A rapid and flexible microneutralization assay for serological assessment of influenza viruses

Abstract Background Serological responses from influenza vaccination or infection are typically measured by hemagglutinin inhibition (HAI) or microneutralization (MN). Both methods are limited in feasibility, standardization, and generalizability to recent strains. We developed a luciferase MN (LMN) assay that combines the advantages of the conventional MN assay with the ease of the HAI assay. Methods Sera were obtained from the HIVE study, a Michigan household cohort. Reverse genetics was used to generate recombinant influenza viruses expressing the hemagglutinin and neuraminidase of test strains, all other viral proteins from an A/WSN/1933 backbone, and a NanoLuc reporter. Serum neutralization of luciferase‐expressing targets was quantified as a reduction in light emission from infected cells. Neutralization titers were measured for cell‐ and egg‐adapted versions of A/Hong Kong/4801/2014 and A/Singapore/INFIMH‐16‐0019/2016 and compared to HAI titers against egg‐grown antigens. Results Three hundred thirty‐three sera were collected from 259 participants between May 2016 and July 2018. Sampled participants were 7–68 years of age, and >80% were vaccinated against influenza. HAI and LMN titers were correlated for A/Hong Kong/4801/2014 (ρ = 0.52, p ≤ 0.01) and A/Singapore/INFIMH‐16‐0019/2016 (ρ = 0.79, p ≤ 0.01). LMN titers were lower for cell strains compared to egg strains (A/Hong Kong/4801/2014 mean log2 fold change = −2.66, p ≤ 0.01 and A/Singapore/INFIMH‐16‐0019/2016 mean log2 fold change = −3.15, p ≤ 0.01). Conclusions The LMN assay was feasible using limited sample volumes and able to differentiate small antigenic differences between egg‐adapted and cell‐derived strains. The correspondence of these results with the commonly used HAI confirms the utility of this assay for high‐throughput studies of correlates of protection and vaccine response.


| INTRODUCTION
Influenza causes 9 million to 36 million illnesses and 4000 to 60 000 deaths in the United States (US) each year. 1 Vaccination is one of the best ways to prevent influenza infection and the large public health burden associated with it. 2,3 Influenza vaccination has been shown to have variable effectiveness from year to year, a phenomenon that has been attributed to egg-adaptations occurring during the manufacturing process that hamper the mounting of antibodies to circulating strains. [4][5][6][7][8][9] For this reason, it is critical to supplement observational studies investigating vaccine effectiveness, with serological analysis of vaccinated and unvaccinated individuals that are able to distinguish between antibodies elicited by egg versus cell-grown antigens.
Serological responses from influenza vaccination or infection can be investigated through multiple assays. [10][11][12] The hemagglutinin inhibition (HAI) assay measures the ability of antibodies to inhibit the binding of the virus's hemagglutinin (HA) surface glycoprotein to sialic acids on the surface of red blood cells (RBCs), thereby preventing hemagglutination. 10 The HAI assay is relatively simple and inexpensive, and it has been considered a gold standard of influenza serology for decades. 10,11,13 However, HAI assays are vulnerable to nonspecific interfering factors, rely on availability of mammalian or avian fresh red blood cells, have a subjective readout method, and have limited utility for newer H3N2 strains. [10][11][12][13][14] More recently, microneutralization (MN) assays and focus reduction neutralization tests (FRNT) have become more common, as they are more sensitive than HAI assays and more broadly measure the neutralizing activity of sera. 11,12,14 The MN assay also performs well across more recent H3N2 strains. 13 Although MN assays have advantages over the HAI assay, they are more labor intensive and protocols are not well standardized across laboratories. 11,12,14 Here, we describe a luciferase-based MN (LMN) assay that combines the advantages of the conventional MN assay with the ease of the HAI assay. We generated recombinant influenza virus targets that express a luciferase reporter, providing a sensitive and quantitative read-out of cellular infection and the inhibitory activity of sera. We demonstrate this assay's performance using cell culture-and eggadapted versions of two recent H3N2 strains. We related the LMN assay's detection of neutralization activity against egg-adapted H3N2 targets to the HAI assay detection of seroreactivity for the same targets and sera. The cell-adapted targets were used to compare the celladapted and the egg-adapted LMN neutralization responses.
Together, our results demonstrate the utility of this assay for studies of immune correlates of protection.

| Participants and sera
Sera were provided from individuals enrolled in the Household Influenza Vaccine Evaluation (HIVE) study. The HIVE study was approved by the Institutional Review Board at the University of Michigan Medical School (HUM00198212), and written informed consent was provided by patients or a proxy/surrogate. Methods pertaining to the HIVE study can be found elsewhere. 15 Briefly, individuals were followed and contacted weekly to identify acute respiratory infections (ARI). Blood draws were scheduled at enrollment and twice annually after that. Serum was separated via centrifugation. The sera were stored at minus 20 C until further processing. In total, we used 333 serum samples from 259 individuals between May 2016 and July 2018.

| HAI assays
All serum samples underwent HAI testing as a standard processing method. Samples were treated with receptor-destroying enzyme (RDE) following the manufacturer's instructions. 16 The HAI assays were performed in 96-well plates by combining a standardized quantity of HA from specific A/H3N2, A/H1N1, and B vaccine strains with serially diluted sera and turkey red blood cells as previously described. 15 HA targets were obtained through the International Reagent Resource (IRR) and from Sanofi.  supernatants were centrifuged at 1400Âg for 4 min and stored in a final concentration of 0.5% glycerol at À80 C. These passage 0 stocks were titered and used to infect 1 Â 10 6 MDCK-SIAT1 at an MOI of 0.01. These passage 1 supernatants were harvested as above and stored at À80 C in single use aliquots. Stocks were titered as median tissue culture infectious dose (TCID 50 ) in 96-well plates, by scoring positive wells as luciferase expression that exceeded two times the cellular background. This titering method was consistent with conventional TCID 50 measurements that scored cytopathic effect.

| LMN assay
Human (positive) control serum was obtained from a vaccinated donor and stored at À20 C. Lyophilized sheep (negative) control serum was obtained from IRR, reconstituted using sterile water, and stored at À20 C. Before use, all serum aliquots were thawed at room temperature and resuspended by vortexing. Human control, sheep control, and patient sera were heat inactivated (HI) at 56 C for 30 min but not treated with RDE prior to use.
Control and patient sera were added to column 1 at a volume of 10 μL and serially diluted across the plate in 100-μL total volume ( Figure 1A). For each target virus, the P1 stock was diluted to a con- After incubation, the supernatants were aspirated from the wells and 50 μL of a 1:8000 concentration of ViviRen (Promega, #E6491) in viral media were added to each well. The plate was incubated at room temperature for 3 min, and luminescence was measured in a Synergy HTX multi-mode luminometer using a gain of 160. All the steps involving ViviRen were performed in a darkened room.

| RESULTS
The LMN assay is modeled after a traditional MN assay, which mea-  Table S1). All viruses replicated to high titer in MDCK cells.
As in the traditional MN assay, the LMN assay is performed in 96-well plates with 100 infectious units of target virus in each well and serial dilutions of test sera ( Figure 1A). 11 The plate has dedicated wells for "virus only controls (VC)" to capture maximum luminescent output, "cell only controls (CC)" to measure background, and twofold serial dilutions of the target virus as a "back titer" to ensure that 100 infectious units were used. For a plate to pass quality control, the luminescent output (RLU) in the back titer wells should be less than two times the CC average in at least one well between wells 5 and 8 (viral dilutions of 1:32-1:256).
The 75% neutralization titer (NT 75 ) is calculated as the serum dilution at which luminescence is reduced 75% relative to the VC (maximum luminescence in the absence of serum) and CC (background cellular luminescence) wells ( Figure 1A). In preliminary testing, we found that the 75% cutoff was more discriminatory than a 50% cutoff given cellular background and occasional flat, as opposed to sigmoidal, curves with non-neutralizing control sera.
We evaluated the performance of the LMN assay using 333 serum We compared LMN NT 75 to HAI titer for each of these sera.
Because these assays differ in their sensitivity, specificity, and dynamic range, both were ordinalized using a common approach. The NT 75 for a negative control serum (e.g., sheep serum) was set to 0 and each twofold serial dilution of the test sera was therefore 1, 2, 3, and so forth.  HAI titers are widely used as a serologic correlate of protection, although some have expressed concerns regarding their utility. 12,13,[18][19][20] The LMN holds key advantages over the HAI assay. Specifically, the HAI assay has a subjective readout method, which can limit reproducibility, and it can produce variable results using recent H3N2 influenza strains. 13 The LMN assay addresses reproducibility issues by using the back titer, a built-in standardization component that provides a secondary check to ensure that the proper concentration of virus was added to each well on the plate. 11 If the back titer fails, the assay results are invalid and the assay must be repeated. This prevents extrapolating information from plates that were subject to experimental error. Additionally, the readout from the luminometer and calculation of the NT 75 is performed objectively using prespecified RLU cutoffs, eliminating subjective assessments of positive hemagglutination, a proxy of virus-cell binding, the LMN assay measures inhibition of cell entry, a proxy of neutralization. 10,11 The LMN assay is more sensitive and has a broader dynamic range compared to the HAI assay ( Figure S1).

DATA AVAILABILITY STATEMENT
All data are available upon reasonable request by contacting the corresponding author.