Intracellular Concentrations of Borrelia burgdorferi Cyclic Di-AMP Are Not Changed by Altered Expression of the CdaA Synthase

The second messenger nucleotide cyclic diadenylate monophosphate (c-di-AMP) has been identified in several species of Gram positive bacteria and Chlamydia trachomatis. This molecule has been associated with bacterial cell division, cell wall biosynthesis and phosphate metabolism, and with induction of type I interferon responses by host cells. We demonstrate that B. burgdorferi produces a c-di-AMP synthase, which we designated CdaA. Both CdaA and c-di-AMP levels are very low in cultured B. burgdorferi, and no conditions were identified under which cdaA mRNA was differentially expressed. A mutant B. burgdorferi was produced that expresses high levels of CdaA, yet steady state borrelial c-di-AMP levels did not change, apparently due to degradation by the native DhhP phosphodiesterase. The function(s) of c-di-AMP in the Lyme disease spirochete remains enigmatic.

Borrelia burgdorferi, the Lyme disease spirochete, encounters numerous microenvironments during its vector-host infectious cycle. Efficient survival and transmission requires that the spirochete produces optimal levels of specific proteins and other components necessary for each step of the cycle. Upon sequencing the B. burgdorferi genome, it was surprising that The cdaA open reading frame was PCR amplified using oligonucleotide primers CDAA-1 and CDAA-2 (Table 1). Primer CDAA-1 introduces a strong AGGAGG ribosome-binding site upstream of the cdaA initiation codon. The resultant amplicon was cloned in pCR2.1 (Invitrogen, Carlsbad, CA), and transformed into E. coli DH5α. The insert of the resultant plasmid was sequenced on both strands to confirm that mutations were not introduced during cloning methods, and that the cdaA ORF was oriented such that transcription could be driven by the vector's lac promoter. This E. coli strain was designated CRS-0. Transcription of cdaA was induced in mid-exponential cultures of CRS-0 by addition of isopropyl-thiogalactoside (IPTG) to a final concentration of 60 μg/ml.
The cdaA ORF was then PCR amplified using primers CDAA-11 and CDAA-12 (Table 1). The B. burgdorferi-E. coli shuttle vector pSZW53-4 [35] was PCR amplified using primers CDAA-13 and CDAA-14. The two amplicons were annealed together by isothermal assembly [37], and E. coli DH5α was transformed with the assembly reaction mixture. The resultant plasmid, pAG1, was purified and the insert sequenced to confirm that no mutation had been introduced, and that the cdaA ORF was in the correct orientation. That construct was introduced into B. burgdorferi B31-e2, and transformant strain AG1 was selected by addition of kanamycin to 200 μg/ml [38].
For studies of the effects of cdaA hyperexpression, mid-exponential phase cultures (approximately 10 7 bacteria/ml) of AG1 were equally divided into two tubes. Transcription of cdaA was induced by addition of 0.5 μg/ml (final concentration) anhydrotetracycline (ATc) to one tube,  and both were incubated for 24h at 35°C. For each pair of induced/uninduced AG1 bacteria, equivalent aliquots were processed for total protein, RNA, and/or cytoplasmic extracts.

SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting
Bacterial protein contents were assessed by electrophoresis in SDS-PAGE and staining with Coomassie brilliant blue. For immunoblot analyses, equal loading of B. burgdorferi cell extracts was assessed by immunoblot against the constitutively-expressed FlaB subunit of the flagella, using monoclonal antibody H9724 [39]. Rabbit polyclonal antisera directed against CdaA was obtained from NeoBioLab (Woburn, MA), who used as antigen a polypeptide consisting of CdaA residues 193-205, NVDSISKAFGTRH, using their standard protocol. Bound antibodies were detected using appropriate horseradish peroxidase-conjugated secondary antibodies and SuperSignal West Pico chemiluminescence reagent (Thermo Scientific).

Analyses of c-di-AMP
E. coli lacks a native c-di-AMP synthetase, and is therefore a useful tool to determine whether or not a protein can produce c-di-AMP [5,9,13]. Thus, cytoplasmic extracts of IPTG-induced E. coli CRS-0 were produced to assess production of c-di-AMP by CdaA. Cytoplasmic extracts were also produced from induced and uninduced B. burgdorferi AG1. For all such analyses, equal volumes of cultures with equivalent concentrations of bacteria were harvested by centrifugation. Bacterial pellets were resuspended in equal volumes of extraction buffer (40:40:20 mixture of methanol, acetonitrile, and 0.1 N formic acid [by volume]), and incubated at -20°C for 30 min. Cellular debris was pelleted by centrifugation, supernatant decanted into a fresh tube, then stored at -80°C. c-di-AMP was quantified by ultra performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) of equal volumes of each bacterial extract, as previously described [5,40].

Quantitative reverse-transcription PCR (q-RT-PCR)
Total RNA was extracted from each set of induced and uninduced bacteria, and cDNA prepared according to previously described methods [41]. For each RNA sample, controls lacking reverse transcriptase were included to confirm absence of contaminating genomic DNA. Each strain and culture condition was independently replicated three times.
Oligonucleotide primer pairs were designed to specifically amplify the B. burgdorferi cdaA, ospC, rpoS, rpoN, bosR, csrA, dhhP, flaB and recA transcripts ( Table 1). The specificity of each primer pair was tested by PCR of B. burgdorferi B31-MI total genomic DNA, and subsequent agarose gel electrophoresis and ethidium bromide staining. The borrelial flaB is generally considered to be constitutively expressed, and is commonly used as an internal standard against which expression levels of other transcripts are determined [41][42][43][44]. Ye et al. used an alternative internal standard, recA, for their analyses of the transcription effects of DhhP levels [30]. Both flaB and recA were used in the current study, in part to compare validity of the two targets as internal standards.
Levels of each target mRNA were assessed by Q-RT-PCR from each sample condition, and performed in duplicate. Transcript fold changes between uninduced and induced cultures of KS50 and AG1 were determined by the ΔΔCt method [45], using both flaB and recA as the standard. Multiple t tests between each transcript fold-changes were performed to determine significance, which were presented graphically (GraphPad Prism version 6.0 for Mac OS X, GraphPad Software, San Diego CA, www.graphpad.com).

B. burgdorferi CdaA synthesizes c-di-AMP
The GenBank bacterial genome database was analyzed by BLAST-P, using the C. trachomatis LGV-L2 c-di-AMP synthase as query. Only one potential homolog was identified in B. burgdorferi type strain B31, ORF BB0008, with an E value of 2 x 10 -30 . Significantly, the borrelial protein contains a consensus DAC domain (Fig 1). Alignment of the predicted borrelial gene product demonstrated extensive homology with other bacterial c-di-AMP synthases (Fig 1).
E. coli does not naturally carry a gene for c-di-AMP synthase, so expression of an exogenous protein in E. coli is a simple means to determine that protein's ability to produce c-di-AMP [5,9]. To that end, the identified borrelial ORF was cloned into E. coli vector pCR2.1, such that its transcription is directed by the vector's lac promoter. The resultant plasmid was introduced into E. coli DH5α, producing strain CRS-0. Cytoplasmic extracts were prepared from induced CRS-0, then analyzed for presence of c-di-AMP by liquid chromatography coupled with tandem mass spectrometry. E. coli expressing the borrelial gene produced readily detectable levels of c-di-AMP (Fig 2).
Thus, it can be concluded that the B. burgdorferi gene encodes a c-di-AMP synthase. A recent proposal has been put forth that DAC domain proteins similar to the borrelial enzyme be named DacA [9]. However, that designation had long ago been given to bacterial D-alanyl-Dalanine carboxypeptidase [46], and B. burgdorferi possesses a gene for that enzyme (ORF BB0605) [27]. We decided not to unnecessarily confuse matters by giving the same name to two unrelated genes/proteins. Among the bacterial proteins with extensive similarities to the borrelial c-di-AMP synthase is the Bacillus subtilis CdaA (formerly YbbP) (Fig 1) [13]. A recent structural analysis of the L. interrogans c-di-AMP synthase also used the name CdaA [7]. We adopted that unambiguous name for the borrelial homolog.

CdaA over-expression in B. burgdorferi
Mass spectrometric analyses of wild-type B. burgdorferi cytoplasmic extracts indicated that cultured borreliae produce very low levels of c-di-AMP, which were barely above the threshold of detection (Fig 3A). Similarly low concentrations of cytoplasmic c-di-AMP were also observed by another research group [30]. Consistent with those observations, CdaA protein levels in cultured B. burgdorferi were found to be below the threshold of immunoblot detection (Fig 3B). Examination of published transcript array data of B. burgdorferi cultured under various conditions, or of regulatory mutants, failed to identify a condition or mutation that significantly altered cdaA expression [e.g., [47][48][49][50][51][52]]. Analyses of our published and unpublished data from RNA sequencing studies of additional B. burgdorferi mutants also did not identify significant regulation of cdaA expression [53] and unpublished results].
Production of c-di-AMP is essential for the survival of previously-studied bacterial species [13,14]. Noting also the low cellular levels of CdaA and c-di-AMP in cultured B. burgdorferi and our demonstration that increased production of CdaA in E. coli resulted in high-level synthesis of c-di-AMP, we examined the effects of hyperexpression of CdaA on B. burgdorferi. Depletion of the DhhP phosphodiesterase blocks borrelial growth [30], so we avoided use of a dhhP mutant for these studies. To that end, strain AG1 was produced, in which cdaA transcription is under control of the TetR-regulated hybrid Post promoter [35,54]. Q-RT-PCR analysis indicated that induction of cdaA in AG1 increased its mRNA levels by 6-fold, and immunoblot analysis confirmed greatly enhanced production of the CdaA protein (Fig 3B,  3C and 3D). However, analyses of cytoplasmic extracts from induced AG1 indicated wildtype levels of c-di-AMP (Fig 3A). The insert of the cdaA-expression plasmid was purified from AG1, re-sequenced, and found to be identical to the native cdaA gene, indicating that the continued low levels of c-di-AMP were not due to a mutation in the introduced enzyme. Hyperexpression of CdaA did not produce any detectable effects of borrelial growth rate, cell size or survival (data not shown). There were also no significant effects on mRNA levels of dhhP, ospC or the regulatory factors rpoS, rpoN, bosR or csrA (Fig 3C and 3D). Bacteria were cultured to mid-exponential phase (approximately 10 7 bacteria/ml), divided equally divided into two tubes, then cdaA transcription was induced by addition of 0.5 μg/ml (final concentration) anhydrotetracycline (ATc) to one tube, and both were incubated for 24h at 35°C. Equal volumes of borrelial cell extracts were analyzed. B. Immunoblot analyses of KS50 and AG1, without and with inclusion of 0.5 μg/ml anhydrotetracycline (ATc) inducer (-and +, respectively). Membranes were probed with antibodies directed against CdaA or the constitutively-expressed FlaB subunit of the flagella. Wild-type and uninduced AG1 bacteria produced substantially less CdaA than did induced AG1, and the immunoblot signal was not detectable for those strains/conditions at the illustrated exposure. Analyses of mRNA levels also indicated that cdaA is expressed at low levels by uninduced AG1 (data not shown). C and D. Q-RT-PCR analyses of the effects of CdaA hyperexpression on transcription of select B. burgdorferi mRNAs. Transcript fold changes are shown as the difference between uninduced and induced cultures for both strains KS50 and AG1, relative to control flaB or recA, respectively [30]. Multiple t tests were performed for each strain and examined transcript. Only the differences in levels of cdaA transcripts in induced cultures of AG1 were significant (indicated by **, p = 0.0012 when compared with flaB, and p = 0.0023 when compared with recA).

Discussion
Bacterial production of c-di-AMP has been detected in some firmicute species, the actinomycetes Mycobacterium tuberculosis and smegmatis, the chlamydian C. trachomatis, and the spirochete B. burgdorferi. DAC motif-containing CdaA homologs are found throughout the spirochete phylum, including the syphilis agent Treponema pallidum and other members of that genus (e.g., T. pallidum Nichols ORF TP0826), and Leptospira interrogans and other leptospires (e.g., L. interrogans Copenhageni ORF LIC10844 and L. biflexa Patoc 1 ORF LEP-BI_I0735) [55][56][57]. It is not obvious why production of this modified nucleotide is restricted to only a few phyla, but absent from proteobacteria and so many others [6].
Since expression of CdaA in E. coli led to significant accumulation of c-di-AMP by that bacterium, we hypothesized that enhanced CdaA levels in B. burgdorferi would similarly lead to increased c-di-AMP production. However, increased levels of the CdaA enzyme in B. burgdorferi did not measurably affect steady-state cytoplasmic c-di-AMP levels. In contrast, depletion of the B. burgdorferi DhhP phosphodiesterase led to increased cytoplasmic levels of c-di-AMP [30]. Those data suggest that DhhP and/or some other enzymatic activity is responsible for maintaining the constant, low levels of c-di-AMP in both wild-type and induced AG1 borreliae.
The results of these studies and those of Ye et al. [30] raise an important question about the function of c-di-AMP in B. burgdorferi: why is this molecule, which uses up 2 ATP molecules, produced by CdaA but then destroyed? To date, no signal has been identified that affects expression levels of CdaA. B. burgdorferi does control expression of dhhP [30]. However, conditional depletion of DhhP led to an approximately 40-fold increase in c-di-AMP concentration, along with a cessation of growth, while ectopic modulation of DhhP that yielded a 5-fold increase in c-di-AMP levels did not have any noticeable effects on growth or cell division [30]. Thus, there is an apparently broad window of c-di-AMP levels that can be tolerated by B. burgdorferi without having a detectable impact on the bacteria. Whether c-di-AMP directly controls B. burgdorferi growth, division, and/or regulatory factors remains to be determined, since the observed phenotypes may be indirect responses to stresses induced by disruption of another bacterial function(s). It is also possible that the DhhP phosphodiesterase acts on substrates other than c-di-AMP, which may be responsible for the growth defects when DhhP is depleted.
The field of bacterial c-di-AMP signaling is still in its infancy, and is not well understood in any species. Of the c-di-AMP-binding proteins that have been identified in other bacteria, homologs of the following are present in B. burgdorferi: ORF BB0724 is orthologous to Streptococcus pneumoniae CabP (E = 6x10 -27 ), BB0725 to Staphylococcus aureus KtrA (E = 7x10 -21 ), and both ORFs BB0216 and BB217 to Staphylococcus aureus PstA (E = 7x10 -21 , and E = 7x10 -21 , respectively) [16,27,32,[58][59][60]. Those streptococcal and staphylococcal proteins are all involved with potassium transport, so the similarities with borrelial proteins may simply be due to that function. Nonetheless, examination of interactions between c-di-AMP and B. burgdorferi phosphate transport proteins, and the significance of any such binding, may be a fruitful venue for future studies.
Riboswitches dependent upon c-di-AMP have been identified in some bacterial species, which may affect gene expression [61][62][63][64]. To the best of our knowledge, the possibility of riboswitches being present in B. burgdorferi has yet to be explored. The oral spirochete Treponema denticola contains a thymidine pyrophosphate-dependent riboswitch [65], suggesting that such regulatory mechanisms may exist in other spirochetes.
Another potential role for CdaA and DhhP is production and degradation of di-AMP (pApA), which is the initial c-di-AMP breakdown product. That dinucleotide may serve as a nanoRNA, which could have wide-ranging impacts upon transcription initiation [66][67][68]. We note also that many different nanoRNAs are produced and degraded in other bacterial species by DHH-motif enzymes, supporting the possibility that B. burgdorferi DhhP might degrade a broader variety of nucleic acids than just c-di-AMP [66][67][68][69][70][71].
Borrelial c-di-AMP may have impacts beyond the bacterium itself. c-di-AMP produced by L. monocytogenes and C. trachomatis activates a type I interferon response by host cells [2,5,[22][23][24][25]. Although those bacteria invade host cells, while B. burgdorferi is an extracellular pathogen, it is possible that a portion of the observed type I interferon responses observed during B. burgdorferi infection might be linked to the spirochete's c-di-AMP [29,[72][73][74].
In summation, these studies demonstrated that B. burgdorferi produces an enzyme, CdaA, that synthesizes c-di-AMP. Homologs of CdaA are found throughout the spirochete phylum. We hypothesize that this modified nucleotide is rapidly broken down by the DhhP phosphodiesterase. Thus, regulation of CdaA did not significantly affect cytoplasmic levels of c-di-AMP, and we predict that other mechanisms, such as factors that control the activity of CdaA or altered expression of DhhP, are the major drivers of altering c-di-AMP levels.