Quantitative virus infectivity titration and dose-response analyses using cell-culture based methods and python code-assisted data analysis

TL;DR

This paper introduces cell-based methods and Python scripts for quickly measuring virus infectivity and drug efficacy, offering a faster and more efficient alternative to traditional techniques.

Contribution

The novel contribution is the integration of cell-based assays with Python scripts for automating data analysis in virus infectivity and drug screening.

Findings

The described methods produce results in under 72 hours, faster than existing techniques.

Python scripts reduce manual data handling and enable analysis of large datasets without commercial software.

The methods are suitable for TCID50, IC50, EC50, and CC50 assays.

Abstract

Enteroviruses greatly affect public health worldwide. Therefore, methods to reliably and quickly identify virus infections and screen treatments or preventatives against them are needed. Here, we describe protocols for analyzing the concentrations of infective enteroviruses and neutralizing enterovirus antibodies and the efficacy of antiviral compounds against enteroviruses and compound cytotoxicity with a cell-based method in a microtiter plate format accompanied with python scripts for data and statistical analyses. Detailed instructions for performing the methods and running the scripts are provided, and scripts are available in a public repository. The assays described here are faster and less labor intensive than state-of-the-art methods, producing results in less than 72 h. Developed scripts can reduce the amount of manual handling of the data and calculations, enabling analysis of large datasets without the use of commercial software. The developed methods can be utilized by researchers and healthcare providers for diagnostic and screening purposes for viruses growing in cell cultures.

•Cell-based screening assays for viruses growing in cell cultures•Developed python scripts arrange, analyze, and plot data•Scripts are available for TCID50, IC50, EC50, and CC50 assays•Fast and quantitative methods for efficient assay optimization and routine use

Cell-based screening assays for viruses growing in cell cultures

Developed python scripts arrange, analyze, and plot data

Scripts are available for TCID50, IC50, EC50, and CC50 assays

Fast and quantitative methods for efficient assay optimization and routine use

Virology

Figures

Illustration of the assay workflowA set of protocols are used to quantify the infective virus titer, using endpoint dilution assay, as well as to measure the concentration of neutralizing enterovirus antibodies, evaluate the efficacy of antiviral compounds, and assess compound cytotoxicity through cell-based, dose-response analyses. These assays are supported by python scripts that perform data processing and statistical analysis. Step 1 presents the method for sample preparation. Step 2 shows assay pipelines for the (1) TCID_50_ and (2) DORA methods as well as the choice of fluorescence-based

Graphical results for the determination of neutralizing antibodies generated by the 4-PL python script(A–E) The results for groups 1–5, respectively. Data points in each graph are the averages of two normalized replicate values for each sample and the curve is the output of four-parametric logistic (4-PL) regression based on Equation 2.Figure 3Graphical results for the determination of neutralizing antibodies generated by commercial software(A–E) The results for groups 1–5, respectively. Data points in each graph are the averages of two normalized replicate values for each sample, and the erro

Comparison of CC_50_ and EC_50_ results for pleconaril assay analyzed with python scripts and GraphPad PrismThe efficacy of pleconaril against CVB1 virus and cytotoxicity to green monkey kidney (GMK) cells were determined with 50% cytotoxic concentration (CC_50_) and 50% effective concentration (EC_50_) assays. The data were analyzed with python scripts and GraphPad Prism, using four-parametric logistic (4-PL) regression to fit the data.(A and B) Fitting of CC_50_ (A) and EC_50_ (B) with python scripts.(C and D) Fitting of CC_50_ (C) and EC_50_ (D) with GraphPad Prism.Data points in each graph

Full text

Introduction

Enteroviruses circulate across humans on a daily basis, causing mild-to-severe, and even life-threatening diseases. Thus, there is a fundamental need to be able to diagnose and quantify infective enteroviruses from clinical samples reliably and quickly to target therapies and preventative measures effectively. This need also applies to other important human and animal viruses such as coronaviruses and influenza viruses. Moreover, the efficacy of treatment and prevention methods such as vaccines and antivirals against viruses from clinical and preclinical samples needs to be studied.

Quantification of infectious virus particles in a sample can be achieved through cell-based techniques, including plaque assay and endpoint dilution (EPD) assays such as the 50% tissue culture infectious dose (TCID_50_) assay. Both assays rely on the detection of cytopathic effects (CPEs) in vitro and require a susceptible cell line that can be infected and either killed or damaged by the virus in question.1 The plaque assay has been widely regarded as the golden standard for quantifying infectious viruses.1^,^2^,^3 It is typically done in a 6- or 12-well plate format where a fully confluent monolayer of cells is infected with serial dilutions of a virus sample. After infecting the cells, a semi-solid overlay (e.g., agarose) is layered on top of the cells, restricting the spread of newly produced viruses to neighboring cells only, where localized infection leads to the formation of discrete plaques (clear zones where infected cells have died), each originating from a single infectious virus particle. After an appropriate incubation period, the plaques that have formed into the cell monolayer due to cell lysis caused by the infection1^,^2 are counted manually, and the results are presented as plaque-forming units per milliliter (PFU/mL).

Although plaque assay is considered the gold standard for enterovirus quantification, it has many disadvantages. First, it is suitable only for viruses that form plaques, which not all enteroviruses do. Second, it is time consuming as the assay run time can range from seven to ten days, depending on the virus in question. Third, it is labor intensive, requiring extensive manual handling and experience in the hands-on work, and the results are based on subjective visual interpretation. Although there might be some automated imaging systems available for plaque quantification,4 how largely they have been adopted in research or clinical settings has not been clearly reported to our knowledge. Finally, the assay requires large sample volumes, which limit the ability to run analyses, the number of samples analyzed per assay, and the number of replicates per assay. Therefore, it may take several years to screen the presence of neutralizing antibodies from large sample sets with plaque assay. Laitinen et al.5 published a study, where serum samples from 183 children with type 1 diabetes (T1D) and 366 matched controls were screened against enterovirus-neutralizing antibodies. However, analyses for the study mentioned were conducted for several years because the serum samples were analyzed against 44 enteroviruses (41 serotypes).5 Laitinen et al. conducted the study to identify enterovirus serotypes that could be involved in the initiation of T1D process. These results were utilized in choosing the serotypes included in an experimental vaccine that could protect against T1D.6 Because all six coxsackie B viruses were found to have association with the T1D initiation process, Stone et al.6 produced formalin-inactivated hexavalent coxsackie B virus vaccine, and it was found to be highly immunogenic and had strong protective capacity in mice and non-human primates. In their study, neutralizing antibody levels were again assessed using plaque assays, reinforcing the method’s importance despite its limitations in scalability. To address challenges in formalin-inactivated enterovirus vaccines, we have continued the development of virus-like particle (VLP)-based enterovirus vaccines that could be applied against any enterovirus. In our recent study, we focused on comparing coxsackie B1 (CVB1) virus VLP vaccine immunogenicity with that of formalin-inactivated virus.7 In the study by Soppela et al.,7 we further developed the protocols for analyzing the concentration of infective enteroviruses and neutralizing enterovirus antibodies in the microplate format, but, here, we have streamlined the process and developed full protocols accompanied with python scripts.

Infective virus quantification for viruses growing in cell cultures can also be done with TCID_50_ assay. TCID_50_ assay aims to find the endpoint where 50% of the plated cells are infected.2^,^8 In this method, virus or tissue samples are serially diluted and inoculated onto cell monolayers or mixed with the cells typically in a 96-well plate. Following incubation, each well is evaluated for the presence or absence of CPE. The dilution at which 50% of the wells show infection (i.e., endpoint) is used to mathematically calculate the TCID_50_ titer, using statistical methods such as Reed-Muench or Spearman-Kärber method.2^,^9 Additionally, TCID_50_ value can be further converted to PFU/mL or multiplicity of infection (MOI), which is the average number of virus particles per volume that are able to infect an individual cell.10 Compared to plaque assay, TCID_50_ method is significantly faster and less laborious, with the typical assay run time being approximately 48–72 h. Furthermore, results can be measured with microplate readers that provide numerical data, eliminating the need for data analysis by subjective observations. Importantly, this method can be applied to viruses growing in cell cultures, including viruses that form plaques. In addition, it enables the use of small sample volumes and larger sample sets at the same time. However, setting up a new cell-based assay will always require optimization, as the assay performance can vary not only with the virus in question but also across different laboratories and experimental conditions. In addition to finding a cell line that is susceptible to the virus, other assay parameters such as cell number and virus infection time need to be carefully assessed before the new assay can be applied to experimental studies.

A common way to observe CPE is by staining wells with crystal violet, which stains and fixes live cells attached to the wells.11 After excess stain is washed away and the plates are dried, wells can be scored to determine the 50% infectious dose by visual inspection or absorbance measurement. Another option is to use fluorescence-based methods to measure the infective virus titer by detecting the viability of cells. For instance, non-toxic cell viability reagents such as AlamarBlue can measure cell viability without affecting the wellbeing of cells. AlamarBlue is a cell viability reagent that contains cell-permeable and non-fluorescent blue indicator dye called resazurin (7-hydroxy-10-oxidophenoxazin-10-ium-3-one), which is converted to bright red fluorescent resorufin by metabolically active cells.12 Living cells continuously convert resazurin to resorufin, increasing the overall fluorescence and color of the media surrounding the cells, which makes it a direct indicator of cell wellbeing proportional to the number of living cells.13

In studying the inhibitory effect of antibodies or the efficacy of antiviral compounds against virus replication, first, the infective virus concentration must be determined using plaque assay or TCID_50_ assay. Here, the term dose-response analyses (DORAs) refer to assays such as the inhibitory concentration 50% (IC_50_), effective concentration 50% (EC_50_), and cytotoxic concentration 50% (CC_50_) assays. These assays are all based on the inhibition of virus growth. For example, the IC_50_ assay, traditionally designated as a neutralization assay, is used to determine the presence of neutralizing antibodies in clinical or preclinical samples (such as serum). Antibodies that bind to infectious virus particles can block infection, reducing or preventing cell death in susceptible cells. The quantity of neutralizing antibodies in a sample is determined by serially diluting the sample and challenging each dilution with a standard amount of virus. In the traditional plaque assays and TCID_50_ assay, the last dilution that shows 50% neutralization defines the endpoint, and the dilution of the sample is reported as the neutralizing antibody titer.

We have developed quantitative and precise methods for determining the concentration of infective viruses or neutralizing enterovirus antibodies and the efficacy of antiviral compounds. These assays are combined with our self-developed python scripts, which enable semi-automated data processing and statistical analysis. For the TCID_50_ assay, the scripts calculate the infective virus titer based on the CPE detected via fluorescence or absorbance. For the DORAs, the scripts first normalize the data and then apply a four-parameter logistic (4-PL) regression model. Based on this fit, the IC_50_ value is reported. This represents the antibody concentration (in the analyzed sample) that reduces the virus signal by 50% compared to the control. Using the same approach, the python scripts are compatible with EC_50_ assays, which evaluate the EC_50_ of compounds such as antivirals that can prevent virus infection in cells. Our python scripts can also be used for CC_50_ assays, where serial dilutions of a compound are analyzed for their impact on cell viability, identifying the concentration that reduces viability by 50%, indicating cytotoxicity.

Here, we developed cell-based microtiter plate assays (TCID_50_ and DORA) to measure the concentration of infective viruses and neutralizing antibodies, as well as the efficacy of antiviral compounds against viruses or their cytotoxicity. We provide two different spectroscopic methods to quantify the results, and our self-developed python scripts arrange, analyze, plot and store data from the assays. The developed methods can be applied for virus screening, diagnostic purposes, and for analyzing the efficacy of possible treatments and prevention methods (such as immunogenicity of vaccines) against viruses that can be grown in cell culture. The developed scripts automate data organization and analysis, eliminating the need for commercial software to complete the assays. Furthermore, they provide a valuable tool in the optimization process, as the data analysis can be handled more effectively and the statistics can be used to guide practical laboratory work.

Results

Cell-based microtiter plate method offers a fast way to measure the concentration of infective virus and neutralizing antibodies, and the efficacy and cytotoxicity of antiviral compounds, enabling large-scale screening effectively

Step 1: Sample preparation

Figure 1 illustrates the assay workflow (detailed description is available in STAR Methods), beginning with sample preparation for the TCID_50_ assay. Here, any preclinical or clinical cell-culture compatible samples can be used. The figure shows samples collected from mice as an example. To release potential virus particles from the cells, tissue samples are first homogenized using ceramic beads, followed by several freeze-thaw cycles to ensure that the virus is released into the supernatant. Next, the samples undergo centrifugation to remove cell debris, preventing unwanted cell material from being carried into the subsequent assay steps.Figure 1. Illustration of the assay workflowA set of protocols are used to quantify the infective virus titer, using endpoint dilution assay, as well as to measure the concentration of neutralizing enterovirus antibodies, evaluate the efficacy of antiviral compounds, and assess compound cytotoxicity through cell-based, dose-response analyses. These assays are supported by python scripts that perform data processing and statistical analysis. Step 1 presents the method for sample preparation. Step 2 shows assay pipelines for the (1) TCID_50_ and (2) DORA methods as well as the choice of fluorescence-based method (A) and absorbance-based method (B) for quantification. Step 3 illustrates the process of data handling and analysis with python scripts.See text and STAR Methods for a detailed description. The image was created with Biorender.com.

Step 2 Option 1: Determination of infective virus titer

The TCID_50_ EPD assay is performed in a 96-well plate format, where the infective virus can be quantified based on cell viability. The TCID_50_ titer is measured using spectroscopic methods and calculated with improved Kärber’s formula. First, the sample potentially containing the infective virus is serially diluted on a 96-well plate and allowed to incubate. This incubation period induces the opening of the virus, which enables it to deliver its genome inside the cells to start reproductive infection. After this, the cell monolayer is seeded on top of the virus and incubated for 46 h. The progression of viral infection can be assessed by measuring viability based on fluorescence or absorbance.

To quantify fluorescence signals from the plates, commercial cell viability reagents such as AlamarBlue can be used. These reagents are directly added to the 96-well plates and incubated for an optimal period. Sufficient incubation period allows for the conversion from resazurin to resorufin (when using AlamarBlue) to occur, while minimizing background noise that could arise from, for example, interactions with the medium or the reagent itself. Additionally, the plates have negative control wells that contain only cells to establish limit for the highest (100%) viability. After incubation, a small sample from each well is transferred to a 384-well plate, from which fluorescence measurements are taken. For accurate data analysis using the python script, it is essential that the 384-well plate is filled in the correct order, as outlined in the STAR Methods, because the script processes data according to this predefined order.

An alternative approach is to measure absorbance from the plates after performing crystal violet staining for the cells. In this method, live cells are first fixed to the plate, followed by staining with crystal violet. The stain is then dissolved, and absorbance is measured directly from the 96-well plate. The quantified data are analyzed using the python script.

Step 2 Option 2: Determination of dose responses

The same assay principle can be applied to dose-response analyses, including IC_50_, EC_50_, and CC_50_ assays. In these assays, clinical or preclinical samples (such as serum) or antiviral compounds are serially diluted and incubated with a standard amount of virus (here, 50 TCID-units per well). For the CC_50_ assay, however, the virus is omitted to assess the compounds’ effects solely on the cells’ viability. Similar to the TCID_50_ assay, the plates include positive control wells containing only cells, which establish the highest viability (100%) and serve as the upper limit for the data. Additionally, negative control wells containing both the virus and cells confirm successful infection and define the lowest viability (0%) and the lower limit for the data. The data are adjusted between these 0% and 100% limits during normalization. After incubating the samples with or without virus, virus-susceptible cells are seeded onto the plates and incubated for 46 h. The quantification can be performed using fluorescence or absorbance measurement, using the same protocols as in the TCID_50_ assay. Data and statistical analyses are conducted using python scripts or, alternatively, using commercial software applications.

Step 3: Data and statistical analyses

For data analysis, we have developed different scripts tailored to different assay formats (EPD and DORAs), well-plate formats (96- and 384-well plates), and specific analysis steps for the DORA data (normalization, 4-PL fitting, and statistical analysis). As a prerequisite, all script files must be stored in the same directory alongside the following subfolders: master_files, data_files, result_files, and config_files. Master_files folder contains the master file where TCID_50_/mL results are stored automatically by the script. Data_files folder stores raw data files for analysis that are updated by the script to include results. Result_files folder is updated by the script to include outputs such as graphs generated by the 4-PL script. Config_files folder holds configuration files required for EPD, DORA, and 4-PL scripts, which can be modified by the user. For EPD scripts, the configuration file contains different tissue types. For DORA scripts, the file contains different analysis types. For the 4-PL script, the file contains the analysis types and initial estimate for HillSlope value.

In EPD assay, after fluorescence or absorbance measurements have been obtained and raw data saved in Excel format, data analysis can be performed using a script matching the well count of the plate used. The script guides users through selecting the appropriate master and data files, followed by defining the data with parameters such as tissue type, dilution series, sample count, and number of replicates. Users can choose between 4 or 8 parallels per sample, which determines how many samples can be analyzed per plate, for example, two samples on a 96-well plate and eight on a 384-well plate. These constraints help improve assay accuracy and reduce variability, making the assay more reliable and robust. While the number of parallels and samples is fixed, users have flexibility in setting dilutions and other parameters. Additionally, the script supports inputs of multiple tissue samples or assay conditions, provided that sufficient data remains for analysis and all data are defined in the end. Based on user-provided inputs, the script collects and organizes the data and calculates TCID_50_/mL values based on improved Kärber’s formula described by Lei et al.9 The formula is as follows:

$$[eqn]$$

where ∑ is the total sum of the number of wells at each dilution showing CPEs divided by the number of total wells at each dilution. The results are stored in the master file and displayed in new sheets within the input data file, showing wells with CPE scoring and step-by-step TCID_50_/mL calculations. Additionally, the script computes average, 50% average, and standard deviation for the positive control wells to score sample wells for CPEs

In DORAs, the data analysis is divided into three main steps: data arrangement and normalization, 4-PL curve fitting, and statistical analysis. To arrange and normalize data in DORAScript, the user provides a raw data file, selects the analysis type, assigns group numbers, and specifies dilutions and other attributes. Similarly, as in EPD scripts, users can define multiple datasets as long as sufficient data remain for analysis and all data are defined. Specific data points can also be excluded to prevent them from being analyzed or shown in the final results. The number of parallels has been set to 1, 2, 4, and 8, which allows a maximum of 32 samples per 384-well plate. These options not only enhance the reliability of the assay but also offer flexibility in customizing the analysis. However, the number of parallels and samples must match the number of data points in the raw dataset. Once the data have been defined, the script organizes it, groups samples according to the specified parameters, and generates new sheets containing normalized data and logarithmic dilution values for 4-PL fitting, well-plate statistics, and preliminary graphs for each defined group. The well plate statistics include averages, 50% averages, and standard deviations for the control wells for the whole 384-well plate and subplates. Variations between different assay plates (inter-assay variation) can be observed from the determined subplate values that show how the control values differ between the 96-well plates that were measured.

To determine the IC_50_, CC_50_, or EC_50_ for the samples, the 4PLScript is used. It analyzes data generated by DORA script, focusing only on the groups provided. If the data are modified, it must retain the same format as the original export sheet to ensure compatibility. As with previous scripts, users provide inputs, specify the groups for analysis, and select the appropriate analysis type. The script references the configuration file with default HillSlope values set to −1. However, users can adjust this either by editing the configuration file or by choosing a custom mode during the script run, which allows manual adjustment to the parameters. The script then performs 4-PL regression to determine IC_50_, CC_50,_ or EC_50_ values for each sample. The fitting is based on the following equation:

$$[eqn]$$

where y is the percent inhibition (effect or response), LogIC_50_ is the 10-based logarithm value of IC_50_, CC_50,_ or EC_50_, i.e., the concentration of the sample that gives 50% inhibition of the maximum response, x is the 10-based logarithm value of sample dilution, and HillSlope is the steepness of the curve. The script generates individual graphs for each group and stores them in the result_files folder, while computed values (IC_50_, CC_50,_ or EC_50_, LogIC_50_, and HillSlope) are saved to the input data file

Lastly, STATScript is used to perform statistical analysis. It performs non-parametric statistical tests on data generated by the 4-PL script. Upon execution, the script prompts the user to choose between Kruskal-Wallis test (with Dunn’s test for pairwise comparisons) and Mann-Whitney U test. In both cases, Bonferroni correction is applied to p values. Like the 4-PL script, statistical analysis is conducted only on the data provided, allowing the user to define which groups are included before execution. The final output is stored as a new sheet in the input data file, displaying results in a table format with corresponding p values and significance symbols. Additionally, for Kruskal-Wallis analysis, the test p value is shown prior to pairwise comparisons.

TCID50 assay with fluorescence and absorbance methods

To compare fluorescence- and absorbance-based quantification, the samples were analyzed using two different methods: AlamarBlue cell viability reagent and crystal violet staining. As shown in Table 1, both methods yielded either identical or comparable TCID_50_/mL titers for each tissue or virus sample. The differences in values derived from raw fluorescence and absorbance data corresponded to one-well difference when scoring the CPE on the well plate. The raw fluorescent or absorbance signal measured from each well was compared to the 50% average value calculated from control wells. Wells with values exceeding this average were classified as containing live, highly viable cells, while those with values below the average contained infected, dead cells with low viability. These infected wells were counted for the TCID_50_/mL determination and contributed to the differences observed between the two methods. In cases where the raw values were close to the cut-off threshold, one method might indicate value above the 50% average while the other might register below the threshold, leading to slight variations in results. Fluorescence-based quantification provided a broader spectrum of intensity values compared with absorbance-based method. Although fluorescence-based detection methods are generally more sensitive and accurate, crystal violet staining might better represent the actual condition of the well plate by indicating the presence of viable, adherent cells. Additionally, the results show that cell density did not impact the accuracy of the measurement results. The limit of detection is 49 TCID_50_ units/mL for this assay.14Table 1. Comparison of the determined infective virus titers (TCID_50_/mL)Tissue typeReplicates per sampleSample dilutionCells per wellTCID_50_/mL AlamarBlue™TCID_50_/mL crystal violetLungs418,0002.25 × 10^3^1.44 × 10^3^Lungs418,0001.44 × 10^3^1.44 × 10^3^Pancreas4208,0001.34 × 10^9^1.34 × 10^9^Pancreas4208,0003.73 × 10^7^5.87 × 10^7^Small intestine418,0001.27 × 10^5^1.27 × 10^5^Small intestine418,0001.44 × 10^3^9.20 × 10^2^CVB1-10796810,0008,0004.14 × 10^8^3.31 × 10^8^CVB1-1079681,00016,0006.49 × 10^7^5.18 × 10^7^

Python script-assisted data and statistical analyses are a competitive alternative to commercial software in IC50 assay

To analyze neutralizing antibodies in serum samples, an IC_50_ assay was performed on 30 different serum samples. The data were analyzed using both self-developed python scripts and GraphPad Prism software for comparison with the results presented in Table 2. The table contains the original values computed by both methods as to highlight the possible differences observed in results. Across all three values calculated (IC_50_, LogIC_50_, and HillSlope), the results are nearly identical. Here, the threshold for positivity was set as IC_50_ ≥ 16. For certain group 1 samples, GraphPad was unable to calculate one or more of the values. For samples 1–5, the values denoted as no data (N/D) are shown in GraphPad as “unstable”. However, the python script successfully computed values for all samples except for sample 6. For sample 6, both GraphPad and python showed extreme LogIC_50_ values and determined IC_50_ as 0. HillSlope values differed in that GraphPad also showed its extreme value, while the python script denoted it as 0 as it was unable to calculate it due to the high LogIC_50_ value. Group 1 contained control samples that should not have neutralizing antibodies. This held through the analysis, and the results support that, and it is the reason why software, especially GraphPad, was unable to calculate the values for all three parameters. Particularly with the python script, a good initial estimate for the HillSlope value proved to be essential for accurately determining the results. Additionally, with the other groups, we can see some samples with high neutralizing antibody titers and some with lower titers.Table 2. Comparison of IC_50_ results for detection of neutralizing antibodies from serum samples analyzed with python scripts and GraphPad PrismGroup numberSample numberIC_50_ (GraphPad)LogIC_50_ (GraphPad)HillSlope (GraphPad)IC_50_ (Script)LogIC_50_ (Script)HillSlope (Script)1113.21.12ND12.91.11−23.42NDNDND12.51.10−21.630.00789−2.10−0.6940.008−2.09−0.697412.81.11ND12.41.09−22.0512.81.11ND0−5.95−1.586NDNDNDNDNDND271712.23−2.091722.23−2.0981962.29−3.831962.29−3.8392102.32−1.202102.32−1.20102052.31−2.292062.31−2.291124.31.39−2.3124.31.39−2.311254.61.74−1.7754.61.74−1.7731369.01.84−2.0863.71.80−1.811465.41.82−1.9060.81.78−1.71153132.50−4.343042.48−3.81161302.11−1.351152.06−1.20175102.71−1.116262.80−1.83183522.55−3.833662.56−4.3641922.91.36−2.3022.91.36−2.292024.91.40−1.7424.91.40−1.742130.81.49−1.3530.91.49−1.352253.01.72−1.3753.01.73−1.372372.21.86−0.99372.21.86−0.9932434.41.54−1.4934.31.54−1.495257562.88−1.787172.86−1.57262562.41−2.082452.39−1.862710973.04−0.45311803.07−0.4932862.71.80−1.2965.91.82−1.352913203.12−3.0913403.13−3.243017203.24−2.0117803.25−2.19IC_50_: 50% inhibitory concentration; LogIC_50_: 10-base logarithmic value of IC_50_; ND: no data. Values were rounded to three significant figures.

In addition to the data presented in Table 2, the 4-PL python script generated a graphical illustration of the fitting (Figure 2). Within the script, the y-axis values were locked to a range of −20 to 120, while the x-axis dynamically adapts to the logarithmic values of the dilution specified by the user. The resulting fit closely resembles the one produced using Prism (Figure 3). Differences observed in IC_50_, LogIC_50_, and HillSlope values and graphical outputs may come from variations in the normalization method. In the python script, normalization is calculated based on control averages specific to each group, and decimal values can be used. In contrast, Prism restricts normalization values to whole integers set as the 0% and 100% limits. While Prism provides users the flexibility to manually adjust normalization parameters, the python script performs normalization automatically without user input.Figure 2. Graphical results for the determination of neutralizing antibodies generated by the 4-PL python script(A–E) The results for groups 1–5, respectively. Data points in each graph are the averages of two normalized replicate values for each sample and the curve is the output of four-parametric logistic (4-PL) regression based on Equation 2.Figure 3. Graphical results for the determination of neutralizing antibodies generated by commercial software(A–E) The results for groups 1–5, respectively. Data points in each graph are the averages of two normalized replicate values for each sample, and the error bars show the variation between these replicates as standard deviation. The curve is the output of four-parametric logistic (4-PL) regression based on Equation 2.

For statistical analysis, Kruskal-Wallis with Dunn’s test for pairwise comparisons was performed, as the data were not normally distributed, but were continuous, and comprised independent samples. Additionally, the data contained more than two groups with sufficient sample sizes for comparison. To facilitate statistical analysis in Prism and script, the lower limit of detection, which served as the cut-off for the analysis, was determined as 8, which is half of the first serum dilution in plates. Thus, IC_50_ values below 16 in Table 2 were set to 8. The results from the statistical analyses conducted using both the python script and Prism are presented in Table 3. Overall, both methods yielded similar statistical outcomes for the IC_50_ assay, though slight differences were observed in p values that are likely due to variations in the calculated rank sums. As the statistical analyses were done based on IC_50_ values, differences observed in Table 2 also affected the following statistical analyses. One notable difference in significance estimates occurred between groups 4 and 5, where Prism and python script analyses produced slightly different results. However, given that both p values are close to the 0.05 significance threshold, they would be considered non-significant taking into account possible errors. Overall, the python scripts provide a reliable alternative to the commercial GraphPad Prism software for data and statistical analyses.Table 3. Kruskal-Wallis and Dunn’s test results with Bonferroni correctionp value (GraphPad)Significancep value (Script)SignificanceTest result0.0002∗∗∗0.00015∗∗∗Group 1 vs. 20.07ns0.069nsGroup 1 vs. 30.007∗∗0.009∗∗Group 1 vs. 4>0.99ns1.0nsGroup 1 vs. 50.0002∗∗∗0.00014∗∗∗Group 2 vs. 3>0.99ns1.0nsGroup 2 vs. 4>0.99ns1.0nsGroup 2 vs. 5>0.99ns1.0nsGroup 3 vs. 40.56ns0.65nsGroup 3 vs. 5>0.99ns1.0nsGroup 4 vs. 50.051ns0.042∗Values were rounded to two significant figures. ns, p > 0.05; ∗, p ≤ 0.05; ∗∗, p ≤ 0.01; ∗∗∗, p ≤ 0.001.

Pleconaril shows antiviral activity against CVB1 virus in the DORA

Pleconaril is a small-molecule antiviral compound that has been shown to target receptor-binding pocket of enteroviruses and rhinoviruses, inhibiting receptor binding and/or uncoating.15^,^16 However, it has not been approved for clinical use as no added benefit has been observed due to the side effects.17 Because pleconaril has been well characterized in previous studies, it was selected as a reference to demonstrate the functionality of our assay and show how the python scripts can also be applied to compound testing. In EC_50_ and CC_50_ assays, the efficacy and cytotoxicity of pleconaril against CVB1 virus was analyzed. At the same time, the data analysis was compared using the python scripts and GraphPad Prism as with IC_50_ assay. Previously, Honkimaa et al.15 reported for pleconaril that CC_50_ was 3,111 μM and EC_50_ was 1.19 μM, using CVB1-10796 virus strain and A549 cells. Here, we used the same virus strain with green monkey kidney (GMK) cells. The results in Figure 4 E differ from those previously reported. The EC_50_ was much higher but the CC_50_ was lower than those expected. The difference between the antiviral’s CC_50_ and EC_50_ was small. Additionally, the amount of virus per well was 20.4× TCID50 units, which is slightly lower than that aimed for (i.e., 50× TCID50 units). Compared to those by Honkimaa et al.,15 our assay setups are vastly different. In addition to different cell lines, Honkimaa et al. used higher cell density and virus amount for infection, and they preincubated the cells before adding the virus-compound mixture. On the other hand, data analyses with GraphPad Prism and python scripts showed identical results (Figure 4). In both cases, the 4-PL fits were identical as seen from the graphs, as are the CC_50_ and EC_50_ results. Moreover, the values computed for LogCC_50_, LogEC_50_, and HillSlope were the same with both Prism and python script analyses (data not shown).Figure 4. Comparison of CC_50_ and EC_50_ results for pleconaril assay analyzed with python scripts and GraphPad PrismThe efficacy of pleconaril against CVB1 virus and cytotoxicity to green monkey kidney (GMK) cells were determined with 50% cytotoxic concentration (CC_50_) and 50% effective concentration (EC_50_) assays. The data were analyzed with python scripts and GraphPad Prism, using four-parametric logistic (4-PL) regression to fit the data.(A and B) Fitting of CC_50_ (A) and EC_50_ (B) with python scripts.(C and D) Fitting of CC_50_ (C) and EC_50_ (D) with GraphPad Prism.Data points in each graph are the averages of four normalized replicate values for each sample, and the error bars in (C) and (D) show the variation between these replicates as standard deviation.(E) Comparison of data analysis results from python script and GraphPad Prism. Values were rounded to four significant figures.

Discussion

Here, we have described cell-based assays for measuring infective virus titer from samples, quantifying neutralizing antibodies in serum, and assessing both the effective and cytotoxic concentrations of an antiviral compound. Compared with the plaque assay, our TCID_50_ method has a more streamlined assay pipeline for analysis of larger sample sets. We demonstrated the functionality of the assays by using coxsackie B1 enterovirus and optimized the performance of the methods in GMK cells. However, for successful application of the developed methods for other viruses and pathogens, the first requirement is to find a susceptible cell line for the virus to be analyzed. After that, adjusting the incubation time to ensure sufficient infection and modifying cell density within wells might be needed. In screening for infective viruses, neutralizing antibodies, or efficacy of antiviral compounds, plaque assays remain currently the state-of-the-art method in enterovirus-based analyses. However, plaque assays are time-consuming processes, particularly when large sample sets and multiple virus serotypes are involved, like in studies where neutralizing antibodies had to be screened from clinical5 and preclinical6^,^7 samples. Here, we developed quicker and more robust assays that could be used alongside plaque assays in enterovirus diagnostics and also utilized with viruses that can be grown in cell culture.

When a virus has been optimized to grow on microplates in a susceptible cell line, successful assay performance requires proper sample handling. Here, we optimized proper handling of the pancreatic, lung, and small intestine samples extracted from mice with homogenization, keeping in mind that virus could be recovered with the highest possible yield from the tissues. We determined that tissue samples need to be thoroughly homogenized, and centrifugation needs to be performed carefully to prevent cell debris from being carried onto the well plates. We detected unwanted cell growth originating from the mouse tissue, if the homogenization and/or the centrifugation of disrupted tissue was not adequate, yielding false negative result, especially in crystal violet staining. Like we have demonstrated previously,18^,^19 we detected that incorporating Tween80 into buffer where the tissues are homogenized could help preserve virus in the samples. In addition to tissue samples described here, we have previously analyzed whole blood for detection of infective virus in the TCID_50_ assay. It was noted that the sample preparation step can impact quantification, especially when using crystal violet staining. It was concluded that the sample should be diluted enough before use to prevent blood from coagulating and forming a film on top of the cells. This film disrupts the fixation and staining of cells, leading to false-negative results because it causes the film, instead of the cells, to be stained (data not shown). Moreover, this also affected fluorescence-based quantification, highlighting the importance of using adequate dilutions.

Unlike plaque assay, where plates are stained with crystal violet and plaques are manually counted on a light table, our method automates the quantification process by using a plate reader to measure fluorescence or absorbance straight from the plate. This enables a faster and more objective analysis. Furthermore, both our crystal violet staining method and the cell viability reagent could effectively quantify virus titers and produce comparable results. In this study, we utilized AlamarBlue and AlamarBlue HS as viability reagents. While other commercial alternatives are available, optimization might be required to ensure reliable outcomes. In some cases, the background signal from certain reagents can interfere with the assay, causing fluctuations in fluorescent signals and thus leading to potential inaccuracies. To mitigate this, optimization of the reagent incubation time is recommended. In addition, examination of the plates with a microscope can aid in confirming that the measured values correspond to the observed cell state and CPEs. When screening compounds for antiviral efficacy and cytotoxicity, it is important to consider potential fluorescence activity of the compound itself, as this may affect the results. In such cases, crystal violet staining might be a more suitable quantification method. Compared to crystal violet staining, cell viability reagents tend to be more sensitive, and applying them in the method gives a more streamlined protocol with fewer steps. However, these reagents and additional materials often come at a higher cost. In contrast, the crystal violet method relies on more cost-effective materials, though it requires a longer assay run time. Additionally, this method is generally less affected by background noise and often correlates more directly with visual observations. Also, the differences between these two methods are largely caused by the way the signal or value is measured, and the variation in the range of values for both methods affects the quantification of results.

The python scripts provide a straightforward, user-friendly, and efficient approach to data analysis that can be applied without any prior experience in programming or data fitting. They have been designed to significantly reduce the need for manual data handling and to complement the cell-based assays described in this study, but they remain flexible and allow users to modify and customize them as needed. Additionally, these python scripts serve as a valuable tool for data and statistical analyses, alongside commercial licensed software such as GraphPad Prism. The scripts facilitate structured data analysis by enabling users to process the data in sections, preserving the original data while systematically saving new results. This kind of functionality is particularly advantageous for handling large datasets and for assays requiring optimization and validation. The scripts provide data analysis and statistics that can be used as a measure to estimate how well the assay works or how well the practical laboratory work has succeeded. For example, DORAScript calculates the statistics for control wells in each 96-well plate and for the whole 384-well plate analyzed in each assay. From the statistics, the variations between plates can be assessed, while the use of multiple replicates gives insight into the variation occurring in each plate. While the scripts allow customization, commercial software offers a broader selection of analytical options and supports various assay types, beyond those applicable to the scripts. However, as long as other assays follow similar principles or can be adapted to the cell-based assays described here, the python scripts are a viable option for data and statistical analyses. Additionally, as the scripts have different basic functions, it is possible to use the scripts independently from others if the prerequisites are met for the data. Moreover, because these scripts are available in a public repository, they can serve as a foundation for further development. Users can modify the code to introduce new functions or customize existing ones to better suite the specific assay requirements. Furthermore, these scripts are available free of charge, while with commercial software, users are often required to purchase a license, which can limit the availability.

In our DORAs, we use a 4-PL function to model the dose-response relationships and quantify sample responses. This non-linear model is often used to analyze dose-response relationships in bioassays. An alternative model is the five-parameter logistic (5-PL) function, which introduces an additional parameter to account for the degree of asymmetry around the sigmoid curve’s inflection point.20 The 5-PL model has been considered to offer more accurate results and representation of the data than the 4-PL model when the data are clearly skewed, helping to reduce bias in estimates caused by a lack of fit.20^,^21^,^22 However, in some cases, even with asymmetric data, the 4-PL model may perform better than the 5-PL model.20 One advantage of the 4-PL model is its simpler implementation. By adding a sufficient number of replicates and data points per sample, random variation can be minimized, allowing for better approximation of the curve.21 We have considered this variation both in the assay design and in the data analysis by offering options for the number of replicates and sample quantities in the script. These constraints help improve reliability and consistency across assays while still allowing customization. Additional flexibility is provided through user-defined parameters such as dilution factors, dilution rates, and working volumes, giving users control over key aspects of the analysis.

Additionally, to evaluate the versatility of DORA and python data analyses, we assessed the cytotoxicity and efficacy of pleconaril in terms of CC_50_ and EC_50_, respectively. The determined values deviated from those in previous studies.15 The EC_50_ was exceedingly higher while CC_50_ was considerably lower than what have been reported. Honkimaa et al.15 employed a substantially different assay setup compared to our assay setup, differing in the cell line used, cell numbers, and plating style. Different cell lines exhibit differences in susceptibility to virus infection, and their ability to replicate in each cell line varies.23^,^24 Moreover, it has been determined that even one amino acid change in the virus capsid can affect the efficacy of an antiviral drug. Schmidtke et al.25 showed that a resistant CVB3 virus could become susceptible to pleconaril after the amino acid at position 1092 of VP1 was mutated to isoleucine. Thus, even though Honkimaa et al. had the same CVB1 strain, at some point during the passaging, our CVB1 strain might have accumulated a mutation that has reduced its susceptibility to pleconaril. Confirmation of this would require a sequencing analysis to determine possible differences in the original sequence. Additionally, one key difference in our assay setup compared to that of Honkimaa et al. is that they plated the cells first and let the cells adhere to confluent monolayers before they added the virus-compound mixture. Often, drug testing is initiated on a confluent monolayer of cells, and this preincubation period can affect the efficacy of the compound in preventing infection.26^,^27 Similarly, cell density affects the infectivity threshold of the virus and can alter cell response to compounds as cell confluency affects not only the medium nutrient consumption but also the expression of proteins, communication between cells, and the overall state of the cells.26 Even though with this cell culture assay setup, we have been successful in IC_50_ assay to determine the neutralizing antibody content of samples, it might not be the optimal setup for compound testing. Thus, the end result of an assay is affected by all assay parameters, and different cell-based assays vary a lot in their sensitivities; also, the sensitivity can differ depending on the application for which the assay is being used. Furthermore, the sensitivity limit for the test is determined by factors such as the MOI at which virus can infect cells, how quickly the replication occurs, and how quickly the cells lyse and release new viruses into the media that can infect new cells.

In conclusion, the quick and robust assays developed here may significantly enhance the throughput and reproducibility of enterovirus diagnostics, while their broader application to other viruses will depend on identifying suitable cell lines that are susceptible to virus growth. Developed scripts reduce the amount of manual handling of data and calculations, enabling the analysis of large datasets without the need for commercial software. Additionally, they are a valuable tool when assay validation is in question. Overall, these innovations contribute to improved viral diagnostics, enhanced screening of treatment and prevention methods, and global health preparedness.

Limitations of the study

We developed the assays and protocols by using coxsackie B1 enterovirus. The developed assays are not standardized assay formats, and their adoption will require optimization into each research setting. Due to limited resources and access to only biosafety level 2 (BSL2) facilities, we could not test the developed assays and scripts with viruses from other species such as influenza or severe acute respiratory syndrome coronavirus 2 (SAR-CoV-2). In principle, the assays should work with viruses that can grow in cell cultures. However, the sensitivity of the assay is dependent on the susceptibility of the cell line for the virus. Therefore, definitive conclusions on the applicability of these assays to other viruses are not made. To successfully apply the developed assays for viruses other than CVB1, the virus being assayed needs to be optimized for assay parameters such as optimal cell line, cell number, virus incubation time, and assay readout. No plaque neutralization assay was conducted to compare the IC_50_ results obtained with the presented method.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Henna-Maarit Kyröläinen (henna-maarit.kyrolainen@tuni.fi).

Materials availability

This study did not generate new unique reagents. The detailed information of all materials and reagents used is given in the key resources table. The assay pipeline protocols for TCID_50_ and DORAs are described in Methods S1 and S2, respectively.

Data and code availability

• •Complementary data to python scripts have been deposited at GitHub repository: https://github.com/VIVAgroupTUNI/TiterDORATools.git and are publicly available.
• •GraphPad Prism (version 10.3.1) software is available at GraphPad Software (www.graphpad.com).
• •Source code is publicly available at GitHub repository: https://github.com/VIVAgroupTUNI/TiterDORATools.git.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

The authors would like to thank Niina Ikonen for her skillful technical assistance (Virology and Vaccine Immunology Group, Tampere University). Tampere University Preclinical and Virus Facilities are thanked for providing equipment and assistance for this study.

The research was funded by the 10.13039/501100002341Research Council of Finland, grant numbers #309455, #335870, and #355414 (M.M.H.); 10.13039/501100004012Jane and Aatos Erkko Foundation grant number #240002 (M.M.H), the 10.13039/501100003125Finnish Cultural Foundation (M.M.H.), 10.13039/501100014438Business Finland (M.M.H.), 10.13039/501100007417Paulo Foundation (M.M.H.), Tampere Institute of Advanced Studies (M.M.H.), 10.13039/501100013500Diabetes Research Foundation (M.M.H. and I.M.), Tampere University Graduate School (I.M.), and 10.13039/501100006706Tampere Tuberculosis Foundation (M.M.H., H-M.K., and I.M.).

Author contributions

Conceptualization, M.M.H; data curation, H.-M.K., J.E., and M.M.H.; formal analysis, H.-M.K., I.M., and M.M.H.; funding acquisition, H.-M.K., I.M., and M.M.H.; investigation, H.-M.K. and I.M.; methodology, H.-M.K., I.M., and M.M.H.; software, H.-M.K., J.E., and M.M.H; supervision, H.-M.K. and M.M.H.; visualization, H.-M.K. and I.M.; writing – original draft, H.-M.K. and M.M.H.; writing – review & editing, H.-M.K. and M.M.H.

Declaration of interests

M.M.H. is a shareholder of Innofend Oy.

STAR★Methods

Key resources table

REAGENT or RESOURCESOURCEIDENTIFIERBacterial and virus strainsCoxsackievirus B1-10796Centers for Disease Control and Prevention, Division of Viral DiseasesGenBank: PP782006Biological samplesMice serum and tissue (lung, pancreas, small intestine) samplesThis paper–Chemicals, peptides, and recombinant proteinsMEM: Minimal Essential Medium Eagle with Eagle’s salts, L-glutamine and sodium bicarbonateSigmaCAT#M4655-500mLFBS: Fetal Bovine SerumSigmaCAT#F9665-500mLPenicillin-Streptomycin (10,000 U/mL penicillin, 10 mg/mL streptomycin)SigmaCAT#P0781-100mLTrypLE™ Express Enzyme (1X)GibcoCAT#12604-013Trypan Blue Solution, 0.4%GibcoCAT#15250061AlamarBlue™ HS Cell Viability ReagentThermo ScientificCAT#A50101-100mLAlamarBlue™ Cell Viability ReagentThermo ScientificCAT#DAL1100Crystal Violet 2 w/v-% H_2_O SolutionOy FF-Chemicals AbCAT#FF407-500mLGlutaraldehyde solutionSigma-AldrichCAT#340855-1LAcetic acid (glacial) 100%MerckCAT#1000632511Sodium phosphate dibasic (Na_2_HPO_4_)Sigma-AldrichCAT#S0876-1KGHEPES solution, 1M pH 7.0-7.6SigmaCAT#H0887-100mLHanks‘ Balanced Salt Solution with sodium bicarbonateSigmaCAT#H9269-500mLTween80Sigma-AldrichCAT#P1754PleconarilMedChemExpressCAT#HY-19952DMSO: Dimethyl sulfoxidePanReac AppliChemCAT#A3672Deposited dataSource code and complementary materialThis paperZenodo: https://doi.org/10.5281/zenodo.18456468Experimental models: Cell linesGMK: Green Monkey Kidney cell lineFinnish Institute for Health and Welfare, Department of Virology–Experimental models: Organisms/strainsMice: BALB/c-JRjJanvier Labs–Software and algorithmsGraphPad Prism version 10.3.1 for WindowsGraphPad Software (www.graphpad.com)–OtherBioLite Cell culture treated flask 75cm^2^Thermo ScientificCAT#130190Nunc™ MicroWell™ 96-Well, Nuclon Delta-Treated, Flat-Bottom MicroplateThermo ScientificCAT#161093AlphaPlate 384-shallow well (ProxiPlate)RevvityCAT#6008350BD Microtainer® blood collection tubeBecton DickinsonCAT#365968Precellys® lysing kit - Tissue homogenizing CKMixBertin TechnologiesCAT#P000918-LYSK1-APrecellys® 24 HomogenizerBertin Technologies–Nunc™ Sealing tapeThermo ScientificCAT#236707EnVision® 2104 Multilabel ReaderPerkin Elmer–VICTOR^3^TM 1420 Multilabel CounterPerkin Elmer–Linear Shaker SO2Stuart Scientific–Optima™ XPN-100 ultracentrifugeBeckman CoulterCAT#B10048

Experimental model and study participant details

Cell line

Green Monkey Kidney (GMK) cells (provided by the Finnish Institute for Health and Welfare, Department of Virology, Dr. Soile Blomqvist) were grown in Minimum Essential Medium Eagle (MEM) with Earle’s salts, L-glutamine and sodium bicarbonate supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (all from Sigma) at +37°C with 5% CO_2_. The cells were cultured in T-75 cell culture flasks and detached using TrypLE™ Express Enzyme (Gibco).

Virus strain

The assay was developed using a wild CVB1 field isolate (CVB1-10796; isolated from Argentina 1983; described in Hämäläinen et al.,28 provided by Centers for Disease Control and Prevention, Division of Viral Diseases), GenBank accession number PP782006. The virus was propagated in GMK cells and recovered from supernatant with slightly modified protocol described in Hankaniemi et al.18 Briefly, GMK cells were infected with MOI 10-20 and virus was harvested from supernatant 3-5 days post infection as follows: 0.1% Tween80 was added in the supernatant, cell debris was pelleted and discarded and virus was concentrated by pelleting through 30% sucrose in PBS-0.1% Tween80 using ultracentrifugation (175 k g, 14 h at 4°C). The pellets were dissolved in PBS-0.1% Tween80 and filtered through 0.2 μm filter. The virus stock was aliquoted in 0.5 ml aliquots and aliquots were stored at -80°C.

Preclinical tissue and serum sample preparation

Female BALB/c-JRj mice (14-week-old, Janvier Labs) tissue samples (two lungs, pancreases and small intestines) were weighted and inserted into 2 ml Precellys® tubes (Bertin Technologies) and 1 ml of PBS-0.1% Tween 80 was added with the lungs and 1 ml of MEM with pancreases and small intestines. Homogenization was done using sterile ceramic beads (1.4/2.8 mm) by vigorous shaking with Precellys® 24 Homogenizer (Bertin Technologies) for two 30 second repeats at 6500 rpm. Samples were freeze-thawed three times and then centrifuged at +4°C and 16,000 g for 15 minutes. The supernatant was collected and stored at -80°C.

Mice blood samples (30 samples) were collected to BD Microtainer® blood collection tubes (Becton Dickinson). The tubes were centrifuged for 2 min at 8,000 g, RT. Serum was collected and stored at -80°C. For more detailed protocols for sample handling, see Methods S1.

All animal experiments in this study were carried out following the Finnish Act on the Protection of Animals Used for Scientific or Educational Purposes (497/2013) or in compliance with Swedish national laws and the National Institutes of Health guidelines for laboratory animal care. Procedures were approved by the Regional State Administrative Agency, Pirkanmaa, Finland (decision number ESAVI/48887/2023) or by the Stockholm Southern Animal Ethics Board.

Method details

Infective virus titer

To quantify infective virus titer from preclinical and virus samples, TCID_50_ endpoint dilution assay was conducted using GMK cells (Methods S1). First, a solution of HEPES-Hanks-FBS (H-Hanks-FBS) was made by supplementing Hanks’ Balanced Salt Solution with 0.01M HEPES and 0.6% FBS (all from Sigma) which was added to all wells in 96-well plate (25μl) (Thermo Scientific). Samples were diluted to H-Hanks-FBS and a 6-fold dilution series with four replicates for preclinical samples (lung, small intestine or pancreas) and eight replicates for virus control were made starting from column three. The first two columns were left without sample as a positive control for cell viability. The plate was sealed with adhesive tape and incubated for one hour at +37°C with 5% CO_2_. After incubation, GMK cells in MEM with 5% FBS were seeded on the plates at a density of 8 000 cells per well (70 μl) and the plate was incubated for 46 hours at +37°C with 5% CO_2_. The result is analyzed according to cell viability utilizing either homogenous cell viability reagent (fluorescence quantification) or absorbance measurement from crystal violet stained cells.

Fluorescence quantification

After incubating the cells with the virus for 46 hours, 10 μl of AlamarBlue™ Cell Viability Reagent (Invitrogen) was added to all wells and the plate was incubated at +37°C with 5% CO_2_ for two hours. For the measurement, the well contents were mixed with multichannel pipette one column at a time and 15 μl sample from each well was transferred to a 384 well plate. 15 μl samples from the first 96-well plate were pipetted to the 384-well plate starting from row A (columns 1-12) and the next plate to rows starting from B (columns 1-12). 15 μl samples from 96-well plates three and four were pipetted similarly as plates one and two but to columns 13-24 on 384 well-plate. The plate was centrifuged with 2,000 rpm for 2 minutes and the fluorescence was measured with EnVision® multiplate reader (PerkinElmer) with excitation and emission wavelengths of 560 nm and 590 nm, respectively. The data was analyzed with a python script.

Absorbance quantification

The protocol is modified from Kueng et al.29 After 46-hour incubation, cells were fixed by adding 10 μl of 11% glutaraldehyde solution (diluted to 0.1M Na_2_HPO_4_ buffer, pH 7.5, from 25% solution, Sigma-Aldrich). The plate was on a shaker at RT for 15 minutes after which the liquid was removed, and the plate was washed three times by submersing in deionized water. Excess water was tapped on a cellulose wipe and left to dry in fume cupboard until dry. The cells were stained by adding 50 μl of 0.1% crystal violet solution (diluted in deionized water from 2% solution, Oy FF-Chemicals Ab) and plate was put on a shaker for 20 minutes. Excess dye was first tapped away on a cellulose wipe and then washed four times by submerging in deionized water. The plate was left to dry in fume cupboard. The dye was dissolved by adding 100 μl of 10% acetic acid (diluted to deionized water from 100% solution, Merck) and put on a shaker for 15 minutes. Then, absorbance was directly measured from the plate at wavelength 590 nm using VICTOR^3^TM 1420 Multilabel Counter (PerkinElmer). The data was analyzed with a python script.

Neutralizing antibodies

To analyze CVB1 neutralizing capacity of mice serum samples, neutralizing antibodies against a live virus were measured by IC_50_ assay using GMK cells (Methods S2). First, a two-fold serial dilution series of serum was prepared in duplicate in H-Hanks-0.6% FBS solution into a 96-well plate to columns 3-11. 50 x TCID_50_ units of the CVB1 virus was added per well to columns 3-12. The plate was sealed with tape and incubated at +37°C with 5% CO_2_ for one hour. After incubation, GMK cells in MEM with 5% FBS were seeded at a density of 16,000 per well (62μl) and the plate was incubated for 46 hours at +37°C with 5% CO_2_. After incubation, the plate was analyzed with AlamarBlue™ HS cell viability reagent (Thermo Scientific) like described for TCID_50_ assay. The raw data was analyzed with python scripts and GraphPad Prism version 10.3.1 for Windows (GraphPad Software, www.graphpad.com). Additionally, a TCID_50_ control plate with 16,000 cells per well and 1:1,000 virus sample was prepared.

EC50 assay to analyze the efficacy of a compound

Pleconaril (MedChemExpress) was dissolved in DMSO in concentration of 7.63 mg/ml and stored at -80°C. The assay was done like described for the IC_50_ assay replacing the serum sample with 5,000 μM pleconaril sample (Methods S2). The plates were analyzed with the AlamarBlue™ HS cell viability reagent like described for TCID_50_ assay. The raw data was analyzed with python scripts and GraphPad Prism version 10.3.1 for Windows (GraphPad Software, www.graphpad.com). Additionally, a TCID_50_ control plate with 16,000 cells per well and 1:1,000 virus sample was prepared.

CC50 assay to analyze the cytotoxicity of a compound

The assay was done like described for the EC_50_ assay, except no virus was added to the plate and it was replaced by adding the same volume of H-Hanks-0.6% FBS solution (Methods S2). The plates were analyzed with AlamarBlue™ HS cell viability reagent and the raw data was analyzed with python scripts as well as GraphPad Prism version 10.3.1 for Windows (GraphPad Software, www.graphpad.com). Additionally, a TCID_50_ control plate with 16,000 cells per well and 1:1,000 virus sample was prepared.

Quantification and statistical analysis

TCID50 assay data analysis with python scripts

To analyze the raw data from the TCID_50_ endpoint dilution assay, the data was stored in .xls or .xlsx format to data_files folder. The master file where the TCID50/ml results are stored was updated to contain label and tissue type for each sample. The script (EPDScript) was started in Windows PowerShell and user inputs were provided as prompted by the script. First, master file, data file and data sheet containing the raw data was selected from a list provided by the script. Next, the data was defined by selecting the tissue type (virus, lungs, small intestine or pancreas), dilution rate (6 i.e. 6-fold dilution), first dilution (1, 20, 1,000 or 10,000), working volume (0.025 i.e. volume before cells as milliliters), sample count (1 or 2), and number of replicates (4 or 8). The definitions for the data varied depending on the sample (see Table 1). The script retrieved the raw data, arranged it and calculated the TCID_50_/ml values based on improved Kärber’s formula (Equation 1), averages and standard deviations before storing the results to the given raw data file as a new sheet and additionally, the TCID_50_/ml values to the master file.

DORA data and statistical analysis with python scripts

The data analysis for IC_50_, EC_50_, and CC_50_ assays was done utilizing three different python scripts: first, DORAScript was used to arrange and normalize the raw data, second, 4PLScript was used to fit the data through four-parametric logistic (4PL) regression model, and third, STATScript was used to perform statistical analysis.

The raw data files were stored in .xlsx format to data_files folder. The first script (DORAScript) was started in Windows PowerShell and user inputs were provided as prompted by the script. First, the data file and the data sheet containing the raw data were selected from a list provided by the script. Next, the data was defined by selecting the analysis type (IC_50_, EC_50_, or CC_50_), defining group number (IC_50_: 1-5, EC_50_: EC50, CC_50_: CC50), initial dilution (IC_50_: 16, EC_50_: 5000, CC_50_: 5000), dilution factor (IC_50_: 2, EC_50_ and CC_50_: -2 i.e. 2-fold dilution), number of samples per group (IC_50_: 6, EC_50_: 1, CC_50_: 1), and number of parallels (IC_50_: 2, EC_50_: 4, CC_50_: 4). The script retrieved the data and arranged it based on the user inputs to groups, analyses, and samples. Additionally, the script calculated averages, and standard deviations and normalized the data between 0% (wells containing virus; no viable cells) and 100% (wells containing cells; viable cells) limits calculated from the control wells.

The 4PL fitting was done with a second script (4PLScript) and it utilized the normalized values and 10-logaritmic dilution values generated by the first script. The script was started in Windows PowerShell and user inputs were provided as prompted by the script. First, the data file and the data sheet containing the raw data were selected from a list provided by the script. Next, the data to be analyzed was selected by first selecting the groups for analysis (IC_50_: 1-5, EC_50_: EC50, CC_50_: CC50) and then the analysis type (IC_50_, EC_50_, or CC_50_). The script generated the IC_50_/ EC_50_/CC_50_, 10-base logarithm values and HillSlope values for each sample and stored them into the raw data file provided. Additionally, script generated image files containing graphical illustration of each group with data points for each sample and the fitted curve (Equation 2).

Statistical analysis was done with the third script (STATScript) and it utilized the IC_50_ values generated by the second script. The script was started in Windows PowerShell and user inputs were provided as prompted by the script. First, the data file and the data sheet containing the raw data were selected from a list provided by the script. Next, the statistical test was chosen (IC_50_: Kruskal-Wallis). The script retrieved the data and ran the Kruskal-Wallis t-test followed by Dunn’s Multiple Comparison test with Bonferroni correction and generated p-values for the test and for the multiple comparisons with statistical significance indicated with symbols (ns: p > 0.05; ∗: p ≤ 0.05; ∗∗: p ≤ 0.01; ∗∗∗: p ≤ 0.001).

DORA data and statistical analysis with Prism

To compare data analysis performed with python scripts, the same was done with GraphPad Prism 10.3.1. for Windows (GraphPad Software, www.graphpad.com). First, the raw data was arranged with the DORAScript like described in python script analysis. The non-normalized, transposed values were transferred to GraphPad Prism. The data was normalized between the 0% and 100% values calculated from the control wells by the python script. In the normalized data, 4PL curve fitting was done using “log(inhibitor) vs. normalized response – Variable slope” model for IC_50_ and CC_50_ data and “log(agonist) vs. normalized response – Variable slope” model for EC_50_ data to determine the inhibitory, cytotoxic, or effective concentration for each sample. The 50% titer was defined by the midpoint of the sample-specific curve. The geometric means of the titers (with geometric standard deviation) were plotted on the graphs presenting the neutralizing antibody or compound responses.

For the statistical analysis, the data for neutralizing antibody responses was treated as non-parametric. The comparison between multiple groups in IC_50_ assay was done using Kruskal-Wallis t-test followed by Dunn’s Multiple Comparison test with multiplicity adjusted p-value (ns: p > 0.05; ∗: p ≤ 0.05; ∗∗: p ≤ 0.01; ∗∗∗: p ≤ 0.001).