Antiphospholipid autoantibodies in Lyme disease arise after scavenging of host phospholipids by Borrelia burgdorferi

Abstract

A close association with its vertebrate and tick hosts allows Borrelia burgdorferi, the bacterium responsible for Lyme disease, to eliminate many metabolic pathways and instead scavenge key nutrients from the host. A lipid-defined culture medium was developed to demonstrate that exogenous lipids are an essential nutrient of B. burgdorferi, which can accumulate intact phospholipids from its environment to support growth. Antibody responses to host phospholipids were studied in mice and humans using an antiphospholipid ELISA. Several of these environmentally acquired phospholipids including phosphatidylserine and phosphatidic acid, as well as borrelial phosphatidylcholine, are the targets of antibodies that arose early in infection in the mouse model. Patients with acute infections demonstrated antibody responses to the same lipids. The elevation of antiphospholipid antibodies predicted early infection with better sensitivity than did the standardized 2-tier tests currently used in diagnosis. Sera obtained from patients with Lyme disease before and after antibiotic therapy showed declining antiphospholipid titers after treatment. Further study will be required to determine whether these antibodies have utility in early diagnosis of Lyme disease, tracking of the response to therapy, and diagnosis of reinfection, areas in which current standardized tests are inadequate.

Trial registration: ClinicalTrials.gov NCT00028080.

Keywords: Diagnostics; Immunoglobulins; Infectious disease; Microbiology; Molecular biology.

Conflict of interest statement

Conflict of interest: PJG and LTH have filed a provisional patent titled Antiphospholipid Antibodies for the Diagnosis of Lyme Disease. ARM has a patent (US 8,926,989) for an unrelated Lyme disease diagnostic.

Figures

Figure 1. Growth of B. burgdorferi is dependent upon exogenous lipids.

(A) Growth of B. burgdorferi in lipid-free medium supplemented with fatty acids and cholesterol (2:2:1 palmitic acid/oleic acid/cholesterol) to a final concentration of 5–500 μM. At 500 μM, cell density equivalent to that of the unmodified BSK medium was reached. No growth was observed in the absence of lipid (dBSK only). (B) Equivalent growth over 10 days of B. burgdorferi in medium supplemented with fatty acids (FA) (200 μM each of palmitic and oleic acids) and phospholipids (PL) (100 μM each of PG and PC). Cholesterol (Ch) was present at 100 μM in all media. Data plotted in A and B show the mean of 3 biological replicates, with error bars indicating the SD.

Figure 2. Uptake of fluorescence-labeled phospholipids by B. burgdorferi.

The canonical phospholipids of B. burgdorferi PC and PG, as well as the noncanonical membrane components PA, PE, and PS, were acquired from the medium. Untreated cells (UT) did not fluoresce in the NBD channel. Cells were incubated in dBSK plus 25 μM NBD-labeled phospholipid analogs over a 6-hour period, and then imaged by fluorescence microscopy at ×63 magnification. DAPI stained the nucleic acids nonspecifically. DAPI (Ex = 405 nm, detector = 410–466 nm) and NBD (Ex = 470 nm, detector 550–600 nm) channels were acquired sequentially. Fluorescence intensity (Ex = 465 nm, Em = 540 nm, normalized to OD600) was quantified by fluorimetry. Graphs plot the mean of 3 biological replicates, with error bars indicating the SD. Images are representative of biological triplicate experiments.

Figure 3. The membrane of B. burgdorferi is more accessible than that of other species.

B. burgdorferi (A), S. aureus (B), and E. coli (C) were all incubated with 25 μM NBD-PC for 4 hours. Only B. burgdorferi acquired the fluorescent phospholipid. DAPI stained nucleic acids nonspecifically. DAPI (Ex = 405 nm, detector = 410–466 nm) and NBD (Ex = 470 nm, detector 550–600 nm) channels were acquired sequentially at a magnification of ×63. Images are representative of biological triplicate experiments.

Figure 4. The borrelial membrane reflects the surrounding medium.

B. burgdorferi was incubated with 2 phospholipid fluorophores — NBD-PC and Texas Red–PE (TXR-PE) — at different ratios for 4 hours. The ratio of fluorescence intensity in washed cells (F540/F625) closely matched the ratio of fluorophores available in the culture medium (μM NBD-PC vs. μM TXR-PE). The total concentration of the phospholipid was 25 μM for all conditions. Bars represent the mean of 3 biological replicates, with error bars indicating the SD.

Figure 5. Phospholipids are retained in the membrane in the absence of an environmental source.

B. burgdorferi was labeled with NBD-PC in dBSK medium for 4 hours, washed, and resuspended in rich medium. Fluorescence intensity (F540/OD600) was measured daily and declined with a half-life of approximately 24 hours, with 56% of the day 0 fluorescence intensity on day 1 and approximately 7% on day 4. Data points represent the mean of 3 biological replicates, with error bars indicating the SD.

Figure 6. Anti-lipid antibodies are raised in a mouse model of B. burgdorferi infection.

(A) Endpoint titers were determined for 7 antibodies (anti-PA [αPA], αPC, αPE, αPG, αPS, αCL, and anti-galactosylcholesterol [αgC]) in naive and infected 4-week-old C3H mice. Data (black lines indicate the mean of 6 replicates, and error bars indicate the SD) are plotted on a log2 scale to reflect binary dilution series. *P < 0.05 and **P < 0.01, by unpaired, 2-tailed t test. (B) In a different group of 6 mice, 3 antibodies were measured over the course of a 28-day (d) infection. Data are plotted as in A.

Figure 7. Antiphospholipid antibodies in human sera during infection.

IgG anti-PA, anti-PC, and anti-PS were raised during infection, while anti-CL was not. Untreated and post-treatment (PT) groups were compared with the naive controls (N). Each group contained 12 individuals, with black lines representing the mean of the group and error bars the SD. All samples were normalized to the average of at least 6 naive controls run in parallel. An index of greater than 1 indicates that the antibody titer was above the level of that in the naive controls. Gray dashed lines represent the cutoff value (the mean of the naive group + 2.291 SDs) above which a sample was considered positive. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001), by Tukey’s test for multiple comparisons.

Figure 8. Independence of the 3 antibody responses.

(A) Heatmap showing the rank score for each elevated antibody in each serum sample. (B–D) Linear regression analyses for pairwise comparisons of the 3 antibody responses. Only anti-PC versus anti-PS was significant, as determined by unpaired, 2-tailed t test (P < 0.05). Trend lines are shown in black and 95% CIs in gray dashed lines.

Figure 9. Antibody titers decline after treatment.

A group of 10 individuals were sampled serially (2, 3, or 4 times) at time points 0–365 days from the beginning of treatment. Antibody titers (anti-PA, anti-PC, and anti-PS) for each patient are shown as percentages of the highest value for each patient/antibody. In A, samples are grouped by time point, with n = 5 for day 0, n = 4 for days 1–30, n = 6 for days 31–99, n = 5 for days 100–199, and n = 7 for days 200–365. Error bars represent the SD. In B, the samples were plotted individually against the day of collection, with trend lines shown in black and 95% CIs in gray dashed lines.