Self-powered rapid antigen-specific T-cell response assay for Mycobacterium tuberculosis infections

Abstract

Interferon-gamma release assays (IGRAs) that evaluate an individual's T-cell activation response to Mycobacterium tuberculosis (M.tb)-specific peptides serve an important role in diagnosing tuberculosis (TB). However, there are substantial challenges to the use of IGRAs in resource-limited settings. Further, IGRA diagnostic performance can also be compromised in anergic individuals. Here we describe a microfluidic chip-based antigen-specific T-cell response assay (ASTRA) that automates the detection of M.tb-specific T-cell activation responses to facilitate screening for latent M.tb infection and TB. We observe that ASTRA demonstrates high specificity for M.tb infection in independent patient cohorts. Compared with IGRA, ASTRA shows greater diagnostic sensitivity in individuals with HIV-1 co-infections (93.8% versus 67%), comparable diagnostic sensitivity in HIV-negative individuals (92.8%) and faster detection (4 h versus 24-48 h). We also find that a self-powered ASTRA chip that analysed microsample (~25 μl) whole-blood samples produced comparable results. ASTRA holds the potential to facilitate efforts to control the global TB epidemic and serve as a versatile platform for analysing T-cell responses across various infectious diseases and immunotherapeutic interventions.

Extended Data Fig. 1 |. Comparison of ASTRA and IGRA based assays.

a. Venn diagram of IGRA results and ASTRA 4–1BB/OX-40 and IFNγ results for individuals with all three tests results. b. IGRA and ASTRA IFNγ and 4–1BB/OX-40 sensitivity and specificity estimates for M.tb infection overall and in individuals with and without HIV-1 co-infection.

Extended Data Fig. 2 |. Images indicating fluid movement across the self-contained ASTRA chip.

Cell samples (pink) and the two ASTRA antibody reagents (green) and nuclear stain to migrate across the chip upon loading, allowing the cells to be captured on the antibody-conjugated cell loading channel, and segregating the antibodies from these samples until after the peptide-mediated stimulation reaction is complete. Sliding the cases all material to migrate “upwards” along the length of the chip at a constant rate so that the staining reagents interact with the captured cells for a set interval determined by the chip flow rate before being washed away by PBS loaded in three of the downstream channels of the chip.

Extended Data Fig. 3 |. Nanozyme-mediated hydrogen peroxide conversion.

Representative images of oxygen bubble formation (dark spheres) detected in wells containing H₂O₂ alone, Pt-Ni nanozyme alone, H₂O₂ with increasing concentration of Pt-Ni nanozyme after 10-minute incubation at 22 °C.

Fig. 1 |. ASTRA optimization and evaluation with PBMCs from an HIV-negative M.tb cohort from Vietnam.

a, HIV-1 infection inhibits T-cell IFNγ expression but not 4–1BB or OX-40 expression. Created with BioRender.com/0ufe6mv. b,c, Surface 4–1BB and OX-40 and intracellular IFNγ expression detected in (b) PBMCs treated with and without PHA (3 technical replicates) and (c) PBMCs from IGRA (T-SPOT.TB)-positive HIV-negative individuals (n = 30 biological replicates) collected from Vietnam (Vietnam cohort) treated with PHA, M.tb-specific CFP-10 and ESAT-6 peptides, or an Ebola virus-specific peptide. d, OX-40- and 4–1BB-positive cell percentages detected in cells captured on glass slides conjugated with polylysine, CD4- and CD8-specific antibodies, or left bare (n = 30). Centre and upper/lower box lines indicate median and upper/lower quartiles, respectively, and whiskers indicate upper/lower extremes. e, Effect of microchip flow rate on cell capture (n = 3 technical replicates per condition). f, ASTRA IFNγ and ASTRA 4–1BB + OX-40 responses (biomarker-positive cell percentages) detected with PBMCs from individuals in the HIV-negative Vietnam cohort with IGRA-positive and -negative results (n = 44). Dashed line indicate the lower threshold for positive signal. g, Pearson correlation of Vietnam cohort ASTRA (blue dots and shade) and IGRA (T-SPOT.TB, red dots and shade) results (n = 22 biological replicates). h, ROC AUC values for the ability of ASTRA IFNγ and ASTRA 4–1BB + OX-40 results to distinguish IGRA-positive and -negative samples from the Vietnam cohort (n = 14). Graphs display mean ± s.d. ****P < 0.0001; NS, no significant difference; two-way ANOVA.

Fig. 2 |. ASTRA results detected in the HIV-affected Eswatini cohort.

a, Eswatini cohort (n = 44) HIV status, microbiologic (Xpert, smear, culture) and clinical (chest X-ray and NIH-defined symptoms) evidence of TB, immunologic evidence of M.tb infection (IGRA) and ASTRA 4–1BB + OX-40 results. b,c, M.tb-peptide-stimulated ASTRA 4–1BB + OX-40 and IFNγ PBMC responses (positive-cell percentages) for (b) HIV-positive and -negative TB cases (n = 18 biological replicates) and (c) individuals with LTBI or no evidence of M.tb infection (non-M.tb controls) (n = 24 biological replicates) d,e, M.tb-peptide-stimulated ASTRA 4–1BB + OX-40 results detected with PMBCs from Eswatini participants with (d) TB (n = 20 biological replicates) and (e) LTBI with and without HIV-1 co-infection (n = 10 biological replicates). Bar graphs display mean ± s.d. Violin plots depict the distribution of data density (outline), the median (dashed line) and the first and third interquartile boundaries (dotted lines). Dashed red lines indicates the lower threshold for positive signal. **P < 0.01, ***P < 0.001, ****P < 0.0001; two-way ANOVA among ASTRA 4–1BB/OX-40 and ASTRA IFNγ.

Fig. 3 |. ASTRA independent validation with the Houston cohort.

a, Houston cohort (n = 38) HIV status, microbiologic evidence of TB (Xpert, culture) and immunologic evidence of M.tb infection (IGRA) and ASTRA 4–1BB + OX-40 results, in HIV-positive and -negative TB, LTBI and TB suspect cases and healthy control and other disease control groups. b, ASTRA IFNγ and 4–1BB + OX-40 results for HIV-positive and -negative Houston cohort individuals after stimulation with or without (negative control; NC) M.tb peptides or PHA. c, M.tb-peptide-stimulated ASTRA 4–1BB + OX-40 and ASTRA IFNγ PBMC responses for HIV-positive and -negative Houston cohort participants diagnosed with TB or LTBI (n = 16 biological replicates). Bar graphs display mean ± s.d. Dashed red lines indicates the lower threshold for positive signal. ****P < 0.0001, two-way ANOVA among ASTRA 4–1BB/OX-40 and ASTRA IFNγ.

Fig. 4 |. Self-contained point-of-care ASTRA chip design, characteristics and performance.

a, Motive force of the self-contained ASTRA microfluidic chip. b, Left: schematic of loading channels for different assay components. Middle: the path of these materials after the chip is adjusted to initiate the chemical reaction that controls their movement across the chip. Right: representative images of fluorescent signal from all cells (blue) and biomarker-positive cells (green) captured on the chip. Illustration created with BioRender.com. c, TEM image and d, size distribution of Ni-Pt nanozyme particles. e, ASTRA chip flow rates with the indicated nanozyme or catalase concentrations (n = 3 technical replicates). f, Nanozyme- or catalase-driven flow rates after 10-min ASTRA chip storage at the indicated temperatures (n = 3 technical replicates). g, ASTRA chip signal (green shade) versus noise (grey shade) curves with indicated nanozyme concentrations (n = 3 technical replicates). h, Self-contained and pump-driven ASTRA chip 4–1BB + OX-40 results for PBMCs from 3 TB cases and 3 non-TB controls (n = 6 biological replicates). i, ASTRA results from paired whole blood and PBMC samples from 20 HIV-negative M.tb-infected individuals (HIV-negative New Orleans cohort, n = 10 biological replicates). j, Pearson correlation lines and their 95% CIs (shaded regions) for mean ASTRA 4–1BB + OX-40 and ASTRA IFNγ signal with IGRA signal (international units (IU) per ml) for 19 Houston cohort individuals. k. ROC AUC values for the ability of ASTRA 4–1BB + OX-40 and ASTRA IFNγ responses to distinguish individuals with positive and negative IGRA results (n = 8 biological replicates). Bar graphs display mean ± s.d. In boxplots (i), centre and upper/lower box lines indicate median and upper/lower quartiles, respectively, and whiskers indicate upper/lower extremes. ****P < 0.0001, two-way ANOVA between the (e) nanozyme and catalase-loaded chip, (h) self-contained and pump-driven chip and (i) ASTRA 4–1BB/OX-40 and ASTRA IFNγ chip results.