Thirty samples of edible mealworms and 30 samples of grasshoppers (boiled, dried, and salted) were purchased via the internet from dealers located in Europe (Belgium and the Netherlands) and Asia (Thailand). Ten mealworm and ten grasshopper samples from each country were collected and marked as follows: TN1-TN10 (mealworms from the Netherlands, Producer 1), TB1-TB10 (mealworms from Belgium, Producer 2), TT1-TT10 (mealworms from Thailand, Producer 3), GN1-GN10 (grasshoppers from the Netherlands, Producer 1), GB1-GB10 (grasshoppers from Belgium, Producer 2), and GT1-GT10 (grasshoppers from Thailand, Producer 3). All the samples were provided in sealed plastic containers and delivered at ambient temperature via international shipping. No information was available on the rearing and hygiene conditions of processing, transport and storage applied to these edible insects before marketing.

Total microbial DNA was extracted directly from the insect samples using PowerFood Microbial DNA Isolation Kit (Mo Bio Laboratories, Carlsbad, CA, USA) as described by Osimani et al. (2017a). The extracted DNA was quantified and checked for the purity using a NanoDrop ND 1000 (Thermo Fisher Scientific, Wilmington, DE, USA) and then standardized to 2 ng μL⁻¹ for qPCR assays and to 25 ng μL⁻¹ for PCR-DGGE analysis. The effective extraction of bacterial DNA was confirmed by conventional PCR amplification of 2 μL (50 ng) of extracted DNA suspensions in a My Cycler Thermal Cycler (BioRad Laboratories, Hercules, CA, USA) using universal prokaryotic primers 27F and 1495R as described by Osimani et al. (2015).

The equal portions of DNAs extracted from insects were mixed together with the goal of obtaining six pooled samples (TB, TN, TT, GB, GN, and GT), each representing an insect type (mealworms and grasshoppers) and country of origin (Belgium, the Netherlands, and Thailand). The amplification products obtained from the 27F-1495R primer pair as described above were purified using the Illustra GFX PCR DNA and Gel Band Purification Kit (GE Healthcare Life Sciences, Buckinghamshire, UK). Subsequently, 2 μL of the purified PCR products was reamplified using the universal prokaryotic primers U968GC (added with GC clamp) and L1401 (Muyzer et al., 1993; Randazzo et al., 2002) following the PCR conditions previously described by Aquilanti et al. (2013). Following amplification, 5 μL of the PCR reaction was loaded on a 1.5% agarose gel together with a 100 bp molecular weight marker (HyperLadder™ 100 bp) to check for the expected product size of 480 bp. Twenty microliters of these PCR amplicons were analyzed by DGGE (30–60% urea-formamide denaturing gradient; 4 h at 130 V) using DCode Universal Mutation Detection System (Bio-Rad Laboratories) as described by Garofalo et al. (2015). All DGGE bands visible under UV light were excised from the gel, and the DNA was eluted overnight at 4°C in 50 μL of molecular biology grade water (Garofalo et al., 2008) and reamplified via PCR as described above, but with the forward primer U968 lacking the GC clamp. The PCR products were sent to Genewiz (Takeley, UK) for purification and sequencing, and the obtained sequences were identified at species level as described above by Osimani et al. (2018c).

DNA extracted from five reference strains (Table 1), each carrying one of the carbapenem resistance genes under study, was used as positive control in the qPCR reactions and for the construction of qPCR standard curves.

Absolute quantification of each carbapenemase gene (bla_NDM−1, bla_VIM, bla_GES, bla_OXA−48, and bla_KPC) in the insect samples was performed by qPCR in a Mastercycler® ep realplex machine (Eppendorf, Hamburg, Germany) using the qPCR primers and cycling conditions described by Monteiro et al. (2012). To check for product specificity, all cycles were followed by a melt curve step analysis with temperature gradually increasing from 65 to 95°C by 0.2°C/s. Each qPCR reaction consisted of 4 μL (8 ng) of the extracted DNA; 5 μL of Type-it 2X HRM PCR Master Mix (Qiagen, Hilden, Germany) containing HotStarTaq Plus DNA Polymerase, EvaGreen Dye, an optimized concentration of Q-solution, dNTPs and MgCl₂; 0.2 μM of forward and reverse primers for each gene; and nuclease-free molecular biology grade water to a final reaction volume of 10 μL. The exogenous standards for each gene were prepared by qPCR amplification of the DNA extracted from the reference strains (Table 1) as described above but in a final reaction volume of 25 μL. The correct melting temperatures (Tm) and sizes of the obtained PCR products were checked by melting curve analysis and electrophoresis on a 1.5% agarose gel, respectively. The Illustra GFX PCR DNA and Gel Band Purification Kit (GE Healthcare Life Sciences) was used for the purification of the amplicons following the manufacturer's instructions. The quantity and the purity of the products were determined using the NanoDrop ND 1000 (Thermo Fisher Scientific). The calculation of each gene copy number was performed using an online calculator (www.idtdna.com) based on the mass and size of the purified qPCR products. The standard curves were created by qPCR amplification of 10-fold serial dilutions of exogenous standards. The amplification efficiencies were estimated from the slopes of the standard curves, and the correlation coefficients (R²) (Stolovitzky and Cecchi, 1996) were calculated automatically by Mastercycler® ep realplex software. To determine the qPCR detection limit for each gene, the standard curves were generated in the range from ~1 to 10⁷ gene copies per reaction.

For the absolute quantification of CEG, the DNAs extracted from mealworms and grasshoppers were run along with the 10-fold serial dilutions of the standards prepared as described above. The gene copy number of each gene detected in the analyzed insect samples was determined using the slope of the corresponding standard curves. The baseline and threshold calculations were performed automatically by the Mastercycler® ep realplex software. In addition to melting curve analysis, the correct sizes of the amplification products were checked by electrophoresis on 1.5% agarose gels using 100 bp DNA Ladder (HyperLadder™ 100 bp, Bioline, UK) as a molecular weight marker. Moreover, the accuracy of the amplification reactions was validated by the sequencing (Genewiz) of randomly selected positive samples (TB6, GT7, and GN1 for the bla_OXA−48 gene; GT2 and TT8 for the bla_NDM−1 gene; GT7 and GT8 for the bla_VIM gene). The resulting sequences were compared with those from the GenBank database (http://www.ncbi.nlm.nih.gov) using Basic Local Alignment Search Tool (BLAST) (Altschul et al., 1990). All qPCR reactions were performed in triplicate and the results were expressed as the mean gene copy number per ng of DNA ± standard deviation for each gene.

Descriptive statistics, calculated on 20 samples for each producer and on 30 samples for each insect species were carried out for the bla_OXA−48, bla_NDM−1, and bla_VIM gene copies by computing the means ± standard deviation.

After first checking for conformance to a normal distribution and identification of outliers, analysis of variance (ANOVA) was carried out using JMP statistical software version 11.0.0 (SAS Institute Inc., Cary, NC, USA) to test the following main effects: producers (Belgium, Thailand, the Netherlands), insect species (mealworm, grasshoppers) and producer × species.

Principal component analysis (PCA) was applied to discriminate among mealworms and grasshoppers coming from different producers (Belgium, the Netherlands, and Thailand), and the presence of genes related to resistance to carbapenems (bla_NDM−1, bla_VIM, bla_GES, bla_OXA−48, and bla_KPC). PCA was carried out using the Unscrambler 7.5 software (CAMO ASA, Oslo, Norway). The mean data were normalized to neutralize any influence of hidden factors. PCA provides a graphical representation of the overall differences in terms of distribution of genes between the insects coming from different producers.

To have an overview of the predominant bacterial species found in the edible insects considered in this study, the total microbial DNA was extracted from the samples, the DNAs were mixed in order to obtain six pooled samples, each representing an insect type (mealworms, grasshoppers) and country of origin (Belgium, the Netherlands, Thailand) and then analyzed by a culture-independent PCR-DGGE method. The results obtained are reported in Table 2.

The dominant species found in mealworms from Belgium and grasshoppers from the Netherlands belonged to the genus Staphylococcus, and species in the genus Bacillus were found in mealworms and grasshoppers from Thailand. Grasshoppers from Belgium were positive for Weissella cibaria/confusa/spp., while bacterial species with a percentage identity below 97% were found in mealworms from the Netherlands. It is interesting to note that the predominance of lactic acid bacteria (LAB), and in particular the Weissella spp., has been reported for processed and fresh grasshoppers from Belgium and the Netherlands (Stoops et al., 2016; Garofalo et al., 2017; Osimani et al., 2017a), thus suggesting that the specific rearing conditions may have selected for this microbial group or that this bacterial species is intrinsically associated with this edible insect.

The genera Staphylococcus and Bacillus identified among the other pooled samples of mealworms and grasshoppers may contain pathogenic species such as Staphylococcus aureus and Bacillus cereus, and these data are in agreement with other studies on the microbiota in fresh and processed mealworms and grasshoppers (Stoops et al., 2016; Garofalo et al., 2017). The presence of Staphylococcus members may be due to an environmental contamination occurring during human handling or processing. Indeed, these insects were boiled and salted, and since Staphylococcus spp. is a halophile bacterium usually predominating in environments with low microbial competition, it could have found conditions suitable for growth.

The lack of detection of bacterial DNA belonging to the Enterobacteriaceae has been already reported by Osimani et al. (2018c), although it is generally reported that Enterobacteriaceae represents a predominant bacterial group in the edible insect gut microbiota (Stoops et al., 2016; Garofalo et al., 2017; Osimani et al., 2017b). These data suggest that good manufacturing practices were applied during rearing and processing of the insects and/or that a degutting step may have been applied. A further explanation is that members of the Enterobacteriaceae family could be present within the processed insect samples but with a lower abundance in respect with other microbial groups. This latter hypotesis is supported by the fact that these microorganisms had previously found below the detection limit of microbial counts (<1 Log cfu g⁻¹) in the same samples (Osimani et al., 2017b,c).

Additionally, the lack of detection of bacterial DNA belonging to Pseudomonadaceae is unusual, but it is possible, as suggested for Enterobacteriaceae, that PCR-DGGE was not able to detect members of this microbial group if they were in the minority relative to the others.

This study represents the first report on the screening of five carbapenemase encoding genes (bla_NDM−1, bla_VIM, bla_GES, bla_OXA−48, and bla_KPC) in processed edible mealworms (T. molitor L.) and edible grasshoppers (L. migratoria migratorioides) from producers located in Europe (Belgium and the Netherlands) and Asia (Thailand). The identification of the genes coding for these carbapenemases in the samples of edible insects under investigation was conducted by using qPCR. As previously underlined by Monteiro et al. (2012), molecular assays are considered the best solutions for the rapid detection of carbapenem resistance genes and for the identification of the resistance mechanism. The detection limit, defined as the lowest concentration at which 95% of the positive samples are detected was <10 gene copies per reaction for all the genes. The efficiencies of the qPCR reactions were 1.00 for the genes bla_OXA−48, bla_VIM, and bla_GES; 1.01 for the gene bla_KPC and 0.96 for the gene bla_NDM−1. The R² was 1.000 for the genes bla_OXA−48 and bla_NDM−1; 0.999 for the gene bla_GES, 0.996 for the gene bla_KPC and 0.995 for the gene bla_VIM. Moreover, the specificity of the primers used for the amplification of carbapenem resistance genes was confirmed by the results of the sequencing of randomly selected positive samples, which showed >97% similarity with the corresponding gene sequences deposited in the GenBank database. In more detail, the results of the BLAST analysis for the bla_OXA−48 gene showed 98% of the similarity with the sequences deposited in GenBank such as K. pneumoniae (CP031374), Citrobacter freundii (MG430338), Enterobacter ludwigii (MG436907), Pantoea agglomerans (MG436898), Escherichia coli (NG_055499), Shewanella sp. (NG_055475), and Proteus mirabilis (KT175900); for the bla_NDM−1 gene 99% similarity with the Enterobacter hormaechei (AP018835), K. pneumoniae (AP018834), Raoultella planticola (MH257689), E. coli (CP021206), Pseudomonas aeruginosa (KT364224), and Acinetobacter baumannii (KU180703); and for the bla_VIM gene 98% similarity with the K. pneumoniae (MH071811), C. freundii (NG_061412), Paenibacillus sp. (KR822172), E. coli (MF169879), E. hormaechei (LT991955), Kluyvera cryocrescens (MG228427), Alcaligenes faecalis (KY623659), Klebsiella oxytoca (NG_050362), and Enterobacter cloacae (CP030081).

The results of the qPCR quantification of carbapenem resistance genes in samples of ready-to-eat edible mealworms and grasshoppers produced in the Netherlands, Belgium and Thailand are reported in Tables 3, 4.

Regarding mealworms, none of the samples were positive for the genes bla_GES, bla_KPC, and bla_VIM, while only sample TB6 was found positive for bla_OXA−48 (3% positivity) and samples TN2, TN8, and TT8 were positive for bla_NDM−1 (10% positivity) (Table 3).

Regarding grasshoppers, the genes bla_GES and bla_KPC were not detected in any of the analyzed samples while only two samples from Thailand (GT7 and GT8) were positive for the bla_VIM gene (7% positivity). Interestingly, a high prevalence of bla_OXA−48 was noted (57% positivity), followed by bla_NDM−1 (27% positivity) (Table 4). Specifically, bla_OXA−48 was prevalent in 80% of the samples from Belgium, in 50% of the samples from the Netherlands and in 40% of the samples from Thailand. The highest frequency of bla_NDM−1 was found among samples from Thailand (40%), followed by samples from Belgium (30%) and the Netherlands (10%) (Table 4).

All the insect samples analyzed in this study were previously screened for the presence of 12 selected genes coding for resistance to tetracyclines [tet(M), tet(O), tet(S), and tet(K)], macrolide-lincosamide-streptogramin B (MLS_B) [erm(A), erm(B), erm(C)], vancomycin (vanA and vanB), beta-lactams (blaZ and mecA) and aminoglycosides [aac(6′)-Ie aph(2″)-Ia referred as aac-aph] through PCR and nested PCR assays (Osimani et al., 2017b,c). It is interesting to note that the mealworms samples TN2 and TN8 from the Netherlands were also found to be positive for the presence of genes coding for resistance to tetracyclines [tet(M), tet(K), tet(S)] and erythromycin [erm(B), erm(C)], while sample TB6 was positive for genes coding for resistance to tetracyclines [tet(M), tet(K)], and sample TT8 from Thailand was positive for genes coding for resistance to tetracyclines [tet(K)], erythromycin [erm(B)], and aminoglycosides (aac-aph) (Osimani et al., 2017b). Moreover, among grasshoppers, almost all of the samples that were found to be positive for bla_OXA−48, bla_NDM−1, and bla_VIM previously showed positivity for AR genes coding for resistance to tetracyclines [tet(M), tet(S), tet(K)], erythromycin [erm(B), erm(C)], aminoglycosides (aac-aph) and beta-lactams (blaZ) (Osimani et al., 2017c).

The average levels of gene copies ng⁻¹ in the 60 samples of edible insects were as follows: 0.59 ± 2.39 with a minimum value of 0 and a maximum value of 16.56 for bla_OXA−48, 2.94 ± 11.53 with a minimum value of 0 and a maximum value of 70.37 for bla_NDM−1, 7.25 ± 50.85 with a minimum value of 0 and a maximum value of 392.25 for bla_VIM.

Descriptive statistics on 20 samples from each producer are shown in Table 5. In addition, descriptive statistics on 30 samples from each insect species are reported in Table 6.

The analysis of variance (Table 7) showed that all the variables (producers, species, and producers × species) had significant effects (P < 0.05) on the frequency of bla_NDM−1, whereas for bla_OXA−48 and bla_VIM, no significant effects were detected for the same source of variation.

Regarding the distribution of bla_OXA−48 and bla_VIM, multiple comparisons between ACC Least Square Means (LSM) carried out using the Tukey test showed no significant differences among samples from different producers or insect species. Regarding bla_NDM−1, multiple comparisons (Tukey HSD) showed that insect species, but not the origin of the sample, had a significant correlation (P < 0.05) with the frequency of the gene.

PCA did not discriminate between the presence of genes encoding resistance to carbapenems among mealworms and grasshoppers coming from different producers. In contrast, in the previous studies on the occurrence of transferable ARs in ready-to-eat edible insects, PCA showed a differentiation among producers, thus suggesting that different rearing and clinical practices associated with different countries may have played a role in the variability observed (Milanović et al., 2016; Osimani et al., 2017b,c).

As reported by Schlüter et al. (2017), it is presumed that the rearing and processing conditions applied to edible insects will comply with the same food safety regulations as for livestock farming. The use of carbapenems is prohibited in food-producing animals in all countries (OIE, 2015; Webb et al., 2016). Notwithstanding, scientific studies reporting CPE and CEG in livestock and their environment are progressively more frequent (Guerra et al., 2014; Webb et al., 2016; Zurfluh et al., 2016; Fischer et al., 2017). Furthermore, in the last EFSA report on antimicrobial resistance in zoonotic and indicator bacteria from humans, animals and food in 2015, the presumptive extended-spectrum beta-lactamase (ESBL)-/AmpC-/carbapenemase-production in Salmonella and E. coli was monitored in humans, meat (pork and beef), fattening pigs and calves for the first time (EFSA and ECDC, 2017). Varying occurrence/prevalence rates of ESBL-/AmpC-producers were observed between countries, and carbapenemase-producing E. coli were detected in single samples of pig meat and from fattening pigs from two Member States (EFSA and ECDC, 2017). These data indicate that other antimicrobial classes could indirectly select CPE outside the hospital setting and that the rapid dissemination of CPE is also promoted by CEG located on plasmids transmissible by horizontal gene transfer events (Tzouvelekis et al., 2012; Woodford et al., 2014). As reviewed by Caniça et al. (2015), AR comprises a dynamic network that involves several environmental niches (e.g., water, soil, and plants) and different reservoirs (e.g., husbandry, hospitals, wild animal, human settings, human hand, food and global trade in foodstuffs) in which the path of dissemination and dynamics of AR genes has to be taken into consideration in order to understand and prevent the AR transmission and spread. Therefore, it is possible to hypothesize that, irrespective of the use of carbapenems in the edible insect rearing, the CEG may derive from the substrates used for feed or from surfaces and hands of operators or from treatments applied for processing, in addition to transport and storage. It is also interesting to note that grasshoppers and mealworms have different dietary habits since grasshoppers are grass-feeders whereas mealworms are usually reared on cereal-based matrices; therefore, the differences in terms of presence and distribution of CEG among these insect species may derive from different rearing practices and substrates.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.