Host-specific functional compartmentalization within the oligopeptide transporter during the Borrelia burgdorferi enzootic cycle

Borrelia burgdorferi must acquire all of its amino acids (AAs) from its arthropod vector and vertebrate host. Previously, we determined that peptide uptake via the oligopeptide (Opp) ABC transporter is essential for spirochete viability in vitro and during infection. Our prior study also suggested that B. burgdorferi employs temporal regulation in concert with structural variation of oligopeptide-binding proteins (OppAs) to meet its AA requirements in each biological niche. Herein, we evaluated the contributions to the B. burgdorferi enzootic cycle of three of the spirochete’s five OppAs (OppA1, OppA2, and OppA5). An oppA1 transposon (tn) mutant lysed in the hyperosmolar environment of the feeding tick, suggesting that OppA1 imports amino acids required for osmoprotection. The oppA2tn mutant displayed a profound defect in hematogenous dissemination in mice, yet persisted within skin while inducing only a minimal antibody response. These results, along with slightly decreased growth of the oppA2tn mutant within DMCs, suggest that OppA2 serves a minor nutritive role, while its dissemination defect points to an as yet uncharacterized signaling function. Previously, we identified a role for OppA5 in spirochete persistence within the mammalian host. We now show that the oppA5tn mutant displayed no defect during the tick phase of the cycle and could be tick-transmitted to naïve mice. Instead of working in tandem, however, OppA2 and OppA5 appear to function in a hierarchical manner; the ability of OppA5 to promote persistence relies upon the ability of OppA2 to facilitate dissemination. Structural homology models demonstrated variations within the binding pockets of OppA1, 2, and 5 indicative of different peptide repertoires. Rather than being redundant, B. burgdorferi’s multiplicity of Opp binding proteins enables host-specific functional compartmentalization during the spirochete lifecycle.

Though ABC transporter SBPs typically engage their ligands via side chain-specific interactions within the binding pocket, bacterial OppAs bind in a sequence-independent manner via the peptide backbone [22][23][24][25]. B. burgdorferi OppAs use this same binding mechanism with variations in cavity volume, configuration, and electrostatics [14,26], theoretically enabling each OppA to accommodate a unique repertoire of peptides. Our previous study suggests that B. burgdorferi couples structural diversity with differential expression of OppAs to optimize peptide uptake at each phase of its enzootic cycle [14,17,[27][28][29][30][31][32][33]. oppA1, oppA2, oppA3 and oppA4 are expressed within the feeding tick, while oppA2 and oppA5 are predominantly expressed in the mammal [14]. Evidence to date suggests that this intricate regulatory scheme involves all three of the LD spirochete's known global regulatory networks [5,12,[34][35][36]. Rel Bb generates the alarmone, (p)ppGpp, which regulates the stringent response within the tick and promotes expression of oppA1 and oppA3 [27,30]. The alternative sigma factor RpoS promotes global transcriptional changes needed for mammalian host-adaptation, including positive regulation of oppA5 [29,32,33]. Activation of the Hk1/Rrp1 two-component system during the blood meal results in the production of c-di-GMP, inducing the expression of tick phase genes, including oppA4 [37]. Conversely, c-di-GMP represses oppA5, potentially explaining why this RpoS-dependent gene is expressed only during the mammalian phase [14,37].
Herein, we further explored how individual oligopeptide-binding proteins of the B. burgdorferi Opp system maintain spirochete viability in ticks and mice. We identified sharply divergent phenotypes for oppA1 and oppA2 transposon (tn) mutants. The oppA1tn mutant was lysed during the tick blood meal, a phenotype that we attribute to deficient uptake of amino acids required for osmoprotection. The oppA2tn mutant, on the other hand, displayed a profound defect in hematogenous dissemination in mice, yet persisted within skin while inducing a minimal antibody response. We hypothesize that this phenotype represents a link between peptide uptake and chemotaxis during the mammalian phase of the cycle. Our previous study [38] identified a role for OppA5 in spirochete persistence within the mammalian host; in this report we show that the oppA5tn mutant displayed no defect during the tick phase of the cycle and could be transmitted. Rather than being redundant, B. burgdorferi's multiplicity of Opp binding proteins enables host-specific functional compartmentalization during the spirochete lifecycle.

Genetic and in vitro characterization of oppA1 and oppA2 tn mutants
As noted in the Introduction, our previous expression and structural data inferred that oppA1 and oppA2 play distinct roles within the spirochete lifecycle [14]. Along these lines, Troy et al. reported that a tn mutant for oppA1 was infectious in mice [39], while oppA2tn was attenuated [40]. We conducted experiments to further these phenotypic characterizations. We selected from the STM tn library [41] the oppA1 and oppA2 tn mutants with the most 5' tn insertions (T11TC050 and T06TC269, respectively; Fig 1A). T06TC269 was missing lp21; while the importance of this plasmid is not known, we elected to reconstruct the mutant in the wt strain (see Methods). We next confirmed by qRT-PCR that the tn insertions in oppA1 and oppA2 affected transcription of just the targeted genes (Fig 1B and 1C). Compared to wt, both mutants were morphologically indistinguishable (S1A Fig), exhibited normal motility in BSK-II (S1-S3 Movies), and demonstrated no in vitro growth defects (S1B Fig). Zhou et al.
[42] recently noted that expression of OspC is dysregulated in an oppA4 mutant; therefore, we confirmed that our mutants temperature-shifts normally (S1C Fig). For complementation, we inserted the oppA1 or oppA2 coding regions into cp26, each preceded by the 500 bp upstream of oppA1, which contains the native operonic promoter (Fig 1A) [17]. Complementation of either mutant restored transcription of their respective genes (Fig 1B and 1C).

oppA1 is essential for survival of B. burgdorferi in feeding ticks
To evaluate infectivity of the oppA1tn mutant, cohorts of mice were needle-inoculated with 1 x 10 4 wt or oppA1tn and assessed for infection four weeks later. As previously reported [39], oppA1tn did not display a virulence defect (S2A Fig and Table 1 -Pilot Experiment). Two weeks after syringe-inoculation, spirochete burdens in ticks were assessed by semi-solid plating of replete larvae fed on the infected mice. In stark contrast to larvae fed on wt-infected mice, larvae fed on oppA1tn-infected mice contained no live spirochetes (S2B Fig). Mutant spirochetes also were not recovered by plating from flat nymphs after the larvae had molted (S2C Fig). These experiments were repeated with the oppA1c strain and, as expected, the mutant and complement were comparably infectious in mice (Table 1 -Complementation Experiment). Once again, there was no recovery of viable oppA1tn spirochetes from replete larvae, while complementation restored their ability to survive the larval blood meal (Fig 2A). Midguts and oppA2tn mutants and the corresponding complements in cp26. The transposon (tn) insertion sites (bp of coding sequence) are indicated. aacC and aadA confer gentamycin and streptomycin resistance, respectively. qRT-PCR primers specific to the tn insertion site are noted with small black arrows. (B-C) Transcript copy numbers of oppA1-3 (mean ± SEM, normalized to flaB) in (B) wt, oppA1tn, and oppA1c and (C) wt, oppA2tn, and oppA2c determined from triplicate biological replicates and quadruplicate technical replicates. Statistical analyses were performed using unpaired Student's t test. https://doi.org/10.1371/journal.ppat.1009180.g001

PLOS PATHOGENS
Host-specific functions of OppAs of acquiring larvae contained equivalent bacterial burdens by qPCR (Fig 2B), confirming that oppA1tn was taken up at levels comparable to wt and complement. Immunofluorescence analysis (IFA) of midguts from larvae fed on oppA1tn-infected mice revealed mainly spirochete remnants (Fig 2C), which, in combination with the qPCR results, indicate that the mutant lysed in the midgut after acquisition. We used larval immersion feeding [43] to confirm that oppA1 is essential for structural integrity of the spirochete during the blood meal. As with natural acquisition, viable mutant spirochetes were not recovered, though bacterial burdens of all three strain were indistinguishable by qPCR (Fig 2D and 2E).

Loss of OppA2 impairs dissemination within the mammal following needle-inoculation
In preliminary studies, mice were needle-inoculated with 1 x 10 4 wt or oppA2tn spirochetes and evaluated by culture (pinnae, inoculation site, tibiotarsal joint, bladder, and heart) and

PLOS PATHOGENS
Host-specific functions of OppAs (Table 2 -Skin Sampling Experiment) were culture-negative; as before, oppA2tn infection elicited minimal serological response ( Fig 3D). In contrast to Pilot Experiment 2 (above), complementation in this experiment was not complete, as hearts from oppA2c-infected mice were culture negative. The wt and oppA2c strains were recovered from the large majority of dorsal and ventral sites (139/160 and 149/160 positive sites, respectively; Fig 3A and 3C), whereas the oppA2tn mutant was recovered from significantly fewer sites (58/160, Fig 3B) distributed radially from the inoculation sites (also culture positive). Notably, oppA2tn spirochetes were recovered from only a handful of sites on the ventral flanks. Cultures from positive skin sites were darkfield-positive within the same time frame (one week) for all three strains.

Lack of OppA2 does not impair spirochete survival in murine blood ex vivo
The above negative blood culture results (Table 4) suggested that oppA2tn is either unable to invade the vasculature or survive within the blood compartment. To evaluate the latter possibility, we devised an ex vivo assay to evaluate spirochete viability during prolonged incubation in blood. wt, oppA2tn, and oppA2c cultures were washed in PBS and incubated for 48 h in triplicate in undiluted mouse blood at a concetration of 1 x 10 6 spirochetes/ml. As darkfield microscopy could visualize only red blood cells, at 24 h, spirochete viability was assessed qualitatively by live/dead staining and epifluorescence microscopy ( Fig 4A). Microscopy revealed viable spirochetes for each strain which displayed normal cell morphology. At 24 and 48 hrs, samples were collected for semi-solid plating ( Fig 4B). None of the strains replicated during culture in blood, and all exhibited essentially identical decreases in viability over time, confirming that oppA2tn fared no worse than wt and oppA2c spirochetes during blood culture.

oppA2tn can survive and be acquired depending on tick placement
In two initial experiments, we evaluated the ability of oppA2tn to be acquired by and survive in the arthropod vector. In experiment one, larvae fed on mice infected with wt or oppA2tn contained equivalent spirochete burdens ( Fig 5A, Experiment 1, left panel). However, in experiment two, which included oppA2c, two of three pools of oppA2tn-infected larvae had no viable spirochetes, while spirochete numbers in the third pool were~1 log lower than the wt and oppA2c pools ( Fig 5A, Experiment 2, right panel).
To explain this disparity, we first conducted experiments using the larval immersion technique [43] to evaluate whether the oppA2tn mutant could survive within the midgut during feeding; as shown in Figs 5B and S4, the mutant survived the blood meal as well as wt and oppA2c. The experiments depicted in Fig 5A were done by whole body infestation with feeding larvae randomly distributed on the mouse. We, therefore, considered the possibility that the

PLOS PATHOGENS
Host-specific functions of OppAs differences in acquisition in the two preliminary experiments were due to non-uniform distribution of the mutant. To test this, we conducted acquisition experiments by dorsal placement of larvae in feeding capsules. In addition to the three groups of dorsally infected mice described in Fig 3A-3C, we included a fourth cohort of mice inoculated ventrally with oppA2tn ( Fig 5D). Skin cultures of the ventrally inoculated oppA2tn mice at the time of sacrifice (4 weeks; Fig 5D) demonstrated a converse pattern to that obtained following dorsal inoculation; namely, spirochetes reached only the dorsal flanks. Larvae capsule-fed on the dorsally infected mice acquired equivalent bacterial burdens, whereas larvae fed on mice infected ventrally with oppA2tn did not acquire spirochetes ( Fig 5C). Thus, the mutant does not display an

PLOS PATHOGENS
Host-specific functions of OppAs

PLOS PATHOGENS
Host-specific functions of OppAs innate acquisition defect; rather, acquisition of the mutant depends upon the placement of ticks in relation to the site of needle-inoculation.

oppA2tn displays defective dissemination via nymphal inoculation
We used the oppA2tn-infected larvae from our first natural acquisition experiment (Fig 5A, Experiment 1, left panel) to assess the ability of the mutant to be transmitted and disseminate by nymphal inoculation. Survival of the molt was confirmed by recovery of equivalent numbers of wt and oppA2tn from flat nymphs by plating ( Fig 6A). When wt-and oppA2tn-infected nymphs were fed to repletion on naïve mice (~15 nymphs/mouse), the resulting colony counts were virtually identical (Fig 6B), further confirming that loss of OppA2 does not affect survival in the tick. At two-weeks post-drop-off, inoculation sites were culture positive in four of the nine oppA2tn nymph-infected mice, while samples from all distal tissues were culture negative ( Table 2 -Nymphal Transmission Experiment); oppA2tn nymph-infected mice also failed to seroconvert (S5 Fig). Samples from wt infected mice were consistently culture positive. We next sought to determine whether tick inoculation with oppA2tn fully replicates the dissemination defect observed following infection by needle. Therefore, we inoculated naïve mice with nymphs infected with wt, oppA2tn, and oppA2c and assessed infection by culture of distal sites and extensive skin-sampling four-weeks post-feeding. As expected, colony counts of replete nymphs fed on naïve mice (~15 nymphs/mouse) were virtually identical among the three strains (Fig 6C) confirming, once again, that oppA2tn survives normally in feeding ticks. As in prior experiments, pinnae and organs from the oppA2tn-infected mice were culture

PLOS PATHOGENS
Host-specific functions of OppAs negative ( Table 2 -Nymphal Skin Sampling Experiment). With the exception of hearts, distal tissues from oppA2c infected mice were culture positive. The wt and oppA2c strains were recovered from the large majority of dorsal and ventral sites (151/160 and 144/160 positive sites, respectively; Fig 6D and 6F). Though all inoculation sites were culture positive for the oppA2tn infected mice, significantly fewer skin samples yielded positive cultures (84/160, Fig  6E). In addition, skin mapping from mice infected with oppA2tn displayed a similar radial dissemination pattern their needle-inoculated counterparts (compare Figs 3B and 6E), with slightly increased coverage on the ventral aspects; in contrast, wt and oppA2c infected mice demonstrated robust cutaneous dissemination (Fig 6D and 6F). The slight increase in recovery from ventral sites of oppA2tn infected mice, as compared to needle-inoculation, could be due to the larger surface area of the 'inoculation site' with capsule feeding. As with needle-inoculated mice, positive skin cultures for all three strains were darkfield positive within the same time frame (one week).

oppA5 is dispensable within the tick but required for persistence within the mouse
Recently, we demonstrated that an oppA5tn mutant displays an impaired ability to persist in needle-inoculated mice which could be restored by complementation [38]. oppA5, the only RpoS-regulated Opp component [17,29,32,33,38], is not expressed in feeding nymphs [14], an RpoS-ON state [5], because it is repressed by c-di-GMP [37]. These expression data lead to the prediction that OppA5 is not required for survival within ticks. To confirm this, we assessed the tick phenotype of the oppA5tn mutant using immersion feeding to circumvent its attenuation in mice. Spirochete burdens in immersion-fed larvae ( Fig 7A) and flat nymphs following the molt (Fig 7B) were essentially identical by plating. Flat nymphs infected with wt or oppA5tn spirochetes were fed on naïve mice; Fig 7C shows that nymphal burdens at repletion, also by plating, were comparable for the two strains. At four weeks, inoculation sites in all mice were culture positive, confirming transmission. Consistent with the previously described phenotype observed by needle-inoculation [38], the oppA5tn-infected mice showed mild attenuation at the four week timepoint (9/12 vs 12/12 positive cultures for oppA5tn and wt, respectively; Table 5).

Discussion
Though B. burgdorferi cannot utilize AAs as carbon sources [7], they are required for biosynthesis of proteins and peptidoglycan (PG). The absence of AA biosynthetic pathways obligates the spirochete to procure these essential nutrients using a small repertoire of dedicated AA transporters in concert with a complex oligopeptide uptake system [7]. As the singular ATPase is the energetic lynchpin of the Opp system, we previously targeted this component via conditional mutagenesis to demonstrate that peptide uptake is essential for B. burgdorferi viability, even within an enriched cultivation medium containing a full complement of AAs [14]. Peptide starvation of the ATPase cond mutant in vitro resulted in a novel, pleomorphic phenotype characterized by dysregulated cell envelope biogenesis and arrested cell division. ATPase cond spirochetes cultivated in DMCs host-adapted but displayed similar growth defects, providing an in vivo morphological explanation for the mutant's inability to establish a foothold following needle inoculation. It is unclear, however, whether these maladaptive responses are strictly nutritional, the result of AA deprivation and as yet undefined mechanisms for sensing diminished intracellular AA pools [45], or reflective of aberrant environmental sensing [46,47]. Transport of signaling peptides by Opp systems is well described in Gram-positive bacteria [48][49][50][51], although there is no evidence to date for peptide-based cell-cell communication in B.

PLOS PATHOGENS
Host-specific functions of OppAs burgdorferi. In the present manuscript, we used oppA mutants to confirm that peptide acquisition is essential within the tick as well as the vertebrate reservoir, demonstrating that B. burgdorferi has evolved a unique 'brand' of nutritional virulence spanning its entire lifecycle. We also confirmed the conjecture emerging from our previous studies [14] that B. burgdorferi

PLOS PATHOGENS
Host-specific functions of OppAs employs temporal regulation in concert with structural variation of Opp components to ensure acquisition of sufficient AAs in every biological niche in which it persists. It is noteworthy that none of the oppA mutants examined herein exhibited major growth or morphologic defects in vitro, underscoring a striking dichotomy between the redundancy of the Opp system in vitro and the functional specialization of its components in vivo. An unexpected facet of this specialization may be a signaling function linking peptide acquisition to dissemination during the mammalian phase of the life cycle.
The major theme to have emerged from this study is that individual borrelial OppAs appear to function in either the tick or murine environment. Of the five OppAs, OppA1 and OppA2 illustrate a clear dichotomy, with unique transcriptional profiles throughout the cycle and the greatest structural divergence in binding cavity size, electrostatics (Fig 8), and non-synonymous liganding residue diversity [14]. Prior qRT-PCR analysis revealed that transcription of oppA1 predominates in feeding ticks [14]. Experiments herein using an oppA1tn mutant brought to light a striking, tick-specific phenotype associated with this transcriptional profile. While needle-inoculated oppA1tn spirochetes established persistent murine infection comparable to wt B. burgdorferi, semi-solid plating and IFA revealed that the mutant not only lacked viability in ticks but also lysed in larval midguts. qPCR unequivocally ruled out the possibility that the mutant had failed to transit to the vector; furthermore, immersion feeding, which bypasses the infectious phase, yielded an identical phenotype. The PG sacculus maintains the integrity of bacterial cell envelopes against high turgor pressures [49,[52][53][54]. Thus, one plausible explanation for the oppA1tn phenotype is that insufficient transport of peptides rich in AAs essential for PG synthesis (e.g., arginine/ornithine, alanine, glycine, and glutamate [55])

PLOS PATHOGENS
Host-specific functions of OppAs renders the mutant vulnerable to lysis from changes in osmolality within the midgut during the blood meal [56]. Another possibility, which is not mutally exclusive, is that OppA1 imports peptides containing AAs that serve as osmoprotectants (e.g., proline and glutamate) [18]. Our thinking that the PG sacculus is involved in the oppA1tn phenotype is based on its resemblance to the lytic phenotype of rrp1 and hk1 mutants during tick feeding [18,37, [57][58][59]. Rrp1 and Hk1 form a two-component system responsible for generation of c-di-GMP, an important regulator for tick phase genes, including importers of chitobiose and N-acetylglucosamine, carbohydrates required for PG synthesis [11,12,31,58]. While the mechanism(s) underlying the protective function of OppA1 during tick feeding remains to be determined, our results argue that evolution has tailored the binding capacity of OppA1 to the range of peptides available within this unique milieu. The blood meal provides an abundance of proteins not just from serum but also intraluminal lysis of erythrocytes (e.g., hemoglobin and serum albumin) [60][61][62]. Furthermore, tick saliva contains a wide array of small peptides, which, upon re-ingestion, provide an arthropod-specific source of peptides [63]. HtrA, a B. burgdorferi surface protease [64][65][66], and/or host-derived proteases such as plasminogen bound to the borrelial surface [67] also may contribute to protein degradation within the midgut. In contrast to relapsing fever spirochetes, which are acquired by their soft tick and louse vectors from blood, Ixodes spp. acquire LD spirochetes from the dermis [68][69][70]. Thus, maintenance of the B. burgdorferi enzootic cycle depends upon widespread dissemination of spirochetes within the reservoir host, often thought to occur predominantly via hematogenous spread [44,[71][72][73][74]. However, two lines of evidence argue for the importance of intracutaneous migration: (i) bioluminescence imaging of infected laboratory mice shows extensive lateral spread at early timepoints (i.e., within two weeks) [75][76][77], and (ii) the hallmark cutaneous lesion of LD in humans, erythema migrans, which can be expansive, is the result of lateral migration of spirochetes [78][79][80][81]. One might surmise, therefore, that extensive cutaneous dissemination also occurs within the reservoir host. Using a Tn-seq format in which tissue samples were pooled, Troy et al. [40] previously showed that an oppA2tn mutant was significantly attenuated. An unexpected outcome from our study is the discovery that the loss of OppA2, a protein with a presumptive nutritive function, results in a striking defect in dissemination. Our finding that oppA2tn retained viability at wt levels in our ex vivo blood assay argues that the failure of the mutant to disseminate hematogenously was due to an inability to access the vascular compartment. Although spirochetes lacking OppA2 could not spread hematogenously, they could migrate intracutaneously from the inoculation site, eventually gaining access to approximately 50% of the mouse's surface area. Notably, the oppA2tn mutant was never recovered from pinnae, a site widely accepted as an indicator of hematogenous dissemination [71]. The oppA2tn phenotype, therefore, provides the first genetic evidence that efficient and complete dissemination of B. burgdorferi throughout the skin requires a combination of intracutaneous migration from the site of inoculation and hematogenous seeding of distal skin. Further experiments tracking spirochete migration in skin and visualizing individual organisms in situ will be needed to determine whether (and, if so, the extent to which) impaired intracutaneous migration contributes to the diminished spread of oppA2tn in skin. Complementation confirmed that the dissemination defect was due to loss of oppA2, although infection of hearts was not fully restored. oppA2 expression in vivo is a combined result of transcription via the operon's oppA1 promoter and an internal oppA2 promoter [14,17,19]. The partial complementation observed with oppA2c likely reflects sub-optimal expression of oppA2 from just the oppA1 promoter in the mammalian environment ( Fig 1A).
Unexpectedly, we could not definitively demonstrate a nutritive function for OppA2 in vitro or in vivo. While the mutant displayed a modest growth defect during DMC cultivation, it maintained wt burdens and viability in the skin at least four weeks post-inoculation and survived as well as wt during ex vivo blood cultivation. Therefore, if OppA2 does contribute to nutrient acquistion, oppA2tn presumably can compensate for its loss via its other OppAs and/ or limited reportoire of free AA transporters. In any event, it seems difficult to attribute the mutant's dissemination defectly solely to AA deficiency. Indeed, our collective results lead us to consider the intriguing possiblity that OppA2 also functions in a signaling capacity during the mammalian phase of the life cycle. We can envision two mechanisms by which this might occur. One is that peptides (or their degradation products) imported by OppA2 activate an as yet unidentified regulatory pathway [82] that orchestrates expression of adhesins and other molecules B. burgdorferi needs to access and penetrate dermal vasculature [83]. Conceivably, this pathway would be part of the program for mammalian host adaptation B. burgdorferi undergoes that is not RpoS-mediated. Another is that the inability of oppA2tn to transit from the dermis to the dermal vasculature reflects impaired chemotaxis. Studies in E. coli provide a clear precedent as well as a mechanism for peptide-mediated chemotaxis. E. coli mutants in either the methyl-accepting chemotaxis protein, Tap, or the dipeptide SBP, DppA, fail to chemotax toward dipeptides. The accepted interpretation of these results is that liganded DppA transduces a chemotactic signal via interaction with Tap [84][85][86]. In the case of B. burgdorferi, liganded OppA2 could interact with one of its five annotated MCPs. In light of the promiscuous nature of OppA-peptide interactions, OppA2 would then serve as part of a 'generalized' sensor of mammalian phase chemoattractant peptides originating from the blood. To date, chemotaxis of B. burgdorferi towards a number of stimuli has been demonstrated [87][88][89][90], though chemotactic responses toward peptides have not been evaluated. Whereas B. burgdorferi motility mutants are cleared from the inoculation site within several days of infection [91][92][93][94], oppA2tn survives for weeks in the skin, a major immune organ [95], inducing only a negligible antibody response. To the best of our knowledge, this is the first report of long term survival of B. burgdorferi without immune recognition.
The reservoir competence of a mammalian host for LD spirochetes is defined by its tolerance of the bacterium [96][97][98]. However, in order to exploit this tolerance, B. burgdorferi must employ a complex parasitic strategy that involves virulence determinants [4,5,73,83,99], immune evasion mechanisms [100][101][102][103], and usurpation of the host's metabolic output [6][7][8][9][11][12][13]. Previous transcriptional analyses [14] led to the prediction that OppA2 and OppA5 function primarily within the mammal and, along with our prior study [38], we now have evidence that these two OppAs are, indeed, indispensable for widespread and persistent murine infection. Moreover, their dispensability within the tick furthers our concept of host-specific compartmentalization of OppA function. Instead of working in tandem, however, OppA2 and OppA5 appear to function in a hierarchical manner: OppA5's ability to promote persistence relies upon OppA2's ability to facilitate dissemination. Although the conformations of the OppA2 and OppA5 binding pockets are similar, differences in electrostatics (Fig 8) and residues lining the pocket [14] suggest divergent peptide repertoires. A major unsolved question is the source of peptides during infection, especially when one considers that the mammalian reservoir is a non-inflammatory milieu [104]. Regardless, in comparison with the feeding tick, the mammal is a peptide-limited environment. Conceivably, peptide limitation establishes a 'set' point for borrelial replication within the mammal, serving as a contributing determinant of the paucibacillary nature of B. burgdorferi infection [96,105,106].

Ethics statement
All animal experiments were performed in strict accordance with protocols approved by the UConn Health Center Institutional Animal Care and Use Committee (Animal Welfare

PLOS PATHOGENS
Host-specific functions of OppAs Assurance No. A347-01) and in compliance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.

Bacterial strains and culture conditions
All strains used in this study are listed in S1 Table.  . Tissues and blood were cultured in BSK-II containing Borrelia antibiotic cocktail (BAC; 0.05 mg/ml sulfamethoxazole, 0.02 ml/ml phosphomycin, 0.05 mg/ml rifampicin, 0.01 mg/ml trimethoprim, and 2.5 μg/ml amphotericin B). Temperature-shift experiments were carried out as previously described [109]. Plasmid content of B. burgdorferi strains was determined using the multiplex approach as described in Bunikis et al. (S6 Fig) [110] Generation of mutant and complement strains oppA2tn. The oppA2tn mutant was reconstructed by amplification of the bb0328-9 region from the original oppA2tn mutant (T06TC269) with~1 kb of DNA flanking the tn insertion. As the tn insertion includes an E. coli origin of replication, the amplified fragment was selfligated to generate pEcAG233 (Fig 1A), which then was transformed into wild-type B31 5A18 NP1. Subsequent clones were confirmed by PCR to carry the tn insertion in oppA2 and screened for plasmid content (S6 Fig) as described above. oppA1c. The coding region for oppA1 was amplified along with the upstream 500 bp to include the oppA1 promoter (PoppA1) flanked by AatII restriction enzyme sites. We modified a cp26 crossover vector containing aacC and PflaB-GFP (pMC2498) [37] in order to utilize the cp26 site for complementation. The antibiotic marker and gfp cassette were removed by inverse PCR and self-ligation via engineered AatII sites (pEcAG265). The aadA marker [111] was introduced into the MCS via a BamHI-restriction site (pEcAG326) The complementation fragment was cloned into the AatII site of pEcAG326 to generate pEcAG341 (Fig 1A). The plasmid was confirmed by sequencing and transformed into the oppA1tn mutant. Subsequent clones were screened by PCR for wild-type oppA1 and plasmid content (S6 Fig). oppA2c. The coding region for oppA2 was amplified and fused with the 500 bp PoppA1 with AatII sites flanking the PoppA1-oppA2 fragment. The complementation fragment was cloned into the AatII site of the cp26 crossover vector pEcAG326 to generate pEcAG342 (Fig 1A). The plasmid was confirmed by sequencing and transformed into the oppA2tn mutant. Subsequent clones were screened by PCR for wild-type oppA2 and plasmid content (S6 Fig).

Growth curves
B. burgdorferi strains were inoculated into BSK-II containing appropriate antibiotics at 1 x 10 3 spirochetes/ml and incubated for up to 6 days at 37˚C. Spirochetes were enumerated daily by dark-field microscopy as previously described [14]; all experiments were performed in triplicate.

PLOS PATHOGENS
Host-specific functions of OppAs

qRT-PCR
Primers used for qRT-PCR assays are listed in S2 Table. Total RNA was isolated as previously described [114] from triplicate cultures of temperature-shifted spirochetes at late logarithmic phase of growth. cDNAs were generated with and without reverse transcriptase using the SuperScript III First Stand Synthesis System (ThermoFisher Scientific). qRT-PCR assays were developed to measure oppA1 and oppA2 transcripts in the tn mutants by flanking the transposon insertion site in each gene (Fig 1A). oppA transcripts were quantified using SsoAdvanced Universal SYBR Mix (Bio-Rad, Hercules, CA) and normalized to flaB transcripts [115] (S2 Table) using SsoAdvanced Universal Probe Mix (Bio-Rad). All assays were performed in quadruplicate with three biological replicates. Generation of internal standards for each assay were previously described [14].

Generation of host-adapted spirochetes in DMCs
Mammalian-host-adapted spirochetes were generated by cultivation in DMCs as previously described [116,117]. Briefly, B. burgdorferi strains were grown to mid-logarithmic phase and diluted to 1 x 10 4 spirochetes/ml. SpectraPor dialysis membrane tubing with a 6-8 kDa cutoff (ThermoFisher Scientific) containing 10 ml of diluted spirochetes were implanted into the peritoneal cavities of female Sprague-Dawley rats (175-200 g). DMCs were explanted after two weeks; spirochetes then were enumerated via dark-field microscopy and evaluated for host adaptation by silver stain.

In vitro blood culture assay
Mouse blood was collected by heart stick and clotting was prevented by the addition of 0.1M sodium citrate. 100ul blood was aliquoted into a 96-well tissue culture plate and 1 x 10 6 spirochetes/ml were added to each well. The input volume was plated in triplicate for each strain to determine the number of viable spirochetes at timepoint 0 hrs. Plates were incubated at 37C at 5% CO2 and samples were collected at 24 hrs for microscopic evaluation. Samples were incubated for 10 min at room temperature with Hoechst (10μg/ml) and propidium iodide (20μg/ ml) for live/dead staining, respectively. Spirochetes were visualized on an Olympus BX41 microscope with a Retiga Exi camera (QImaging, Surrey, BC, Canada). Images were acquired with a 40x (1.4NA) objective and QCapture software v. 2.1 (QImaging). For epifluorescence, an X-Cite Xylis light source and a DAPI HYB filter or HQ:Rdil filter were used to detect

Host-specific functions of OppAs
Hoechst and propidium iodide, respectively. Images were processed using ImageJ v 1.8.0; and Hoechst and propidium iodide images were colorized green and red, respectively. Samples also were collected at 24 and 48 hrs for semi-solid plating and colony counts.

Infection studies
Five-to-eight-week-old female C3H/HeJ mice (Jackson Laboratories, Bar Harbor, ME) were used for all infections. Mice were injected intradermally with 1 x 10 4 spirochetes in a central dorsal location, with the exception of the oppA2tn mutant infections that included an additional cohort inoculated ventrally. Between two and five weeks post-inoculation, tissues were collected for culture as described above; mouse sera were collected for Western blots as described above. To evaluate hematogenous dissemination, blood was collected for culture at 5, 6, and 7 days post-inoculation (peak spirochetemia) [44] and 10 μl was inoculated into 5 ml BSK-II with BAC. To measure cutaneous dissemination, thirty-two skin samples, including the inoculation site, were collected as shown in the sample maps (Figs 3A-3C, 5D and 6D-6F) and cultured as described above.

Tick studies
Larval acquisition. At two weeks post-infection, naïve mice were used as a blood meal for pathogen-free Ixodes scapularis larvae (Oklahoma State University, Stillwater, OK). Larvae were allowed to feed either by whole body infestation or by feeding within a capsule mounted on the dorsal side of the mouse [114]. At repletion, pools of ten larvae from each mouse were evaluated by semi-solid plating for colonies [118] and qPCR using a 1:10 dilution of DNA and the Taq-man flaB assay [115] (S2 Table) for quantification of DNA burdens [119]. Immunofluorescence microscopy of crushed ticks was performed as previously described [114] using KPL anti-Bb FITC-conjugated antibody (1:400; SeraCare Life Sciences, Milford, MA). Spirochetes were visualized on an Olympus BX41 microscope with a Retiga Exi camera (QImaging, Surrey, BC, Canada). Images were acquired with a 40x (1.4NA) oil immersion objective and QCapture software v. 2.1 (QImaging). Images were processed using ImageJ v. 1.8.0.
Immersion fed larvae. Immersion fed larvae were generated as previously described [43]. Briefly,~200 naïve larvae were immersed in 2 x 10 8 spirochetes/ml for 1 hr and washed with PBS. Larvae were allowed to recover overnight in an environmental incubator and then fed on naïve mice by the capsule feeding method [114]. At repletion, triplicate pools of ten larvae were evaluated for spirochete burdens as detailed above.
Flat nymphs. Replete larvae were allowed to molt (~3-6 months post-repletion), and triplicate pools of ten molted, unfed nymphs were evaluated for spirochete burdens as noted above.
Fed nymphs. Fifteen infected nymphs were fed via the capsule feeding method [114] on the dorsum, at repletion nymphs were split into multiple pools of 3-5 nymphs per mouse and evaluated for spirochete burdens as detailed above. At two weeks post-feeding, mice were evaluated for infection as detailed above for tissue culture and skin sampling.

Statistics
All statistical analysis was performed using Prism 8 (GraphPad, Software, Inc.) with an unpaired Student's t test with two-tailed p values and a 95% confidence interval, and data are presented as mean ± standard error of the mean (SEM). p < 0.05 was considered statistically significant.

PLOS PATHOGENS
Host-specific functions of OppAs Supporting information S1 Fig. The oppA1 and oppA2 tn mutants display normal in vitro growth and  is missing lp56 and lp28-4, as previously published, as are all subsequent strains. oppA5tn has lost lp5, a commonly lost plasmid with no known effects on mammalian or tick infectivity. oppA1c is missing lp21, which is not required for mouse infectivity [120]. Little is known about the requirement for lp21 during tick infection, though it appears to be unnecessary due to wt levels of tick colonization by oppA1c. (TIF) S1

PLOS PATHOGENS
Host-specific functions of OppAs