All sera used in this study came from: i) other studies previously approved by the Ethics Committee for the Use of Animals in Research at the Universidade de Passo Fundo (CEUA no. 10/2020, 19/2020 & 20/2020); ii) routine of diagnosis of the AFK Imunotech Laboratory; iii) other studies previously approved by the Ethics Committee for the Use of Animals in Research at the Universidade Federal de Minas Gerais (CEUA n° 133/2018, 36/2016).

The in vitro cultivation of L. intracellularis is extremely difficult, which limits the production of this bacterium to few laboratories in the world. Since our objective was to develop a serological assay that could be easily performed in any laboratory, we used the vaccine strain of L. intracellularis from Enterisol^® Ileitis (Boehringer Ingelheim), a live attenuated vaccine, as a source of antigen. This licensed vaccine is distributed worldwide, and all new batches of live antigen are qualified by the manufacturer, an ideal situation to guarantee antigen quality and ensure repeatability between antigens batches and laboratories that will perform this diagnosis.

A total of 246 serum samples obtained from pigs with different serological condition were used to determine the preliminary cut-off values of this flow cytometry antibody test ( Table 1 , Phase I). The second set of specimens included 1,200 serum samples collected between 2018 and 2021 from 32 farms located in Southern region of Brazil (Paraná, Santa Catarina and Rio Grande do Sul) ( Table 1 , Phase II); this panel of sera was used to analyze the ability of the established cut-off to identify conventional pigs with or without anti-L. intracellularis IgGs.

The stability of the reconstituted antigen was evaluated daily for a period of 30 days. Three parameters were analyzed by flow cytometry: i) absolute count of L. intracellularis; ii) bacterial morphology; and iii) antigenicity. For the first two analyses, a daily aliquot of the antigen was diluted 1:1,000 in filtered PBS pH 7.4 and acquired in the cytometer. The number of bacteria per microliter of acquired samples was calculated automatically by the cytometer which was equipped with flow sensor used for volumetric measurement. Therefore, to calculate the total number of bacteria per mL of the reconstituted vaccine, we used the following mathematical formula: number of events per μL × inverse of the dilution factor (1,000) × per 1,000 (to convert μL to mL). The morphometric analysis was conducted analyzing the parameters of Forward Scatter (FSC) vs Side Scatter (SSC) of the daily aliquot of the antigen. For the antigenicity analysis, 3 sera from pigs immunized with the Porcilis^® Ileitis vaccine were used; briefly, 10⁶ bacteria were incubated into the wells of a 96 well conical-bottom polystyrene plates (cat. # K30-6096V, Olen, China) together with 100 µL of porcine serum diluted 1:100 in filtered (0,22 µm, cat. # GPWP04700, Millipore, Ireland) PBS pH 7.4 containing 1% bovine serum albumin (cat. # A2153, Sigma-Aldrich, USA) (FACS buffer) for 1 hour at 37°C. After three washing steps with 200 μL of FACS buffer (plate centrifuged at 1,200 × g for 5 min; centrifuge 5810R Eppendorf, Germany), 100 μL of FACS buffer containing 1 μg of Goat anti-Porcine IgG(H+L)-PE (cat. # 6050-09, SouthernBiotech, USA) was added and incubated for 1 hour at 37°C. Then, the washing steps were repeated, and the bacterial pellet resuspended in 350 µL of PBS for analysis. All parameters were analyzed by Flow Cytometry using a FACSVerse Cytometer (Becton Dickinson, USA) equipped with a 405 nm violet laser, 488 nm blue laser, 640 nm red laser and flow sensor. A total of 100,000 events were acquired in P1 region and analyzed. Each sample was analyzed in triplicate.

To determine the optimal amount of antigen that would be used in the assay, three different concentrations (10⁵, 10⁶ and 10⁷) of L. intracellularis were evaluated. Each concentration was incubated with a panel of sera (n=5) with different titers of anti-L. intracellularis IgG as defined previously by the IPMA ( Table 1 ): i) highly positive sera (titer between 1:960 and 1:1,920); ii) moderately positive sera (titer between 1:120 and 1:480); and iii) weakly positive sera (titer of 1:30). The immunostaining was performed as described above (subsection 2.3) and the ideal bacteria concentration was defined as the smallest number of bacteria that: i) allowed the statistical differentiation of three sera categories and ii) ensured enough bacteria to be analyzed after the washing steps.

To define the optimal dilution of the primary antibody (porcine sera) different dilutions of a panel of highly positive (n=5, titer > 960), moderately positive (n=5, titer between 1:120 and 1:480, weakly positive (n=5, titer of 1:30) and negative (n=5) sera were analyzed. The dilutions evaluated were 1:50, 1:100 and 1:200 in a final volume of 100 µL. The immunostaining was performed as described above (subsection 2.3).

To investigate the shortest incubation period necessary to guarantee the optimal interaction between the molecules involved in the assay, three incubation periods were evaluated: 60 min, 30 min and 20 min. These periods were evaluated for primary (10 different positive samples) and secondary (Goat Anti-Porcine IgG (H+L)-PE) antibodies. All incubations were performed at 37°C. The immunostaining was performed as described above (subsection 2.3).

A total of 100 samples ( Table 1 ) were analyzed for the presence of anti-L. intracellularis antibodies using a commercial blocking ELISA Kit (SVANOVIR^®, Boehringer Ingelheim Svanova, Sweden). The protocol was performed following the manufacturer’s recommendations.

The immunoperoxidase monolayer assay (IPMA) was used to detect porcine anti-L. intracellularis IgG. This assay was carried out by the laboratory of Prof. Roberto Guedes (Universidade Federal de Minas Gerais), according to the methodology previously described (20).

A total of 10⁶ L. intracellularis suspended in 200 µL was added per well of a 96 well conical bottom polystyrene plates. The plate was centrifuged (3,200 × g, 5 min) and the bacterial pellet was resuspended in 100 µL of porcine serum diluted 1:100 in FACS buffer and incubated at 37°C during 20 min. After three washing steps with 200 μL of FACS buffer (step of centrifugation at 1,200 × g for 5 min between washes), 100 μL of FACS buffer containing 1 μg of Goat Anti-Porcine IgG(H+L)-PE was added and incubated for 20 min at 37°C. Then, the bacteria were washed 3 times, resuspended in 350 µL of FACS buffer and transferred into a round-bottom tube (Falcon^®, cat. # 352052). A total of 100,000 events were acquired in a FACSVerse cytometer. The bacterial population was initially identified through a dot plot crossing the FSC and SSC parameters; where the study region, designated P1, was established. Subsequently, a second dot plot was created, in this case, crossing the FL-2 (Phycoerythrin, PE) and FL-4 (Allophycocyanin, APC) channels and having P1 as the study region. The events considered positive were those enclosed within the quadrant representative of PE (low right). The results are expressed as the percentage of positive bacteria in relation to the total bacteria contained in the P1 region. The step-by-step execution of this diagnosis as well as the analysis strategy are illustrated in Figure 1 .

The determination of the cut-off was performed using a panel of 198 porcine sera ( Table 1 ). Briefly, sera from 78 animals known to be negative for anti-Lawsonia intracellularis IgG and molecularly (qPCR) negative for L. intracellularis were obtained. We also obtained sera from 40 animals from a genetic nucleus herd with no history of Ileitis; these animals were molecularly negative for L. intracellularis; collectively sera from a total of 118 negative animals. We used sera from 80 animals serologically positive for L. intracellularis; of these total, 60 SPF pigs were vaccinated (qPCR negative), and 20 were experimentally infected (and consequently qPCR-positive). Therefore, the cut-off calculation, performed using the Receiver Operating Characteristic (ROC) Curve, included 118 negative and 80 positive animals. All sera were tested by this new assay and the results were used to generate a Receiver Operating Characteristic (ROC) Curve. The selected cut-off was based on the sensitivity and specificity values. The ROC Curve analysis was performed using Graph Pad Prism Software Version 9.2.0.

The specificity analysis was conducted to evaluate the ability of this assay to identify IgG specifically to L. intracellularis, which is the analyte of interest. The analysis was performed as recommended by Selliah, Eck (22). Briefly, a panel of 48 sera ( Table 1 , Specificity) from SPF pigs immunized with Glaesserella parasuis (bacterin prepared with SV7 (21), n=8), Mycoplasma hyopneumoniae (Safesui Mycoplasma, Ourofino, n=8), Pasteurella multocida A (Bacterin formulated with a Brazilian clinical strain of P. multocida A, pfhA ⁺, Vaxxinova, n=8), Salmonella enterica, serovar Choleraesuis (Enterisol^® SC-54 vaccine, Boehringer Ingelheim, n=8), Clostridium perfringes type C and Escherichia coli (Porcilis^® Coliclos, MSD, n=8) and Porcine circovirus type 2 (Porcilis^® PCV ID, MSD, n=8) were assessed using the protocol described in subsection 2.8.

Precision is one of the most critical parameters in flow cytometry assay (22). In order to know the intra-assay precision (precision), samples from 3 positive pigs (vaccinated with Porcilis^® Ileitis) and negative SPF pigs were tested twice under the same conditions by a single analyst. With the values obtained from each sample, the mean and coefficient of variation were calculated. For reproducibility analysis (inter-assay precision) the same samples were analyzed by 2 different analysts on 2 different days. Again, with the values obtained from each sample, the mean and the coefficient of variation were calculated. The accuracy acceptance criterion was established by accepting a coefficient of variation between the analyses of the same analyst or between the analysts of a maximum of 25%.

To analyze the level of agreement between the FCAT and IPMA, we used a panel of sera from two experimental groups: G1) pigs (n = 8) immunized with the Porcilis^® Ileitis vaccine at 21 days of life and G2) pigs (n = 8) inoculated with PBS. Serum samples were collected prior to vaccination (D0) and at the following post-vaccination times: D7, D14, D21, D28 and D55 contributing to 88 sera for comparative purposes. Agreement analysis (Kappa) was performed as described by Landis and Koch (23).

All data were analyzed using GraphPad Prism™ (GraphPad Software, San Diego, California, USA). One or two-way ANOVA with Tukey’s multiple comparisons test were used to assess significance between the different variables analyzed in this study. The specific test used in each analysis, as well as the significance, is indicated in the figure legends.

L. intracellularis obtained from the attenuated vaccine Enterisol^® Ileitis was used as a diagnostic antigen. The antigen used in our study is destined to immunizing pigs and should be immediately used upon reconstitution. In the test designed here, the amount of antigen used in each assay is low and a single reconstituted bottle contains enough antigen to be used along a few weeks. Thus, the stability and quality of the reconstituted and refrigerated (4 – 8°C) antigen was analyzed daily along a 30 days period. During this time, changes in the morphology of the bacterial population (analyzed in the P1 region) were observed after 15 days of antigen storage ( Figure 2B ); a subpopulation of larger and more complex L. intracellularis was clearly detected, which may represent loss of bacteria cell wall integrity. Although we observed morphological changes in the bacterial population, the number of bacteria detected at each time point was similar with no significant reduction at any time ( Figure 2A ). The antigenicity analysis revealed that antigen stability follows the same trend as bacteria reconstituted for more than two weeks were significantly more antigenic compared to previous time points ( Figure 3 ), which indicated that storage time alters the antigenic characteristics of the reconstituted antigen; thus, for all assays the reconstituted antigen was used for 15 days only.

The amount of antigen used in a diagnostic test must be evaluated with two objectives: i) functionally (to ensure maximum resolution of the test) and ii) economically (antigen saving). Taking these premises into account, we analyzed 3 concentrations of antigens (10⁵, 10⁶, and 10⁷) immunostained with 3 categories of sera, weakly, moderately, and highly positive as determined by IPMA. As illustrated in Figure 4 , regardless of the bacterial concentration it is possible to differentiate the 3 serum categories, except for the concentration of 10⁷, in which weakly and moderately positive sera could not be statistically differentiated. There is a reduction in the percentage of detection of L. intracellularis between concentrations of 10⁶ and 10⁷, when using weakly and moderately positive sera ( Figure 4 ). Additionally, we noticed that the number of bacteria acquired from immunostaining using 10⁵ bacteria was lower (long acquisition period) compared to 10⁶ and 10⁷, indicating that part of the bacteria was lost during the washing steps; therefore, the concentration of 10⁶ was defined as the ideal concentration of antigen.

The dilution rate of the serum (primary antibody) is essential in any diagnostic test and should be set to mitigate false positive results, usually associated with polyreactive IgGs. Three serum dilutions were analyzed; 1:50, 1:100 and 1:200. We observed that there was no reduction in the percentage of L. intracellularis associated to porcine IgGs between the 1:50 and 1:100 dilutions, regardless of the serum category. On the other hand, the percentage of L. intracellularis with associated IgGs was significantly lower (p < 0.0001) when sera diluted 1:200 were used, and the number of sera considered positive (cut-off set at 15.15% as described in the section 3.5) at lower dilutions (5/5) decreased when the weakly positive sera category was tested (2/5) ( Figure 5 ). Therefore, the dilution of the primary serum was set at 1:100.

Evaluating antibody incubation time (primary and secondary) is essential in any diagnostic test under development; in general, it is always recommended to select the shortest time, as long as there are no changes in the final result of the diagnosis. Three incubation times (20, 30 and 60 min) for both the primary (pig IgG anti-L. intracellularis) and secondary (Goat anti-Porcine IgG-PE) antibody were tested. As illustrated in Figure 6 , the mean percentage of L. intracellularis with associated IgGs at the 3 different incubation times was 47.5%, 47.1% and 48.0%; additionally, the same dispersion profile of the results was observed, and no significant differences were found between the analyzed times. In view of these results, an incubation time of 20 min was selected for the primary and secondary antibodies.

Once the definition of all parameters and conditions of the Flow Cytometry Antibody Test (FCAT) was concluded, a wide panel of swine sera was used ( Table 1 , Phase I) aiming to find the cut-off value capable of truly differentiating a positive from a negative serum sample. For this purpose, a ROC curve analysis was conducted using 118 serum samples negative and 80 serum samples positive to L. intracellularis antibodies ( Figure 7A ). The ROC curve analysis of the FCAT data generated paired estimates of relative sensitivity and relative specificity at different cut-off values. A cut-off of 15.15% (percentage of L. intracellularis with associated IgG) was recommended; and at this cut-off value, the relative sensitivity and specificity estimates were 98.8% [95% confidence interval (CI) = 93.2% to 99.9%] and 100% (95% CI = 96.9% to 100%), respectively. The ROC curve had an Area Under the Curve (AUC) value of 0.9963 (95% CI = 0.99 to 1.0), which indicated a high level of accuracy for this FCAT ( Figure 7B ). Thereafter, using the established cut-off value, a second panel consisting of 48 sera ( Table 1 , Specificity) was tested on the FCAT. As illustrated in Figure 7C , none of the sera reached the cut-off of 15.15%; therefore, all sera were considered negative for L. intracellullaris. This result demonstrates that there is no cross-reactivity between L. intracellularis and G. parasuis, M. hyopneumoniae, P. multocida A, Salmonella enterica, serovar Choleraesuis, E. coli, C. perfringes type C and Porcine circovirus type 2. Finally, a third panel of sera (n=1,200, Table 1 , Phase II) from: i) non-vaccinated conventional pigs [L. intracellularis (qPCR) negative feces]; ii) non-vaccinated conventional pigs [feces positive for L. intracellularis (qPCR)]; and iii) Vaccinated (Porcilis^® Ileitis, MSD) conventional pigs, was evaluated in the FCAT. As shown in Figure 7D , all animals vaccinated with the Porcilis^® Ileitis vaccine developed antibodies against L. intracellularis and were therefore considered positive. Similarly, 85.3% of the animals that were shedding L. intracellularis in the feces had anti-L. intracellularis IgG at a level higher than the FCAT cut-off; and 68.7% of the animals that were not shedding L. intracellularis at the time of blood collection had previous contact with L. intracellularis.

The precision (intra-assay) and reproducibility (inter-assay) analysis were performed by two different analysts (A and B). As described in the Table 2 , the coefficient of variation (CV) of the samples analyzed in triplicate by each of the analysts was always less than 10%, indicating that analysts performed the immunostaining and acquisition of samples with precision. Furthermore, when the same samples were repeated by the same analyst on the following day, a CV of less than 10% was again observed. It is important to highlight that the results of immunostaining performed by the two analysts varied little, with the smallest variation being 0.9% and the maximum 13%; these data indicate that FCAT has high accuracy.

To analyze the level of agreement between FCAT and IPMA, sera collected from two groups of pigs (vaccinated and unvaccinated) at different post-vaccination periods were analyzed by these two methodologies in two different laboratories (FCTA conducted at the Frandoloso’s lab and IPMA conducted at Guedes’s lab). To avoid any bias, the analysis was blind, and the history of the sera was not revealed to the analysts. As described in Table 3 , out of the total number of samples tested (n=88), 74 had the same result in both methodologies, which represents a direct agreement of 84.09%. Using all the data (samples with the same and different results according to the technique used) the Kappa index was generated resulting in a general agreement of 0.66 ± 0.08 (95% CI = 0.50 to 0.82) ( Table 4 ), which represent a substantial agreement taken into consideration the Kappa scale (0 to 1).

Considering that the pigs used in this experiment were Specific Pathogen Free and kept in a controlled facility (BSL-2) throughout the study, our expectation was that all animals, regardless of the diagnostic technique, would be negative for L. intracellularis. As illustrated in Table 3 , we observed that a considerable number of pigs from group G2 (non-vaccinated) were positive for anti-L. intracellularis IgG as determined by the IPMA day (D) 7 (n=2, 25%), D14 (n=4, 50%) and D21 (n=1, 12.5%) post inoculation with PBS ( Table 3 ). We understand that these IPMA results are false positives, and this interpretation is supported by two observations: i) samples collected from the same pigs at days D35 and D55 were negative by IPMA ( Table 3 ); ii) all sera samples from this group were negative for anti-L. intracellularis IgG by FCAT.

The sensitivity to specifically detect the analytical, in our case the anti-L. intracellularis IgG during the genesis of humoral response induced by vaccination is one of the most important features. We noticed that both techniques are very similar in terms of sensitivity; however, FCAT standing out slightly over IPMA on D14, at that moment the FCAT classified more animals as positive than the IPMA.