Upconverting nanoparticle reporter–based highly sensitive rapid lateral flow immunoassay for hepatitis B virus surface antigen

Detection of hepatitis B Virus surface antigen (HBsAg) is an established method for diagnosing both acute and chronic hepatitis B virus (HBV) infection. In addition to enzyme immunoassays (EIAs), rapid diagnostic tests (RDTs) are available for the detection of HBsAg in resource-poor settings. However, the available RDTs have inadequate sensitivity and therefore are not suitable for diagnosis of patients with low levels of HBsAg and for blood screening. To provide a high-sensitivity RDT, we developed a lateral flow immunoassay (LFIA) for HBsAg utilizing upconverting nanoparticle (UCNP) reporter. The UCNP-LFIA can use whole blood, serum, or plasma and the results can be read in 30 min using a reader device. When compared with a commercial conventional visually read LFIA, the developed UCNP-LFIA had a Limit of Detection (LoD) of 0.1 IU HBsAg/ml in spiked serum, whereas the LoD of the conventional LFIA was 3.2 IU HBsAg/ml. The developed UCNP-LFIA fulfills the WHO criterion for blood screening (LoD ≤ 0.13 IU HBsAg/ml) in terms of LoD. The UCNP-LFIA and conventional LFIA were evaluated with well-characterized sample panels. The UCNP-LFIA detected 20/24 HBsAg-positive samples within the HBsAg Performance Panel and 8/10 samples within the Mixed Titer Performance Panel, whereas the conventional LFIA detected 8/24 and 4/10 samples in these panels, respectively. The performance of the assays was further evaluated with HBsAg-positive (n = 108) and HBsAg-negative (n = 315) patient samples. In comparison with a central laboratory test, UCNP-LFIA showed 95.4% (95% CI: 89.5–98.5%) sensitivity whereas sensitivity of the conventional LFIA was 87.7% (95%CI: 79.9–93.3%). Supplementary Information The online version contains supplementary material available at 10.1007/s00216-020-03055-z.


Details about Reference Assays
. Samples used in UCNP-LFIA evaluation. Fig. S1. Reactivity of hybridoma culture supernatants with HBsAg. Fig. S2. Screening of tracer antibody. Fig. S3. Competitive binding inhibition assay. Table S2. Inhibition percentages resulted from the competitive binding inhibition assay. Fig. S4. Reactivity of different capture antibodies to genotype panel members. Fig. S5. Grouping of mAbs based on competitive binding inhibition assay and genotype detection. Table S3. The analytical sensitivities obtained with different capture mAbs combinations. Fig. S6. The effect of the UCNP-mAb conjugate drying. Table S4. Nitrocellulose membranes used for the comparison.

Reference Assays
The status of the sample as HBsAg positive was based on results in the central laboratory assays as listed in Table S1. The ARCHITECT HBsAg assay (Abbott Laboratories, USA) is an automated chemiluminescent microparticle immunoassay for the detection of HBsAg in human serum or plasma. The LIAISON® XL MUREX HBsAg assay (DiaSorin Inc.) is an automated chemiluminescence immunoassay platform for the detection of HBsAg in human serum or plasma samples. VIDAS HBsAg (BioMérieux, France) is an automated enzyme linked fluorescent assay for the detection of HBsAg in human serum or plasma samples.
Genetic Systems HBsAg assay (Bio-Rad Laboratories Inc., USA) is a manual (compatible with automation) sandwich ELISA test for the detection of HBsAg in human serum or plasma samples.  Reactivity of hybridoma culture supernatants with HBsAg. The reactivity of mouse anti-HBsAg antibodies, from hybridoma culture supernatants, were determined using indirect ELISA method. X-axis represents the clone # and the y-axis represents the OD at 450nm

Fig. S2
Screening of tracer antibody. Eight mAbs were conjugated to the UCNP surface and screened in a dipstick format. Goat polyclonal anti-HBsAg antibody was used as a capture. The produced mAb-UCNP-conjugates were initially screened using blank goat serum (blue) and 100 ng/ml native HBsAg spiked in goat serum (green). The mAb-3D3 was observed to result in the highest specific photoluminescence signal and the highest signal-to-background ratio. In the strip profiles of the lateral flow strips, 3D3-UCNP-conjugate had the best flow properties on the nitrocellulose, and no photoluminescence signal at the nitrocellulose entry point (indicating that the non-specific retention of UCNPs to be minimal). X-axis represents the mAb clones and y-axis represents the test line signal from the strip 0 0 Photoluminescence (A.U.) antibody

Fig. S3
Competitive binding inhibition assay. The mAbs were tested against each other both as free inhibitors and Eu(III)-chelate labeled tracers to investigate whether they bind to the same epitope on HBsAg. Eu(III)-chelate-labeled mAb was allowed to compete with an excess of free mAb in the reaction with HBsAg. Biotinylated anti-HBsAg goat polyclonal antibody was used as capture on streptavidin coated wells. The amount of competing mAb was optimized to ensure a sufficient level of inhibition prior to performing the competitive binding inhibition assay. The inhibition percentage was calculated by subtracting the signal obtained from the competitive reaction from the signal from control reaction (no competing mAb present) and dividing by the control reaction signal. In case of high inhibition percentage, it was assumed that the competing antibody binds to overlapping epitope as the Eu(III)-labeled tracer-mAb, whereas with low inhibition percentage it was assumed that the competing antibody binds to non-overlapping epitope

Fig. S5
Grouping of mAbs based on competitive binding inhibition assay and genotype detection. Competitive binding inhibition assay resulted in two different groups: mAb-2505, mAb-2508, mAb-4G9, mAb-1H1, mAb-3F10 and mAb-3G6 (group 1, green box) and mAb-3D3, mAb-3G8, mAb-1F8 and mAb-3E10 (group 2, yellow box). Group 1 was further divided into two groups based on the results from evaluation of the 1st WHO reference panel (PEI): mAbs with weak reactivity to genotype D2 (4G9, 1H1, 3F10 and 3G6; presented in purple box) and strong reactivity to genotype D2 (2505 and 2508; presented in blue box). Capture antibody combinations were selected from different groups to increase the probability of targeting most HBsAg variants. Overall ability to detect multiple genotypes of mAb candidates was considered when mAbs were selected for further studies Preliminary analytical sensitivities of the assay were determined by using HBsAg-spiked goat serum within the concentration range of 0.05 to 100 ng/ml. The cut-off was calculated as 3*SD of the blank when specific photoluminescence signals were used. The equation was obtained with linear regression of the standard curve. b The highest sensitivity was achieved using the combination of monoclonal antibodies 3G8, 4G9 and 2508 as capture.

Fig. S6
The effect of the UCNP-mAb conjugate drying. Performance of the liquid UCNPreporters are presented with grey bars and that of dried tracer with blue bars. The mAb-3D3-UCNP-bioconjugate was dried on the conjugate pads of the LF strips with 5 % sucrose drying buffer. The effects of the transition to the dry-reagent format were compared to a dipstick assay with UCNP-reporters in solution. The assay was tested with HBsAg-spiked serum in concentrations of 0, 5 and 50 ng/ml. The drying of the reporters did not have an impact on the signal levels  The LFIA was performed with spiked whole blood samples with HBsAg concentrations of 0 ng/ml (blue lines) and 4 ng/ml (green lines) in three replicates per concentration.The strips were measured at 10 minutes intervals and once the strips were dried overnight. Some of the NCMs tested had an initially high signal peak at the 10-minute time point, which then decreased over time. These signals were disregarded when choosing the most reliable and sensitive NCM, since the peak was probably caused by transient weak binding to the test line and was washed away at the later time-points. The chosen nitrocellulose membrane CNPH-N SS60 had a consistent signal increase over reasonable turn-around-time with maximum difference in signal intensity of positive and negative sample at the 20-minute time point. Reading at 30 minutes was considered optimal because of good combination between speed and antigen-specific signal shown. There is a slight increase in the strip baseline signal from left to right, which is because we have read the strip in the wet form (30 minutes from the addition of samples), and some particles may be still moving towards the absorbent pad (right side). However, this difference in the baseline signal along the strip's length does not affect the test result as we subtract the baseline signal from both sides of the peak