Bockenstedt's group used the mouse model of Lyme disease for the study. To ensure that the tissues harbored enough B. burgdorferi spirochetes to be visible by intravital microscopy, the mice were genetically deficient in the intracellular signaling protein MyD88. MyD88 links the recognition of microbial parts by most Toll-like receptors to activation of certain nuclear genes whose products are involved in the inflammatory process. Mice lacking MyD88 are unable to control the proliferation of a number of bacterial pathogens, including B. burgdorferi. The load of B. burgdorferi in tissues is about 100-fold higher in MyD88-deficient mice than in mice with a complete immune system.
The spirochetes were genetically altered to express green fluorescent protein (GFP). The GFP+ B. burgdorferi was introduced into the MyD88-deficient mice by tick inoculation. 21 days later some of the mice were treated for one month with doxycycline, one of the antibiotics used to treat Lyme disease in humans.
The researchers next peered into the thin layer of skin covering the ear by intravital microscopy. In the mice that were left untreated, they saw lots of spirochetes scurrying about in the dermis.
At the deepest depths of the dermis, they noticed immobile specks and patches of green material deposited near the cartilage. They also saw the deposits in the doxycycline-treated mice. The material was detected by immunofluorescence of ear sections with antibody against B. burgdorferi up to 10 weeks after antibiotic treatment was completed, indicating that the immune system was unable to clear the deposits.
There was no evidence that any spirochetes survived antibiotic treatment. The researchers did not see any motile spirochetes in the skin by intravital microscopy. In addition, tissues were culture negative, ticks that fed on the treated mice were culture negative (xenodiagnosis), and transplantation of skin from the treated mice failed to transmit the infection to recipient mice. Based on these results, the authors concluded that the deposits were remnants of dead spirochetes. As expected, untreated mice tested positive by these assays.
Since chronic infection can lead to Lyme arthritis, the investigators also examined the joints. In another set of mice, the infection was allowed to proceed for four months. The mice were then treated with the antibiotic ceftriaxone for 18 days. When the researchers looked in the joints by intravital microscopy, they again saw the green material (see figure below).
A critical issue to address is whether the amorphous material left behind following antibiotic treatment inflames the joints. The authors could not answer this question directly because of the limitations of the mouse model. Histopathology is unlikely reveal joint inflammation, even in the untreated animals, because laboratory mice do not reliably exhibit joint inflammation so late (4-5 months) during B. burgdorferi infection. Instead, the authors conducted a test tube experiment to see whether the deposits had inflammatory potential. They ground up joint tissue from antibiotic-treated mice in buffer and applied the homogenate to cultured mouse macrophages. The macrophages responded by producing TNF, a key cytokine that promotes inflammation. The more tissue that was added, the more TNF that was produced by the macrophages. In contrast, joint tissue from uninfected mice did not promote TNF production by the macrophages. Therefore, the deposits had the potential to spark inflammation, even after motile spirochetes were eliminated by antibiotics. The debris would continue to inflame the tissues even after antibiotics killed all live spirochetes, explaining why symptoms persist in ~10% of Lyme arthritis cases even after antibiotic treament.
The relevance of the deposits to Lyme disease in humans could be questioned because the MyD88-deficient mice did not have a complete immune system. The authors addressed this concern in the Discussion by mentioning a recent study that described a TLR1 variant linked to severe inflammation and treatment failure in Lyme arthritis patients. Although the gene encoding MyD88 has never been examined in Lyme disease patients, it is conceivable that the TLR1 variant or different forms of other immune genes lead to deposits of Borrelia antigen in the joint and other host tissues.
The authors also addressed the possibility that the deposits are really biofilms, which generally resist killing by antibiotics. Biofilms are believed to be populated by persister cells, which are in a nondividing state that allows bacteria to tolerate antibiotics. According to the authors, if the deposits had harbored persister cells, those cells should have resumed growing when conditions became favorable for growth again. Because the skin and joints from the treated mice were culture negative and because the skin also tested negative by xenodiagnosis and transplantation assays, the authors quickly dismissed the biofilm hypothesis.
Stricly speaking, the authors are correct. Persister cells should start multiplying again in fresh culture medium. However, it's hard to dismiss the biofilm hypothesis completely given the known examples of culture-negative chronic infections associated with biofilms (see this review for one example). Electron microscopy of the joint tissue could reveal whether these deposits are intact spirochetes or debris.
Regardless of their exact nature, deposits of antigen have never been detected within the joints of Lyme arthritis patients. Allen Steere's group failed to find such deposits in pieces of synovial membrane removed from 26 patients with antibiotic-refractory Lyme arthritis. The findings of Bockenstedt and colleagues, who detected the deposits in a location outside of the synovial membrane, suggest that Steere's group was looking in the wrong place.
Bockenstedt, L., Gonzalez, D., Haberman, A., & Belperron, A. (2012). Spirochete antigens persist near cartilage after murine Lyme borreliosis therapy Journal of Clinical Investigation, 122 (7), 2652-2660 DOI: 10.1172/JCI58813
Bolz DD, Sundsbak RS, Ma Y, Akira S, Kirschning CJ, Zachary JF, Weis JH, and Weis JJ (August 1, 2004). MyD88 plays a unique role in host defense but not arthritis development in Lyme disease. The Journal of Immunology 173(3):2003-2010. Link
Strle K, Shin JJ, Glickstein LJ, and Steere AC (May 2012). Association of a Toll-like Receptor 1 polymorphism with heightened Th1 inflammatory responses and antibiotic-refractory Lyme arthritis. Arthritis and Rheumatism 64(5):1497-1507. DOI: 10.1002/art.34383
Bakaletz LO (October 2007). Bacterial biofilms in otitis media, evidence and relevance. The Pediatric Infectious Disease Journal 26(10):S17-S19. Link
Carlson D, Hernandez J, Bloom BJ, Coburn J, Aversa JM, Steere AC (December 1999). Lack of Borrelia burgdorferi DNA in synovial samples from patients with antibiotic treatment-resistant Lyme arthritis. Arthritis and Rheumatism 42(12):2705-2709. DOI: 10.1002/1529-0131(199912)42:12<2705::aid-anr29>3.0.CO;2-H2705::aid-anr29>