Evidence That Two ATP-Dependent (Lon) Proteases in Borrelia burgdorferi Serve Different Functions

The canonical ATP-dependent protease Lon participates in an assortment of biological processes in bacteria, including the catalysis of damaged or senescent proteins and short-lived regulatory proteins. Borrelia spirochetes are unusual in that they code for two putative ATP-dependent Lon homologs, Lon-1 and Lon-2. Borrelia burgdorferi, the etiologic agent of Lyme disease, is transmitted through the blood feeding of Ixodes ticks. Previous work in our laboratory reported that B. burgdorferi lon-1 is upregulated transcriptionally by exposure to blood in vitro, while lon-2 is not. Because blood induction of Lon-1 may be of importance in the regulation of virulence factors critical for spirochete transmission, the clarification of functional roles for these two proteases in B. burgdorferi was the object of this study. On the chromosome, lon-2 is immediately downstream of ATP-dependent proteases clpP and clpX, an arrangement identical to that of lon of Escherichia coli. Phylogenetic analysis revealed that Lon-1 and Lon-2 cluster separately due to differences in the NH2-terminal substrate binding domains that may reflect differences in substrate specificity. Recombinant Lon-1 manifested properties of an ATP-dependent chaperone-protease in vitro but did not complement an E. coli Lon mutant, while Lon-2 corrected two characteristic Lon-mutant phenotypes. We conclude that B. burgdorferi Lons -1 and -2 have distinct functional roles. Lon-2 functions in a manner consistent with canonical Lon, engaged in cellular homeostasis. Lon-1, by virtue of its blood induction, and as a unique feature of the Borreliae, may be important in host adaptation from the arthropod to a warm-blooded host.


Introduction
In nature, Borrelia burgdorferi, the agent of Lyme disease [1,2], must adapt to disparate hosts, alternating between Ixodes ticks and various small rodent species. It is thought that this adaptation is made possible through the remodeling of the spirochete outer surface in response to environmental cues such as temperature [3,4,5], blood [6], pH [7,8,9,10], and microbial density [11,12,13]. A recognized example of this is the reciprocal expression of outer surface lipoprotein (Osp) AB and OspC. OspA and OspB are dominantly expressed when the spirochete is in culture or in the midgut of a flat unfed tick, then downregulated upon feeding and subsequent exposure to blood, increased temperature, and a drop in pH. OspC is concomitantly upregulated [3,14,15,16]. At this time the spirochetes experience a period of vigorous growth and migrate from the tick midgut to the salivary glands via the hemolymph [17]. This is followed by transmission to the mammalian host. The coordinated expression of OspAB and OspC has been proposed as an example of spirochete-vector interaction, with OspAB and OspC being implicated in spirochete adhesion to the tick midgut [18,19,20] and salivary gland respectively. At the time of feeding, abundant, surface exposed OspA and OspB need to be broken down to remodel the outer membrane. We have had a long-standing interest in the proteases of B. burdorferi and their functions. In the absence of known secreted proteases in the genome of B. burgdorferi [21], we have documented the reliance of this organism on borrowed proteolytic activity. The plasminogen activation system is used by the spirochete to cross cellular and extracellular matrices by inducing the production of and/or incorporating enzymatically active plasmin, urokinase plasminogen activator, and metalloproteases onto its surface [15,22,23,24,25] in both vectors and vertebrate hosts. While plasmin was important in promoting the migration of B. burgdorferi through the tick, this mammalian protease did not have an effect on the remodeling the outer surface of the spirochete at this crucial time. However, OspC, which is the upregulated lipoprotein during tick feeding is a plasminogen receptor [26] providing further indication that this system is associated with migration of the spirochete in the vector.
In a previous study, our interest in the proteolytic remodeling of outer surface lipoproteins during the transition of the spirochete from tick to mammalian host led us to examine the B. burgdorferi transcriptome after exposure in vitro to increased temperature in the presence and absence of blood [6]. The changes that we observed for the blood condition included the upregulation of OspC, other lipoproteins, and many genes of known or unknown function. One intriguing observation was the significant upregulation of a putative ATP-dependent protease La (Lon-1, BB0253) [21], a homolog of the Escherichia coli lon gene [27] and identical to the lon gene described for B. burgdorferi by Cloud et al. [28]. The B. burgdorferi genome also codes for a second putative ATP-dependent lon homolog, BB0613 (lon-2) [21], which was not differentially expressed in the array [6].
The canonical E. coli Lon (Lon-Ec), a conserved and much studied protease important for intracellular degradation of shortlived and abnormal proteins [27], has also been identified as a DNA-binding protein [29], and as a part of the heat shock regulon [30,31]. Besides bacteria, Lon homologs have also been found in other organisms including archaea [32], yeast [33,34,35], and animals [36,37]. It is an 87-kDa oligomeric serine protease with a structure consisting of three functional domains: the heterogeneous NH 2 -terminal (LAN) domain, the ATPase domain, and the C-terminal proteolytic domain. Structural studies have demonstrated that Lon is a ring-shaped complex composed of multiple identical 87-kDa subunits, with Lon-Ec functioning as a hexamer [38]. Lon plays an important role in homeostasis by targeting abnormal proteins and unstable regulatory proteins such as the SulA division regulator [39] and RcsA, the positive regulator of capsular polysaccharide [40,41], and degrading tmRNA-tagged proteins during trans translation [42]. In addition, Lon has been shown to be a virulence factor for infection in mice [43,44]. Our interest in B. burgdorferi Lon-1 stemmed from the idea that as a protease, it could take part in the remodeling of the spirochete outer surface during the period following a blood meal in the tick. With this in mind, we sought to identify functional roles for Lon-1 and Lon-2 through phylogenetic analysis, and biochemical and genetic means. Our results support the idea that Lon-2 fulfills the role of canonical Lon and is involved in cellular quality control and homeostasis, while Lon-1 plays a role in host adaptation during the transition from tick to mammal.

Results/Discussion
Spirochetal Lon-1 and Lon-2 are only distantly related to canonical Lon B. burgdorferi codes for two putative ATP-dependent Lon homologs designated Lon-1 and Lon-2 [21]. In a previous study, Lon-1 was upregulated in response to blood, while Lon-2 was not [6], though each is expressed by B. burgdorferi during culture in BSK-H medium ( Figure 1A). The presence of more than one lon gene homolog is rare in the eubacteriae, but appears to be a unique characteristic of the genus Borrelia, as the genomes for B. afzelii [45], B. garinii [46], B. recurrentis [47], B. duttoni [47], and B. hermsii [48] also code for two Lon homologs. Our interest in Lon-1 stemmed from its hypothetical function as a protease, and we hypothesized that it could be involved in the remodeling of spirochete outer surface lipoproteins during the transition from the tick to the mammal, thus warranting its investigation as a potential virulence factor. By extension, our hypothesis would imply that Lon-2 would likely function in a manner similar to that of Lon-Ec. The canonical Lon-Ec resides on the chromosome, adjacent to and downstream from two other ATP-dependent proteases, clpP and clpX, the proteolytic subunits comprising the ATP-dependent clpXP complex thought to share its role in cellular housekeeping. Examination of the chromosomal location of B. burgdorferi lon-2 reveals that it is arranged identically, immediately downstream of clpX and clpP, while Lon-1 is on a distant part of the chromosome ( Figure 1). The lon-2 arrangement shown for B. burgdorferi in Figure 1 is maintained in other Borrelia, while Treponema and Leptospira code for single lon genes that are not arranged in the same fashion as canonical lon (data not shown).
Because the similarity of the chromosomal location of Lon-2 to canonical E. coli lon suggested that they could serve a similar function, we tested the premise that Lon-1 of B. burgdorferi may have other functions that could be unique to spirochetes, as transcription of this gene was upregulated on blood induction. Our assessment of the premise began with comparison of amino acid sequences (defined by Pfam, PF00004, PF02190, and PF05362) of Borrelia Lons with the Lon sequences of other spirochetes and more distantly related bacteria through the construction of phylogenetic trees ( Figure 2). The trees we created included Lon amino acid sequences from spirochetes coding for two Lons (Lyme disease and relapsing fever Borreliae) and single Lons (Treponema, Leptospira, E. coli, and B. subtilis). In Figure 2A, a tree based on the entire amino acid sequences of Lon proteins is shown. One evident feature from the tree is that while all of the Lons are related to some extent, they began to diverge in the distant past. The Borrelia Lon-1s cluster together, indicating that they are closely related. Within the Lon-1 group, the relapsing fever Borreliae, consisting of B. recurrentis, B. duttoni, and B. hermsii form a subcluster, while the Lyme disease group, B. burgdorferi, B. afzelii, and B. garinii form another subcluster. Although distinct, Borrelia Lon-1 is also related to Lons of Treponema and Leptospira, suggesting a common ancestor. Lon-2 sequences from Lyme disease and relapsing fever Borreliae cluster separately from all the other bacteria, indicating a close relationship exclusive to the genus Borrelia, and form subclusters identical to those of Lon-1. We also performed the phylogenetic analyses using lon nucleotide sequences, with virtually identical results (data not shown).
The carboxy-terminal proteolytic domain and the ATPase domain of Lon are conserved across different bacterial species, whereas the NH 2 -terminal domain, which is responsible for substrate binding, is heterogeneous, and thus, the most useful for phylogenetic analysis. To determine a possible functional difference between Lon-1 and Lon-2, we tested the phylogenetic

Author Summary
Lyme disease, caused by the spirochete Borrelia burgdorferi, is the most prevalent arthropod-borne disease in North America. In nature, the bacterium oscillates between its tick vector host (Ixodes spp.) and small rodents (Peromyscus spp.). B. burgdorferi is able to persist in these two very different host environments by modulating the expression of surface lipoproteins proteins, or other proteins, in response to host factors or environmental cues such as temperature and pH. Our interest in this process led to the identification of a homolog of the E. coli ATP-dependent lon protease (lon-1) in B. burgdorferi that was upregulated in response to blood. The prototypical Lon of E. coli is a conserved protease important for the destruction of abnormal and short-lived proteins. B. burgdorferi is unusual in that it also codes for a second lon homolog, lon-2, that was not upregulated in response to blood. In this study, we sought to clarify the roles for Lon-1 and Lon-2 in B. burgdorferi. We present evidence that Lon-1 is an ATP-and Mg 2+ -dependent protease but does not function in a manner consistent with a prototypical Lon. Lon-2, however, functionally complemented Lon in E. coli. Thus, Lon-1 and Lon-2 appear to have distinct roles in B. burgdorferi; Lon-1 by virtue of its blood induction may be important in host adaptation, while Lon-2 is the functional homolog of E. coli Lon.
relatedness of the NH 2 -terminal domains of the bacteria analyzed in Figure 2A ( Figure 2B). Overall, the clustering pattern for the NH 2 -terminal domain sequences closely resembles that of the whole molecule, implying that the clustering is driven by differences in the NH 2 -terminal domain alone. Perhaps the most significant outcome of the phylogenetic analysis is that the NH 2terminal domains of Lon-1 and Lon-2 are shown to be especially dissimilar, which may correlate with substrate specificity. Thus, Lon-1 and Lon-2 may each recognize different substrates. Another significant outcome of the phylogenetic analysis is that neither Lon-1 nor Lon-2 are closely related to the canonical Lons of E. coli and B. subtilis, preventing us from making sequence-driven inferences about the function of either Borrelia Lon.

Recombinant Lon-1 is enzymatically active
To further examine B. burgdorferi Lon function, we characterized the Lon-1 gene product to study its structure and test it for biological activity. For this purpose, we generated wild-type rLon-1 (strains and PCR primers are given in Tables 1 and 2). Proteolytic domains of Lons typically contain a lysine-serine dyad critical for catalysis with the lysine appearing 43 residues downstream of the catalytic serine [38]. Inspection of the deduced Lon-1 amino acid sequence shows that it contains a serine residue a position 714 (Swiss-Prot predicted active site S714) and a lysine at position 757 (Swiss-Prot predicted active site K757), therefore, we also generated a recombinant with the putative catalytic serine714 mutated to alanine (S714A) to abolish proteolytic activity and confirm its identity as part of the catalytic dyad. Recombinant wild-type (rLon-1) and mutant Lon-1 (rLon-1 S714A ) was analyzed by SDS-PAGE and western blot ( Figure 3) to assess their purity and confirm their size. Figure 3 panels A and B show that rLon-1 and rLon-1 S714A were purified to homogeneity and their SDS-PAGE profiles are consistent with their predicted size (native Lon-1 has a predicted MW of 90,701 and a predicted MW of 92,864 with the attached His-tag).
To examine the catalytic activities of wild-type rLon-1 and rLon-1 S714A we conducted ATPase and caseinolytic assays ( Figure 4). Analysis of ATPase activity by a colorimetric dyebinding assay showed that both the rLon-1 and the rLon-1 S714A exhibited similar levels of activity ( Figure 4A). The ATPase activity of rLon-1 S714A confirmed that the ATPase and the proteolytic domains function independently and that introduction of a mutation into the proteolytic domain does not affect the ability of Lon-1 to hydrolyze ATP. To test for rLon-1 and rLon-1 S714A ATP-dependent proteolytic activity we measured the degradation of a-casein-fluorescein isothiocyanate (FITC-casein), in the presence of 10 mM ATP and 4 mM ATP [33,37,49]. In contrast to the ATPase assay, the rLon-1 S714A did not show activity levels above that of the buffer control while the wild-type rLon-1 degraded FITC-casein in a time-dependent manner ( Figure 4B). This confirms that through targeted mutagenesis of serine714, the catalytic serine residue was identified correctly. Omission of ATP from the assay verified that it is required for the degradation of FITC-casein by rLon-1, as would be expected for an ATPdependent protease ( Figure 4C, left panel).
The presence of Mg 2+ is critical for assembly of both E. coli and Mycobacterium smegmatis Lon subunits into proteolytic multimeric structures [50,51]. To investigate whether the caseinolytic activity of Lon-1 was Mg 2+ -dependent we carried out the caseinolytic assay with wild-type rLon-1 in the presence and absence of Mg 2+ (for the absence of Mg condition, MgCl 2 was omitted from the reaction buffer). In the presence of Mg 2+ the catalytic activity was retained, however, the absence of Mg 2+ reduced the activity to background levels ( Figure 4C, right panel).  This result implies that the assembly of Lon-1 monomers into multimeric structures is essential for proteolytic activity. We also examined the caseinolytic activity of wild-type rLon-1 over a gradient of pH conditions. The pH optimum for wild-type rLon-1 ranged from 9.0 to 9.5, which is similar to that of Lon-Ec [52] ( Figure 4D). Exhaustive attempts to overexpress rLon-2 proved unsuccessful, as the detergent-soluble product was proteolytically inactive, precluding further use.

Lon-1 does not degrade Borrelia-SsrA tagged reporter protein in vitro
A variety of in vivo Lon substrates have been identified to date, including SulA, RcsA, and UmuD [40,53,54,55]. In addition, rLon-Ec was reported to degrade SsrA-tagged proteins, a characteristic that could be directly examined in vitro [42]. SsrA (also known as tmRNA) is a unique RNA molecule that is highly conserved in Eubacteria and plays a major role in dealing with stalled ribosomes caused by aberrant mRNAs. In this process, SsrA co-translationally places a peptide tag on the carboxyl termini of the incomplete polypeptides. These marked polypeptides are recognized by cellular proteases, including Lon, and readily degraded [56]. The ssrA sequence of B. burgdorferi has been identified and the presence of SsrA RNA product has been confirmed [57]. To test whether Lon-1 can degrade proteins marked with a Borrelia-SsrA tag, we constructed and purified a reporter protein, l-CI-N-SsrA Bb , which carries a Borrelia-SsrA tag on the C-terminus. When purified product was analyzed, we observed two distinct protein bands of similar sizes, most likely artifacts of the purification procedure ( Figure 5B, low MW proteins). Analysis of each protein by MALDI-TOF mass spectroscopy showed that the higher molecular weight protein is l-CI-N-SsrA Bb containing a full-length SsrA tag, and the lower molecular weight is the same protein with a truncated SsrA tag ( Figure 5A). Next, we performed in vitro degradation assays using purified rLon-1 or rLon-Ec and the reporter protein to directly evaluate Lon activity. Assays showed that rLon-1 degraded neither tagged l-CI-N reporter nor truncated tagged l-CI-N reporter proteins ( Figure 5B,C). Interestingly, rLon-Ec could effectively degrade tagged l-CI-N reporter despite a significant amino acid sequence difference between Borrelia and E. coli SsrA tags. Truncated SsrA tagged reporter protein was not degraded by rLon-Ec, suggesting that it specifically recognized the SsrA tag, not the l-CI-N protein itself. Furthermore, rLon-Ec required ATP for the degradation of the l-CI-N-SsrA Bb reporter protein ( Figure 5C). To further analyze the protease activity of Lon in vitro, we used insulin as a substrate in an in vitro assay similar to that of the tagged l-CI-N reporter. Assays showed that both Lon-1 and Lon-Ec could effectively degrade insulin ( Figure 5D). These results suggest that unlike Lon-Ec, Lon-1 of B. burgdorferi is an active protease that does not recognize Borrelia-SsrA tagged proteins in vitro.

Mutant Lon-1 inhibits the aggregation of insulin b-chain
In addition to its function as an ATP-dependent protease, Lon exhibits a chaperone-like role in mammalian mitochondria, yeast, and bacteria [58,59,60,61,62]. Cloud et al. [28] reported that complementation of an E. coli lpxA temperature-sensitive mutant with sequence corresponding to the N-terminus and ATP-binding site of B. burgdorferi lon-1 allowed the mutant to grow at the nonpermissive temperature. One reason for the suppression of the temperature-sensitive phenotype could be a novel chaperone-like activity conferred to lon-1 through the absence of the proteolytic domain from the construct, thus allowing stabilization of the mutant lpxA gene. To further characterize Lon-1 we investigated its capacity to act as a chaperone by use of an in vitro assay employing bovine insulin. Insulin is a 6-kDa polypeptide consisting of two chains, A and B, connected by two disulfide bridges. Reduction of the molecule results in aggregation of the B-chain, while the A-chain remains soluble [63]. The chaperone assay is based upon the prevention of B-chain aggregation through interaction with an added chaperone. Assay results are measured via light scattering caused by insulin aggregates [59,61]. Therefore, to determine if rLon-1 possessed chaperone-like activity, we tested the ability of rLon1 and rLon-1 S714A to inhibit the aggregation of insulin B-chain under reducing conditions. Kinetic spectrophotometric analysis revealed that in the absence of added Lon protein (buffer only), an irreversible formation of Bchain aggregates was the result. Addition of rLon-1 S714A without ATP did not significantly affect aggregate formation ( Figure 6A,  top panel). However, addition of rLon-1 S714A in the presence of increasing amounts of ATP resulted in a dose-dependent inhibition of aggregate formation ( Figure 6A, lower panels). Wild-type rLon-1 enzymatically degraded the insulin and thus was not suitable for use in the assay ( Figure 6B).
The preservation of chaperone-like activity despite mutagenesis of the proteolytic site in Lon-1 is consistent with that found in yeast and mammalian Lon [58,60]. The critical requirement for ATP in our Lon-1 chaperone assay results is in agreement with that reported for chaperones in general [64] and for yeast Lon [58,60], but differs with one study in which the chaperone-like activity of a bacterial Lon was investigated. In that report, Lon from Brevibacillus thermoruber prevented insulin aggregation in an ATPindependent manner [59].
Protein stabilization and catalysis would seem to be opposing actions; nevertheless, ATP-dependent chaperone-proteases of the Clp and Lon families are well documented [62]. In the case of Lon-1, the demonstration of a chaperone-like activity in vitro, while interesting, does not definitively establish a role as a chaperone in vivo. On the other hand, differential recognition of potential substrates l-CI-N-SsrA Bb and insulin or casein by rLon-1 demonstrates an ability by the enzyme to discriminate, implying that it could act as a chaperone for one protein and degrade another.

Lon-2 complements an E. coli lon mutant
The implied redundancy of retaining two Lon genes necessarily raises the question of whether they are in actuality functionally equivalent, complementary, distinct, or overlapping, which may be reflected by similarities or differences in composition or expression pattern. B. burgdorferi Lons-1 and -2 are comparable in size (,90 kDa) and domain structure. However, Lon-2 may be more chemically similar to the archetypal Lon-Ec, as suggested by the similarity of their isoelectric points (Lon-2: pI = 6.04/Lon-Ec: pI = 6.32), and the similarity in its chromosomal arrangement to Lon-Ec. Lon-1, on the other hand, is an exceedingly basic protein (pI = 9.85). In our previous report, lon-1 was transcriptionally upregulated in vitro when exposed to blood and temperature shift in comparison to temperature shift alone. Lon-1 upregulation in response to temperature shift alone was not determined in this study [6]. However, in other studies, lon-1 was not activated by temperature [65,66], pH shift [66], or by cultivation in dialysis membrane chambers in rats [66,67]. In line with our own findings, none of the above microarray studies reported the differential expression of lon-2 as a result of any condition [65,66,67]. We further investigated the respective roles of lon-1 and lon-2 by testing experimentally whether either gene is functionally equivalent to lon-Ec. Use of a complementation strategy allowed us to investigate and compare the role(s) of lon-1 and lon-2 in an in vivo context and circumvent the inability to produce rLon-2. Lon-Ec specifically degrades RcsA, a positive activator for transcription of capsular polysaccharide (cps) genes [40,41]. Lon-deficiency in E. coli causes an accumulation of RcsA and a concommitant excess of CPS, resulting in a mucoid phenotype. Lon-1 or lon-2 was inserted into the multiple cloning site of plasmid pBAD24, to produce plasmids pBADlon-1 and pBADlon-2. Each plasmid was transformed into lon-deficient strain HDB98 (lon510), which contains a cpsB::lacZ fusion reporter [68]. Measurement of bgalactosidase activity allowed us to assess the ability of Lon-1 or Lon-2 to correct the defect in cpsB regulation by RscA. As shown in Figure 7A, after culture at 30uC in the presence of 0.1% arabinose, HDB98 containing pBADlon-1 (HDB98pBADlon-1) exhibited b-galactosidase activity at levels similar to parental HDB98 and HDB98 containing vector alone (HDB98pBAD24). In HDB98pBADlon-2, however, b-galactosidase activity levels were significantly reduced in comparison to HDB98, HDB98pBAD24, and HDB98pBADlon-1 (P,0.001). The reduction in b-galactosidase levels did not occur in the absence of arabinose ( Figure 7B). Expression of soluble Lon-1 and Lon-2 by E. coli containing plasmids pBADlon-1 and pBADlon-2 was verified by western blot (Figure 7C, D, E).
Another substrate for Lon is cell division inhibitor, SulA, which functions along with FtsZ protein to prevent disadvantageous cell division [39,69]. Accumulation of SulA due to a Lon deficiency in E. coli causes impaired cell division resulting in a filamentous morphology, a phenotype exhibited in culture by lon 510 mutant HDB98 (data not shown). In the presence of 0.2% arabinose, HDB98pBADlon-1 produced long filamentous forms similar to the bacteria grown without arabinose and to HDB98, while in HDB98pBADlon-2, arabinose induction of Lon-2 expression resulted in greatly reduced filament formation ( Figure 8A). Expression of Lon-1 and Lon-2 was verified by western blot (Figure 8B). The ability of Lon-2 and not Lon-1 to complement HDB98 for filamentation and dysregulation of cps implies a differential capacity for these proteases to degrade RscA and SulA and supports a Lon-Ec-like role for Lon-2 in Borreliae. We have shown that B. burgdorferi Lons-1 and -2 have characteristics in common with canonical Lon and with each other. Lon-1 protein displayed features of an ATP and Mg 2+dependent chaperone-protease, but lon-1 did not complement an E. coli lon mutant, while lon-2 corrected two distinct lon mutant phenotypes. Primary structural analysis showed that their respective ATPase and proteolytic domains are similar and characteristically Lon-like, but their substrate-binding domains are very different, suggesting that they may recognize divergent substrates. Taken together, this suggests that despite the structural similarities of Lon-1 and Lon-2, they appear to have distinct functional roles. The presence of Lon-1 as a unique feature of Borrelia would indicate that it has functions relevant to this genus as arthropod-transmitted pathogens. Specifically to B. burgdorferi, Lon-1 may function in host adaptation where it could be important as a protease during periods of rapid spirochete proliferation during a bloodmeal with its associated increases in protein biogenesis. It is also possible that during this period, Lon-1 could act as a chaperone for molecules that are not enzymatic substrates. Lon-2, on the other hand, functions as the canonical Lon, engaged in cellular homeostasis.

Ethics statement
All animal procedures were performed in compliance with the guidelines and with the approval of the Institutional Animal Care and Use Committee (IACUC) of Stony Brook University. Truncated Borrelia SsrA tag refers to the reporter protein carrying a partial SsrA tag (determined by mass spectroscopy). (B) In vitro proteolysis assay was performed as described and protein samples taken at indicated time points were resolved on 15% polyacrylamide Tris-tricine gel followed by Coomassie blue staining. Protein bands labeled as tagged and trunc. tagged refer to the l-CI-N reporter protein carrying a full-length or truncated Borrelia SsrA tag, respectively. Creatine kinase, present in the reaction mixture, was used as a loading control. (C) Protein gels were quantitatively analyzed using the Odyssey infrared imaging system (LI-COR) and remaining reporter proteins were calculated and presented as a percentage (average of three independent experiments). Error bars represent standard deviations. (D) Proteolysis assay of insulin was performed in a manner similar to l-CI-N reporter in vitro assays. doi:10.1371/journal.ppat.1000676.g005

Generation of polyclonal antisera to Lon-1 and Lon-2
For rabbit antiserum to Lon-1, New Zealand white rabbits were inoculated subcutaneously with 100 mg (400 ml) of recombinant Lon-1 in Freund's Complete Adjuvant. The rabbits were boosted three times over a period of a month (100 mg in 400 ml) in incomplete Freund's Incomplete Adjuvant followed by test a bleed and serological testing by ELISA and western blot to confirm reactivity to recombinant and native B. burgdorferi Lon-1. A final boost in PBS was given and the serum was collected four days later. For mouse antiserum to Lon-2, BALB/c mice were inoculated subcutaneously with 100 mg (200 ml) of recombinant Lon-2 in Complete Freund's Adjuvant. The mice were boosted subcutaneously (100 mg in 200 ml) two times with Freund's Incomplete Adjuvant followed by test a bleed and serological testing by ELISA and western blot to confirm reactivity to recombinant and native B. burgdorferi Lon-2. A final boost in PBS was given and the serum was collected three days later.

Site-directed mutagenesis of the Lon proteolytic domain
The QuicKChange II Site-Directed Mutagenesis Kit (Stratagene) and primers P3F and P4R (Table 2) were used to introduce a point mutation in the proteolytic domain of the lon-1 open reading frame, converting the catalytically active serine residue to alanine (S714A). The resulting construct in pET30a(+), pCK004.3 (Table 1) was confirmed by sequencing and used to transform expression host BL21 Star TM (DE3)pLysS. Soluble mutant rLon-1 (rLon-1 S714A ) was expressed and purified as described above for wild-type rLon-1.

In vitro proteolysis of Borrelia-SsrA tagged reporter protein
The purification of l-CI-N reporter protein was performed as described by Choy et al [42] with some modifications. Briefly, plasmid pPW500 [70], which encodes the l-cI-N-trpAt gene with no stop codon, was first modified by site directed mutagenesis. In this process, the trpAt terminator sequence of the reporter gene was replaced by the B. burgdorferi ssrA-tag sequence that was obtained from the tmRNA website (http://www.indiana.edu/ ,tmrna/). Thus, the final plasmid construct carried the l-cI-N gene with the B. burgdorferi ssrA-tag sequence (l-cI-N-ssrA Bb ) and was transformed into an E. coli clpP clpX lon triple mutant. Bacterial cells were cultured in LB broth at 37uC and reporter expression was induced for 3 h with 1 mM IPTG when cultures reached mid-log phase. Finally, the l-CI-N-SsrA Bb reporter protein was purified by affinity chromatography with the aid of an internal His 6 epitope as described [42]. The quality of the purified product was determined by SDS-PAGE followed by Coomassie blue staining, and the size of each protein band was verified by MALDI-TOF mass spectroscopy. In vitro proteolysis assays were carried out with a final concentration of 1.5 mM of either rLon-1 or rLon-Ec, 10 mM l-CI-N-SsrA Bb or 0.3 mg/ml insulin as substrates, an ATP regeneration system (50 mM creatine phosphate, 80 mg/ml creatine kinase [56], 4 mM ATP), and a minimal activity buffer (50 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 1 mM dithiothreitol). An additional 20 mM of dithiothreitol was included in assays with insulin. The reaction mixtures were incubated at 30uC and a fraction of each sample was collected at various time points for analysis by 15% Tristricine SDS-PAGE. Protein bands were quantified by using an Odyssey infrared imaging system (LiCor Biosciences, Lincoln NE).

Phylogenetic amino acid sequence analysis
Sequences and their annotations were obtained from the Swiss-Prot Protein knowledgebase. The Clustal W algorithm [71] was used for sequence alignments and Mega 3 software version 3.1 [72] for phylogenetic analyses. Neighbor joining methods [73] were used to build phylogenetic trees. Percentage support values were obtained through a bootstrap procedure. Analyses were also carried out using nucleotide sequences.

Complementation of E. coli lon mutant HDB98
Wild-type E. coli HDB97 and lon mutant HDB98 (lon510) were the gift of Dr. Harris D. Bernstein, Genetics and Biochemistry Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health. B. burgdorferi lon-1 was amplified from B. burgdorferi strain B31A3 [74] by PCR using primers P50F and P51R and cloned into the NcoI restriction site of plasmid pBAD24 [75] which confers ampicillin resistance (American Type Culture Collection, Manassas, VA). The insert was placed eight nucleotides downstream from the optimized ribosomal binding site to form plasmid pBADlon-1. Lon-2 was amplified by PCR using primers P52R F and P53R and cloned into a region between the NheI and KpnI restriction sites of pBAD24, also leaving eight nucleotides between the pBAD24 ribosomal binding site and the lon-2 ATG start site to yield plasmid pBADlon-2. HDB98 was subsequently transformed with pBADlon-1 and pBADlon-2 to form strains HDB98pBADlon-1 and HDB98pBADlon-2. Both constructs were confirmed by nucleotide sequencing.

b-galactosidase assay
Measurement of b-galactosidase activity of E. coli strains was done as previously described [76,77]. HDB97 and HDB98, which contain a cpsB::lacZ reporter fusion, were cultured in LB medium. HDB98pBAD24 (containing empty pBAD24 vector), HDB98pBADlon-1 and HDBpBADlon-2 were cultured in LB medium with ampicillin and 0.1% L-arabinose to induce expression of B. burgdorferi Lons. Activity units were calculated as follows:

Units~1000x
Abs 600 OD 600 xVxRT , where OD 600 C represents the optical density of the culture, V, the culture volume analyzed (ml), RT, the reaction time (min), and Abs 420 , the absorbance reading after color development. Each strain was analyzed in triplicate and the experiment was repeated a total of five times. Data was tested for statistical relevance using InStat 3.0 Software (GraphPad Software, Inc., San Diego, CA). Assays of enzymatic activity Hydrolysis of casein by B. burgdorferi rLon-1 was used as a measure of proteolytic activity and was based on the method of Twining et al. [78]. Preparations of rLon-1 and rLon-1 S714A contained 20 mM Tris-HCl, 0.5M NaCl, 0.5M imidazole, 0.02% NP40, and 10% glycerol. To assay for proteolytic activity, 5 mg of enzyme was incubated in 1.5 ml microcentrifuge tubes for 3 hours at 37uC with 100 mg of FITC-labeled casein, type I (Sigma). Final reaction conditions after addition of Lon proteins were: 50 mM Tris-HCl, pH 8, 10 mM MgCl 2 , 4 mM ATP, 25 mM NaCl, 25 mM imidazole, 0.001% NP40, and 0.5% glycerol in a total volume of 100 ml. In assays designed to test for the influence of Mg 2+ on rLon-1 proteolytic activity, 10 mM MgCl 2 was omitted from the reaction buffer. Proteolysis was terminated by the addition of 10 ml of bovine serum albumin (10 mg/ml stock solution) immediately followed by 100 ml of 10% trichloroacetic acid (TCA), and 10 minutes of incubation on ice. The reaction contents were then centrifuged at 16,0006 g to separate TCA insoluble material. For each sample replicate, 125 ml of the TCA soluble supernatant fraction was transferred to the well of a 96-well plate (Falcon -Becton Dickenson) and neutralized by the addition of 125 ml of 0.5 M CHES-Na, pH 12. Fluorescence was measured in a SpectraMax M2 microplate reader (Molecular Devices, Sunnyvale, CA) at wavelengths of 490 nm (excitation) and 525 nm (emission). The assay for measurement of optimum pH for casein degradation was carried out at 26uC for 16 hours and the following buffers were used to achieve the desired assay pH: 50 mM MES (pH 6.0, 6.5, 7.0), 50 mM Tris-HCl (pH 7.5, 8.0, 8.5), 50 mM 2-aminomethanepropanediol-HCl (AMPD) (pH 9.0, 9.5, 10.0), 50 mM CAPS (pH 10.5) [52].
The ATPase activity of rLon-1 and rLon-1 S714A was measured colorimetrically by use of a kit-based assay which utilized the dye malachite green to detect the liberation of inorganic phosphate resulting from enzymatic ATP hydrolysis (Innova Biosciences, Cambridge, UK). The manufacturer's protocol was followed for all assays and utilized 5 mg of enzyme per assay reaction. Components of the enzyme purification buffer resulted in final concentrations of 12.5 mM NaCl, 12.5 mM imidazole, 0.0005% NP40, and 0.25% glycerol in the reaction. Absorbances were measured in a SpectraMax M2 microplate reader at a wavelength of 635 nm.

Assay of chaperone-like activity
Reduction of its disulfide bonds induces the separation of insulin A and B chains with the attendant aggregation of the B chain [63]. The assay for chaperone-like activity is based upon the protection of reduced B-chain from aggregation by an added chaperone-like molecule. The degree of protection conferred is monitored by the measurement of resultant light scattering. In our assay, a modification of Farahbakhsh et al [79], 15 mg of bovine insulin was added to 25 mg of wild-type rLon-1 or mutant rLon-1 S714A protein in the presence of 10 mM sodium phosphate, pH 7.2, 20 mM dithiothreitol, 500 mM NaCl, 50 mM imidazole, 1% glycerol, 20 mM MgCl 2 , and either 4 mM or 10 mM ATP (total volume of 500 ml) in a 1-cm plastic cuvette. Light scattering was measured over time at 20uC in an Ultrospec 4000 UV/Vis spectrophotometer (Pharmacia Biotech, Piscataway, NJ) at a wavelength of 360 nm. A water-jacketed cell changer in conjunction with a heater/circulator (Neslab Instruments, Newington, NH) was used to ensure that the temperature was maintained. The data were collected and analyzed by SWIFT software in the kinetics mode (Pharmacia Biotech). Wild-type rLon-1 and mutant rLon-1 S714A were also analyzed for their ability to degrade reduced insulin under chaperone assay conditions. These digestions were carried out in 1.5 ml microcentrifuge tubes under the conditions described for the chaperone assay, with the exceptions that 150 mg of insulin was used. Incubation was done for 2 hours at 37uC.

SDS-PAGE and western blotting
Tris-glycine SDS-PAGE of recombinant B. burgdorferi Lon proteins (1 mg) was conducted as described previously [80] in gels of 12.5% polyacrylamide. For E. coli used in complementation experiments, 1 ml of culture was centrifuged and resuspended in 100 ml 16 SDS-PAGE sample buffer, and boiled. Samples (10 ml) were electrophoresed by 12.5% Tris-glycine SDS-PAGE. In all cases, electrophoretic transfer of protein was carried out over a period of 16 hours in 0.01M CAPS buffer, pH 11. For analysis of Lon-1 insulin degradation in the chaperone assay, 100 ml of reaction mixture was added to 50 ml of 36 SDS-PAGE sample buffer and boiled. For each condition, 10 ml was electrophoresed by Tris-tricine SDS-PAGE using a 16.5% separating gel, a 10% spacer gel, and a 4% stacking gel. Equivalent volumes of degraded tagged reporter protein (2.2 mg) and insulin (4.5 mg) from SsrA experiments were analyzed using a 15% Tris-tricine separating gel alone. For direct staining, the gels were fixed in 50% methanol-12% acetic acid then stained with 0.1% Coomassie blue R-250 in 50% methanol-12% acetic acid. For western blotting, Lon-1 and Lon-2 in lysates were detected using rabbit and mouse polyclonal antiserum, respectively. Infrared-conjugated secondary antibodies were goat anti-rabbit IgG (H+L) and goat anti-mouse IgG (H+L) (Rockland, Gilbertsville, PA). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was detected using goat polyclonal antibody (GenScript Corp., Piscataway, NJ) and donkey anti-goat IgG (H+L) secondary antibody. Bands were visualized by use of an infrared scanner (Odyssey Infrafred Imaging System, LiCor Biosciences).
Accession/ID numbers for proteins sequences mentioned in the text