Borrelia burgdorferistimulation of chemokine secretion by cells of monocyte lineage in patients with Lyme arthritis

For this study, which was carried out in 2007, frozen PBMCs were tested from 24 Lyme arthritis patients in whom large numbers of cells were available. Concomitant SFMCs were also available in 11 of the 24 patients. In the few patients in whom samples were obtained on more than one date, the first date was used for testing. The 24 patients were seen over a 12-year period, from September 1994 through January 2007. In initial experiments, frozen PBMCs that were collected from 4 normal control subjects from June 2006 through March 2007 were also tested. We used frozen cells from normal subjects so that the cells would be comparable with those from patients.

All patients met the Centers for Disease Control and Prevention criteria for the diagnosis of Lyme disease [15] and were entered into a study called 'Immunity in Lyme Arthritis'. The patients were treated with antibiotic therapy in accordance with the guidelines of the Infectious Diseases Society of America [16]. The Human Investigations Committees at Tufts Medical Center (Boston, MA, USA) (1987 to 2002) and Massachusetts General Hospital (2002 to 2009) approved the study, and all patients (or the parents of patients who were minors) provided written informed consent.

On the day of testing, frozen PBMCs (5 × 10⁶ cells/mL) or SFMCs from patients or normal control subjects were thawed in a 37°C water bath and washed with 50 mL of phosphate-buffered saline (PBS). The viable cells were counted using trypan blue exclusion. The average yields of viable cells were 62% (range of 40% to 90%) from patient samples and 59% (range of 47% to 70%) from normal subjects. With PBMCs, 1 × 10⁶ cells were set aside for stimulation; in patients in whom enough cells were available, the remaining cells were used for isolation of CD14⁺ monocytes/macrophages with magnetic-activated cell sorting (MACS) columns (Miltenyi Biotec Inc., Auburn, CA, USA) in accordance with the instructions of the manufacturer. Briefly, cells were suspended in 50 mL of MACS buffer (PBS, 0.5% bovine serum albumin, and 2 mM EDTA [ethylenediaminetetraacetic acid]) and were labeled with primary antigen-presenting cell (APC)-conjugated CD14 antibody (BD Biosciences, San Jose, CA, USA) and incubated with MACS anti-APC microbeads (Miltenyi Biotec Inc.). The CD14⁺ cells bound with anti-APC microbeads were isolated using a magnetic separation column (Miltenyi Biotec Inc.).

The average yields of CD14⁺ cells were 11% (range of 3% to 30%) from the PBMCs of patients and 20% (range of 4% to 32%) from those of normal controls. These percentages from frozen cells were similar to the proportion of CD14⁺ cells (14% ± 3%) recovered from fresh PBMCs in a previous study [17]. In two control subjects in whom sufficient cells were collected, flow cytometric analysis showed that CD14⁺ cells made up, respectively, 82% and 90% of this cell fraction. Because of limited cell numbers, fluorescence-activated cell sorting (FACS) analysis to determine the proportion of monocytes and macrophages in the CD14⁺ cell fraction was not done, and CD14⁺ cells were not isolated from SFMCs.

Whole PBMCs (2 × 10⁵ cells per well), CD14⁺ cells from PBMCs (4 × 10⁴ cells per well), or SFMCs (2 × 10⁵ cells per well) were suspended in cell culture medium (RPMI, 1% HEPES, 10% human serum, and 1% glutamine) and plated singly in 96-well culture plates (Corning Incorporated, Corning, NY, USA). The number of spirochetes per milliliter of culture was determined using spectrophotometry of spirochetal DNA, as previously described [18]. Live B. burgdorferi was used in all experiments.

To determine optimal culture conditions for the stimulation of chemokine secretion, PBMCs from 6 normal control subjects were stimulated with multiplicities of infection (MOIs) ranging from 6 to 500 spirochetes per host cell with durations of incubation ranging from 4 to 120 hours. Although similar results were seen with MOIs from 6 to 100 for all four chemokines, the highest secretion of CCL4 occurred with an MOI of 50. Secretion of CCL4 increased steadily during 5 days in culture, but secretion of CXCL9 and CXCL10 did not appear until the fifth day. This suggested that stimulated cells were viable throughout the 5 days in culture. On the basis of these preliminary findings, cells were incubated with B. burgdorferi (strain EM70, an OspC type A strain, MOI of 50), IFN-γ (50 ng), or B. burgdorferi and IFN-γ at 37°C in 5% CO₂ for 5 days.

After the collection of supernatants, non-adherent cells were resuspended and collected to determine the expression of CXCR3 or CCR5 and T cells in PBMCs or SFMCs. Harvested cells were washed with FACS buffer (PBS, 1% fetal bovine serum, and 0.1% Na Azide) and stained with the antibodies specific for CD3 (APC; BD Biosciences), CD4 (PerCP; BD Biosciences), CXCR3 (FITC; BD Biosciences), or CCR5 (PE; BD Biosciences). The cell surface expression of these receptors was analyzed by flow cytometry with a FACSCalibur (BD Biosciences).

The 24 patients with Lyme arthritis, who were representative of our previously published cohort of 117 patients with this type of arthritis [19], had a median age of 36 years (range of 12 to 67 years); 17 were male and 7 were female. The diagnosis of Lyme arthritis was usually made within 1 to 4 weeks after the onset of knee swelling. Eleven of the 24 patients (46%) were first seen prior to the initiation of antibiotic therapy, 10 were first evaluated during 1- to 4-month courses of oral or intravenous (i.v.) antibiotic treatment, and 3 were not seen until the post-antibiotic period, 7 to 11 months after the start of antibiotics. At their first visit, 13 of the 24 patients (54%) had a positive polymerase chain reaction (PCR) result for B. burgdorferi DNA in joint fluid. Of the 13 patients, 6 were seen prior to antibiotic treatment and 7 were evaluated during antibiotic therapy. Seven of the 24 patients had the resolution of arthritis within 3 months after the initiation of antibiotics, defined as antibiotic-responsive arthritis, whereas the remaining 17 patients had persistent arthritis for more than 3 months after at least 8 weeks of oral antibiotics or at least 4 weeks of i.v. antibiotics or both, defined as antibiotic-refractory arthritis [19]. This distribution of refractory and responsive cases is reflective of our role as a referral center.

The chemokine data from cell cultures were compared in patients seen before, during, or after antibiotic therapy, in those with positive or negative PCR results for B. burgdorferi DNA in joint fluid, and in those with antibiotic-responsive or antibiotic-refractory arthritis. Although B. burgdorferi-stimulated PBMCs from patients with positive PCR results tended to secrete higher levels of CCL2 and CXCL9 in culture than PBMCs from patients with negative PCR results, none of the differences between the groups was statistically significant. In addition, since patients' cells had been stored for as long as 12 years, the chemokine data were analyzed according to the length of time that the cells had been frozen, but no correlation was found. Because timing of cell collection, PCR results, and duration of freezing did not result in differences in chemokine secretion ex vivo, the data from all 24 patients are presented together here.

In initial experiments, PBMCs were first tested from 4 healthy control subjects from the laboratory. Unstimulated PBMCs from control subjects secreted only basal levels of CCL4, CCL2, CXCL9, or CXCL10 during 5 days in culture (Figure 1a). With B. burgdorferi stimulation, the levels of each of the four chemokines tended to be higher; with IFN-γ stimulation, the levels of CCL2, CXCL9, and CXCL10 tended to be greater, and there was no additive effect from stimulation with B. burgdorferi and IFN-γ together. However, differences in chemokine secretion among stimulated or unstimulated cells were not statistically significant.

As with PBMCs, unstimulated CD14⁺ monocytes/macrophages secreted only basal levels of the four chemokines in culture (Figure 1b). In several cases, B. burgdorferi alone induced CD14⁺ cells to secrete small amounts of CCL4; B. burgdorferi and IFN-γ together stimulated CCL2 secretion, and IFN-γ seemed to induce CXCL9 and CXCL10 secretion. However, as with PBMCs, the differences between stimulated and unstimulated cells were not statistically significant (Figure 1a, b).

The patterns of chemokine secretion from PBMCs or CD14⁺ monocytes/macrophages from the 4 normal control subjects were similar to those from the 24 patients with Lyme arthritis. However, the larger number of patients tested allowed a clearer picture of the differences between B. burgdorferi- or IFN-γ-stimulated cells.

Unstimulated PBMCs from the 24 patients secreted basal levels of CXCL9, CXCL10, CCL2, or CCL4 during 5 days in culture, except in rare cases (Figure 2a). In comparison, B. burgdorferi stimulation of PBMCs resulted in the secretion of CCL4 and CCL2, and B. burgdorferi and IFN-γ each stimulated the production of CXCL9 and CXCL10. These differences among stimulated and unstimulated cells were statistically significant (P < 0.001) (Figure 2a).

In 14 of the 24 patients, large enough numbers of PBMCs were available to separate a fraction containing CD14⁺ monocytes/macrophages. With the CD14⁺ cell fraction, B. burgdorferi alone stimulated the secretion of CCL4; B. burgdorferi and IFN-γ together induced CCL2 secretion, and IFN-γ alone stimulated the secretion of CXCL9 and CXCL10 with no additive effect from B. burgdorferi and IFN-γ together (Figure 2b). These differences among stimulated and unstimulated cells were statistically significant (P < 0.001) (Figure 2b).

Since the types of cells, their numbers, and their activation state would presumably be different in SFMCs than PBMCs, cells from these two sites might respond differently to stimuli. For this analysis, concomitant PBMCs and SFMCs were available in 11 of the 24 patients. Owing to limited numbers of cells, CD14⁺ cells were not separated from SFMCs. Without stimulation, SFMCs or PBMCs secreted basal levels of CXCL9, CXCL10, CCL2, and CCL4, except in one case (Figure 3). B. burgdorferi, but not IFN-γ, induced the secretion of CCL4 and CCL2 from PBMCs and SFMCs, whereas B. burgdorferi or IFN-γ each induced cells from either compartment to secrete CXCL9 and CXCL10. There were no significant differences between PBMCs and SFMCs in the secretion of these chemokines.

CXCR3, the receptor for CXCL9 and CXCL10 on T cells, and CCR5, the receptor for CCL4 on monocytes, dendritic cells, and some T cells, are preferentially expressed in T_H1-like immune responses [7‐9]. In 6 of the 11 study patients in whom concomitant PBMCs and SFMCs were available, the expression of these receptors on T cells was determined by flow cytometry (Figure 4a). In these patients, the percentage of unstimulated T cells expressing CXCR3 or CCR5 was significantly greater in SFMCs than PBMCs (Figure 4b). With PBMCs, B. burgdorferi or IFN-γ appeared to stimulate greater expression of CXCR3 or CCR5 compared with unstimulated cells. With SFMCs, CCR5 expression tended to increase with both stimuli, but the high expression of CXCR3 was rarely modulated further by either stimulus. However, none of the differences between stimulated and unstimulated cells was statistically significant.