In order to determine what the cellular targets of ROS in B. burgdorferi are, we first needed to determine the sensitivity of Borrelia cells to various oxidants. Microaerobic cultures of B. burgdorferi strain B31A3 were grown to a cell density of 5 ? 10⁷ cells ml^?1, treated with varying concentrations of H₂O₂ (0?50 mM) or t-butyl hydroperoxide (0?50 mM) and the number of surviving cells determined by plating. The results are shown in Fig. 1. When cells were exposed to 10 mM t-butyl hydroperoxide, approximately 50% of the cells survive (Fig. 1A). In contrast, when E. coli cells were exposed to 1 mM t-butyl hydroperoxide, approximately 1% of the cells survive. B. burgdorferi cells were much more resistant to t-butyl hydroperoxide than E. coli cells, with a survival rate of approximately 100% at 1 mM t-butyl hydroperoxide. This trend was observed when cells were exposed to H₂O₂. E. coli cells exposed to 1 mM H₂O₂ have approximately 10% survivability, while 100% of the B. burgdorferi cells survive at this concentration, and approximately 80% survive when exposed to 10 mM H₂O₂ (Fig. 1B). Similar results were obtained when cells were exposed in HN (Hepes-NaCl) buffer, suggesting that the high-level resistance to ROS was not due to the interaction of ROS with components of Barbour-Stoenner-Kelly (BSK-II) growth medium. Taken together, these data indicated that B. burgdorferi strain B31A3 was highly resistant to exposure to both t-butyl peroxide and H₂O₂.

For most bacteria, DNA is the major target of ROS causing a wide variety of DNA lesions. This is in part due to the localization of ?free? Fe²⁺ along the phosphodiester backbone of nucleic acids, putting the DNA in close proximity to the active species formed via the Fenton reaction. However, B. burgdorferi has been shown to harbour few genes encoding orthologues of known iron-containing proteins, does not require Fe for growth and has intracellular Fe concentrations estimated to be < 10 atoms per cell (^{Posey and Gherardini, 2000}). Taken together, these observations strongly suggest that B. burgdorferi DNA is not a major target for ROS. To test this hypothesis, different techniques were used to measure DNA damage in B. burgdorferi cells after exposure to ROS.

One reliable indicator of DNA damage by ROS in a cell is an increase in the spontaneous mutation rate. In B. burgdorferi, mutations that confer resistance to coumermycin A1, which targets the ? subunit of DNA gyrase, have been mapped to gyrB, the gene encoding DNA gyrase B (^{Samuels et al., 1994}). In each case, a single point mutation correlated with this drug resistance. To determine if exposure to oxidants increases DNA damage by increasing point mutations, B. burgdorferi B31A3 cells grown under microaerobic conditions were treated with 5 mM H₂O₂ and plated in the presence and absence of 250 ng ml^?1 coumermycin A1. The mutation frequency was calculated as the number of colonies that are Cou^R per total number of cells plated. The spontaneous resistance frequency of treated cells was approximately equivalent to that of untreated cells, 8.8 ? 10^?8 and 1.33 ? 10^?7, respectively, indicating no increase in the number of point mutations (Fig. 2A). Also, no increase in point mutations was observed when cells were treated with higher concentrations of H₂O₂ (10 mM) or when treated with t-butyl hydroperoxide (5 and 10 mM) (data not shown).

Another effective way to determine DNA damage is by measuring the number of apurinic/apyrimidinic sites (AP). AP sites in DNA, where the DNA base is lost, can be generated spontaneously under physiological conditions by hydrolysis of the N-glycosylic bond, or can also be formed by DNA-damaging agents (^{Lindahl and Nyberg, 1972}), such as UV, alkylating agents or OH. They are also intermediates in the base excision repair pathway (BER) (^{Weiss and Grossman, 1987}; ^{Friedberg and Hanawalt, 1988}; ^{Wallace, 1988}). Thus, the cellular steady state level of AP sites would increase as a consequence of base modifications and their subsequent repair. As AP sites are bypassed inefficiently by DNA polymerase in bacterial cells, DNA lesions can result in a significant block in DNA replication. Therefore, the number of AP sites can serve as a sensitive indicator of DNA damage resulting from oxidative stress (^{Kubo et al., 1992}). To determine whether ROS can damage B. burgdorferi DNA, strain B31A3 cells grown under microaerobic conditions were treated with H₂O₂ (1 or 10 mM), t-butyl hydroperoxide (1 or 10 mM) or lipoxidase (an enzyme which specifically catalyses the hydroperoxidation of lipids containing a cis, cis-pentadiene structure, such as linoleic acid). The DNA was isolated and assayed for AP sites. The results are shown in Fig. 2B. In all cases, the numbers of AP sites per 10⁵ bp DNA were equivalent, indicating that the addition of oxidants did not increase the number of DNA base lesions. In contrast, when E. coli strain TA4315 cells (ahpCF) (^{Storz et al., 1989}) were treated with 100 ?M H₂O₂, the number of AP sites increased ?10-fold over untreated cells. These data suggest that B. burgdorferi DNA was not a major target for oxidative damage under these conditions.

It is of note to mention that the total number of AP sites per 10⁵ bp DNA is approximately 10-fold higher in untreated Borrelia B31A3 DNA than in untreated E. coli DNA. The Borrelia genome consists of a linear chromosome and multiple linear and circular plasmids. Numerous observations indicate that functional telomeres require interaction with DNA damage repair proteins, suggesting that the DNA damage repair machinery, including the BER, is involved in replication of telomeres and protection of functional chromosome ends (^{Verdun and Karlseder, 2007}). An important intermediate in BER is the apurinic or abasic site. Therefore, it was possible that the higher numbers of AP sites in untreated Borrelia DNA was due to the number of telomeres present in the genome. The B. burgdorferi strain B31A3 harbours 11 linear plasmids and a linear chromosome with ?100 unpaired bases in the telomere loops per genome or 10 bases per 10⁵ bp DNA (^{Hinnebusch and Barbour, 1991}; ^{Chaconas, 2005}). Untreated Borrelia B31A3 contains 11 ? 3.8 AP sites per 10⁵ bp DNA, suggesting that the elevated number of AP sites in Borrelia was due to the number of telomeres. To test this hypothesis, the AP sites were measured in DNA isolated from untreated B. burgdorferi strain B314 which harbours no linear plasmids (^{Sadziene et al., 1993}). The number of unpaired bases in the telomere loops per genome was estimated to be 10, which corresponds to 1 base per 10⁵ bp DNA, and only 3.6 ? 0.2 AP sites per 10⁵ bp DNA were detected, supporting the hypothesis that the high number of AP sites in untreated Borrelia DNA was due to the number of unpaired bases in the telomeres. The analyses of the AP sites strongly suggested that B. burgdorferi DNA was not the major target for oxidative damage.

In addition to the generation of abasic sites, oxygen radicals often damage DNA through the formation of 8-oxoguanine lesions (^{Nakamura et al., 2000}). 8-Oxoguanine, through its ability to mispair with bases other than cytosine, likely plays a role in DNA mutagenesis. Consequently, 8-oxoguanine is often used as a marker of oxidized DNA damage. To determine whether ROS can damage B. burgdorferi DNA and cause 8-oxoguanine lesions, strain B31A3 cells grown under microaerobic conditions were treated with H₂O₂ (5 or 10 mM), t-butyl hydroperoxide (5 or 10 mM) or lipoxidase. The DNA was isolated and assayed for 8-oxoguanine using an Enzyme-Linked ImmunoSorbent Assay. The results are shown in Fig. 2C. In all cases, the amount of 8-oxoguanine in Borrelia cells was below the detection limit of the assay. In contrast, when a MutM (the specific glycosylase that removes the 8-oxoguanine)-deficient E. coli strain was treated with 100 ?M H₂O₂, the amount of 8-oxoguanine sites increased ?fivefold over untreated cells. It is important to point out that no MutM homologue has been identified in the genome of B. burgdorferi. Taken together, these data suggested that B. burgdorferi DNA was not a major target for oxidative damage under these conditions.

In eukaryotes, membrane lipids are a major target of ROS. Free radicals can attack polyunsaturated fatty acids, such as linoleic acid and linolenic acid, in membranes and initiate lipid peroxidation. This reaction can cascade throughout the membrane to adjacent polyunsaturated fatty acids, decreasing membrane fluidity and generating more toxic products such as aldehydes (^{Imlay, 2003}). Because most bacterial membranes contain saturated and monounsaturated fatty acids rather than ?reactive? polyunsaturated lipids, peroxidation of lipids in bacterial membranes is not considered a problem. As B. burgdorferi cannot synthesize their own lipids, they must instead scavenge them. Therefore, it seems likely that their membrane composition would reflect the host's lipid profile or that of their growth medium (^{Barbour, 1984}; ^{Fraser et al., 1997}) and would contain some polyunsaturated fatty acids. To determine if B. burgdorferi contains polyunsaturated lipids, B. burgdorferi strain B31A3 was grown under anaerobic conditions and analysed for fatty acid composition by Lipid Technologies (Austin, MN). The results are shown in Table 1 and are reported as percentage of total fatty acid content. Linoleic acid comprised ?10% of the total lipid content and the linolenic acid content was measured to be ?1%, indicating that Borrelia cells do contain lipids that are susceptible to ROS damage. The amount of linoleic acid and linolenic acid present in Borrelia reflected the amount present in the media. Therefore, these results suggested that the amount of these fatty acids in the membranes would vary as availability varies.

To determine if Borrelia polyunsaturated lipids can undergo lipid peroxidation, B. burgdorferi B31A3 cells grown microaerobically were treated with 1 mM t-butyl hydroperoxide or 250 mg of lipoxidase, and the cell pellets were analysed for fatty acid composition (Industrial Laboratory). The results are shown in Table 2 and are reported as the percentage of total cell mass. The per cent of linoleic acid in the total cell mass decreased with treatment, while the levels of oleic acid (c18:1n9) and pentadecanoic acid (c15:0) remained relatively constant. Untreated cells contained 0.04% linoleic acid, while cells treated with t-butyl hydroperoxide contained 10-fold less (0.004%) linoleic acid and cells treated with lipoxidase had no detectable linoleic acid present in the sample. These data indicated that the linoleic acid present in B. burgdorferi membranes can be oxidized by ROS.

Malondialdehyde (MDA) is generated as a relatively stable end-product from the oxidative degradation of polyunsaturated fatty acids. MDA has thus been used as an indicator of lipid peroxidation (^{Gutteridge and Halliwell, 1990}; ^{Esterbauer et al., 1991}). To further demonstrate that Borrelia lipids can undergo lipid peroxidation, B31A3 cells grown microaerobically were treated with 5 mM AAPH (free radical generator) or 250 mg of lipoxidase and MDA measured (^{Seljeskog et al., 2006}). The results are shown in Fig. 3. Untreated cells contained ?16.5 ?M of MDA per mg of protein. When the cells were treated with AAPH or lipoxidase, the amount of MDA increased ?1.5-fold (27.3 ?M of MDA per mg of protein) and approximately twofold (33 ?M of MDA per mg of protein) respectively (Fig. 3A). As a control, eukaryotic cells (mouse myeloma cells SP2) were treated with AAPH and MDA measured (Fig. 3B). Mouse myeloma cells have been used as a model system for the determination of phospholipid hydroperoxides (^{Chotimarkorn et al., 2005}). In this case, the amount of MDA increased approximately threefold (36.3 ?M of MDA per mg of protein) in the treated cells versus untreated cells (11.7 ?M of MDA per mg of protein). Taken together, these data suggested that, like eukaryotic membranes, Borrelia membrane lipids were capable of undergoing lipid peroxidation.

As shown in Fig. 3A, there is a measurable quantity of MDA present even in untreated cells, suggesting that the membrane lipids are damaged without the addition of exogenous ROS. Oxidative damage is an unavoidable by-product of growth in an oxygen environment because superoxide anions and H₂O₂ are formed whenever molecular oxygen chemically oxidizes electron carriers. To determine if the Borrelia lipids are damaged from growth in an oxygen environment, B. burgdorferi B31A3 cells were grown under anaerobic, microaerobic and aerobic conditions and MDA measured. Figure 3C demonstrates that as the oxygen concentration increased, the amount of MDA increased. Cells grown under anaerobic conditions contained the lowest amount of MDA (7.6 ?M of MDA per mg of protein), approximately twofold less than the measured amount in microaerobic cells (16.5 ?M of MDA per mg of protein). Aerobically grown cells contained the highest amount of MDA, ?1.5-fold greater than that observed in microaerobic cells (28.8 ?M of MDA per mg of protein) and ?3.7-fold greater than that observed in anaerobically grown cells. These data suggested that Borrelia lipids can be damaged during aerobic growth.

Our High Performance Liquid Chromatography analyses of the MDA present in anaerobically grown B. burgdorferi cells showed a small peak with a retention time similar to that of the MDA standard. This was puzzling as little or no lipid peroxidation should occur under these conditions. To further characterize this ?MDA? peak, a three-dimensional diode array spectra was generated by scanning each sample during the elution of the peak (Fig. 4, lower sections). An authentic MDA standard was also scanned as a control (Fig. 4A, lower section). In the anaerobically grown cells, the spectrum shows that two compounds with different absorbance maximums (Fig. 4B, lower section) comprised the single retention time peak from the HPLC chromatogram (Fig. 4B, upper section). Based on this spectrum, the amount of actual MDA contributed < 15% of the total amount of material detected in the HPLC peak while the second contaminating peak contributed > 85%. Therefore, the amount of MDA in untreated anaerobically grown cells was considerably less than the 7.6 ?M of MDA per mg of protein actually measured. However, in cells grown under microaerobic conditions, the MDA peak contributed more to the overall retention time peak when compared with the anaerobic spectrum, while the second contaminating peak stays relatively constant (Fig. 4C, lower section). These spectra demonstrated that the increase in the MDA retention time peak between anaerobically and aerobically grown cells was due to the increase in the amount of authentic MDA present.

The fluorescent probe diphenyl-1-pyrenylphophine (DPPP) has been used for detection of lipid hydroperoxides in cell membranes (^{Okimoto et al., 2000}; ^{Takahashi et al., 2001}). In this method, the hydroperoxides are reduced with DPPP, resulting in the formation of the fluorescent arylphosphine oxide. To visualize the lipid hydroperoxides, B. burgdorferi B31A3 cells were labelled with DPPP and observed by fluorescence microscopy (Fig. 5B). E. coli Top10 cells and mouse myeloma SP2 cells were also labelled and visualized to serve as negative and positive controls respectively (Fig. 5A and C). Red Fluorescent dye was used to visualize the cell membranes. A strong fluorescence of DPPP was observed for B31A3 cells (Fig. 5B) and for the mouse myeloma cells (Fig. 5C), but not for the E. coli cells (Fig. 5A). An overlay of the two dyes demonstrates that the DPPP fluorescence of both the Borrelia and myeloma cells corresponds to the areas of red fluorescence. This indicated that lipid hydroperoxides were present on the Borrelia cell membranes and suggested that the membranes were damaged.

To further demonstrate Borrelia membrane damage, B. burgdorferi B31A3 cells were grown under anaerobic and microaerobic conditions and visualized by negative stain using an electron microscope. Additionally, cells grown under microaerobic conditions were treated with 250 mg of lipoxidase and visualized by Electron Microscopy. Intact membranes were observed in cultures of B31A3 grown under anaerobic and microaerobic conditions (Fig. 6A and B respectively). However, in cultures treated with lipoxidase (Fig. 6C), a significant number of membrane blebs were seen surrounding the spirochetes, indicating membrane damage. Taken together, these data indicated that Borrelia membranes were a target for oxidative stress.

Borrelia burgdorferi strain B31A3 and strain B314 (^{Sadziene et al., 1993}) were grown in modified BSK-II (^{Barbour, 1984}) medium at 34?C under an atmosphere of 0?20% O₂ with 5% CO₂ and the balance N₂. Cells density was determined using a dark-field microscope. All reagents were purchased from Sigma Chemicals, St. Louis, MO unless stated otherwise. E. coli strain Top10, strain TA4315 (ahpCF) (^{Storz et al., 1989}) and CM1319 (mutM) (^{Bridges et al., 1996}) were grown in Luria?Bertani (LB) at 37?C with shaking.

Borrelia burgdorferi strain B31A3 was grown to a cell density of 5 ? 10⁷ cells ml^?1 in BSK II under microaerobic (3% O₂) conditions, the culture split and the cells treated with varying concentrations of oxidants (0?50 mM t-butyl hydroperoxide or H₂O₂) at 34?C for 4 h. After the incubation, cells were diluted in fresh BSK II, plated on BSK plates and incubated 7?14 days at 34?C. Per cent survivability was calculated as the number of colonies on the treated plates versus the number of colonies on the untreated plates.

To determine spontaneous resistance to coumermycin A1, B. burgdorferi B31A3 cells were grown under microaerobic conditions to a cell density of 5 ? 10⁷ cells ml^?1 and treated with 5 mM H₂O₂ for 1 h at 34?C. The cells were plated on BSK plates containing 0 or 250 ng ml^?1 coumermycin A1 and incubated 7?14 days at 34?C. The resistance frequency was calculated as the number of colonies that are Cou^R per total number of cells plated.

Borrelia burgdorferi B31A3 cells were grown in BSK-II under microaerobic conditions to a cell density of 5 ? 10⁷ cells ml^?1, treated with various oxidants (1 or 10 mM t-butyl hydroperoxide, 1 or 10 mM H₂O₂, 10 mM paraquat or lipoxidase) for 4 h and total DNA was isolated using Wizard Genomic DNA Purification Kit (Promega Corp., Madison, WI). The number of base lesions was determined using the DNA Damage Quantification Colorimetric Assay kit (Oxford Biomedical Research, Oxford, MI) following the manufacturer's protocol. Briefly, 500 ng of DNA was mixed with an equal volume of 10 mM biotinylated aldehyde reactive probe (ARP) reagent and incubated for 1 h at 37?C. The DNA-ARP product was ethanol-precipitated using glycogen as a carrier, washed three times with 70% ethanol and resuspended in Tris-EDTA to give a final concentration of 0.5 ?g ml^?1. The DNA-ARP product was allowed to bind to the wells of 96-well microplate overnight at 37?C. After the binding, the wells were washed four times with TPBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₃HPO_4, 2 mM KH₂PO₄, 0.5% Tween 20, pH 7.4). The HRP-streptavidin conjugate was diluted to 0.5 ?g ml^?1 in Assay Buffer (0.15 M NaCl, 10 mM Na_sHPO₄, 1.5 mM KH₂PO₄, 2.5 mM KCl, 5 mg ml^?1 BSA, 0.1% Tween, pH 7.5), 100 ?l was added to each well and the plate incubated at 100 r.p.m. for 1 h at room temperature (RT). After incubation, the wells were washed four times with TPBS, 100 ?l of substrate was added to each well and incubated for 1 h at 37?C. The reaction was then quenched with 100 ?l of 1 M sulphuric acid and the reaction was monitored at 450 nm. The number of aldehyde reactive probe (DNA base lesions) per 10⁵ bp DNA was determined using a standard curve. As a control, E. coli TA4315 cells were grown in minimal media to OD₆₀₀ of 0.4, treated with 0 or 100 ?M H₂O₂ for 30 min and DNA isolated. The number of base lesions was determined as described above. B. burgdorferi strain B314 cells were grown under microaerobic conditions and DNA isolated. The number of DNA lesions was determined as described.

To determine the amount of 8-oxoguanine in B. burgdorferi DNA, cells were grown and treated as described above and DNA isolated. The DNA was converted to single-stranded DNA by boiling the sample for 5 min followed by rapid chill on ice. The DNA was then digested with nuclease P1 for 2 h and then treated with alkaline phosphatase for 1 h following manufacturer's protocols. The resultant mixture was then centrifuged for 5 min at 6000 g and the supernatant used for the 8-oxoguanine ELISA assay (Oxford Biomedical Research, Oxford, MI). As a control, E. coli CM1319 (mutM) cells were grown in LB to OD₆₀₀ of 0.4, treated with 0 or 100 ?M H₂O₂ for 1 h and DNA isolated. The concentration of 8-oxoguanine was determined as above.

To determine fatty acid content in B. burgdorferi total membranes, B. burgdorferi strain B31A3 was grown under anaerobic conditions to a cell density of 5 ? 10⁷ cells ml^?1, harvested by centrifugation (5000 g, 15 min, 4?C) and washed two times with HN (20 mM NaCl, 50 mM Hepes, pH 7.6) buffer. Cell pellets were analysed for fatty acid composition by fatty acid methyl ester (FAME) gas chromatography (Lipid Technologies, Austin, MN) and results are reported as percentage of total fatty acid content. To determine the effects of oxidants on fatty acid composition, a 1.5 l culture of B. burgdorferi strain B31A3 was grown under microaerobic conditions to a cell density of 5 ? 10⁷ cells ml^?1. The culture was split into 500 ml aliquots, the first was treated with 1 mM t-butyl hydroperoxide, the second with 0.25 mg of lipoxidase (17 700 units) and the third was untreated. All cultures were incubated for 12 h at 34?C. Cells were harvested by centrifugation (5000 g, 4?C, 15 min) and washed three times with HN buffer. The fatty acids present in the cell pellets were analysed by FAME gas chromatography (Industrial Laboratory, Wheat Ridge, CO). Fatty acids were reported as percentage of total cell mass.

Borrelia burgdorferi strain B31A3 was grown aerobically, microaerobically and anaerobically as described above to a cell density of 5 ? 10⁷ cells ml^?1. The microaerobic culture was split and treated with either 5 mM AAPH [2,2?-azobis(2-methylpropionamidine) dihydrochloride] or 250 mg of lipoxidase for 4 h at 34?C. To measure the amount of MDA, the cells were derivatized with thiobarbituric acid and analysed by HPLC as described by ^{Seljeskog et al. (2006}). After the incubation, all cells were harvested by centrifugation (1000 g, 5 min, 4?C) and washed three times with HN buffer. Each sample was resuspended in 50 ?l of HN, mixed with 150 ?l of 0.1 N perchloric acid, 150 ?l of 40 mM thiobarbituric acid and 35 ?l of 20% SDS, vortexed and heated at 97?C for 60 min. After cooling at ?20?C for 20 min, 300 ?l of methanol and 100 ?l of 20% trichloroacetic acid was added and the samples were mixed vigorously and centrifuged (13 000 g, 6 min). The samples (10 ?l) were then analysed with an Agilent Technologies 1200 series HPLC system using a C18 4.6 ? 150 mm column with mobile phase 72:17:11 (50 mM KPO4, pH 6.8 : methanol : acetonitrile). Absorbance was monitored at 532 nm. Pure MDA standards (0?10 ?M) were prepared in methanol for comparison. As a negative control, E. coli Top10 cells were grown to OD₆₀₀ of 0.4, treated with 0 or 5 mM AAPH for 30 min and MDA measured as above. As a positive control, mouse myeloma SP2/O cells were cultured with HYQ-CCM1 (HyClone) medium at 37?C in a humidified 5% CO₂ atmosphere, treated with 1 mM AAPH at 37?C for 4 h (^{Chotimarkorn et al., 2005}) and MDA measured as above.

Borrelia burgdorferi strain B31A3 cells were grown microaerobically as described above until a cell density of 5 ? 10⁷ cells ml^?1 was obtained. The culture was divided into two equal aliquots and the cells treated with 5 mM AAPH for 4 h at 34?C. After the incubation, all cells were harvested by centrifugation (1000 g, 5 min, 4?C) and washed three times with HN buffer. To visualize Borrelia cells, the cells were stained with PKH26 Red Fluorescent Cell Linker Dye (Sigma Aldrich) following the manufacturer's protocol. To visualize the lipid hydroperoxides in the cell membrane, the cells were counterstained with DPPP (Cayman Chemicals, MI) (^{Okimoto et al., 2000}). Briefly, after cells were stained with the Red Fluorescent dye, 2 ml of rabbit serum was added to stop the reaction and the mixture incubated for 1 min at RT. Next, 4 ml of HN buffer was added, the cells harvested by centrifugation (1000 g, 10 min, RT) and washed three times with HN buffer. The cells were then resuspended in 1 ml of HN buffer, incubated at 34?C for 5 min and 30 ?l of 2.5 mM DPPP was added. The incubation was then continued at 34?C for 5 min in the dark. After incubation, the mixture was centrifuged (1000 g, 10 min, RT) and the cells washed three times with HN buffer. The cells were then resuspended in 100 ?l HN buffer and observed by fluorescence microscopy with excitation wavelengths 551 nm (Red Fluorescent) and 351 nm (DPPP), and emission wavelengths 567 nm (Red Fluorescent) and 380 nm (DPPP). As a positive control, mouse myeloma SP2/O cells were cultured with HYQ-CCM1 (HyClone) medium at 37?C in a humidified 5% CO₂ atmosphere, treated with 1 mM AAPH at 37?C for 4 h (^{Chotimarkorn et al., 2005}) and stained as described above. As a negative control, E. coli Top10 cells were grown to OD₆₀₀ of 0.4, treated with 0 or 5 mM AAPH for 30 min and stained as described above.

Borrelia burgdorferi cells were grown under microaerobic or anaerobic conditions to a cell density of 5 ? 10⁷ cells ml^?1. The microaerobic cultures were split and treated with lipoxidase for 4 h at 34?C. Cells were harvested by centrifugation (5000 g, 10 min, 4?C), washed and resuspended in HBSS (Lonza Group Ltd, Switzerland). The cells were fixed with Karnovsky's phosphate for 5 min at RT adsorbed to Formvar/carbon-coated grids (Ted Pella, Redding, CA) for 5 min and washed three times in H₂O. The grids were stained with 1% ammonium molybdate and allowed to air-dry. Samples were examined using a Hitachi H7500 electron microscope (Hitachi High Technologies America, Pleasanton, CA).