Interaction of amphiphilic lipoarabinomannan with host carrier lipoproteins in tuberculosis patients: Implications for blood-based diagnostics

Lipoarabinomannan (LAM), an amphiphilic lipoglycan of the Mycobacterium tuberculosis cell wall, is a diagnostic target for tuberculosis. Previous work from our laboratory and others suggests that LAM is associated with host serum lipoproteins, which may in turn have implications for diagnostic assays. Our team has developed two serum assays for amphiphile detection: lipoprotein capture and membrane insertion. The lipoprotein capture assay relies on capture of the host lipoproteins, exploiting the biological association of host lipoprotein with microbial amphiphilic biomarkers to “concentrate” LAM. In contrast, the membrane insertion assay is independent of the association between pathogen amphiphiles and host lipoprotein association, and directly captures LAM based on its thermodynamic propensity for association with a supported lipid membrane, which forms the functional surface of an optical biosensor. In this manuscript, we explored the use of these assays for the detection of LAM in sera from adults whose tuberculosis status had been well-characterized using conventional microbiological tests, and endemic controls. Using the lipoprotein capture assay, LAM signal/noise ratios were >1.0 in 29/35 (83%) individuals with culture-confirmed active tuberculosis, 8/13 (62%) individuals with tuberculosis symptoms, but no positive culture for M. tuberculosis, and 0/6 (0%) symptom-free endemic controls. To evaluate serum LAM levels without bias associated with potential differences in circulating host lipoprotein concentrations between individuals, we subsequently processed available samples to liberate LAM from associated host lipoprotein assemblies followed by direct detection of the pathogen biomarker using the membrane insertion approach. Using the membrane insertion assay, signal/noise for detection of serum LAM was greater than that observed using the lipoprotein capture method for culture-confirmed TB patients (6/6), yet remained negative for controls (2/2). Taken together, these results suggest that detection of serum LAM is a promising TB diagnostic approach, but that further work is required to optimize assay performance and to decipher the implications of LAM/host lipoprotein associations for diagnostic assay performance and TB pathogenesis.

A typical ELISA experiment for the detection of LAM at various concentrations using FIND clones 24, 29 and 31 as the capture, and clone 171 as the reporter is shown in Fig S1. The limit of detection (LoD) for LAM using this assay was found to be 62.5 nM (assuming a molecular weight of 19 KDa for LAM) (n=3, per experimental condition) using clone 31 as the capture and clone 171 as the reporter. LoD was defined as 3 (s/n) where s/n is the standard deviation in observed measurements, and summary of s/n for the measurements is shown in Fig S1. Statistical significance was determined using one-way ANOVA with Fisher's least significant different test used for post-hoc analysis (**P<0.05).
Antibodies were assessed for suitability for use in the lipoprotein capture assay based on three parameters (Table S1): a) stability (>6 weeks after labeling with either horse radish peroxidase (HRP) or Alexa fluor 647, under refrigeration), b) specificity (no cross reactivity with M. smegmatis LAM), and c) sensitivity (use of <30 nM of antibody to obtain a limit of detection of  62.5 nM of LAM, which is the maximal sensitivity obtainable for this antigen by ELISA in our hands). Antibodies that did not satisfy all three criteria were not used in the waveguide-assays. For instance, the best sensitivity for detection of LAM using this method incorporated clone 31 as capture, and clone 171 as the reporter (Fig S1).
However, clone 31 was unstable upon HRP labeling, and demonstrated cross reactivity with M. smegmatis LAM, and therefore, was not selected for further development. Using a similar assessment metric (Fig S1), FIND clones 171 and 24 were chosen for use in the lipoprotein capture assay. S1

Characterization, selection, labeling and preparation of antibodies
Five different FIND monoclonal antibodies (clones 24, 27, 29, 31 and 171) were evaluated for sensitivity and specificity of detection of LAM by Enzyme Linked Immunosorbent Assays (ELISA), and optimal antibodies were selected for further use.
For ELISA screening, a sandwich assay protocol was used. The conditions described for this screening assay were shown to provide good sensitivity for detection from early screening measurements at various antibody concentrations using the BEI antibodies, spectroscopy. Three different batch aliquots were prepared and characterized. The degree of labeling for FIND 171 was 6, 7.6, 5.2 for each of these batches, and for FIND 24 was 7.5, 9.1 and 5.03. Aliquots of each antibody were prepared and stored at 4 °C.
They were combined to make a cocktail of 15nM total antibody concentration before each experiment. Lipoprotein capture assay is a sandwich immunoassay, where the capture antibody targets the coat-protein of HDL nanodiscs, namely Apolipoprotein A1.
This capture antibody, goat polyclonal anti-ApoA1 (1mg/ml), was purchased in biotinylated form and activity was measured by ELISA, and subsequently evaluated by immunoblot at periodic intervals (once every month). Aliquots of the capture antibody were stored at -20 °C, 100nM of the antibody being used as effective concentration in each experiment. . represents missing data, Urine LAM result-positive=1, negative=0