Post by LymeEnigma on May 27, 2008 11:24:02 GMT -8
Outer Surface Protein B Is Critical for Borrelia burgdorferi Adherence and Survival within Ixodes Ticks
Girish Neelakanta1, Xin Li1, Utpal Pal1¤, Xianzhong Liu1, Deborah S. Beck1, Kathleen DePonte2, Durland Fish3, Fred S. Kantor2, Erol Fikrig1,3
Survival of Borrelia burgdorferi in ticks and mammals is facilitated, at least in part, by the selective expression of lipoproteins. Outer surface protein (Osp) A participates in spirochete adherence to the tick gut. As ospB is expressed on a bicistronic operon with ospA, we have now investigated the role of OspB by generating an OspB-deficient B. burgdorferi and examining its phenotype throughout the spirochete life cycle. Similar to wild-type isolates, the OspB-deficient B. burgdorferi were able to readily infect and persist in mice. OspB-deficient B. burgdorferi were capable of migrating to the feeding ticks but had an impaired ability to adhere to the tick gut and survive within the vector. Furthermore, the OspB-deficient B. burgdorferi bound poorly to tick gut extracts. The complementation of the OspB-deficient spirochete in trans, with a wild-type copy of ospB gene, restored its ability to bind tick gut. Taken together, these data suggest that OspB has an important role within Ixodes scapularis and that B. burgdorferi relies upon multiple genes to efficiently persist in ticks.
Author Summary
Lyme disease is the most common vector-borne disease in North America and Europe. The causative agent Borrelia burgdorferi is a bacterium that is maintained in an enzoonotic cycle between Ixodes ticks and a large range of mammals. Accidental encounters of infected Ixodes ticks with humans results in the transmission of B. burgdorferi and subsequent Lyme disease. Given that global control efforts have met with limited success, the need for developing novel interventions to combat this infection has become all the more vital. A better understanding of how B. burgdorferi interacts with its vector might lead to new ideas for combating the Lyme disease. B. burgdorferi upregulates outer surface protein (Osp) A and B during entry into ticks, and OspA contributes to the colonization of bacterium within the vector gut. We now demonstrate that OspB also facilitates the colonization and survival of B. burgdorferi in ticks. This work provides the basis for future studies as to how this protein facilitates interaction of B. burgdorferi to the tick gut and thus ultimately a basis for the development of novel strategies to interrupt the spirochete life cycle.
Introduction
Lyme disease is the most common tick-borne disease in the United States [1]. The causative organism, Borrelia burgdorferi, is a microaerophilic spirochete that contains a 910-kb linear chromosome and at least 21 linear and circular plasmids [2]. B. burgdorferi is composed of several genospecies known collectively as B. burgdorferi sensu lato (s.l.) of which B. burgdorferi sensu stricto, Borrelia garinii, and Borrelia afzelii are responsible for most cases of Lyme borrelliosis worldwide [3,4]. B. burgdorferi s.l. is maintained in an enzootic cycle that primarily involves Ixodes ticks and a large range of transmission-competent vertebrate hosts [1, 3–6]. Ticks of the Ixodes ricinus species complex, including Ixodes scapularis and Ixodes pacificus in eastern and western North America, respectively [3], and I. ricinus and Ixodes persulcatus in Europe and Eurasia [4], respectively, are competent vectors for the transmission of B. burgdorferi s.l. during engorgement on a reservoir host [1,3–6]. Studies reported so far tend to show similar means of transmission and modes of pathogenesis of the B. burgdorferi s.l. in this group of ticks [3–8]. After entry into the ticks, B. burgdorferi s.l. replicates and persists within the gut, then during a subsequent blood meal, migrates through the vector and is transmitted to a new host [9]. In humans, B. burgdorferi s.l. initially establishes a localized infection in the skin at the site of the tick bite known as erythema migrans, then disseminates via the blood stream and can chronically infect distant organs, resulting in arthritis, carditis, and neurological disease [1,10]. Laboratory mice can be infected with B. burgdorferi s.l. and serve as a reliable model for the study of Lyme borreliosis [11].
Variation in the synthesis of outer surface proteins (Osps) is a primary strategy by which B. burgdorferi evades the host immune system and adapts to various host microenvironments, such as those in a mammal or a tick vector [12–15]. Numerous studies have shown that B. burgdorferi selectively expresses specific Osps in distinct phases of its life cycle and in specific tissue locations. For example, the expression of B. burgdorferi OspA and OspB is immediately turned on when the spirochetes enter and reside within the arthropod vector. However, during transmission from the arthropod vector to a vertebrate host, B. burgdorferi downregulate OspA and OspB expression and upregulate the expression of proteins such as OspC, DbpA, and BBK32 [16–21]. This selective and temporal gene expression of OspA and OspB in ticks suggests that these two proteins may function during early spirochete colonization and persistence within the tick vector. Indeed, a recent study showing that OspA mediates spirochete adherence within the tick gut by binding to the I. scapularis TROSPA protein [22] supports this contention and indicates how stage-specific gene expression contributes to the maintenance of the natural cycle of the spirochete
The genes ospA and ospB are highly conserved among B. burgdorferi isolates in the United States [23,24]. They are encoded on the linear plasmid (lp) 54 and are generally expressed by a common promoter [25,26]. Both OspA and OspB are surface-exposed lipoproteins that are closely related in terms of sequence and structure [2,27,28]. Since the discovery of B. burgdorferi as the Lyme disease agent, OspA has been a subject of intensive investigation [29]. In contrast, less is known about the role of OspB in the life cycle of B. burgdorferi. Previous studies have identified an ospB escape mutant in a clonal population of infectious B. burgdorferi with a single base change in the consensus ribosomal binding sequence and a single nucleotide deletion in the open reading frame of ospB gene [30]. These changes in the ospB gene resulted in reduced expression and truncation of this protein that diminished the penetration capability and infectivity of the spirochete in the human umbilical vein endothelium cells [30]. Several reports have indicated that OspB is present on the surface of B. burgdorferi within unfed ticks [31–33] and that OspB antibody inhibits B. burgdorferi colonization in I. scapularis gut [34]. Targeted gene disruption of ospAB operon inhibited B. burgdorferi colonization and persistence in the tick gut [35]. While this study highlighted an important role of OspA in spirochete-tick interaction in vivo, the independent role of OspB in the life cycle of spirochetes remains unclear.
Progress in the methodologies for genetic manipulation of virulent B. burgdorferi allows researchers to study the importance of specific B. burgdorferi genes that are required throughout the spirochete life cycle [36]. In the current study, we have analyzed the functional role of OspB during spirochete-tick interactions by generating an OspB-deficient B. burgdorferi, genetically complementing the mutant, and performing in vivo studies in ticks and mice.
Results
Generation of OspB-Deficient B. burgdorferi
To understand the function of B. burgdorferi OspB, an isogenic OspB-deficient mutant was generated from infectious B. burgdorferi B31 clone 5A11 by replacing a 554-base pair (bp) internal fragment of the ospB gene with a Borrelia-adapted kanamycin resistance cassette, kanAn, through homologous recombination (Figure 1). The construct used to inactivate the ospB gene, pXLF11303, is schematically shown in Figure 1A. After electroporation of clone 5A11 with pXLF11303, three kanamycin-resistant B. burgdorferi transformants were obtained, of which two were confirmed to be OspB-deficient by immunoblot (unpublished data). The plasmid content of these two mutants was analyzed and compared to that of the wild-type parental clone using an array-based assay [37], which showed that one of the mutants lost cp9 and the other lost lp28–1. Because the loss of cp9 has been shown to have little effect on B. burgdorferi infectivity in mice [38], we we chose the ospB mutant lacking cp9 for further studies
To verify the desired genomic arrangements in the ospB mutant, a series of PCRs were performed (Figure 1B). The two PCRs using primer combinations N21/N27 and N28/N17 verified the left and the right junctions of the replacement in the ospB mutant. PCR with primer combination N21/N17 generated different sized DNA fragments from the wild-type strain and the ospB mutant, which is consistent with the replacement of the 554-bp internal fragment of the ospB gene with the 1,248-bp kanAn cassette. Collectively, these PCR results show that a double-crossover event had occurred in the mutant, resulting in the inactivation of the ospB gene. RT-PCR and immunoblot further showed that the ospB mutant lacked ospB mRNA and was OspB-deficient (Figure 1C and 1D). The ospB mutant had a similar total protein profile as well as comparable OspA mRNA and protein levels to that of the wild-type isolate (Figure 1C and 1D).
The OspB-Deficient B. burgdorferi Is Infectious and Pathogenic in Mice
To evaluate whether the loss of OspB expression affected the pathogenicity of the spirochete, we examined both the wild-type and the ospB mutant spirochetes in the murine model of Lyme borreliosis [39]. Groups of three C3H/HeN mice were challenged intradermally with in vitro-grown spirochetes, either the mutant or the wild-type B. burgdorferi, at a dose of 105 spirochetes/mouse (see Materials and Methods for details). At 25 d post-inoculation, mice were sacrificed and spirochete infection was assessed by serology and in vitro culturing of the bladder and the spleen. The results of three independent experiments using a total of 18 mice, nine infected with wild-type strain and the other nine infected with the mutant strain, indicated that all mice seroconverted and were culture-positive for spirochetes (unpublished data). The ospB mutant spirochetes recovered from murine tissues remained OspB-deficient, indicating that no reversion had occurred (Figure 1B, 1C, and 1D). Mice infected with the wild-type or the ospB mutant B. burgdorferi had similar spirochete burdens in various tissues, including the bladder, heart, joints, and skin (p > 0.05, Figure 2A). Joint swelling and inflammation were similar in both groups of mice (p > 0.05, Figure 2B). Taken together, these data indicate that at the dose of 105 spirochetes/mouse the ospB mutant
is fully infectious and pathogenic in mice.
The OspB-Deficient B. burgdorferi Enters but Cannot Colonize or Survive in I. scapularis
To determine whether the lack of OspB expression influenced the arthropod phase of the B. burgdorferi life cycle, uninfected nymphs were allowed to engorge on mice infected with either the wild-type or the OspB-deficient B. burgdorferi. Infection in mice was confirmed by positive flaB PCR from an ear punch biopsy. Three independent experiments were carried out with a total of 18 mice (nine infected with wild-type and the other nine infected with the mutant strain) and a total of 900 naïve nymphs (50 nymphs/mice). Five nymphs were removed from the murine skin at various time points during feeding, in order to examine the kinetics of spirochete migration, and the remaining nymphs were allowed to feed to repletion. Nymphs were subjected to Q-RT-PCR analyses to determine the spirochete burden (see Materials and Methods for details). At 8, 24, and 48 h during feeding, the viable spirochete burden in the ticks fed on the wild-type or the ospB mutant B. burgdorferi-infected mice were comparable (p > 0.05, Figure 3A). From 72 h (during feeding) onward, the ospB mutant spirochete levels were
dramatically reduced in ticks compared with controls (p < 0.03, Figure 3A). RT-PCR for flaB and ospA transcripts yielded similar results (Figure 3B)
The observed reduction of spirochetes in ticks fed on mice infected with the OspB-deficient B. burgdorferi was further evaluated by immunofluorescence analysis (IFA) of a subset of the nymphs that were collected at various time points. Tick gut luminal contents (including the blood meal) as well as gut tissues washed free of luminal contents were prepared and subjected to IFA (see Materials and Methods for details). At 72 h of tick feeding, a striking number of wild-type spirochetes were detected in both blood meal (13 spirochetes/microscopic field) and gut (8 spirochetes/microscopic field) tissue samples (Figure 3C and 3D). In contrast, the spirochete number was drastically reduced (4-fold in blood meal and 16-fold in gut tissue) in nymphs fed on the ospB mutant-infected mice (See 72-h panel in Figure 3C and 3D). Of note, the number of OspB-deficient spirochetes adhering to the tick gut was significantly less (7-fold, p < 0.0001) in comparison to the number of the OspB-deficient spirochetes in the blood meal sample (Figure 3D). IFA of ticks from the 48-h post-feeding phase further corroborated these findings (Figure 3C and 3D).
Furthermore, a subset of fully engorged nymphs was allowed to molt to the adult stage to determine whether the diminished capacity of ospB mutant to colonize the tick gut was also reflected in the adult stage. The wild-type spirochetes were readily detected in both luminal content and gut sample (Figure 3C and 3D). In contrast, ospB mutant spirochetes were not detected in the gut sample and out of 15 microscopic field observations only one ospB mutant spirochete was seen in the luminal content sample (p < 0.0001, Figure 3C and 3D). A quantitative RT-PCR analysis also corroborated the results (p < 0.02, Figure 3A). Thus, these observations collectively show that the ospB mutant spirochetes and the wild-type B. burgdorferi are acquired by ticks at the same rate (note results at nymphal stage 8, 24, and 48 h during feeding), but the ospB mutant spirochetes are
not able to persist within the luminal content or fully adhere to the tick gut
Complementation of the ospB Mutant
To further address whether the loss of OspB expression results in a defect of B. burgdorferi colonization and survival in the ticks, we constructed a strain of the mutant complemented in trans with a wild-type copy of the ospB gene. The promoter of the ospAB operon was fused to the ospB gene and cloned into the shuttle vector pKFSS1 [40], resulting in the construct designated as pFGN1 (Figure 1A). Electroporation of the ospB mutant with pFGN1 resulted in three positive clones that grew in the BSK-H media supplemented with streptomycin and kanamycin, of which one clone had lost both lp25 and lp28–1 plasmids and the other two clones showed identical endogenous plasmid profiles as its parental isolate (unpublished data). Furthermore, to determine whether these two clones harbored pFGN1 and were indeed from the ospB mutant, total DNA of these complemented clones was examined by PCR amplification. The three PCRs using primer combinations N21/N27, N28/N17, and N21/N17 confirmed the inactivated ospB locus (representative result shown in Figure 1B). PCR with primer combination N23/N86 generated different sized DNA fragments from the wild-type and the complemented strains, which confirmed the presence of PospAB-ospB gene fusion, and PCR amplification of the internal sequences of the aadA gene with primer combinations N83/N84 further confirmed the presence of pFGN1 in the transcomplemented strains (representative result shown in 1E). Collectively, these PCR results revealed the expected amplicons with both complemented clones in all the PCR reactions in comparison to the wild-type and the ospB mutant. We chose one of these two clones, designated as the OspB complemented strain (ospB−/pFGN1) for further analysis. OspB expression was detected at both the mRNA and the protein levels in the OspB complemented strain (Figure 1C and 1D). Furthermore, to confirm that the OspB complemented strain contained the pFGN1 plasmid, whole-cell lysates of kanamycin- and streptomycin-resistant cells were used to transform Escherichia coli DH5α-competent cells. Plasmid was then rescued from these E. coli transformants, and restrictive digestions were performed to verify the recovery of pFGN1 (Figure 1F). Immunoblot with mAb B22J was also performed to confirm the presence of OspB protein in these transformed E. coli cells (unpublished data).
Experimentally infected nymphs were prepared by microinjection of cultured spirochetes into the rectal aperture of uninfected nymphs as previously described [35]. Three independent experiments were carried out with a total of nine C3H/HeN naïve mice (three mice for feeding nymphs microinjected with wild-type, three for ospB mutant, and three for OspB complemented strain) and a total of 135 nymphs (15 nymphs/mice). Eight nymphs were forcibly removed during feeding (48 h) and analyzed by IFA and Q-RT-PCR. The results of the confocal microscopy revealed that, in contrast to the ospB mutant, the OspB complemented strain readily colonized the tick gut tissue (p < 0.0001), albeit to a lower level than the wild-type isolate (Figure 4A and 4B). No significant difference was seen in the blood meal samples of the nymphs infected with the wild-type, the ospB mutant, or the OspB complemented strain (Figure 4A and 4B). Quantitative RT-PCR analysis of cDNA samples also supported the microscopic observations that showed a significant higher level in the persistence of the OspB complemented strain in comparison to the ospB mutant (p < 0.001, Figure 4C).
The OspB-Deficient B. burgdorferi Show Reduced Binding In Vitro to I. scapularis Gut
To further support the role of OspB in the attachment of B. burgdorferi to tick gut tissue, we performed an in vitro binding assay to the tick gut extract (TGE) prepared separately from flat-nymphal ticks and fed-nymphal ticks with the wild-type, the ospB mutant, and the OspB complemented strains (See Materials and Methods for details). The results revealed that the binding of the ospB mutant to TGE from flat nymphs and fed nymphs is significantly reduced by 40% and 60%, respectively, in comparison to the wild-type B. burgdorferi (p < 0.0001, Figure 4D). In contrast, the OspB complemented strain showed a significant increase in the binding to both TGE in comparison to the ospB mutant (p < 0.0001) and was comparable to the wild-type spirochetes (Figure 4D). Taken together, these data from Figure 4 show that genetic complementation of the ospB mutant with a wild-type copy of the ospB gene restores the defects seen in the colonization and survival of the ospB mutant inside the ticks and can restore B. burgdorferi binding to the TGE.
Discussion
B. burgdorferi present an amazing variety of Osps that enable them to invade, colonize, and persist in environmental niches such as those inside vertebrates or ticks [23,26,41]. OspA, OspB, OspC, and DbpA are several of the major lipoproteins of B. burgdorferi that are differentially expressed in response to the varying environmental conditions [23,26,41]. B. burgdorferi upregulates OspA and OspB upon entry into ticks, and OspA contributes to the colonization of spirochetes within the vector gut [22]. Since ospA and ospB are cotranscribed [25,26] and colocalized on the bacterial surface [42], we speculated that OspB might also function for B. burgdorferi within ticks. To determine the precise role of OspB in the life cycle of B. burgdorferi, we have generated an OspB-deficient isogenic isolate of B. burgdorferi. Our data show that OspB facilitates the colonization and survival of B. burgdorferi within ticks.
While significant research has focused on the biological role of OspA in spirochete life cycle, relatively little information is available on the role of OspB in the life cycle of B. burgdorferi. Our in vivo studies with the OspB-deficient B. burgdorferi show that OspB is essential for the colonization and persistence of B. burgdorferi in ticks. During tick feeding, the ospB mutant and the wild-type B. burgdorferi enter the ticks from infected mice at the same rate. However, after feeding, the ospB mutant spirochetes are unable to persist within the blood meal or fully adhere to the tick gut, which also leads to a significant reduction in the number of spirochetes in the molted adult ticks (Figure 3). The binding of residual OspB-deficient spirochetes to the tick gut could be attributed to OspA, as the level of the OspA is unaltered in the OspB-deficient spirochetes (Figure 1). The luminal face of the gut epithelium is covered by a dense array of glycoproteins that may act as “receptor-buffet” for many pathogens [43]. Some of these glycoproteins are involved in general tissue structure and digestion [43,44] and some are involved in innate immunity [45,46]. Our surprising finding that the reduced ability of the ospB mutant to attach to the gut epithelium and its subsequent clearance in the gut may suggest that adherence to tick gut cells also is critical for some as yet unknown aspect(s) of spirochete viability.
Our in vivo analysis showed that in contrast to the ospB mutant, the OspB complemented strain readily colonized tick gut tissue and showed a drastic increase in its persistence within ticks, which was comparable to the wild-type isolate (Figure 4). Furthermore, our in vitro binding assays with the TGE also supported the in vivo analysis, indicating that in contrast to the ospB mutant, the transcomplemented strain binds with a greater affinity to the TGE. In addition, the difference between the wild-type and the ospB mutant spirochetes in binding to the fed TGE is significantly higher in comparison to the unfed TGE, suggesting that the levels of expression of putative OspB gut receptor proteins/glycolipids might increase during feeding. Furthermore, our in vitro binding data correlated with a previous study [34] showing that the B. burgdorferi N40 OspB protein binds significantly to the TGE. Overall, our studies solidify a great body of experimentation implicating an important role of OspB in the attachment of B. burgdorferi to the tick gut
Yang and co-workers recently examined the role of the ospAB locus in the infectious life cycle of B. burgdorferi [35]. This was accomplished by the generation of an ospAB double mutant from B. burgdorferi strain BbAH130 (infectious clone recovered after plating B. burgdorferi strain 297), and it was found that disruption of both the ospA and ospB genes had no observable effect on the ability of spirochetes to establish infection in mice, whereas the locus is critically essential for colonization of the tick gut [35]. Spirochetes deficient for both OspA and OspB entered ticks but were unable to persist within ticks for a long time [35]. Furthermore, complementation of the ospAB double mutant with both OspA and OspB expression restores the ability of B. burgdorferi to colonize the gut [35]. On the other hand, complementation with the ospA gene alone could only partially restore (50%–60% in comparison to the wild-type) the colonization defect of the ospAB mutant [35], suggesting that OspB expression is also required for the complete restoration of the defect. A comparison of our data with the prior study [35] indicated that in contrast to the ospAB double mutant complemented with the ospA gene alone, the ospB mutant analyzed in our study was significantly impaired in its persistence in the tick gut. These variations could have been the results of (i) the different B. burgdorferi strains used in the studies (B31 5A11 versus 297 BbAH130) and (ii) the relative OspA expression in the ospAB double mutant complemented with the ospA gene (on a circular plasmid) compared to the ospB mutant analyzed in our study. Overall, our studies in conjunction with the previous studies [35] show that (i) absence of OspB alone could result in severe impairment in colonization and persistence, and (ii) absence of both OspA and OspB could lead to the complete impairment in colonization and persistence of B. burgdorferi in ticks. In an evolutionary perspective, the conservation of ospB in the genome of B. burgdorferi is the result of positive selection pressure [23,24], and thus OspB must be of intrinsic value to the organism. Our studies suggest that the function of OspB and OspA are codependent
OspA and OspB share approximately 50% identity and 62% similarity in their amino acid sequences [2]. The crystal structures of OspA and the C-terminal region of OspB have been determined [27,28]. Comparison of the crystal structure of OspA and C-terminal region of OspB shows that these two molecules are quite similar [27,28]. The C-terminal region of OspB adopts the same fold as is observed for the C-terminal half of OspA [28]. Li and co-workers have identified that the C-terminal barrel domain in OspA is a trio of partially buried charged residues: Arg 139 from beta-strand 10, Glu-160 from beta-strand 12, and Lys-189 from beta-strand 15 [27]. The barrel domain of the OspA/B fold features a prominent cavity, in which the first two residues are strictly conserved in both OspA and OspB; position 189 is nearly always Lys in OspA and Arg in OspB [27,28]. Studies from Li et al. (1997) and Becker et al. (2005) have proposed that the cavity in the OspA/B barrel domain might be a ligand binding site for a small peptide, linear saccharide, or an exposed protein loop [27,28]. Furthermore, the mapping of amino acid sequences required for OspA binding to the tick gut showed that the residues 85–103 and 229–247 are important [47]. The percent similarity (identity) for the two amino acid stretches are 63 (68) and 84 (79) in OspB amino acid sequence at positions 110–128 and 252–270, respectively [34,47]. Given the structural and amino acid sequence conservation of OspA and OspB, it is possible that both lipoproteins recognize either the same target or closely related targets. It has recently been shown that OspB antibodies prevent B. burgdorferi colonization of I. scapularis gut [34]. Because of the high structural similarities between OspA and OspB, it could be reasoned that the OspB antibodies may bind to several epitopes of OspB on the B. burgdorferi surface, and steric hindrance might then interfere with OspB binding to the tick gut; or it is possible that steric hindrance by OspB antibodies also affected OspA-mediated binding of spirochetes to the tick gut [34]. Thus, our studies in conjunction with the previous reported study [34] raises interesting questions regarding the potential of antibody binding interfering with spirochete adherence in ticks. In the in vitro-grown B. burgdorferi cultures, the expression of OspB is lower than OspA (Figure 1D). Since OspA is also involved in the attachment of spirochetes to the tick gut and since we have found that OspB-deficient spirochetes are unable to attach to the tick gut, it is possible that disruption of the OspB resulted in the interference of the OspA-mediated attachment to TROSPA. Three scenarios may be envisioned that may elucidate the possible involvement of OspB in the OspA-TROSPA interactions. Firstly, OspB may directly associate with either OspA or TROSPA and may form a complex structure that is required for the tight attachment of B. burgdorferi to the tick gut. Secondly, OspB may bind to its own receptor within the gut and this interaction might be required for TROSPA to interact with OspA. Finally, OspA and OspB might bind to separate TROSPA molecules on the gut epithelium and both these interactions might be required for the tight attachment of B. burgdorferi to the tick gut. With any of these three models, our finding that OspB-deficient spirochetes were unable to colonize or persist in tick gut is significant because it suggests a possible synergistic interaction between OspA, OspB, and TROSPA.
In summary, these data suggest that OspB plays a critical role for B. burgdorferi adherence and persistence in ticks. These studies are not only important in understanding significant roles of spirochete ligands (such as OspB) in spirochete colonization and survival at arthropod-pathogen interface, but they also enhance our knowledge in the development of new therapeutic strategies, such as new transmission blocking vaccines that may be useful to combat B. burgdorferi infection.
pathogens.plosjournals.org/perlserv/?request=get-document&doi=10.1371%2Fjournal.ppat.0030033&ct=1
Girish Neelakanta1, Xin Li1, Utpal Pal1¤, Xianzhong Liu1, Deborah S. Beck1, Kathleen DePonte2, Durland Fish3, Fred S. Kantor2, Erol Fikrig1,3
Survival of Borrelia burgdorferi in ticks and mammals is facilitated, at least in part, by the selective expression of lipoproteins. Outer surface protein (Osp) A participates in spirochete adherence to the tick gut. As ospB is expressed on a bicistronic operon with ospA, we have now investigated the role of OspB by generating an OspB-deficient B. burgdorferi and examining its phenotype throughout the spirochete life cycle. Similar to wild-type isolates, the OspB-deficient B. burgdorferi were able to readily infect and persist in mice. OspB-deficient B. burgdorferi were capable of migrating to the feeding ticks but had an impaired ability to adhere to the tick gut and survive within the vector. Furthermore, the OspB-deficient B. burgdorferi bound poorly to tick gut extracts. The complementation of the OspB-deficient spirochete in trans, with a wild-type copy of ospB gene, restored its ability to bind tick gut. Taken together, these data suggest that OspB has an important role within Ixodes scapularis and that B. burgdorferi relies upon multiple genes to efficiently persist in ticks.
Author Summary
Lyme disease is the most common vector-borne disease in North America and Europe. The causative agent Borrelia burgdorferi is a bacterium that is maintained in an enzoonotic cycle between Ixodes ticks and a large range of mammals. Accidental encounters of infected Ixodes ticks with humans results in the transmission of B. burgdorferi and subsequent Lyme disease. Given that global control efforts have met with limited success, the need for developing novel interventions to combat this infection has become all the more vital. A better understanding of how B. burgdorferi interacts with its vector might lead to new ideas for combating the Lyme disease. B. burgdorferi upregulates outer surface protein (Osp) A and B during entry into ticks, and OspA contributes to the colonization of bacterium within the vector gut. We now demonstrate that OspB also facilitates the colonization and survival of B. burgdorferi in ticks. This work provides the basis for future studies as to how this protein facilitates interaction of B. burgdorferi to the tick gut and thus ultimately a basis for the development of novel strategies to interrupt the spirochete life cycle.
Introduction
Lyme disease is the most common tick-borne disease in the United States [1]. The causative organism, Borrelia burgdorferi, is a microaerophilic spirochete that contains a 910-kb linear chromosome and at least 21 linear and circular plasmids [2]. B. burgdorferi is composed of several genospecies known collectively as B. burgdorferi sensu lato (s.l.) of which B. burgdorferi sensu stricto, Borrelia garinii, and Borrelia afzelii are responsible for most cases of Lyme borrelliosis worldwide [3,4]. B. burgdorferi s.l. is maintained in an enzootic cycle that primarily involves Ixodes ticks and a large range of transmission-competent vertebrate hosts [1, 3–6]. Ticks of the Ixodes ricinus species complex, including Ixodes scapularis and Ixodes pacificus in eastern and western North America, respectively [3], and I. ricinus and Ixodes persulcatus in Europe and Eurasia [4], respectively, are competent vectors for the transmission of B. burgdorferi s.l. during engorgement on a reservoir host [1,3–6]. Studies reported so far tend to show similar means of transmission and modes of pathogenesis of the B. burgdorferi s.l. in this group of ticks [3–8]. After entry into the ticks, B. burgdorferi s.l. replicates and persists within the gut, then during a subsequent blood meal, migrates through the vector and is transmitted to a new host [9]. In humans, B. burgdorferi s.l. initially establishes a localized infection in the skin at the site of the tick bite known as erythema migrans, then disseminates via the blood stream and can chronically infect distant organs, resulting in arthritis, carditis, and neurological disease [1,10]. Laboratory mice can be infected with B. burgdorferi s.l. and serve as a reliable model for the study of Lyme borreliosis [11].
Variation in the synthesis of outer surface proteins (Osps) is a primary strategy by which B. burgdorferi evades the host immune system and adapts to various host microenvironments, such as those in a mammal or a tick vector [12–15]. Numerous studies have shown that B. burgdorferi selectively expresses specific Osps in distinct phases of its life cycle and in specific tissue locations. For example, the expression of B. burgdorferi OspA and OspB is immediately turned on when the spirochetes enter and reside within the arthropod vector. However, during transmission from the arthropod vector to a vertebrate host, B. burgdorferi downregulate OspA and OspB expression and upregulate the expression of proteins such as OspC, DbpA, and BBK32 [16–21]. This selective and temporal gene expression of OspA and OspB in ticks suggests that these two proteins may function during early spirochete colonization and persistence within the tick vector. Indeed, a recent study showing that OspA mediates spirochete adherence within the tick gut by binding to the I. scapularis TROSPA protein [22] supports this contention and indicates how stage-specific gene expression contributes to the maintenance of the natural cycle of the spirochete
The genes ospA and ospB are highly conserved among B. burgdorferi isolates in the United States [23,24]. They are encoded on the linear plasmid (lp) 54 and are generally expressed by a common promoter [25,26]. Both OspA and OspB are surface-exposed lipoproteins that are closely related in terms of sequence and structure [2,27,28]. Since the discovery of B. burgdorferi as the Lyme disease agent, OspA has been a subject of intensive investigation [29]. In contrast, less is known about the role of OspB in the life cycle of B. burgdorferi. Previous studies have identified an ospB escape mutant in a clonal population of infectious B. burgdorferi with a single base change in the consensus ribosomal binding sequence and a single nucleotide deletion in the open reading frame of ospB gene [30]. These changes in the ospB gene resulted in reduced expression and truncation of this protein that diminished the penetration capability and infectivity of the spirochete in the human umbilical vein endothelium cells [30]. Several reports have indicated that OspB is present on the surface of B. burgdorferi within unfed ticks [31–33] and that OspB antibody inhibits B. burgdorferi colonization in I. scapularis gut [34]. Targeted gene disruption of ospAB operon inhibited B. burgdorferi colonization and persistence in the tick gut [35]. While this study highlighted an important role of OspA in spirochete-tick interaction in vivo, the independent role of OspB in the life cycle of spirochetes remains unclear.
Progress in the methodologies for genetic manipulation of virulent B. burgdorferi allows researchers to study the importance of specific B. burgdorferi genes that are required throughout the spirochete life cycle [36]. In the current study, we have analyzed the functional role of OspB during spirochete-tick interactions by generating an OspB-deficient B. burgdorferi, genetically complementing the mutant, and performing in vivo studies in ticks and mice.
Results
Generation of OspB-Deficient B. burgdorferi
To understand the function of B. burgdorferi OspB, an isogenic OspB-deficient mutant was generated from infectious B. burgdorferi B31 clone 5A11 by replacing a 554-base pair (bp) internal fragment of the ospB gene with a Borrelia-adapted kanamycin resistance cassette, kanAn, through homologous recombination (Figure 1). The construct used to inactivate the ospB gene, pXLF11303, is schematically shown in Figure 1A. After electroporation of clone 5A11 with pXLF11303, three kanamycin-resistant B. burgdorferi transformants were obtained, of which two were confirmed to be OspB-deficient by immunoblot (unpublished data). The plasmid content of these two mutants was analyzed and compared to that of the wild-type parental clone using an array-based assay [37], which showed that one of the mutants lost cp9 and the other lost lp28–1. Because the loss of cp9 has been shown to have little effect on B. burgdorferi infectivity in mice [38], we we chose the ospB mutant lacking cp9 for further studies
To verify the desired genomic arrangements in the ospB mutant, a series of PCRs were performed (Figure 1B). The two PCRs using primer combinations N21/N27 and N28/N17 verified the left and the right junctions of the replacement in the ospB mutant. PCR with primer combination N21/N17 generated different sized DNA fragments from the wild-type strain and the ospB mutant, which is consistent with the replacement of the 554-bp internal fragment of the ospB gene with the 1,248-bp kanAn cassette. Collectively, these PCR results show that a double-crossover event had occurred in the mutant, resulting in the inactivation of the ospB gene. RT-PCR and immunoblot further showed that the ospB mutant lacked ospB mRNA and was OspB-deficient (Figure 1C and 1D). The ospB mutant had a similar total protein profile as well as comparable OspA mRNA and protein levels to that of the wild-type isolate (Figure 1C and 1D).
The OspB-Deficient B. burgdorferi Is Infectious and Pathogenic in Mice
To evaluate whether the loss of OspB expression affected the pathogenicity of the spirochete, we examined both the wild-type and the ospB mutant spirochetes in the murine model of Lyme borreliosis [39]. Groups of three C3H/HeN mice were challenged intradermally with in vitro-grown spirochetes, either the mutant or the wild-type B. burgdorferi, at a dose of 105 spirochetes/mouse (see Materials and Methods for details). At 25 d post-inoculation, mice were sacrificed and spirochete infection was assessed by serology and in vitro culturing of the bladder and the spleen. The results of three independent experiments using a total of 18 mice, nine infected with wild-type strain and the other nine infected with the mutant strain, indicated that all mice seroconverted and were culture-positive for spirochetes (unpublished data). The ospB mutant spirochetes recovered from murine tissues remained OspB-deficient, indicating that no reversion had occurred (Figure 1B, 1C, and 1D). Mice infected with the wild-type or the ospB mutant B. burgdorferi had similar spirochete burdens in various tissues, including the bladder, heart, joints, and skin (p > 0.05, Figure 2A). Joint swelling and inflammation were similar in both groups of mice (p > 0.05, Figure 2B). Taken together, these data indicate that at the dose of 105 spirochetes/mouse the ospB mutant
is fully infectious and pathogenic in mice.
The OspB-Deficient B. burgdorferi Enters but Cannot Colonize or Survive in I. scapularis
To determine whether the lack of OspB expression influenced the arthropod phase of the B. burgdorferi life cycle, uninfected nymphs were allowed to engorge on mice infected with either the wild-type or the OspB-deficient B. burgdorferi. Infection in mice was confirmed by positive flaB PCR from an ear punch biopsy. Three independent experiments were carried out with a total of 18 mice (nine infected with wild-type and the other nine infected with the mutant strain) and a total of 900 naïve nymphs (50 nymphs/mice). Five nymphs were removed from the murine skin at various time points during feeding, in order to examine the kinetics of spirochete migration, and the remaining nymphs were allowed to feed to repletion. Nymphs were subjected to Q-RT-PCR analyses to determine the spirochete burden (see Materials and Methods for details). At 8, 24, and 48 h during feeding, the viable spirochete burden in the ticks fed on the wild-type or the ospB mutant B. burgdorferi-infected mice were comparable (p > 0.05, Figure 3A). From 72 h (during feeding) onward, the ospB mutant spirochete levels were
dramatically reduced in ticks compared with controls (p < 0.03, Figure 3A). RT-PCR for flaB and ospA transcripts yielded similar results (Figure 3B)
The observed reduction of spirochetes in ticks fed on mice infected with the OspB-deficient B. burgdorferi was further evaluated by immunofluorescence analysis (IFA) of a subset of the nymphs that were collected at various time points. Tick gut luminal contents (including the blood meal) as well as gut tissues washed free of luminal contents were prepared and subjected to IFA (see Materials and Methods for details). At 72 h of tick feeding, a striking number of wild-type spirochetes were detected in both blood meal (13 spirochetes/microscopic field) and gut (8 spirochetes/microscopic field) tissue samples (Figure 3C and 3D). In contrast, the spirochete number was drastically reduced (4-fold in blood meal and 16-fold in gut tissue) in nymphs fed on the ospB mutant-infected mice (See 72-h panel in Figure 3C and 3D). Of note, the number of OspB-deficient spirochetes adhering to the tick gut was significantly less (7-fold, p < 0.0001) in comparison to the number of the OspB-deficient spirochetes in the blood meal sample (Figure 3D). IFA of ticks from the 48-h post-feeding phase further corroborated these findings (Figure 3C and 3D).
Furthermore, a subset of fully engorged nymphs was allowed to molt to the adult stage to determine whether the diminished capacity of ospB mutant to colonize the tick gut was also reflected in the adult stage. The wild-type spirochetes were readily detected in both luminal content and gut sample (Figure 3C and 3D). In contrast, ospB mutant spirochetes were not detected in the gut sample and out of 15 microscopic field observations only one ospB mutant spirochete was seen in the luminal content sample (p < 0.0001, Figure 3C and 3D). A quantitative RT-PCR analysis also corroborated the results (p < 0.02, Figure 3A). Thus, these observations collectively show that the ospB mutant spirochetes and the wild-type B. burgdorferi are acquired by ticks at the same rate (note results at nymphal stage 8, 24, and 48 h during feeding), but the ospB mutant spirochetes are
not able to persist within the luminal content or fully adhere to the tick gut
Complementation of the ospB Mutant
To further address whether the loss of OspB expression results in a defect of B. burgdorferi colonization and survival in the ticks, we constructed a strain of the mutant complemented in trans with a wild-type copy of the ospB gene. The promoter of the ospAB operon was fused to the ospB gene and cloned into the shuttle vector pKFSS1 [40], resulting in the construct designated as pFGN1 (Figure 1A). Electroporation of the ospB mutant with pFGN1 resulted in three positive clones that grew in the BSK-H media supplemented with streptomycin and kanamycin, of which one clone had lost both lp25 and lp28–1 plasmids and the other two clones showed identical endogenous plasmid profiles as its parental isolate (unpublished data). Furthermore, to determine whether these two clones harbored pFGN1 and were indeed from the ospB mutant, total DNA of these complemented clones was examined by PCR amplification. The three PCRs using primer combinations N21/N27, N28/N17, and N21/N17 confirmed the inactivated ospB locus (representative result shown in Figure 1B). PCR with primer combination N23/N86 generated different sized DNA fragments from the wild-type and the complemented strains, which confirmed the presence of PospAB-ospB gene fusion, and PCR amplification of the internal sequences of the aadA gene with primer combinations N83/N84 further confirmed the presence of pFGN1 in the transcomplemented strains (representative result shown in 1E). Collectively, these PCR results revealed the expected amplicons with both complemented clones in all the PCR reactions in comparison to the wild-type and the ospB mutant. We chose one of these two clones, designated as the OspB complemented strain (ospB−/pFGN1) for further analysis. OspB expression was detected at both the mRNA and the protein levels in the OspB complemented strain (Figure 1C and 1D). Furthermore, to confirm that the OspB complemented strain contained the pFGN1 plasmid, whole-cell lysates of kanamycin- and streptomycin-resistant cells were used to transform Escherichia coli DH5α-competent cells. Plasmid was then rescued from these E. coli transformants, and restrictive digestions were performed to verify the recovery of pFGN1 (Figure 1F). Immunoblot with mAb B22J was also performed to confirm the presence of OspB protein in these transformed E. coli cells (unpublished data).
Experimentally infected nymphs were prepared by microinjection of cultured spirochetes into the rectal aperture of uninfected nymphs as previously described [35]. Three independent experiments were carried out with a total of nine C3H/HeN naïve mice (three mice for feeding nymphs microinjected with wild-type, three for ospB mutant, and three for OspB complemented strain) and a total of 135 nymphs (15 nymphs/mice). Eight nymphs were forcibly removed during feeding (48 h) and analyzed by IFA and Q-RT-PCR. The results of the confocal microscopy revealed that, in contrast to the ospB mutant, the OspB complemented strain readily colonized the tick gut tissue (p < 0.0001), albeit to a lower level than the wild-type isolate (Figure 4A and 4B). No significant difference was seen in the blood meal samples of the nymphs infected with the wild-type, the ospB mutant, or the OspB complemented strain (Figure 4A and 4B). Quantitative RT-PCR analysis of cDNA samples also supported the microscopic observations that showed a significant higher level in the persistence of the OspB complemented strain in comparison to the ospB mutant (p < 0.001, Figure 4C).
The OspB-Deficient B. burgdorferi Show Reduced Binding In Vitro to I. scapularis Gut
To further support the role of OspB in the attachment of B. burgdorferi to tick gut tissue, we performed an in vitro binding assay to the tick gut extract (TGE) prepared separately from flat-nymphal ticks and fed-nymphal ticks with the wild-type, the ospB mutant, and the OspB complemented strains (See Materials and Methods for details). The results revealed that the binding of the ospB mutant to TGE from flat nymphs and fed nymphs is significantly reduced by 40% and 60%, respectively, in comparison to the wild-type B. burgdorferi (p < 0.0001, Figure 4D). In contrast, the OspB complemented strain showed a significant increase in the binding to both TGE in comparison to the ospB mutant (p < 0.0001) and was comparable to the wild-type spirochetes (Figure 4D). Taken together, these data from Figure 4 show that genetic complementation of the ospB mutant with a wild-type copy of the ospB gene restores the defects seen in the colonization and survival of the ospB mutant inside the ticks and can restore B. burgdorferi binding to the TGE.
Discussion
B. burgdorferi present an amazing variety of Osps that enable them to invade, colonize, and persist in environmental niches such as those inside vertebrates or ticks [23,26,41]. OspA, OspB, OspC, and DbpA are several of the major lipoproteins of B. burgdorferi that are differentially expressed in response to the varying environmental conditions [23,26,41]. B. burgdorferi upregulates OspA and OspB upon entry into ticks, and OspA contributes to the colonization of spirochetes within the vector gut [22]. Since ospA and ospB are cotranscribed [25,26] and colocalized on the bacterial surface [42], we speculated that OspB might also function for B. burgdorferi within ticks. To determine the precise role of OspB in the life cycle of B. burgdorferi, we have generated an OspB-deficient isogenic isolate of B. burgdorferi. Our data show that OspB facilitates the colonization and survival of B. burgdorferi within ticks.
While significant research has focused on the biological role of OspA in spirochete life cycle, relatively little information is available on the role of OspB in the life cycle of B. burgdorferi. Our in vivo studies with the OspB-deficient B. burgdorferi show that OspB is essential for the colonization and persistence of B. burgdorferi in ticks. During tick feeding, the ospB mutant and the wild-type B. burgdorferi enter the ticks from infected mice at the same rate. However, after feeding, the ospB mutant spirochetes are unable to persist within the blood meal or fully adhere to the tick gut, which also leads to a significant reduction in the number of spirochetes in the molted adult ticks (Figure 3). The binding of residual OspB-deficient spirochetes to the tick gut could be attributed to OspA, as the level of the OspA is unaltered in the OspB-deficient spirochetes (Figure 1). The luminal face of the gut epithelium is covered by a dense array of glycoproteins that may act as “receptor-buffet” for many pathogens [43]. Some of these glycoproteins are involved in general tissue structure and digestion [43,44] and some are involved in innate immunity [45,46]. Our surprising finding that the reduced ability of the ospB mutant to attach to the gut epithelium and its subsequent clearance in the gut may suggest that adherence to tick gut cells also is critical for some as yet unknown aspect(s) of spirochete viability.
Our in vivo analysis showed that in contrast to the ospB mutant, the OspB complemented strain readily colonized tick gut tissue and showed a drastic increase in its persistence within ticks, which was comparable to the wild-type isolate (Figure 4). Furthermore, our in vitro binding assays with the TGE also supported the in vivo analysis, indicating that in contrast to the ospB mutant, the transcomplemented strain binds with a greater affinity to the TGE. In addition, the difference between the wild-type and the ospB mutant spirochetes in binding to the fed TGE is significantly higher in comparison to the unfed TGE, suggesting that the levels of expression of putative OspB gut receptor proteins/glycolipids might increase during feeding. Furthermore, our in vitro binding data correlated with a previous study [34] showing that the B. burgdorferi N40 OspB protein binds significantly to the TGE. Overall, our studies solidify a great body of experimentation implicating an important role of OspB in the attachment of B. burgdorferi to the tick gut
Yang and co-workers recently examined the role of the ospAB locus in the infectious life cycle of B. burgdorferi [35]. This was accomplished by the generation of an ospAB double mutant from B. burgdorferi strain BbAH130 (infectious clone recovered after plating B. burgdorferi strain 297), and it was found that disruption of both the ospA and ospB genes had no observable effect on the ability of spirochetes to establish infection in mice, whereas the locus is critically essential for colonization of the tick gut [35]. Spirochetes deficient for both OspA and OspB entered ticks but were unable to persist within ticks for a long time [35]. Furthermore, complementation of the ospAB double mutant with both OspA and OspB expression restores the ability of B. burgdorferi to colonize the gut [35]. On the other hand, complementation with the ospA gene alone could only partially restore (50%–60% in comparison to the wild-type) the colonization defect of the ospAB mutant [35], suggesting that OspB expression is also required for the complete restoration of the defect. A comparison of our data with the prior study [35] indicated that in contrast to the ospAB double mutant complemented with the ospA gene alone, the ospB mutant analyzed in our study was significantly impaired in its persistence in the tick gut. These variations could have been the results of (i) the different B. burgdorferi strains used in the studies (B31 5A11 versus 297 BbAH130) and (ii) the relative OspA expression in the ospAB double mutant complemented with the ospA gene (on a circular plasmid) compared to the ospB mutant analyzed in our study. Overall, our studies in conjunction with the previous studies [35] show that (i) absence of OspB alone could result in severe impairment in colonization and persistence, and (ii) absence of both OspA and OspB could lead to the complete impairment in colonization and persistence of B. burgdorferi in ticks. In an evolutionary perspective, the conservation of ospB in the genome of B. burgdorferi is the result of positive selection pressure [23,24], and thus OspB must be of intrinsic value to the organism. Our studies suggest that the function of OspB and OspA are codependent
OspA and OspB share approximately 50% identity and 62% similarity in their amino acid sequences [2]. The crystal structures of OspA and the C-terminal region of OspB have been determined [27,28]. Comparison of the crystal structure of OspA and C-terminal region of OspB shows that these two molecules are quite similar [27,28]. The C-terminal region of OspB adopts the same fold as is observed for the C-terminal half of OspA [28]. Li and co-workers have identified that the C-terminal barrel domain in OspA is a trio of partially buried charged residues: Arg 139 from beta-strand 10, Glu-160 from beta-strand 12, and Lys-189 from beta-strand 15 [27]. The barrel domain of the OspA/B fold features a prominent cavity, in which the first two residues are strictly conserved in both OspA and OspB; position 189 is nearly always Lys in OspA and Arg in OspB [27,28]. Studies from Li et al. (1997) and Becker et al. (2005) have proposed that the cavity in the OspA/B barrel domain might be a ligand binding site for a small peptide, linear saccharide, or an exposed protein loop [27,28]. Furthermore, the mapping of amino acid sequences required for OspA binding to the tick gut showed that the residues 85–103 and 229–247 are important [47]. The percent similarity (identity) for the two amino acid stretches are 63 (68) and 84 (79) in OspB amino acid sequence at positions 110–128 and 252–270, respectively [34,47]. Given the structural and amino acid sequence conservation of OspA and OspB, it is possible that both lipoproteins recognize either the same target or closely related targets. It has recently been shown that OspB antibodies prevent B. burgdorferi colonization of I. scapularis gut [34]. Because of the high structural similarities between OspA and OspB, it could be reasoned that the OspB antibodies may bind to several epitopes of OspB on the B. burgdorferi surface, and steric hindrance might then interfere with OspB binding to the tick gut; or it is possible that steric hindrance by OspB antibodies also affected OspA-mediated binding of spirochetes to the tick gut [34]. Thus, our studies in conjunction with the previous reported study [34] raises interesting questions regarding the potential of antibody binding interfering with spirochete adherence in ticks. In the in vitro-grown B. burgdorferi cultures, the expression of OspB is lower than OspA (Figure 1D). Since OspA is also involved in the attachment of spirochetes to the tick gut and since we have found that OspB-deficient spirochetes are unable to attach to the tick gut, it is possible that disruption of the OspB resulted in the interference of the OspA-mediated attachment to TROSPA. Three scenarios may be envisioned that may elucidate the possible involvement of OspB in the OspA-TROSPA interactions. Firstly, OspB may directly associate with either OspA or TROSPA and may form a complex structure that is required for the tight attachment of B. burgdorferi to the tick gut. Secondly, OspB may bind to its own receptor within the gut and this interaction might be required for TROSPA to interact with OspA. Finally, OspA and OspB might bind to separate TROSPA molecules on the gut epithelium and both these interactions might be required for the tight attachment of B. burgdorferi to the tick gut. With any of these three models, our finding that OspB-deficient spirochetes were unable to colonize or persist in tick gut is significant because it suggests a possible synergistic interaction between OspA, OspB, and TROSPA.
In summary, these data suggest that OspB plays a critical role for B. burgdorferi adherence and persistence in ticks. These studies are not only important in understanding significant roles of spirochete ligands (such as OspB) in spirochete colonization and survival at arthropod-pathogen interface, but they also enhance our knowledge in the development of new therapeutic strategies, such as new transmission blocking vaccines that may be useful to combat B. burgdorferi infection.
pathogens.plosjournals.org/perlserv/?request=get-document&doi=10.1371%2Fjournal.ppat.0030033&ct=1