Parsing innate immune response to borrelia burgdorferi using a hydrogel-based model of the skin microvasculature

Introduction: Lyme disease is caused by infection of the tick-transmitted bacteria Borrelia, and is associated with an erythema migrans skin rash and symptoms that overlap with the common flu. While Lyme disease can be treated with an aggressive course of antibiotics, chronic Lyme disease and Lyme-associated arthritis are complications that can persist for years, sometimes leading to debilitating neurological symptoms. Single nucleotide polymorphism (SNP) TLR1-1805GG is present in approximately 50% of the population (TLR1-1805TG/GG accounting for the remaining 50%), and is associated with antibiotic refractory Lyme arthritis, Lyme neuroborreliosis, and debilitating symptoms. While the skin microvasculature is the initial site of infection, models do not exist that enable the study of borrelia interactions with the dermal microenvironment, which is rich with microvasculature and macrophages. Macrophages have been shown to increase ~2 fold in the erythma migrans rash compared to healthy skin and are among the first cells to encounter borrelia. Moreover, murine models deficient of adaptive immune systems retain the ability to acquire severe Lyme arthritis, suggesting a key role of innate immune dysfunction in disease pathogenesis. Elucidation of mechanisms of early host response to borrelia infection are necessary to parse immune interactions between patients to both advance treatment options beyond broad spectrum antibiotics and avoid chronic complications.

Materials and

Methods: We used a microfluidic platform to produce formation of a dermal microvasculature microenvironment mimic from primary human cells. Human umbilical vein endothelial cells, dermal fibroblasts, macrophages, and Borrelia burgdorferi B31 (“ B31”) were co-cultured in our synthetic extracellular matrix (ECM) to visualize cell-cell interactions and elucidate crosstalk between Bb and host cells, which formed microvascular networks. Our synthetic ECM comprised 8-arm poly(ethylene glycol) (PEG) crosslinked with matrix metalloproteinase-degradable peptides. Adhesive peptides were covalently linked to PEG prior to crosslinking to mimic some features of native ECM proteins and to sequester cell-produced ECM. Cells along with Bb were mixed within the hydrogel precursor on day 0 and injected into the central gel channel of the microfluidic device (Figure 1a). Microvascular networks emerged over 6 days and their geometric features were quantified via confocal microscopy and subsequent MATLAB analyses. Media was changed daily and frozen for ELISA analysis of IL6 and CCL2. Macrophages were derived from buffy coats of healthy donors using CD14+ positive isolation and subsequent differentiation with macrophage colony stimulating factor (M-CSF) over 6 days. To parse the role of TLR1-1805GG in model phenotypes, CD14+ cells were genotyped prior to differentiation by amplifying the TLR1 fragment and subsequently digesting with restriction enzyme PstI before running products on a 1.5% agarose gel. Donors with TLR1-18055GG would produce 2 bands (~850 bp and ~400 bp), TLR1-1805TG would produce 3 bands (~1250 bp, ~850 bp, and ~400 bp), and TLR1-1805TT would produce 1 band (~1250 bp, i.e., no digestion with PstI).

Results, Conclusions, and Discussions: Co-cultures with B31 enhanced mean microvascular network diameter compared to networks without B31, an effect which was consistent across 5 macrophage donors (Figure 1b,c); however, macrophage donor variability was evident in baseline (ie., B31-) network formation. This diameter increase with B31 was not observed in the absence of macrophages (i.e., with only endothelial cells and dermal fibroblasts). Moreover, B31 presence induced inflammatory cytokine secretion 24 h after co-culture initiation, mirroring clinical observations of acute and chronic Lyme disease.
Six donors genotyped for TLR1 SNP revealed one TLR1-1805TT, two TLR1-1805TG, and three TLR1-1805GG. These frequencies mirror literature citing the presence of TLR1-1805GG in ~50% of the population. Microvasculature cultures with and without B31 were prepared, and mean network diameter (day 6), IL-6 secretion (day 1), and CCL2 (day 1) were compared. While TLR1 SNP did not dictate mean network diameter or CCL2 secretion (Figure 1d), TLR1-1805GG enhanced IL-6 secretion compared to TLR1-1805TT/TG (Figure 1e, significant difference in slopes p=0.0001). These data recapitulate serum sample data from patients with erythema migrans stratified by TLR1-1805GG polymorphism and borrelia genotype. Patients with TLR1-1805GG and RST1 borrelia (i.e., B31) had elevated serum IL6, whereas CCL2 was indistinguishable.
Together, these studies are important findings toward validating our dermal microphysiological borrelia infection model, toward our goals of: 1) performing mechanistic studies of B31infection, and 2) investigating novel treatments for Lyme disease. We were able to stratify clinically relevant responses in our model based on TLR1 polymorphisms, and future work will validate these findings using proteomics analyses. Furthermore, we will use our model to stratify response to antibiotics and novel therapeutics to treat acute and chronic Lyme disease.

Acknowledgements (Optional): We thank the Fairbairn Fund, NIH, NSERC (PDF awarded to LCB), and the donors at Massachusetts General Hospital for making this work possible.