￼Genetic Manipulation of Borrelia.

Genetic studies in Borrelia require special consideration of the highly segmented genome, complex growth requirements and evolutionary distance of spirochetes from other genetically tractable bacteria. Despite these challenges, a robust molecular genetic toolbox has been constructed to investigate the biology and pathogenic potential of these important human pathogens. In this review we summarize the tools and techniques that are currently available for the genetic manipulation of Borrelia, including the relapsing fever spirochetes, viewing them in the context of their utility and shortcomings. Our primary objective is to help researchers discern what is feasible and what is not practical when thinking about potential genetic experiments in Borrelia. We have summarized published methods and highlighted their critical elements, but we are not providing detailed protocols. Although many advances have been made since B. burgdorferi was first transformed over 25 years ago, some standard genetic tools remain elusive for Borrelia. We mention these limitations and why they persist, if known. We hope to encourage investigators to explore what might be possible, in addition to optimizing what currently can be achieved, through genetic manipulation of Borrelia.


Introduction
Spirochetes in the genus Borrelia have been known to cause human disease for well over a century, but in-depth investigations of these pathogenic bacteria awaited many developments, including the advent of molecular genetics. Since transformation of Borrelia burgdorferi was first described in 1994 (Samuels et al., 1994b;Samuels, 1995), a molecular genetic approach has been applied with growing success to study the biology and pathogenic potential of these tick-borne spirochetes Cabello et al., 2001;Rosa et al., 2005;Battisti et al., 2008;Fine et al., 2011;Brisson et al., 2012;Raffel et al., 2014;Krishnavajhala et al., 2017;Raffel et al., 2018;Samuels et al., 2018). Both in vitro and in vivo analyses have thus been conducted to identify and describe the basic functions of many Borrelia genes and gene products and their contributions to the relationship of these important pathogens with their tick vectors and mammalian hosts. Direct applications of genetic tools developed for Gram-positive and Gram-negative bacteria have encountered log phase cultures, which have a greater number of cells, yield the most transformants (Samuels, 1995;Tilly et al., 2000). Briefly, cells are harvested sterilely by centrifugation, washed several times, and finally resuspended in cold electroporation solution as an approximately 100-fold concentrated cell slurry. Some B. burgdorferi strains aggregate during the washes and cannot be resuspended in electroporation solution, which prevents or severely limits transformation ; this undesirable trait is intensified when bacteria are grown to higher cell density. The presence of outer surface protein B (OspB) contributes to this clumping phenotype (Tilly et al., 2016); hence most strains selected for use in genetic studies are spontaneous mutants that lack OspB (Elias et al., 2002). Aliquots of competent cells can be transformed by electroporation immediately or they can be stored at −80°C, which decreases transformation efficiency by about 50% (Samuels, 1995;Samuels and Garon, 1997;Samuels et al., 2018).

Electrotransformation
Electroporation conditions for transformation of B. burgdorferi are similar to those used for other bacteria (Samuels et al., 1994a;Samuels, 1995;Samuels et al., 2018). A high-voltage electric pulse generates pores in the cellular membranes through which DNA can enter the cytoplasm (Shigekawa and Dower, 1988;Trevors et al., 1992;Nickoloff, 1995). Transmission electron microscopy of electroporated B. burgdorferi reveals darkly stained regions on the membrane along the length of the cell that are thought to be disruptions in the surface (Samuels and Garon, 1997). These putative pores are no longer visible after a 30 minute incubation in culture medium.
As mentioned above, a significant improvement to the original protocol was increasing by approximately 1000-fold the amount of DNA used per electroporation, from the nanogram levels typically used with other bacteria, to the microgram quantities needed to routinely transform B. burgdorferi Tilly et al., 2000;Hübner et al., 2001;Eggers et al., 2002;Elias et al., 2002). However, even when successful, relatively few B. burgdorferi transformants are typically recovered per electroporation, particularly when working with infectious clones. Aside from R/M systems, which can be eliminated, the basis for the extremely low transformation frequency and efficiency of B. burgdorferi remains a mystery. Perhaps very little of the transforming DNA actually gets into the cell, the DNA is rapidly targeted for degradation once it enters, or some other feature of the spirochete limits stable retention of transforming DNA. Whatever the basis, this unfortunate characteristic limits the routine construction of saturated libraries that are needed to apply powerful forward genetic screens in Borrelia (Lin et al., 2012;Ellis et al., 2013).

Transformation substrate
In addition to the amount of DNA, its source and form have an effect on transformation efficiency. Allelic exchange transformation was first demonstrated by introducing sitedirected mutations that conferred antibiotic resistance, using PCR-generated linear DNA substrates (Samuels et al., 1994a). Short oligonucleotides also can be used as transformation substrates (Samuels and Garon, 1997), although this has limited utility because the sitedirected mutation encoded by the oligonucleotide must confer a selectable advantage (currently limited to antibiotic resistance). A circular suicide plasmid was previously shown to recombine into the genome at the same frequency as linear DNA (Stevenson et al., 1998); hence linearized plasmid DNA or PCR-amplified DNA fragments are typically used when allelic exchange is the desired outcome. This prevents integration of suicide plasmids and generation of merodiploids, which are recovered more frequently when working with Borrelia clones lacking loci encoding restriction-modification (R/M) systems. As described in a later section, circular shuttle vectors have been constructed that replicate autonomously in B. burgdorferi and E. coli Eggers et al., 2002;Stewart et al., 2003;Byram et al., 2004). In addition, linear shuttle vectors have been constructed using replication functions from both native circular and linear plasmids (Chaconas et al., 2001;Beaurepaire and Chaconas, 2007). Transformation of autonomously replicating shuttle vectors into strains lacking R/M systems (see below) is generally more efficient than that of allelic exchange constructs (Elias et al., 2003), presumably because the latter require the additional step of recombination. Remarkably, intact native plasmids up to 36 kbp in length have been successfully transformed into B. burgdorferi. Jewett et al., 2007b).
The requirement for microgram quantities of substrate DNA per Borrelia transformation can present a problem in itself. Plasmids containing Borrelia sequences or Borrelia codonoptimized genes can limit E. coli growth, and rapid propagation of E. coli to high cell densities (as typically done for large scale plasmid preparations) can select for mutations in sequences conferring a growth phenotype. Propagating E. coli carrying recombinant plasmids at a lower temperature, such as 30°C, (Takacs et al., 2018), lessens the selection for mutations that confer a growth advantage, and plasmid DNA preparations should be sequenced to confirm the integrity of the insert.

DNA modification
One of the identified barriers to efficient transformation of B. burgdorferi is the presence of plasmid-encoded restriction-modification (R/M) systems (Lawrenz et al., 2002;Kawabata et al., 2004). Transformation of a set of closely related B. burgdorferi strain B31 clones with distinct plasmid complements demonstrated that loss of two linear plasmids, lp25 and lp56, correlated with higher transformation efficiency of a shuttle vector (Lawrenz et al., 2002). These two plasmids carry genes (bbe02 and bbq67) that are predicted to encode type IV R/M enzymes. Disruption of bbe02 increased shuttle vector transformation efficiency approximately 40-fold (Kawabata et al., 2004). Subsequent studies confirmed these initial observations and demonstrated that bbe02 and bbq67 of strain B31 encode distinct R/M enzymes that methylate endogenous DNA and restrict foreign DNA lacking sequencespecific modifications (Chen et al., 2008;Rego et al., 2011). The motifs modified by bbe02 and bbq67 were recently inferred by SMRT sequence analysis of genomic DNA from B31 clones that differed in R/M gene content (Casselli et al., 2018). This information should facilitate the design of optimized constructs for more efficient transformation of B31 clones containing intact bbe02 and bbq67 loci. Analysis of genomic sequences in the database indicates that some B. burgdorferi strains contain additional R/M loci that likely recognize and modify novel sequence motifs . However, a third plasmid-borne R/M locus in strain B31, bbh09 on lp28-3, does not negatively impact shuttle vector transformation (Lawrenz et al., 2002;Kawabata et al., 2004), suggesting that the presence of R/M loci does not necessarily indicate a barrier to transformation in these strains.
In addition to the bbe02 R/M locus, lp25 also carries genes required for mouse and tick infectivity (Purser and Norris, 2000;Labandeira-Rey and Skare, 2001;Purser et al., 2003;Grimm et al., 2005;Revel et al., 2005;Strother et al., 2005;Gilmore et al., 2014). The linkage of these disparate functions on lp25 is the primary reason that infectious B31 derivatives (containing lp25) are difficult to genetically transform. Transformation efficiency into low-passage infectious strains can be increased by first transforming the DNA into a strain carrying lp56 but lacking lp25 (Jewett et al., 2009). The resulting genomic DNA purified from Borrelia is expected to be partially modified and thus presents less of an obstacle when introduced in subsequent transformations into strains containing lp25. The transformation barrier imposed by putative restriction enzymes on lp56 and lp25 applies primarily to shuttle vector transformations rather than gene inactivation by allelic exchange (Elias et al., 2002;Lawrenz et al., 2002;Hübner et al., 2003;Kawabata et al., 2004;. This likely reflects the requirement for intact circular DNA for autonomous plasmid replication, whereas linearized DNA remains an effective template for allelic exchange. Consequently, trans-complementation on a shuttle vector presents a greater challenge than gene inactivation in infectious clones retaining lp25. Fortunately, spirochetes do not require bbe02 or bbq67 for growth in vitro or at any stage of the experimental mouse/tick infectious cycle. Therefore, fully infectious B31 variants have been constructed in which the bbe02 locus on lp25 has been deleted (Lawrenz et al., 2002;Kawabata et al., 2004;Rego et al., 2011). These modified strains typically lack the (nonessential) lp56 plasmid that carries bbq67, thereby further enhancing their suitability for genetic studies. As an added bonus, the presence of an antibiotic resistance cassette in bbe02 provides selection for lp25 during in vitro growth and genetic manipulation. The bbe02 locus on lp25 has been chosen as an optimal site for insertion of reporter genes, components of inducible systems, and trans-complementation (Gilbert et al., 2007;Chan et al., 2015;Kasumba et al., 2015), as targeted insertion at this site guarantees stable retention of introduced DNA at normal copy number throughout the mouse/tick infectious cycle. This is discussed more fully in subsequent Complementation and Reporter gene sections.

Alternative methods for transformation
In addition to electroporation, B. burgdorferi has been successfully transformed using a simple protocol that entails resuspension of bacteria and DNA together in a solution of PEG and DMSO (Chung et al., 1989;Elias et al., 2002). However, this method has not been widely used or optimized, and its general utility remains unexplored. Likewise, Eggers and colleagues have demonstrated that the endogenous prophage of Borrelia can mediate relatively efficient horizontal gene transfer between co-cultured spirochetes (Eggers et al., 2016), but phage transduction has not been developed as a standard genetic tool for B. burgdorferi. A substantial advantage of both protocols over electroporation is that they require fewer bacteria at the outset of an experiment because neither technique results in significant cell death, thereby greatly reducing (~100-fold) the initial culture volume needed per transformation.

Selectable markers
Antibiotic resistance markers are one of the most effective selective tools used to single out rare bacterial variants containing new genetic properties, as well as to select for and screen mutants deficient in the expression of specific genes (Pich et al., 2006). B. burgdorferi is exquisitely susceptible to many antibiotics (Hunfeld et al., 2002;Ruzic-Sabljic et al., 2005;Hunfeld and Brade, 2006), but the choice of antibiotic resistance genes as genetic markers is somewhat limited by biosafety considerations: markers encoding resistance to ampicillin and tetracycline are prohibited for use in B. burgdorferi because these antibiotics are clinically useful in the treatment of Lyme disease (Steere, 2001;Wormser, 2006). The selectable markers currently available for genetic studies in B. burgdorferi and their relevant features are listed in Table 1.

Endogenous genes
To date, only antibiotic resistance has been used to select Borrelia transformants (Samuels et al., 1994a;Bono et al., 2000;Sartakova et al., 2000;Elias et al., 2003;Frank et al., 2003). An allele of the endogenous B. burgdorferi gyrB gene (Samuels et al., 1994b), which encodes the B subunit of DNA gyrase, was the first selectable marker used in Borrelia (Samuels, 2006). However, the gyrB301 marker, which confers resistance to the antibiotic coumermycin A 1 , has limited utility for several reasons, including pleiotropic effects due to relaxed DNA supercoiling (Alverson and Samuels, 2002;Alverson et al., 2003). Nonetheless, the use of gyrB301 was seminal in the development of transformation and genetic manipulation of B. burgdorferi (Samuels et al., 1994a;Rosa et al., 1996).

Hybrid genes
A major breakthrough in genetic analysis of B. burgdorferi was the construction of hybrid genes using native B. burgdorferi promoters fused to foreign open reading frames. This was first used to assay transcription by fusing various borrelial promoters to cat, encoding chloramphenicol acetyl transferase (Sohaskey et al., 1997). Bono et al. (Bono et al., 2000) constructed the first efficient selectable marker for Borrelia by fusing the strong and constitutive flgB and flaB promoters of B. burgdorferi to the aphI gene from Tn903, which confers resistance to the aminoglycoside kanamycin; the resulting cassette confers kanamycin resistance in both E. coli and Borrelia. Several other chimeric genes were constructed in a similar fashion, notably selectable markers conferring resistance to gentamicin with the aacC1 gene from Tn1696 (Elias et al., 2003) and resistance to spectinomycin and streptomycin with the aadA gene from Shigella flexneri plasmid R100 (Frank et al., 2003). Most recently, this approach was applied to the hph gene of E. coli and the bsd gene of Aspergillis terreus, yielding codon-optimized selectable markers fused to the flgB promoter that confer resistance to hygromycin and blasticidin, respectively (Takacs et al., 2018). Although seemingly straightforward, not all antibiotic resistance cassettes constructed in this fashion function as selectable markers in B. burgdorferi. The cat and pac genes, fused to the flgB promoter, do not confer resistance to chloramphenicol and puromycin, respectively (Elias et al., 2003). Hybrid cassettes conferring resistance to various antibiotics and used as selectable markers in B. burgdorferi are summarized in Table 1.

Selection parameters
Transformants were originally isolated (Samuels et al., 1994a) as sub-surface colonies in semi-solid medium using a modification (Rosa and Hogan, 1992) of the plating protocol described by Kurtti et al. (Kurtti et al., 1987). Following electroporation, cells are allowed to recover in liquid culture for approximately 18 hours in the absence of selection before plating in semi-solid medium containing the selective agent and incubating in a humidified microaerobic atmosphere, typically a CO 2 incubator. Stably transformed bacteria form colonies in the semi-solid medium in five days to two weeks and can be isolated by aspirating a plug of agarose containing the colony to liquid medium. A PCR-screen for the presence of the selectable marker can be used to distinguish transformants from spontaneous antibiotic resistant mutants.
A method of selecting for clonal transformants by limiting dilution was developed by Yang et al. (Yang et al., 2004), who were unable to isolate ospA mutants in solid medium.
Recovered transformants are diluted into 100 ml of liquid medium with selection and distributed among several 96-well plates. Again, the plates are incubated in a humidified atmosphere containing CO 2 and positive wells are detected by the change in color of the medium from red to yellow, which is due to the production of lactic acid by the bacteria during growth, as detected by phenol red in the medium. Putative transformants can be expanded in a larger volume of medium containing antibiotic for confirmation and subsequent analysis. Regardless of the method used to isolate transformants, there is the possibility that two spirochetes are intertwined and thus a colony in semi-solid medium or a positive well at the endpoint of limiting dilution may not be derived from a single cell; therefore, extensive mixing, vortexing, and/or pipetting is required to separate cells before plating.

Not currently available
Effective counter-selectable markers have not been widely applied to B. burgdorferi. The use of the rpsL gene for counter-selection in Borrelia has been described but not utilized in any genetic studies to date (Drecktrah et al., 2010). The lack of defined culture medium, coupled with an inherently limited metabolic capability, has impaired the development of auxotrophic selection as an in vitro genetic tool for Borrelia. However, Richards and colleagues recently demonstrated that B. burgdorferi mutants lacking genes encoding the initial enzymes in acetate utilization are non-infectious in mice and have an absolute requirement for the addition of mevalonate to culture medium, suggesting a means of auxotrophic selection in vitro and potential utility as selectable markers in vivo (Richards et al., 2015).

Gene inactivation
A number of reviews have outlined various strategies and provided detailed protocols for gene inactivation in Borrelia (Hyde et al., 2011b;Drecktrah and Samuels, 2018;Latham and Blevins, 2018;Radolf and Samuels, 2021;Samuels et al., 2018), and this information will not be reiterated in detail in this review. As described above, targeted gene inactivation by homologous recombination is straightforward but inefficient in B. burgdorferi, requiring microgram quantities of DNA and ranging in frequency from 10 −5 to 10 −8 , depending upon the recipient organism (Samuels et al., 1994a;Tilly et al., 2000;Elias et al., 2002). Fortunately, illegitimate recombination (insertion of transforming DNA at a nonhomologous site) does not commonly occur in B. burgdorferi and thus does not impede the recovery of intended mutants. Simple gene disruption can be accomplished by integration (through a single crossover event) of a circular suicide vector carrying an appropriate selectable marker and a fragment of homologous DNA. Alternatively, if homologous sequences flank the selectable marker in the inactivation construct, allelic exchange can occur with the targeted gene through a double crossover event. Depending upon the design of the inactivation construct, allelic exchange can result in simple disruption or deletion of the targeted gene. A somewhat more elegant form of allelic exchange entails modifying the coding sequence of the targeted gene rather than deleting it, to assess the resulting mutant phenotype. This approach has been used for site-directed mutagenesis of a number of B. burgdorferi genes (Knight et al., 2000;Yang et al., 2003;Earnhart et al., 2010;Earnhart et al., 2011;Drecktrah et al., 2013).
As mentioned previously, a single crossover event can occur with allelic exchange constructs if they are introduced as circular DNA. This can result in the unintentional introduction into B. burgdorferi of other antibiotic resistance markers present on the cloning vector. As stated above, some selectable markers commonly used in E. coli, such as those conferring resistance to ampicillin and tetracycline, are not permitted for genetic studies in B. burgdorferi due to the clinical utility of these antibiotics in treating Lyme disease. At the minimum, cloning vectors carrying these markers should be linearized prior to transformation into B. burgdorferi. Preferably, allelic exchange constructs should be cloned into vectors that do not contain such restricted markers or the genes encoding them should be inactivated prior to transformation of B. burgdorferi.
Several approaches for engineering large deletions in Borrelia have been described.
Chaconas and colleagues co-opted an endogenous cellular activity involved in the formation of telomeres in Borrelia to engineer staggered deletions from either end of a linear plasmid to targeted internal sites (Beaurepaire and Chaconas, 2005;Bankhead and Chaconas, 2007). Bestor et al. have adapted the Cre-lox site-specific recombination system to Borrelia and used it to generate large internal deletions on a linear plasmid (Bestor et al., 2010). Finally, a basic understanding of the genes/sequences required for autonomous replication and plasmid incompatibility has been applied to displace entire plasmids in Borrelia Eggers et al., 2002;Stewart et al., 2003;Byram et al., 2004;Stewart et al., 2005;Jewett et al., 2007a;Jewett et al., 2007b;Dulebohn et al., 2011;Dulebohn et al., 2013). Conditional mutants, as described elsewhere in this review, can now be generated with any of several inducible systems available for Borrelia, (Blevins et al., 2007;Gilbert et al., 2007;Whetstine et al., 2009;Latham and Blevins, 2018) Conditional mutants provide an important means to assess the contributions of genes that encode essential functions or for which unmodulated expression is lethal. Finally, all forms of gene inactivation or modification in Borrelia entail insertion of a selectable marker at a targeted site, with potential polar effects on adjacent sequences. Interrupted or enhanced transcription of upstream or downstream sequences following insertion of a constitutively expressed antibiotic resistance cassette is not uncommon and should be considered when designing gene inactivation constructs and interpreting experimental outcome. Promoter-less resistance cassettes have been used successfully within constitutive operons of B. burgdorferi to target individual genes without impacting downstream genes (Motaleb et al., 2011;Showman et al., 2016).

Complementation
Complementation plays a crucial role in genetic studies of B. burgdorferi due to the extremely low frequency with which targeted gene inactivation occurs and the inherent instability of the segmented genome, as described above. Although measures can be taken to ensure that wild-type and derivative clones contain the same plasmids (Purser and Norris, 2000;Labandeira-Rey and Skare, 2001;Eggers et al., 2002;Elias et al., 2002;Parveen et al., 2006;Bunikis et al., 2011;Norris et al., 2011), screening for more subtle genomic alterations elsewhere in the genome is not routinely feasible. To control for unforeseen or unidentified second site mutations, or polar effects on flanking sequences, the goal of complementation is to leave the original mutation unaltered, but to re-introduce a wild-type copy of the targeted gene in trans on an autonomously replicating shuttle vector or elsewhere in the genome, or in cis by allelic replacement. As described below, various technical considerations, such as copy number, DNA supercoiling, genomic organization and placement of the antibiotic cassette, should be considered when selecting which complementation approach to use.

Endogenous plasmid shuttle vectors
A number of shuttle vectors have been derived from the endogenous plasmids of B. burgdorferi. As reviewed in Radolf and Samuels (2021) the complete genome sequence of B. burgdorferi identified a set of paralogous gene families (PGFs) present on all plasmids, suggested to be involved in plasmid replication and maintenance, although lacking strict sequence homology to plasmid genes with comparable functions in other bacteria (Zückert and Meyer, 1996;Casjens et al., 2000). An early shuttle vector derived from a small, nonessential endogenous circular plasmid (cp9) of B. burgdorferi demonstrated that the region encompassing these PGFs did not support replication in E. coli, but was sufficient for autonomous replication in B. burgdorferi . This shuttle vector, termed pBSV2, and its derivatives have become the workhorse for plasmid-based transcomplementation and other applications in B. burgdorferi. In the absence of strong counterselection (Tilly et al., 2006), pBSV2 is relatively stable both in vitro and in vivo , and present at approximately 5-10 copies per cell (Tilly et al., 2006). However, sequences inserted on a shuttle vector (such as for complementation or fluorescent protein expression) can destabilize the shuttle vector, leading to its loss if antibiotic selection is not maintained. Depending on the mutation, the increased copy number of a wild-type gene on the shuttle vector has the potential for ineffective complementation due to deleterious effects of overproduction of the target protein Eggers et al., 2002;Tilly et al., 2006;Ristow et al., 2012;Kasumba et al., 2015). Additional shuttle vectors have been derived from other endogenous plasmids (lp25, lp28-1, cp26, and a cp32) by cloning their PGF counterparts together with a functional selectable marker (Eggers et al., 2002;Stewart et al., 2003;Byram et al., 2004;Beaurepaire and Chaconas, 2005). Along with autonomous replication, the PGF regions also mediate incompatibility, resulting in displacement of the endogenous B. burgdorferi plasmid from which they were derived Eggers et al., 2002;Stewart et al., 2003;Beaurepaire and Chaconas, 2005), unless selection exists for the coexistence of both replicons (Byram et al., 2004). Displacement with incompatible shuttle vectors thus has been used to experimentally manipulate plasmid content in B. burgdorferi Byram et al., 2004;Grimm et al., 2004;Jewett et al., 2007a;Jewett et al., 2007b;Dulebohn et al., 2011;Dulebohn et al., 2013).

Genetic reconstitution at the original locus
An alternative complementation strategy is genetic reconstitution, in which the mutated copy of the gene is replaced with a wild-type copy and another closely linked marker (Hübner et al., 2001;Tilly et al., 2001;Li et al., 2007b), similar to the strategy used to introduce point mutations (Knight et al., 2000;Yang et al., 2003). This approach, however, does not leave the original mutation unaltered. A potential advantage to complementation by genetic reconstitution, at the site of the mutated locus or elsewhere on the same replicon, is that the reintroduced wild-type gene copy is at the same copy number and similar supercoiling state as the original gene. Some attempts to complement mutations in trans on a shuttle vector have only partially or unsuccessfully restored the wild-type phenotype Ristow et al., 2012), raising questions about whether the context of the complementing gene has significantly altered its expression. Alternatively, such results could indicate the presence of an unrecognized second site mutation, which is responsible for the defective phenotype of the complemented mutant. Distinguishing among the possible causes of unsuccessful complementation is addressed below.

Trans-complementation in the genome
An improvement over trans-complementation on a shuttle vector is to insert the complementing gene elsewhere in the genome, such as on an essential plasmid (cp26, lp25, lp28-1 or lp36) (Byram et al., 2004;Kawabata et al., 2004;Grimm et al., 2005;Blevins et al., 2007;Gilbert et al., 2007;Jewett et al., 2007b;Caimano et al., 2019). This approach leaves the original mutation undisturbed and maintains in vivo selection for the reintroduced gene by virtue of insertion on an essential plasmid. A logical target for this approach is the bbe02 locus on lp25 as the insertion site. Currently, many research labs employ B. burgdorferi clones in which bbe02 has been inactivated as a means to increase transformation efficiency, yet maintain infectivity (Kawabata et al., 2004;Jacobs et al., 2006;Rego et al., 2011). Further, because lp25 is required for B. burgdorferi survival throughout the enzootic cycle, this provides biological selection for in vivo maintenance of the complementing gene. The allelic exchange suicide vectors pKBE-Strep and pKBE-Kan have been engineered with a multiple cloning site for insertion of the complementing gene upstream of a linked antibiotic cassette, and flanking DNA homologous to sequences upstrem and downstream of the bbe02 R/M locus (Kasumba et al., 2015). Validation of this approach via complementation of an ospC mutant demonstrated copy number on lp25 consistent with the copy number of the native gene on cp26, as well as functional and stable maintenance of the ospC gene in trans on lp25 (Kasumba et al., 2015). Similarly, the intergenic region between oppositely transcribed bb0445 and bb0446 genes on the chromosome has been used successfully for trans-complementation (Li et al., 2007a;Zhang et al., 2009). Innocuous sites on lp25, lp28-1 and lp36 have been useful for insertion of an antibiotic resistance cassette to provide antibiotic selection to ensure in vitro maintenance (Blevins et al., 2007) or to restore to B. burgdorferi clones lacking each of the critical virulence plasmids Jewett et al., 2007b). Given that this approach effectively preserved or reestablished the wild-type infection phenotype, these sites like bbe02 on lp25, provide preferable alternatives for stable trans-complementation for in vivo studies.

Limitations, failures and alternatives
Complementation, while critical to demonstrating the link between a mutant phenotype and gene, presents a substantial challenge when working with infectious B. burgdorferi clones retaining the full set of endogenous plasmids. As mentioned above, this is primarily due to the activities of restriction enzymes encoded by genes bbe02 and bbq67 on lp25 and lp56, respectively (Lawrenz et al., 2002;Kawabata et al., 2004). The DNA of B. burgdorferi contains N 6 -methylated adenine only when bbe02 and/or bbq67 are present, indicating that both genes encode adenine methyltransferase activity , and recently the sequence motifs recognized by these restriction enzymes were identified (Casselli et al., 2018). Two genetic approaches to address the barrier that these R/M loci present for transformation of infectious B. burgdorferi are (i) to use a genetic background that lacks lp56 and gene bbe02 (Kawabata et al., 2004;Jacobs et al., 2006;Rego et al., 2011;Jain et al., 2012;Showman et al., 2016;Aranjuez et al., 2019), or (ii) to perform genetic manipulations in a non-infectious, yet transformable, B. burgdorferi genetic background lacking lp25 and subsequently restore the in vivo-essential lp25-encoded pncA (bbe22) gene prior to mouse infection studies (Labandeira-Rey and Skare, 2001;Labandeira-Rey et al., 2003;Purser et al., 2003;Seshu et al., 2006;Weening et al., 2008). However, this second approach does not allow tick studies due to the requirement for lp25-encoded genes in addition to pncA for tick colonization and persistence Revel et al., 2005;Strother et al., 2005;Gilmore et al., 2014).
When repeated attempts to obtain an isogenic complemented mutant retaining essential plasmids are unsuccessful, an argument can be made linking a mutant phenotype and targeted gene if several independently derived mutants, obtained in separate transformation experiments, are compared with an isogenic wild-type clone. This argument is based on the low probability that second site mutations resulting in a similar mutant phenotype would have arisen independently in separate transformations. Occasionally, a quandary arises when a wild-type copy of the gene has been reintroduced successfully into the mutant strain, yet the defective phenotype persists, leading to the conclusion that complementation has failed. As discussed above, one explanation is that the complementing gene may not be appropriately expressed in trans due to changes in supercoiling, copy number or absence of key regulatory sequences. This issue may be addressed by employing one of the alternative complementation strategies. If persistence of the mutant phenotype in the complemented strain cannot be explained by dysregulated gene expression, the next possibility to investigate is polar effects of the original mutation on the transcription of adjacent sequences. If expression of adjacent transcripts appears to be altered by the mutation, then a different gene inactivation strategy should be employed that minimizes potential polar effects. Finally, if the complementing gene is intact and appropriately expressed, and there is no evidence that the mutation impacts adjacent sequences, then the most likely explanation for the defective phenotype of the complemented mutant strain is an unlinked and unrecognized mutation elsewhere in the genome. In this case, complementation has failed because the phenotype of the mutant strain does not derive from the introduced mutation, thus illustrating the essential role of complementation in genetic studies.

Use of published 5′ RNA-seq data to predict endogenous promoters
When designing a complementation construct it is important to incorporate sufficient sequence upstream of the complementing gene to ensure its appropriate regulation. Similarly, the genomic and transcriptional context of the gene should be taken into account, including whether or not the gene is transcribed alone or with other genes as a part of an operon. The B. burgdorferi 5′ end transcriptome has provided significant information for refined annotation of the B. burgdorferi genome, including a genome-wide map of transcription start sites and a comprehensive analysis of 5′ UTR length (Adams et al., 2017b). These data provide a valuable resource (https://lymedisease.med.ucf.edu/) for predicting endogenous promoters and putative regulatory sequence, critical for the design of effective mutagenesis and complementation approaches.

Surrogate genetics
Limitations on the genetic manipulation of B. burgdorferi have stimulated researchers to perform surrogate genetics in E. coli because of the well-developed genetic systems of the latter organism and the availability of extensively characterized mutants (Beckwith, 1991). Numerous B. burgdorferi genes have been expressed in E. coli from either their endogenous promoters or from cloning vector promoters, primarily as a source of purified recombinant protein. A number of B. burgdorferi genes have been used to complement E. coli or Salmonella mutants deficient in the homologous genes, which has permitted partial functional characterization of these borrelial genes (Tilly et al., 1993;Margolis et al., 1994;Knight and Samuels, 1999;Lin et al., 2001;Purser et al., 2003;Eggers et al., 2004;Liveris et al., 2004;Putteet-Driver et al., 2004;Bugrysheva et al., 2005;Jewett et al., 2011). Similarly, cloning of B. burgdorferi genes encoding phosphatidylcholine and phosphatidylglycerolphosphate synthase in E. coli permitted the characterization of these enzymatic activities (Wang et al., 2004). Expression in E. coli of a phage display library led to the identification of a putative integrin-binding adhesin of B. burgdorferi (Coburn et al., 1999;Coburn and Cugini, 2003). Surrogate genetics in E. coli also permitted the characterization of B. burgdorferi promoter elements involved in transcriptional regulation (Alverson et al., 2003;Boylan et al., 2003;Caimano et al., 2004;Eggers et al., 2004;Burtnick et al., 2007;Medrano et al., 2010). However, differences in metal acquisition and homeostasis between B. burgdorferi and E. coli have been cited as a reason why E. coli may not be an appropriate surrogate system for studying B. burgdorferi gene expression and regulation (Ouyang et al., 2010).The advances in genetic manipulation of B. burgdorferi described in this review suggest that the need to perform surrogate genetic experiments in other bacterial hosts may decrease in the future. However, given its complex culture medium and slow growth rate, it is unlikely that B. burgdorferi will ever be as easy to manipulate as E. coli.

Engineered gene expression Endogenous promoter fusions
The first tools employed to control gene expression in B. burgdorferi were its endogenous flaB and flgB promoters (Sohaskey et al., 1997;Bono et al., 2000;Sartakova et al., 2000).
These two promoters have been used extensively to constitutively express endogenous and foreign genes in B. burgdorferi, and to develop new genetic markers and genetic tools, as described throughout this review. Enzootic stage-specific expression has been achieved using the ospA and ospC promoters, for tick stage and mammalian stage expression, respectively (Tilly et al., 2016;Adams et al., 2017b). Constitutive expression mediated by flaBp has also been utilized to study the role of ospC-repression in the ability of B. burgdorferi to evade mammalian host defenses (Xu et al., 2006). Similarly, B. burgdorferi containing flaBp-ospAB gene fusions were used to demonstrate that constitutive expression of these proteins led to rapid clearance of Borrelia in mice by the immune system (Strother et al., 2007). For many applications, strong gene expression driven by the flaB and flgB promoters is experimentally useful; however, overexpression of certain proteins can be limited by possible toxic effects, mis-localization and even altered function (Hayes et al., 2014). Recently, to address these challenges, constitutive promoters of low and intermediate strengths were identified from a global transcriptome data set that measured B. burgdorferi gene expression at multiple in vitro growth phases (Arnold et al., 2016;Takacs et al., 2018).
This panel of cloned and validated promoters provides a range of levels of constitutively controlled expression of genes of interest (Takacs et al., 2018). Future analysis of the activities and utility of these promoters in vivo will be important.

Inducible gene expression
Inducible promoter systems that can regulate gene expression in a graded manner offer certain advantages over the total ablation of gene expression generated by null mutants, or constitutive expression generated by endogenous B. burgdorferi promoters. For example, information about the function of unknown but essential genes can be obtained through modulating gene expression and monitoring the phenotypic changes that occur when gene products become limiting. In B. burgdorferi, endogenous genetic switches have yet to be identified. However, well characterized genetic switches, such as the E. coli lactose operon (Jacob and Monod, 1961) and the E. coli tetracycline resistance gene of Tn10 (Gossen and Bujard, 1992;Bujard, 1999), have been adapted for use in B. burgdorferi.

lac system
The lac system has been adapted to experimentally regulate gene expression in B. burgdorferi (Blevins et al., 2007;Gilbert et al., 2007). Gilbert et al. (Gilbert et al., 2007) constructed the flacp inducible promoter by integrating the synthetic lacO id operator (Oehler et al., 1994) into the B. burgdorferi flgB promoter (Ge et al., 1997). This inducible promoter was fused to ospC and rpoS by directly incorporating flacp into the genome and replacing the native promoters of the respective genes (Gilbert et al., 2007). These investigators also Rosa and Jewett Page 15 generated B. burgdorferi strains that produce high levels of the LacI repressor by fusing the flgB promoter to lacI and inserting the hybrid into lp25, a plasmid required for infectivity (Purser and Norris, 2000;Labandeira-Rey and Skare, 2001). The expression of flacpcontrolled genes in this background is tightly regulated in response to the inducer isopropyl β-D-thiogalactopyranoside (IPTG); genomic flacp-ospC and flacp-rpoS fusions bypass the normal regulation by environmental factors (Gilbert et al., 2007). In addition, flacp-rpoS was used to demonstrate that RpoS was not sufficient to downregulate OspA in vitro (Caimano et al., 2007). This lac system was used in vivo (Gilbert et al., 2007) to further test the hypothesis that OspC was required at an early time point in infection (Tilly et al., 2006;Tilly et al., 2007). This experiment indicated that the inducible system functions in vivo and can be used to test genes required for infectivity (Gilbert et al., 2007).
An alternative system of gene regulation based on the E. coli lac operon was also shown to be functional in B. burgdorferi (Blevins et al., 2007). Gene regulation by the lac operator was demonstrated using a codon-optimized luciferase reporter (luc Bb ) and through expression of the endogenous B. burgdorferi protein BptA. In this system, the B. burgdorferi lac expression system consisted of a single plasmid based on the pKFSS1 shuttle vector with the T5/lac hybrid promoter pQE30 from plasmid QE30 (QIAGEN); this promoter contains two lac operators (Blevins et al., 2007). On the same plasmid, a codon-optimized lacI gene was present under the control of the flaB promoter (Blevins et al., 2007). Adding IPTG to B. burgdorferi carrying this plasmid with either luc Bb or bptA cloned under the control of the pQE30 promoter resulted in increased expression of these gene products. This lac system worked relatively well in B. burgdorferi, but induction was incomplete, which may result from either insufficient IPTG or excessive LacI inside the cell, or weak activity of the pQE30 promoter in Borrelia.

tet system
Like the lac system, the tetracycline-inducible system has been adapted for controlling gene expression in B. burgdorferi (Cabello et al., 2006;Whetstine et al., 2009). These approaches use hybrid tetracycline-responsive B. burgdorferi promoters and constitutively expressed tetR. Placement of the tet operator upstream of the −10 and −35 sequences of the bmpA promoter resulted in inducible, but significantly leaky expression of the target gene (Cabello et al., 2006). In contrast, the hybrid P ost promoter includes the ospA promoter and 5′ UTR sequence with the first 19 nucleotides of the 5′ UTR replaced with the tetO sequence (Whetstine et al., 2009). The P ost promoter, along with the constitutively expressed tetR gene, are engineered on a single kanamycin-resistant shuttle vector, pCRW53 (Whetstine et al., 2009). This system demonstrates little to no background expression in the absence of anhydrotetracycline (ATc) and titratable expression up to 2 μg/ml ATc. Yet, maximal P ost promoter activity was not achieved, perhaps due to read through-transcription of tetR. In its current design, the pCRW53 tet-inducible system is limited to environments in which ospA is expressed and further adaptation is required for its application during mammalian infection (Whetstine et al., 2009).
These results demonstrate that gene expression can be artificially regulated in B. burgdorferi using either the lac system or the tet system. Modifications yielding stronger and tighter induction of genes under the control of these inducible promoters should be possible in the future. The availability of inducible expression systems for B. burgdorferi has allowed isolation of conditional lethal mutants through regulated repression of a number of essential genes. (Lenhart and Akins, 2010;Groshong et al., 2012;Bandy et al., 2014;Ye et al., 2014;Chu et al., 2016;Groshong et al., 2017;Drecktrah et al., 2020). Conversely, increased expression of rpoS from an inducible lac promoter resulted in cell death, consistent with a failure to engineer over-expression of rpoS from the strong flaB promoter (Chen et al., 2013). Further optimization of these genetic switches, as well as development of endogenous genetic switches, will improve manipulation of B. burgdorferi gene expression in vivo and open a new avenue for studies of pathogenesis in Lyme borreliosis.

Random tagged mutagenesis
The use of transposon mutagenesis as a tool for bacterial genetics represents an important advance that has generally superseded chemical and radiation mutagenesis (Hayes, 2003;Bosse et al., 2006;Salama and Manoil, 2006;Burrack and Higgins, 2007). Mutagenized bacteria are easily isolated on the basis of antibiotic resistance or other selectable markers carried on the transposable element, and the mutated gene can be readily identified by routine molecular procedures (Judson and Mekalanos, 2000;Hayes, 2003;Bosse et al., 2006;Salama and Manoil, 2006;Burrack and Higgins, 2007).
Genome-wide transposon mutagenesis of B. burgdorferi was first demonstrated with a suicide plasmid that contained a modified Himar1 element of the mariner family . In this system, called pMarGent, the inverted repeats of the transposable element flank a ColE1 origin of replication and a gentamicin-resistance cassette that functions in both E. coli and B. burgdorferi, while the Himar1 transposase is carried on a portion of the suicide plasmid that is not mobilized, and hence only transiently present in B. burgdorferi Stewart and Rosa, 2008). This approach yielded a saturated mutant library in a non-infectious B. burgdorferi clone and permitted the isolation and characterization of a large number of mutants, with random transposon insertions in the linear chromosome and all linear and circular plasmids . These experiments demonstrated the strength and feasibility of transposon mutagenesis in B. burgdorferi.
The utility of transposon mutagenesis as a tool to study B. burgdorferi virulence was illustrated by using the pMarGent system to generate a limited number of transposon mutants in an infectious B. burgdorferi clone and subsequently assessing their ability to infect mice (Botkin et al., 2006a;Botkin et al., 2006b). However, this approach was limited due to the instability of the pMarGent plasmid in E. coli and the low transformation frequency of infectious B. burgdorferi. To reduce the possibility of recovering inactive plasmid from E. coli due to transposition of the gentamicin marker, a kanamycin resistance cassette was added adjacent to the transposase, creating the vector pGKT (Stewart and Rosa, 2008). However, this modification does not fully stabilize pGKT because both markers can be retained in the same cell at different genomic locations following transposition. The limitation on transposon mutagenesis imposed by B. burgdorferi's low transformation frequency has been partially addressed, as in other genetic approaches, with B. burgdorferi clones lacking lp56 and gene bbe02 on lp25, thereby avoiding the R/M systems they encode (Kawabata et al., 2004;Rego et al., 2011;Lin et al., 2012;Lin et al., 2014;Lin and Gao, 2018). Lin et al. further adapted the pGKT vector for signature tagged mutagenesis (STM) by inserting eleven distinct 7-bp tags in the region between the ColE1 and inverted terminal repeat 2 (Lin et al., 2012). This molecular genetic advance allowed for generation of a genome-scale STM library of 4,479 sequence-defined mutants, which can be interrogated in groups of clones to identify genes important for in vivo and in vitro phenotypes (Lin et al., 2012).
Isolation of a random, saturated mutant library requires an insertion every 200 base pairs on average, thus a completely saturated transposon library of B. burgdorferi would contain at least 7,500 independent mutants, since the B. burgdorferi genome is approximately 1,500,00 bp (Fraser et al., 1997;Casjens et al., 2000;Salama and Manoil, 2006). Generation of a genome-wide STM library necessitated development of a multiplex, high-throughput method for plasmid content analysis of the individual clones. The Luminex plasmid analysis assay employs a combination of three multiplex PCRs using 24 primer sets specific for each of the 21 B. burgdorferi B31 plasmids, and Luminex xMAP technology for microfluidic bead-based detection of the B. burgdorferi replicons (Norris et al., 2011). This approach allows simultaneous analysis of 30 B. burgdorferi clones in a 96-well plate format and has been used to assess the plasmid content of 44 B31 clones and 4,464 transposon mutants in an STM library (Norris et al., 2011). To date, 434 STM mutants from the library of 4,479 unique mutants have been screened for infectivity in mice (Lin et al., 2012), resulting in the identification of infection-relevant candidate genes (reviewed in Radolf and Samuels, 2021). Initially, comparison of the input and output STM mutant pools to identify those mutants that were attenuated for infection involved a semi-quantitative Luminex-based multiplex PCR procedure (Lin et al., 2012). More recently, Tn-seq has been used to identify genes contributing to a phenotype. Using this approach, the fitness of STM mutants is determined by sequencing the genomic DNA flanking the transposon insertion sites in the entire library, before and after undergoing selection in the environment of interest (Troy et al., 2016). Tnseq has been applied in B. burgdorferi to identify gene candidates important for mouse infection, survival in ticks, resistance to reactive oxygen and nitrogen species and carbohydrate utilization (Troy et al., 2016;Ramsey et al., 2017;Phelan et al., 2019), (see Radolf and Samuels, 2021), as well as to identify when and where B. burgdorferi encounters barriers to mouse infection (Troy et al., 2016).
STM library screens are powerful molecular genetic tools to identify candidate genes of interest. It is important to note, however, that genetic screens of this nature do not assess the fitness of a single mutant alone in the challenge environment, rather the fitness of all mutants in the input pool relative to each other. For this reason, the phenotype of a single mutant alone may differ from that in the context of a pool of other clones. Further studies, including targeted inactivation and complementation, are required to validate the roles of the implicated genes in the phenotype of interest. Despite technical advancements in the transformation frequency of B. burgdorferi, the challenge of achieving a saturated transposon mutant library remains daunting. The increased technical feasibility and reduced cost of whole genome re-sequencing raises the possibility for the resurgence of chemical and radiation mutagenesis approaches as alternatives for genetic screens in B. burgdorferi.

Reporter genes
Reporter genes encode proteins whose function, such as enzymatic activity, fluorescence or selectable phenotype, can be readily assayed or detected; they have been used extensively to monitor gene expression, to track cells in complex environments, and to determine the subcellular locations of proteins. Different types of reporters are typically better suited for distinct applications. The reporter genes currently available for Borrelia and their pertinent features are summarized in Table 2.

In vitro gene expression
A number of reporter genes have been used successfully as transcriptional fusions with B. burgdorferi promoters to study gene expression, including cat (chloramphenicol acetyl transferase), luc (luciferase), lacZ (β-galactosidase), and gfp (green fluorescent protein) (Sohaskey et al., 1997;Sartakova et al., 2000;Carroll et al., 2003;Eggers et al., 2004;Blevins et al., 2007;Hayes et al., 2010). The initial studies in B. burgdorferi used promoter:cat fusions to assay gene expression (Sohaskey et al., 1997;Sohaskey et al., 1999), but cat has largely been replaced by other reporter genes that are easier to use or have broader utility. β-galactosidase, encoded by the lacZ gene, is a widely used reporter whose activity can be easily monitored in bacteria grown in liquid and solid media without specialized equipment or radioactivity (Sambrook et al., 1989). A codon-optimized lacZ reporter was developed for B. burgdorferi, and β-galactosidase activity measured in lysates of cultured spirochetes carrying lacZ transcriptional fusions with a standard Miller assay (Hayes et al., 2010). The omission of phenol red from BSK medium facilitates detection of lacZ-expressing (blue) B. burgdorferi colonies in plates containing X-Gal (Hayes et al., 2010). Use of a lacZ reporter in a blue-white colony screen confirmed the identity of a negative regulator of ospC encoded on linear plasmid lp17 (Sadziene et al., 1993;Hayes et al., 2014). The availability of lacZ further expands the repertoire of tools for investigating mechanisms of gene regulation in B. burgdorferi.
Whereas cat and lacZ reporters are routinely used to quickly measure gene expression in a population of cells, fluorescent protein reporters offer other advantages, such as directly monitoring gene expression at the individual cell level by microscopy or flow cytometry. Green fluorescent protein (GFP) was the first fluorescent protein reporter used to monitor gene expression and identify promoter elements in B. burgdorferi (Carroll et al., 2003;Eggers et al., 2004;Falkow, 2004;Valdivia et al., 2006;Srivastava and de Silva, 2008). GFP signal can be influenced by temperature, pH and oxygen concentration, but is relatively stable, which limits its utility for detecting rapid changes in gene expression (Valdivia et al., 2006). An expanded range of alternative fluorescent reporter genes that have been expressed in B. burgdorferi include yellow fluorescent protein, yfp; cyan fluorescent protein, cfp (Eggers et al., 2002); monomeric red fluorescent protein, mRFP (Schulze and Zuckert, 2006) and dTomato (Lee et al., 2010). Recently, a number of B. burgdorferi codon-optimized reporters have been developed, including CFP variants, mCerulean and msfCFP; GFP variants, mEGFP and msfGFP; YFP variants, mCitrine and msfYFP and a monomeric improved variant of mRFP1, mCherry (Takacs et al., 2018). In addition, an infrared fluorescent protein (iRFP) reporter was developed. iRFP is an extrinsically fluorescent protein that covalently binds biliverdin as the fluorophore (Filonov et al., 2011). Growth of B. burgdorferi expressing iRFP in medium supplemented with the biliverdin cofactor results in near-infrared fluorescence of the cells (Takacs et al., 2018). In addition to in vitro studies, iRFP may be useful for in vivo live imaging of infected mice and/or ex vivo imaging of infected tissue explants, as described elsewhere in this review, due to the high excitation light penetrance and low tissue auto-fluorescence in the infrared region of the spectrum. Further, the endogenous levels of biliverdin in animal tissues are predicted to be sufficient for maximal fluorescence of iRFP-expressing B. burgdorferi (Takacs et al., 2018), although this has yet to be tested experimentally. Combinatorial imaging studies demonstrated that CFP variants can be co-imaged with mCitrine, mCherry and iRFP, and GFP variants can be co-imaged with mCherry and iRFP, suggesting the possibility for simultaneous imaging of up to four proteins in the same spirochete (Takacs et al., 2018).

Subcellular protein localization
In addition to their utility as transcriptional fusions for gene expression studies, translational fusions of fluorescent proteins represent important tools for cell biology, as described in greater detail in Radolf and Samuels (2021). Monomeric red fluorescent protein (mRFP) has been used in this capacity to investigate the lipoprotein export pathway of Borrelia (Schulze and Zuckert, 2006) and to develop a FACS-based lipoprotein localization screen (Kumru et al., 2010). More recently, an mCherry fusion protein was employed in a microscopy-based approach, coupled with quantitative image analysis, to investigate the dynamic cellular localization of an envelope protein of Borrelia (Takacs et al., 2018). A key feature of mRFP and mCherry as fusion partners for protein localization studies is that they continue to fluoresce outside of the cytoplasm, which earlier variants of GFP did not allow (Schulze and Zuckert, 2006). However, this previous limitation has been overcome with super-folding variants of GFP (sfGFP) that maintain fluorescence in periplasmic and extracellular locations (Pedelacq et al., 2006;Dinh and Bernhardt, 2011). A shortcoming of fluorescent proteins as translational fusions is their relatively large physical size, which can hinder the function, folding or localization of the protein to which they are fused. This potential drawback can be avoided with an alternative fluorescent labeling approach that entails tagging a cellular protein with a small tetra-cysteine motif that can bind fluorescent biarsenical dyes (Griffin et al., 1998). This approach has been used to label membrane proteins in B. burgdorferi and while avoiding the size constraints of fluorescent proteins, it is considerably less sensitive, requires a labeling step and can be limited by non-specific binding of the fluorescent dye (Hillman et al., 2019).

In vivo tracking and imaging
Beyond applications with spirochetes grown in vitro, reporter genes are important tools for detecting B. burgdorferi in ticks and mice (see Radolf and Samuels, 2021, for an in-depth review). GFP has permitted study of gene expression of individual spirochetes in ticks (Bykowski et al., 2006;Miller et al., 2006), and this reporter was used to visualize the interaction of the spirochete with the mouse vasculature in real time with high resolution imaging (Moriarty et al., 2008). Furthermore, GFP-expressing B. burgdorferi have been visualized in the intact skin of living mice and time-lapse images taken early after infection have allowed analysis of spirochete motility in vivo (Sultan et al., 2015;Novak et al., 2016).
GFP-expressing B. burgdorferi were used to track spirochete migration from an infected mouse to a feeding tick (Bockenstedt et al., 2014) and spirochete dissemination within ticks (Dunham-Ems et al., 2009).
Further expansion of in vivo live imaging of B. burgdorferi during infection has come with adaptation of a codon-optimized luciferase gene (luc Bb ) from the firefly Photinus pyralis for use in B. burgdorferi (Blevins et al., 2007). Luciferase activity of the firefly luc Bb reporter, which requires an exogenous substrate and is measured with a luminometer, was first used to monitor the activity of endogenous Borrelia promoters and an ITPG-induced lac expression system (Blevins et al., 2007), as described above. Subsequently, in vivo bioluminescent imaging of mice infected with B. burgdorferi constitutively synthesizing luciferase has been used for temporal and spatial tracking of spirochete infection at the population level (Hyde et al., 2011a;Chan et al., 2015;Wager et al., 2015;Skare et al., 2016;Adams et al., 2017b;Hyde and Skare, 2018). The luc Bb gene, driven by the strong constitutive flaB promoter and expressed on a multi-copy shuttle vector, allows detection of bioluminescent B. burgdorferi in live mice as early as three days post-inoculation (Adams et al., 2017b). This technique works well using both BALB/c and C3H/HeN mice (Hyde et al., 2011a;Adams et al., 2017b). The relative luminescence units (RLUs) detected by imaging mice infected with bioluminescent B. burgdorferi correlates with spirochete load in tissues of the same mice, as measured by quantitative PCR, allowing assessment of the ability of B. burgdorferi mutant clones to survive and disseminate in a live animal over time (Hyde et al., 2011a;Wager et al., 2015).

Quantifying in vivo gene expression
Application of in vivo bioluminescent imaging for quantitation of B. burgdorferi promoter activity as a means to measure expression of target genes during infection is less well developed. Currently this approach involves normalizing the luminescence signals of mice infected with B. burgdorferi carrying luc Bb fused to the target gene promoter, with those of other mice infected with B. burgdorferi constitutively expressing luc Bb , and subsequent qPCR analysis to demonstrate that all mice carry equivalent bacterial loads at the endpoint of the study (Skare et al., 2016). A dual reporter system that allows simultaneous measurement of the luminescence signal of the target promoter and a control constitutive promoter from the same population of spirochetes would provide a more straightforward and robust method to quantitate gene expression during infection. The first dual luciferase reporter system of B. burgdorferi was recently developed (Adams et al., 2017a). For this reporter system, constitutive expression of the B. burgdorferi codon-optimized Renilla reniformis (sea pansy) luciferase gene (P flaB -rluc Bb ) allows normalization of the activity of a second promoter fused to the B. burgdorferi codon-optimized firefly luciferase (fluc Bb ) gene on the same shuttle vector plasmid. The bioluminescence signals of the two reporters are distinguishable due to their respective specificities for different substrates, coelenterazine and luciferin. This dual luciferase system is a simple and robust approach for quantitative analysis of B. burgdorferi promoter activity under in vivo growth conditions and in infected ticks (Adams et al., 2017a). Application of this dual reporter system to B. burgdorferi promoter activity during mammalian infection remains a challenge, however, because coelenterazine demonstrates high background fluorescence in the mouse that was not circumvented through extensive modification of experimental parameters (Adams et al., 2017a). Future development of a dual reporter system for in vivo live imaging of B. burgdorferi infection may include a combination of green-and red-shifted luciferase enzymes (Branchini et al., 2005) or application of iRFP (Takacs et al., 2018), as discussed above.

Identifying in vivo promoters
Reporter genes are not limited to those whose expression can be measured as an enzymatic activity, fluorescence, luminosity or antibiotic resistance. A gene that is required for survival in a specific environment can also be used as a reporter of promoter activity. This concept is the basis of in vivo expression technology (IVET), a powerful and versatile method for genome-wide identification of promoters that are active in a host environment, by complementation of an auxotrophic mutation through gene fusion (Mahan et al., 1993). IVET has been adapted for B. burgdorferi using the pncA gene (Ellis et al., 2013;Casselli and Bankhead, 2015), which encodes a nicotinamidase activity that is required during infection of a mammalian host (Purser et al., 2003). The pncA promoter trap provides strong selection for in vivo-active promoters, as only those clones from the IVET library that express pncA during infection will survive in the mouse. Moreover, when expressed, pncA partially complements the phenotype of B. burgdorferi lacking virulence plasmid lp25 (on which pncA is encoded), allowing the IVET library to be generated in a more readily transformable genetic background, as described above (Lawrenz et al., 2002;Purser et al., 2003;Ellis et al., 2013). The initial IVET method in B. burgdorferi used the pncA reporter alone and identified promoters that were active both in vitro and during infection, as well as those that were specifically active during infection (Ellis et al., 2013;Adams et al., 2017b). This approach identified 233 unique infection-active promoters across a majority of the B. burgdorferi replicons (Ellis et al., 2013;Adams et al., 2017b). These sequences have led to the identification of novel genes critical for B. burgdorferi infectivity (Jain et al., 2015;Showman et al., 2016), as well as the discovery of previously unannotated intergenic, intragenic and antisense transcripts throughout the B. burgdorferi genome (Adams et al., 2017b).

In vivo phenotype
An aspect of genetic studies with B. burgdorferi that requires recognition (but is not unique to this bacterial pathogen) is the phenotype of the organism and the context in which gene function is addressed. Cultured spirochetes look and respond quite differently from genetically identical organisms present in infected ticks or mice (Iyer et al., 2015). Thus in vitro assays may not accurately reflect or predict the in vivo function of a particular B. burgdorferi gene product because other differentially synthesized bacterial proteins could mask or expose its function. There is no ready solution to this potential problem, but interpretation of results obtained from in vitro assays should include this caveat. Propagation of spirochetes in implanted chambers in the peritoneal cavity of rats results in bacteria exhibiting a partially host-adapted state, as assessed by altered patterns of protein synthesis and gene expression (Akins et al., 1998;Revel et al., 2002;Brooks et al., 2003;Caimano, 2005) and presents a somewhat more relevant source of bacteria for testing the in vitro properties of mutants.
Offsetting the inefficiencies and challenges of genetic studies in B. burgdorferi is the availability of relevant animal models in which to analyze mutants. Rodents are natural reservoir hosts, and both ticks and mice can be experimentally infected, making the entire infectious cycle amenable to investigation. Another publication (Radolf and Samuels, 2021) describes the animal models of Lyme disease in greater detail. Several parameters of the in vivo model that merit attention with respect to mutant characterization are infectious dose, mode and route of challenge, mouse strain, number of animals and tissues analyzed, quantitation of tissue burden and gene expression, and the statistics used to evaluate the data. Competitive index may be a useful parameter for discovery and quantitation of more subtle phenotypes (Bestor et al., 2012;Dulebohn et al., 2013;Arnold et al., 2015). Experimental design typically is influenced by the mutant phenotype and the focus of the investigation.
When designing, executing and interpreting in vivo genetic screens that use a highly complex inoculum (such as a saturated library of mutants) in which the ability to be recovered (or not) is a defining trait (such as IVET or Tn-seq), it is important to keep in mind the population dynamics of B. burgdorferi infection. Population bottlenecks during infection have been experimentally demonstrated using both Tn-Seq and genetic bar-coding approaches (Troy et al., 2013;Rego et al., 2014). A significant population bottleneck was found to occur at the site of infection, due in part to early innate immune responses (Troy et al., 2013;Casselli and Bankhead, 2015). Population bottlenecks were also detected during persistent infection of mice, as well as during spirochete acquisition and transmission by ticks (Rego et al., 2014;Phelan et al., 2019). These data highlight the potential misinterpretation of a mutant phenotype in the context of a mixed infection (in vivo genetic screen or competition assay) because clones can undergo stochastic elimination at a bottleneck independently of their genotype.

Genetic manipulation of relapsing fever spirochetes
A significant advance in the Borrelia field has been the development of a genetic system for the relapsing fever (RF) spirochetes (Battisti et al., 2008;Fine et al., 2011;Guyard et al., 2013;Lopez et al., 2013;Fine et al., 2014;Raffel et al., 2014;James et al., 2016;Krishnavajhala et al., 2017;James et al., 2018;Jackson-Litteken et al., 2019). The availability of genetic tools and complete genomic sequences for multiple RF species (Lescot et al., 2008;Elbir et al., 2012;Miller et al., 2013;Barbour and Campeau Miller, 2014;Barbour, 2016;Wilder et al., 2016;Elbir et al., 2017;Kuleshov et al., 2020) has permitted investigation of genetically modified RF spirochetes in their respective tick vectors or rodent hosts, and informative comparisons with other Borrelia, as detailed in Radolf and Samuels (2021). As with LD spirochetes, there is considerable variation among RF strains in the efficiency with which they can be grown, transformed or recovered as colonies in solid medium (Raffel et al., 2018). A noteworthy feature of genetic studies with RF spirochetes is retention of infectivity following prolonged cultivation (Lopez et al., 2008), suggesting a more stable segmented genome or a stronger requirement for plasmidencoded functions during in vitro growth, relative to the LD Borrelia. Additional research is needed to determine if this attribute is true for all RF strains and clones, and how it is influenced by media and culture conditions (Raffel et al., 2018). Another advantage of RF Borrelia for genetic studies is the relatively high density they reach in the blood of an infected mouse during a relapse, permitting direct visualization and quantitation of in vivo organisms, and sufficient material for downstream analyses.
The genetic approaches and tools developed for the LD spirochetes provided a strong foundation for molecular genetic manipulation of the closely related RF spirochetes. Generation of the first site-directed allelic-exchange deletion mutant in Borrelia hermsii was reported in 2008 (Battisti et al., 2008). Furthermore, this work included a method for genetic transformation by electroporation and isolation of individual B. hermsii clones in liquid medium by limiting dilution in 96-well plates. In addition, antibiotic resistance cassettes and a B. hermsii shuttle vector for trans-complementation and expression studies were described (Battisti et al., 2008). Further expansion of the genetic tools for RF spirochetes came with the generation of an infectious B. hermsii clone that constitutively expresses GFP, allowing visualization of fluorescent spirochetes both in vitro and in vivo (Fine et al., 2011). The methods for genetic manipulation of B. hersmii have been applied successfully to another RF spirochete, B. turicatae (Lopez et al., 2013;Krishnavajhala et al., 2017). Most recently, a method for reliable growth of B. hermsii and B. turicatae in solid medium has been developed, allowing isolation of single colonies following genetic manipulation, as well as enumeration of colony forming units from both in vitro and in vivo samples (Raffel et al., 2018). The application of reverse genetics to the RF spirochetes has opened the door to broad investigation of these tick-borne pathogens (Raffel et al., 2014;James et al., 2016James et al., , 2018Jackson-Litteken et al., 2019;Schwan et al., 2020).

Conclusion
The genetic toolbox for B. burgdorferi continues to grow, expanding the spectrum of approaches available for investigating the genetic mechanisms driving the biology of this fascinating pathogen. An increased understanding of the R/M systems that limit transformation efficiency has resulted in B. burgdorferi clones with enhanced capacity for transformation while retaining infectivity. These B. burgdorferi clones have been critical for improved recovery of targeted mutant and complemented clones, as well as the generation of genome-wide mutant and promoter-trap libraries. The development of a wide array of new reporter genes have opened the door for novel microscopy, flow-cytometry, genetic screening, in vivo bioluminescence, and in vivo selection assays. Moreover, new constitutive and inducible promoters allow for various levels of engineered expression of genes of interest. Molecular genetic approaches pioneered in B. burgdorferi have facilitated the development of tools for genetic manipulation and investigation of the closely related RF spirochetes. Transformation, plating in solid medium, reverse genetics, and florescencetagging of B. hermsii and B. turicatae are critical advances for understanding these medically important spirochetes. In sum, although there are still challenges and obstacles to overcome, significant strides have been made in Borrelia genetics, and persistent innovation and creativity will continue to increase and enhance the genetic tools available for the study of LD and RF spirochetes. reporter genes for Borrelia REPORTER APPLICATION(S)