Assembly of a Sequence-defined, High Density Signature-tagged Mutagenesis Library
transposon vector pGKT 
was used as the basis for the construction of suicide Himar1
transposon vectors for use in B. burgdorferi
. The vectors pGKTSTM1 through pGKTSTM11 were constructed by inserting eleven 7-bp tags into a region between the ColE1 origin and inverted terminal repeat 2 (ITR2) in plasmid pGKT ( in File S1). The infectious, moderately transformable B. burgdorferi
strain 5A18NP1 was transformed with these plasmids to create an STM library in an infectious background. Between 100 and 300 clones were obtained with each transformation of 5×109
organisms with 10 µg DNA, yielding an electroporation frequency of 2 x10−8
. A library of 6,625 STM mutants has been created, and the insertion sites have been determined for 4,479 of these mutants (; Spreadsheet S1).
Maps of the STM transposon insertion points in representative plasmids of B. burgdorferi B31 5A18NP1.
Distribution of transposon insertions among the B. burgdorferi replicons.
Mapping of the transposon insertion sites permitted the assessment of their distribution. Insertion sites were distributed on all 20 of the linear and circular replicons present in the parent strain, 5A18NP1 (, Spreadsheet S1). However, the number of insertions per kb of DNA varied widely from 0.91 in the circular plasmid cp32-1 to 23.97 in cp9; the average value for the genome was 3.11. The chromosome had a relatively low value of 1.63 insertions per kb DNA, consistent with a high proportion of essential genes in that replicon. The library included 3,865 unique insertion sites, i.e.
614 clones had insertion sites identical to those of other transposon mutants. These included 124 sibling clones (isolated from the same transformation); most of the siblings were obtained in early transformations which utilized longer incubation periods (e.g.
48 h) in liquid medium prior to plating. The remaining 490 clones with shared insertion sites (n
230) were independently derived from different transformations. No indication of sequence bias beyond the recognition of the well-characterized 5′-TA-3′ insertion site was observed. Anomalous (non-TA) insertion points were present in 55 (1.2%) of the clones, with sequences of TT and AT (12 clones each), AA (9), GA, TC, and TG (5 each), AG, CT, and GT (2 each), and AC (1).
Analysis was further extended to the number of genes with transposon insertions (). Overall, 790 (45.5%) of the predicted protein-encoding genes in the genome were disrupted in the STM transposon library. This value was approaching saturation toward the end of the transposon mutant selection process ( in File S1); for example, in the final transformation only 11 of 260 clones examined (4.2%) had insertions in genes that had not been disrupted previously. As expected, a lower proportion of genes were disrupted in the chromosome, with only 287 of 857 (34%) of predicted protein-encoding genes having insertions. In contrast, over 90% of the genes were disrupted in cp9, cp26, and cp32-6, indicating a small number of the encoded products of these plasmids are needed for in vitro growth of B. burgdorferi or plasmid maintenance. Each of the two 23S rRNA-encoding loci (22 mutants total) and one of two 5S rRNA genes (1 mutant) had insertions, but the single 16S rRNA locus and the many tRNA loci were not disrupted. A much lower proportion of chromosome intergenic regions had insertions than those in the plasmids (14% as compared to 38%), perhaps related to the smaller average size of chromosomal intergenic regions and the higher density of promoters and other regulatory elements that could affect borrelial survival and growth. Several of the clones with transposon insertions in cp32 plasmids could not be assigned unambiguously to a single plasmid; in these cases, the sequence obtained from circularized genome fragments containing the transposon (typically ~500 bp) was identical to homologous regions of two or more plasmids in the highly conserved cp32 sequences. For those genes with insertions, there was a median of 4.7 insertions per kb DNA and 6.4 insertions per gene; ten genes had more than 30 transposon insertions. There were no overall differences in the number of insertions per kb DNA in the genes of the chromosome, linear plasmids, or circular plasmids ( in File S1).
Proportion of predicted protein genes and intergenic regions with transposon insertions in the B. burgdorferi replicons.
Luminex-based procedure for detection of STM clones in tissues and in cultures.
We used the insertion ratio (distance in bp from the insertion site to the 5′ end of the gene divided by the total gene length) to provide a convenient measure of the location of the transposon insertion within the gene. The insertion ratio was also determined for intergenic regions, but in this case was simply oriented relative to the 5′ end of the replicon. There were 234 genes with only a single transposon disruption: 87 in the chromosome and 147 in the plasmids. Plotting the percentage of insertions at different insertion ratios ( in File S1) indicated that insertions in the last 10% of the gene length were recovered at a higher rate (33.33% of the chromosomal genes and 15.65% of the plasmid genes). This result is consistent with improved preservation of functional gene products when insertions occur late in the gene 
. Examples of probable essential genes in which single insertions occurred in the final 10% of the reading frame include those encoding the cell division protein FtsH, replicative DNA helicase DnaB, ribosomal protein S21, glucose-6-phosphate isomerase (Pgi), a C-terminal protease (Ctp), two pantothenate transport and utilization proteins (PanF and Dfp), two enzymes involved in peptidoglycan synthesis (MurD and MurG), and the cp26-encoded telomere resolvase ResT. Other genes with only late insertions may also be essential ( and in File S1).
There was also a subset of genes that had an unexpectedly high number of transposon insertions, including 105 genes >100 bp in length that had more than 10 insertions per kb DNA. This set comprises the genes for histidine kinase 1 (bb420; 49 unique insertion sites, 10.93 per kb DNA), the cp26-encoded oligopeptide transporter periplasmic binding protein OppAIV (bbb16, 29 insertions, 18.20 per kb DNA), conserved integral membrane protein bb0017 (26 insertions, 27.08 per kb DNA), and cp9-encoded conserved hypothetical protein bbc12 (31 insertions, 23.98 per kb DNA).
The ordered transposon library described here is not saturating in most replicons, so essential genes cannot be identified simply as those without insertions. However, functional groups of genes that are either essential or nonessential for 1) in vitro growth in BSK-II medium or 2) plasmid replication can be discerned (; in File S1). The functional groups that were found to be nonessential for in vitro growth included some regulatory factors, genes of the phosphoenolpyruvate phosphotransferase system (PEP-PTS), additional transport systems, genes involved in glycerol transport and incorporation, DNA repair and recombination genes, chemotaxis and motility genes, and several known virulence determinants and candidates. As expected, only a few genes involved in DNA replication, transcription and translation, peptidoglycan biosynthesis and intermediate metabolic pathways had transposon insertions, and these insertions were commonly at the 3′ ends of the genes or were in genes that encoded accessory functions. A detailed listing of the genes with assigned functions and their transposon disruption patterns is provided in in File S1.
Summary of essential gene group candidates, and apparent infectivity phenotypes of nonessential genes (based on STM results).
Some of the plasmids had a high number of insertions per kb of DNA, so the distribution of insertions in these cases may more reliably identify candidate essential genes. The insertion maps of plasmids of interest are shown in ; the maps for all plasmids are provided in the Supporting Information ( in File S1). The circular plasmid cp26 will be used as an example of the identification of candidate essential genes here. Of the 29 annotated genes in cp26, 26 genes (90%) were disrupted in the transposon mutagenesis library (). No insertions were observed in 3 ORFs: bbb10
, and bbb26
. BBB10 and BBB11 are predicted proteins belonging to the Paralogous Families (PFs) 62 and 50, and are likely involved in plasmid replication or partitioning. bbb10, bbb11
, and bbb13
were shown recently to support the maintenance of a partition-defective MiniF plasmid in a heterologous E. coli
plasmid system 
. Jewett et al. 
determined that BBB26 and BBB27 are highly homologous, membrane-associated periplasmic proteins, and that one or the other gene had to be intact to promote bacterial viability. In our mapping of transposon insertion sites in cp26, bbb27
had three independent insertions (two at the same location), whereas no insertions were identified in bbb26
. In the STM mutant T08TC298, the gene encoding the telomere resolvase ResT (bbb03
) was disrupted at the 3′ end (insertion ratio
0.94). ResT is critical for the replication of the linear chromosome and linear plasmids 
. We did not notice an obvious in vitro
growth defect in T08TC298; indeed, this resT
mutant was both viable and infectious. It is likely that T08TC298 expresses a truncated but functional ResT protein.
Procedure for high-throughput STM infectivity analysis of B. burgdorferi transposon mutants in this study.
Plasmid Profiles of STM Transposon Mutants
A total of 4464 STM mutants have been examined for plasmid content to date 
. In this PCR-based Luminex analysis, the fluorescence output obtained for each plasmid was evaluated as positive, intermediate, or negative (Spreadsheet S1) 
. The intermediate category was included because the median fluorescent intensity (MFI) values obtained for some plasmids could not be clearly subdivided into positive and negative groups; the clones that have intermediate MFI values may contain a mixture of cells that have retained or lost the plasmid in question. For the 4464 STM clones examined, 1799 (40.3%) retained all plasmids. Plasmid loss occurred most commonly with lp5 (1916 clones, 42.9%), cp9 (602, 13.5%), lp21 (301, 6.7%), lp28-1 (291, 6.5%), and cp32-6 (135, 3.0%). One hundred three mutants (1.6%) had negative or intermediate plasmid analysis results for lp36. Only one clone had lost lp25 due to consistent inclusion of both kanamycin and gentamycin in the culture medium for the transposon mutants, resulting in selection for the kanamycin resistance cassette present in lp25 of the parent clone 5A18NP1; the one lp25-negative clone may be a spontaneous kanR
mutant. Loss of other plasmids occurred in less than 2% of examined STM mutants; cp26 and lp54 were consistently present, and cp32-1, cp32-8, cp32-9, and lp28-3 were absent from 0.3% or less of the clones. Prior studies have shown that lp25, lp28-1, and lp36 are required for full infectivity of B. burgdorferi
, so clones lacking these plasmids were excluded from infectivity analysis. lp5 and lp21 are not essential for infection of mice 
. The potential role of cp32-6 (deficient in 3% of transposon mutants) in B. burgdorferi
infectivity of mice or ticks is not known.
Screening of STM Mutants for Infectivity in Mice
Initial efforts to screen the STM mutants for infectivity utilized PCR and gel electrophoresis (data not shown). However, examination of a large number of mutants from input and output pools using this method is time-consuming and laborious, and generally provides only a subjective measure of infectivity. To overcome this barrier and facilitate the high-throughput detection of B. burgdorferi STM mutants in mammalian hosts, we developed a Luminex-based multiplex PCR procedure for semi-quantitative detection of STM mutants (). The Luminex protocol was performed in parallel using DNA extracted directly from the mouse tissue, and DNA extracted from B. burgdorferi cultured from the same tissue specimen. Because each tissue specimen required only one well in a 96-well plate, it was also possible to evaluate multiple tissue sites (bladder, ear, heart, tibiotarsal joint, and inoculation site) and sets of three mice per time point. In this manner, we were able to assess the occurrence of localized and disseminated infection at 2 and 4 weeks post inoculation in five tissues from 3 mice with two DNA preparation methods (). Thus, in a typical analysis, there were 60 MFI values collected for each clone inoculated.
In this article, we report the STM analysis results from 434 STM transposon mutants, representing 28,371 individual MFI measurements (Spreadsheet S2). To provide an example of the results obtained, the evaluation of mutants in the plasmid cp26 is provided in and in File S1. Negative and positive controls consisted of mutants in pncA and the chromosomal conserved hypothetical protein gene bb0051, respectively; the results obtained for these controls in three infectivity experiments are shown at the bottom of each panel. We found that the cumulative (or average) MFI value provided the most reliable indicator of the infectivity of a given mutant (see Materials and Methods). In general, there was good correlation between the cumulative MFI values obtained with the tissue DNA extraction and culture DNA extraction methods (, and ) as well as between the 2 week and 4 week post inoculation results (compare , and ). Additional information regarding the clones tested and tabular results using a positive/negative cutoff of 100 MFI for each tissue are provided in in File S1.
Luminex STM infectivity analysis of B. burgdorferi clones with transposon insertions in 26 of the 29 cp26 genes (bbb01-bbb29).
Although infectivity patterns for individual clones could be discerned easily by examining the set of 60 data points, there was considerable variability among the individual MFI values obtained (). Even in clones with high infectivity phenotypes, 8% of individual samples processed by culturing the organisms from the tissue specimens had MFI values less than 100 ( in File S1); the proportion was higher (14% to 17%) in samples prepared by direct extraction of tissue, presumably due to the lack of ‘amplification’ resulting from multiplication of the organisms in culture. As expected, the percentage of MFI<100 samples increased in intermediate and low infectivity clones ( in File S1). To determine whether the Luminex procedure yielded reproducible values, we repeated the Luminex assay 6 times with the same inoculation site specimen from STM set 51. The results indicated that the MFI values were well clustered for each of the 11 clones in the set, and reinforced the occurrence of low, intermediate and high MFI values ( in File S1). It was possible that the variation observed was due to uneven distribution of organisms, in keeping with the paucibacillary nature of B. burgdorferi infection. Indeed, there was a low quantitative correlation between the culture- and direct DNA extraction-derived MFI values obtained from the same tissue, which utilized neighboring tissue samples. To examine this possibility, we subdivided STM set 51 inoculation site specimens from 6 mice (three each at 2 weeks and 4 weeks) into 9 neighboring samples. The DNA was extracted from each piece and then tested using the Luminex assay. The data in in File S1 indicates that there is considerable variability in the MFI values for high infectivity clones in each of the 6 animals. The low MFI values obtained for the pncA mutant T01P01A01 were all within the range obtained with ‘no DNA’ controls. These results suggest that sampling error due to the uneven distribution of organisms accounts for much of the MFI value variation observed for individual clones.
A simple heat map representation of the mouse infectivity of transposon mutants in cp26 genes.
Transposon mutations in Protein Family (PF) genes postulated to be involved in plasmid maintenance.
A summary of the gene disruption and infectivity results obtained is provided in . The essentiality of certain gene function groups required for in vitro culture is based on the initial isolation of clones, as described above. In terms of mouse infectivity, a given clone was considered to be defective in infectivity if less than 20% of specimens yielded MFI values above 100. The requirement of several genes in a functional group provided strong evidence that the associated activities are essential for mouse infection under the conditions tested.
Roles of Plasmid-encoded Genes in Mouse Infectivity
Plasmid-encoded genes play a vital role in the enzootic cycle of Lyme Borrelia. As shown in and in File S1, the results obtained with mutants in cp26 provide evidence that a number of genes encoded in this plasmid are needed for full infectivity in mice.
The STM analysis provides a large amount of data that is challenging to interpret en masse. To provide a clearer view of genes required for infectivity in mice, a simple heat map approach was developed that provides a visual overview of the infectivity of mutants in these gene groups. We analyzed groups of mutants based on the genes associated with certain plasmids or with different functional groups. An example of a heat map for genes associated with cp26 is shown in ; additional maps for lp25, lp28-1, lp36, and lp54 are shown in Fig. 7 in File S1. In the upper portion of each panel, the coloration is based on the percentage of samples with an MFI greater than 100 (percent positive); very similar patterns were obtained using the mean MFI, i.e. the mean of the 12 to 15 MFI values for each time point and DNA preparation method (lower portion of each panel).
Overall, a surprisingly high number of the plasmid genes appear to be required for full mouse infectivity. Using the conservative cutoff of ≤20% of tissue sites positive and an overall mean MFI ≤100, transposon insertions in the following cp26 genes appeared to engender defective infectivity: bbb02
(PF32 plasmid maintenance protein), bbb13
(PF49 plasmid maintenance protein), bbb14
, GMP synthase), bbb19
), and bbb22
, purine permease G1) ( and , Spreadsheet S2). In addition, mutations in other genes, including bbb07
(a putative outer surface protein), bbb17
, inosine-5-monophosphate dehydrogenase), and bbb23
, purine permease G2) yielded reduced MFI values in most tissues. Thus 12 of the 26 cp26 genes examined (46%) are virulence determinant candidates based on this STM analysis; 5 of these genes had been identified as virulence determinants previously 
. Similarly, a high proportion of potential virulence determinants were also identified in the other plasmids examined to date (Fig. 7 in File S1, Spreadsheet S2).
Functional Groups in Infectivity
Transposon mutations in genes associated with DNA recombination and repair, chemotaxis, motility, the PEP-PTS, other transport systems, plasmid maintenance, gene regulation, sRNA genes, carbohydrate, amino acid, and nucleic acid and other important metabolic pathways, predicted lipoproteins, conserved hypothetical proteins, hypothetical proteins (in the chromosome and some plasmids), protease, and complement regulator-acquiring surface proteins (CRASPs) of B. burgdorferi were selected for STM screening. A representative number of mutants in intergenic regions were also tested for infectivity to evaluate potential polar effects, small RNA genes, and regulatory regions. The results for some of these functional groups are depicted in the heat map format in Fig. 8 in File S1. Mutations in 10 of 14 chemotaxis genes resulted in an severe loss of infectivity (Fig. 8A in File S1). Mutants in cheR-2 and cheB-1 appeared to have reduced infectivity, whereas a mutant in cheW-3 exhibited near wild-type infectivity. In B. burgdorferi B31, these genes have a total of 2, 2, and 3 paralogs, respectively. The results obtained for two methyl-accepting chemotaxis protein 5 (mcp5) mutants were quite different, so the findings for this gene were inconclusive.
Similarly, mutations in 4 of 7 genes associated with flagellar structure or assembly showed reduced infectivity, with only mutations in fliD and flgB yielding a high infectivity phenotype (Fig. 8B in File S1). Although most STM mutants did not have obvious in vitro growth defects or morphologic changes, mutations in 7 motility and chemotaxis genes (fliH, fliI, flbA, flaA, cheA-2, cheB-2, and cheR-2) exhibited slow growth and elongated, string-like or rod-shaped morphology. Additionally, 7 mutants (flaA, flgI, fliG-1, fliW-1, cheA-2, cheR-2, and mcp-5) had reduced motility, and string-like mutants (flbA, fliH, and fliI) were nearly non-motile, ‘trembling’ in a few sites of the cell. Overall, the results obtained were consistent with a requirement for chemotaxis and motility activities for mouse infection.
The results obtained with representative mutations in PEP-PTS genes are shown in Fig. 8C in File S1. ptsG encodes a glucose-specific IIBC component and appears to be required for mouse infection. In contrast, mutation of ptsH-1 (bb0448, one of two genes encoding phosphocarrier protein homologs) had little apparent effect on infectivity. No mutants in ptsH-2 (bb0557) or genes encoding other ‘core’ PEP-PTS components (e.g. crr, ptsI, and ptsP) were isolated. Mutations in the glucose and maltose-specific IIABC component homolog genes malX-1 (bb0116) and malX-2 (bbb29, located in cp26) yielded intermediate to low infectivity results. Mutants in the cp26-encoded chitobiose IIC and IIA components chbC and chbA were infectious, although the chbA mutant may have reduced infectivity. Two different mutants in the neighboring chitobiose IIB component chbB had low mouse infectivity in this system. Low infectivity was also obtained with a mutant in fruA-1 (bb0408, fructose-specific IIABC component). Inconsistent results were obtained with two mutants in fruA-2 (bb0629), precluding interpretation in the absence of additional information.
In terms of other transporter systems (Fig. 8D in File S1), mutation of the genes encoding lactose permease (lctP
), Na+/H+ antiporter proteins (nhaC-1
), the ATP-binding protein of the methylgalactose transporter (mglA
), and another ABC transporter ATP-binding protein (bb0573
) appeared to result in loss of infectivity. Mutation at the very end of proX
, insertion ratio
0.99) resulted in a low infectivity phenotype, suggesting that the C-terminus is required for function; the downstream conserved hypothetical protein (CHP) gene bb0143
is oriented in the opposite direction, ruling out polar effects. While Borrelia
efflux system genes besC
appear to be required for full infectivity, mutation of the downstream gene besB
had no apparent effect on infectivity. Intermediate infectivity results were generally obtained for mutations in the oligopeptide periplasmic binding protein genes oppA-1
(Fig. 8D in File S1). No transposon insertions were observed in the 6 other genes for this ABC transport system; thus it is likely that the oligopeptide transport function is required for in vitro
growth, but that the periplasmic binding proteins have the ability to partially complement one another, permitting in vitro
culture of clones having a mutation in one oppA
Analysis of Genes Implicated in Plasmid Maintenance
Genes potentially required for plasmid replication or maintenance were investigated based on the mapping of transposon insertions in plasmid genes previously postulated to be involved in plasmid replication, partitioning, or retention 
. Genes representing five Paralogous Families (PFs) are present in both the linear and circular plasmids: PF32, PF49, PF50, PF57, and PF62 (). These PF genes are not consistently present in every plasmid, but all plasmids have either a PF57 gene or a PF62 gene. There are also many PF pseudogenes (gene fragments) in the plasmids; these are presumably nonfunctional and were not considered in this analysis. Transposon insertions were identified in 20 of the 72 intact PF genes, but in two cases (cp9 PF57 and lp25 PF49) the genes had insertions only within the last four percent of the reading frame and were likely to be functional (). An apparently functional copy of either PF57 or PF62 was retained in each plasmid for all transposon mutants recovered, consistent with the proposed requirement for one or the other of the corresponding protein products for the initiation of plasmid-specific replication 
. Transposon insertions in PF32, PF49, and PF50 genes were obtained, but the pattern was not consistent in the different plasmids. The occurrence of multiple transposon insertions in the lp17 PF32 gene bbd21
is in agreement with a prior study showing that bbd21
is not required for lp17 replication and maintenance 
We were surprised to find that very low MFI signals were recovered from infected mice in nearly all of the mutants in these PF genes examined to date (Fig. 8E in File S1). The sole exception to this trend was in a mutant of the lp28-1 PF32 gene bbf13, which exhibited full infectivity. However, in this case, bbf13 is a redundant copy of another intact lp28-1 encoded PF32 gene, bbf24 (, in File S1).
Tick Inoculation Studies
To date, attempts to implement the STM procedure in tick transmission studies have not been successful due to insufficient recovery of organisms from ticks infected by capillary feeding with mixtures of in vitro
cultured B. burgdorferi
, as well as from mice that were exposed to these ticks. To determine whether results obtained by needle inoculation/STM analysis were comparable to those obtained with tick transmission, we inoculated I. scapularis
nymphs with individual transposon mutant clones by capillary feeding and then fed the ticks on C3H/HeN mice, as previously described 
. Results obtained with transposon mutants in genes involved in flagellar assembly and chemotaxis are shown in . For each of the clones tested, a high proportion of ticks contained viable B. burgdorferi
after capillary feeding with suspensions of in vitro
cultured organisms, as determined by culture of tick tissue and detection of spirochetes in tick smears by immunofluorescence (“unfed ticks”). With the exception of the methyl-accepting chemotaxis protein-5 (mcp5
) mutant MG064, all of the mutants proliferated after the infected ticks were fed on naïve C3H/HeN mice. For MG064, none of 21 ticks were positive for B. burgdorferi
by culture or immunofluorescence after the blood meal was taken. Cultures from multiple tissues (ear, heart, joint, bladder) and seroconversion for antibodies against the VlsE C6 peptide indicated that most of the motility and chemotaxis mutants had profound defects in infectivity, as compared to consistent culture and serologic positivity in the B31 5A18NP1 parental strain controls (). For comparison, the corresponding STM analysis results obtained by needle inoculation with these mutants (and one additional mcp5
mutant) are provided on the right side of ; the MG064 mcp5
transposon mutant was not tested by STM analysis because it was created prior to the construction of the STM library 
and therefore lacks a signature tag. In the STM studies, each of the motility and chemotaxis mutants in this group exhibited reduced infectivity (right side of ); although the mutants in flaA, cheR-2,
genes had several sites with MFI values >100, the overall mean MFI values for these clones were low relative to the positive control (bb0051
mutant T02P01A01, shown at the bottom of the table).
Effects of transposon mutations in genes involved in flagellar assembly and chemotaxis on I. scapularis nymph transmission of B. burgdorferi clones to C3H/HeN mice.a