Lyme disease, a Zoonosis, is the most commonly reported vector-borne disease in the United States and approximately affects 300,000 individuals annually in North America [1
] and is spread by the spirochete Borrelia burgdorferi
sensu stricto (hereafter termed B. burgdorferi or Bb
). The clinical manifestations of Lyme disease include three phases [2
]. Early infection involves localized erythema migrans, followed within days or weeks by dissemination to the nervous system, heart, or joints. Without antibiotics treatment, 60% of patients with Lyme disease in the United States develop arthritis, which may recur at intervals and last for months or years. A fewer number of patients (4 to 10%) suffer carditis, which is an early and nonrecurring manifestation of the infection [3
]. The antibiotic treatment using oral doxycycline is effective for most patients at the early localized stage of Lyme disease [4
]. Several studies indicate that disseminated infection is not eradicated by conventional antibiotics such as tetracycline, doxycycline, amoxicillin, or ceftriaxone in animal models such as mice [5
], dogs [8
], ponies [9
], and non-human primates [10
]. Several reports also showed that antibiotics daptomycin and cefoperazone in combination with doxycycline or amoxicillin effectively eliminated B. burgdorferi
]. However, these antibiotic combinations failed to act against B. burgdorferi
biofilm forms [13
]. Macrolides are also less effective against B
, and resistance of these spirochetes to erythromycin has been reported [14
]. As a result, macrolides including azithromycin, clarithromycin and erythromycin are recommended only for patients who are intolerant to the first-line therapy. Therefore, based on these observations, new mechanistic classes of antibiotics need to be developed to treat infections arising from various forms of B. burgdorferi
. Hence, the discovery of new antimicrobials that could be used alone or in combination with other antibiotics will be highly beneficial for drug-intolerant patients and potentially for patients suffering from chronic Lyme disease that is refractory to other agents.
One approach to expedite the development of new antibiotics is to repurpose preexisting drugs that have been approved for the treatment of other medical conditions. Previously, we screened drugs (80% of them are FDA approved, with a total of 4366 chemical compounds from four different libraries) with high efficacy against the log and stationary phase of B. burgdorferi
by BacTiter-Glo™ Assay. Amongst them, disulfiram (AntabuseTM
), an oral prescription drug for the treatment of alcohol abuse since 1949, was found to have the highest anti-persister activity against B. burgdorferi
]. In addition, disulfiram and its metabolites are potent inhibitors of mitochondrial and cytosolic aldehyde dehydrogenases (ALDH) [16
]. Recent U.S. clinical trials using repurposed disulfiram treatments include: methamphetamine dependence (NCT00731133); cocaine addiction (NCT00395850); melanoma (NCT00256230); muscle atrophy in pancreatic cancer (NCT02671890); HIV infection (NCT01286259); a modulator of amyloid precursor protein processing (NCT03212599); and a recently initiated for Lyme disease treatment (NCT03891667) [17
]. In the area of infectious diseases, disulfiram has been shown to have antibacterial [18
] and anti-parasitic [20
] properties. A recent study showed that disulfiram has beneficial effects in the treatment of patients with Lyme disease and babesiosis [21
]. Disulfiram is an electrophile that readily forms disulfides with thiol-bearing substances. B. burgdorferi
possesses a diverse range of intracellular cofactors (e.g., coenzyme A reductase) [22
], metabolites (e.g., glutathione) and enzymes (e.g., thioredoxin) [23
] containing thiophilic residues that can be modified by disulfiram through thiol-disulfide exchanges to evoke antimicrobial effects. Therefore, disulfiram has the potential to inhibit B. burgdorferi
metabolism by forming mixed disulfides with metal ions [24
] and our group previously showed that B. burgdorferi
requires zinc and manganese as co-factors for key biological processes [25
]. Recently, we have shown that the drug azlocillin can be effective in controlling doxycycline-persisters of B. burgdorferi sensu stricto JLB31
in both in vitro and in vivo studies [26
In the present study, we evaluated the antibacterial activities of disulfiram against log- and stationary phase cultures of B. burgdorferi sensu stricto B31 MI. Furthermore, the bactericidal activity of disulfiram in vivo was determined using the C3H/HeN mouse model of Lyme disease at days 14 and 21 B. burgdorferi sensu stricto B31 MI infection, a timeline that denotes the early onset of chronic infection.
Since antibiotics are the top-of-the-line options to treat infections, there remains a dire need and a practical approach to bring more efficient antibiotics to the clinic. The repurposing of FDA approved antibiotics through fast-track approvals can be an excellent solution. In the current study, we evaluated the borreliacidal potential of an FDA approved drug, disulfiram, using in vitro and in vivo experimental models based on our previous high-throughput screening hits [15
]. We performed preliminary in vitro antimicrobial assays by Bac-titer Glo assay with a wide range of disulfiram concentrations (0.625 µM to 100 µM). Later, we confirmed the preliminary results by comparing the antimicrobial effect of disulfiram to that of doxycycline and used reliable quantitative methods to establish the bactericidal activity [33
]. Disulfiram in both soluble forms (DMSO or cyclodextrin) inhibited the growth of B. burgdorferi
strain B31 MI at an MIC90
range of 0.74 to 2.97 µg/mL in case of log-phase cultures (~94%) and at 1.48 µg/mL in case of stationary phase cultures (~90%), (Figure 1
and Figure 2
), with MBC varying from 1.48 µg/mL to 2.97 µg/mL for log and stationary phase cultures. The immediate deceleration in log and stationary phases of B. burgdorferi
growth at the low dose of disulfiram treatment is attributed to the rapid cleavage of disulfiram by thiophilic residues in intracellular cofactors (e.g., coenzyme A reductase) [22
], enzymes (e.g., thioredoxin) [23
], metal ions (e.g., zinc and manganese) [25
], and cofactors of B. burgdorferi
, which are hypothesized to instigate an abrupt halt in B. burgdorferi
metabolism, thereby killing B. burgdorferi
. A similar mechanism of action is proposed for pathogens like Giardia
, drug-resistant Mycobacterium
, and multidrug-resistant Staphylococcus
]. Disulfiram drug efficacy has been enhanced to a maximum level with the concentration range from 2.5 µM to 10 µM (Figure 1
and Figure 2
), above which the effect is diminished (25 µM to 100 µM), atypical bell-shaped curve. We have also found that more than 1000 citations of molecules demonstrated a similar efficacy pattern to ours in numerous previous studies [36
]. Anticancer drugs such as fulvestrant, sorafenib, and crizotinib, have critical aggregation concentrations (CACs) of 0.5−20 μM. Below their respective CACs, these drugs exist in a classic monomeric form where, at sufficiently high (monomeric) concentrations, they are toxic to cells; in contrast, above their CACs, these drugs form colloidal aggregates that are substantially less cytotoxic in cell assays [37
]. In addition, it has been reported that disulfiram producing a biphasic cytotoxic response in some breast cancer cell lines [39
], human tumor cell lines [40
], and neuronal cells [41
]. In these cell lines 1 μM disulfiram was toxic involving consistent loss of cell viability. These effects disappeared at 10 μM where morphology was comparable to diluent controls; then, the increase of the disulfiram concentration to 100 μM restored the toxic 1 μM phenotype. These studies prompted us to perform DLS and AFM imaging to investigate the aggregation formation of disulfiram in DMSO and CD. Our DLS results clearly indicates that disulfiram in DMSO and CD has a critical aggregation concentrations (CACs) of 10 μM and this was supported by the AFM imaging showing above 25 μM aggregation of disulfiram occurs. Based on these results we concluded that at high concentrations (25 μM to 100 μM) of disulfiram aggregate into large particles that cannot diffuse across the cell membrane, and as their concentration rises they act as sinks for even the free monomer, leading to a bell-shaped concentration−response shown in Figure 1
and Figure 2
. Further extensive study is underway to establish the possibility of target-based mechanisms.
Disulfiram is an oral medication that is approved by the U.S. Food and Drug Administration (FDA) for the administration of up to 500 mg daily [42
]. Pharmacokinetic studies in humans have shown that disulfiram has a half-life (t1/2
) of 7.3 h and a mean plasma concentration of 1.3 nM, although significant intersubjective variations are noted [43
]. The toxicity of both disulfiram and its metabolites have also been broadly investigated in cell and animal studies, which yielded no evidence for teratogenic, mutagenic, or carcinogenic effects [44
]. DMSO is toxic at a low dose in vivo [45
] so, we have used non-toxic cyclodextrin [46
] as a solubilizing agent for disulfiram in vivo studies. Based on these observations, we conducted our preliminary in vivo mouse efficacy studies by administering (I.P.) low dose of disulfiram (i.e., 10 mg/kg of body weight) to infected C3H/HeN mice for 5 days and found that these mice were not able to clear the B. burgdorferi
from tissues (unpublished data). As shown in the current study, however, when we repeated in vivo C3H/HeN mouse efficacy studies by administering (I.P.) 75 mg/kg of body weight disulfiram to infected mice for 5 days, all infected mice have either reduced or cleared the bacteria in most of the tissues at 21 and 28 post infection (Table 1
and Table 2
and Figure 3
). C3H mice develop bradycardia and tachycardia beginning on day 7 through 60 days after B. burgdorferi
inoculation and triggers severe inflammatory responses particularly in C3H mice on days 15 to 21 post infection [28
]. Thus, we have chosen the C3H/HeN mouse model for our efficacy studies and day 14 or day 21 post infection as time points for antibiotics treatments. Lyme carditis, a macrophage-mediated pathology, is not directly influenced by B. burgdorferi
specific antibodies, but by inflammatory micro environment coming from pro-inflammatory Th1 cytokines (IL-1β, TNF, and IFN-γ), anti-inflammatory Th2 cytokine (IL-10), and other M1/M2 macrophage-polarizing factors such as iNOS and NOS2 derived from macrophages and T cells [48
]. Similarly, chemokines (e.g., MIP-2, KC, and RANTES) preferentially attract monocytes and lymphocytes significantly contributing to the inflammation and tissue damage in Lyme disease [51
]. We have shown that in disulfiram-treated mice groups there is a significant reduction in the infiltration of leucocytes in the heart wall and no inflammation (inactive carditis) compared to the doxycycline-treated group (active mild carditis) and PBS infected group (active severe carditis) at day 21 or day 28 post infection (Figure 3
). This implies that disulfiram treatment reduced the inflammatory microenvironment by reducing the inflammatory chemokines (MIP-2 and RANTES), and cytokines (IL-10, IL-1β, TNF, and IFN-γ) and further reduces the disease severity in the heart. Macrophage phenotype is flexible, and once the infection is cleared and a more anti-inflammatory environment is created, macrophages switch to a pro-resolution M2 phenotype [52
]. Henceforth, in disulfiram-treated mice, the level of NOS2 (M2 polarizing factor) was elevated compared to those of the doxycycline-treated and PBS infected groups at day 21 or day 28 post infection (Figure 4
). However, the underlying mechanism involved in differential expression of chemokines and cytokines and their effect on disease severity needs to be further investigated.
In addition, our study found a lower bacterial burden in the ear, heart and bladder of disulfiram-treated mice when compared to PBS-treated infected mice at 21 days post infection. This result indicates that the disulfiram administration has promoted the antibody-mediated killing early in the infection, and thus it not only limited the B. burgdorferi
colonization in tissues but also altered the development of adaptive immune response, which is aligned to the reduction of tissue inflammation as observed in heart samples [53
]. In fact, B. burgdorferi
infection leads to strong and sustained IgM response and delayed development of long-lived antibody and B cell memory [55
]. So, disulfiram-treated mice might have accelerated long lived antibody and B cell memory development, which resulted in statistically lower amount of Bb specific IgM, total IgM, IgG and IgG1 at day 21 post infection (Figure 5
and Figure S6
). A similar pattern with a statistically lower amount of both Bb specific IgM and IgG was observed for disulfiram treated mice on day 28. On the other hand, at day 28 post infection, disulfiram-treated mice have higher amounts of total IgM, IgG1 and IgG3 isotypes similar to the saline-treated mice, which all bind to C1q and activate the classical pathway, whereas IgG2a and IgG2b bind to the Fc receptor [56
]. As such, it is likely that those immuno-complexes formed with C1q-binding antibodies cannot be opsonized by the complement system during infection due to the absence of C1q, thus fail to be engulfed by phagocytes and accumulated within the circulation system. We have not further studied Bb-IgG subsets in detail because the current study involves short term experiments (21 and 28 days post infection are last time points). Therefore, our data indicate that memory to B. burgdorferi
infection may have not be formed until late during infection. Moreover, to what extent long-lived plasma cells contribute to immune protection remains to be studied, which is the subject of our future study using chronic Lyme arthritis model. Lymphoadenopathy observed during Lyme borrreliosis is caused by a massive increase in lymph node cellularity triggered by the accumulation of live B. burgdorferi
spirochetes into the lymph nodes. This increase in cellularity is caused by accumulation of CD19+ B cells [30
]. Disulfiram treatment alleviates lymphoadenopathy by reducing the percentage of CD19+ B cells in day 28 post infected mice (Figure 6
). An important function of CD4+ T cells is their ability to enhance antibody-mediated immunity by driving affinity maturation and the development of long-lived plasma cells and memory B cells [57
]. However, it appears that the response of protective B cells to B. burgdorferi
, a highly complex pathogen expressing many immunogenic surface antigens, is confined to T-independent antibody responses alone. Even though disulfiram-treated mice induced an increase in percentage of CD3+ CD4+, Naïve, effector and memory T cells, further studies are needed to understand the role of these increased T cells in disease resolution and bacteria clearance.
In summary, the disulfiram drug not only successfully cleared the bacteria but also suppressed the inflammatory responses in heart tissues of C3H/HeN mice on day 28 post infection. Furthermore, disulfiram reduced antibody titers followed by nullifying lymphoadenopathy. The preclinical data offered here is beneficial in ascertaining the effectiveness of disulfiram and aids in performing future mechanistic and translational research studies. Moreover, the inhibitory effects of disulfiram on borrelial metabolism leave a room to exploit multiple mechanisms that can be used as therapeutic targets. Although the results from our in vivo study cannot be extrapolated directly to clinical practice at this point, we strongly believe they form a strong basis for future follow-up studies, and promote the development of effective formulations of disulfiram for clinical management of Lyme disease.
4. Materials and Methods
4.1. Culturing and Growth Conditions of B. burgdorferi B31 MI
Borrelia burgdorferi sensu stricto low passage strain B31 MI was (obtained from Luciana Richer, US biologics, Memphis, TN, USA) used for MIC tests and all infection studies in C3H/HeN mice. Bacteria cultures were started by thawing −80 °C glycerol stocks of B. burgdorferi (titer, ~107 CFU/mL) and diluting 1:40 into fresh Barbour-Stoner-Kelly (BSK) complete medium with 6% rabbit serum followed by incubating at 33 °C. After incubation for 4–5 days log phase, and 8–9 days stationary-phase B. burgdorferi culture (~106 borrelia/mL) were transferred into a 48-well plate for evaluation with the drugs.
4.2. Drug Formulations
The disulfiram (Sigma, St. Louis, MO, USA) stock solution (50 mM) was made by dissolving in sterile 30% hydroxypropyl β-cyclodextrin (Sigma) and also another disulfiram stock solution (20 mM) was made by dissolving in sterile 100% DMSO (Tocaris bioscience,Bristol, UK). A stock solution of 100 mM of doxycycline (as a positive control) was made by dissolving the doxycycline powder in ultra-pure Milli Q water. All drug stocks were passed through 0.22 μm filters (Millipore-Sigma, St. Louis, MO, USA), used within 72 h of preparation and were not subject to freezing temperatures. Working solutions were made by mixing the desired volume of stock solutions in the desired volume of ultra-pure MilliQ water. Furthermore, the vehicle for hydroxypropyl β-cyclodextrin (cyclodextrin) and DMSO controls were made similarly and it is important to note that the vehicle controls were identical to the test formulation in every single aspect except for the active ingredient. This measure was strictly followed for vehicle control wherever used in the entire study.
4.3. In-Vitro Testing of Antibiotics by Microdilution and Dark Field/Fluorescent Methods
A standard microdilution method was used to determine the minimum inhibitory concentration (MIC) of the antibiotics tested before [58
]. Approximately, 1 × 106 B. burgdorferi
(log and stationary phase respectively) were inoculated into each well of a 48-well tissue culture microplate containing 900 μL of BSK medium per well. The cultures were then treated with 100 μL of each drug at varying concentrations ranging from 0.625, 1.25, 2.5, 5, 10 and 20 μM. Control cultures were treated with respective vehicles, and all experiments were run in triplicate. The well plate was covered with parafilm and placed in the 33 °C incubator with 5% CO2
for 4 days. Spirochetes proliferation was assessed using a bacterial counting chamber (Petroff-Hausser Counter) after the 4–5 days incubation followed by dark-field microscopy respectively.
SYBR® Green I/PI by Fluorescent Microscopy
As a confirmation test, the SYBR Green/PI method was used for cell growth by directly counting live and dead bacteria by fluorescent microscopy. To evaluate live and dead cells, standard SYBR Green I/propidium iodide (SYBR Green I/PI) was performed as previously described [34
]. To 1 mL of sterilized distilled water, 10 μL of SYBR Green I (10,000× stock, Invitrogen, Grand Island, NY, USA) and 30 μL of propidium iodide (Thermo Scientific, Waltham, MA, USA) were briefly mixed. The staining mixture (10 μL) was added to all the wells containing B. burgdorferi
and was incubated in the dark for 15 min. The standard equation was determined from 1 × 106
cells (logarithmic phase) and 5 × 106
cells (stationary phase). A live and dead population was prepared. For the dead cell population, the cells were killed by adding 300 μL of 70% iso-propyl alcohol (Fisher Scientific, Santa Clara, CA, USA). We counted (200×) live (SYBR green) or dead (PI) bacteria cells in each condition of control or treatment (took at least 6 fields per condition) by fluorescent microscope. We combined the averages of live or dead bacteria cells per each condition to obtain total cells per condition. To generate a standard curve, different ratios of live and dead cell suspensions (live:dead ratios = 0:10, 2:8, 5:5, 8:2, 10:0) were added to the wells and stained as aforementioned in methods. Using the least square fitting analysis, the regression equation was calculated between the percentage of live bacteria and green/red fluorescence ratios. The regression equation was used to calculate the percentage of live or dead cells in each sample. Also, images of the treated sample were taken using fluorescent microscopy.
To further determine the minimum bactericidal concentration (MBC) of the antibiotics tested (the minimum concentration beyond which no motile spirochetes can be sub cultured after a 21 days incubation period), wells of a 48-well plate were filled with 1 mL of BSK medium and 20 μL of antibiotic-treated spirochetes were added into each of the wells. The well plate was wrapped with parafilm and placed in the 33 °C incubator with 5% CO2 for 21 days. After the incubation period, the plate was removed and observed for motile spirochetes in the culture by dark-field and further cell proliferation was assessed using the SYBR Green I/PI assay fluorescence microscopy. All these experiments were repeated at least three times. Statistical analyses were performed using Student’s t-test.
Semisolid plating method: We performed a semisolid plating procedure as described [60
]. The 2× BSK-II medium was prepared in the following manner. To the 500 mL of CMRL-1066 medium: 50 g of bovine serum albu-min (Sigma), 5 g neopeptone (BD), 6.6 g HEPES acid (Sigma), 0.7 g sodium citrate (Sigma), 5 g glucose (Sigma), 2 g yeastolate (BD), 2.2 g sodium bicarbonate (Fisher), 0.8 g sodium pyruvate (Sigma), 0.4 g N-acetyl-glucosamine (Sigma) were added and mixed thoroughly. Finally, the pH of the medium was adjusted to 7.6 and filtered through 0.2 µm filter units. For plating the medium is mixed in the following way. The 250 mL of 2× BSK-II medium prewarmed at 55 °C was mixed with 250 mL 1.75 mL of agarose (55 °C) and 35 mL sterilized rabbit serum and equilibrated to 55 °C. Then 8 mL of equilibrated BSK-II medium was dispensed into 100-mm Petri dishes as bottom agar and allowed to solidify. Finally, the sample was resuspended in 0.5 mL fresh BSK-II medium and mixed with 8 mL of BSK-II agarose medium (55 °C) and poured as a top agar. The plates were incubated in the incubator with 5% CO2
at 35 °C for a minimum of 21 days. The white visible colonies were counted after 21 days for the analysis. Finally, semisolid plating was chosen to obtain the exact count of the growing borrelial colonies as colony forming units (CFU).
4.4. Dynamic Light Scattering
Since the disulfiram is insoluble in water, the stock solutions of 1M disulfiram were prepared either in DMSO or in 30% (w/v) hydroxypropyl β-cyclodextrin (CD). Disulfiram was then diluted in bovine serum albumin (BSA) solution to obtain disulfiram concentration 0.125 µM, 0.25, 0.5, 10, 25, 50, and 100 µM, and 5% (w/v) BSA in the final solution for DLS. The measurements were obtained from the Brookhaven 90-Plus particle size analyzer (Brookhaven instruments corporation) at an angle of 90° with 10% dust cutoff filter. The results represent an average of three measurements.
4.5. Atomic Force Microscopy
Atomic force microscopy (AFM) samples were prepared from drugs disulfiram-CD and disulfiram-DMSO solutions of respective concentrations (100 μM, 25 μM, 10 μM and 5 μM) on clean silicon wafers that were plasma-treated to increase hydrophilicity. Then, 10 µL droplets were deposited, spreading for most of the surface of 1 cm2 wafers and were quickly dried in a desiccator under vacuum to minimize additional aggregation due to local increase in concentrations. AFM imaging was performed with NX-10 AFM (Park Systems, Suwon, Korea) operating in non-contact mode with Micromasch NCS15 AL BS tips (NanoandMore, Watsonville, CA, USA) at 0.8 Hz with 256 pixels per line.
4.6. Animal Experiments Ethical Statement
All mice were maintained in the pathogen-free animal facility according to animal safety protocol guidelines at Stanford University under the protocol ID APLAC-30105. All experiments using animals were conducted according to the Administrative Panel on Laboratory Animal Care guidelines at Stanford University.
4.7. In Vivo Testing of Drugs in Immunocompetent C3H/HeN Mice
Four-week-old female C3H/HeN mice were purchased from Charles River Laboratories, Wilmington, Massachusetts. The mice (5 weeks) were infected subcutaneously close behind the neck with 0.1 mL BSK medium containing log phase 105 B. burgdorferi
B31 MI. For in vivo studies, we used only disulfiram soluble in cyclodextrin. On the 14 and 21 days post Bb infection, the mice were intraperitoneally administered a daily dose of drugs, disulfiram (75 mg/kg) and doxycycline (50 mg/kg) for 5 consecutive days (Figure 3
A). After 48 h of the last dose of administering compounds, both groups (day 21 and day 28 post Bb infection) of mice were terminated and their urinary bladders, ears, and hearts were collected. The DNA was extracted from the urinary bladder, ear and heart. The absence of B. burgdorferi
marked the effectiveness of the treatment in these organisms. Quantification of important pro/anti-inflammatory immune marker transcripts and histopathology of heart was also done. At termination on day 28 post infection, spleen and peripheral lymph nodes (axillary, brachial, cervical and inguinal) were also collected for immunophenotyping.
4.8. Quantitative (Q-PCR) and Real-Time PCR (RT-PCR) Analysis
Urinary bladder, ear punches, heart bases were homogenized, and DNA was extracted using the NucleoSpin tissue kit according to the manufacturer’s instructions (Macherey-Nagel, Düren, Germany). Q-PCR from the above tissues were performed in blinded samples using B. burgdorferi Fla-B gene-specific primers and a probe. These primers were listed as follows: Fla-B primers Flab1F 5′-GCAGCTAATGTTGCAAATCTTTTC-3′, Flab1R 5′-GCAGGTGCTGGCTGTTGA-3′ and TAMRA Probe 5′-AAACTGCTCAGGCTGCACCGGTTC-3′ according to the published protocol. Reactions were performed in duplicate for each sample. Results were plotted as the number of Fla B copies per microgram of tissue. The lower limit of detection was 10 to 100 copies of B. burgdorferi Fla-B DNA per mg of tissue. In addition to standard laboratory measures to prevent contamination, negative controls (containing PCR mix, Fla-B primers, probe, and Taq polymerase devoid of test DNA) were included.
Total RNA was extracted from tissues using the RNeasy mini kit (Qiagen, Germantown, MD, USA) and reverse-transcribed using a high-capacity cDNA reverse transcription kit (Invitrogen, USA). cDNA was subjected to real-time PCR using primer and TAMRA probes (Stanford Protein and Nucleic acid Facility) previously described [61
]. PCR data are reported as the relative increase in mRNA transcript levels of CxCL1 (KC), CxCL2 (MIP-2), CCL5 (RANTES), IL-10, TNF-α, IFN-γ, and iNOS/NOS2 normalized to respective levels of GAPDH.