Overall, 359 strains of C. jejuni isolated from 114 refrigerated chicken carcasses, slaughtered in establishments registered with the Federal Inspection Service by the Ministry of Agriculture, Livestock, and Supply (MAPA), were analyzed. The carcasses were analyzed from October 2017 to July 2018 to meet the Exploratory Program for the Research and Estimation of the Prevalence of Campylobacter spp., the same ones used for the official verification of the Salmonella spp. Control and Monitoring Program, determined by IN n° 20 of 21 October 2016 (MAPA, 2016). The collections of carcasses were carried out in establishments located in the states of Paraná (A), Santa Catarina (B), and Rio Grande do Sul (C), distributed across 43 municipalities (Rodrigues et al., 2021).

The strains were isolated and identified as C. jejuni in official MAPA laboratories using the methodology proposed by ISO 10272-1: 2017 (ISO, 2017) and mass spectrometry (MALDI TOF^®) for the identification of the genus and species, respectively. Additional information, such as establishment and place and date of isolation, was received together with the cryopreserved strains and deposited in the culture bank of the Molecular Epidemiology Laboratory of the Faculty of Veterinary Medicine of the Federal University of Uberlândia (LEPIMOL-FAMEV-UFU).

The samples were prepared according to the direct transfer protocol (Direct Transfer Method) for the plate from the isolated colony in three spots. After application to the plate, the sample was covered with 1 μl of alpha-cyano-4-hydroxycinnamic acid (HCCA) solution. Reference strains belonging to LANAGRO-RS were used as controls. The Brucker platform, model Autoflex Speed, was used to identify the isolates. The spectrum bank was the MALDI Biotyper RTC/OC 3.1. The species decision criteria are as follows: equal to or >1.7 for genus and equal to or >2.0 for genus and species. The data acquisition control program was Brucker FlexControl 3.4. Calibration was performed using the Protein Standard I calibrator, with a calibration error tolerance of <200 ppm.

After reactivation on Campylobacter Blood-Free Selective Agar (mCCDA) and incubation in a microaerobic atmosphere (5–15% O₂ and 10% CO₂) using a Microaerobac (Probac do Brasil, São Paulo, Brazil) at 37°C for 44 ± 4 h, followed by strain confirmation by oxidase, catalase, motility, and typical morphology tests (ISO, 2017), genomic DNA was extracted using the Wizard Genomic DNA Purification Kit (Promega, cat no. 0000110703, São Paulo, Brazil), following the protocol provided by the manufacturer. Purified DNA (10 ng) was used as a template for all PCR reactions. The PCR conditions and primers used in this study are described in Table 1.

Overall, 14 genes were studied and divided into five virulence categories: i) biofilm formation – flaA (motility) and luxS (quorum-sensing mechanism); ii) secretion systems – cdtABC (distensive cytotoxin secretion) and hcp (type VI secretion system); iii) invasion and colonization – cadF (intracellular colonization), ciaB (cell invasion), and pldA (invasion/colonization); iv) adaptation to stress – dnaJ (thermotolerance), htrA (aid in growth under stress), and cbrA (resistance to osmotic shock); and v) induction of Guillain-Barré Syndrome – cstII (GBS) and neuA (GBS).

PCR reactions were performed using the GoTaq^® Green Master Mix kit (Promega, cat no. 0000112705, São Paulo, Brazil), according to the manufacturer's instructions. As positive controls, strains of C. jejuni ATCC 33291, IAL 2383, and NCTC 11351 were used. The amplified products were subjected to electrophoresis in a 1.5% agarose gel, using the TBE 0.5× running buffer (Invitrogen) and, as a molecular weight standard, the 100 bp marker (Invitrogen).

A criterion adapted from the analysis of the antimicrobial multi-resistance index (Biswas et al., 2011) and the five categories described in Table 1 were used to classify the virulence level of the strains. Thus, the strains were classified as virulent (V) when they had at least one gene from each category and multi-virulent (MV) when they had two or more genes from each category.

The variability of the strains within a given group (sample, establishment, municipality, or state) was verified using the analysis of the number of distinct virulence profiles identified. The variability, virulence, and multi-virulence indices were determined by calculating the relative frequency for each group of strains evaluated. The classification for the variability index was adapted (Krumperman, 1983), being low for values <0.5, medium for values between 0.5 and 0.7, and high for values >0.7.

The analysis of the results of the virulence panel was performed based on the number of strains isolated in each analyzed sample, the establishment of origin, the place (city and state), and the seasonal aspects for the percentage description. For comparative analyses, the normality of the data was verified, followed by the application of Fisher's exact test or a Student's t-test to compare two variables and ANOVA or Kruskal–Wallis in the comparison of three or more variables. The program used was GraphPad Prism 8.0.1 with a 95% confidence interval.

A total of 14 genes associated with virulence were studied in all 359 strains of C. jejuni isolated from chicken carcasses. The evaluation by virulence category found that the concomitant frequency of genes linked to adaptation to stress was significantly higher (86.07%; 309/359) compared to the remaining categories, and those related to secretion were the least frequent (131/359; 36.49%) (Table 2). The majority of the isolates (292/359; 81.34%) had at least eight of the 14 studied genes (p < 0.05, Fisher's test), which portrayed the high percentage of virulent strains (229/359; 72.14%), of which 47.60% (109/229) were classified as multi-virulent. We found a total of 124 virulence profiles, of which five represented 40.1% (144/359) of the strains, with 76 strains belonging to a single multi-virulent profile (cdtA, cdtB, cdtC, luxS, pldA, flaA, dnaJ, cadF, neuA, ciaB, cstII, cbrA, htrA, and hcp). This fact determined the low variability index (I.Var. = 0.35) considering the 359 strains.

Profiles including genes linked to the biofilm formation/motility category were the least identified (60.5%; 75/124), and those associated with secretion systems (95.2%; 118/124) and adaptation to stress (96.0%; 119/124) were the most prevalent. A total of 16 clusters (A1 to A16) of virulence categories were identified, with A16 comprising 57 profiles (46%), of which 14 were classified as multi-virulent (Table 3).

The 359 strains used in our study came from the analysis of 114 carcasses, of which one to ten distinct colonies of C. jejuni were isolated per sample (Md = 3). Samples in which we obtained isolates from three or more colonies (59/114; 51.75%) were grouped according to the number of isolates to determine the variability index (I.Var.), whose value identified did not differ between groups (p = 0.4841). The elevated I.Var. average found (0.82) justified the individual analysis of each strain throughout the study. The I.Var. identified within each group differed significantly (p < 0.05), and this value was >0.75 for all samples (Figure 1).

Our study included the analysis of 46 slaughterhouses under federal supervision in 43 municipalities. We obtained a variation of one to 22 isolates in each location, and those establishments that presented three or more strains (36/46; 78.3%) were used to determine the general and individual I.Var. scores. In the interpretation of virulence and multi-virulence indices, all establishments were considered.

The average I.Var. score of all establishments was 0.87, with the majority of them (28/36; 77.78%) showing values above 0.7 (Table 4).

The distribution of establishments according to the classification of virulence and multi-virulence is described in Table 5. In general, and according to our expectations, the number of virulent strains per establishment and between establishments was significantly higher than that of multi-virulent strains (Student's t-test, p < 0.001). The significant majority of establishments (21/46; 45.7%) had 70–100% of the strains classified as virulent, and two had the same percentage range of multi-virulent strains, both located in state A. Percentages < 40% of virulent and multi-virulent strains were identified in 11/46 (23.9%) and 33/46 (71.8%) of the establishments, respectively.

The characteristics of C. jejuni were analyzed according to their states of origin: A, B, and C. State A had the lowest number of isolates (81/359; 22.6%) and profiles (42/124; 33.9%) compared to the other two states (p < 0.05), whose numbers were similar (Table 6).

Despite presenting more isolates, state C had the lowest relative frequency (Fisher's test, p < 0.05) of strains classified as virulent and multi-virulent compared to states A and B, as well as multi-virulent profiles (Figure 2).

The construction of the heat graph led to the identification of more evident hot zones in state A. Genes linked to biofilm formation were the least frequent but predominant in all states in December and most common in state A. The highest concentration of virulent and multi-virulent strains was also more expressive in state A, especially in November and December. Thus, state A presents itself as the main hotspot in terms of potential for the maintenance of virulent and multi-virulent strains of C. jejuni compared to the other states. The other genetic categories (SS, IC, and SA) showed a similar frequency in the three states (Figure 3).

Even with a significant number of profiles that contemplate only one strain in the three states, no state had a high I.Var. due to the large number of strains being grouped into a few profiles (Table 6). In state C, the variability below 0.7 is due to the existence of two profiles comprising 16 and 18 strains. In state B, the lower value is linked to a profile that includes 37 multi-virulent strains, and in state A, to a multi-virulent profile with 25 strains.

The georeferencing of the strains made it possible to determine the specific regions within each state where there is greater densification in terms of the number of isolates (Figure 4A). High relative frequencies of virulent (Figure 4B) and multi-virulent (Figure 4C) strains were evident in the geocoded hot tags.

In state A, the distribution of the number of isolated strains showed a more dispersed characteristic, but the greatest densification came from the north, northwest, and west regions. Paradoxically, this high dispersion masks its high rates of virulence and multi-virulence, especially in the central and metropolitan regions (Figure 4).

States B and C had a high and significant number of isolates, with a more convergent disposition. In Santa Catarina, the predominance was evident in the western, plateau, and southern regions. In Rio Grande do Sul, the strains were found exclusively in the northwest, northeast, and central-east regions of the state. Virulence characteristics are evident in both states, but the hot zones referring to the multi-virulence index are restricted only to the west and south of SC (Figure 4).

There was a variation of one to 23 isolates per city, with a median of eight isolates. To determine the I.Var. and the distribution of virulence profiles according to the classification of V and MV, the selected municipalities were those that presented three or more isolates, which corresponded to 79.1% (34/43) of the total, with seven in state A, 16 in state B, and 11 in state C. The average I.Var. identified in the municipalities was 0.78 and considered high. The significant number of cities with I.Var. >0.7 (23/34; 70.6%) determined the high mean value identified (Fisher's test, p < 0.05). The number of cities/states with virulent strains did not fluctuate significantly, regardless of the percentage range of virulent strains. However, a significant minority of cities (4/34; 11.8%) had 40% or fewer of their virulent strains. Controversially, cities with 0–40% of the multi-virulent strains were predominant (24/34; 70.6%) in our study. This high number is because 100% of the municipalities in state C fall within this range (Table 7).

The I.Var. and the virulence and multi-virulence characteristics were evaluated for the 359 strains, considering the months from October to December 2017 and January and March to July 2018.

High I.Var. are directly related to low rates of multi-virulence due to the odds ratio (Table 7). Our study showed that we had the lowest rates of multi-virulence in the months from January to July 2018, with a decreasing characteristic over time. This determined the high genetic variability between the strains due to the diversity of virulence profiles. In the months from October to December 2017, there was greater balance and equivalence in these indexes (Figure 5).

In general, the months considered hot and humid (October to December) in the southern hemisphere determine the highest number of isolates associated with high rates of virulence and multi-virulence, indicating the seasonal characteristic of C. jejuni (Table 8).