Peptide Uptake Is Essential for Borrelia burgdorferi Viability and Involves Structural and Regulatory Complexity of its Oligopeptide Transporter

ABSTRACT Borrelia burgdorferi is an extreme amino acid (AA) auxotroph whose genome encodes few free AA transporters and an elaborate oligopeptide transport system (B. burgdorferi Opp [BbOpp]). BbOpp consists of five oligopeptide-binding proteins (OBPs), two heterodimeric permeases, and a heterodimeric nucleotide-binding domain (NBD). Homology modeling based on the crystal structure of liganded BbOppA4 revealed that each OBP likely binds a distinct range of peptides. Transcriptional analyses demonstrated that the OBPs are differentially and independently regulated whereas the permeases and NBDs are constitutively expressed. A conditional NBD mutant failed to divide in the absence of inducer and replicated in an IPTG (isopropyl-β-d-thiogalactopyranoside) concentration-dependent manner. NBD mutants grown without IPTG exhibited an elongated morphotype lacking division septa, often with flattening at the cell center due to the absence of flagellar filaments. Following cultivation in dialysis membrane chambers, NBD mutants recovered from rats not receiving IPTG also displayed an elongated morphotype. The NBD mutant was avirulent by needle inoculation, but infectivity was partially restored by oral administration of IPTG to infected mice. We conclude that peptides are a major source of AAs for B. burgdorferi both in vitro and in vivo and that peptide uptake is essential for regulation of morphogenesis, cell division, and virulence.

A mino acids (AAs) are essential for function and survival of the bacterial cell. In addition to providing the building blocks for protein synthesis, AAs are involved in peptidoglycan synthesis, nitrogen metabolism, energy generation, cell-cell communication, and environmental sensing (1,2). Cellular AA homeostasis can be maintained by de novo biosynthesis, interconversion from other AAs or products of intermediary metabolism, and acquisition from the environment (3,4). Free-living bacteria usually possess extensive biosynthetic pathways along with a large repertoire of free AA and/or di-, tri-, and oligopepeptide (Opp) transporters (5). Obligate pathogens, on the other hand, have varying capacities for de novo synthesis and interconversion, depending instead on repertoires of transporter evolutionarily tailored to exploit the range of nutrients available in the various niches that they inhabit (4,6). Opp transporters are more energy efficient than single-AA transporters, enabling the import of multiple AAs per molecule of ATP (7). The canonical Opp system consists of an oligopeptide-binding protein (OBP; OppA), a heterodimeric cytoplasmic membrane permease (OppBC), and a nucleotide-binding domain (NBD) heterodimer (OppDF) that drives transport by hydrolyzing ATP (7). In Gram-positive bacteria, OBPs are lipoproteins tethered to the external leaflet of the plasma membrane by N-terminal lipids and are capable of binding peptides as large as 35 AAs from the extracellular milieu, though only peptides of up to 18 AAs can be transported (7). In Gram-negatives, OBPs are periplasmic and nonlipidated and generally bind peptides of 2 to 5 AAs; the relatively small size of the peptide ligand is dictated by the dimensions of the porin channels through which they traverse the outer membrane permeability barrier (8). Although spirochetes have a double membrane ultrastructure resembling that of Gram-negatives, their OBPs are lipid modified as in Gram-positives, presumably to keep the binding protein in proximity to the permease to facilitate peptide import in these slowly growing organisms. According to structural data, OBPs contain an extended hinge region which facilitates binding of large substrates (7). The OBP binding cavity can accommodate a diverse range of peptides due to the presence of negatively charged residues lining their ligand binding pockets that interact with the peptide ligand backbone instead of specific AA side chains (9). However, some OBP homologues contain subtle structural modifications that allow binding of highly specific ligands (e.g., pheromones and muropeptides), which function as cues for the activation of signal transduction, cell competence, and gene regulation pathways (7). Conformational changes accompanying closure of the globular domains of the OBP around the ligand, a mechanism often described as a "Venus fly trap," enable the OBP to dock to its cognate permease (7).
Bacterial Opp systems contain a great amount of variability with respect to gene arrangement and component multiplicity. Escherichia coli contains a canonical opp locus (oppABCDF) for support of nutrition along with mppA (murein peptide permease A), encoding an OppA-like OBP orphan that recycles muropeptides derived from the turnover of peptidoglycan via the Opp permease (see Fig. S1 in the supplemental material) (10). Streptococcus pneumoniae harbors a canonical opp locus (amiA1BCDF) as well as two separately located oppA genes (aliA1 and aliA2) to import AAs for nutrition and the regulation of competence via the global regulator CodY (1,11). Lactococcus lactis uses two separate ABC transporters, Opp and Opt, arranged in noncanonical gene order (oppDFBCA and optSABCDF), to support exponential growth from milk-derived peptides; OptS and OptA also transport milk-derived signaling peptides to regulate opp expression (12). Staphylococcus aureus contains four complete opp loci (opp1ABCDF, opp2BCDF, opp3BCDFA, and opp4ADFBC). Of these, only opp3BCDFA is required for uptake of peptides under AA-limiting conditions (6). The spirochete Treponema denticola has a complicated Opp system involving multiple OBPs, permeases, and NBD proteins arranged in noncanonical order (13). This oral commensal has limited AA biosynthetic capacity and instead uses an array of proteases to create a peptide-rich microenvironment in gingival tissues and energy generation via fermentation (14). In contrast, the syphilis spirochete, T. pallidum, an invasive, obligate human pathogen with almost no AA biosynthetic capability, contains an orphan OBP with no discernible cognate transporter, apparently relying on an assortment of AA ABC transporters and symporters to meet its AA requirements (15).
Borrelia burgdorferi, the Lyme disease spirochete, must obtain essential nutrients from its arthropod vector and mammalian host (16,17). Bioinformatics analyses reveal that B. burgdorferi contains only a few free AA transporters, no capacity for de novo AA synthesis, and can interconvert only serine and glycine (Table 1). Collectively, these data suggest that the spirochete's elaborate Opp transport system (B. burgdorferi Opp [BbOpp]) (Fig. 1) is a principal means by which the spirochete satisfies its AA requirements. BbOpp consists of five OBPs (OppA1 to OppA5; three tandemly encoded in the chromosome and two encoded in separate plasmids), two heterodimer permeases (OppB1C1 and OppB2C2), and a heterodimer NBD domain (OppDF) (16,18,19). Since its first description by Bono et al. (18), studies have revealed that the BbOpp system can import peptides when expressed in a heterologous host (i.e., E. coli), that the OBPs have overlapping specificities for small (3-to-7 AA) peptides (20,21), and that differential expression of OBPs involves all of the spirochete's known major gene regulatory pathways (Rrp2/RpoN/RpoS, Hk1/Rrp1, and Rel Bb ) (18,(22)(23)(24)(25)(26)(27)(28)(29)(30). Infectivity studies performed with OBP transposon mutants have suggested that OppA2 (31) and OppA5 (32) are required for murine infection but that OppA1 is not (33). Nevertheless, despite this large body of work, we possess only a limited understanding of the contribution of this system to borrelial physiology. In this study, we utilized a combination of phylogenetics, structural biology, transcriptomics, and mutagenesis to develop a comprehensive understanding of how the individual Opp components function and of the overall importance of this system for spirochete survival in vitro and in vivo. Our approach has allowed us to identify unique as well as common features of the OBPs that help explain their collective ability to import a wide range of peptide substrates. By mutagenizing the ATP-hydrolyzing NBD domain to incapacitate the entire BbOpp system, we demonstrated that peptides are essential for spirochete growth in vitro, as well as within the mammalian host, a milieu in which free AAs should be available (34). Most surprisingly, we found that spirochetes deprived of peptides via ablation of the Opp system not only failed to replicate but also displayed an elongated morphotype and dysregulated septation. Moreover, many elongated organisms lacked a planar waveform near midcell due to an absence of flagellar filaments, suggesting abnormalities in synthesis or insertion of new flagellar apparatuses near future division sites. We conclude that peptides are a major source of AAs for B. burgdorferi both in vitro and in vivo and that peptide uptake serves an intracellular signaling function regulating morphogenesis and cell division. with OppA2 occupying its own node. OppA1 and OppA2 are further segregated into Lyme disease and relapsing fever sister groups. From these relationships, we surmise that the Opp system of Lyme disease spirochetes and other borrelial species evolved from a series of gene duplications. Structural features distinguish BbOppA4 from Gram-negative and Grampositive OBPs. We took advantage of the crystal structure of BbOppA4 (PDB ID: 4GL8) (39) to determine if B. burgdorferi OBPs possess structural elements compatible with promiscuous peptide binding, the functional hallmark of bacterial OBPs (40). The "closed" structure of BbOppA4 (Fig. S3A), determined at a resolution of 2.2 Å, displays a well-defined, bilobed globular structure (domains II and III) with an extended hinge region (domain I) (7); the latter, a characteristic feature of "cluster C" substrate-binding proteins, provides extra space within the ligand binding cavity to accommodate peptides (5). Domain I of BbOppA4 is formed from residues 30 to 62, 187 to 285, and 504 to 530 and contains seven centrally located ␤-sheets. Residues 63 to 186 form domain II, consisting of four stranded ␤-sheets and two ␣-helices. Residues 286 to 503 form domain III, which consists of four antiparallel ␤-strands sandwiched between three ␣-helices and two parallel ␤-strands with an ␣-helix. The final ␤-strand of domain III merges with the ␤-strand of domain I. In accord with the phylogenetic analysis, the lower C␣ root mean square deviation (RMSD) values of BbOppA4 with respect to E. coli OppA (EcOppA) (1.27 Å) and Salmonella enterica OppA (SeOppA) (1.18 Å), compared to L. lactis OppA (LlOppA) (6.57 Å), indicate that borrelial OBPs are structurally closer to their Gram-negative counterparts.
Fortuitously, the structure for BbOppA4 was liganded to an endogenous X-Ala-Ala-Ala tetrapeptide (X denotes an N-terminal AA with unresolved electron density after the C␥ atom) in an extended conformation (39). BbOppA4 contains a conserved Asp431 (Asp445 in E. coli and Asp419 in S. enterica) containing a side chain that hydrogen bonds to the amine group of the N-terminal residue of the peptide (Fig. S3B); this docking mechanism is characteristic of OBPs of Gram-negative bacteria (9). OppA from the Gram-positive L. lactis, in contrast, uses a large hydrophobic pocket in the ligand binding cavity to anchor a centrally located hydrophobic residue within the peptide (Fig. S3C) (40). BbOppA4 contains a similar hydrophobic pocket which could accommodate a hydrophobic residue at position 2 in a bound peptide (Fig. S3C) and, therefore, could be a determinant of ligand specificity. In the crystal structure, BbOppA4 engages its tetrapeptide via peptide backbone interactions (Fig. S3B). This finding indicates that BbOppA4 is capable of sequence-independent binding and suggests that cavity size and geometry define the range of ligands. EcOppA and SeOppA have the smallest binding cavities at 1,480 Å 3 and 1,600 Å 3 , respectively, and are known to bind ligands of 2 to 5 AAs ( Fig. S3C and D) (41). LlOppA has the largest cavity at 4,785 Å 3 , consistent with its known capacity to bind and transport peptides as large as 18 AAs (7). The intermediate cavity size of BbOppA4, 2,893 Å 3 ( Fig. S3C and D), suggests a capacity to bind peptides larger than tri-and tetrapeptides as shown elsewhere (21). Water molecules play a large role in determining OBP ligand size by filling empty space that can accommodate larger peptides and AA side chains. The presence of 26 water molecules within the cavity of BbOppA4 (Fig. S3C) further suggests that larger peptides can be accommodated. Moreover, unlike EcOppA and SeOppA, BbOppA4 contains a water-filled extended region of the cavity near the conserved Asp with the potential to accommodate long-chain AAs or other bulky moieties at position 1 (Fig. S3C, right side of cavity).
The charge distribution within the ligand binding cavity can influence the peptide repertoire of a given OBP. To assess the electrostatics of the BbOppA4 binding cavity, we modeled BbOppA4 against the unliganded or "open" structure of EcOppA (PDB ID: 3TCH). The EcOppA cavity is primarily negatively charged and tends to prefer positively charged peptides (9,42); BbOppA4 is lined with positively charged residues, suggesting a preference for negatively charged peptides (Fig. S3E).
Structural homology models explain diversity of peptide binding by B. burgdorferi OBPs. Using the BbOppA4 structure as a template, we created in silico structural homology models of the other B. burgdorferi OBPs to understand how variation in the OBP sequences could broaden the peptide ligand repertoire of the spirochete's Opp system. RMSD values based on C␣ deviations (0.05 to 0.07 Å) for the homology models demonstrate strong structural similarity to BbOppA4 (Fig. S4A). Four B. burgdorferi OBPs contain the conserved N-terminal docking aspartate (BbOppA1, Asp429; BbOppA2, Asp428; BbOppA4, Asp431; BbOppA5, Asp429), while BbOppA3 contains a glutamic acid at position 437, which would function analogously ( Fig. 2A and S4B). The multiplesequence alignment of B. burgdorferi OBPs (Fig. S4C) demonstrates that the putative ligand binding residues of the modeled OBPs align with the binding residues identified in the BbOppA4 structure. The projection of the BbOppA4 ligand into the BbOppA2 and BbOppA3 binding sites shows the greatest conservation of predicted ligand-interacting residues across the five OBPs, while BbOppA4, BbOppA1, and BbOppA5 contain one, two, and three unique substitutions, respectively, within their known or putative ligand-binding cavities. The variable residues are situated at different contact points along the peptide; BbOppA1 contains substitutions along the linear edge of the peptide, while BbOppA4 has a substitution at the C-terminal end of the peptide. There is variability in the distribution of hydrophobic residues along the pockets near residue 2 of the B. burgdorferi OBPs, suggesting a dichotomy with respect to the preferred residue(s) at position 2 of the liganded peptide (Fig. 2B). BbOppA2, BbOppA4, and BbOppA5 are predicted to accommodate bulky hydrophobic residues at this position, while BbOppA1 and BbOppA3 may accept both large hydrophobic and polar side chain residues. The volumes of the B. burgdorferi OBP binding cavities range from 2,174 Å 3 to 2,957 Å 3 ( Fig. 2C and D). The smallest of these, BbOppA2, has been shown to bind heptapeptides (21), suggesting that all B. burgdorferi OBPs can minimally bind peptides of seven AAs. The extended region near the conserved aspartate of BbOppA4 is present in BbOppA1, BbOppA3, and BbOppA5, but not BbOppA2 (Fig. 2B, right side of cavity). Modeling of the B. burgdorferi OBPs using the open conformation of EcOppA reveals variability in electrostatic distributions within the ligand binding cavity (Fig. 2D). The binding cavity of BbOppA1, like that of BbOppA4, is positively charged, while those of BbOppA2 and BbOppA5 are negatively charged. BbOppA3 is unique, presenting a neutral binding cavity. Projecting the AA variability of BbOppA1 to BbOppA5 onto the BbOppA4 structure, we find that the highest variability is within the ligand binding region, likely reflected by the variation in electrostatics, while the outer surface regions are more highly conserved (Fig. S5). Taken together, these comparisons strongly argue that the B. burgdorferi OBPs are not functionally redundant.
Closure of an OBP around its ligand creates a docking site for the cognate permease. The BbOppA2, BbOppA4, and BbOppA5 surface domains present predominantly negative surface charges, while the surfaces of BbOppA1 and BbOppA3 are more neutral and positively charged, respectively (Fig. 2D). These results suggest that the B. burgdorferi OBPs have differing affinities for the two permeases.
Differential expression of opp genes throughout the enzootic cycle. While the expression profiles of individual opp genes have been examined previously (23)(24)(25)(26)(27)(28)30), regulation of the system throughout the enzootic cycle remains poorly understood. To investigate this, we used quantitative reverse transcriptase PCR (qRT-PCR) to compare the transcript levels for all 11 opp genes in flat and feeding ticks and within the mammal using our dialysis membrane chamber (DMC) model (Fig. 3). Not surprisingly, spirochetes displayed a unique opp profile under each condition. As in other bacterial species (7), B. burgdorferi oppAs are transcribed at relatively high levels compared to other transporter components, presumably to provide the permeases with a steady stream of cargo. Previously, Bono et al. (18) and Wang et al. (30) reported that oppA1 to oppA3 can be transcribed as an operon or individually during in vitro cultivation. We detected relatively similar transcript levels for all three genes in fed larvae and nymphs, consistent with operonic transcription ( Fig. 3A and B), but substantially higher levels of oppA2 in flat nymphs and DMCs, suggesting individual regulation ( Fig. 3B and D). Expression of the plasmid-borne OBPs was host specific, with oppA4 expressed only in feeding ticks ( Fig. 3A and C) and oppA5 expressed only in mammals (Fig. 3D).

Loss of Opp transporter function abrogates spirochete growth in vitro.
While previous studies have confirmed that the B. burgdorferi oligopeptide transport (opp) system is capable of importing peptides (20,21), the importance of the system for spirochete viability and pathogenesis has not been established. We devised a genetic strategy to address this issue-deletion of the heterodimeric NBD domain (oppDF) to deprive the Opp system of its energy source. Repeated attempts to create oppDF deletion mutants were unsuccessful, suggesting that the BbOpp system is essential. As an alternative approach, we generated a conditional NBD mutant (BbAG132) in which expression of oppDF can be induced with IPTG (isopropyl-␤-D-thiogalactopyranoside) (Fig. 4A). To ensure tight repression, we utilized the IPTG-inducible construct developed by Blevins et al. (43), which contains two lac operator sequences. The mutant was grown in Barbour-Stoenner-Kelly II medium (BSK-II) supplemented with IPTG (0 to 1 mM) using starting inocula of 1 ϫ 10 3 spirochetes/ml, and triplicate cultures were enumerated for 6 days by dark-field microscopy (Fig. 4B). No growth was observed at IPTG concentrations below 0.08 mM. At 0.08 mM IPTG, there was a noticeable lag in growth, and spirochetes failed to reach peak density. At higher IPTG concentrations, organisms grew at wild-type levels. We next repeated the growth curves with higher inocula (~1 ϫ 10 6 spirochetes/ml) to allow enumeration of spirochetes throughout the period of incubation. We saw no replication at concentrations below 0.04 mM IPTG and a noticeable lag in growth with substantially lower peak densities at 0.04 and 0.06 mM IPTG (Fig. 4C). Last, we tested whether the addition of IPTG to uninduced mutant cultures would restore growth (Fig. 4D). Spirochetes were initially cultured to midlogarithmic phase with 1 mM IPTG, washed, and resuspended in BSK-II lacking IPTG. Addition of 1 mM IPTG on day 0 and day 1 after resuspension restored normal growth. Growth following addition of IPTG on day 2 was suboptimal, while addition of IPTG at later time points did not restore replication.
Peptide starvation induces dramatic morphological changes. Notably, NBD mutants grown without IPTG progressively elongated (Fig. 5A), with considerable heterogeneity in the degree of elongation at each day of observation (Fig. 5B). From day 2 onward, the differences in lengths between induced and uninduced mutants, measured using the Metamorph integrated morphometry analysis tool, were highly significant (P Յ 0.0001, Fig. 5B). By day 8, mutants grown without IPTG were nonmotile,  Table S1. and some showed a semitranslucent "ghostlike" appearance, indicative of nonviability (Fig. 5A, -IPTG D8). While some elongated spirochetes were helical along their entire lengths, others demonstrated loss of helicity near mid-cell (Fig. 5A). Spirochetes in the latter group displayed motility at the cell poles but propagated waveforms weakly or not at all toward the cell centers (see Movies S1 and S2 in the supplemental material). ability of IPTG-starved cultures to recover following the addition of IPTG. The mutant, grown to mid-logarithmic phase with 1 mM IPTG, was washed and inoculated into 15 ml of BSK-II (~1 ϫ 10 6 spirochetes/ml) without IPTG; 1 mM IPTG was added to starved cultures at daily intervals, following which growth was monitored by dark-field microscopy for 5 days. In all experiments, growth curve analyses were performed in triplicate. Dotted lines denote cultures with densities below the limits of detection by dark-field microscopy.
Peptide Uptake Essential for B. burgdorferi Viability ® The planar wave morphology of B. burgdorferi arises from the force exerted by rigid periplasmic flagellar filaments against the elastic peptidoglycan sacculus (44). We used transmission electron microscopy to determine whether the central flattening was due to the absence of flagella. Similar to ΔflaB spirochetes (45), cultures of the NBD mutant without IPTG contained numerous cross sections without outer membrane protrusions containing flagellar bundles (Fig. 5C). Regardless of whether a waveform was present, elongated mutants lacked discernible septal invaginations. We confirmed this by comparison with ΔflaB spirochetes, which septate but do not separate (45) (Fig. 5D). Elongated NBD mutant spirochetes that retained a planar waveform oscillated in segments of approximate wild-type length (see Movie S3). Together, these data confirm that peptide starvation results in gross morphological anomalies and perturbation of normal cell division events.
Peptide uptake is essential for viability, morphogenesis, and virulence within the mammalian host. We used the DMC system to assess the importance of the Opp transporter for viability within the mammal. Chambers inoculated with 1 ϫ 10 4 wild-type or NBD mutant spirochetes/ml (gently washed to remove residual IPTG) were implanted for 2 weeks. Wild-type organisms grew normally and were host adapted ( Fig. 6A and B), while no NBD mutants were visualized by dark-field microscopy. We next determined whether peptide-starved mutants cultivated in DMCs develop morphological abnormalities comparable to those observed in vitro. To ensure a sufficient density of organisms for evaluation by microscopy, we modified our standard protocol by (i) using a 10-fold-greater inoculum (1 ϫ 10 5 B. burgdorferi spirochetes/ml); (ii) inoculating DMCs with NBD mutant organisms cultured in vitro to mid-logarithmic phase with 1 mM IPTG; and (iii) explanting DMCs at an early time point (7 days) when viable organisms are more likely to be present. Spirochetes explanted from rats given IPTG-supplemented water appeared morphologically similar to wild-type spirochetes, while organisms recovered from animals given untreated water were markedly elongated, although with normal waveforms (Fig. 6A). SDS-PAGE revealed that the DMCcultivated spirochetes host adapted with or without IPTG supplementation (Fig. 6B).
Last, we assessed whether the Opp system is required for infectivity. Three days prior to syringe inoculation, we began administering IPTG-supplemented water to C3H/HeJ mice (n ϭ 8). On day 0, IPTG-treated and untreated mice were subjected to Peptide Uptake Essential for B. burgdorferi Viability ® needle inoculation with NBD mutants (1 ϫ 10 4 spirochetes). For both groups of mice, cultures of the inoculation site, ear, tibiotarsal joint, and heart collected 2 weeks postinoculation in BSK-II containing IPTG were negative. However, whereas sera from NBD-infected mice given plain water displayed no immunoreactivity, all eight mice given IPTG prior to and during infection seroconverted (Fig. 6C).

DISCUSSION
In most bacteria, Opp systems are ancillary and used to supplement pathways for biosynthesis of AAs and mechanisms for procuring them from the environment. The Lyme disease spirochete is an extreme AA auxotroph whose limited repertoire of AA transporters is unable to provide a full complement of these essential nutrients. Previous investigators surmised that the BbOpp system could play a central role in maintaining borrelial homeostasis (18,20,21,30). While facets of the BbOpp system have been explored (18,20,21,27,(31)(32)(33), its essentiality has never been demonstrated, nor have strategies been developed to clarify how its various components functionally interrelate. In this study, we confirmed that, even with the capacity to transport some free AAs, B. burgdorferi requires peptides for growth in vitro, within the mammal and, presumably, within the tick. Moreover, we show that the B. burgdorferi Opp system possesses a modular framework that enables adaptation to changing microenvironments via a combination of differential oppA expression and variations on the structural theme of sequence-independent peptide binding, the functional hallmark of prototypical OBPs (40).
Bacteria often encode multiple OBPs as a means of exploiting the full range of peptides available in a given niche (7); OBP diversity can be achieved by lateral gene transfer and/or gene duplication with subsequent sequence variation. Our phylogenetic and structural analyses, taken together, suggest that OBP diversity in Borrelia spp. arose by a combination of gene duplication events and mutations within domains II and III that modified the geometry and electrostatics of the ligand binding cavities as well as the surface residues that form the permease docking sites. Hu and colleagues (20,21) established an approximate size range (3 to 7 AAs) of the peptides bound by B. burgdorferi OBPs and demonstrated that individual B. burgdorferi OBPs have both overlapping and distinct specificities for peptide substrates. Given the combinatorial complexity of peptide-binding studies involving all five B. burgdorferi OBPs, we reasoned that a structural approach would inform our understanding of the binding preferences for individual B. burgdorferi OBPs as well as of their collective ability to meet the spirochete's AA requirements. The structure of BbOppA4 confirmed that, as with other bacterial OBPs (9), B. burgdorferi OBPs achieve ligand promiscuity by binding to peptide backbones. Then, using the BbOppA4 structure as a template, we demonstrated that modifications in cavity binding parameters, such as volume, configuration, and electrostatics, likely determine the ligand preferences of each OBP. Particularly striking is the variation in electrostatics of the binding cavities, implying that peptides are bound or excluded on the basis of their overall charge. We surmise, therefore, that each OBP accommodates a unique array of peptides that collectively enable the spirochete to meet its AA needs throughout the enzootic cycle, including the microenvironments that it occupies to persist within the mammal. We also noted differences in the surface electrostatics of the permease docking sites, most notably for BbOppA3, that likely alter the ability of a given OBP to compete for access to either (or both) permease(s).
As seen in previous studies (22)(23)(24)(25)(26)(27)(28), differential regulation of the BbOpp system primarily involves modulation of oppA gene expression. This transcriptional versatility derives, in large part, from the modular organization of the transporter genes within the B. burgdorferi genome. The expression profiles of oppA4 and oppA5, orphan OBPs encoded on separate genetic elements, represent excellent cases in point. oppA4 is expressed during the larval and nymphal blood meals and has been shown to be positively regulated by c-di-GMP, a signaling molecule generated by the Hk1/Rrp1 two-component system active in the feeding tick (23). oppA5, on the other hand, is absolutely RpoS dependent but is also repressed by c-di-GMP; these results, taken together, may explain why this gene is expressed exclusively in mammals (23,24,27,28). Among the five B. burgdorferi OBPs, OppA5 has the most divergent ligand binding cavity with respect to ligand binding residues, suggesting a distinct functionality during mammalian infection. The three chromosomally encoded OBPs are organized as an operon but exhibit expression profiles that clearly are not simply polycistronic. Bono et al. (18) identified mono-, bi-, and tricistronic transcripts by Northern blot analysis, while Wang et al. (30) confirmed that each BbOppA gene has its own promoter. oppA1 to oppA3 are expressed in the feeding tick at comparable levels, whereas only oppA2 is highly expressed in flat ticks and DMCs. A potential regulator of this locus is the alarmone (p)ppGpp, a product of Rel Bb which has been shown to modulate expression of oppA1, oppA2, and oppA3 and to be required for survival of spirochetes in ticks (22,25). Last, our finding that OppA2 is expressed at relatively high levels throughout the enzootic cycle strongly implies that this OBP is a functional centerpiece of BbOpp. This inference is buttressed by the fact that OppA2, whose ligand binding cavity features are highly conserved among Borrelia species, is the only full-length OBP in the louse-borne relapsing fever spirochete B. recurrentis. Furthermore, a transposon mutant for oppA2, but not oppA1, is attenuated for murine infection (31,33).
In recent years, investigators have made considerable inroads elucidating B. burgdorferi's mechanisms for carbon uptake and their regulation as carbon sources change during the enzootic cycle (31,(46)(47)(48)(49)(50)(51)(52)(53)(54). On the other hand, AA uptake by the spirochete has received much less attention and, not surprisingly, is much more poorly understood. Using a state-of-the-art mutagenesis technique (43,55,56) to target OppDF, the energy lynchpin of the Opp permease, we definitively demonstrated that peptides are essential for replication of B. burgdorferi in vitro and in vivo and that the Opp system is the principal if not the only means by which the bacterium can acquire them. Indeed, we saw a direct correlation between the concentration of IPTG used for induction of the NBD domain and spirochete replication in vitro. The relatively low expression levels of the NBD domains, determined by qRT-PCR analysis, likely explain why the growth rate of the constitutive mutant was exquisitely sensitive to small changes in the IPTG concentration. That limited replication occurred at a minimal concentration of IPTG (i.e., 0.04 mM) suggests a low threshold for stimulation of cell division by cytosolic peptides and/or AA pools.
Unexpectedly, in the absence of IPTG, cells not only ceased to replicate but progressively elongated, and wild-type growth patterns could not be restored following a relatively short period of cultivation without inducer. Thus, in contrast to the adaptive slowing of replication during peptide limitation, peptide starvation causes irreversible dysregulation of morphogenesis and cell division, a complex set of responses that are obviously maladaptive and not easily explained given our current knowledge of B. burgdorferi cell biology, physiology, and gene regulation. Residual NBD may enable some peptide transport during the initial period of cultivation without IPTG. A recent study using an inducible FtsH, an ATP-dependent cytoplasmic membrane-bound protease, demonstrated that spirochetes had to be cultured for 48 h without inducer for full degradation of preexisting protein (55). Even with complete cessation of peptide uptake, limited protein synthesis could be supported by uptake of some free AAs by dedicated transporters, "cannibalization" of preexisting proteins, and redirection of protein synthesis away from nonessential polypeptides. A striking feature of the peptide starvation phenotype is the inhibition of septum formation. Septum formation has long been known to be triggered by signals involving protein synthesis (57). More recently, direct links have been established between carbon and nitrogen metabolism and Z-ring formation (58,59). Thus, it is conceivable that depletion of intracellular peptide and/or AA pools results in failure to provide one or more of the metabolic signals essential for divisome assembly. Another fascinating aspect of the morphotype is the heterogeneity in helicity of the elongated organisms, which could be attributed to the presence or absence of flagellar filaments near mid-cell. Separation of dividing spirochetes requires motive force provided by newly assembled flagellar apparatuses flanking the division site (45). Perhaps some elongated spirochetes become "arrested" in a preseptation stage of the cell cycle such that new flagellar motors are not inserted (60). A diminished capacity to synthesize flagellar motor and filament proteins is likely also contributory. Experiments are under way to define the dysregulated steps in divisome and flagellar assembly and how they relate to the metabolic perturbations caused by peptide depletion. Given the poor growth and elongation of mutants cultivated in DMCs implanted into rats not given IPTG, we surmise that the same phenotype explains the avirulence of the NBD mutant and that incomplete complementation reflects the difficulty in maintaining adequate levels of IPTG by ad libitum oral administration, limited penetration of IPTG into tissues, and/or or suboptimal dosing.
The major implication of our study is that peptides must be available to support spirochete metabolism throughout the enzootic cycle. During both the larval and nymphal blood meals, when spirochetes undergo rapid expansion (61), salivary proteases produce a glut of peptides from host proteins, such as serum albumin and hemoglobin (62,63). As the molt progresses and spirochetes become quiescent, protein homeostasis is likely maintained by residual blood meal components (64). Our demonstration that peptides are an absolute requirement for B. burgdorferi in the mammal presents a conundrum regarding the source of the peptides, particularly within the noninflammatory milieu associated with persistence in the reservoir host, Peromyscus leucopus (65). The literature provides little to no evidence for the presence of free peptides in blood and interstitial fluid at concentrations presumably needed for bacterial replication. One can envision two complementary mechanisms by which B. burgdorferi could generate the requisite peptides. One is by secreting an extracellular protease, such as HtrA (66,67). The other is by decorating its surface with host-derived proteolytic enzymes, such as plasminogen and urokinase-type plasminogen activator (68,69). The paucibacillary nature of Lyme disease is well recognized and is the root cause for many of the diagnostic dilemmas associated with B. burgdorferi infection (70). It is tempting to speculate that the ability of B. burgdorferi to scavenge peptides, a physiological cornerstone of its persistence strategy, also serves as a novel form of in vivo growth regulation that facilitates its ability to "fly below the radar" of immune surveillance.

MATERIALS AND METHODS
Ethics statement. All animal experiments described here were performed in strict accordance with protocols reviewed and approved by the UConn Health Center Institutional Animal Care and Use Committee (Animal Welfare Assurance no. A347-01) and followed the recommendations of the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (71).
Sequence alignments and phylogeny. Multiple-sequence alignments of the five BbOppA proteins were generated using MUSCLE (78) and displayed with ESPript 3.0 (76). The MUSCLE output file was submitted to PhyML program (79) for phylogenetic analysis and the output phylogenetic tree generated was displayed using the Interactive Tree of Life (iTOL) (80).
Surface electrostatics representations of B. burgdorferi OBPs. To calculate surface charge, PDB files were converted to PQR files using the PDB2PQR server (87) and entered into the APBS Web server (88) for electrostatic calculations. The resulting electrostatic potential maps were visualized using the electrostatic surface coloring tool in UCSF-Chimera (86). Color saturation was reached at Ϫ5 kT/e to ϩ5 kT/e (k ϭ Boltzmann constant; T ϭ temperature; e ϭ charge on an electron). The default temperature for APBS is 298.15 K. Charges were assigned on each atom using the PDB2PQR server, which sets the pH value at 7.0. Final electrostatic potential maps were colored at a range of Ϫ7 kT/e to ϩ7 kT/e. Volume calculations for B. burgdorferi OBP ligand binding cavities. The CASTp server (89) was used to calculate binding cavity volumes using the closed structures of BbOppA4, EcOppA, and LlOppA and the closed homology models for the other B. burgdorferi OBPs. Cavity volumes were determined using a minimum solvent probe radius of 1.4 Å; the resulting images were visualized as cavities in UCSF-Chimera (86). The cavities of liganded EcOppA, LlOppA, and BbOppA4 are displayed as water filled.
Evolutionary conservation of B. burgdorferi OBPs by ConSurf analysis. The ConSurf Web server (consurf.tau.ac.il) (90) was used to project the evolutionary conservation of AA residues from B. burgdorferi B31 OppA1 to OppA5 onto the structure of BbOppA4. ConSurf employs an empirical Bayesian algorithm to determine position-specific evolutionary conservation scores. These conservation scores are converted into a discrete visual scale ranging from the most variable positions (grades 1 to 4) to the intermediately conserved positions (grades 5 and 6) and the most conserved positions (grades 7 to 9) and visualized in UCSF-Chimera.
Larval acquisition of B. burgdorferi and generation of fed nymphs for RNA extractions. Five-to-eight-week-old female C3H/HeJ mice (Jackson Laboratories, Bar Harbor, ME) were inoculated intradermally with 1 ϫ 10 4 temperature-shifted (23°C to 37°C) B31 5A18 NP1 spirochetes. At 3 weeks postinoculation, infected mice were used as a blood meal source for pathogen-free larvae (Oklahoma State University, Stillwater, OK) (200 to 300 per mouse). Following drop-off, replete larvae were allowed to molt to nymphs over supersaturated potassium sulfate in an environmental incubator. Twenty infected flat nymphs were applied to each naive C3H/HeJ mouse via the capsule feeding method (91) and allowed to feed to repletion. Nymphs were collected at drop-off for RNA extraction. Transmission of spirochetes to mice was confirmed 2 weeks postinfestation via ear culture and serology.
Generation of mammalian-host-adapted spirochetes in DMCs. Mammalian-host-adapted spirochetes for RNA were generated by cultivation in DMCs implanted into the peritoneal cavities of female Sprague-Dawley rats (175 to 200 g) and evaluated for host adaptation as previously described (92). To assess peptide requirements for growth within the mammalian host, wild-type B31 5A18 NP1 and BbAG132 spirochetes were cultivated at 37°C in BSK-II without and with 1 mM IPTG, respectively, and used for routine DMC implants as noted above. BbAG132 cultures were washed in phosphate-buffered saline (PBS) and resuspended in BSK-II prior to implantation to remove IPTG. To assess the morphology of peptide-starved mammalian-host-adapted spirochetes, DMCs were implanted with unwashed NBD mutant spirochetes at 1 ϫ 10 5 spirochetes/ml and incubated for 1 week. Three days prior to implantation, rats were given water alone or IPTG-treated water (2% sucrose solution containing 80 mM IPTG) (93) and were then maintained under the same condition for the remainder of the experiment.
Transcriptional analysis of opp genes. Total RNA was isolated as previously described (91) from samples of 200 replete larvae, 150 flat nymphs, 20 replete nymphs, and~5 ϫ 10 7 DMC-cultivated spirochetes. cDNAs, prepared with and without reverse transcriptase, were assayed for opp gene transcripts using the primer pairs listed in Table S2 in the supplemental material. Optimized amplification conditions for each gene were determined using SsoAdvanced Universal SYBR mix (Bio-Rad, Hercules, CA). Expression of opp genes was determined using a TaqMan-based assay and SsoAdvanced Universal Probe Mix (Bio-Rad) and normalized to flaB transcripts (94). All assays were performed in quadruplicate with three biological replicates. Internal standards for each assay were generated by cloning the corresponding amplicon into pCR-TOPO 2.1 (Invitrogen) using designated gene-specific primers (Table S2) according to the manufacturer's instructions.
(iii) Construction of the NBD mutant in B. burgdorferi. Electrocompetent B31 5A18 NP1 was transformed with 4 g of pEcAG137, and transformants were recovered in BSK-II under selection with Strep and Kan. Recovered clones were screened first by PCR for the Strep r gene (pless Strep F and pless Strep R) and then evaluated for plasmid content. A positive clone with wild-type plasmid content (BbAG128) was further confirmed to carry pEcAG137 by demonstrating recovery of the plasmid following transformation of total genomic DNA from BbAG128 into E. coli TOP10 recovery. Electrocompetent BbAG128 was subsequently transformed with 4 g of pEcAG138 and recovered overnight in BSK-II containing 1 mM IPTG. The following day, transformants were plated in BSK-II with Kan, Strep, Erm, and 1 mM IPTG. Recovered clones were first screened for Erm r by PCR (pless Erm F and pless Erm R); Erm r -positive clones were subsequently screened using bb0334 and bb0335 flanking primers 5= bb0334-35 null F1 and 3= bb0334-35 null F2. A positive clone with wild-type plasmid content (BbAG132) was further confirmed to carry pEcAG137 by demonstrating recovery of the plasmid following transformation of total genomic DNA into E. coli TOP10 recovery.
Growth curves. BbAG132 cultures were washed and inoculated into 15 ml of BSK-II (approximately 1 ϫ 10 3 or 1 ϫ 10 6 spirochetes/ml) with IPTG at concentrations ranging from 0 to 1 mM and incubated for up to 7 days at 37°C. For the recovery experiment (Fig. 4D), BbAG132 cultures were washed and inoculated into 15 ml of BSK-II (approximately 1 ϫ 10 6 spirochetes/ml) without IPTG. Each day (starting at day 0), 1 mM IPTG was added to 15-ml aliquots of washed BbAG132 and the reaction mixtures were incubated up to 5 days at 37°C. In all experiments, spirochetes (triplicate samples) were enumerated daily by dark-field microscopy using a Petroff-Hausser counting chamber (Hausser Scientific, Horsham, PA).
Evaluation of spirochete morphology and motility by light microscopy. Dark-field microscopy was performed on an Olympus BX41 microscope equipped with a Retiga Exi camera (QImaging, Surrey, British Columbia, Canada); images were acquired using a 40ϫ or a 100ϫ oil objective with QCapture software (v2.1; QImaging). All images were processed with ImageJ v1.50i (97). Videos (see Movies S1, S2, and S3 in the supplemental material) of spirochetes in BSK-II were acquired with a 40ϫ objective for 20 frames for a total duration of 2.755 s (0.13775 s per frame). Video frame rates were slowed to 10 frames/s using AVI FR Changer v1.10 (Inmatrix; http://inmatrix.com/files/avifrate_download.shtml). Videos can be viewed using Windows Media Player or VLC media player for Mac OS X (available from http://www .videolan.org/vlc/download-macosx.html).
Measurement of spirochete length. BbAG132 cultures were washed and inoculated into 15 ml of BSK-II (1 ϫ 10 6 spirochetes/ml) with 0 or 1 mM IPTG and incubated for 5 days at 37°C. Images (magnification of ϫ400) were acquired for samples daily. A minimum of 100 spirochete lengths for each sample on each day of observation were determined using the Metamorph (v7.8.2.0; Molecular Devices, LLC) integrated morphometry analysis tool, and the data were graphed as whisker plots using Prism software (v5.00; GraphPad Software, Inc., San Diego, CA).
Transmission electron microscopy. Samples were prepared for transmission electron microscopy as previously described (98). BbAG132 was cultured in BSK-II-1 mM IPTG to late-logarithmic growth and washed twice with PBS. Spirochetes were then resuspended and incubated in BSK-II with or without IPTG. After 7 days, spirochetes were washed in PBS and incubated overnight in 0.1 M sodium cacodylate (CAC) buffer containing 4% paraformaldehyde and 2.5% glutaraldehyde. Samples were then washed in CAC and centrifuged. The pellet was rinsed with fresh CAC and postfixed with 1% OsO 4 -0.8% potassium ferricyanide-CAC, followed by en bloc staining with 1% uranyl acetate-distilled water, dehydration in ascending ethanol solutions, and embedding in PolyBed812 (Polysciences, Warminster, PA). Ultrathin (70-nm-thick) sections were cut on an ultramicrotome (Leica), collected onto formvar-coated 200-mesh copper grids (Electron Microscopy Sciences, Hatfield, PA), stained with 6% uranyl acetate-50% methanol, washed, and then stained with 0.4% lead citrate. The stained sections were viewed on a Hitachi H-7650 transmission electron microscope at 80 kV of accelerating voltage. Images were acquired using an AMT camera and Image Capture Engine v6.01 software (Advanced Microscopy Techniques) and processed using ImageJ v.1.42 software (97).
Virulence testing of the conditional NBD mutant. Five-to-eight-week-old female C3H/HeJ mice (Jackson Laboratories, Bar Harbor, ME) were inoculated intradermally with 1 ϫ 10 4 temperature-shifted spirochetes. Testing of the NBD mutant was done in two independent experiments consisting of two cohorts of four mice each. Beginning 3 days prior to inoculation, cohorts of mice were given water alone or IPTG-treated water (2% sucrose solution containing 80 mM IPTG) (93) and then maintained on the same regimen throughout the duration of the study (93). Two weeks postinoculation, infection was evaluated by serology and tissues (inoculation site skin, ear, tibiotarsal joint, bladder, and heart) were collected for culture in BSK-II containing 1 mM IPTG.
Statistics. Levels of expression of individual opp genes under the different experimental conditions and with the different spirochete lengths were compared using Prism software using the unpaired t test with two-tailed P values and a 95% confidence interval.