Possible role of glial cells in the onset and progression of Lyme neuroborreliosis

B. burgdorferi strain B31 clone 5A19 spirochetes, passage one, isolated from an ear biopsy of a previously infected mouse, were grown in Barbour-Stoenner-Kelly (BSK)-H medium supplemented with 6% rabbit serum and antibiotics (Rifampicin at 45.4 mg/ml, Phosphomycin at 193 mg/ml and Amphotericin at 0.25 mg/ml) (Sigma, St. Louis, MO) to late logarithmic phase under microaerophilic conditions. Spirochetes were pelleted at 2000 × g for 30 min at RT. At the end of the run the rotor was left to coast without breaking so as to minimize damage to the live spirochetes. The culture was washed twice using phosphate buffered saline (PBS) pH 7.2 (Invitrogen, Grand Island, NY) and resuspended in RPMI 1640 medium (Biowhittaker, Walkersville, MD) to contain a suspension of 1 × 10⁸ spirochetes/ml.

Seven 3–7-year old male rhesus macaques (Macaca mulatta) of Chinese origin were used in this study. The inoculation protocol was approved by the Institutional Animal Care and Use Committee of the Tulane National Primate Research Center. Anesthetized animals received up to 1 ml of RPMI 1640 medium with 1 × 10⁸ spirochetes (n = 5), or no spirochetes (n = 2) into the cisterna magna, after removal of an equivalent volume of CSF. This dose was found to be safe in studies involving intrathecal inoculation of Treponema pallidum, another spirochete, in rhesus macaques [30]. Similar inoculum doses were used previously in the rhesus monkey [31‐34]. To monitor CNS inflammation following inoculation, CSF (0.5–1.0 ml) and serum (5 ml of blood) were collected on a weekly basis for 4 weeks, and then once every two weeks until the end of the study for each animal. Two inoculated animals (DR50, EL81) and one control (EL66) were necropsied at six weeks post-inoculation (PI) and three inoculated animals (DH50, EP51, CH82) and one control (EJ86) at 12 weeks PI. The procedure used for euthanasia was consistent with the recommendations of the American Veterinary Medical Association's Panel on Euthanasia.

Serum antibody levels to the VlsE C6 peptide (B. burgdorferi B31) were quantified by ELISA using a procedure described previously [29]. As is customary, the cut-off line for positive ELISA values was set at the mean value of all of the pre-immune serum specimens plus three times the standard deviation of that mean. The probability that an ELISA value above this cut-off line is negative is p = 0.003.

CSF and serum cytokines and chemokines were quantified with the Bio-Plex Human Cytokine 17-Plex Panel following the manufacturer's instructions (Bio-Rad, Hercules, CA). The analytes detected by this panel are: Hu IL-1beta, Hu IL-2, Hu IL-4, Hu IL-5, Hu IL-6, Hu IL-7, Hu IL-8, Hu IL-10, Hu IL-12, Hu IL-13, Hu IL-17, Hu G-CSF, Hu GMCSF, Hu IFN-gamma, Hu MCP-1/CCL2, Hu MIP-1beta/CCL4, Hu TNF-alpha. The multiplex plate was read using a Bio-Plex 200 Suspension Array Luminex System (Bio-Rad). CXCL13 concentration was measured with a sandwich ELISA (R&D, Minneapolis, MN). As with the serology ELISA, the cut-off line for positive Bio-Plex and sandwich ELISA values was set at the mean value for each mediator of all of the pre-infection and control CSF specimens plus three times the standard deviation of that mean.

Tissues were collected from various regions of the brain and spinal cord and DRG, and fixed in formalin or Z-fixative (Anatech, Battle Creek, MI) for routine histopathological evaluation. For the purpose of detecting intracytoplasmic localization of cytokines, fresh unfixed tissues were collected from various regions of the brain and spinal cord, as well as DRG at necropsy, and immediately processed for blocking of intracytoplasmic cytokines (modification of the method previously described [35], by incubation of tissue slices for four hours in RPMI medium (Invitrogen) containing 10% fetal bovine serum (Invitrogen) and the fungal metabolite brefeldin A (Invitrogen), which blocks export of intracellular proteins via the endoplasmic reticulum. Tissues were then fixed in 2% paraformaldehyde in PBS pH 7.0 (USB, Cleveland OH), and cryopreserved as described previously [35]. Tissue samples from various regions of the brain and spinal cord were also collected at necropsy in BSK-H medium, supplemented with 6% rabbit serum and antibiotics (Sigma) for detection of B. burgdorferi by culture.

For in situ analysis, the cryopreserved tissues were cryosectioned into 16-μm sections as previously described [35]. Tissues collected in formalin or fixed with Z-fixative were also sectioned into 10–15 μm sections, depending on the availability of tissue. Presence of inflammatory lesions was determined by routine histopathological evaluation of tissues sectioned at 5 μm and stained with hematoxylin and eosin (H&E). Immunohistochemical staining was done using several monoclonal and polyclonal antibodies (anti-CD3, CD20, and CD68, Table 1). All histological sections were incubated with the primary antibody for 1 hour at room temperature, followed by biotinylated anti-rabbit or anti-mouse secondary antibodies (Dako) as appropriate for 30 minutes. Finally, sections were incubated with avidin-biotin complex (ABC) (Vector labs, Burlingame, CA) for 30 minutes, and the reaction was visualized with 3,3'-diaminobenzidine (Dako) as the chromogen. Sections were subjected to immunofluorescence staining as previously described [22]. The primary antibodies against various immune mediators, cell phenotypic markers and B. burgdorferi are listed in Table 1. Relevant isotype controls (Sigma) at the concentration of the corresponding primary antibodies were also included.

Glial and neuronal apoptosis was assessed by the in situ-TUNEL assay in tissues collected from the brain and spinal cord as well the DRG, that were directly fixed in 2% paraformaldehyde at collection and cryopreserved and frozen as previously described [22]. Sections were stained for any one of the following brain cell markers: NeuN (neurons), IBA-1 (microglia), GFAP-cy3 (astrocytes), or S-100 (astrocytes, oligodendrocytes, Schwann cells, satellite cells) by incubation with the appropriate primary antibody (Table 1) followed by secondary antibodies conjugated with Alexa Flour 568. Sections were fixed in 2% paraformaldehyde for 15 min, followed by a 15-min wash in PBS. Sections were then subjected to the terminal deoxynucleotidyl transferase mediated UTP nick end labeling (TUNEL)-ApopTagPlus fluorescein in situ apoptosis assay (Chemicon, Temecula CA) as per the manufacturer's instructions. Sections were also stained with anti-B. burgdorferi antibody as described above followed by secondary antibody conjugated to Alexa Fluor 633. Contiguous sections stained with isotype controls at the respective concentrations of the primary antibody were included. To confirm the identity of S-100-staining cells, S-100-stained (Alexa Fluor 568) sections were also stained with anti-GFAP followed by secondary antibody conjugated to Alexa Fluor 633 (blue), before doing the TUNEL assay. Cells that were only positive for S-100 (labeled red) were regarded as oligodendrocytes in the CNS and Schwann cells and/or satellite cells in the DRG [36], while those that were labeled pink due to an overlap of S-100 (red) and GFAP (blue) were considered to be astrocytes. Slides were washed and mounted as described above and stored at 4°C in the dark until viewed. The percentage of apoptotic cells from ten fields were counted from each section (more than 500 cells in all cases). The total number of NeuN or S-100 positive cells respectively in each section was ascertained, followed by the percentage of cells that colocalized with the TUNEL signal for each cell marker, All counts were made by viewing slides under a fixed magnification of 63 × (corresponding to an area of 0.05 mm²) using the confocal microscope (see below).

Confocal microscopy was performed using a Leica TCS SP2 confocal microscope (Leica Microsystems, Exton, PA) as previously described [22]. Photoshop CS3 (Adobe systems Inc., San Jose, CA) was used for image processing.

The statistical significance of the apoptosis data was evaluated using the one-way ANOVA non-parametric analysis, followed by the Tukey's stringency test using PRISM software (IBM).

Live B. burgdorferi spirochetes were recovered from the CSF cell pellets of all but one of the inoculated animals at various time points PI, as follows: animals DR50 and EL81 at week 6, DH50 at week 2 and EP51 at week 10, except for animal CH82, whose CSF cell-pellet cultures yielded no spirochetes. Tissues harvested at necropsy from the dura mater and the cervical spinal cord of one of the two animals (EL81) that were inoculated with B. burgdorferi and euthanized 6 weeks PI were also culture positive for B. burgdorferi. CSF cell-pellet or tissue cultures of the control animals were negative throughout.

CSF pleocytosis was evident in all of the B. burgdorferi-inoculated animals as early as one week PI; specimens from the control animals were essentially free of cells (Table 2). Pleocytosis was minimal in CH82, which peaked to about 35 leukocytes/μl by week two PI. The evaluation of the percentage of various leukocytes (differential counts) of cytospins prepared from the CSF samples showed a predominance of lymphocytes and macrophage/monocytes. The differential counts of leukocytes in the CSF of all of the animals with a pleocytosis > 50 cells/μl is shown in Fig. 1.

CSF from infected animals showed levels above the cut-off value for the cytokine IL-6 in animals DR50, EL81 and DH50, between week 1–3 (Fig 2a). Animal EP51 showed levels of IL-6 just exceeding the cut-off value at weeks 1 and 2 PI. The chemokine IL-8 was significantly elevated in the CSF of all of the inoculated animals (Fig 2b), reaching maximum levels ranging from 10.9 to 35 pg/ml between weeks 1 and 2 PI. Monocyte chemoattractant protein-1MCP-1/CCL2 was significantly elevated in all of the infected animals (Fig 2c) except CH82, ranging between 350 pg/ml to 500 pg/ml by week 2 PI. The levels of IL-6 in the CSF were higher than those observed in the serum of all of the infected animals, except CH82, (Table 3), indicating that this cytokine was produced primarily intrathecally. On the other hand, IL-8 levels were much higher in serum than they were in CSF (Table 3). The levels of CCL2 were considerably higher in the CSF compared to those found at the corresponding time points in serum, suggesting, once again, a predominantly CNS origin for this chemokine. The concentration of the B-lymphocyte chemokine CXCL13/BLC was elevated above the cut-off value in animals DR50, DH50, EL81, and EP51, reaching peak values of around 5500 pg/ml between weeks 2 and 4 PI (Fig 2d). The CSF levels of CXCL13 were higher in three out of four of the inoculated animals (DR50, EL81, and DH50) as compared to those in paired serum specimens (Table 3), implying a possible CNS origin for this mediator as well. The very early appearance of inflammatory mediators (week 1–2 PI) in the CSF indicates that innate immune responses are at play. In addition, the very high serum concentrations of IL-8 detected as early as week 1 PI, much higher than the corresponding CSF levels, are data to suggest that the infection became systemic soon after the intrathecal inoculation.

The evaluation of the time of onset of acquired immunity in the periphery was done by serially quantifying anti-VlsE serum antibodies. VlsE, the antigenic variation lipoprotein of B. burgdorferi, is expressed by the spirochete uninterruptedly upon the initiation of infection in mammals. We measured the anti-VlsE antibody response by quantifying antibodies to the VlsE C6 peptide. The earliest that this response was detected was by week 3 PI (animal DR50), Fig. 3A. In all of the other animals that had received a B. burgdorferi inoculation the anti-C6 response appeared between weeks 4 and 6 PI, except for animal CH82 (Fig. 3E), who did not have a detectable anti-C6 response. No anti-C6 response was detected in control animals (Fig. 3F, 3G).

Histopathological evaluation of tissues collected from the 2 animals that were euthanized at 6 weeks PI revealed severe multifocal lymphocytic, monocytic and plasmacytic leptomeningitis in the brain and spinal cord. Radiculitis was evidenced by inflammatory infiltrates in the nerve roots of the cervical, thoracic, lumbar and sacral spinal cord. Similar lesions were seen in animal DH50, which was euthanized at 12 weeks PI. A representative image of brain leptomeningitis is shown in Fig 4A. The cellular infiltrates contained, in decreasing order of abundance, B lymphocytes (CD20⁺), T lymphocytes (CD3⁺), and monocyte/macrophages (CD68⁺). Figure 4B depicts a confocal micrograph of a lesion found in the dura mater of animal DR50 showing the presence of T cells, B cells and monocyte/macrophages. Inflammatory infiltrates in the vicinity of B. burgdorferi antigen were observed in the dura mater (not shown) and frontal cortex of animals DR50 (Fig. 4C), EL81 and DH50. As with the meninges, the parenchymal cell infiltrates were composed chiefly by B and T lymphocytes and few macrophages. B. burgdorferi antigen was also present in tissues collected from the periventricular areas and spinal cord of infected animals (not shown). Radiculitis seen as inflammation of the dorsal roots collected from the cervical, thoracic, lumbar, and sacral spinal cord was observed in animals DR50, EL81 (6-week-long infection) and DH50 (12-week). A representative image of radiculitis is depicted in Fig. 4D. Inflammation and presence of B. burgdorferi antigen was also observed in the DRG, in animals DR50 and DH50. A chronic type of inflammation of DRG from animal DH50 is shown in Fig 4E. The presence of B. burgdorferi antigen in the vicinity of CD3-labeling-T cells in the DRG is shown in Fig. 4F. No inflammatory lesions were detected at necropsy in the remaining animals that were given a B. burgdorferi inoculation and were euthanized at 12-weeks PI (EP51 and CH82), or in the control animals.

IL-6 was produced by both astrocytes and neurons, as visualized by intracytoplasmic staining of spinal cord tissues from three of the five B. burgdorferi-inoculated animals (DR50, EL81 and DH50). Fig. 5A shows IL-6 produced by astrocytes in the spinal cord. IL-6 was also found in neurons as well as extracellular in the DRG of animals DR50 and DH50 (Fig. 5B). CXCL13 and CCL2, two of the mediators produced chiefly in the intrathecal compartment, were localized to microglia in the spinal cord (Fig. 5C and Fig. 5D respectively). CXCL13 was also detected in CD68-labeled macrophages in the lesions (not shown). In addition, CCL2 and CXCL13 were present in endothelial cells, especially in the periventricular areas of the brain (not shown). IL-8 was not detectable in any of the tissues at the time of necropsy.

IL-6, CCL2 and CXCL13 were undetectable in comparable tissues from control animals (Fig. 5E, F, and 5G). No IL-6 was detected in astrocytes from the spinal cord or in neurons from the DRG of a control animal (Fig 5E and 5F, respectively). Similarly, microglia from the spinal cord of control animals were devoid of mediators (IL-6, IL-8, CCL2 or CXCL13). A representative image of microglia lacking CCL2 is shown in Fig 5G. Non-specific signals were absent in contiguous sections that were stained with immunoglobulin isotype controls at the concentrations of the respective primary antibodies, followed by the corresponding secondary antibodies.

All of the infected animals except CH82 showed significant levels of both Schwann/satellite cell and neuronal apoptosis in the DRG as compared to that seen in the respective controls. The percentage of DRG Schwann/satellite cells (S-100⁺/GFAP^- staining cells) and neurons undergoing apoptosis in each of the animals in the study, as assessed by the in situ TUNEL assay is shown in Fig. 6A. Schwann/satellite cells showed significant levels of apoptosis ranging from 4.35 ± 2.39% (p < 0.0001) to 10.35 ± 3.11% (p < 0.0001). Neuronal apoptosis was less prevalent and ranged from 3.83 ± 1.77% (p < 0.001) to 6.58 ± 1.99% (p < 0.0001). Higher levels of both Schwann/satellite cell and neuronal apoptosis were observed in 2 of the animals necropsied at 12 weeks PI (EP51 and DH50) as compared to those necropsied at 6 weeks PI (Fig. 6A). A qualitative view of the extent of the apoptosis around neurons stained with antibody to NeuN observed in the DRG is shown in Fig 6B. Fig. 6C shows the TUNEL signal (green) seen in Schwann cells/satellite cells labeled with antibody to S-100 followed by secondary antibody conjugated to Alexa fluor 568 (red) that are surrounding a neuron that appears unstained as it does not take up S-100, but is also showing the TUNEL signal in the center and traces of B. burgdorferi antigen (blue) stained with antibody to whole B. burgdorferi followed by secondary antibody conjugated to Alexa fluor 633. Apoptotic DRG neurons in the vicinity of B. burgdorferi antigen are shown in Fig. 6D. No significant oligodendrocyte or neuronal apoptosis was detected in the tissues collected from the various regions of the brain and spinal cord. Apoptosis was not detected in astrocytes or microglia in any of the tissues.