The Diadenylate Cyclase CdaA Is Critical for Borrelia turicatae Virulence and Physiology

ABSTRACT Relapsing fever (RF), caused by spirochetes of the genus Borrelia, is a globally distributed, vector-borne disease with high prevalence in developing countries. To date, signaling pathways required for infection and virulence of RF Borrelia spirochetes are unknown. Cyclic di-AMP (c-di-AMP), synthesized by diadenylate cyclases (DACs), is a second messenger predominantly found in Gram-positive organisms that is linked to virulence and essential physiological processes. Although Borrelia is Gram-negative, it encodes one DAC (CdaA), and its importance remains undefined. To investigate the contribution of c-di-AMP signaling in the RF bacterium Borrelia turicatae, a cdaA mutant was generated. The mutant was significantly attenuated during murine infection, and genetic complementation reversed this phenotype. Because c-di-AMP is essential for viability in many bacteria, whole-genome sequencing was performed on cdaA mutants, and single-nucleotide polymorphisms identified potential suppressor mutations. Additionally, conditional mutation of cdaA confirmed that CdaA is important for normal growth and physiology. Interestingly, mutation of cdaA did not affect expression of homologs of virulence regulators whose levels are impacted by c-di-AMP signaling in the Lyme disease bacterium Borrelia burgdorferi. Finally, the cdaA mutant had a significant growth defect when grown with salts, at decreased osmolarity, and without pyruvate. While the salt treatment phenotype was not reversed by genetic complementation, possibly due to suppressor mutations, growth defects at decreased osmolarity and in media lacking pyruvate could be attributed directly to cdaA inactivation. Overall, these results indicate CdaA is critical for B. turicatae pathogenesis and link c-di-AMP to osmoregulation and central metabolism in RF spirochetes.

intracellular c-di-AMP, (ii) significantly elongated morphology, and (iii) decreased gene expression and protein production of the virulence regulators BosR and RpoS (64). In the second study, Savage et al. showed that overexpression of cdaA in vitro led to no identifiable changes in expression of several known regulators and virulence factors (78). Although overexpression of cdaA led to an increase in CdaA levels, no increase in intracellular c-di-AMP was observed. These studies have provided insight into functions and regulation of c-di-AMP signaling in Borrelia, but several questions remain. Specifically, the physiological role of CdaA has yet to be defined in any Borrelia species. Additionally, given the evolutionary divergence, unique pathologies, and distinct enzootic cycles of LD and RF spirochetes, it is possible that c-di-AMP could have unique functions in these two groups of Borrelia (3,4,(79)(80)(81).
Herein, we aimed to investigate the role of CdaA in the TBRF spirochete Borrelia turicatae. Using a murine model of RF, we demonstrated that cdaA mutants are significantly attenuated and that this phenotype is reversible by genetic complementation. Next, due to the essentiality of c-di-AMP for in vitro viability in many bacterial systems, whole-genome sequencing (WGS) was performed on independently generated cdaA mutants to identify potential suppressor mutations. These analyses revealed single-nucleotide polymorphisms (SNPs) in genes potentially involved in membrane transport, metabolism, and translation. A conditional cdaA mutant was then used to demonstrate that depletion of CdaA is detrimental to bacterial growth and physiology, supporting the hypothesis that compensatory suppressor mutations are required for normal bacterial physiology in vitro upon cdaA mutation. We next investigated phenotypes associated with cdaA mutation in vitro to identify potential reasons for the infection defect. cdaA mutation had no effect on bosR or rpoS expression or protein production, suggesting that c-di-AMP-mediated regulation may differ between LD and RF spirochetes. Interestingly, the cdaA mutant exhibited significant growth defects when treated with salt, cultured at decreased osmolarity, or grown in media lacking pyruvate. The salt treatment phenotype of the cdaA mutant was not reversed by genetic complementation, possibly due to suppressor mutations, but growth defects at decreased osmolarity and in media lacking pyruvate could be attributed directly to mutation of cdaA. Overall, these data link c-di-AMP signaling to virulence, osmoregulation, and central metabolism in RF spirochetes.

RESULTS
Generation and complementation of the BtDcdaA mutant. c-di-AMP-dependent signaling is important for the virulence of several bacteria (44,45,54,(67)(68)(69)(70)(71). The chromosomally encoded protein BT0008, referred to as CdaA herein, is predicted to be the only DAC in B. turicatae, and homologs are encoded in all Borrelia spirochetes (33,34,64,77,78). B. turicatae CdaA is a 258-amino-acid, 29.02-kDa protein with a predicted pI of 7.74. CdaA is annotated as a TIGR00159 family protein, and the NCBI Conserved Domain Database identified amino acids 49 to 258 as belonging to the DisA_N superfamily (E value = 1.52e-83), both of which are consistent with CdaA homologs in other bacteria (82)(83)(84). Additionally, TMHMM-2.0 predicted CdaA to have three transmembrane domains (amino acids 15 to 32, 39 to 61, and 66 to 85) close to the N terminus (85,86). The N terminus and C terminus would be located in the periplasm and cytoplasm, respectively, which is characteristic of CdaA proteins (34,87). Finally, as is the case with other CdaA proteins, two probable coiled-coil domains were identified within the DisA_N superfamily domain (amino acids 109 to 149 and 213 to 251) by Waggawagga coiled-coil prediction analyses (34,88). To determine the contribution of the B. turicatae cdaA homolog during mammalian infection, allelic exchange mutagenesis was used to replace an internal region of the open reading frame (ORF) in wildtype B. turicatae (BtWT) with an aacC1 gentamicin resistance cassette, generating BtDcdaA (Fig. 1A). Genetic complementation of BtDcdaA was achieved by inserting an aphI kanamycin resistance cassette and a copy of cdaA with the putative promoter region into the BtDcdaA chromosome adjacent to the site of mutagenesis, creating the BtDcdaA C-cis strain (Fig. 1B). Genotypic confirmation of BtDcdaA and BtDcdaA C-cis was performed using PCR specific for internal regions of cdaA, aacC1, and aphI genes (Fig.  1C). PCR to amplify a region of the flagellin (flaB) gene was also performed as an amplification control. PCR for cdaA only produced amplicons in BtWT and BtDcdaA C-cis . In addition, PCRs specific for the resistance markers failed to yield amplicons with BtWT, whereas BtDcdaA and BtDcdaA C-cis were positive for the aacC1 gene and aphI gene, respectively. Finally, PCR for flaB resulted in amplicons of the expected size for all strains. Immunoblot analyses were then performed to assess CdaA production in BtWT, BtDcdaA, and BtDcdaA C-cis (Fig. 1D). As expected, CdaA was detected in BtWT and BtDcdaA C-cis but undetectable in BtDcdaA. Importantly, levels of FlaB, which was included as a loading control, were consistent across strains. In all, these results indicate successful mutation of cdaA and genetic complementation in BtDcdaA and BtDcdaA C-cis , respectively.
The BtDcdaA mutant is significantly attenuated in a murine needle-challenge model of RF. Given the importance of c-di-AMP and DACs for virulence of other bacteria, we hypothesized that CdaA would be required for mammalian infection (44,54). To test this hypothesis, a murine model of RF was used in which groups of four mice were intradermally needle inoculated with 10 2 BtWT, BtDcdaA, or BtDcdaA C-cis bacteria (Fig.  2). On days 3 to 14 postinfection, blood samples were taken, and bacteremia was measured by quantitative PCR (qPCR). As expected, all four mice infected with BtWT and BtDcdaA C-cis experienced recurring bouts of spirochetemia, with the first peak The mutational (pUAMS248B) and complementation (pUAMS313B) constructs were included as positive controls, and reactions with no template (H 2 O) served as a contamination control. MW denotes the DNA standard, and numbers to the left indicate molecular weight in kb. (D) Immunoblot confirmation of BtDcdaA and BtDcdaA C-cis . Whole-cell lysates of late-exponential-phase BtWT, BtDcdaA, and BtDcdaA C-cis were separated by SDS-PAGE and transferred to a nitrocellulose membrane. Membranes were then probed with antiserum or antibody against CdaA or FlaB, respectively. Antiserum/antibodies used to detect the respective proteins are indicated to the right. MW denotes the protein standard, and numbers to the left indicate molecular weight in kDa. Two biological replicates were performed, yielding similar results, and a representative blot from one replicate is shown. occurring between days 4 to 6 and maximum bloodstream burden ranging from 10 6 to 10 8 bacteria/ml of blood ( Fig. 2A and C). Conversely, mice infected with BtDcdaA failed to reach detectable levels of spirochetemia by qPCR on any day postinfection (Fig. 2B).
Because the limit of detection (LOD) with our qPCR is 10 4 bacteria/ml of blood, it was possible that BtDcdaA bacteria are still infectious but failed to reach detectable levels (89). Therefore, we also attempted to culture bacteria out of the bloodstream of infected mice during this experiment, which is predicted to provide an LOD of 4 Â 10 2 live bacteria/ml (see Materials and Methods). As expected, days when BtWT-and BtDcdaA C-cis -infected mice were positive by blood culture largely overlapped with days when the mice were positive by qPCR (data not shown). Alternatively, bacteria could not be cultured from the blood of BtDcdaA-infected mice on any day postinfection (data not shown). As another metric to evaluate infection, mice were also screened for seroconversion at 14 days postinfection against BtWT whole-cell lysates; all BtWT-and BtDcdaA C-cis -infected mice seroconverted, but BtDcdaA-infected mice did not (data not shown). This result suggested that BtDcdaA failed to stimulate a strong humoral immune response because the mutant was cleared early during infection. In all, these murine infection studies support the hypothesis that CdaA is critical for B. turicatae mammalian infection.
The BtDcdaA mutational strategy results in polar mutation effects. Because the 39 end of cdaA is located only 4 bp upstream of the coding region for bt0009 and cdaA is encoded on the same strand as the six genes that are immediately downstream, it was possible that our BtDcdaA mutational approach may result in polar mutation effects. Quantitative reverse transcription-PCR (qRT-PCR) was used to measure expression of cdaA and adjacent genes in in vitro-cultured BtWT, BtDcdaA, and BtDcdaA C-cis Cyclic-di-AMP-Mediated Signaling in Borrelia turicatae Infection and Immunity ( Fig. 3). As expected, cdaA transcript was undetected in BtDcdaA spirochetes, while expression was restored in the BtDcdaA C-cis strain, albeit at reduced levels relative to BtWT (0.55-fold change in expression). Expression of bt0007, the gene immediately upstream of bt0008, was unaffected in BtDcdaA, but expression of the downstream genes bt0009 and bt0010 was reduced approximately 10-fold in the BtDcdaA strain relative to BtWT. Importantly, this polar mutation effect was near-completely reversed upon complementation in the BtDcdaA C-cis strain (mean fold change in expression of 0.93 and 0.50 for bt0009 and bt0010, respectively, relative to BtWT). These results implied that the murine infection defect observed with BtDcdaA could be due to either mutation of cdaA or to polar mutation effects associated with the mutational strategy. Therefore, additional experiments were required to determine if CdaA was essential for mammalian infection.
Addressing polar mutation effects associated with the BtDcdaA mutant. Because the BtDcdaA mutational strategy resulted in polar mutation effects on genes downstream of bt0008 in the chromosome, we were unable to directly attribute the mammalian infection defect in BtDcdaA to inactivation of cdaA. As complementation of BtDcdaA in the BtDcdaA C-cis strain resulted in near-complete reversal of the polar mutational effect, we hypothesized that a similar strategy could be used to introduce a copy of cdaA with a point mutation rendering the encoded protein incapable of DAC activity into BtDcdaA, thereby generating a cdaA mutant without the significant polar mutation effects. This mutant could subsequently be genetically complemented by replacing the point mutant cdaA allele with a wild-type copy to maintain similar expression of adjacent genes in mutant and complemented strains. Rosenburg et al. previously identified a conserved glycine residue in CdaA homologs that can be mutated to an alanine to abrogate DAC activity (90). Using an approach analogous to that employed for generation of BtDcdaA C-cis , the BtDcdaA strain was transformed with a construct to insert the aphI kanamycin resistance cassette and a point mutant copy of cdaA, which contained a single-nucleotide change converting the glycine encoded at position 175 to an alanine (G175A), adjacent to the original mutagenesis site, creating BtDcdaA C-cis(G175A) (Fig. 4A). BtDcdaA C-cis(G175A) was then complemented by replacing the kanamycin resistance marker and G175A-mutated allele of cdaA with an aadA streptomycin resistance cassette and wild-type copy of cdaA, designated BtDcdaA C-cis(G175A) :: cdaA C-cis(WT) (Fig. 4B). Consistent with the presence of cdaA alleles in BtDcdaA C-cis(G175A) and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) , both strains were positive by PCR for cdaA, while the BtDcdaA mutant screened negative (Fig. 4C). As expected, BtDcdaA C-cis(G175A) and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) were positive for the kanamycin and streptomycin resistance markers, respectively, whereas BtDcdaA, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) were all PCR positive for the gentamicin marker. Additionally, PCR for flaB resulted in FIG 3 Polar mutation effects associated with the BtDcdaA mutational strategy. cDNA generated from BtWT, BtDcdaA, and BtDcdaA C-cis RNA was used for qRT-PCR analyses to measure expression of bt0007, cdaA, bt0009, and bt0010. Expression was normalized to flaB, and error bars represent SEM. Shown are results from two biological replicates, and fold change relative to BtWT was calculated using the 2 2DDCT method. ND, not detected.
BtDcdaA C-cis(G175A) is significantly attenuated in a murine needle-challenge model of RF. To assess if the G175A point mutation in cdaA resulted in attenuated mammalian infection, groups of four mice were intradermally inoculated with 10 2 BtDcdaA C-cis(G175A) or BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) bacteria (Fig. 6). While BtDcdaA C-cis(G175A) spirochetes were not detected on any day postinfection in the bloodstream by qPCR (Fig. 6A), BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) -infected mice reached levels of 10 6 to 10 8 bacteria/ml of blood during recurrent peaks (Fig. 6B), which is consistent with BtWT infection ( Fig. 2A). Furthermore, bacteria could not be cultured from the blood of BtDcdaA C-cis(G175A) -infected mice, but BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) spirochetes were cultured from the blood on days postinfection when spirochetemia was detectable by qPCR (data not shown). Finally, all four BtDcdaA C-cis(G175A) -infected mice failed to seroconvert when BtWT lysates were probed with serum collected 14 days postinfection, indicating the bacteria were likely eliminated early during infection. Conversely, all four mice infected with the BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) strain seroconverted (data not shown). These results indicate that DAC activity of CdaA is crucial for B. turicatae mammalian infection and that polar mutation effects were not the sole reason for the attenuation seen with BtDcdaA (Fig. 2).
cdaA mutants contain chromosomal single-nucleotide polymorphisms. Because c-di-AMP signaling is essential for growth in many bacterial systems under standard culture conditions, it was possible that suppressor mutations were required to generate the BtDcdaA mutant (34,40,56,(60)(61)(62)(63). To test this, WGS was performed on two independently generated BtDcdaA clones and the parental BtWT strain. These analyses identified SNPs in both mutants (Table 1). BtDcdaA clone 1 had two missense mutations in the chromosomal genes bt0380 and bt0531, which encode an MgtE family Mg 21 transporter and N-acetylmuramoyl-L-alanine amidase, respectively. MgtE family transporters are widespread in eukaryotes and prokaryotes and primarily transport divalent cations, such as Mg 21 and Co 21 (91,92). N-acetylmuramoyl-L-alanine amidases cDNA generated from BtWT, BtDcdaA, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) RNA was used for qRT-PCR analyses to measure expression of bt0007, cdaA, bt0009, and bt0010. Expression was normalized to flaB, and error bars represent SEM. Shown are results from two biological replicates, and fold change relative to BtWT was calculated using the 2 2DDCT method. ND, not detected.
are ubiquitous bacterial proteins involved in turnover of the cell wall (93). In BtDcdaA clone 2, four chromosomal SNPs were discovered. Missense mutations were identified in bt0219 and bt0747, which encode a ZIP family metal transporter and an oligopeptide transport system permease protein (OppB), respectively. ZIP family metal transporters are found in eukaryotes and prokaryotes, and they generally have broad specificity for divalent cations, including Zn 21 , Fe 21 , and Mn 21 (94,95). OppB serves as a membrane permease that is part of the borrelial oligopeptide transport system (96). Additionally, nonsense mutations were found in bt0127 and bt0241B, which encode the 30S ribosomal protein S1 and a glycerophosphoryl diester phosphodiesterase (GlpQ), respectively. The 30S ribosomal protein S1 is a component of the prokaryotic 30S ribosomal subunit that participates in binding and unfolding of specific mRNA structures and allows for correct ribosomal positioning during translation initiation (97). GlpQ cleaves deacylated phospholipids to glycerol-3-phosphate, which can be used for phospholipid biosynthesis or be shunted into the glycolytic pathway via a dihydroxyacetone phosphate intermediate (98). In all, WGS of BtDcdaA clones identified SNPs in proteins involved in membrane transport, metabolism, and translation, possibly representing compensatory suppressor mutations required for normal physiology during in vitro growth. Of note, BtDcdaA clone 1 was used for initial infection experiments (Fig. 2), as well as for derivation of BtDcdaA C-cis(G175A) and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) clones used herein.
cdaA is required for normal growth and physiology in vitro. The identification of SNPs in two independently generated BtDcdaA clones could imply that suppressor  To determine whether depletion of CdaA resulted in an in vitro physiological defect, BtiCdaA-DcdaA was grown to late exponential phase, washed twice in media to remove IPTG, and then inoculated at an initial density of 10 3 bacteria/ml of modified Barbour-Stoenner-Kelly (mBSK) medium with or without 1 mM IPTG. Growth was then BtiCdaA-DcdaA spirochetes were grown to late exponential phase with 1 mM IPTG, washed twice with mBSK media to remove IPTG, and then inoculated at 10 3 bacteria/ml with (1) or without (2) 1 mM IPTG. Bacteria were then enumerated by dark-field microscopy daily beginning on day 3 postinoculation. Displayed are results from two biological replicates, and error bars represent SEM. *, P , 0.05. (C) Impact of CdaA depletion on morphology. BtiCdaA-DcdaA spirochetes were grown with (1) or without (2) 1 mM IPTG to late exponential phase and examined by dark-field microscopy. One hundred spirochetes from two biological replicates were examined for membrane blebbing, and error bars represent SEM. *, P , 0.05. (D) Imaging of morphological defects associated with depletion of CdaA. Representative bright-field images of BtiCdaA-DcdaA spirochetes grown with (1) or without (2) 1 mM IPTG from the experiment in panel C are displayed. Two biological replicates were examined, and a representative image from one replicate is shown. The scale bar equals 10 mm. (E) Immunoblot analyses of BtiCdaA-DcdaA cultured with (1) or without (2) IPTG. Whole-cell lysates were prepared after completion of growth curve experiments in panel B. Proteins were separated by SDS-PAGE and then transferred to a nitrocellulose membrane. Membranes were then probed with antiserum or antibody for CdaA or FlaB, respectively. Antiserum/antibodies used to detect the respective proteins are indicated to the right. MW denotes the protein standard, and numbers to the left indicate molecular weight in kDa. Two biological replicates were performed, yielding similar results, and a representative blot from one replicate is shown. quantified daily by dark-field microscopy ( Fig. 7B). Interestingly, BtiCdaA-DcdaA grown in the absence of IPTG exhibited a growth defect, with significantly reduced densities on days 7 to 10 postinoculation relative to BtiCdaA-DcdaA cultured with IPTG. BtiCdaA-DcdaA cultured without IPTG also exhibited extensive membrane blebbing visible by bright-field microscopy ( Fig. 7C and D). In fact, 53.5% of BtiCdaA-DcdaA spirochetes grown without IPTG had visual membrane blebbing, while only 10% of BtiCdaA-DcdaA bacteria grown with IPTG had membrane blebs. Finally, to confirm successful depletion of CdaA, immunoblot analyses were performed (Fig. 7E). As expected, CdaA was readily detectable in BtiCdaA-DcdaA cultured with IPTG, but the protein was undetectable in BtiCdaA-DcdaA grown without IPTG. Blots for FlaB were comparable regardless of IPTG treatment. Overall, these results indicated that conditional mutation of cdaA does not necessarily result in cell death in vitro, but CdaA is required for normal bacterial growth and physiology. Importantly, given that the two BtDcdaA mutants used for WGS did not have an in vitro growth defect or any noticeable morphological changes by darkfield microscopy (data not shown), it is likely that the identified SNPs represent suppressor mutations which contribute to the c-di-AMP-responsive physiology of these clones in vitro.
Inactivation of cdaA does not impact bosR or rpoS expression or protein production. In B. burgdorferi, the RpoS alternative sigma factor controls expression of several known virulence determinants and is essential for virulence (101)(102)(103). Borrelia oxidative stress regulator (BosR) in B. burgdorferi binds upstream of rpoS and serves as a transcriptional activator (104)(105)(106)(107). Not surprisingly, BosR is also required for mammalian infection by B. burgdorferi (104,106). Ye et al. demonstrated that conditional depletion of the c-di-AMP PDE, DhhP, in B. burgdorferi resulted in increased levels of cdi-AMP, as well as modestly reduced expression of bosR and significantly reduced expression of rpoS (64). BosR and RpoS protein levels were also markedly reduced, resulting in decreased production of virulence factors (64). This led us to hypothesize that mutation of cdaA could impact BosR and/or RpoS production in B. turicatae, possibly contributing to the observed infection defect. To evaluate this hypothesis, bosR and rpoS transcription (Fig. 8A) and protein production ( Fig. 8B and C) were measured in in vitro cultured BtWT, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) . Interestingly, there were only modest changes in expression of either regulator in BtDcdaA C-cis(G175A) or BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) relative to BtWT (1.12-and 0.81-fold change, respectively, for bosR; 0.94-and 1.03-fold change, respectively, for rpoS). Immunoblot analyses also revealed no substantial changes in BosR or RpoS production between BtWT, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) , and levels of FlaB, which served as a loading control, were equivalent in the three strains. To confirm the specificity of the BosR and RpoS antiserum used in immunoblot analyses, a B. turicatae shuttle vector carrying a lacI cassette and either bosR or rpoS under transcriptional control of a lac-inducible promoter was transformed into BtWT, generating the strains BtiBosR and BtiRpoS, respectively. As expected, when BtiBosR and BtiRpoS were treated with 1 mM IPTG, increases in BosR and RpoS protein levels, respectively, were observed. In all, mutation of cdaA appeared to have no effect on bosR and rpoS transcription or protein production in vitro. This observation strongly suggests that c-di-AMP-mediated gene regulation differs between RF and LD spirochetes.
The cdaA mutant is sensitive to increased salt. Considering the critical role of cdi-AMP signaling in osmoregulation in many bacterial systems, it is not surprising that mutation of c-di-AMP signaling pathway components results in altered susceptibility to salt treatment (42-44, 50, 58, 65, 108). To determine if mutation of cdaA in B. turicatae also results in altered resistance to salt treatment, the maximum concentrations of NaCl and KCl at which BtWT could grow was first determined using previously described MIC plating assays (109,110). BtWT was able to grow in concentrations up to 50 mM for each salt (data not shown), so these concentrations were used in subsequent growth curve analyses with BtWT, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) :: cdaA C-cis(WT) strains ( Fig. 9). In standard mBSK media, BtDcdaA C-cis(G175A) and BtDcdaA C-cis(G175A) :: cdaA C-cis(WT) strains showed modestly increased growth rates relative to BtWT, with , and BtDcdaA C-cis(G175A) :: cdaA C-cis(WT) RNA was used for qRT-PCR analyses to measure expression of bosR and rpoS. Expression was normalized to flaB, and error bars represent SEM. Shown are results from four biological replicates, and fold change relative to BtWT was calculated using the 2 2DDCT method. (B) Effect of cdaA mutation on BosR production. BtiBosR was grown to mid-exponential phase and treated with 1 mM IPTG (1) or left untreated (0 mM IPTG) (2) for 48 h, and BtWT, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) were grown to late exponential phase. Whole-cell lysates were prepared, and proteins were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and probed with antiserum or antibody against BosR or FlaB, respectively. Antiserum/antibodies used to detect the respective proteins are indicated to the right. MW denotes the protein standard, and numbers to the left indicate molecular weight in kDa. Four biological replicates were performed, yielding similar results, and a representative blot from one replicate is shown. (C) Effect of cdaA mutation on RpoS production. BtiRpoS was grown to mid-exponential phase and treated with 1 mM IPTG (1) or left untreated (0 mM IPTG) (2) for 24 h, and BtWT, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) strains were grown to late exponential phase. Whole-cell lysates were prepared, and proteins were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and probed with antiserum or antibody against RpoS or FlaB, respectively. Antiserum/antibodies used to detect the respective proteins are indicated to the right. MW denotes the protein standard, and numbers to the left indicate molecular weight in kDa. Four biological replicates were performed, yielding similar results, and a representative blot from one replicate is shown.
BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) having a significantly increased density on day 5 postinoculation (see Discussion) ( Fig. 9A). With 50 mM NaCl supplementation, all strains grew slower than that seen in mBSK media (Fig. 9B). However, BtDcdaA C-cis(G175A) had a growth defect relative to BtWT, exhibiting significantly lower numbers of bacteria on 6 to 14 days postinoculation, and genetic complementation of BtDcdaA C-cis(G175A) failed to reverse this growth phenotype. Significant growth defects were also observed in BtDcdaA C-cis(G175A) and BtDcdaA C-cis(G175A) :: cdaA C-cis(WT) relative to BtWT when grown with 50 mM KCl; both strains had significantly reduced numbers of bacteria on days 6 to 7 postinoculation (Fig. 9C). The inability to complement the salt-dependent growth defect in BtDcdaA C-cis(G175A) suggests that (i) suppressor mutations, (ii) decreased c-di-AMP levels, or (iii) modestly reduced expression of genes downstream of cdaA in BtDcdaA C-cis(G175A) and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) strains could be responsible for the phenotype. Therefore, more evidence is required to elucidate a possible role for CdaA in survival under conditions of increased osmolarity (see Discussion).
Mutation of cdaA has no effect on growth under nutrient-limited conditions but results in a significant growth defect at decreased osmolarity. Because c-di-AMP signaling and the bacterial stress response have been linked in S. aureus and L. monocytogenes (55)(56)(57), it was possible that CdaA may play a role in survival under nutrient-limited conditions. To assess this hypothesis, growth curve analyses were performed with BtWT, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) strains in either standard mBSK media (Fig. 10A) or in media diluted 1:10 in 1Â phosphate-buffered saline (PBS) to simulate nutrient-limited conditions (Fig. 10B) (111,112). As expected, BtDcdaA C-cis(G175A) and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) strains exhibited an increased growth rate in standard mBSK relative to the BtWT strain (see Fig. 9A). In media diluted 1:10 in 1Â PBS, however, BtDcdaA C-cis(G175A) failed to reach numbers detectable by darkfield microscopy. Alternatively, BtWT and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) strains were able to grow in the dilute media, indicating that CdaA is required for growth under these conditions.
It should be noted that the osmolarity of mBSK medium is 450 mOsm, and the osmolarity of 1Â PBS is 300 mOsm (137 mM NaCl, 2.7 mM KCl, and 11.9 mM phosphates) (113). Therefore, dilution of the media with PBS alters both the nutrient composition and osmolarity of the media. Given the role of c-di-AMP signaling in osmoregulation in other bacteria, a possibility for the observed growth defect in media diluted 1:10 in 1Â PBS could be the decrease in osmolarity (58). To address this alternative, growth curves were performed with BtWT, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) in mBSK diluted 1:10 in 1.25Â PBS (375 mOsm 171.25 mM NaCl, 3.38 mM KCl, and 14.88 mM phosphates)  (Fig. 10D). BtWT and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) grew similarly in mBSK diluted 1:10 in 1.25Â PBS, but BtDcdaA C-cis(G175A) exhibited a growth defect with significantly reduced densities on days 5 to 8 postinoculation. BtDcdaA C-cis(G175A) was not able to grow in mBSK diluted in 1Â PBS, but detectable bacterial densities of BtDcdaA C-cis(G175A) (e.g., maximum of 10 6 bacteria/ml) were observed in mBSK diluted in 1.25Â PBS. Interestingly, in mBSK diluted 1:10 in 1.50Â PBS, which is the same osmolarity as mBSK, growth was similar between all three strains. These data imply that, rather than being required to tolerate nutrient-limited conditions in vitro, CdaA is essential for growth at decreased osmolarity, suggesting a critical role for c-di-AMP in osmoregulation in B. turicatae.
The cdaA mutant requires pyruvate for growth in vitro. c-di-AMP signaling is important for resistance to ROS in other bacterial systems (41,(51)(52)(53). Therefore, experiments were performed to assess if mutation of cdaA in B. turicatae led to altered susceptibility to oxidative stresses. Experiments to measure the sensitivity of Borrelia spirochetes to ROS are typically performed using media lacking pyruvate, as pyruvate can act as an ROS scavenger and decrease the assay sensitivity (89,114,115). When attempting to test the susceptibility of the cdaA mutant to oxidative agents, it was serendipitously found that BtDcdaA C-cis(G175A) failed to reach an adequate density in mBSK medium lacking pyruvate. Following this observation, growth curve analyses were performed to quantify differences in growth of BtWT, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) in mBSK with or without pyruvate (Fig.  11). As expected, in the normal media formulation, BtDcdaA C-cis(G175A) and BtDcdaA C-cis(G175A) :: cdaA C-cis(WT) had modestly increased growth rates relative to the BtWT strain (Fig. 11A). However, BtDcdaA C-cis(G175A) exhibited a growth defect in mBSK lacking pyruvate, reaching significantly lower bacterial densities than BtWT on days 5 to 8 postinoculation (Fig.  11B). Importantly, genetic complementation in BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) reversed this phenotype. Therefore, while the contribution of cdaA to ROS resistance could not be determined using media lacking pyruvate (see Discussion), pyruvate was found to be critical for growth of the cdaA mutant in vitro. Given the critical role of pyruvate in bacterial metabolism, these results possibly link c-di-AMP signaling to an important function in central metabolism in B. turicatae.

DISCUSSION
RF is a significant global public health concern, but no literature exists regarding regulatory pathways required for virulence of the causative Borrelia spirochetes (4,13,14,116,117). Given the essential role of c-di-AMP signaling in bacterial virulence, we sought to assess the function of this dinucleotide second messenger in the TBRF spirochete B. turicatae by inactivating the sole DAC, CdaA (44,45,54,64,(67)(68)(69)(70)(71). In a murine needle-challenge infection model, cdaA mutants were unable to establish bloodstream infection like BtWT (Fig. 2 and 6). Furthermore, mice infected with cdaA mutants failed to seroconvert, indicating that the bacteria were likely eliminated soon after inoculation and before mice could develop a strong B-cell response (118)(119)(120)(121). Importantly, this infection defect was reversed by genetic complementation, indicating that putative suppressor mutations important for in vitro growth in the absence of CdaA (Table 1) and modestly decreased expression of genes downstream of cdaA in BtDcdaA C-cis(G175A) (Fig.  5) were not responsible for the defect. Of note, it is possible that the infection defect exhibited by cdaA mutants could be overcome by increasing the dose, as only 10 2 spirochetes were used for the murine model described herein. However, Boyle et al. demonstrated that only 1 to 10 spirochetes can be transmitted during feeding by Ornithodoros turicata, the vector for B. turicatae, during natural infection (122). We use an inoculum of 10 2 spirochetes because this was the closest dose to natural infection that we could accurately deliver while achieving a 100% infection rate with consistent bacteremic relapses in BtWT-infected mice. Therefore, our dose was approximately 10-to 100-fold higher than could be possibly transmitted by the tick vector, and increasing the inoculum would be further inconsistent with respect to the enzootic cycle of B. turicatae. It is worth noting that infection experiments performed with the BtDcdaA mutant at a dose Cyclic-di-AMP-Mediated Signaling in Borrelia turicatae Infection and Immunity of 10 4 spirochetes (1,000-to 10,000-fold higher than numbers transmitted during natural infection) demonstrated that this strain was still unable to infect mice like BtWT (data not shown). This suggests that the infection defect observed in cdaA mutants could likely not be overcome by increasing the inoculum. Ultimately, our data indicate that CdaA is critical for infection in a biologically relevant needle-inoculation model of infection, but it is still necessary to assess the role of CdaA in transmission from tick to mammal using the experimental murine-tick infection model. Intriguingly, c-di-AMP signaling pathways are absent in many Gram-negative organisms, but they are conserved in the Spirochaetes phylum (78). This conservation, as well as the results reported herein and in B. burgdorferi, may suggest a possible role for the signaling pathway in the virulence of other pathogenic spirochetes, such as Leptospira and Treponema species (64). The initial BtDcdaA mutational strategy, which replaced an internal region of the cdaA ORF with a gentamicin resistance cassette, resulted in approximately 10-fold reduced expression of the downstream genes bt0009 and bt0010. The fact that this polar mutation was almost completely reversed upon genetic complementation (Fig. 3  and 5) could indicate that a promoter region controlling expression of these downstream genes is located within the cdaA ORF. However, in Firmicutes, cdaA is the first gene of a conserved three-gene operon. cdaA is typically followed by genes encoding the cyclic di-AMP synthase A regulator (CdaR) and the phosphoglucosamine mutase GlmM (48,65,(123)(124)(125)(126). In these bacteria, the N-terminal domains anchor both CdaA and CdaR to the membrane, and the C-terminal domains of CdaA and CdaR are located intracellularly and extracellularly, respectively. CdaR interacts with the transmembrane domain of CdaA to regulate DAC activity (48,65,(123)(124)(125). GlmM, an intracellular enzyme involved in cell wall biosynthesis, also directly interacts with CdaA to modulate DAC activity (124)(125)(126). In B. turicatae, the gene immediately downstream of cdaA, bt0009, has been annotated as a YbbR domain-containing protein (a domain found in CdaR proteins of Firmicutes), possibly indicating that BT0009 could function as a CdaR homolog (48,65,(123)(124)(125)(126). However, bt0010 does not encode a GlmM homolog; rather, the gene encodes a holo-(acyl-carrier-protein) synthase, an enzyme involved in fatty acid metabolism (127). RT-PCR analyses revealed that cdaA, bt0009, and bt0010 are transcriptionally linked (data not shown). Additionally, the four genes downstream of bt0010 are encoded on the same strand with no more than 4 bp separating the ORFs, suggesting these genes could also be part of the cdaA-containing operon. These genes encode two hypothetical proteins, tRNA pseudouridine synthase A and primosomal protein N9. Future studies will aim to define the cdaA-containing operon and examine transcriptional regulation at this locus. Additionally, future work will assess if BT0009, BT0010, and other possible proteins encoded in the operon have a function related to c-di-AMP signaling.
In this study, we found that the genomes of two independently generated cdaA mutant clones contained SNPs, which likely represent suppressor mutations that compensate and allow normal growth and physiology in the absence of CdaA in vitro (Table 1 and Fig. 7). The presence of suppressor mutations upon dac mutation is consistent with reports in other bacteria (54,56,61,(128)(129)(130)(131)(132). BtDcdaA clone 1 had SNPs in genes encoding an MgtE family transporter and an N-acetylmuramoyl-L-alanine amidase. Interestingly, an MgtE homolog in Bacillus subtilis binds c-di-AMP via the conserved cystathionine beta-synthase (CBS) domain (130). This could suggest that the identified missense mutation enables appropriate MgtE-mediated regulation of Mg 21 or Co 21 homeostasis in vitro when the protein is unable to bind c-di-AMP. To our knowledge, N-acetylmuramoyl-L-alanine amidase proteins have not been specifically linked to c-di-AMP signaling. However, due to the extensive role of c-di-AMP in cell wall metabolism in other bacteria, the identification of this SNP is not surprising (133). Of note, BtDcdaA clone 1 was used for generation of BtDcdaA C-cis(G175A) , which was subsequently complemented to make BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) . All three of these strains exhibit a significant increase in growth rates relative to BtWT, while BtDcdaA clone 2 exhibits no difference in growth (Fig. 9A, Fig. 10A, and Fig. 11A; data not shown). This observation suggests that a SNP or combination of SNPs in BtDcdaA clone 1 may be responsible for the increased growth rate. Interestingly, mutation of mgtE in the bacterium Shewanella oneidensis results in an increased growth rate in the presence of certain divalent metals (134). Future examination of the growth of BtDcdaA clone 1 in different media formulations, as well as generation and phenotypic analysis of a B. turicatae mutant containing the point mutation in mgtE, will address the possibility that this SNP is responsible for the altered growth kinetics.
BtDcdaA clone 2 had SNPs in genes encoding the 30S ribosomal protein S1, a ZIP family metal transporter, a glycerophosphoryl diester phosphodiesterase (GlpQ), and an oligopeptide transport system permease protein (OppB). Interestingly, suppressor mutations in genes encoding subunits of Opp oligopeptide transport systems have also been noted upon dac mutation in L. monocytogenes and S. agalactiae (54,56,129). Given that oligopeptides can serve as osmolytes, this conserved relationship between c-di-AMP signaling and oligopeptide transport could be related to bacterial osmoregulation (54,135). Alternatively, this relationship could suggest roles for c-di-AMP in amino acid metabolism. Further studies are required to examine these possibilities. To our knowledge, suppressor mutations in genes encoding ZIP family transporters, glycerophosphoryl diester phosphodiesterase proteins, or ribosomal proteins have not been reported upon dac mutation in other bacteria. As MgtE and ZIP family transporters both transport diverse divalent cations, the mutation in the gene encoding the ZIP family protein in BtDcdaA clone 2 could serve an analogous compensatory function to the mutation in mgtE in BtDcdaA clone 1 in the absence of CdaA (91,92,94,95). The potential suppressor mutation in glpQ may support a role for c-di-AMP in lipid metabolism or glycolysis, and the growth defect in the cdaA mutant in the absence of pyruvate (Fig. 11) may further implicate c-di-AMP in control of the glycolytic pathway (discussed below) (98). Finally, the nonsense mutation in the gene encoding the 30S ribosomal protein S1 may suggest a role for c-di-AMP in translation initiation. Of note, all attempts to transform BtDcdaA clone 2 with pUAMS351B to generate a second DcdaA C-cis(G175A) clone failed to result in insertion of the aphI resistance marker and mutated cdaA allele. However, kanamycin-resistant bacteria were readily isolated. Subsequent MIC plating assays revealed that BtDcdaA clone 2 was hyperresistant to kanamycin, with an MIC .300 mg/ml (data not shown). This observation implicates one or more of these suppressor mutations in resistance to kanamycin. Interestingly, mutation of Opp family proteins has been implicated in increased resistance to aminoglycoside antibiotics, as Opps can facilitate antibiotic permeation of the Gram-negative inner membrane (136)(137)(138). Further investigation is required to determine if the SNP in oppB results in kanamycin hyperresistance. In all, WGS of BtDcdaA clones identified SNPs in genes encoding proteins involved in membrane transport, metabolism, and translation. Additional studies are required to elucidate whether these individual mutations in BtDcdaA clones 1 and 2 can reverse the physiological and growth defects seen upon depletion of CdaA (Fig. 7). Future work will also seek to identify if c-di-AMP regulates expression or activity of these genes or gene products, respectively.
During complementation, we confirmed that the BtDcdaA C-cis(G175A) mutant lacked c-di-AMP, but approximately 2-fold less c-di-AMP was measured in the BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) -complemented strain than BtWT (Fig. 4D). This observation could be due to the modest reduction in cdaA expression in BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) relative to BtWT (mean, 0.54-fold change) (Fig. 5), which correlated with the measured fold difference in c-di-AMP production. However, Savage et al. reported that significantly altering production of CdaA in B. burgdorferi had little effect on c-di-AMP levels, indicating that DAC activity is tightly controlled through an unknown mechanism (78). It is likely, given the critical role of c-di-AMP in B. turicatae, that RF spirochetes also tightly regulate c-di-AMP production. This observation leads to two alternative possibilities for decreased c-di-AMP levels in BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) . First, modestly reduced expression of bt0009 is observed in the complemented strain relative to BtWT (0.71-fold change). Heterologous co-overexpression of the cdaA and cdaR homologs of B. subtilis in Escherichia coli results in significantly increased c-di-AMP levels relative to expression of cdaA alone, indicating that CdaR could serve as a positive regulator of CdaA DAC activity (123). Therefore, it is logical that reduced bt0009 expression in the complemented strain may lead to decreased c-di-AMP levels, assuming that BT0009 acts as a CdaR homolog. A second, and more intriguing, explanation for reduced c-di-AMP levels in BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) could be due to the acquisition of additional suppressor mutations. As mutations are required for normal physiology and growth in the absence of CdaA (Fig. 7), it is possible that restoration of CdaA in a mutant adapted for growth in the absence of c-di-AMP could also lead to detrimental effects. It is therefore possible that additional suppressor mutations were required for generation of the complemented mutant. These suppressor mutations could lie within genes encoding proteins which regulate CdaA activity either directly or indirectly, thus leading to altered DAC activity. Although outside the scope of the current manuscript, future studies will evaluate the possibility of additional suppressor mutations in BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) . Importantly though, despite the inability to fully restore c-di-AMP levels in BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) , all phenotypes associated with cdaA mutation, aside from the growth defect upon treatment with salts, were reversed using our complementation approach.
Although c-di-AMP regulates global gene expression in B. burgdorferi through control of bosR and rpoS expression, mutation of cdaA had no discernible effect on expression or protein production of these regulatory orthologs in B. turicatae (Fig. 8) (64). In B. burgdorferi, expression of numerous virulence-associated genes is controlled via the RpoN-RpoS alternative sigma factor pathway (101-103, 139, 140). The pathway(s) regulating virulence gene expression in RF spirochetes remains to be determined, but B. turicatae and other New World RF spirochetes have a point mutation in the rpoN gene, which significantly truncates the protein (141). Therefore, RF spirochetes may have evolved a divergent signaling pathway to control expression of virulence genes, and it is possible that c-di-AMP impacts this pathway. It should be noted, however, that, while depletion of DhhP in B. burgdorferi led to changes in the proteome visible by SDS-PAGE and Coomassie brilliant blue (CBB) staining, no changes were visible via these analyses upon mutation of cdaA in B. turicatae (data not shown) (64). While this observation could imply that c-di-AMP does not impact global expression in B. turicatae, it is also possible that c-di-AMP impacts expression of genes which encode proteins not produced at high enough levels to be detected by CBB staining. Therefore, future studies will include transcriptomic and proteomic approaches to measure the impact of cdaA mutation in B. turicatae on global gene expression and protein production, respectively.
The most well-characterized function of c-di-AMP in bacteria is osmoregulation (58). Herein, we found two lines of evidence supporting a role for c-di-AMP in osmoregulation in B. turicatae. First, the cdaA mutant had a significant growth defect when cultured in the presence of NaCl and KCl (Fig. 9). Interestingly, genetic complementation failed to reverse this phenotype, which implicates the identified SNPs, reduced levels of c-di-AMP, and/or modestly decreased expression of genes downstream of cdaA in BtDcdaA C-cis(G175A) and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) as possible contributors for the growth defect. While these caveats do not allow us to directly link CdaA to survival during salt treatment, the possibility that a suppressor mutation required for normal growth in the absence of CdaA impacts survival during salt treatment could indirectly link c-di-AMP signaling to osmoregulation. Interestingly, BtDcdaA clone 1 (as well as strains derived from this clone) and BtDcdaA clone 2 harbor completely different SNPs, but they exhibited similar growth defects with NaCl and KCl treatment (data not shown). These data could imply that the unique combinations of suppressor mutations in these clones may result in similar, but each with narrower ranges of osmolarity resistance, with the clones being the most well adapted to the osmolarity of mBSK media, 450 mOsm (113). More evidence is required, however, to address this possibility. The second line of evidence supporting a role for c-di-AMP in osmoregulation in B. turicatae was the growth phenotype of the cdaA mutant at decreased osmolarity (Fig. 10). The mutant was unable to grow at an osmolarity of 300 mOsm, and, importantly, genetic complementation reversed this growth defect. Growth at 375 mOsm partially rescued this phenotype, and no growth defect was noted at 450 mOsm (the osmolarity of mBSK). Of note, the osmolarity of mammalian blood is 300 mOsm (113). Given that the cdaA mutant is unable to grow at 300 mOsm in vitro or establish bloodstream infection (Fig. 6), it is tempting to speculate that the infection phenotype is due to this defect in osmoregulation. However, given that c-di-AMP plays roles in other virulence-associated bacterial phenotypes, defects in osmoregulation could represent just one of several reasons why the cdaA mutant is significantly attenuated during murine infection (44,46,64,69,(71)(72)(73)(74)(75)(76).
In this study, we serendipitously discovered that CdaA is required for normal growth of B. turicatae in mBSK medium lacking pyruvate (Fig. 11), which suggests a role for c-di-AMP signaling in central carbon metabolism, consistent with reports in other bacteria (41,54,142,143). However, these previous studies have revealed only one direct role for c-di-AMP in regulation of central metabolism, which is inhibition of pyruvate carboxylase activity (142,143). Interestingly, Borrelia spirochetes do not encode a pyruvate carboxylase homolog or any protein involved in the tricarboxylic acid (TCA) cycle (77,144,145). In fact, the only known fate of pyruvate in Borrelia following glycolysis is conversion to lactate by the lactate dehydrogenase enzyme (144). This reaction is one of only three in Borrelia, along with the reduction of the disulfide form of coenzyme A (CoA) by CoA disulfide reductase and the nicotinamidase reaction in the nicotinamide salvage pathway that is capable of regenerating NAD 1 from NADH (144,146,147). Importantly, glycolysis has an absolute requirement for the cofactor NAD 1 (148). Therefore, the growth defect observed in media lacking pyruvate could imply that the cdaA mutant is unable to generate sufficient NAD 1 levels for continued glycolysis and that, by increasing pyruvate levels, the NAD 1 /NADH ratio can be restored to a value necessary for survival. Future studies are required to elucidate the exact role of c-di-AMP in central carbon metabolism in B. turicatae.
c-di-AMP-dependent signaling is important for resistance to oxidative stresses in other bacterial systems (41,(51)(52)(53). While we were unable to assess the role of CdaA in resistance to oxidative stresses using media lacking pyruvate, no significant growth defect was observed in the cdaA mutant in complete mBSK medium supplemented with oxidizing agents using MIC plating assays or growth curve analyses (data not shown) (89,109,110,115). Another phenotype associated with deletion of c-di-AMP signaling pathway components in other bacteria is altered resistance to b-lactam antibiotics, presumably due to altered cell wall metabolism (38,(40)(41)(42)(43)(44)(45)(46)(47)(48)(49)(50). However, using MIC plating assays, we failed to detect any significant changes in susceptibility of the cdaA mutant to this class of antimicrobials (data not shown). Finally, c-di-AMP has been linked to survival during heat shock and in low pH, but no defect was observed when growing the cdaA mutant in either of these conditions (data not shown) (37)(38)(39)43). The absence of several phenotypes commonly associated with altered c-di-AMP levels in other bacteria is not entirely unexpected since several of these works utilized Gram-positive bacteria (primarily Firmicutes). As such, it is important to examine this pathway in an evolutionarily divergent Gram-negative bacterium, as these studies could reveal key functional differences.
In summary, this work represents the first characterization of the DAC, CdaA, in the context of any Borrelia spirochete. Importantly, we found that cdaA mutants were significantly attenuated in a murine needle-challenge model of RF. Hypothesizing that CdaA could be important for in vitro viability, WGS was performed on two mutant clones, which identified SNPs in genes involved in membrane transport, metabolism, and translation, which represent potential suppressor mutations. We also found that depletion of CdaA resulted in a significant growth defect and extensive membrane blebbing. Although c-di-AMP signaling impacts global regulation through control of expression of the virulence regulators bosR and rpoS in B. burgdorferi, mutation of cdaA in B. turicatae had little effect on their respective transcript or protein levels (64). This observation implied that c-di-AMP may impact gene regulation differently in RF and LD spirochetes. Finally, in vitro characterization of the cdaA mutant revealed significant growth defects with salt treatment, at decreased osmolarity, and in media lacking pyruvate. While the salt treatment phenotype was not reversed after genetic complementation, growth defects at decreased osmolarity and in media lacking pyruvate could be directly attributed to the cdaA mutation. These latter observations imply a role for c-di-AMP signaling in osmoregulation and central metabolism in B. turicatae. Future studies will aim to elucidate the mechanism(s) through which c-di-AMP impacts B. turicatae virulence, osmoregulation, and central metabolism by examining the effects of c-di-AMP on global gene regulation, identifying effectors of c-di-AMP, and determining the effect of c-di-AMP binding on these individual effectors. Furthermore, future studies will elucidate how the c-di-AMP signaling pathway is regulated. Lastly, future work will determine if c-di-AMP plays a role in the vector phase of the enzootic cycle.

MATERIALS AND METHODS
Bacterial strains and culture conditions. Bacterial strains used in this study are listed in Table 2. E. coli strain TOP10F9 (Life Technologies, Carlsbad, CA) was used for cloning and plasmid propagation, and E. coli strain C41(DE3) (Lucigen, Middleton, WI) was used for expression of recombinant protein. E. coli was cultured at 37°C in Luria-Bertani (LB) medium supplemented with 100mg/ml ampicillin, 5mg/ml gentamicin, 50mg/ml kanamycin, 100mg/ml spectinomycin, or 30mg/ml chloramphenicol when appropriate. Low-passage B. turicatae strain 91E135 (Oz1), designated BtWT, was used in this study (149,150). B. turicatae strains were passaged no more than twice beyond the original frozen stock. B. turicatae was cultured at 35°C with 3% CO 2 in modified Barbour-Stoenner-Kelly (mBSK) medium with 12% rabbit serum at pH 7.6 unless noted otherwise (151,152). mBSK was supplemented with 40mg/ml gentamicin, 150mg/ml kanamycin, or 150mg/ml streptomycin when appropriate.
Generation of constructs used in this study. Plasmids and primers used in this study are detailed in Tables 2 and 3, respectively. PrimeStar Max DNA polymerase (TaKaRa Bio, Mountain View, CA) was used for PCR, and all amplicons were TA cloned into pGEM-T Easy (Promega Corp., Fitchburg, WI) and confirmed by Sanger sequencing. For expression of recombinant protein, the full bosR ORF (primers 59 BtBosR ORF BamHI and 39 BtBosR ORF SpeI), amino acids 120 to 131 of the rpoS ORF (primers 59 BtRpoS-trun_BamHI and 39 BtRpoS-trun_SpeI), and the cdaA ORF lacking the N-terminal region encoding the predicted signal sequence and transmembrane domains (amino acids 1 to 90; primers 59 bt0008 w/o TM BamHI and 39 bt0008 ORF BamHI) were amplified from genomic BtWT DNA (gDNA). The putative transmembrane domains of CdaA were predicted using TMHMM-2.0 (85,86). cdaA, rpoS, and bosR fragments were excised with their respective restriction enzymes and ligated into linearized pProEX-HTb (Life Technologies) to generate pUAMS338, pUAMS233, and pUAMS159, respectively.
Allelic exchange mutagenesis was used for generation of the BtDcdaA mutant and subsequent complementation. To make the DcdaA mutational construct, 59 (primers 59 F1 bt0008 KO and 39 F1 bt0008 KO_AscI) and 39 (primers 59 F2 bt0008 KO_AscI and 39 F2 bt0008 KO_BssHII) flanking regions were amplified from BtWT gDNA. These flanking regions were then ligated together with the B. turicatae-adapted gentamicin resistance cassette from pUAMS4 between them, generating the DcdaA construct pUAMS248B (89). In the final mutational construct, the aacC1 marker replaces a 419-bp internal region between nucleotides 86 and 504 of the bt0008 ORF. For complementation, a B. turicatae-adapted kanamycin resistance cassette was made by amplifying the putative promoter region for the flaB gene from BtWT gDNA (primers 59 BtFlaB-aphI-BamHI and 39 Kan ORF int -NdeI v.2) and the Tn903-derived aminoglycoside phosphotransferase ORF (aphI) from pBSV2 (primers 59 Kan ORF int -NdeI v.2 and 39 BtFlaB-aphI-BamHI) (153). These amplicons were then fused together by overlap extension PCR, and the resulting product was ligated into pGEM-T Easy, generating the plasmid pUAMS309. To create the DcdaA C-cis complementation construct, a segment encoding the 59 flanking region and disrupted cdaA ORF (primers 59 F1 bt0008 KO and 59 BtflgB-BamHI) was amplified from pUAMS248B, and a segment encoding the intact cdaA ORF with its adjacent upstream region, as well as a downstream segment to facilitate recombination (primers 59 F2 bt0008 Comp-BamHI and 39 F2 bt0008 Comp-BglII), was amplified from BtWT gDNA. These segments were then ligated together with the PflaB-aphI resistance cassette from pUAMS309 between them, yielding the final DcdaA C-cis construct, pUAMS313B.
To complement the BtDcdaA mutant with a mutated copy of cdaA containing a point mutation converting a glycine at position 175 to an alanine, an approach similar to the BtDcdaA C-cis complementation strategy was used. To introduce the G175A point mutation into cdaA, overlap extension PCR was performed with pUAMS313B serving as the template (primer pairs 59 DAC/cis bt0008 and 39 bt0008 G175A; 59 bt0008 G175A and 39 DAC/cis bt0008). The full-length product was then ligated into pUAMS313B, replacing the analogous region and yielding the DcdaA C-cis(G175A) construct, pUAMS351B.
For complementation of BtDcdaA C-cis(G175A) , a B. turicatae-adapted aadA streptomycin resistance cassette was first made by amplifying the putative promoter region for the flaB gene from BtWT gDNA (primers 59 BtPflaB_BamHI and 39 PflaB NdeI) and the aminoglycoside adenylyltransferase ORF (aadA) from pJD7 (primers 59 BtflaB-aadA junc_NdeI and 39 aadA ORF_BamHI) (154). The promoter and aadA ORF were fused by ligating the two amplicons together using the common NdeI restriction site. The final ligated PflaB-aadA cassette in pGEM-T Easy was designated pUAMS402. The PflaB-aadA cassette was subsequently ligated into pUAMS313B in place of the PflaB-aphI cassette using the BamHI sites, generating the final DcdaA C-cis(G175A) ::cdaA C-cis(WT) construct pUAMS406B.
To confirm BtDcdaA mutant and BtDcdaA C-cis clones, PCRs were performed to amplify a region within the replaced segment of cdaA (primers 59 bt0008 int diag and 39 bt0008 int diag), an internal segment of aacC1 (primers 59 aacC1 diag and 39 aacC1 diag), an internal region of aphI (primers 59 aphI diag and 39 aphI diag), and an internal segment of flaB as an amplification control (primers 59 BtFlaB and 39 BtFlaB). To confirm complementation of the BtDcdaA strain with the mutated cdaA ORF [BtDcdaA C-cis(G175A) ] and subsequent complementation with the wild-type copy of cdaA [BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) ], PCRs were conducted to amplify an internal segment of cdaA (see above), an internal segment of aphI (see above), an internal segment of aadA (primers 59 aadA diag and 39 aadA diag), and an internal segment of flaB (see above). A region of cdaA containing the point mutation was additionally amplified (primers 59 bt0008 int diag and 39 bt0008 ext diag), gel purified, and Sanger sequenced to confirm the G175A mutation.
The cdaA conditional mutant (BtiCdaA-DcdaA) was generated similar to previously described methods in B. burgdorferi (99,100). BtWT was first transformed with pUAMS346 to generate BtiCdaA. A confirmed BtiCdaA clone was then transformed with the DcdaA mutational construct pUAMS248B, and bacteria were recovered after electroporation in the presence of 1 mM IPTG to maintain expression of cdaA in spirochetes that acquired the mutation. Following the 24-h recovery period, bacteria were passed into media containing kanamycin, gentamicin, and 1 mM IPTG to select for bacteria with the DcdaA mutation. BtiCdaA-DcdaA clones were then isolated using serial dilution plating (156). The conditional mutant was confirmed to harbor pUAMS346 by plasmid recovery in E. coli, and the mutation was confirmed by differential PCR using primers flanking the mutation (primers 59 bt0008 ext diag and 39 bt0008 ext diag), as well as PCRs for internal segments of the aacC1 marker (see above) and flaB gene (see above). Expression and purification of recombinant protein. Expression from pProEX-HTb generates recombinant protein with an N-terminal His 6 tag and tobacco etch virus (TEV) cleavage site. pUAMS338 was transformed into C41(DE3) E. coli, and expression of recombinant CdaA lacking the transmembrane domains was achieved by induction with 1 mM IPTG for 3 h at 37°C. For expression of recombinant truncated RpoS, pUAMS233 was also transformed into C41(DE3) E. coli, and expression was induced with 1 mM IPTG for 3 h at 37°C. For expression of recombinant BosR, pUAMS159 was cotransformed into C41 (DE3) E. coli with pRARE (Novagen, Madison, WI), a plasmid isolated from RosettaBlue(DE3) cells that encodes tRNAs to express genes with rare E. coli codons (157). Expression of recombinant BosR was then achieved by induction with 1 mM IPTG for 3 h at 37°C. Recombinant CdaA, RpoS, and BosR were purified using HisPur nickel-nitrilotriacetic acid (Ni-NTA) resin (Thermo Fisher Scientific) under nonnative conditions. Briefly, cells were lysed using BugBuster protein extraction reagent (MilliporeSigma, Burlington, MA) in conjunction with Lysonase (MilliporeSigma) per the manufacturer's instructions. Following cell lysis, soluble and insoluble fractions were separated by centrifugation at 24,000 Â g for 15 min. Inclusion bodies were then solubilized in resin wash buffer (20 mM Tris, 20 mM NaCl, 5% glycerol, pH 7.5) supplemented with 0.3% N-lauryl-sarcosine. Ni-NTA resin was washed two times in this same buffer prior to combining with the solubilized inclusion body. Binding was then performed using an end-over-end rotator for .30 min. Following binding, the resin was washed in 10 bed volumes of buffer A (20 mM Tris, 20 mM NaCl, 20 mM imidazole, and 5% glycerol, pH 7.5) supplemented with 0.3% N-lauryl-sarcosine, followed by a wash in 10 resin bed volumes of buffer B (20 mM Tris, 1 M NaCl, and 5% glycerol, pH 7.5) supplemented with 0.3% N-lauryl-sarcosine. A final wash was then performed in 10 resin bed volumes of buffer A with 0.3% N-lauryl-sarcosine. Finally, the protein was eluted from the resin using buffer C (20 mM Tris, 200 mM NaCl, 250 mM imidazole, and 5% glycerol, pH 7.5) with 0.3% N-lauryl-sarcosine. Following purification, concentrations of purified recombinant proteins were calculated using the DC protein assay kit (Bio-Rad Laboratories, Hercules, CA).
Generation of CdaA-, RpoS-, and BosR-specific antiserum. Rat immunizations were performed in accordance with the recommendations of the Public Health Science (PHS) Policy on Humane Care and Use of Laboratory Animals, the Guide for the Care and Use of Laboratory Animals, and the Animal Welfare Act, and the utilized protocol was approved by the University of Arkansas for Medical Sciences (UAMS) Institutional Animal Care and Use Committee (IACUC). Generation of CdaA-, RpoS-, and BosRspecific rat antiserum was performed as previously described (158) SDS-PAGE and immunoblotting. Immunoblot analyses were performed as previously described (159). Briefly, whole-cell lysates were prepared from B. turicatae cultures grown to late exponential phase. A volume of lysate equivalent to 2 Â 10 7 spirochetes was then separated by SDS-PAGE and transferred to a nitrocellulose membrane. Membranes were subsequently probed with rat antiserum recognizing CdaA, BosR, or RpoS (see above). Horseradish peroxidase (HRP)-conjugated goat anti-rat IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) was used as a secondary antibody. As a positive control, membranes were additionally probed for FlaB using chicken anti-B. burgdorferi FlaB IgY as the primary antibody and HRP-conjugated donkey anti-chicken IgY (Jackson ImmunoResearch Laboratories) as a secondary antibody (159). To assess seroconversion of mice from murine infection experiments (see below), BtWT lysates were probed with serum from infected or naive mice, and HRPconjugated goat anti-mouse IgG (Jackson ImmunoResearch Laboratories) served as the secondary antibody (data not shown). For all blots, colorimetric detection was achieved using 4-chloro-1-naphthol as a substrate, and Precision Plus Protein all blue prestained protein standard (Bio-Rad Laboratories) served as the molecular weight standard.
Murine infection experiments. Murine infection experiments were performed in accordance with the recommendations of the PHS Policy on Humane Care and Use of Laboratory Animals, the Guide for the Care and Use of Laboratory Animals, and the Animal Welfare Act, and the utilized protocol was approved by the UAMS IACUC. Infections were performed as previously described (89). Briefly, 4-to 6week-old female Swiss Webster mice (Charles River Laboratories) were intradermally/subcutaneously injected with 10 2 bacteria in the thoracic region. On days 3 to 14 postinfection, 2.5 ml of blood was taken by tail vein venipuncture, combined with 47.5 ml of SideStep lysis and stabilization buffer (Agilent Technologies, Santa Clara, CA), and stored at 280°C until qPCR analyses were performed to quantify bacteremia (see below). An additional 2.5 ml of blood was then collected and added to mBSK medium supplemented with Borrelia antibiotic mixture (BAM) (Monserate, San Diego, CA) to assess presence of live bacteria in the bloodstream. On day 14, mice were euthanized, blood was collected by brachial artery bleed, and serum was isolated for immunoblot analyses for seroconversion. Two weeks following collection of daily blood samples, cultures were assessed for presence of spirochetes; 10 fields of view were scanned by dark-field microscopy, and presence of one or more spirochetes was considered culture positive. Given the volume of blood taken during the murine infection experiments, the LOD for this analysis is equivalent to one spirochete/2.5 ml of blood (4 Â 10 2 bacteria/ml). For use in qPCR standards, naive mice were euthanized, and blood was collected by brachial artery bleed and added to lysis and stabilization buffer at a blood-to-buffer ratio of 1:18. Blood/buffer mixtures from naive mice were stored at 280°C until standard preparation (see below). qPCR for bacterial burdens. qPCR analyses to detect spirochetemia in mice were performed as previously described (89). Briefly, in a 96-well real-time PCR plate, 3ml of blood in lysis and stabilization buffer (see above) was added in a 20-ml reaction mixture containing 10ml of SsoAdvanced universal probes supermix (Bio-Rad Laboratories) and final concentrations of 400 nM and 300 nM for primers (BtflaB F and BtflaB R) and probe (BtflaB-Probe), respectively. For qPCR standards, late-exponential BtWT cultures were pelleted, washed twice in PBS supplemented with 5 mM MgCl 2 (PBS-MgCl 2 ), and then resuspended in PBS-MgCl 2 and quantified by darkfield microscopy. This suspension was then used to make serial dilutions in PBS-MgCl 2 from 10 4 to 10 8 bacteria/ ml. For use in the no-template control (NTC), nuclease-free water was diluted 10-fold in PBS-MgCl 2 . These dilutions were then added to naive blood in lysis and stabilization buffer (see above) at a 1:19 ratio, and 3ml of these preparations were added in a 20-ml reaction mixture, as described for murine blood samples above, to generate a standard curve. qPCRs for all samples and standards were performed in triplicate. The QuantStudio 6 Flex realtime PCR system (Thermo Fisher Scientific) was used for real-time PCR, and the reaction conditions consisted of an initial 2 min, 50°C hold followed by a 10 min, 95°C polymerase activation step. Forty cycles of DNA denaturation at 95°C for 15 s and primer annealing/DNA extension at 60°C for 60 s were then performed for DNA amplification. Data were then imported into GraphPad Prism version 8 (GraphPad Software, San Diego, CA) for analysis.
qRT-PCR analyses. RNA extraction, cDNA synthesis, and qRT-PCR were performed as previously described (89,154,158). Briefly, B. turicatae cultures were grown to late exponential growth phase followed by addition of 10% RNA stop solution (154,160). Bacteria were then collected by centrifugation and stored at 280°C until RNA isolation was performed. RNA was isolated by TRIzol extraction (Thermo Fisher Scientific), followed by purification with the RNeasy minikit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Treatment with RNase-free DNase I (Qiagen) was then performed to degrade possible contaminating DNA, and absence of B. turicatae gDNA was confirmed using PCR to amplify an internal segment of the flaB gene (see above). The iScript cDNA synthesis kit (Bio-Rad Laboratories) was then used to reverse transcribe purified RNA into cDNA via the manufacturer's protocol. As a negative control, mock reactions were performed in the absence of reverse transcriptase. Successful cDNA synthesis was then confirmed by PCR amplification of an internal segment of the flaB gene (see above).
qRT-PCR was used to measure expression of bt0007 (primers S-bt0007_IDT-SYBR and AS-bt0007_IDT-SYBR), cdaA (primers S-bt0008_IDT-SYBR_V2 and AS-bt0008_IDT-SYBR_V2), bt0009 (primers S-bt0009_IDT-SYBR and AS-bt0009_IDT-SYBR), bt0010 (primers 2_bt0010_IDT-SYBR FWD and 2_bt0010_IDT-SYBR REV), bosR (primers bt0647 SYBR/IDT FWD Set 1 and bt0647 SYBR/IDT REV Set 1), rpoS (primers BtrpoS SYBR FWD and BtrpoS SYBR REV), and flaB (primers 59 BtflaB SYBR/ABI and 39 BtflaB SYBR/ABI). SYBR reactions were performed with SsoAdvanced universal SYBR green supermix (Bio-Rad Laboratories) per the manufacturer's instructions with 100 ng of cDNA serving as the template, and all reactions were performed in triplicate. The QuantStudio 6 Flex real-time PCR system was used for real-time PCR with reaction conditions consisting of an initial polymerase activation step at 95°C for 30 s and 40 cycles of DNA denaturation at 95°C for 10 s and primer annealing/DNA extension at 60°C for 30 s. Results were imported into GraphPad Prism version 8 for analysis.
Whole-genome sequencing and SNP analysis. For WGS analyses, B. turicatae strains of interest were grown to late exponential growth phase, and DNA was purified using the DNeasy blood and tissue kit (Qiagen) via the manufacturer's instructions. gDNA quality was then analyzed using the TapeStation 2200 system (Agilent Technologies), and only samples with DNA integrity numbers (DINs) above 8 were utilized. The Nextera XT DNA library preparation kit (Illumina Inc., San Diego, CA) was used for library preparation, and the MiSeq system (Illumina Inc.) was used for next-generation sequencing using 500-cycle, 2 Â 250-bp chemistry. The paired-end FASTQ files generated were processed using fastp (v0.20.0) for quality filtering (Phred score $20), adapter trimming, and read correction (161). All sequencing data sets had a coverage of approximately Â400 to Â500 postfiltering and correction. Variant analysis was performed using an "all-in-one" variant calling pipeline, Snippy (v4.6.0), using default parameters (-mapqual 60, -basequal 13, -mincov 10, -minqual 100) with -cpus 30 (162). The updated, complete B. turicatae 91E135 genome generated by the Job Lopez laboratory at Baylor College of Medicine was used as the reference (J. E. Lopez and A. R. Kneubehl, unpublished data). BtWT was used to assess any differences between the B. turicatae 91E135 from our lab and what was sequenced by the Lopez laboratory. This analysis yielded only eight SNPs/variants between BtWT and the Lopez B. turicatae 91E135 genome. Variants were then assessed in BtDcdaA clones 1 and 2 using the same reference genome. The variants reported herein were those found to be different from the BtWT variants, indicating differences between the mutant clones and their parental cell line compared to the reference genome. The variant analyses were performed on a System76 Thelio Massive Linux machine with an Intel Xeon Gold 6230 processor and 126 Gb ECC DDR4 2933 MHz RAM (System76, Denver, CO). All SNPs were confirmed by amplifying a segment containing the SNP, gel purifying the product, and Sanger sequencing. The following primer pairs were used for amplification of the regions containing the SNPs in bt0380, bt0531, bt0127, bt0219, bt0241B, and bt0747: 59 BT0380 SNP Diag and 39 BT0380 SNP Diag, 59 BT0531 SNP Diag and 39 BT0531 SNP Diag, 59 BT0127 SNP Diag and 39 BT0127 SNP Diag, 59 BT0219 SNP Diag and 39 BT0219 SNP Diag, 59 BT0241B SNP Diag and 39 BT0241B SNP Diag, and 59 BT0747 SNP Diag and 39 BT0747 SNP Diag, respectively.
Quantification of c-di-AMP. For quantification of c-di-AMP, the B. turicatae strains of interest were grown to late exponential growth phase, and 10 10 cells were pelleted and washed two times in filtered saline. Following the last wash, cells were resuspended in 50 ml of Milli-Q H 2 O, heat inactivated for 10 min at 99°C, and stored at 280°C until c-di-AMP quantification was performed. Cell resuspensions were mixed with 25 mM heavy (C 13 N 15 ) c-di-AMP, followed by 500 ml of methanol, and sonicated (20 s, 80% amplitude). Methanol was collected following centrifugation of cell lysates. The remaining cell lysates were then resuspended in 50 ml Milli-Q H 2 O, mixed with 500 ml methanol, and sonicated (10 s, 80% amplitude) again. The second methanol fraction was pooled with the first following centrifugation. Fractions were dried in a speed vacuum concentrator, and resultant pellets were resuspended in 30 ml Milli-Q H 2 O for mass spectrometry analysis. Quantification of c-di-AMP was based on the 659/136 mass transition (689/146 for the internal standard) as previously described (68).
Growth curve analyses. For in vitro growth curve analyses of BtWT, BtDcdaA C-cis(G175A) , and BtDcdaA C-cis(G175A) ::cdaA C-cis(WT) , cultures were inoculated at a density of 10 4 spirochetes/ml in normal mBSK medium or in the following conditions: (i) mBSK medium supplemented with 50 mM NaCl or 50 mM KCl; (ii) mBSK medium diluted 1:10 in 1Â PBS, 1.25Â PBS, or 1.50Â PBS; or (iii) mBSK medium prepared without pyruvate. For in vitro growth curve analysis of BtiCdaA-DcdaA, an initial 1-ml starter culture was grown to late exponential growth phase in mBSK medium supplemented with 1 mM IPTG, followed by centrifugation at 5,283 Â g for 5 min and resuspension in 1 ml mBSK without IPTG supplementation. This wash step was repeated twice to remove any remaining IPTG from the culture. Following the washes, BtiCdaA-DcdaA was inoculated into media with or without 1 mM IPTG at an initial density of 10 3 bacteria/ml. On the indicated days postinoculation, bacteria were quantified by dark-field microscopy. Ten microscopic fields were counted for each biological replicate, and two biological replicates for each condition were analyzed.
Imaging of Borrelia spirochetes and quantification of membrane blebbing. BtiCdaA-DcdaA was grown in the presence or absence of 1 mM IPTG to late exponential growth phase, centrifuged at 9,391 Â g for 5 min, and resuspended in PBS-MgCl 2 . This step was repeated one time, and the washed spirochetes in PBS-MgCl 2 were used for microscopy. To immobilize the spirochetes for visualization and imaging, bacterial resuspensions were spotted onto a 1% agarose pad on a microscope slide and covered with a coverslip (163). The Keyence BZ-X800 microscope (Keyence Corp., Itasca, IL) was used to visualize bacteria and capture bright-field images at Â60 magnification with oil immersion. Two biological replicates were analyzed, and 100 spirochetes from each replicate were assessed for membrane blebbing.
Statistical methods. All statistical analyses were performed using GraphPad Prism version 8. Twoway analysis of variance (ANOVA) models with Dunnett's test for multiple comparisons were used to compare growth of mutant and complemented strains to the BtWT parent. To compare growth of BtiCdaA-DcdaA grown with and without IPTG, a two-way ANOVA model with Bonferroni's test for multiple comparisons was used. For comparison of numbers of bacteria exhibiting membrane blebbing in BtiCdaA-DcdaA grown with and without IPTG, an unpaired Student's t test was used. P values of 0,0.05 were considered statistically significant. For qRT-PCR analyses, the threshold cycle (2 2DDCT ) method was used to calculate fold change in gene expression relative to BtWT as previously described (89,164).