Abstract: Although most patients with Lyme disease can be cured with a 2-4 week antibiotic therapy, about 10-20% of patients continue to suffer prolonged persistent symptoms, a condition called post-treatment Lyme disease syndrome (PTLDS). The cause for PTLDS is unclear and hotly debated. B. burgdorferi develops morphological variants under stress conditions but their significance is not clear. Here we isolated the biofilm-like microcolony (MC) and planktonic (spirochetal form and round body) (SP) variant forms from the stationary phase culture and showed that the MC and SP were not only more tolerant to the current Lyme antibiotics but also caused more severe arthritis in mice than the log phase spirochete form (LOG). We propose to divide the persistent Lyme disease into two categories: (1) early development of persistent disease from inoculation with persister/biofilm at the beginning of infection introduced by tick bites, or Type I persistent disease (i.e., PTLDS); and (2) late development of persistent disease due to initial infection not being diagnosed or treated in time such that the infection develops into late persistent disease, or Type II persistent disease. Importantly, we show that the murine infection caused by LOG could be eradicated by ceftriaxone whereas the persistent infection established with MC could not be eradicated by doxycycline (Doxy), ceftriaxone (CefT), or vancomycin (Van), or Doxy+CefT or Van+CefT, but could only be eradicated by the persister drug combination daptomycin+doxycycline+ceftriaxone. Our studies demonstrate that varying levels of persistence and pathologies of Borrelia infection can be established with heterogeneous inocula with different morphologies and have different treatment responses. These observations may have broad implications for understanding pathogenesis and treatment of not only persistent Lyme disease but also other persistent infections in general and call for studies to evaluate if treatment of persistent infections with persister drug combination regimens is more effective than the current mostly single-antibiotic monotherapy.

Introduction

Lyme disease (LD) is the most common tick-borne illness and an important emerging zoonosis in the United States (Radolf et al., 2012) with an estimated 300,000 cases per year (CDC, 2017a). The causative agents of Lyme disease are pathogenic Borrelia species including B. burgdorferi, B. afzelii, and B. garinii, where B. burgdorferi is the predominant cause of human LD in North America. In humans, Lyme disease may cause a local erythema migrans (EM) rash at the site of the tick bite and then readily disseminates through the bloodstream to other tissues, setting up an infection that can last for months to years (Steere et al., 2016). Patients with Lyme disease are routinely treated with 2-4 week doxycycline, amoxicillin, or cefuroxime, which effectively hastens the resolution in most cases (Wormser et al., 2006). However, about 10-20% of patients, according to CDC, continue to suffer lingering symptoms of fatigue, pain, or joint and muscle aches, and neurocognitive manifestations that last 6 months or more despite treatment, a condition called “post-treatment Lyme disease syndrome” (PTLDS) (CDC, 2017b). A more recent large-scale analysis suggests that 36-63% of treated patients develop symptoms of persistent disease (Adrion et al., 2015). While the cause of PTLDS is complex and remains to be determined, one of the possible explanations is persistent B. burgdorferi infection due to persisters not being killed by the current antibiotic treatment, which evade host immune clearance and drive immunological responses continually as shown in various animal models (Embers et al., 2012; Hodzic et al., 2008; 2014; Straubinger et al., 1997). A recent study in humans demonstrated the recovery of B. burgdorferi DNA by xenodiagnoses in patients despite antibiotic treatment (Marques et al., 2014). Findings indicate that current Lyme disease treatment may not sufficiently eliminate B. burgdorferi persisters or that the immune system fails to clear persisting organisms or bacterial debris, which may be the underlying cause for those who suffer from unresolved Lyme disease symptoms. In contrast to other bacterial pathogens that cause persistent infections such as M. tuberculosis (Zhang et al., 2012), an unusual feature of the in vivo persistence of B. burgdorferi is the lack of culturability of the persisting organisms despite the demonstration of its DNA and even increased DNA content by PCR or by xenodiagnosis (Embers et al., 2012; Hodzic et al., 2014).

In addition to the above in vivo persistence, B. burgdorferi has recently been shown to develop persisters in vitro in cultures as shown by tolerance to current Lyme antibiotics doxycycline, amoxicillin, and cefuroxime (Caskey and Embers, 2015; Feng et al., 2014a; Sharma et al., 2015). In addition, B. burgdorferi can develop morphological variants including spirochetal form, round body form, or cystic form, and aggregated biofilm-like microcolonies as the culture grows from log phase to stationary phase or under stress conditions in vitro (Brorson et al., 2009; Feng et al., 2015a). The variant forms such as cystic and round body forms have been found in vivo in brain tissues of Lyme borreliosis patients (Miklossy et al., 2008) but their role in the persistent form of the disease is controversial (Lantos et al., 2014). We showed that stationary phase cultures of B. burgdorferi contain different morphological variants including planktonic spirochetal form, round body form, and aggregated microcolony form (Feng et al., 2015a), which have varying levels of persistence or can be considered different types of persisters in comparison to the log phase culture which mainly consists of growing spirochetal form with no or few persisters (Feng et al., 2015a). However, the previous study that showed different degrees of tolerance to antibiotics by the variant forms was performed with mixed stationary phase culture (Feng et al., 2015a), but not with individually isolated variant forms. In addition, although different variant forms of B. burgdorferi seem to elicit different host cytokine response (Merilainen et al., 2016), their ability to cause disease has not been evaluated. In this study, we isolated three different forms of B. burgdorferi, including growing log phase spirochetes (LOG), non-growing stationary phase planktonic (SP) form, and aggregated biofilm-like microcolony (MC) form and compared their susceptibility to different antibiotics in vitro and found indeed the MC form and SP form to be more tolerant to antibiotics. Interestingly, we found that the MC form and the SP form caused more severe arthritis in mice than the log phase spirochetal form. More importantly, we show that the current Lyme antibiotic treatment with doxycycline or even ceftriaxone was unable to eradicate such persistent infections in the mouse model introduced by MC persister inocula but could be completely eradicated by the persister drug regimen Daptomycin+Doxycycline+Ceftriaxone, which was previously shown to do so in vitro (Feng et al., 2015a).

Materials and Methods

Strain, media, and culture techniques

Borrelia burgdorferi strain 297 and strain N40 were kindly provided by Dr. Steve Dumler and Dr. Emir Hodzic, respectively. B. burgdorferi strains were grown in BSK-H medium (HiMedia Laboratories Pvt. Ltd.) with 6% rabbit serum (Sigma-Aldrich, St. Louis, MO, USA). All culture medium was filter-sterilized by 0.2 μm filter. Cultures were incubated in a microaerophilic incubator (33°C, 5% CO₂) without antibiotics.

Antibiotics and drug susceptibility assay

Antibiotics including doxycycline, ceftriaxone, and cefuroxime were purchased from Sigma-Aldrich (St. Louis, MO, USA) and dissolved in water at appropriate concentrations to form stock solutions. All antibiotic stocks were filter-sterilized using a 0.2 μm filter. The residual viability of B. burgdorferi cells treated with antibiotics or drug combinations was evaluated using the SYBR Green I/PI assay combined with fluorescence microscopy as described previously (Feng et al., 2014b). Briefly, the ratio of live and dead cells was calculated by the ratio of green/red fluorescence, the regression equation, and the regression curve with least-square fitting analysis.

Separation and preparation of microcolony form and planktonic form from stationary phase culture

After incubation for 10 days, 1 ml stationary phase B. burgdorferi culture (~10⁷ spirochetes/mL) was centrifuged at 800 × g for 10 min, and the supernatant was transferred to a new tube as the stationary phase spirochetal form. The bottom 50 μl microcolony rich culture was resuspended and centrifuged at 800 × g 3 times to remove the planktonic spirochetes. The stationary phase spirochetal form and microcolony form were checked with a fluorescence microscope to ensure their morphologies before being used for the infection (see below).

Animals and infections

Five-to-six week-old female C3H/HeN mice purchased from Charles River were infected by subcutaneous inoculation of the dorsal thoracic midline with about 10⁵ cells of different forms of B. burgdorferi strain N40 or strain 297, i.e., log phase spirochetal form, stationary phase planktonic form, and stationary phase aggregated biofilm-like microcolony form. The inoculum size of 10⁵ cells of B. burgdorferi was chosen based on previous studies (Ma et al., 1998). To control the potential problem that aggregated microcolony form may contain more bacteria in the inocula, we gauged for the same amount of fluorescence using SYBR Green/PI staining for both the MC form and the spirochete form in our inoculation preparation. Infection was confirmed by the culture of B. burgdorferi from the ear punch. The infection was followed over a period of 90 days to monitor the degree of arthritis induced by different morphological forms. All experiments were performed in accordance with the Animal Care and Use Committee of Johns Hopkins University.

Assessment of arthritis and myocarditis severity

Caliper measurements of tarsus and knee joints were performed every two days during the infection by an investigator blinded to the experimental groups. For assessment of histopathologic lesions, the heart and tarsus were separately fixed in 10% neutral buffered formalin (Sigma-Aldrich, St. Louis, MO, USA). Fixed tissues were decalcified and embedded in paraffin, sectioned at 6 mm, and stained with H&E.

Histopathology scoring criteria

Tissues were scored blindly by two pathologists who reviewed all the slides at one sitting and assigned relative scores based on the severity of inflammation. Overall lesion composite score was based on all elements of the tarsus or heart and 4-5 slides per mouse were examined using a scoring system modified from Crandall et al. (2006). Hearts were scored for the degree/extent of inflammation in myocardium, great vessels (pulmonary artery and aorta), and valves, using the following categories: no inflammation (score of 0), mild inflammation (score of 1), moderate inflammation (score of 2), and severe inflammation (score of 3). In general, the inflammation infiltrates in the heart consisted of lymphocytes and macrophages. Mice infected with either stationary phase or microcolony form had a higher degree of inflammation compared to the mice infected with log phase bacteria.

The tarsus from both legs was scored based on the degree of inflammation involving the tendon, joint capsule, and joint space, using the following categories: tarsus with no inflammation (score of 0), mild inflammation (score of 1), moderate inflammation (score of 2), and severe inflammation (score of 3). The inflammation was a mixture of rare polymorphonuclear leukocytes with more mononuclear cell (lymphocyte and macrophage) infiltration. Tendon sheaths were thickened due to hypertrophy and hyperplasia of surface cells and/or underlying dense sheets of cells resembling immature fibroblasts, synovial cells, and/or granulation tissue. In some bones of the tarsus, the reactive/reparative response was present including periosteal hyperplasia and new bone formation and remodeling. Mice infected with either stationary phase or microcolony form had a higher degree of inflammation compared to the mice infected with log phase bacteria.

Treatment study in the persistent infection mouse model using persister inocula

Table 1. Drug dosage and route of administration in the treatment experiment.

Mice (5 mice per group) were infected with B. burgdorferi strain N40 microcolony persister form or log phase spirochetal form as described above. The treatment was started one week after infection. The antibiotic doses and regimens and the route of drug administration are shown in Table 1. The infected mice were treated daily for 30 days per previous treatment studies in the persistent B. burgdorferi mouse model that demonstrated the persistence phenomenon (Hodzic et al., 2014) after standard Lyme antibiotic treatment of 2-4 weeks (Wormser et al., 2006). Eight weeks after the treatment, mice were sacrificed with CO₂ asphyxiation. To evaluate the effect of treatment, ear punch culture was performed in BSK-H medium and incubated for 27 days for mice that were subjected to different treatments.

Results

Separation of aggregated microcolony form from planktonic form in stationary phase culture

Figure 1. Representative images of different forms of B. burgdorferi at different growth stages. Three-day old log phase (A), 10-day old stationary phase microcolony form (B), and 10-day ols stationary phase planktonic form (C) of B. burgdorferi. Cells were stained with SYBR Green I/PI and observed using fluorescence microscopy at 400X magnification.

Previous studies demonstrated that B. burgdorferi develops multiple morphological forms including spirochetal form, round bodies (cysts), and aggregated microcolony form (Feng et al., 2015a; Miklossy et al., 2008). Log phase B. burgdorferi culture is mainly in spirochetal form, but stationary phase culture is dominated by coccoid or round-body forms and aggregated micro-colony forms (Feng et al., 2015a; 2014a). We showed that the current clinically-used antibiotics for treating Lyme disease had high activity against log phase B. burgdorferi but had very limited activity against the stationary phase cells (Feng et al., 2014a). We also demonstrated that aggregated microcolony forms are more tolerant to antibiotics than planktonic spirochetal and round body forms in the stationary phase culture (Feng et al., 2015a). To assess the antibiotic susceptibility of individual variant forms, here we first separated the aggregated microcolony form from the planktonic form (spirochetal and round body) in stationary phase B. burgdorferi culture according to their different densities using low-speed centrifugation. After multiple differential centrifugation separations, microcolony form and planktonic form were separated effectively (Figure 1B and 1C). Consistent with our previous study (Feng et al., 2015a), here we found that the 5-day old log phase cells were mostly in spirochetal form and it was more like a log phase culture (Figure 1A), while the 10-day old stationary phase culture contained, in addition to aggregated biofilm-like microcolonies (Figure 1B), planktonic forms made up of not only spirochetes but also many round body cells (Figure 1C).

Different B. burgdorferi forms have different drug susceptibility in vitro

Figure 2. Susceptibility of log phase (LOG), stationary phase planktonic (SP), and microcolony form (MC) B. burgdorferi to 5 µg/mL to clinically used antibiotics after 7-day treatment. (A) The percentages of residual live cells (means ± standard deviation) were determined with SYBR Green I/PI assay. The asterisks indicate significant susceptibility difference (triple asterisks, P < 0.001; quadruple asterisks, P < 0.0001) as compared to the log phase group. (B) Representative images of the different forms of B. burgdorferi treated with doxycycline and ceftriaxone for 7 days followed by staining with SYBR Green I/PI assay (40X magnification).

Since our previous studies showed that stationary phase B. burgdorferi cells are more tolerant to antibiotics (Feng et al., 2015a; 2014a), here we compared the drug susceptibility of the individually isolated stationary phase planktonic form, aggregated microcolony form, and the log phase B. burgdorferi. After 7-day drug treatment, the stationary phase planktonic (SP) form and microcolony form (MC) of B. burgdorferi were found to be much more tolerant to the current clinically-used antibiotics doxycycline, cefuroxime, and ceftriaxone (5 μg/ml) (Figure 2). The residual viability after drug treatment showed a statistically significant difference between the log phase form and two stationary phase forms SP and MC (Figure 2A). By microscope analysis, we found that most of the log phase spirochetal B. burgdorferi cells were eradicated by the 7-day doxycycline or ceftriaxone treatment. In contrast, most of the stationary phase planktonic and microcolony form of B. burgdorferi cells were still alive as shown by green fluorescence after the 7-day doxycycline or ceftriaxone treatment (Figure 2B).

Stationary phase planktonic form and microcolony form of B. burgdorferi cause more severe persistent Lyme arthritis than the log phase spirochete form

To investigate the effects of different forms of B. burgdorferi on the development of arthritis in the mice, C3H/HeN mice were infected by subdermal injection with 10⁵ cells of N40 strain. All injected mice became infected, as determined by ear punch and bladder culture at the end of experiments. During the 90-day infection period, we used caliper measurements to monitor the swelling or edema of ankle and knee joint, which can directly reflect the degree of the inflammatory response to B. burgdorferi infection (Ma et al., 1998). We found that the different forms of B. burgdorferi showed different effects on inducing ankle joint swelling in mice, especially in the early stage of 21 days (Figure 3). The microcolony (MC) form of B. burgdorferi showed the most obvious pathogenic effect in terms of the ability to cause joint swelling compared to the other forms and uninfected control (Figure 3). The MC infected group showed significant swelling (p<0.005) compared with the uninfected group as early as 9 days (Figure 3A); however, the log phase spirochete form and stationary phase planktonic form did not cause significant swelling (p>0.05) (Figure 3A). At 21 days, the joint swelling of the MC group dramatically increased and reached a peak (Figure 3); but the SP form showed significant swelling (p<0.005) at this time point but less severe than the MC form infected group (Figure 3). The log phase B. burgdorferi infected mice were just beginning to develop joint swelling at day-21 post infection, with only slight swelling (Figure 3A and C). At 30-day post infection, the SP group mice were also found to develop similar severity of joint swelling as the MC group, but the log phase group still did not show significant swelling (p>0.05) (Figure 3A). However, at 35-day post infection, all three infected groups showed a similar degree of significant joint swelling (p<0.05) compared to the uninfected control group (Figure 3A and C). After 35-day infection, the joint swelling of the three infected groups waxed and waned during the 90-day observation, while the severity of ankle joint swelling did not have a significant difference between the three different B. burgdorferi infected groups post 35-day infection (Figure 3B).

Figure 3. Ankle joint measurements of C3H/HeN mice infected with different forms of B. burgdorferi and uninfected control. Measurements of every ankle joint were made for each mouse every week for 90 days. (A) At the early stage infection (before 35 days), mice infected with the group of microcolony form (MC), stationary phase planktonic form (SP), and log phase form (LOG) showed different degrees of severity on ankle swelling. Measurement values, means, and standard deviations (error bars) were shown for each group. Statistical analyses were performed using the standard t-test, comparing the joint measurements of the infected groups with those of the uninfected group. (B) Ankle joint swelling mean values of different groups during the 90-day measurements. The colored arrows indicate the earliest time of reaching the most severe swelling level of the corresponding groups. (C) The images of the most severe swollen ankles in different groups at day 21 and day 35, respectively.

Mice infected with persistent forms of Borrelia (stationary phase planktonic and stationary phase microcolony forms) have increased disease pathology

Histopathology analysis of infected mice with B. burgdorferi showed that all mice had low-grade aortitis and myocarditis compared to the healthy heart (average range of 0.66 to 2.33). In mice that were infected with strain 297, an inoculum-dependent severity of disease was observed. Mice infected with microcolony forms had the most severe disease with an average severity score of 2.3 (p<0.05) whereas mice infected with stationary planktonic or log phase forms have severity scores of 1.3, and 0.67, respectively. Mice infected with stationary phase planktonic or microcolony forms showed inflammation in the epicardium which was not noted in the infection of the log phase group or healthy control group.

Similar to the histopathology findings of the heart, inflammation in the joints was the most severe in mice infected with the microcolony form. Mice infected with microcolony form from the 297 strain had a severity score of 2.2 while infection with log phase bacteria produced a severity score of 1.3. Similarly, a difference was observed in mice infected with the N40 strain (a severity score of 1.5 in microcolony-infected mice versus a severity score of 1.2 in log phase infected mice). Mice infected with stationary planktonic or microcolony forms showed moderate to severe mononuclear cell infiltration (with some fibroblast proliferation) into the joint capsule. Infection with either strain 297 or strain N40 showed an inoculum-dependent severity of disease in the joints, with infection with the stationary phase/more persistent forms having increased inflammation more than mice infected with log phase form of the respective strains.

The healthy control mice (Figure 4, Panel A) (40X) exhibited normal features of aorta, valves, and myocardium with no inflammation or fibrosis. Mice infected with log phase B. burgdorferi strain 297 bacteria (Figure 4, Panel B) (40X) exhibited a mild degree of lymphohistiocytic inflammation in the aorta and accompanying valve and myocardium in heart base. Mice infected with stationary phase B. burgdorferi strain 297 planktonic form bacteria (Figure 4, Panel C) (40X) had a moderate amount of lymphohistiocytic inflammation in the aorta and thickening of valves. Mice infected with stationary phase B. burgdorferi strain 297 microcolony form bacteria (Figure 4, Panel D) (40X) exhibited a moderate degree of lymphohistiocytic inflammation in the aorta and accompanying valve and myocardium.

Figure 4. Images of heart base histopathology taken from mice infected with different forms of B. burgdorferi. Mice were infected with log phase (LOG) (B), stationary phase planktonic form (SP) (C), and stationary phase aggregated microcolony form (MC) (D) for 12 weeks when the heart base was sectioned for hematoxylin and eosin (HE) staining and examined under the microscope (40X). Representative heart base examples (A-D panels) show the difference in severity across treatment groups. The healthy control mice (Panel A) (40X) showed normal features of aorta, valves, and myocardium with no inflammation or fibrosis. Mice infected with log phase B. burgdorferi strain 297 (Panel B) (40X) exhibited a mild degree of lymphohistiocytic inflammation in the aorta and accompanying valve and myocardium. Mice infected with stationary phase B. burgdorferi strain 297 planktonic form (Panel C) (40X) had a moderate level of lymphohistiocytic inflammation in the aorta and thickening of valves. Mice infected with stationary phase B. burgdorferi strain 297 microcolony form (Panel D) (40X) exhibited a moderate degree of lymphohistiocytic inflammation in the aorta and accompanying valve and myocardium. Bar with arrows denotes aorta vessel; # denotes aortic valve leaflet; ## denotes AV valve. Lower right corner bar = 100 microns.

The healthy control mice (Figure 5, Panel A) (40X) exhibited features including normal overlying skin, tendon, thin layer of synovium within the joint, and tarsal bones without any inflammation or fibrosis. The tarsus of mice infected with log phase B. burgdorferi strain 297 (Figure 5, Panel B) (100X) exhibited a mild degree of synovial hyperplasia within the joint with mild to moderate tendon fibroplasia with lymphohistiocytic infiltrates within tendon and joint capsule. The tarsus of mice infected with stationary phase B. burgdorferi strain 297 planktonic form bacteria (Figure 5, Panel C) (40X) had a moderate amount of synovial hyperplasia, edema, fibroplasia, and lymphohistiocytic inflammation associated with the joint capsule and tendon sheath. The tarsus of mice infected with stationary phase B. burgdorferi 297 strain microcolony form bacteria (Figure 5, Panel D) (100X) had severe fibroplasia and lymphohistiocytic inflammation associated with the joint capsule and tendon sheath.

A triple persister drug combination regimen Daptomycin + Doxycycline + Ceftriaxone completely eradicated the persistent infection in the mouse arthritis model while the current standard of care antibiotics failed to do so.

Since we established the rapid persister-inocula mouse arthritis model, we wanted to evaluate the ability of single drugs and drug combinations to eradicate the persistent infection in this model. Mice were infected with different forms of B. burgdorferi strain N40 and the infection was allowed to establish for 1 week when the treatment started. We treated the infected mice with doxycycline (Doxy), ceftriaxone (CefT), vancomycin (Van), Doxy+CefT, or Doxy+CefT+Daptomycin (Dap), for 30 days, and waited for 8 weeks to allow any residual bacteria to revive or relapse before the sacrifice of the mice for assessing the effect of the treatments. The results with ear punch culture (27 days) showed that doxycycline was unable to eradicate the infection caused by log phase or microcolony Borrelia. In addition, vancomycin did not clear the infection caused by either log phase inocula or stationary phase microcolony inocula (Table 2). However, CefT was more effective and could eradicate the infection with log phase inocula but not stationary phase microcolony inocula (Table 2). This finding is consistent with our in vitro studies (Feng et al., 2015a) and also clinical studies (Dattwyler et al., 1988; Wormser et al., 2006), indicating our new mouse model is a useful model to evaluate drug combinations. Importantly, we found that the persister drug combination (Doxy+CefT+Dap) which was previously shown to completely eradicate B. burgdorferi microcolony forms in vitro (Feng et al., 2015a) is more effective than single drugs or two-drug combinations and completely eradicated the microcolony form infection from the mice (Table 2). Thus, our findings suggest that there is a hierarchy of persistence with different bacterial inocula where infection created by aggregated microcolony form of B. burgdorferi is more persistent and difficult to cure than the infection caused by log phase (growing) Borrelia, a finding that supports our recent hypothesis of persister inocula (Zhang, 2014) that PTLDS could be due to persister inocula introduced during the tick bite (Feng et al., 2018).

Discussion

Figure 5. Images of tarsus joint histopathology taken from mice infected with different forms of B. burgdorferi. Mice were infected with log phase (LOG) (B), stationary phase planktonic form (SP) (C), and stationary phase aggregated microcolony form (MC) (D) for 12 weeks when the joints were sectioned for hematoxylin and eosin (HE) staining and examined under the microscope (40X). Representative joint examples (A-D panels) show the difference in severity across treatment groups. The joint from healthy control mice (Panel A) (40X) showed features including normal overlying skin, tendon, thin layer of synovium within the joint, and tarsal bones without any inflammation or fibrosis. The tarsus of mice infected with log phase B. burgdorferi strain 297 (Panel B) (100X) exhibited a mild degree of synovial hyperplasia within the joint with mild to moderate tendon fibroplasia with lymphohistiocytic infiltrates within tendon and joint capsule. The tarsus of mice infected with stationary phase B. burgdorferi strain 297 planktonic form bacteria (Panel C) (40X) had a moderate level of synovial hyperplasia, edema, fibroplasia, and lymphohistiocytic inflammation associated with the joint capsule and tendon sheath. The tarsus of mice infected with stationary phase B. burgdorferi strain 297 microcolony form (Panel D) (100X) had the most severe fibroplasia and lymphohistiocytic inflammation associated with the joint capsule and tendon sheath. * denotes the tendon. Lower right corner bar = 100 microns.

In this study, we showed that different variant forms of B. burgdorferi are different types of persisters with varying degrees of persistence and are not killed readily by current Lyme antibiotics (Figure 2). However, the significance of the in vitro persisters in relevance to the in vivo persisters and disease mechanisms was not clear. Here, we found that the variant forms MC and stationary phase planktonic form (SP) caused more severe disease (arthritis) than the log phase spirochetal form of B. burgdorferi (Figures 3 and 4). In addition, B. burgdorferi in the tick could develop variant forms that may represent different forms of persisters (Cabello et al., 2017), and that biofilm bacteria have been found in skin biopsies in humans (Sapi et al., 2016). Previously, we hypothesized that persistent bacterial infections could be caused by “persister-seeding” theory (Zhang, 2014). In view of our new findings in this study that support this theory, we propose a new theory of persistent Lyme disease or PTLDS to better explain the persisting symptoms despite the standard Lyme antibiotic treatment: i.e., a biofilm/persister “seeding” theory. This theory can be called “Inocula-Dependent Severity of Disease” determined by the status/quality of the inocula bacteria, ranging from spirochetes to round bodies, microcolonies, and biofilms. The characteristics of the inocula persisters matter, that is, the outcome of the infection is to a large extent dependent on the discreteness of the inoculum. In other words, PTLDS is made to happen at the time of the tick bite by the variant of B. burgdorferi that is injected into the human body. A small inoculum may be sufficient to cause disease if it is mainly microcolony or persister/mother cells or stem-like cells. Persistent infections can be caused by B. burgdorferi morphologic variant forms, and different types of persisters (spirochetal form, round body, cyst, microcolony, biofilm) in a mixture; different forms may be able to cause disease with different severities depending on the proportion of the different variant forms, with spirochetal form causing mainly active or acute form of the disease that is more easily curable, while round bodies and microcolonies, and biofilms may cause a more severe and persistent disease that is more difficult-to-treat by the current treatment of Lyme disease with a single antibiotic.

This new model of biofilm/persister bacteria seeding explains some mysterious features of post-treatment Lyme disease syndrome (PTLDS), i.e., inability of the current Lyme antibiotics to eradicate or cure the 10-20% patients despite the initial standard 2-4 week antibiotic treatment as a result of persister inocula; as such one could develop PTLDS from the very beginning after a tick bite and continue to have PTLDS despite early diagnosis and timely treatment. Thus, an infection caused by biofilm/persister bacteria will not respond well to the current standard Lyme antibiotics, which is essentially a treatment failure. In terms of the disease severity, it is interesting to note that persister inocula (microcolony) produced a more severe disease than the spirochete form of B. burgdorferi from early on from 9 to 21 days in the mouse model (Figures 3 and 4). Interestingly, similar findings of a severe disease with early onset have recently been observed in PTLDS patients that “start severe and stay severe” (Brian Fallon and John Aucott, personal observations). On the other hand, there is another type of persistent disease, i.e., the initial infection is not diagnosed and treated in time such that more persisters and biofilms could develop, leading to a more persistent infection that is more difficult to treat or cure. Thus, while persistent Lyme disease can be a very heterogeneous condition, we can broadly divide persistent Lyme disease (PLD) [or chronic Lyme disease (CLD)] into two types or categories: Type I persistent Lyme disease (PLD) which is essentially PTLDS, or early development persistent disease where persister/biofilm inocula cause severe disease from the beginning; and our biofilm microcolony induced mouse arthritis model (Feng-Zhang model) is a representative of this Type I PLD; and Type II persistent Lyme disease, or late development persistent disease, where the long-term mouse persistence model which allows an acute infection to develop into a persistent phase of the disease through prolonged infection before treatment (Hodzic-Barthold model) (Hodzic et al., 2014). The development of this biofilm/persister-inocula seeding model or Type I PLD model may allow us to more efficiently evaluate persister drug combinations as well as certain essential oils which we found to be effective in vitro (Feng et al., 2015a). In addition, our finding that persister forms cause a more severe disease may have implications for better understanding disease pathogenesis and developing better diagnostics and more importantly better treatments for PTLDS as well as chronic or persistent Lyme disease.

Table 2. Evaluation of drug combinations using the persistent arthritis model.

Although our mouse study provides a feasible explanation for PTLDS condition with persister inocula, our treatment study was not done on previously treated mice to develop a treatment for PTLDS per se but has implications for treating this condition. With the new mouse model we developed in this study that starts with different types of inocula, the two forms of PLD (Type I and Type II) seen in vivo including both untreated and previously treated conditions can now be addressed in this model for future studies in addition to studying persister cell pathology in the mice. While we believe there is chronicity or persistence of Lyme disease in some patients, we have not used the term “chronic Lyme disease” in this study as it is a controversial term that has been used to include unclear chronic conditions that may not be Lyme disease (Feder et al., 2007; Lantos, 2011; Marques, 2011; Stricker and Fesler, 2018). Thus, we have chosen a more neutral term “persistent Lyme disease” in this study. We are fully aware of other possible theories of PTLDS besides persistent infection, including antigenic debris (Bockenstedt et al., 2012), autoimmunity (Arvikar et al., 2017), residual damage to tissues during infection, and non-specific pains of “daily living.” Our finding that biofilm persister form causes a more severe and persistent disease in the mouse model will allow us to identify persister drug regimens that may help to address the possible role of persistent infection underlying PTLDS and may ultimately help to resolve the controversy if such persister drug regimens that eradicate the persistent infection in mice turn out to “cure” PTLDS in new clinical trials.

The observation that biofilm/persister inocula produce a more severe disease than log phase growing bacteria (Figures 3, 4, 5) has not gained deserved attention previously in the field of bacterial pathogenesis studies in general. It is commonly assumed that all bacteria of a given bacterial pathogen are equally able to cause disease in a host. Most studies use log phase bacteria for infections and pay more attention to the number of bacteria used for infection rather than the quality or metabolic status of the bacterial inocula. Although some studies used stationary phase bacteria as inocula for infection, they did not compare with log phase bacteria in the same study (Liu et al., 2005) such that they did not reach the conclusion we did in this study, i.e., the stationary phase bacteria cause more severe disease than log phase bacteria. Our findings that biofilm/persister inocula produce a more severe disease than log phase bacteria call attention to the quality of the bacterial inocula used for infection, which is an important consideration in disease model and a re-evaluation of the possible differences in pathogenesis and host responses associated with the more severe form of the disease induced by the biofilm/persister inocula. This study calls for a comparative study of different bacterial inocula in their ability to cause disease. Future studies are needed to determine the effect of different inoculum size with different types of bacterial inocula (persisters versus non-persisters and spirochetes versus biofilm, microcolony, and round body persisters) on disease severity and pathogenesis and response to treatment.

There are some caveats with the use of aggregated biofilm-like microcolony (MC) forms for inoculation of infection as they may contain more organisms than the log phase spirochete form inocula. However, we believe this may not be a significant issue as our stationary phase planktonic form (containing both round bodies and spirochetes) without aggregates already produced a more severe disease than the log phase spirochete form (Figures 3, 4, and 5). Nevertheless, to limit this potential confounding problem, we gauged for the same amount of fluorescence using SYBR Green/PI assay for both the MC form and the spirochete form in our inocula preparation. Other ways such as protein or DNA content quantitation to ensure the same inocula size being used for inoculation of infection can be tested in the future.

While the detailed mechanisms by which the biofilm/persister inocula produce a more severe disease remain to be determined, several possibilities do exist. The genes and pathways that are turned on in biofilm/persisters (Zhang, 2014) may allow the persister bacteria to survive the attack by the host immune system such that they can better establish the infection and cause disease. In contrast, log phase bacteria may be killed more readily by the immune systems (innate and adaptive immunity) such that they do not cause as severe a disease as the biofilm/persister inocula that are more difficult to eradicate. Indeed, it has been shown that B. burgdorferi persisters tolerant to antibiotics express more virulence factors such as decorin-binding proteins (DbpAB), CRASP BB_A68, ErpQ, BdrEFMRVW, BB_I29, Clp protease, and DNA repair proteins in RNA-seq analysis (Feng et al., 2015b) and also in our proteomic study (Feng and Zhang, 2018). Thus, there is a molecular basis for the increased virulence of the biofilm/persister inocula in the case of B. burgdorferi, and perhaps for other bacteria. In addition, the persisters and biofilm bacteria may express unique or more of certain antigens or toxic proteins that elicit more severe inflammatory responses and pathologies. The observations of more severe pathology produced by microcolonies than log phase B. burgdorferi were made from susceptible C3H mice, but such observations may vary to some extent in different mouse strains. In addition, the timing of observation would also affect the results and the degree of histopathology of target tissues from mice inoculated with different forms of B. burgdorferi. Future studies with different time points as well as with different mouse strains may be of interest to better characterize the mouse model.

The biofilm/persister seeding model (Zhang, 2014), where the biofilm/persister bacteria cause more severe and persistent disease, has been shown to be valid not only for Borrelia infection in this study, but also for other bacterial infections such as S. aureus persistent infections in mice (Yee et al., 2018). We believe this biofilm/persister seeding model to be valid for different microbial infections and will have broad implications for understanding disease pathogenesis and for developing more effective treatments of persistent infections in general.

PTLDS is a very large unmet medical need in the U.S. and globally as each year 10-20% of 300,000 cases of Lyme disease in the U.S. will develop PTLDS, the cause of which has remained unknown with no proper treatment for such a disease condition. This study offers a new possible explanation for this condition due to Borrelia persister inocula that cause a more severe disease, a finding that has clinical implications. More importantly, we were able to show that there is a varying degree of persistence or persistent infection established with different inocula, where biofilm-like microcolony inocula produced a more severe and a more persistent disease that could not be eradicated by the current Lyme antibiotics or even some two-drug combinations but could be eradicated by the persister drug combination Daptomycin+Doxycycline+Ceftriaxone. In contrast, the disease induced by the log phase spirochete form is more readily eradicated by Ceftriaxone. That the inclusion of persister drug Daptomycin in combination with Doxycycline and Ceftriaxone is critical for eradicating the persistent infection established by persister inocula validates the relevance of our screening for drugs or drug combinations against stationary phase bacteria enriched in persisters in vitro (Feng et al., 2015a; 2014a). Furthermore, the eradication of the persistent infection with the persister drug combination utilizing persister drug Daptomycin that kills non-growing B. burgdorferi persisters and Doxycycline and Ceftriaxone that kill growing and semi-dormant Borrelia bacteria, respectively, validates the importance of persister drugs and represents an example of the principle of applying the persister drug pyrazinamide (PZA) (Zhang et al., 2014) per the Yin-Yang model (Zhang, 2014) for treating persistent infections in general. Our findings may not only provide a new understanding of PTLDS but will also allow us to develop novel persister drug combination regimens that can more effectively cure persistent Lyme disease in the future. Finally, now that we demonstrated that the persister drug combination could eradicate the persistent Lyme disease in the mouse model, clinical studies can be initiated to evaluate if the persister drug combination could more effectively treat or cure patients with PTLDS.

Acknowledgments

We thank Steve Dumler for providing Borrelia burgdorferi strain 297, Emir Hodzic for Borrelia burgdorferi strain N40, and Tim Sellati, Justin Radolf, Mark Soloski, John Aucott, Brian Fallon and Adriana Marques for helpful discussions. We acknowledge the support by Steven & Alexandra Cohen Foundation, Global Lyme Alliance, LivLyme Foundation, NatCapLyme, and the Einstein-Sim Family Charitable Fund.

Disclosure

The authors report no conflicts of interest.

Corresponding Author

Ying Zhang, M.D., Ph.D., Professor, Department of Molecular Microbiology and Immunology, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD 21205, USA.

References

Adrion ER, Aucott J, Lemke KW, Weiner JP. Health care costs, utilization and patterns of care following Lyme disease. PLoS One 10(2):e0116767, 2015.

Arvikar SL, Crowley JT, Sulka KB, Steere AC. Autoimmune arthritides, rheumatoid arthritis, psoriatic arthritis, or peripheral spondyloarthritis following Lyme disease. Arthritis Rheumatol 69(1):194-202, 2017.

Bockenstedt LK, Gonzalez DG, Haberman AM, Belperron AA. Spirochete antigens persist near cartilage after murine Lyme borreliosis therapy. J Clin Invest 122(7):2652-2660, 2012.

Brorson O, Brorson SH, Scythes J, Macallister J, Wier A, Margulis L. Destruction of spirochete Borrelia burgdorferi round-body propagules (RBs) by the antibiotic tigecycline. Proc Natl Acad Sci U S A 106(44):18656-18661, 2009.

Cabello FC, Godfrey HP, Bugrysheva JV, Newman SA. Sleeper cells: the stringent response and persistence in the Borreliella (Borrelia) burgdorferi enzootic cycle. Environ Microbiol 19(10):3846-3862, 2017.

Caskey JR, Embers ME. Persister Development by Borrelia burgdorferi populations in vitro. Antimicrob Agents Chemother 59(10):6288-6295, 2015.

CDC. Lyme Disease. Available online at: //www.cdc.gov/lyme/ (accessed 6/13/2018). 2017a.

CDC. Post-treatment Lyme Disease Syndrome. Available online at: //www.cdc.gov/lyme/postLDS/index.html (accessed 7/3/2018). 2017b.

Crandall H, Dunn DM, Ma Y, Wooten RM, Zachary JF, Weis JH, Weiss RB, Weis JJ. Gene expression profiling reveals unique pathways associated with differential severity of Lyme arthritis. J Immunol 177(11):7930-7942, 2006.

Dattwyler RJ, Halperin JJ, Volkman DJ, Luft BJ. Treatment of late Lyme borreliosis — randomised comparison of ceftriaxone and penicillin. Lancet 1(8596):1191-1194, 1988.

Embers ME, Barthold SW, Borda JT, Bowers L, Doyle L, Hodzic E, Jacobs MB, Hasenkampf NR, Martin DS, Narasimhan S, Phillippi-Falkenstein KM, Purcell JE, Ratterree MS, Philipp MT. Persistence of Borrelia burgdorferi in rhesus macaques following antibiotic treatment of disseminated infection. PLoS One 7(1):e29914, 2012.

Feder HM, Jr, Johnson BJ, O’Connell S, Shapiro ED, Steere AC, Wormser GP, Ad Hoc International Lyme Disease Group, Agger WA, Artsob H, Auwaerter P, Dumler JS, Bakken JS, Bockenstedt LK, Green J, Dattwyler RJ, Munoz J, Nadelman RB, Schwartz I, Draper T, Mcsweegan E, et al. A critical appraisal of “chronic Lyme disease”. N Engl J Med 357(14):1422-1430, 2007.

Feng J, Auwaerter PG, Zhang Y. Drug combinations against Borrelia burgdorferi persisters in vitro: eradication achieved by using daptomycin, cefoperazone and doxycycline. PLoS One 10(3):e0117207, 2015a.

Feng J, Li T, Yuan Y, Yee R, Zhang Y. Biofilm/persister/stationary phase bacteria cause more severe disease than log phase bacteria i biofilm Borrelia burgdorferi not only display more tolerance to Lyme antibiotics but also cause more severe pathology in a mouse arthritis model: implications for understanding persistence, PTLDS and treatment failure. https://www.biorxiv.org/content/10.1101/440461v3, 2018.

Feng J, Shi W, Zhang S, Zhang Y. Persister mechanisms in Borrelia burgdorferi: implications for improved intervention. Emerg Microbes Infect 4(8):e51, 2015b.

Feng J, Wang T, Shi W, Zhang S, Sullivan D, Auwaerter PG, Zhang Y. Identification of novel activity against Borrelia burgdorferi persisters using an FDA approved drug library. Emerg Microb Infect 2014:3, e49; doi:10.1038/emi.2014.1053, Jul. 2, 2014a.

Feng J, Wang T, Zhang S, Shi W, Zhang Y. An optimized SYBR Green I/PI assay for rapid viability assessment and antibiotic susceptibility testing for Borrelia burgdorferi. PLoS One 9(11):e111809, 2014b.

Feng J, Zhang Y. Proteomic analyses of morphological variants of Borrelia burgdorferi shed new light on persistence mechanisms: implications for pathogenesis, diagnosis and treatment. https://www.biorxiv.org/content/10.1101/501080v1, 2018.

Hodzic E, Feng S, Holden K, Freet KJ, Barthold SW. Persistence of Borrelia burgdorferi following antibiotic treatment in mice. Antimicrob Agents Chemother 52(5):1728-1736, 2008.

Hodzic E, Imai D, Feng S, Barthold SW. Resurgence of persisting non-cultivable Borrelia burgdorferi following antibiotic treatment in mice. PLoS One 9(1):e86907, 2014.

Lantos PM. Chronic Lyme disease: the controversies and the science. Exp Rev Anti Infect Ther 9(7):787-797, 2011.

Lantos PM, Auwaerter PG, Wormser GP. A systematic review of Borrelia burgdorferi morphologic variants does not support a role in chronic Lyme disease. Clin Infect Dis 58(5):663-671, 2014.

Liu GY, Essex A, Buchanan JT, Datta V, Hoffman HM, Bastian JF, Fierer J, Nizet V. Staphylococcus aureus golden pigment impairs neutrophil killing and promotes virulence through its antioxidant activity. J Exp Med 202(2):209-215, 2005.

Ma Y, Seiler KP, Eichwald EJ, Weis JH, Teuscher C, Weis JJ. Distinct characteristics of resistance to Borrelia burgdorferi-induced arthritis in C57BL/6N mice. Infect Immun 66(1):161-168, 1998.

Marques A. Chronic Lyme disease. In: Lyme disease: an evidence based approach. pp 248-258. (Ed. JJ Halperin.) CABI, Oxfordshire, UK, 2011.

Marques A, Telford SR, 3rd, Turk SP, Chung E, Williams C, Dardick K, Krause PJ, Brandeburg C, Crowder CD, Carolan HE, Eshoo MW, Shaw PA, Hu LT. Xenodiagnosis to detect Borrelia burgdorferi infection: a first-in-human study. Clin Infect Dis 58(7):937-945, 2014.

Merilainen L, Brander H, Herranen A, Schwarzbach A, Gilbert L. Pleomorphic forms of Borrelia burgdorferi induce distinct immune responses. Microbes Infect 18(7-8):484-495, 2016.

Miklossy J, Kasas S, Zurn AD, Mccall S, Yu S, Mcgeer PL. Persisting atypical and cystic forms of Borrelia burgdorferi and local inflammation in Lyme neuroborreliosis. J Neuroinflammation 5:40, 2008.

Radolf JD, Caimano MJ, Stevenson B, Hu LT. Of ticks, mice and men: understanding the dual-host lifestyle of Lyme disease spirochaetes. Nat Rev Microbiol 10(2):87-99, 2012.

Sapi E, Balasubramanian K, Poruri A, Maghsoudlou JS, Socarras KM, Timmaraju AV, Filush KR, Gupta K, Shaikh S, Theophilus PA, Luecke DF, Macdonald A, Zelger B. Evidence of in vivo existence of Borrelia biofilm in borrelial lymphocytomas. Eur J Microbiol Immunol 6(1):9-24, 2016.

Sharma B, Brown AV, Matluck NE, Hu LT, Lewis K. Borrelia burgdorferi, the causative agent of Lyme disease, forms drug-tolerant persister cells. Antimicrob Agents Chemother 59(8):4616-4624, 2015.

Steere AC, Strle F, Wormser GP, Hu LT, Branda JA, Hovius JW, Li X, Mead PS. Lyme borreliosis. Nature Rev Dis Primers 2:16090, 2016.

Straubinger RK, Summers BA, Chang YF, Appel MJ. Persistence of Borrelia burgdorferi in experimentally infected dogs after antibiotic treatment. J Clin Microbiol 35(1):111-116, 1997.

Stricker R, Fesler M. Chronic Lyme disease: a working case definition. Am J Infect Dis 14:1-44, 2018.

Wormser GP, Dattwyler RJ, Shapiro ED, Halperin JJ, Steere AC, Klempner MS, Krause PJ, Bakken JS, Strle F, Stanek G, Bockenstedt L, Fish D, Dumler JS, Nadelman RB. The clinical assessment, treatment, and prevention of Lyme disease, human granulocytic anaplasmosis, and babesiosis: clinical practice guidelines by the Infectious Diseases Society of America. Clin Infect Dis 43(9):1089-1134, 2006.

Yee R, Yuan Y, Brayton C, Tarff A, Feng J, Shi W, Behrens A, Zhang Y. Biofilm/persister/stationary phase bacteria cause more severe disease than log phase bacteria II infection with persister forms of Staphylococcus aureus causes a chronic persistent skin infection with more severe lesion that takes longer to heal and is not eradicated by the current recommended treatment in mice. https://www.biorxiv.org/content/10.1101/476465v2, 2018.

Zhang Y. Persisters, persistent infections and the Yin-Yang model. Emerg Microb Infect (Nature Publishing Group) 3, e3; doi:10.1038/emi.2014.3, 2014.

Zhang Y, Yew WW, Barer MR. Targeting persisters for tuberculosis control. Antimicrob Agents Chemother 56(5):2223-2230, 2012.

[Discovery Medicine; ISSN: 1539-6509; Discov Med 27(148):125-138, March 2019. Copyright © Discovery Medicine. All rights reserved.]