J774.1 cells (ATCC; TIB-67) were cultured in DMEM medium supplemented with 100 U/ml penicillin, 100 µg/ml streptomycin, 0.3 mg/ml L-glutamine (Gibco), and 10% FBS (Gibco). A10 cells were purchased at ATCC (CRL-1476) and were cultured in DMEM medium supplemented with 1 mM pyruvate (Gibco), 100 U/ml penicillin, 100 µg/ml streptomycin, 0.3 mg/ml L-glutamine (Gibco) and 10% FBS (Gibco). Vero cells were obtained from WHO (10–87) (originally derived from ATCC (CCL-81)) and were cultured in VP-SFM medium (Gibco), supplemented with 2 mM L-Glutamine (Gibco). THP-1 cells used in Figures 1 , 2 , 3A–C , 4A , Supplementary Figures S1 , and S2 were obtained from an in-house stock. To confirm results obtained with this stock, subsequent experiments were performed with THP-1 cells from a stock derived from a vial purchased at ATCC (TIB-202) ( Figures 3D , 4B–D , 5 , 6 , Supplementary Figures S3 – S5 ). THP-1 cells were maintained in RPMI medium (Gibco), supplemented with 50 µM β-mercaptoethanol, 100 U/ml penicillin, 100 µg/ml streptomycin, 0.3 mg/ml L-glutamine (Gibco) and 10% FBS (Gibco). All cell lines were cultivated at 37°C in a humidified atmosphere of 5% CO₂.

THP-1 cells were seeded at 25000 cells/well in a 96-well flat bottom plates (Greiner Bio One) and differentiated with 300 ng/mL PMA (Sigma), unless mentioned otherwise. A10 cells, J774.1 and Vero cells were seeded at 100000, 25000 and 25000 cells/well in 96-well flat bottom plates. For all cell types, viable cells were counted following staining with a 0.4% trypan blue solution and only cell populations with a minimal viability of 70% were used for seeding. After seeding, cells were incubated over night at 37°C in a humidified 5% CO₂ incubator. Subsequently, the cells were exposed to various dilutions of the Cl. perfringens type C non-inactivated antigen preparation, containing active β-toxin (kindly provided by a pharmaceutical company that was part of the VAC2VAC consortium (http://www.vac2vac.eu), hereafter referred to as company B), medium or Triton-X-100 (Merck) for 16-24 hours. Cell viability was verified either by staining with 1) Crystal Violet (Sigma) - which stains living cells; 2) by addition of MTS (MTS assay, Promega) - a substance (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) that is converted into colored formazan by living cells or 3) the LDH assay (Promega) performed on the culture supernatant, measuring the conversion of tetrazolium salt (iodonitrotetrazolium violet) into a red formazan product by cellular lactate dehydrogenase (LDH) released into the supernatant upon cell death. The MTS and LDH assay were performed according to the manufacturer’s protocol and the quantity of formazan product as measured after 3.5-4.5 hours for the MTS assay and 30 minutes for the LDH assay at 490 nm. Crystal Violet (Sigma) staining was also performed according to manufacturer’s protocol and was measured at 562 nm.

Cl. perfringens type C non-inactivated antigen samples were prepared in medium containing 10% FCS (unless mentioned otherwise) at the indicated dilutions within one hour before addition to the cells. To determine the specificity for β-toxin, the Cl. perfringens type C non-inactivated antigen samples were pre-incubated for 30 min incubation at 2-8°C on an orbital shaker (250 rpm) with dilutions of a neutralizing monoclonal antibody (clone 10A2 purchased from the USDA; concentration unknown, produced either in mice (referred to as ‘Ascites’) or in a bioreactor (referred to as ‘Bioreactor’)) or an international antitoxin standard (2CPBETAAT from NIBSC; 4770 IU of antitoxin, reconstituted in 1 mL PBS, referred to as ‘Antitoxin’).

Cl. perfringens type C toxoid (kindly provided by company B) samples were prepared in medium containing 0.5% FCS at the indicated dilutions within one hour before addition to the cells. To neutralize the effect of residual formaldehyde (CH₂O) present in the inactivated toxoid preparation, the undiluted β-toxoid and toxin preparations were incubated with 25% acetoacetamide (C₄H₇NO₂) for 30 minutes at room temperature or dialyzed against PBS overnight at 2-8°C. 0.5 g/L formaldehyde served as a control. If indicated, dialyzed samples were passed through a 0.22 µm filter before being added to the cells. For the spiking experiment, the undiluted toxoid preparation was treated with a 1/54 dilution of the Cl. perfringens type C non-inactivated antigen or left untreated. The toxoid or toxin samples were prepared on the day of the experiment or incubated overnight as indicated.

As the dose-response curves of β-toxin dilutions (x-axis) versus the OD at 490 nm (y-axis) follow a symmetrical sigmoidal shape, 4-parameter curve fitting (GraphPad Prism version 10.1.2) was used to mathematically describe the curves that were based on triplicate measurements ( Figures 3D , Supplementary Figure S5D ) and the curves shown in Figure 5 (duplicate measurements and one single measurement). The linear part of Figure 5 was analyzed using simple linear regression on the average values of duplicate measurements and one single measurement. The results are shown in Supplementary Figure S4 .

Inspired by the in vitro assay based on the Vero cell line in combination with Crystal Violet staining, which proved to be suitable for assessment of toxicity of Cl. septicum vaccines and intermediate products (16), we aimed at using this strategy as the readout for the current Cl. perfringens type C toxicity test, as well. From a wider range of cell lines that was evaluated elsewhere (7, 10, 18), we selected THP-1 cells as one of our candidate cell lines, along with the β-toxin sensitive cell line A10 and the P2X₇ expressing murine J774A.1 cell line (19). Vero cells were included as a control. Both THP-1 and A10 cells ( Figures 1A, B ) were sensitive to Cl. perfringens type C non-inactivated antigen in a concentration-dependent manner, though the measured ODs were relatively low. The measured ODs for the Crystal Violet staining of J774A.1 ( Figure 1C ) were also very low and not concentration-dependent. Whereas the ODs for the Crystal Violet staining were higher for Vero cells ( Figure 1D ), toxicity was only observed for a 100-fold dilution of the Cl. perfringens type C non-inactivated antigen preparation or less diluted, suggesting this cell line has a lower sensitivity compared to the THP-1 and A10 cell lines. THP-1 cells and A10 cells were therefore selected for subsequent experiments.

In Figure 1 , the remaining cells after Cl. perfringens type C non-inactivated antigen exposure were visualized by staining with Crystal Violet as a readout for viability, like the Vero-cell based test for veterinary Cl. septicum vaccines. However, since the measured ODs were relatively low the MTS and the LDH assays were evaluated as alternative methods to estimate cell viability. The MTS assay measures the conversion of the MTS tetrazolium compound into formazan by living cells. The assay is therefore directly proportional to the number of living cells. The LDH assay measures the activity of lactate dehydrogenase, a stable enzyme that is released upon cell lysis. Released LDH converts the tetrazolium salt into formazan and is therefore proportional to the number of dying cells. Exposure of THP-1 cells to Cl. perfringens type C non-inactivated antigen in a dilution range between 25 to 3.9x10⁵ fold resulted in a concentration dependent increase of the MTS signal ( Figure 2A ) and decrease of the LDH signal ( Figure 2B ). The range of the ODs was acceptable for both readout parameters and substantially higher than the Crystal Violet readout ( Figure 1A ) with the dynamic range (highest OD/lowest OD) being 12-fold for MTS and 3.2-fold for LDH ( Figures 2A, B ). Exposure of THP-1 cells to a one- or five-fold dilution of the Cl. perfringens type C non-inactivated antigen preparation resulted in a higher signal of the MTS assay as compared to the 25-fold dilution ( Figure 2A ), which most likely is the result of a direct reaction between the Cl. perfringens type C non-inactivated antigen preparation and the MTS reagent ( Figure 2D ). Remarkably, exposure of THP-1 cells to the one- and five-fold dilutions of the Cl. perfringens type C non-inactivated antigen preparation was associated with a lower LDH signal as compared to the 25-fold dilution, which cannot be attributed to the Cl. perfringens culture medium or a reaction of the preparation with the LDH reagent ( Figure 2E ), but nonetheless should be taken into account. More importantly, the absence of an effect of the Cl. perfringens culture medium (5-3.9x10⁵ fold dilution) on THP-1 viability indicates that the effect of the Cl. perfringens type C non-inactivated antigen can be specifically attributed to the β-toxin or other (toxic) components in the preparation. Exposure of A10 cells to the Cl. perfringens type C non-inactivated antigen preparation in the range of 10-3.9x10⁵ fold dilution resulted in a very slight concentration dependent increase in MTS signal with a dynamic range of 1.5-fold ( Supplementary Figure S1A ). The level of LDH released by A10 cells was only affected by a 10-fold dilution of the non-inactivated antigen preparation to the point of total cell death (Triton-X-100 control), while all other dilutions (50-3.9x10⁵) resulted in the release of LDH at a level of the medium control ( Supplementary Figure S1B ). In contrast the Crystal Violet staining does suggest a toxin concentration dependent effect with a dynamic range of 3.8 ( Figure 2C ), however the effect is not specific since the Cl. perfringens type C culture medium has a similar effect ( Figure 2F ). Consequently, the THP-1 cells in combination with the MTS assay were selected as most promising for further assay development.

To further determine whether the toxicity can be specifically attributed to β-toxin activity, experiments were performed where THP-1 cells were exposed to serially diluted Cl. perfringens type C non-inactivated antigen in the presence of serial dilutions of three different sources of β-toxin specific antibody preparations (ascites-derived and bioreactor-derived monoclonal antibody 10A2, and an international anti-β-toxin standard (2CPBETAAT)). All antibody preparations neutralized the effect of the Cl. perfringens type C non-inactivated antigen preparation, which is reflected in an increase in THP-1 viability ( Figures 3A-D ). The strength of neutralization differed between the three antibody preparations. Whereas mAb 10A2 from ascites had the highest neutralizing capacity, the international antitoxin clearly had the lowest neutralizing capability ( Supplementary Figure S2A , Figure 3D ). Furthermore, cell viability was not affected by any of the included dilutions of mAb 10A2 (both from ascites or bioreactor sources), while the international antitoxin at a one- and ten-fold dilution did negatively affect cell viability ( Supplementary Figure S2B ). Taken together, the results clearly show that the toxicity on THP-1 cells induced by the Cl. perfringens type C non-inactivated antigen can be specifically attributed to the β-toxin. Therefore, effects will subsequently be indicated as being β-toxin dependent.

The experimental setup thus far consisted of 25000 cells per well, 300 ng/ml PMA and 10% FCS. Initial optimization experiments with and without PMA ( Figure 4A ) and with different concentrations of PMA (data not shown), confirmed the necessity of PMA-differentiation of the THP-1 cells for the assay and demonstrated that lower PMA concentrations (75 ng/mL) are associated with a higher maximum OD at 490 nm. Lowering the PMA concentration down to 12.5 ng/mL further increased the OD and the dynamic range to 21-fold ( Figure 4B ). Using a PMA concentration of 12.5 ng/mL, the OD at 490 nm also increased when the number of cells was increased up to 50000 and 100000 cells per well ( Figure 4C ), though the latter cell densities did change the shape of the curve. In a subsequent experiment, a cell density of 50000 and 75000 cell per well (in combination with a PMA concentration of 12,5 ng/ml and 10% FCS, Supplementary Figure S3 ) was examined and a density of 75000 cells per well resulted in the highest OD, with a 11-fold dynamic range ( Supplementary Figure S3B ). After determination of the optimal cell density (75000) and PMA concentration (12.5 ng/ml), several concentrations of FCS were compared to the 10% used thus far. Lowering the FCS concentrations only had a slight effect ( Figure 4D ), 0% FCS being associated with the highest OD but also with a higher background. 0% FCS also changed the shape of the curve and slightly decreased dynamic range to 9,1-fold ( Supplementary Figure S3B ). Based on these results and personal communication with company B, 0.5% was chosen as the most optimal FCS concentration and used for sample preparation in subsequent experiments, while cell culturing was still performed in 10% FCS.

Repeatability is an important aspect of toxicity tests and quality control tests in general and was evaluated here based on five individual experiments in which very similar dilution ranges of β-toxin were used ( Figure 5 ). There is only a very small variability between the five experiments regarding the absolute maximal OD at 490 nm. The fold change between total toxicity (lowest OD) and no toxicity (highest OD) for all experiments was between 12 and 16 fold and the lowest dilution still affecting cell viability, indicating the assay’s sensitivity, was around 10000-fold dilution of the toxin in this R&D setting. Analysis of the data of the five experiments shown in Figure 5 revealed that all adequately fit a four parameter logistic curve with similar r² (0.98-0.99) and a slight range in EC₅₀ (2163-3843 fold dilution) ( Supplementary Figure S4A ). The results were confirmed by fitting the data of Figure 3D and Supplementary Figure S5D (EC₅₀ of 7086 and 1343 and r² of 0.99 and 0.99, respectively) ( Supplementary Figures S4B, C ). Simple linear regression analysis was performed on a part of the curve (875-6650 fold dilution) and revealed that the r² was 0.87 ( Supplementary Figure S4D ). Whereas sensitivity of the assay was sufficient in all of these experiments, the range in EC₅₀ highlights the needed for dedicated experiments to examine the linear range of the curve in more detail. These experiments should be a part of a (pre-)validation study.

The assay developed so far allows detection of Cl. perfringens type C β-toxin activity in a specific and sensitive manner. In addition to the evaluation of β-toxin activity, detection of β-toxin in the context of a toxoided antigen preparation is of key importance because it would allow the assay to be used for non-toxicity testing of inactivated Cl. perfringens type C vaccine intermediate and end-products. We therefore determined the compatibility of the assay with a toxoided Cl. perfringens type C antigen preparation ( Figures 6A-C ) and its ability to detect β-toxin in the presence of this toxoid preparation by spiking with a fixed concentration of toxin (1/54 dilution) ( Figure 6D ). One of the obstacles is the residual free formaldehyde in commercial toxoid preparations. The toxoid itself negatively affects cell viability and a 220x dilution is required to negate this effect ( Figure 6A ), an effect that can primarily be attributed to residual formaldehyde present in the toxoid preparation ( Supplementary Figure S5A ). Various substances were tested for their ability to neutralize this effect (NaHSO₃, Bovine Serum Albumin, Tris, glycine), but only acetoacetamide and dialysis ( Figures 6A, B ) against PBS effectively neutralized the formaldehyde effect. Acetoacetamide has a slight effect on cell viability ( Figure 6A ), but a one in six dilution is sufficient to negate this effect. Dialysis increased the sample volume from 1 to 1.5 mL, which was reflected in the protein content that decreased from 1407 µg/ml to 1092 µg/ml, while filtration had a negligible effect (1086 µg/ml), suggesting a minimal loss of material. This was confirmed by spiking toxoid with a 1/54 dilution of the non-inactivated antigen. Without any treatment, a pilot experiment demonstrated that the effect of the toxoid with a 1/54 dilution of the non-inactivated antigen is similar to the effect of the β-toxin itself and a discrimination between the negative effect of the toxoid and β-toxin is therefore impossible ( Figure 6C ). However, dialyzed spiked toxoid preparations (with and without filtration) caused cell death ( Figure 6D ), though to a smaller extent than the toxin itself. This is most likely not ue to a loss in β-toxin activity overnight, as storage for one day had no effect on the activity of the toxin ( Supplementary Figure S5D ). More importantly, the experiment demonstrates that dialysis of the toxoid is a treatment that allows for the detection of β-toxin in the context of a β-toxoid preparation.