. 2003; Steere et al. 1983; Wang et al. 2001; Wormser et al. 2000; Zeidner et al. 2001). Therefore, during the diversification of the genus Borrelia into two main groups, B. burgdorferi may have lost mechanisms for genomic stability that only become evident when forced to grow repeatedly to high cell numbers during serial in vitro cultivation.
Several of our results conflict with data and conclusions reported elsewhere (Hovis et al. 2006). First and most important is our finding of total congruence between the B. hermsii fhbA type and genomic group. All isolates with fhbA1 were in GGI, and all isolates with fhbA2 were in GGII. Therefore, we found no evidence for the lateral transfer of this locus from spirochetes in one genomic group to the other. Second, we found only three fhbA alleles among the 34 isolates, while Hovis et al. reported seven alleles in 10 of the same isolates (Hovis et al. 2006). We sequenced amplicons produced with primers that flanked the open reading frame (ORF). Hovis et al. sequenced amplicons produced with primers with sequences within the ORF that were designed from the fhbA2 sequence of B. hermsii YOR, a GGII isolate (Hovis et al. 2006). By using primers with sequences unique to fhbA2 to amplify fhbA1 from GGI isolates (Hovis et al. 2006), erroneous bases were likely incorporated into the fhbA1 amplicons. Additionally, we do not know the basis for the discrepancy between our fhbA phylogram and that reported previously (Hovis et al. 2006), but they cannot be due to a different source of material because the same isolates were included for both studies. Regardless, here we demonstrate that the fhbA locus clearly defined the genomic group for each isolate of B. hermsii. The significance of factor H binding by B. hermsii toward the pathogenesis of relapsing fever remains to be determined as high-passage isolates lacking this locus remained infectious. In addition, the genomic stability of relapsing fever spirochetes will be advantageous when applying recently developed gene inactivation tools (Battisti et al. 2008) to study the role of specific genes like fhbA in the pathogenicity of these tick-borne bacteria.
We thank Mollie W. Jewett, Kit Tilly, and B. Joseph Hinnebusch for reviewing this manuscript; Anita Mora and Gary Hettrick for photographic assistance; and Bob Karstens for technical assistance. This work was supported by the Division of Intramural Research, NIAID, NIH.