Introduction

The use of elution-concentration methodologies for recovery and detection of virus from food matrices has implicated enteric viruses as important etiologic agents of diseases associated with contaminated foods worldwide (Havelaar et al., 2015; Mäde et al., 2013). More recently, such methodologies were validated and standardized for norovirus and hepatis A in four food classes including drinking water, fruits and vegetables, surfaces and bivalve molluscs (ISO/TS 15,216–1:2017).

Ready-to-eat foods such as sandwiches, vegetables, deli meat and dairy products have been associated with gastrointestinal outbreaks caused by norovirus in several countries, mainly due to the inappropriate handling of those foods (Stals et al., 2011a, b). Noroviruses, family Caliciviridae, are classified into genogroups (GI-X), which GI and GII causes the majority of illnesses in humans (Chhabra et al., 2019).

Also associated with food contamination (Rodriguez-Manzano et al., 2014), human adenovirus (HAdV), Adenoviridae family, are often found in wastewater worldwide, being considered as a potential indicator of human fecal contamination (Hewitt et al., 2013). In developed countries, periodic testing of raw foods using adenovirus as a bio-indicator of quality is recommended (Lu et al., 2012).

Although Brazilian studies drew attention to the possible role of cheese as a vehicle for enteric viruses affecting humans (de Castro Carvalho et al., 2020; Melgaço et al., 2018), tests for their detection in foods are still not required in Brazil.

Raw milk cheese is the dairy product more associated with foodborne diseases (Verraes et al., 2015). In France, dairy products are responsible for 7.5% of foodborne illnesses and 2.1% of outbreaks due to viral agents (InVS, 2012).

In Brazil, the raw milk Minas artisanal cheese is the oldest and most traditional one representing the main source of revenue of several smallholders, with a production rate of about 70,000 tons per year (Emater, 2004). Artisanal Coalho cheese from Northeastern Brazil is equally important. Thus, the aim of this study was to estimate the prevalence of HAdV and norovirus GI and GII of five artisanal Brazilian cheese types, made with raw milk, using a cross-sectional randomized survey.

Methods

Design, Location of the Study and Sample Planning

From October 2017 to April 2018, 100 ready-to-eat raw milk artisanal Brazilian cheese samples, obtained from 81 different agroindustries from Minas Gerais (MG) and 19 from Piauí (PI), states of Brazil (Fig. 1) were assessed for norovirus GI and GII, and HAdV.

Fig.1
figure 1

Map of Brazil and the location of the agroindustries producing the five cheese types sampled. Serro (circles), Canastra (squares), Cerrado (crosses), Triângulo Mineiro (triangles), Parnaíba (lozenges)

Full size image

The inclusion criteria required the participating agroindustries to be both manufacturers from a set of single-farm milk and to be registered in the state animal health agency.

There was a list of 235 rural family based cheese-processing agroindustries registered officially in five traditional producing regions of Brazil: Canastra (MG), Serro (MG), Cerrado (MG), Triângulo Mineiro (MG) and Parnaíba (PI). Then each one was marked with a specific number, and the sample of the target population to estimate the prevalence of viruses was determined by random drawing of 100 (42.5%) agroindustry samples. The sample was chosen using the Randbetween function (Microsoft Excel). One cheese was collected by agroindustry to represent it, since each one produces cheese from a set of single-farm milk. The sample size was based on a finite population of 235, margin of error of 7.5%, confidence level of 95% and an expected proportion of 50% Viruses positive cheese.

Data and Sample Collection

The cheese samples were collected for analysis in their original packaging and kept in isothermal boxes under refrigeration (< 4 ºC) during transportation to the laboratory.

Information on each agroindustry and respective samples were obtained using a structured questionnaire organized by a content group, focusing on socio-economic features, animal health, Good Agricultural Practices (GAP) and Good Handling Practices (GHP). Data about bacterial analyses of each cheese sample also were obtained from the Minas Gerais Agriculture and Livestock Institute (IMA) to assess possible associations with the presence of viruses.

Viral Concentration Method

Murine norovirus 1 (MNV-1), kindly provided by Herbert W. Virgin (Washington University School of Medicine) was spiked (10uL) in all samples as internal process control (IPC) before processing of viral concentration. It was used MNV-1 stock concentration of 1.13 × 104 genomic copies/Liter (gc/L). The cheese samples were processed using a swab and eluting in PBS buffer as described in the ISO/TS 15,216–1, 2017.

Nucleic Acids Extraction and Synthesis of Complementary DNA

Nucleic acids (RNA/DNA) were extracted from 140 µL of the concentrated samples, with a QIAamp viral RNA mini kit® (Qiagen, Valencia, USA), using an Automated System procedure following the manufacturer’s instructions QIAcube® (Qiagen, CA, USA). For RNA viruses (norovirus GI/GII and MNV-1) synthesis of complementary DNA were performed using random primers and Superscript III® Reverse Transcriptase (Invitrogen—Life Technologies, Carlsbad, CA, USA) following manufacturer instructions.

Quantitative Polymerase Chain Reaction (qPCR)

Human adenovirus (HAdV), norovirus GI and GII and MNV-1 were investigated by qPCR TaqMan® system (ABI PRISM 7500, Applied Biosystems) using a set of protocols previously described (Table 1). All samples were tested using undiluted and 1:10 diluted nucleic acid, both in duplicate, to assess and avoid possible qPCR inhibitors, with four qPCR reactions per sample. Negative and positive controls of the nucleic acid were also included in all procedures. For all determination, a specific DNA standard curve containing the target qPCR region for each of the viruses (gBlock Gene Fragment, Integrated DNA Technologies, Iowa, USA) was generated by a tenfold serial (107–100 genomic copies (gc)/µL). Samples that showed signals crossing the threshold line in both replicates until cycle threshold (Ct) value ≤ 40 presenting a characteristic sigmoidal curve were considered positives. Viral concentrations were expressed as genome copies per liter (GC/L) and calculated as the average of the duplicate data obtained, correcting for the dilution analyzed.

Table 1 Set of primers and probes used for quantitative and qualitative PCR according virus target
Full size table

Molecular Characterization

Positive samples presenting Ct ≤ 35 were also tested by the conventional PCR assay (Table 1) for further Sanger sequencing. Amplicons were purified using Wizard SV Gel and PCR Clean-Up System (Promega™, Madison, WI, USA) following the manufacturer’s recommendations. Nucleotide sequencing was performed using the dideoxynucleotide chain termination method with the ABI Prism Big Dye Terminator Cycle Sequencing Ready Reaction Kit 1, v. 3.1 and the ABI Prism 3730 Genetic Analyzer (Applied Biosystems™, Foster City, CA, USA). Clustal W in Mega X (Kumar et al., 2018) was used for edition and alignment of nucleotide sequences, which were compared to those available in the National Centre for Biotechnology Information (GenBank, http://www.ncbi.nlm.nih.gov/) database using the BLAST (Basic Local Alignment Search Tool). The MEGA X (Kumar et al., 2018) was also used to build the comparative analyses, evolutionary distances and dendrogram, which were performed using Maximum Likelihood method and Tamura 3 parameter model (T92) (Tamura, 1992), Gamma Distributed (G) with 2,000 bootstraps above 70%.

Epidemiological Analyses

The prevalence of viruses-positive artisanal cheese was determined with a 95% confidence interval, and univariate analyses were performed to assess factors associated with viruses-positive artisanal cheese (outcome). Mid-p exact tests were used to assess which explanatory variables were associated with the outcome, with a 5% significance level.

Results and Discussion

From a total of 100 samples, viral genome was detected in 43%, being 26 (26%, CI95% 17.74% to 35.73%) norovirus GI strains, 14 (14%, CI95% 7.87% to 22.37%) HAdV, and 3 (3%, CI95% 0.62%, 8.52%) both viruses. Viral concentrations ranged from 6.17 × 104 to 1.44 × 107 gc/L−1 and MNV-1, used as IPC, showed 100% success rate of recovery with efficiency of 10%. Accordingly, MNV-1 has been described as an efficient IPC (Stals et al., 2011a, b).

Although positivity results followed a strict criterion considering reproducibility of the results with replicas, the high Cts obtained did not allow the molecular characterization of all viruses detected. Out of 17 with qPCR positive results for HAdV, we were successful in the nucleotide sequencing of one (Gen Bank accession number MW713373). This exhibited high similarity with the HAdV group F (HAdV-41) strains, which is one of the most common enteric viruses involved in foodborne outbreaks (Greening & Cannon, 2016; Russell, 2009).

Concerning contamination of dairy products previously, HAdV was detected in the hands of food handlers at some farms suggesting potential source of contamination (Maunula et al., 2013).

Table 2 shows the univariate analysis for both outcomes: positivity for HAdV and norovirus. Noroviruses were more commonly associated with cheese samples with lower total coliform counts (p = 0.014); perhaps due to a confounding effect of the ‘humidity degree’ variable, since when stratified by this variable that association was no longer found (p = 0.12). Accordingly, HAdV-DNA was more commonly found in cheese samples with lower humidity degrees (bordeline p value = 0.056), but, unlikely noroviruses, this was more associated with higher faecal coliform counts (borderline p value = 0.06). This co-contamination of cheese was also found by other Brazilian studies (de Castro Carvalho et al., 2020; Melgaço et al., 2018). We found a trend towards a higher positivity rate for both viruses in the rainy season in this study, which was also showed among both water environment in Mexico (Rosiles-González et al., 2019) and human cases in Brazil (Amaral et al., 2015).

Table 2 – Univariate analysis for human adenovirus (HAdV) and norovirus positivity
Full size table

Norovirus GII strains were not detected in any cheese samples. GI strains may also be underestimated among humans and along the environment, since this genogroup has been associated with lower attack rate (Lopman et al., 2004) and less severe clinical presentation than GII-4 (Friesema et al., 2009; Huhti et al., 2011). Fankhauser et al., (2002) also showed a significant increase of GI strains in US, which was attributed to the application of newly designed primers, called region B primers, that are highly effective at detecting GI strains previously missed with region A primers. We are confident that detecting only GI among the samples tested was not the result of ineffective GII primers or probes, since GI and GII were successfully detected in one cheese (1.1%) each (Melgaço et al., 2018) in a previous study by our team using the same protocol. Thus, further research to access which noroviruses are prevalent in developing countries are needed.

The adenonoviruses, enteroviruses and hepatitis viruses (Johnson et al., 1990), and, more recently, norovirus (Langer et al., 2012), are among the major milk-associated pathogens transmitted by humans, and cheese is the main dairy product involved in foodborne outbreaks (Verraes et al., 2015). Cooling, freezing, thawing, acidification, and pasteurization, appear to be insufficient to inactivate norovirus within a food matrix, or on food surfaces (Mormann et al., 2010; Richards et al., 2012). Adenovirus also is resistant to both tertiary treatment and UV radiation (Thompson et al., 2003); however, temperatures like pasteurization were effective in inactivating this virus (Allard & Vantarakis, 2017).

Our findings of enteric virus contamination in Brazilian artisanal cheese types may be even more worrisome, since these foods are minimally processed and made from raw milk, undergoing no thermal processing, most of which marketed and consumed fresh (median of only 10 days of ripening among our sample). Thus, this study was a first step to assess the risk that such contaminations may pose to the consumer of these products, as well as to emphasize the need for good manufacturing practices, quality control systems in the dairy industry and markets.

Conclusion

HAdV and norovirus GI were found in samples of Brazilian artisanal cheese by a randomized survey. Thus, this work established representative baseline figures for virus prevalence in five types of ready-to-eat raw milk Brazilian artisanal cheese, to allow any monitoring trends, setting control targets and future local risk analysis studies.