Enteroviruses are among the most common pathogens infecting infants and children (Coyne et al., 2021). Enterovirus D68 (EV-D68), primarily a respiratory pathogen, was first detected in children with pneumonia and bronchiolitis in 1962 (Schieble et al., 1967). Most EV-D68 infections are asymptomatic or mild, with reports of severe disease remaining relatively rare until the fall of 2014, when the U.S. experienced a nationwide outbreak of severe respiratory illness associated with EV-D68 (Midgley et al., 2015). Concurrently, cases of acute flaccid myelitis (AFM) cases were also reported, particularly afflicting young children (Sejvar et al., 2016).

AFM is a serious neurologic condition characterized by rapid-onset limb weakness or paralysis in otherwise healthy children (Sejvar et al., 2016). Since 2014, nationwide U.S. surveillance has demonstrated sharp increases in AFM cases every two years, with the exception of 2020, with some cases continuing to occur coincident to outbreaks of enterovirus-associated respiratory disease (Kidd et al., 2020; Park et al., 2021). Despite these trends, diagnostic testing has rarely detected a pathogen in the cerebrospinal fluid (CSF) of AFM patients, making it challenging to conclusively pinpoint the etiology of AFM (Kidd et al., 2020). However, data accumulated over the past six years support enterovirus infection, including EV-D68, as a contributing factor in AFM (Messacar et al., 2018). New avenues of research have sought to identify indirect evidence of infection occurring prior to the onset of AFM. For example, AFM patients were more likely than non-AFM patients to have enterovirus-specific antibodies present in CSF, supporting a role for recent enterovirus infection as an underlying factor for AFM (Mishra et al., 2019; Schubert et al., 2019).

Neutralizing antibodies (nAb) are generally regarded as the primary correlate of protection from enterovirus disease, particularly exemplified by poliomyelitis (McKay et al., 1963; Mermel et al., 1993). While specific factors mediating protection from EV-D68 respiratory disease have not been directly demonstrated in humans, anti-enterovirus antibodies can protect against paralysis and death in murine models of EV-D68 neural infection, supporting the induction of humoral immunity as a key protective factor against EV-D68-associated disease (Hixon et al., 2017; Morrey et al., 2018; Zhang et al., 2018). Despite the important role of antibodies in the outcome of enterovirus infection, levels of EV-D68-specific neutralizing antibodies in the general U.S. population have not been well characterized. A single serosurvey performed in the U.S. demonstrated that antibodies capable of neutralizing EV-D68 were widely present in both adults and children prior to the 2014 outbreak of respiratory disease. However, this study was small in scope and not population-based. Larger, population-based studies have not been feasible with the existing 96-well assay.

Here, we describe an automated high-throughput microneutralization assay that provides a platform with high reproducibility and quantitative output for the screening of EV-D68-specific nAb in human serum specimens against various EV-D68 strains. The assay is a modification of the dried blood spot (DBS) poliovirus microneutralization assay that uses a 384-well plate format and automated liquid handling (Weldon et al., 2016). The 384-well platform also allows for conservation of specimens, particularly valuable in testing specimen collections with limited volume available or precious specimen types. A robust, automated, highly reproducible, and quantitative assay that also reduces the amount of sample required for testing will be critical to complete wide-scale EV-D68 serosurveys. This platform also can easily be modified for future high-throughput screening, including identification of small molecules with antiviral effects on EV-D68 infection.

We developed an automated high-throughput EV-D68 microneutralization assay, adapted from the gold-standard 96-well assay (Harrison et al., 2019). The established, traditional 96-well EV-D68 neutralization assay protocol relies on formation of a complete monolayer by RD cells, in order to use visualization and scoring of cytopathic effect (CPE) as the assay readout. Retaining the use of RD cells in the automated assay would be an advantage, since RD cells are highly permissive to EV-D68 infection (Zhang et al., 2019). We shifted to the use of ATPlite^™ reagent for measurement of cellular viability and assay readout. Traditional reagents like crystal violet often stain cell culture debris in addition to live cells that remain adherent, making it difficult to differentiate dead from live cells (unpublished observation).

The reportable nAb titer range for the 384-well assay was initially set based on optimization of virus input (100 CCID₅₀ titer) for each EV-D68 strain, as well as past experience with the 96-well assay. We used control serum designated as in-house reference serum (IHRS), which is pooled human sera with high neutralizing antibody titers to each EV-D68 strain, to serve as initial test sera to confirm the reportable range (Fig. 2). A dilution scheme of 2-fold serial dilutions, starting with a 1:8 dilution, was used for IHRS replicates, to yield detectable Ab titers falling neatly within a range of 2.5 log₂ (10^2.5; lower limit of quantitation [LLOQ]) to 10.5 log₂ (10^10.5; upper limit of quantitation [ULOQ]). IHRS exhibited a wide distribution of nAb titers to EV-D68 strains Fermon, 14–18953, and 18–23087. Ninety-eight percent (98/100) of the IHRS replicates tested fell within the set reportable range of antibody titers, from 2.5 to 10.5 log₂ (Fig. 2) indicating a high probability that tested samples with antibodies against EV-D68 will fall within this range. Consistent with parameters for the poliovirus neutralization assays (Weldon et al., 2016), the titer threshold was set at ≥ 3 log₂ to consider a sample as positive for nAb activity (as determined by a value ≥ 0.5 log₂ from negative control antisera). It should be noted, however, that the minimum protective serum antibody titer for EV-D68 is unknown.

We next characterized the specificity of potential positive and negative control sera, EV-D68-specific antiserum (T2–3420) and EV-D70-specific antiserum (J670/71), respectively. The two antisera were tested against three EV-D68 strains as well as EV-D70 (Fig. 3). EV-D68 antiserum T2–3420 specifically neutralized all three EV-D68 strains, albeit with differences in the nAb titers against each strain, while no neutralizing activity was detected against EV-D70 (Fig. 3A). Conversely, the J670/71 antiserum raised against EV-D70 neutralized EV-D70 and did not have any neutralizing effect on the three EV-D68 strains tested (Fig. 3B). These data support EV-D68 antiserum T2–3420 as a positive control and EV-D70 antiserum J670/7 as a negative control for the assay.

To assess intra-run variability of the assay, the nAb titers for 50 aliquots of the same IHRS pool were determined against three EV-D68 strains within a single experiment (Fig. 4). The nAb titers were quantified by limiting dilution and endpoint titers calculated. The coefficients of variations, calculated as standard deviation of titers / mean titer x 100, were 7.35 % for EV-D8 strain Fermon, 5.97 % for strain 14–18953, and 2.85 % for strain 18–23087. All three values are well below the recommended cutoff of 20 % CV for accuracy (https://www.fda.gov/files/drugs/published/Bioanalytical-Method-Validation-Guidance-for-Ind ustry.pdf). We used Z-factor analysis (Zhang et al., 1999) as a measure of precision, reliability, and robustness of the assay. For high-throughput screening assays, the recommended Z-factor should be ≥ 0.5 – 1.0. This range indicates that the values of the positive and negative controls are distant enough from each other for high assay reliability. The Z-factors for the nAb titers determined during testing of the 50 samples of IHRS against EV-D68 strains Fermon, 14–18953, and 18–23087 were 0.694, 0.638, and 0.852, respectively (Fig. 4), indicating a high level of precision, reliability, and robustness for the assay.

Replicates of IHRS were tested against each of the three EV-D68 strains. For intra-assay reproducibility,10 replicates were tested within a single run of the assay (Fig. 5A, B) and 50 total replicates in five independent runs were tested to determine inter-assay reproducibility (Fig. 5C, D). Both intra-assay and inter-assay variation are well below the 20 % CV threshold, indicating a high level of reproducibility. Interestingly, EV-D68 strain 18–23087, when tested against IHRS, consistently resulted in higher titers (which were near or at the ULOQ). This may suggest differences in antibody affinity. It should be noted that accuracy, precision and reproducibility, for EV-D68 strain 18–23087, could be positively skewed but remained within an acceptable range.

We explored if EV-D68 antibody T2–3420 could cross neutralize additional enteroviruses. Dilutions of antibody were tested against select viruses (100 CCID₅₀ titer). Viruses representing Enterovirus species A (EV-A71), species B (CVB5), species C (PV1) and species D (EV-D68, EV-D70) were assessed using the 384-well microneutralization assay. Of the viruses tested, only species D (EV-D68) viruses exhibited neutralization (Table 1).

We directly compared the performance of the 384-well assay to the previously validated and published 96-well assay. Fifty human serum samples were tested in parallel using the two versions of the assay. The resulting nAb titers for the two assays were highly concordant, except a couple of data points at the LLOQ, with a correlation coefficient of 0.955 (Fig. 6).

The 384-well EV-D68 neutralization assay developed here met or exceeded FDA-established guidelines (https://www.fda.gov/files/drugs/published/Bioanalytical-Method-Validation-Guidance-for-Industry.pdf) for sensitivity, specificity, accuracy, precision, and reproducibility. Detection of intracellular adenosine triphosphate (ATP) was used as a proxy of cell viability, allowing for a quantitative measure of virus-induced CPE, which is directly related to the level of protection provided by nAbs present in tested serum samples. This approach is ideal for high-throughput assays due to its homogenous mix-and-measure assay format, use with cell lines independent of their adherence properties, and amenability to high-density plate formats that can be automated.

The choice of instrumentation for a high-throughput automated neutralization assay is important for the performance of the assay. Small-footprint liquid handlers and reagent dispensers allow the use of existing biosafety cabinets, an important consideration for work using cells and infectious agents. User-friendly software and custom-written scripts facilitate a minimal learning curve for end users. Automated liquid handling instruments allow miniaturization of the assay from 75 μL in 96-well plates, to 35 μL 384-well plates, resulting in reagent savings and conservation of specimens. Automation of small volume transfers and 384-well serial dilutions resulted in greater accuracy and precision, reducing opportunities for user error associated with manual transfers (data not shown).

The new assay will facilitate larger studies than what is feasible with the current 96-well assay, including population-based serosurveys that may span multiple years and demographics as well as screening against multiple virus strain targets. The automated platform allows a sample throughput of up to 160 samples against three virus strains per day and has streamlined several labor-intensive steps including serial dilutions, bulk reagent additions, and microplate reading. This flexible, automated assay platform promises to support future high throughput testing of antibodies and small molecule targets against picornaviruses and other cytolytic viruses.