Adaption of a Conventional ELISA to a 96-well ELISA-Array for Measuring the Antibody Responses to Influenza virus proteins, viruses and vaccines

We describe an adaptation of conventional ELISA methods to an ELISA-Array format using non-contact Piezo printing of up to 30 spots of purified recombinant viral fusion proteins, vaccine and virus on 96 well high-protein binding plates. Antigens were printed in 1 nanoliter volumes of protein stabilizing buffer using as little as 0.25 nanograms of protein, 2000-fold less than conventional ELISA. The performance of the ELISA-Array was demonstrated by serially diluting n=8 human post-flu vaccination plasma samples starting at a 1/1000 dilution and measuring binding to the array of Influenza antigens. Plasma polyclonal antibody levels were detected using a cocktail of biotinylated anti-human kappa and lambda light chain antibodies, followed by a Streptavidin-horseradish peroxidase conjugate and the dose-dependent signal was developed with a precipitable TMB substrate. Intra- and inter-assay precision of absorbance units among the eight donor samples showed mean CVs of 4.8% and 10.8%, respectively. The plasma could be differentiated by donor and antigen with titer sensitivities ranging from 1 × 103 to 4 × 106, IC50 values from 1 × 104 to 9 × 106, and monoclonal antibody sensitivities in the ng/mL range. Equivalent sensitivities of ELISA versus ELISA-Array, compared using plasma and an H1N1 HA trimer, were achieved on the ELISA-Array printed at 0.25ng per 200um spot and 1000ng per ELISA 96-well. Vacuum-sealed array plates were shown to be stable when stored for at least 2 days at ambient temperature and up to 1 month at 4-8°C. By the use of any set of printed antigens and analyte matrices the methods of this multiplexed ELISA-Array format can be broadly applied in translational research.

measuring binding to the array of Influenza antigens. Plasma polyclonal antibody levels were 23 detected using a cocktail of biotinylated anti-human kappa and lambda light chain antibodies, 24 followed by a Streptavidin-horseradish peroxidase conjugate and the dose-dependent signal was 25 developed with a precipitable TMB substrate. Intra-and inter-assay precision of absorbance units 26 among the eight donor samples showed mean CVs of 4.8% and 10.8%, respectively. The plasma 27 could be differentiated by donor and antigen with titer sensitivities ranging from 1 x 10 3 to 4 x 28 10 6 , IC 50 values from 1 x 10 4 to 9 x 10 6 , and monoclonal antibody sensitivities in the ng/mL 29 range. Equivalent sensitivities of ELISA versus ELISA-Array, compared using plasma and an 30 H1N1 HA trimer, were achieved on the ELISA-Array printed at 0.25ng per 200um spot and 31 1000ng per ELISA 96-well. Vacuum-sealed array plates were shown to be stable when stored for 32 at least 2 days at ambient temperature and up to 1 month at 4-8°C. By the use of any set of 33 The activity of humoral antibodies provide the best correlation to long-term immune memory 37 and protection (Antia et al. 2018). During the first two weeks of exposure to a pathogen, the 38 majority of antibodies found in the serum derive from plasmablasts, either rapidly re-activated 39 from memory B cell pools or expanded from newly stimulated, somatically hypermutated and 40 differentiated B cells upon contact with antigen in lymph tissue (De Silva and Klein 2015). 41 During recovery, some plasmablasts will home to the bone marrow where they terminally 42 differentiate into long-lived plasma cells stably secreting antibodies that circulate in serum for 43 many years ( The enzyme-linked-immunosorbent-assay (ELISA) first described by Engvall and 57 Perlmann (1972), is commonly used to measure specific antibody-antigen binding. Variants and 58 derivatives of the ELISA have become assay workhorses of immunology laboratories and a host 59 of compatible reagents, consumables, plate washers, multichannel pipettes, robotic liquid 60 handlers, and assay formats have been developed and are available from multiple vendors. A 61 conventional antigen ELISA single plex format passively coats antigens on a 96-well high 62 assays we observed that the upper intensity range was never greater than 180 AU and thus set 154 limits of 0-200 AU in 4P curve fitting. Because at high concentrations the hook effect can lead to 155 reduced intensity readings (Tighe et al. 2015), we disregard any decreased values at high analyte  156 concentrations. 157

ELISA-Array Sensitivity and Specificity 159
In each ELISA-Array assay, polyclonal human immune reference plasma and monoclonal 160 antibodies of known binding activity were used to control for assay performance and determine 161 sensitivity. The dose-dependent binding of each of these controls over three independent assays 162 ( Figure 3), IC 50 values and LLOQ are reported in Table 2. mAb A is known to be a neutralizing 163 antibody recognizing a conformationally dependent epitope on the stalk region of HA trimers 164 (Kallewaard et al. 2016) and was detected against the array of Influenza antigens from 10-165 120ng/mL, well below the quantitative ug/mL range of relevant protective antibody levels in vivo 166 (Crum Cianflone et al. 2012; Semenova et al. 2004). Reference plasma showed Influenza 167 antigen binding IC 50 values of 1.4 x 10 5 to 9 x 10 6 and titers of 6.4 x 10 4 to 1 x 10 6 ( Figure 3 and 168 Table 2). These values reflect a polyclonal mixture of antibodies of any isotype since detection 169 was not limited to IgG (a cocktail of anti-kappa and anti-lambda light chain secondary antibodies 170 was used). The correlation of binding titers to protection varies by disease and for Influenza has 171 not been shown to be predictive (Madore et al. 2010 These were aliquoted and stored at -80ºC. Each plasma donor was also aliquoted undiluted and 197 stored at -80ºC. Although not done in these assays, it would be optimal in the future to briefly 198 spin down donor plasma before assaying to clear the sample of lipid and other aggregates. 199 Aliquots were freshly thawed for each assay, and the same lot of assay diluent and TMB 200 substrate were also used throughout all assays. The final concentrations of assay materials are 201 described in the methods section. Intra-assay precision was determined by running n=3 plate 202 assays on the first day after printing the arrays. Inter-assay precision was determined by running 203 an additional two plates one and two weeks later. Precision was calculated by the variance 204 between plates of titer and IC 50 values for reference plasma and each of the eight donors for all 205 array antigens. Intra-and Inter-assay precision data is shown for the reference plasma and two 206 donors in Figure 4, and Tables 3 and 4, respectively, and for all eight plasma samples in 207 Supplementary Tables S1 and S2. The precision of absorbance units among the reference plasma 208 and two donors showed mean CV of 4.8% intra-assay and 10.8% inter-assay, and 6.0% intra-209 assay and 12.5% inter-assay among the eight donor samples. There were a few examples of high 210 variance inter-assay, in samples of low dilutions. This may be due to weak binding or 211 interference from the serum matrix. The plasma titers could be differentiated by donor and 212 antigen with sensitivities ranging from 1 x 10 3 to 4 x 10 6 and IC 50 values from 1 x 10 4 to 9 x 10 6 213 ( Figure 4, Table 4, and Table S2). For example, we measured a robust titer in the reference 214 plasma donor to all array antigens (Figure 4). In contrast, robust titers in donor 1 plasma were 215 measured only to the vaccine itself, to Influenza B viruses and to the individual antigens of 216 H1N1 and H3N2 HA trimers matching the strains used in the vaccine (Table 1 and Figure 4). 217 There was only weak binding to HA trimers not in the vaccine (i.e. H2, H5, H7), indicating 218 insufficient cross-reactive antibodies were elicited in donor 1. A plot of IC 50 values in Figure 5  219 for three donors shows the overall tight standard deviations between three assays performed over 220 three weeks, and visually quantifies differences between antibody binding for each of the array 221 antigens and donors. We cannot differentiate pre-existing antibody immunity from vaccine 222 responses in these samples, but the quantitative nature of the data would allow for this to be done 223 using titer and IC 50 comparisons with pre-vaccine plasma, not included in this study. Three 224 rounds of freeze thaws of the reference plasma from -80ºC showed no change in IC 50 values or 225 titers (data not shown). One plate assay was also run by a second operator to evaluate the 226 robustness of the method, which was determined to be equivalent to intra-assay precision (Plate 227 4 in Figure 6 and Table S3). 228 229

Stability testing of printed plates 230
Printed array plates were covered with a foil plate seal and vacuum-sealed immediately after the 231 overnight curing step and stored at 4ºC. They were found to be stable stored in this manner for 232 up to one month ( Figure 6). At 8 weeks post-print, plates stored at 4ºC showed about a 2-fold 233 drop in IC 50 and the titer shifted to one higher dilution in the ¼ titration series (i.e. a change from 234 1x10 6 to 2.6 x10 5 ). Significant losses in activity were also measured for vacuum-sealed plates 235 stored for 1 week at either ambient temperature or 37 ºC ( Figure 6). All of the stability assay data 236 including variability for each antigen and plasma sample are provided in Supplementary Tables 237 S4 and S5. Using a different print lot, we tested the plates for 2-days at ambient temperature and 238 found them to be stable (data not shown

261
An ability to print microarrays in a format for a 96-well ELISA-Array was first published by 262 Mendoza et al. (1999), and its utility for infectious disease testing has been demonstrated with 263 antibody arrays to encephalitis viruses (Kang et al. 2012) and viral antigen arrays to Flaviviridae 264 (Wang et al. 2015), using a non-contact piezo Bio-Dot Printing System (Biodot, Irvine, CA). As 265 with these two prior studies, we printed using non-contact piezo nozzles, but in smaller volumes 266 using a Scienion S12 instrument. We compared binding data in arrays using a variety of 267 can help identify the best donor for antibody discovery. In this application it is valuable to have 293 functional neutralization assay data on the same sera, to correlate to binding. 294 Overall, we have provided new methods and qualification data to support applications 295 ELISA-Array assay format for infectious disease research. The key advantages we observed with 296 this technology included the passive coating in a protein stabilizing buffer, the low protein 297 reagent consumption with nanoliter printing, and the ability to perform quantitative analyses 298 using nearly the same workflow, reagents and lab equipment as used in the conventional ELISA. General instrument and experimental parameters 310 The Scienion sciFLEXARRAYER S12 instrument (Scienion AG, Berlin, Germany) has been 311 optimized for non-contact, piezo-acoustic dispensing of ultra-low volumes from an inert coated 312 glass capillary in a climate-controlled (temperature, dewpoint and humidity) environment, with 313 precise XYZ axis control and on-board camera and software for QC of each spot in the array. 314 We printed our arrays with the PDC70 type 3 nozzle (Scienion) due to its reduced 315 dispense volume and the specific hydrophobic coating optimized to improve the dispense 316  The sciREADER CL2 (Scienion) is used for the colorimetric reading of the final assay of arrays. 331 After images are taken of each well, the software analysis program aligns the spot pattern to the 332 imaged spots and calculates a median intensity in absorbance units (AU). 333

Initial ELISA-Array assay used for optimization of parameters 334
Initial 96-well printed arrays were printed according to the general instrument parameters 335 described above. All assay steps were performed at ambient temperature. Blocking solution 336 (sciBLOCK Protein D1M solution, Scienion) was added at 200 μL/well with a multichannel 337 pipet and allowed to incubate without agitation for 1 hour. The block solution was manually 338 Protein stocks of recombinant proteins at 0.5 mg/mL in PBS or vaccine stocks were diluted 1:1 384 in D12 buffer, mixed by pipetting, transferred to a 384 well polypropylene plate 385 (sciSOURCEPLATE, Scienion), and centrifuged for 2 min at 1800xg ambient temperature to 386 eliminate debris or air bubbles. The pattern printed was a 6x6 spot array on each well, and each 387 protein or vaccine along with positive and negative controls was printed in triplicate, 3 spots per 388 well, with 3 drops printed per spot. A single lot of twelve 96-well plates were printed in one day, 389 and after overnight curing were either subject to the ELISA-Array assay the next day (plates 1-390 4), subject to temperature variations for one week (plates 5-7), or subject to varying incubation 391 times at 4ºC (plates 8-10). 392 393 Influenza ELISA-Array assay 394 Each 96-well printed array was printed according to the general instrument parameters described 395 above. All assay steps were performed at ambient temperature, incubations except the blocking 396 step were done with low agitation on a Titer Plate Shaker (Lab-line Instruments, Melrose, IL) 397 and washes were done using a BioTek ELx405 plate washer (Bio-Tek Instruments, Winooski, 398 VT). High agitation is avoided as it leads to comets around the spots, which interferes with 399 accurate spot definition during reading of absorbance intensity. Array plates were washed 1x 300 400 μL/well before immediately adding 200 μL/well of assay diluent (PBS, 0.5% BSA, 2% filtered 401 FBS, 0.2% BGG, 0.25% CHAPS, 5mM EDTA, 0.05% Polysorbate-20 and 0.05% ProClin 300, 402 pH 7.2) down the sides of the wells with a multichannel pipette. Plates were allowed to block for 403 1 hour. Human plasma was diluted 1/1000 in assay diluent and serially diluted ¼ for n=8 points. Data is from 3 inter-assay plates, with mean and SD shown. The IC 50 value for donor 1 plasma 462 against H7 is not shown because the titer was greater than the minimum 1/1000 dilution. 463 464 Figure 6. Intra-assay, Inter-assay and Stability Performance of the ELISA-Array.