The dynamic proteome of Lyme disease Borrelia
© BioMed Central Ltd 2006
Published: 17 March 2006
Skip to main content
© BioMed Central Ltd 2006
Published: 17 March 2006
The proteome of the spirochete bacterium Borrelia burgdorferi, the tick-borne agent of Lyme disease, has been characterized by two different approaches using mass spectrometry, providing a launching point for future studies on the dramatic changes in protein expression that occur during transmission of the bacterium between ticks and mammals.
We have all experienced it: the 'deer in the headlights' sensation of dumbfounded wonderment and awe when confronted with our first genome sequence. This is a particularly likely response to the genome of the spirochete Borrelia burgdorferi and its relatives - spiral bacteria that are transmitted by deer ticks of the Ixodes ricinus group and that cause the chronic disease Lyme borreliosis in humans and other animals. Although there were previous indications of its unusual characteristics, no one anticipated that the 1.5 megabase genome of B. burgdorferi would contain an odd mixture of approximately 12 linear and 9 circular plasmids, as well as a 0.9 megabase linear chromosome [1, 2]. The plasmids, which range from 5 to 56 kilobases, are present consistently in the strains examined so far and contain genes required for the spirochete's life cycle; these replicons thus could be considered 'mini-chromosomes', although the term plasmids is typically used for simplicity. The 833 predicted plasmid-encoded open reading frames (ORFs) include 454 hypothetical genes, many of which are members of 107 paralogous families of unknown function. The plasmids are also rife with pseudogenes, leading to the conclusion that the Lyme disease spirochete genome is 'in flux': that is, it is actively evolving . The unique properties of the Borrelia proteome are now starting to be revealed with two recent studies of global protein expression. In the first approach, Jacobs et al.  compared the protein profiles of three strains of B. burgdorferi, while in the second, Nowalk et al.  examined the proteins present in the soluble and membrane-associated fractions of the bacterium.
It has been difficult to examine protein expression directly during mammalian or tick infection, as only a small number of spirochetes are present during most phases of infection, limiting the utility of conventional methods. A model system in which B. burgdorferi cultures are 'incubated' within dialysis tubing in the abdomens of rabbits or rats has been used extensively to study adaptation to the mammalian environment . This set-up excludes contact with host cells and extracellular matrix, however, and some aspects of adaptation (for example, recombination in the antigenic variation gene vlsE, which encodes a variable lipoprotein) do not occur under these conditions. A novel approach to the direct study of bacterial protein expression in infected tissues took advantage of the fact that lipoproteins, prominent in B. burgdorferi and other spirochetes, are selectively partitioned to the detergent phase following solubilization in Triton X-114 . By this means, VlsE, OspC, and the decorin-binding adhesin DbpA were found to be expressed at high levels in mouse joints and dermal tissue, and OspC and DbpA, but not VlsE, were found in heart tissue. These results suggest that protein expression varies between tissues; this pattern may be related to tissue tropism of the bacteria.
Global analysis of protein expression in Lyme disease Borrelia is now under way and will be useful not only for examining changes in gene expression, but also in understanding the biological importance of the multiple paralogous gene families and other unique properties of the predicted proteome. In one approach, Jacobs et al.  compared the protein profiles of three strains of B. burgdorferi using trypsin cleavage of whole-organism preparations followed by a two-step liquid chromatography separation of fragments and tandem mass spectrometry (MS/MS). In a second approach, Nowalk et al.  used two-dimensional gel electrophoresis with tryptic digestion and MS analysis of individual spots to look at the proteins in the soluble and particulate fractions of B burgdorferi strain B31.
Jacobs et al.  compared the three strains B31, N40, and JD-1, which were chosen because they represent three genotypic groups that differ in their patterns of pathogenesis in humans and experimentally infected mice. Whole-cell lysates were treated with trypsin, and the resulting complex mixture of fragments was separated by strong cation exchange resin chromatography. Fractions of that separation were then subjected to reverse-phase capillary chromatography coupled with MS/MS analysis of the most abundant fragments in the initial MS separation. Between 13,500 and 17,300 peptides were isolated from each strain, and a total of 6,982 were confidently identified by comparison with protein predictions from the B31 genome sequence. From these data, 522, 498, and 471 proteins were detected in the B31, N40, and JD-1 preparations, respectively, which together represent 665 proteins, or roughly 38% of the predicted proteome.
There are other possible reasons for the uneven detection of certain proteins in the three strains . Only the B31 DNA sequence was used for analysis, although partial sequences of the N40 and JD-1 genomes are now available . Only 0.5% pairwise nucleotide differences were observed on average in alignments of the three DNA sequences , so relatively few false negatives in tryptic peptide identifications should have resulted from differences between the experimental molecular masses and those of the peptides predicted from the B31 sequence. Other reasons for the occurrence of 'unique' proteins could be low abundance or artifactual differences in abundance. The vast majority of the proteins found in only one or two of the strains were identified from fewer than three peptides (and often from only one). In some cases, the 'missing' proteins are required housekeeping proteins such as tRNA synthetases or a flagellar motor protein, and thus must actually be present in all three strains. Therefore, the uneven detection of low-abundance proteins most probably accounts for most (more than 90%) of the proteins detected in only one or two strains (Figure 2).
The other major approach to proteome characterization is two-dimensional gel electrophoresis followed by identification of individual spots on the gel by excision and MS analysis of tryptic peptides. In 1999, Jungblut et al.  analyzed the antigens of B. garinii using this method and the B. burgdorferi sequence; only a limited number of proteins were identified, however. Nowalk et al.  have now examined the proteins present in the soluble and membrane-associated (or, more accurately, the particulate) fractions of B. burgdorferi B31. One of the underlying problems with two-dimensional gel electrophoresis of Borrelia proteins is the high number and abundance of basic proteins, including the major membrane proteins OspA (pI = 8.3) and OspB (pI = 8.6). Basic proteins often migrate off the end of standard isoelectric focusing gels (even those with a broad pH range) or are poorly resolved. Nowalk et al.  addressed this issue by using either non-equilibrium pH gradient electrophoresis (NEPHGE) or immobilized pH gradients. With some modifications, NEPHGE generally out-performed the immobilized gradients in terms of both spot resolution and pH range.
One of the surprising findings of this study was that the glycolytic pathway enzyme enolase was found in nearly equal concentrations in both the soluble and membrane-associated fractions, and aminopeptidase I and the chaperone protein GroEL were present in larger amounts in the membrane fraction than in the soluble fraction . Enolase has, however, been found on the surface of staphylococci and streptococci, where it has been implicated in the adherence of the intact bacteria to plasmin, plasminogen and laminin. Because B. burgdorferi has both an inner and an outer membrane, however, it is not yet known whether the membrane-associated enolase is exposed on the surface. As a chaperone, GroEL may associate with membrane proteins during translocation and folding. Aminopeptidase I has also been shown to localize in the cytoplasm, periplasm, membrane and cell-wall fractions of other bacteria.
Comparison of the MS/MS- and gel-electrophoresis-derived proteomes [3, 4] indicates, as expected, that the electrophoretic approach primarily identified the more abundant protein species in the MS/MS dataset (based on the number of peptides detected per protein). There were, however, many abundant proteins in the MS/MS group that were not present in the electrophoresis analysis, consistent with the more limited sampling in the electrophoretic approach. The ribosomal proteins were the most prominent under-represented group; others included the periplasmic serine protease DO (BB0104), the putative surface-located lipoprotein Lmp1 (BB0210), glycerol-3-phosphate dehydrogenase (BB0243), the flagellar sheath protein FlaA (BB0668), and some of the RNA polymerase subunits (BB0388, BB0389). Conversely, 11 of the 105 proteins detected by electrophoresis were not represented in the MS/MS dataset, and only one or two peptides were detected by MS/MS for several others. Thus, there are some biases in the two approaches, either in sample preparation or the detection methods themselves.
These initial proteome characterizations lay important groundwork for future studies on the expression patterns and localization of B. burgdorferi proteins. Both methods can be adapted to provide quantitative comparisons of protein expression under different conditions, for example, incubation at different temperatures or pH, or conceivably following host adaptation in dialysis membrane chambers implanted in animals. In the whole-cell MS/MS analysis, parallel cultures can be differentially labeled using stable isotopes (14N and 15N) and cysteine affinity tags to quantitate differences in expression patterns , as in the recent analysis of the heat-shock response in the radiation-resistant bacterium Deinococcus radiodurans . In the electrophoretic method, scanning of gels stained with Coomassie blue or other dyes can provide some quantitation. A more elegant approach, however, is difference gel electrophoresis , which comprises the differential labeling of organisms from two different incubation conditions using Ettan fluorescent labeling. The two preparations are labeled separately with derivitized Cy3 and Cy5, and then mixed together before electrophoresis. Fluorescence associated with the polypeptide spots is quantitated, and the corresponding Cy3 and Cy5 signals are used to determine differences in expression patterns between the two conditions.
The initial proteomes of B. burgdorferi and other organisms will serve as the basis for the global analysis of protein expression in response to different environments. In turn, the interfacing of these proteome datasets with other '-omes', such as genome, transcriptome, interactome, and immunoproteome, may begin to reflect the actual complexity of bacterial physiology and pathogenesis.