Microbiological and safety evaluation of green table olives marketed in Italy

Abstract

The microbiological and safety conditions of green table olives sold on the Italian market were evaluated on 40 samples, 20 loose and 20 packed in containers, purchased at street markets and supermarkets. The olives were analyzed for microflora and food safety indices, and for aflatoxin B₁ and ochratoxin A occurrence, and the results showed acceptable security. There was wide heterogeneity in the microflora, the numerical values being in relation to the olive type. The microbial population was dominated by the yeasts and lactic bacteria responsible for the fermentation process, and their numbers, together with the metabolic activity, led to conditions unfavorable for the development of pathogenic microflora; hence, such microflora were always absent. However, fecal coliforms were occasionally found in cracked olives. Lactic Acid Bacteria and yeast classification was performed at the species level by molecular techniques. The mycotoxin aflatoxin B₁ was detectable in 25% of the olive samples, in the range of 0.40–0.55 μg/kg, while ochratoxin A, present in 58% of the samples, was in the range of 0.20–3.90 μg/kg.

Introduction

The olive (Olea europaea L.) is native to Asia Minor, but has been cultivated in the Mediterranean basin for approximately 6,000 years. Its cultivation is widely diffused throughout the Mediterranean countries, and plays an important role not only through its contribution to the economy but also from the cultural and environmental points of view. Indeed, the olive tree is of nutritional, social and economic importance to the populations of many countries. Furthermore, it is responsible for many elements in the landscape, and contributes to soil conservation in many hilly areas, significantly contributing to the fight against desertification in areas of very hot climate. In the EU, Italy is the third biggest producer, after Spain and Greece, and in 2007-2008 its production was about 80,000 tons, 44% of this being in Sicily (Fig. 1); the per capita yearly consumption is about 3 kg, one-third of the domestic production (UNAPROL 2009).

Fig. 1

Distribution of table olive production in Italy, 2007-2008 (Source: UNAPROL 2009)

Numerous studies have shown that the populations of certain Mediterranean regions, where large amounts of olives and olive oil are consumed, tend to have a decreased incidence of coronary heart disease and some types of cancer, and this have been put down to the olives’ tocoferol and polyphenol content (Dell’Agli et al. 2008; Han et al. 2009; Visioli et al. 2002).

The market offers various kinds of table olives, and there are varieties to satisfy all tastes, even the most sophisticated of palates. The appeal of olives lies in their flavor, texture, and aroma, and there is a complexity of flavors varying from sweet to sour, to bitter, and to pungent. With regard to the safety of such table olives, one of the Mediterranean area’s most important fermented foods, many investigations have been carried out (Belaiche 2001; Bircan 2006; Caggia et al. 2004; Cantoni et al. 2003; El Adlouni et al. 2006; Ghitakou et al. 2006; Le Tutour et al. 1984; Leontopoulos et al. 2003; Roussos et al. 2006). Food quality is evaluated to ensure product safety, and such evaluations highlight situations such as microbial contamination and the absence/limited presence of potentially toxic substances like mycotoxins, namely aflatoxins and ochratoxin A, which are toxic secondary metabolites produced by molds belonging to the Penicillium and Aspergillus genera. The International Agency for Research on Cancer (IARC 1993) has classified aflatoxin B₁ (AFB) as a carcinogenic agent to humans (Group 1), and ochratoxin A (OTA) as a possible human carcinogen (Group 2B); OTA is nephrotoxic and seems to be involved in Balkan Endemic Nephropathy (O’Brien and Dietrich 2005).

Table olive quality is closely connected to the fermentation process, which occurs through microorganisms naturally present on the drupes, in the brine or in starter cultures used to confer the distinctive organoleptic characteristics of the end product (Arroyo-López et al. 2008, 2009; Ciafardini et al. 1994; Comi et al. 2000; Durán-Quintana et al. 1999; Fernández-Gonzalez et al. 1993; Montaño et al. 1993; Panagou and Katsaboxakis 2006; Panagou et al. 2008; Rodriguez-Gómez et al. 2010; Sánchez-Gómez et al. 2006). Traditionally, these microorganisms are not considered harmful; however, poor hygiene conditions during the retail phase can produce conditions favorable to the growth of alterative and pathogenic microorganisms (López-López et al. 2004; Pereira et al. 2008).

The aim of this research was to evaluate the microbial quality and the occurrence of AFB and OTA in several kinds of green olives, paying particular attention to the loose products sold at street markets where hygiene conditions are not always optimal.

Materials and methods

We selected 40 samples, each approximately 500 g, of green table olives of different kinds, sold loose (20), or packed in plastic pouches or glass containers (20) at street markets and in supermarkets, in Milan and Palermo, and assessed them for the presence of microbial populations and the occurrence of AFB and OTA.

Microbiological analysis

Loose olives

Ten g of olive pulp were removed aseptically, put into a sterile bag, homogenized in the bag with 90 mL of sterile 0.85% tryptone salt solution and blended for three minutes at 230g in a Seward 400 circulator Stomacher (PBI International, Milan, Italy). Decimal progressive dilutions were prepared, and the following microbial determinations were performed: Total Bacterial Count (TBC) by pouring plates in Plate Count Agar (PCA) (Merck, Darmstadt, Germany) and incubation at 30 ± 1°C for 48–72 h (ISO 4833 2003); fecal coliforms by double layer pour plates in Violet Red Bile Agar (VRB) (Merck), incubation at 44°C for 24 h (Hitchins et al. 1992); Lactic Acid Bacteria (LAB) pouring plates in Man Rogosa Sharpe (MRS), incubation at 30°C for 48 h under anaerobic conditions (gas pack) (De Man et al. 1960); yeasts and molds by agar spread plates in Yeast extract Glucose Chloramphenicol (YGC) (Merck ), incubation at 25–28°C for 3–5 days (ISO 6611 1992). As safety indicators, Salmonella sp. (ISO 6579 2002), Listeria monocytogenes (ISO 11290-1 2004) and Staphylococcus aureus (ISO 6888-1 2003) were determined.

All the microbiological determinations were carried out in triplicate, and the results were expressed as the average of colony count units per gram (CFU/g).

Packed olives

Commercial sterility was verified on each sample, according to the French normative (AFNOR 1997a).

Microorganism identification

Bacteria

All the colonies that grew on the last countable dilution of MRS (LAB) and PCA (TBC) media were isolated on MRS broth and TSB, respectively, and PCA. The TBC microorganisms were differentiated by their cellular morphology and motility using phase contrast microscopy (×480), Gram staining, catalase and oxidase test, and oxidative-fermentative glucose metabolism. Identification was then made at the genus level using the API System (bioMérieux, Craponne, France). The LAB were grouped on the basis of morphology, CO₂ production from glucose and gluconate, NH₃ production from arginine and temperature of growth. Classification at the species level was by molecular techniques.

Reference strains

The following reference strains were included in this study: Enterococcus faecalis NCDO 585, Enterococcus faecium NCDO 942, Lactobacillus paraplantarum ATCC 14917T, Lactobacillus plantarum ATCC 4008, Lactobacillus pentosus LMG 10755T, Leuconostoc mesenteroides LEUm1 (Department of Food Science Technology and Microbiology collection), Pediococcus acidilactici DSM 20284, Pediococcus damnosus DSM 20331T, Pediococcus pentosaceus ATCC 33314. The reference strains and the isolates were cultured in appropriate media, maintained at 4°C and sub-cultured monthly. Cultures were also stored in 20% glycerol at −70°C.

DNA extraction and PCR protocols

Total DNA extraction was performed on 100 μL of overnight broth culture as described by Mora et al. (2000); DNA solutions were stored at −20°C until analyzed. All the PCR reactions were performed in a 50 μL volume containing approximately 50–100 ng of bacterial genomic DNA, 5 μL of 10X PCR reaction buffer, 200 μM of each dNTP, 2 mM of MgCl₂, 0.5 μM of each primer and 0.5 U of Taq polymerase (Amersham Pharmacia Biotech, Uppsala, Sweden). Primer sequences obtained from a commercial supplier (PRIMM Laboratory, Milan, Italy) were used. 16S rDNA region amplification was performed using the universal primer set 16SF–16SR (16SF, 5′-AGAGTTTGATCCTGGCTCAG-3′; 16SR, 5′-CTACGGCTACCTTGTTACGA-3′) according to Scarpellini et al. (2002). RecA region amplification was performed to differentiate lactic acid rods according to Torriani et al. (2001). Enterococcus spp. was differentiated by means of species-specific primers, as previously described (Franzetti et al. 2004). Following amplification, 5 μL of product was electrophoresed at 5 V/cm (2% agarose gel, 0.2 μg/mL ethidium bromide) in TAE buffer. All amplification reactions were performed in PCR TGradient Thermocycler (Biometra, Göttingen, Germany).

16S rDNA sequence analysis

After amplification of 16S rDNA from extracted DNA, the PCR product was purified according to the instructions of a commercial Qiaquick kit (Qiagen, Valencia, CA, USA); the strains were sequenced with the following primer mix: [6SF, 16SR, and 926F (5′-AAACTY(1)AAAK(2)GAATTGACGG-3′)] (Stackebrandt and Goodfellow 1991). The reaction mix was prepared according to the instructions of a commercial DY Enamic™ ET Terminator Cycle Sequencing kit (Amersham Bioscence, Piscataway, NJ, USA) and the next amplification was performed in a model 310 automatic DNA sequencer (Applied Biosystems, Foster City, CA, USA). The cycle used was 30 cycles of 20 s at 95°C, 15 s at 50°C and 1 min at 60°C. The obtained sequences were elaborated by the software Chromas 2.13 (Technelysium, Helensvale, Queensland, Australia) and the results were compared with the sequences found in the NCBI gene bank (http://www.ncbi.nlm.nih.gov/).

Yeasts

For identification, all the colonies of differing morphology were isolated in malt agar, and initially characterized according to Van der Walt and Yarrow (1984). Identification at the species level was done by molecular techniques.

DNA extraction and PCR protocols

Total DNA extraction was performed on 500 μL of broth culture as described by Querol et al. (1992); DNA solutions were stored at −20°C until analyzed. All PCR reactions were performed in a 100 μL volume, containing 10 μL of 10X PCR reaction buffer, 200 μM of each dNTP, 0.5 μM of each primer and 1 U of Taq polymerase (Amersham Bioscence). Primer sequences obtained from a commercial supplier (PRIMM Laboratory) were used; the 18S–28S intergenic spacer (ITS) region amplification was performed using the universal primer set ITSL1 (5′-GTTTCCGTAGGTGAACCTGC-3′), ITSL2 (5′-ATATGCTTAAGTTCAGCGGGT-3′) according to Montrocher et al. (1998). Following amplification, 5 μL of product was electrophoresed at 5 V/cm (2% agarose gel, 0.2 μg/mL ethidium bromide) in TAE buffer. All the amplification reactions were performed in a PCR TGradient Thermocycler (Biometra, Göttingen, Germany).

18S–28SrDNA sequence analysis

After amplification, the PCR product was purified according to the instructions of a commercial Qiaquick kit (Qiagen, Valencia, CA, USA); the strains were amplified with one primer: ITSL1 or ITSL2. The reaction mix was prepared according to the instructions of a commercial DY Enamic™ ET Terminator Cycle Sequencing kit (Amersham Bioscence,) and the next amplification was performed in a model 310 automatic DNA sequencer (Applied Biosystems,). The cycle used was 30 cycles of 20 s at 95°C, 15 s at 50°C and 1 min at 60°C. The obtained sequences were elaborated by the software Chromas 2.13 (Technelysium,,) and the results were compared with sequences found in the NCBI gene bank (http://www.ncbi.nlm.nih.gov/).

pH measurement

The pH measurement of the olive homogenate was performed with a digital pHmeter model 3510 (Jenway, Danmow, UK) according to the French normative (AFNOR 1997b).

Mycotoxin analysis

Extraction and purification

One hundred grams of each olive sample were destoned by hand and homogenized to a paste by a refrigerated Waring Blender. For AFB and OTA determination, 20-g aliquots of olive paste were extracted and purified according to El Adlouni et al. (2006). The final chloroform extract was evaporated under vacuum, using a rotary evaporator in a 35°C water bath. The residue was solubilized in 2 mL methanol and then subdivided into two aliquots, each of 1 mL. The methanol solutions, diluted with PBS buffer, were passed, for cleanup, through an Easi-Extract™ Aflatoxin and an Ochraprep^® (R-Biopharm Rhône, Glasgow, Scotland) immunoaffinity column, respectively, operating according to the manufacturer’s instructions. The buffer was prepared dissolving 0.2 g KCl, 0.2 g KH₂PO₄, 1.16 g anydrous Na₂HPO₄, 8.0 g NaCl in 950 mL distilled water, and after adjusting the pH to 7.4 the solution was made to 1 L. The flow rate was regulated by a Resprep™ SPE Manifold Vacuum Unit (Restek, Bellefonte, USA). AFB desorption was performed with 3 mL of methanol, and OTA desorption with 3 mL of methanol:acetic acid (98:2) using the back flushing technique; the eluates were evaporated to dryness under nitrogen. For AFB determination, the residue was derivatized with trifluoracetic acid according to Park et al. (1990), the final volume being 200 μL. For OTA determination, the residue was re-suspended in 200 μL of HPLC mobile phase, i.e. sodium acetate 4 mM/acetic acid (19:1):acetonitrile (52:48). Each sample was extracted in triplicate.

Chemicals and reagents

AFB and OTA crystalline standards were purchased from Sigma (St Louis, MO, USA). Reagents were of the highest quality available, and solvents of high performance liquid chromatography grade.

Standard solutions

Stock solutions were prepared and quantified according to AOAC methods (2007). AFB working solutions (ranging from 1 to 200 μg/L) were prepared by evaporating aliquots of the stock solution at 35°C under nitrogen stream and derivatized according to Park et al. (1990). OTA working solutions (ranging from 0.5 to 100 μg/L) were prepared by evaporating aliquots of the stock solution at 35°C under nitrogen stream, and dissolving the residue in an appropriate volume of the LC mobile phase.

HPLC analysis

HPLC determination was done using a Perkin Elmer Series 200 Liquid Chromatograph (Perkin Elmer, Norwalk, CO, USA), equipped with a 20-μL loop, connected to a spectrofluorimetric detector, Perkin Elmer Fluorescence Detector Series 200, and to a TurboChrom workstation. The wavelengths for excitation and emission were 360 and 440 nm for AFB and 330 and 460 nm for OTA, respectively. The separation was performed on a Supelcosil LC-18 (15 cm × 4.6 mm) column, connected to a Supelcosil Cartridge LC-18 (2 cm × 4.6 mm) (Supelco , Bellefonte, PA, USA). Chromatography was conducted isocratically using, as mobile phase, water:methanol:acetonitrile (66:17:17) at a flow rate of 1 mL/min for AFB, and sodium acetate 4 mM/acetic acid (19:1):acetonitrile (52:48) at a flow rate of 1.2 mL/min for OTA elution.

Results

Microbiological analysis

Loose olives

Table 1 shows the results of the microbiological determinations performed on the green table olives. The most interesting microbial indices are TBC, yeasts and LAB. Moulds were always lower than the minimum determinable value, i.e. 10 CFU/g. Faecal coliforms were found occasionally, and only in cracked olives (samples H, L, M, O), the values ranging between 10²-10⁴ CFU/g. No pathogenic forms were found in any of the samples.

Table 1 Microbiological determination (CFU/g) and pH of green table loose olives

In samples A–G, olives in brine, TBC ranged between 3.0 × 10³ and 4.9 × 10⁵ CFU/g, with a mean value of 1.7 × 10⁵ CFU/g. To be noted was the presence of LAB, which was higher than TBC in some samples. Yeasts were the most variable microbial group: their numbers ranged from a minimum value of 2.0 × 10² CFU/g in sample D, to the maximum of 2.1 × 10⁵ CFU/g in sample B, where their numbers were higher than TBC and LAB. In the seasoned olives (H–V), both whole and cracked, the addition of aromatic ingredients (garlic, chili, wild marjoram, etc.) led to an increase in all the microbial indexes. The cracked samples (H–O) had the most significant increase in yeasts, as well as the most abundant microbial group. The mcroorganisms constituting the TBC were classified at the genus level by biochemical tests. Indeed, the Total Bacterial Count was constituted by 80% of Gram-negative fermentative rods belonging to the Enterobacteriaceae family coming from water and the vegetal environment. Pantoea sp. was the most frequently found genus, while the genus Enterobacter was present in minor numbers. The remaining 20% were Gram-positive catalase-positive forms equally subdivided between aerobic cocci belonging to the genus Micrococcus and anaerobic facultative spore-forming rods belonging to the Bacillus genus.

Lactic acid bacteria were mainly represented by homofermentant cocci and facultative heterofermentant rods. All the isolates underwent partial 16S rDNA sequencing, and the results are shown in Table 2. Among the rods, four strains were classified as Lactobacillus casei; all the other isolated rods showed the same nucleotide sequence and homology with the 16S rDNA gene sequences of 99% of L. plantarum, L. pentosus and L. paraplantarum. The discrimination of these species was obtained by amplification of the recA genus. L. plantarum, L. pentosus and L. paraplantarum give three different amplicones of lengths 318 pb, 218 pb, and 107 pb, respectively, in accordance with the literature (Torriani et al. 2001). All the tested strains gave a characteristic band of 218 pb typical of L. pentosus (Fig. 2). Most cocci were classified as Pediococcus acidilactici, 12 strains as Pediococcus parvulus, and 14 as Leuconostoc mesenteroides. For four strains from samples A and F, partial 16S rDNA sequencing failed to distinguish the species. Amplification with species-specific primers gave a band of the 280-bp fragments, including the V9 region characteristic of Enterococcus faecalis (Fig. 3).

Table 2 LAB identification according to 16S rDNA partial sequencing

Fig. 2

Agarose gel electrophoresis of PCR products from DNA using pREV, plantF, paraF, pentF primers. Lane 1 L. pentosus LMG 10755T. Lane 2 L. plantarum ATCC 4008. Lane 3 L. paraplantarum ATCC 14917T. Lanes 4–8 A1, G1, H1, Q5, V5. Lane 9 L. casei (B1). M marker Gene Ruler 100 pb (MBI Fermentas)

Fig. 3

Agarose gel electrophoresis of PCR products from DNA using species-specific primer for Enterococcus faecalis. Lane 1 Enterococcus faecalis NCDO 585. Lanes 2–3 D2, F1. Line 4 Enterococcus faecium NCDO 942. M marker Gene Ruler 100 pb (MBI Fermentas)

None of the isolated yeasts produced gas from glucose, Table 3 shows the results of the partial sequencing of the region 18S–28S rDNA (ITS). The genus Pichia was present on all the samples: Pichia membranifaciens was the most frequently found species, followed by P. fermentans, P. anomala and P. norvegensis. On the seasoned products we found a wide variety of yeasts (Galactomyces geotrichum, Rhodotorula mucilaginosa, Metschnikowia pulcherrima, Filobasidium capsuligenum), probably coming from the added ingredients.

Table 3 Yeast identification according to18S-28SrDNA partial sequencing

Packed olives

All the tested samples were stable at 32 and 55°C and showed a pH value lower than 4.5, important to control the growth of pathogen microorganisms.

Mycotoxin recovery and quantification

The standard curve for AFB was linear, ranging from 1 to 200 μg/L, and the coefficient of linearity (R ²) was 0.999; the standard curve for OTA was linear, ranging from 0.5 to 100 μg/L, and the coefficient of linearity (R ²) was 0.998. The LOD and LOQ were 0.1 and 0.4 μg/kg for AFB and 0.05 and 0.2 μg/kg for OTA, respectively.

Three olive paste samples were spiked with AFB at 0.4, 1.0, and 2.0 μg/kg and with OTA at 0.2, 1.0, and 5.0 μg/kg and extracted in triplicate. The AFB average recoveries were 91.3 ± 2.1 (2.3 RSD%), 93.7 ± 1.5 (1.6 RSD%), 96.3 ± 5.5 (5.7 RSD%) and the OTA average recoveries were 77.7 ± 3.8 (4.9 RSD%), 81.3 ± 2.5 (3.1 RSD%), 79.0 ± 3.6 (4.6 RSD%), respectively. One green olive sample, without stone, was spiked with 2.0 μg/kg of AFB or 5 μg/kg of OTA and analyzed six times on three successive days by the same operator and with the same HPLC system. The average concentrations were 1.8 ± 0.1 μg/kg for AFB and 3.8 ± 0.3 μg/kg for OTA, respectively.

Mycotoxin occurrence

The toxin distribution is shown in Fig. 4.

Fig. 4

Distribution of aflatoxin B₁ and ochratoxin A contamination levels, μg/kg, in green table olives

AFB

The toxin was found in 4 loose olive samples at levels ranging from 0.40 to 0.55 μg/kg, and in 6 packed olives samples at levels ranging from 0.40 to 0.46 μg/kg.

OTA

The toxin was found in 12 loose olive samples at levels ranging from 0.20 to 3.90 μg/kg, and in 11 packed olives samples at levels ranging from 0.48 to 3.40 μg/ kg.

AFB and OTA results were not corrected for recovery.

Discussion

The different kinds of loose olives tested in our work showed similar microbial compositions: LAB and yeasts, despite differing numerical ratios in the samples, were the most important microbial groups. Their numbers, metabolic activity and pH lower than 4.5 pose important hurdles for microbial pathogen growth (Salmonella spp., Listeria monocytogenes and Staphylococcus aureus). The presence of fecal coliforms, found only occasionally in cracked olives, confirms the uncertain hygiene conditions of the marketplace. With regard to the LAB, the dominant forms were homofermentative cocci: the species Pediococcus acidilactici, with a percentage of 38% was the most frequent, followed by Pediococcus parvulus (9%) and Enterococcus faecalis (3%). Not to be overlooked was the presence of heterofermentative cocci, represented by Leuconostoc mesenteroides, found at 11%. Rods were represented by facultative heterofermentative forms, such as Lactobacillus pentosus (36%). All are microbial species involved in the fermentation process, especially Lactobacillus pentosus for its β-glucosidase activity, and in the biological process of olive debittering. Among the yeasts, the no-gas-producer forms, the last to become established during the fermentation, were dominant; furthermore, the yeasts play an important role in the final organoleptic characteristics of the fermentation process (Arroyo-López et al. 2008). These results agree with Resolution No RES 2/91-IV/04 (IOOC 2004), which emphasizes the importance of respecting the GHP during production, and, especially, during the trade phase, permitting the presence of yeast and lactic acid bacteria during fermentation.

The packed olives were revealed to have a good situation: all respected the principle of commercial sterility according to the International Olive Oil Council (IOCC 2004).

Mold growth is a major worldwide problem in the food industry. Fungal invasion leads to loss in product quality, which is manifested in many ways: deterioration of color, texture, and taste, nutritional loss, and formation of toxic metabolic products such as mycotoxins. Molds were never found in our samples; however, it is well known that the presence of a toxigenic mold on a substrate does not automatically mean toxin presence, just as the absence of toxigenic molds does not guarantee toxin absence since toxins can persist long after mold growth has disappeared. The observed mycotoxin values were of the same magnitude as in the literature (El Adlouni et al. 2006; Ghitakou et al. 2006). With regard to AFB and OTA concentration, Student’s t test revealed no meaningful differences (p < 0.05) between the loose and packed olives; in this calculation, we assigned the LOD value to samples below the level of quantification.

Unlike for many other foods, the European Union has not set a maximum contamination level for aflatoxins and OTA in table olives (Commission Regulation 2006, 2010). Nevertheless, it is appropriate to limit the toxin content of food, particularly with respect to AFB, the most toxic compound among fungal secondary metabolites, and for which no tolerable weekly intake (TWI) has yet been established for humans. For OTA, the European Food Safety Authority (EFSA 2006) has indicated a TWI of 120 ng/kg bw, i.e. 7,200 ng for a human weighing 60 kg. In Italy, the weekly olive intake is about 60 g and therefore, considering the mean contamination value of the examined olives, the potential assumption is below 1% of the TWI.

Each effort to prevent mold growth and mycotoxin production along the food chain is important; for this reason, factors like temperature, salt concentration, pH, packaging modality, and hygienic conditions should be controlled to obtain a final product of high quality and safe.

In conclusion, concerning the microbiological situation of green table olives, those examined in the study were found to be of good quality from the food safety point of view. With regard to toxin presence, the observed level was low, but it must be emphasized that there is a constant need for skilful control in this kind of product since olives are main components in the Mediterranean diet.