Urine lipoarabinomannan glycan in HIV-negative patients with pulmonary tuberculosis correlates with disease severity

An accurate urine test for pulmonary tuberculosis (TB), affecting 9.6 million patients worldwide, is critically needed for surveillance and treatment management. Past attempts failed to reliably detect the mycobacterial glycan antigen lipoarabinomannan (LAM), a marker of active TB, in HIV-negative, pulmonary TB–infected patients’ urine (85% of 9.6 million patients). We apply a copper complex dye within a hydrogel nanocage that captures LAM with very high affinity, displacing interfering urine proteins. The technology was applied to study pretreatment urine from 48 Peruvian patients, all negative for HIV, with microbiologically confirmed active pulmonary TB. LAM was quantitatively measured in the urine with a sensitivity of >95%and a specificity of >80% (n = 101) in a concentration range of 14 to 2000 picograms per milliliter, as compared to non-TB, healthy and diseased, age-matched controls (evaluated by receiver operating characteristic analysis; area under the curve, 0.95; 95% confidence interval, 0.9005 to 0.9957). Urinary LAM was elevated in patients with a higher mycobacterial burden (n = 42), a higher proportion of weight loss (n = 37), or cough (n = 50). The technology can be configured in a variety of formats to detect a panel of previously undetectable very-low-abundance TB urinary analytes. Eight of nine patients who were smear-negative and culture-positive for TB tested positive for urinary LAM. This technology has broad implications for pulmonary TB screening, transmission control, and treatment management for HIV-negative patients.


This PDF file includes:
Materials and Methods Fig. S1. The Kd affinity between RB221 and LAM exceeds that of FB28. Fig. S2. Copper dyes outperform copper free dyes such as fast blue B and safranin O. Fig. S3. Nanocages dissociate biomarker from interfering substances, in silico mathematical modeling. Fig. S4. CS-35 mAb is specific for LAM diluted in human urine, batch verification. Fig. S5. Competition assay confirmed the specificity of CS-35 mAb. Fig. S6. Coupling chemistry to covalently incorporate the FB28 dye in the inner volume of the nanocages. Fig. S7. LAM binding to RB221 and depletion from supernatant are independent of pH in a 5 to 7 range. Fig. S8. RB221 binding to LAM is hindered by the presence of a copper-chelating agent (EDTA). Fig. S9. RB221-LAM interaction requires intact diol moieties of LAM as proven by NaIO4 oxidation. Fig. S10. Carbohydrate concentration in the LAM reference standard (0.160 mg/ml) was quantified by a linear colorimetric assay. Fig. S11. Plot of the 95% CI of the sensitivity and specificity of the ROC analysis reported in Fig. 3C. www.sciencetranslationalmedicine.org/cgi/content/full/9/420/eaal2807/DC1 Fig. S12. The RB221 dye is immobilized in the inner volume of the cages and is available for high-molecular weight ligand binding after cross-link degradation and consequent increase of the effective pore size. Fig. S13. CS-35 anti-LAM mAb does not cross-react with purified polysaccharides from N. meningitidis and S. pneumoniae. Fig. S14. Nanocage capturing followed by CS-35 antibody detection is specific for LAM and does not cross-react with M. tuberculosis lipomannan and arabinogalactan. Table S1. Nanocage bait chemistries screened to capture and enrich LAM from human urine. Table S2. Medical characteristics of diseased TB-negative controls. Table S3. Urinalysis results for all study participants.

Theoretical justification for affinity capture and sensitivity
The signal response of an immunoassay can be expressed via Four Parameter Logistic Model.
According to the law of mass action, when the binding reaction between an affinity probe A and the antigen C + = reaches the equilibrium, the ratio between the concentration of product AC and the reactants C and A is constant and can be expressed as Eq.S1 Where R is the expected signal response of the immunoassay, T is the top asymptote of the antibody dose response curve, B is the bottom asymptote of the antibody dose response curve, EC50 is the concentration of antigen at which the signal response of the immunoassay is halfway between T and B, and S is the slope of the of the antibody dose response curve at the mid asymptote point. This mathematical equation reveals that lower KD, equivalent to higher affinity of the probe towards the target analyte, yields higher signal intensity (Fig. 6E). This provides theoretical justification why the high affinity chemical baits increase the sensitivity of the assay compared to the conventional sandwich immunoassay.

Theoretical justification for high affinity sequestration of TB antigens in a urine matrix.
Nanocages sequester low molecular weight or low abundance TB antigens (biomarkers) from complex biological matrices, even though the biomarker is complexed with high abundance  Figure S1, for a harvesting cage with a biomarker affinity that is ten times greater than that of the natural carrier protein, k H /k N =10 The mechanism underlying these model solutions may be explained as follows: with its affinity for the natural urinary carrier protein, K N , the closer this ratio will be to unity (all biomarker in complexed association with the harvesting cage.)

Equilibrium constant of LAM binding to dyes in the nanocages.
The following simple model was used to describe the interaction between LAM and the dyes   concentrations of DMF (90%, 75%, 50%, 25%, and 5%) in water in order to eliminate unreacted dye. The cages were re-suspended in 10 mL of water.

Fast Blue B and Safranin O
Fast Blue B (Sigma Aldrich) and Safranin O (Sigma Aldrich) were incorporated into 2:NBiAc nanocages by mixing 10 mL of nanocage suspension with 60 mg/mL dye solution pre-filtered using a nitrocellulose membrane disk filter (0.45 μm pore size). The mixture was allowed to incubate overnight, then washed with a water solution containing 0.1% SDS (five centrifugations, 19,000 rpm, 50 min, 25 °C), and re-suspended in 10 mL of DI water.

Reagent authentication.
The following reagents were obtained through BEI Resources, NIAID Antibody sources and dilutions.

Competition assay to verify specificity of anti LAM mAb clone CS-35.
In order to verify the specificity of band reactivity of the anti-LAM mAb clone CS-35, a competition assay was developed. Prior to staining, the mAb was incubated with a solution containing excess LAM thus neutralizing and blocking the antibody binding sites on the variable regions. The mAb that was bound to the neutralizing antigen was no longer available to bind to the epitope transferred on the western blot membrane. The blocked mAb and the mAb alone were used to probe duplicate western blots. All other parameters of the western blotting remained the same. The comparison of neutralized mAb to mAb alone showed which staining was specific: the specific staining was absent from the western blot membrane probed with the Analytes were transferred onto an Immobilon PVDF membrane (BioRad) for 60 minutes at 50V.
The PVDF membrane was separated into two sections containing identical amount of LAM.
LAM-saturated and un-modified antibodies were used to probe the PVDF membranes. The membrane was incubated with a peroxidase conjugated goat anti-mouse IgG diluted 1:5,000 in 0.2 % I-Block, 0.1 % Tween 20 in PBS. Three washes of 10 min in 0.2 % I-Block, 0.1 % Tween 20 in PBS were performed. Proteins were detected with an enhanced chemiluminescence system (Supersignal West Dura, Thermo Fischer Scientific) on a Kodak MM4000 Imager.

Urine sample handling prior to analysis
Urinalysis was performed on urine samples using Siemens Multistix 10SG. Urine samples were then centrifuged at 3,700 rcf for 10 minutes at 25 °C to remove cellular debris. Supernatant was transferred in a new tube. Urine pH was measured and adjusted to 6 with 1M HCl when necessary.
Magnetic nanoparticles were incubated with hydrogel nanocages for 15 minutes; the cage suspension was then exposed to neodymium magnets.

Cross link degradation of DHEA containing nanocages.
Cages containing DHEA were subjected to oxidation by NaIO4 solution in order to degrade DHEA cross links. Aliquots of cage suspension were mixed with equivalent volumes of NaIO4 dissolved in 0.05 M citrate buffer pH 5.0 for 10 minutes. DHEA and NaIO4 molar ratio was kept at 1:1.

Cross link degradation of BAC containing nanocages.
1: NBaAl (bis(acryloyl)cystamine, BAC cross-linked) cages were subjected to chemical degradation of cross-linkers in order to increase the effective pore size. 200 µL of 1M dithiothreitol (DTT) was added to 200 µL of 1:NBaAl cage suspension (5 mg/mL) and incubated for 30 minutes. 50 µL of 0.5M iodoacetamide was added to the solution and incubated for 20 minutes in the dark. Cages were centrifuged at 16.1 rcf for 10 minutes. Supernatant was discarded and the pellet was resuspended in 1mL of MilliQ H2O. Washing was repeated for a total of 5 times.

Nanocage incubation with LAM and Anti-LAM antibody
In order to assess the binding availability of dye molecules in the inner volume of the cages, HCl and read at 450 nm using Multiskan plus (Fisher Scientific) plate reader.

Nanocage-integrated lateral flow immunoassay.
Anti ESAT6 mouse mAb (Abcam) was diluted 1:10 in PBS for a final concentration of 0.1 mg/mL. 0.01 mL of the diluted mAb solution was deposited on a glass fiber membrane filter (1 × 4 cm, Millipore) and allowed to dry at 37°C in a forced air oven (Fisher scientific Isotemp).
Glass fiber membrane was incubated with 10 mL of blocking solution (50 mg/mL PEG 8000 in PBS) for 30 minutes at room temperature on a rocker. The membrane was rinsed in washing buffer (PBS supplemented with 0.05% v/v Tween 20) and let dry at 37°C in a forced air oven.
Nanocages (3:NBiDAc and TB, Table 7, 0.5 mL of 10 mg/mL dry weight water suspension) were mixed with 10 mL of human urine containing 1-0.03 ng of ESAT6 and incubated for 15 minutes. Cages were washed twice with PBS and incubated with 0.02 mL of 0.1M NaIO4 dissolved in 0.05 M citrate buffer pH 5.0 for 5 minutes. This step caused the cages to shift shape and display the captured antigen. Cages were then centrifuged (16,100 rcf, 25 °C, 5 minutes) and re-suspended in 1 mL of PBS. Two wicks (1.5 × 2.5 cm, extra thick filter paper, Bio-Rad) were placed on the extremities of the glass fiber membrane filter obtained as previously described.
The cage suspension was deposited on one of the wicks and let flow for 5 minutes. Cages containing the ESAT6 antigen arrested on the line. Cages were imaged with a scanner. HRP labelled anti ESAT6 mAb diluted 1:100 in PBS was allowed to flow through the membrane. The membrane was then incubated with SuperSignal Chemiluminescence substrate and the signal was detected with a Kodak imager.

Saccharide quantification.
Total saccharide concentration of the LAM reference standard (BEI Resources) was quantified by the Anthrone method. D-(+)-Glucose (Sigma-Aldrich) was used as calibrator (0.1 mg/mL, 0.08 mg/mL, 0.06 mg/mL, 0.04 mg/mL, 0.02 mg/mL, 0 mg/mL). 50 L of calibrators and samples were added to 100 L of chilled 75% H2SO4 solutions and to 200 L of Anthrone solution (100 mg/mL in ethanol). Samples were placed at 100 °C for 15 minutes and then in ice.
Absorbance of samples was read at =630 nm with a UV-2501 PC spectrophotometer (Shimadzu).        LAM was successfully eluted from the cages and its antibody binding site was intact. Stronger oxidation conditions (100 mM NaIO4 pH 9 1 hour at room temperature) caused a more extensive diol bond oxidation and loss of antibody binding site (E2).  Nanocages that were subjected to cross link degradation exhibited an increased amount of antibody bound to the inner volume of the cage, more than 40-fold with respect to non-degraded cages.