In 1975, a cluster of children and adults in the small community of Lyme, Connecticut experienced unusual arthritic symptoms. Today, Lyme disease (LD) is the most prevalent vector-borne disease in the US, with an estimated 476,000 annual cases (1). The causative agent, Borrelia burgdorferi sensu lato (Bb), is a spirochete transmitted from the Ixodes tick into the human host dermis during its blood meal (2). The spirochete leaves the blood and disseminates into multiple organ systems in as little as two weeks post-infection (3–5). Disseminated LD is more challenging to diagnose, and delayed treatment can lead to long-term disability or death (6, 7). The disease is endemic in the Northeastern US, and incidence rates continue to rise. The US’s annual treatment and diagnostic cost is over 4.8 billion USD (8).

Antibiotic treatment for LD in the acute phase is often curative (9–12). However, untreated patients and a subset of treated patients progress to disseminated disease (13, 14), Disseminated disease can result in facial nerve palsy (15), Lyme Carditis (16), Lyme Arthritis (17), Lyme Neuroborreliosis (18, 19), and long-term disability (20, 21). Persistent symptoms are reported by 10-20% of patients diagnosed and treated during the acute phase of LD. Persistent symptoms include joint pain, fatigue, and neurocognitive deficits (10, 22). This highlights the need for an accurate early diagnosis and the ability to track disease resolution.

Current diagnosis is complicated because testing relies on indirect methods. Direct PCR and blood culture methods often fail due to the spirochete’s limited presence in the bloodstream, low bacterial counts in circulation, slow replication cycle, the requirement for complex growth media, and specialized microscopy requirements (23, 24). Hence, indirect methods that rely on the patient’s serological response are the principal method of confirming an infection (25). These indirect methods of acute LD diagnostics based on ELISA and western immunoblot technologies suffer from low sensitivity and a high false-positive rate. Thus, while advances in LD detection research are being made (26–29), clinicians currently lack a sensitive method to diagnose early disease. Furthermore, clinical assays are unable to determine treatment efficacy, track disease resolution (30) or diagnose subsequent infections.

Serum protein glycosylation is often altered during inflammatory and autoimmune diseases. The glycosylation profile of immunoglobulins is dynamic and offers a novel immunologic insight into the host’s response (31–34). Glycosylation is the most abundant complex post-translational serum protein modification (35) and plays a significant role in protein structure and function in vivo (36, 37). IgG has a well-characterized N-glycosylation site on the constant fragment (Fc- Asn-297) region (38). This site contains complex biantennary glycans with varying degrees of galactose, bisecting N-acetyl-glucosamine, and sialic acid residues. Most notably, the glycans are highly core-fucosylated. IgG glycosylation is dynamic and the glycans present can affect the binding avidity to various Fcγ receptors, rate of complement activation, and release of cytokines (39–42). Previous studies have profiled IgG N-glycans in sera obtained from patients with inflammatory disease ( Table 1 ). Reports indicate a trend for IgG N-glycan reduction in terminal galactose and sialic acid content during inflammatory diseases and this glycan signature is linked to a pro-inflammatory phenotype (52–58).

Scientists are beginning to understand how glycosylation reflects health status and influences protein structure and function. IgG glycosylation moieties are associated with specific functions. In general, lowered galactose and sialic acid residue content are reported in inflammatory states ( Table 1 ), and conversely, increased galactose is associated with an anti-inflammatory state. Hence, to improve immunotherapy outcomes, pharmaceutical companies glycoengineer monoclonal antibodies with specific glycosylation features to treat cancers and chronic diseases. For example, the effectiveness of intravenous immunoglobulin (IVIG) therapy for autoimmune conditions including Gillian-Barre, Immune Thrombocytopenic Purpura, or Kawasaki disease patients is associated with the increased sialic acid content on the Fc N-glycans of the IVIG (59–61). In addition, therapeutic monoclonal antibodies are glycoengineered to contain specific a-fucosylated, agalactosylated N-glycans to promote superior half-life and treatment efficacy (62–65). IgG N-glycan modulation of the immune response during disease reveals another layer of physiologic crosstalk. Evidence indicates that the repertoire of N-glycans present on IgG produced in response to vaccines is dependent on many factors, including age, inflammatory state, and the type of adjuvant (58).

It is well accepted that N-glycans modulate the function of antibodies and are altered in disease states (66). We hypothesize that the IgG N-glycan profile of LD patients will reflect the immunological response to the acute LD infection. Thus, this first report of total IgG N-glycans associated with LD will provide insight into the inability of the host immune system to resolve Bb bacterial infection.

We demonstrate that total serum IgG N-glycosylation of acutely infected Lyme disease patients contrasts with the typical pro-inflammatory signature often found on other inflammatory diseases. First, a proof-of-concept study was performed using pooled serum to identify the IgG N-glycan signature of healthy control, acute LD, and patient-matched antibiotic-treated LD serum using high-pressure liquid chromatography separation (HPLC) and detection of fluorescently labeled glycans (75). The samples were subsequently analyzed using the recently developed GlycoTyper MALDI method (73) which pairs a specific total-IgG capture antibody with subsequent MALDI-FT-ICR imaging (MALDI). The trends in the glycan signatures were reproducible between both platforms, and thus we proceeded to analyze a larger confirmatory set of individual serum samples using the GlycoTyper platform.

This is the first report of total human IgG N-glycan profiles during early and treated LD. Our initial studies detected an unanticipated shift in IgG N-glycan profile which was confirmed and expanded upon using a larger set of samples. During many inflammatory states ( Table 1 ), IgG N-glycan profiles respond with reduced terminal sialic acids and galactose and increased GlcNAc sugar exposure (39). This agalactosylated IgG phenotype has been demonstrated to promote pro-inflammatory responses and may aid in clearing the infection (52–58, 77, 78). In the case of LD, total IgG N-glycan profiles had the opposite trend. LD increases the galactose and sialic acid content of IgG N-glycans while further decreasing the terminal GlcNAc exposure ( Figures 1 – 4 ). The result of such a shift requires further investigation, but in theory could contribute to ineffective host responses to LD infection by reducing humoral immune activation (77, 78). The ability to detect these changes within total IgG suggests that LD induces a large shift in the IgG N-glycan composition. Thus, this dysregulated immune response may at least in part explain why human IgG produced during an initial LD infection is not effective to clear the current infection or protect from a future re-infection with LD.

The pilot study served to compare the well-established HPLC IgG N-glycan analysis method (79) to the novel MALDI analysis method (73) using pooled cohorts of control, acute LD, and treated LD serum. Due to MALDI’s limited detection of unmodified sialic acids, samples were treated with a sialidase (72). The results presented in Figure 2 demonstrate that both methods detect comparable individual identified N-glycans with similar trends of terminal galactose exposure induced during LD and maintained during the post-antibiotic treatment time points. To confirm the pooled pilot study findings, three cohorts of 18 individual samples were analyzed in triplicate using the MALDI method. The GlycoTyper method was selected to analyze the confirmatory cohort due to its significantly higher throughput compared to HPLC. We found that the demographics of the age- and sex-matched patient cohorts did not lead to inherent differences in IgG N-glycosylation.

While the MALDI results confirmed the original trends observed in the pooled pilot study, a higher number of samples analyzed led to further trends emerging ( Figures 3A, B ). Agalactosylated structures are reduced while total N-glycans terminating with a galactose moiety continued to increase within the acute and treated cohorts as indicated within the pooled pilot experiments. N-glycans containing a core-fucose are markedly lowered in the antibiotic-treated cohort which suggests an increase in Antibody-Dependent Cellular Cytotoxicity (ADCC) abilities of post-treatment IgG (80). The increased level of digalactosylated (G2) N-glycans on IgG at the acute stage and subsequent return to healthy control levels is a potential biomarker reflecting the host’s response to successful antibiotic treatment. Following antibiotic therapy for Tuberculosis (TB), IgG N-glycan digalactose content returned to healthy control levels (43). Grace et al. found G2 N-glycans decreased during acute TB and subsequently increased after effective antibiotic treatment. In the case of LD, IgG N-glycans increase G2 content during infection and return to healthy control levels after effective antibiotic treatment ( Figure 3B ). This once more converse trend observed in LD suggests a subverted immune response during LD compared to TB infection. The implication of the IgG N-glycan response to LD is portrayed in Figure 5 .

LD patients are discriminated from healthy controls with a high degree of sensitivity (75-100%) and specificity (94.7-100%) using total IgG N-glycan measurements ( Figure 4 ). Blinded analysis of acute LD, post-treatment LD, and mimic diseases using these methods should be completed to validate these findings. Future LD tests incorporating total IgG N-glycan analysis could increase acute LD diagnostic sensitivity and track subsequent antibiotic treatment responses.

Mechanisms controlling the dynamic B-cell glycosyltransferase expression operating on IgG N-glycans during LD require examination. B-cell glycosyltransferase expression is known to respond to the cytokine micro-environment (81). Recent studies detected IgG N-glycosylation profiles are impacted by the type of adjuvant present during vaccination (82). Additionally, during the onset of autoimmune disease, T_H17 cells signal newly differentiated B cells with IL-22 and IL-23 to regulate glycosyltransferase expression, resulting in a pro-inflammatory agalactosylated, non-fucosylated IgG N-glycan (83). The increase in terminal galactose of IgG N-glycans during acute LD may be attributed to an upregulation of naïve B-cell beta-1,4-galactosyltransferase expression (66, 84–86). Future studies should determine if LD induces a specific cytokine signaling pathway to affect the glycosyltransferase expression of plasma cells or B cells during LD infection.

Lyme disease has been demonstrated to destroy germinal centers of lymph nodes during early infection (77, 78, 87, 88). These germinal centers are a vital structure that produces long-lived immunoglobulin responses through T-cell-dependent interactions (89). The destruction of the germinal center may explain why patients are liable to become re-infected with LD after treatment. Additionally, Bb has been demonstrated to gain entry inside human endothelial cells (90), alter innate immune responses after phagocytosis (91), suppress lymphocytes growth rate (92), and down-regulate major-histocompatibility complexes expression on Langerhans cells (93). Any one of these effects could be responsible for altering the IgG N-glycan profile during LD. For example, antigens presented in a T-cell independent manner promote increased sialyation on the Fc portion of IgG leading to an immunotolerant, less pro-inflammatory IgG N-glycan repertoire (94). Interestingly, murine studies have demonstrated a reduction in the general humoral response after LD infection as indicated by low titers of anti-viral antibodies produced post-vaccination and Bb-impaired helper T-cell mediated affinity maturation (87, 95). Lastly, because the LD cohort completed 2-3 week of oral doxycycline therapy, the serum N-glycome or immune response could have been altered in part due to changes in gut flora (96, 97).

Using the GlycoTyper MALDI-FT-ICR imaging approach, we detected an unexpected IgG N-glycan signature in humans during LD. We have demonstrated the IgG N-glycans produced during a Lyme disease infection lack the classic highly agalactosylated signature associated with most inflammatory diseases. Instead, LD induces IgG N-glycans containing larger, terminally galactosylated sugar moieties. Moreover, many IgG N-glycans detected at the acute LD timepoint remain altered from healthy control levels at the post-antibiotic treatment timepoint. Of note, the digalactose content of IgG N-glycans in the treated time point was comparable to the healthy baseline levels after antibiotic treatment. Furthermore, we detected a significant decrease in total core-fucosylation at the treated timepoint suggesting a possible increase in ADCC which may aid in clearing Bb from the host. IgG N-glycans offer numerous biomarkers, reflect acute disease state, response to treatment, and may improve the sensitivity of the acute diagnosis of LD above the current two-tiered testing protocol. This first examination of IgG N-glycan signatures associated with LD requires future study.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants was reviewed and approved by the University of Drexel College of Medicine Review Board. This study involves secondary research using de-identified data and biospecimens not collected specifically for this study. The IRB 1808006553 was assigned an exempt status.

BH-G and MC were involved in conceptualization of assay, assay design, data analysis, and manuscript writing. BH-G was responsible for statistical analysis. AM and SG were involved in assay design and data analysis. BW was involved in editing the manuscript. All authors contributed to the article and approved the submitted version.

This research was funded through the Mary K. Dewitt Pettit MD, Fellowship Fund (Award number: 282975) and through the Institute for Molecular Medicine and Infectious Disease at Drexel University College of Medicine.

We thank Mengjun Wang and Dr. Laura Steel for their support. We also thank Dr. Elizabeth Horn for her partnership through the Bay Area Lyme Disease Biobank.

Author AM is a founding partner in GlycoPath, Inc. and holds patents for the GlycoTyper methodology.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.