4.2. Kinetic Infectious Virus Titer (KIT) Assay
Sample dilution and infection procedure
BHK-21 cells were seeded into 96-well flat bottom plates (Thermo Fisher #161093) at a density of 18,000 cells in 100 µl 22±4 hours before infection. All pipetting steps for cell seeding and virus dilutions were either performed manually or in a semi-automated fashion using Integra pipetting robots (Mini96, VIAFLO96, Assist Plus). All dilutions for the reference standard, samples and assay control were prepared using supplemented cell culture medium as diluent.
In our typical assay setup for VSV-GP, the reference standard curve consists of six MOIs with ½ log10 increment, resulting in a 2.5 log10 assay range (e.g., MOI 31.6 – MOI 0.1). To broaden the assay range, lower MOIs can be included in the reference standard curve, but the duration of image acquisition might have to be prolonged to ensure that all curves reach the constant upper asymptote of rounded cells.
In addition to the reference standard, an assay control (optional) can be included. The assay control is a representative virus sample, which is included on every assay plate. It is treated just like all other samples, but the infectious virus titers determined for the assay control should be documented together with other critical assay details (e.g., operator, cell passage, cell pre-culture time, time between seeding and infection, batch ID of critical reagents) to monitor assay performance over time. Just like for all other samples, measurement of the assay control is performed at a single concentration, i.e., MOI. In our assays, the target MOI used for infection was 1.78, which lies in the middle of the standard curve and allows valid measurements within the working range of the assay even if the assumed sample titer differs ±1 log
10 from the actual titer. MOI calculations are made based on expected titers, which are informed by previous knowledge about the sample type or production process. One challenge that comes with the constant target MOI is that different sample volumes would have to be used for infection for each sample, which is error prone and represents a considerable limitation for assay automation. To bypass this issue, expected sample titers are rounded to the next integer log
10 value and only then used to calculate the sample volumes required to reach the target MOI (refer to
Table M1 for an example of possible dilution schemes). Serial 10-fold sample dilutions are used to obtain the required sample concentration in volumes that are easy to pipet. Due to the previous rounding step, the volumes of pure sample required to infect at MOI 1.78 always differ by a factor of 10 between samples with different expected titers. Thus, only the number of 10-fold pre-dilutions must be adapted, but all pipetted volumes remain constant. This is also true for the final dilution step (in our example: dilution of 192 µl pre-diluted virus in 408 µl medium), in which the highest 10-fold pre-dilution is used to prepare the assay dilution, which is used to infect the BHK-21 cells.
Finally, 100 µl of the serially diluted standard as well as the assay dilutions of samples and assay control were transferred to replicate wells (triplicates in the standard assay setup) of the 96-well plate containing the cells seeded on the previous day. Wells in which 100 µl assay medium (no virus) were added served as uninfected control (mock). 100 µl medium was added to unused wells to maintain consistent culturing conditions throughout the whole plate.
The overall infection procedure was the same when titers of Newcastle disease virus (NDV) were determined in DF-1 cells. The only differences were a seeding density of 15,000 cells/well and different MOIs used for infection.
Automated image acquisition
Immediately after infection, incubation of the plate with imaging of all wells in regular intervals (1 h for VSV-GP, 2 h for NDV) using an the automated Cytation5 (Agilent/BioTek) multimode imaging reader in combination with the automated BioSpa (Agilent/BioTek) incubator was started. The total duration of image acquisition depended on the virus, the MOIs used for infection and the specific purpose of the experiment. Imaging was performed using a wide field of view objective with 4x magnification and laser autofocus. Expression of the red fluorescent Katushka protein was monitored at an excitation/emission wavelength of 584±20 nm/625±20 nm.
Image analysis
Analysis of cell rounding other infection-induced morphological changes was performed on bright-field images. If not stated otherwise, image analysis was performed using Gen5 (Agilent/BioTek). Cellular analysis in Gen5 was restricted to an area with 3000 µm diameter in the center of the image, avoiding the outmost areas of the well with possible lower optical quality. Within the total number of identified cells (object size 5-75 µm) the software determined the rounded cell proportion. Cells are analyzed by approximating their shape with an ellipse and defined as rounded if the ratio of its smallest diameter to the largest diameter exceeds 0.3. A different threshold value would shift the absolute value of % rounded cells of the upper and lower asymptote. Importantly, however, the time-dependent progression of cell rounding would be maintained and due to calculation of the titers relative to a reference standard, this would not significantly affect the results.
ColumbusTM (PerkinElmer) was used as a complementary software tool to determine cell rounding and other cellular changes occurring upon infection. Two approaches with different levels of complexity were applied. First, as done by the Gen5 software, cells were defined as rounded if the ratio of their smallest diameter to the largest diameter exceeded 0.3. Second, a linear classifier model based on the ColumbusTM PhenoLOGICTM technology was applied to discriminate elongated versus rounded cells. The linear classifier was trained with both cell populations (around 200 cells each across many different images) and then all images were analyzed. For classification of cells, Columbus integrated different parameters including roundness, width-length-ratio, cell area, intensity and cell surface texture.
Statistical analysis
In the KIT assay the outcome of the image analysis is the percentage of rounded cells in each well at a specific point of time after infection. If these two factors are plotted against each other with time post infection on the x-axis and % rounded cells on the y-axis, this results in a sigmoidal curve (
Figure 3). The following kinetic curve fit model is used to determine the RT50 value, i.e., the time after infection at which the half-maximal cell-rounding is reached, from each of the sigmoidal curves.
In which: LA = lower asymptote. Value is dependent on the baseline of rounded cells per well. UA = upper asymptote. Value is the maximum % rounded cells, that is reached based on the defined image analysis parameters. k = slope at the inflection point of the curve. RT50 = time after infection at which half-maximal extent of cell rounding is reached, which is identical to the inflection point of the curve.
The RT50 values of the different doses of the reference standard are plotted against the corresponding known MOIs, resulting in a standard curve (
Figure 3). For VSV-GP in BHK-21 cells this standard curve is reflected by a semi-log line, whereas a sigmoidal four parameter logistic fit model was applied for NDV in DF-1 cells. Based on the determined RT50 values of the assay control and the samples with unknown titer, their MOIs can be interpolated from this standard curve. The interpolated MOIs, together with the known number of cells seeded per well and the volume of the samples that was added per well, is then used to calculate the infectious titer. The calculated infectious virus titer has the same unit in which infectivity of the standard sample is expressed.